Document zoEJJ0pYBOQoqpkw9RX7o83j6

7 0164cma223 8 February 20, 1986 9 9 9 9 .9 9 11 BEFORE THE 12 UNITED STATES DEPARTMENT OF LABOR 13 OCCUPATIONAL SAFETY AND HEALTH ADMINISTRATION 14 14 /J. 16 COMMENTS OF THE 17 CHEMICAL MANUFACTURERS ASSOCIATION fro}I P"7 3 ' a o u j 18 ON OSHA'S PROPOSED STANDARD FOR 19 20 OCCUPATIONAL EXPOSURE TO BENZENE Po l!y e Sj-O " 20 20 - Cm * " 20 22 23 24 Occupational Exposure to Benzene: ) ) 25 Proposed Rule, 50 Fed. Reg. 50512 ) Docket No. H-059-C 26 (December 10, 1985) ) 27 ) 28 28 28 28 31 Geraldine V. Cox, Ph.D. David F. Zoll, Esq. 32 Vice President and Vice President and 33 Technical Director General Counsel 34 35 Carol R. Stack, Ph.D. Of Counsel: 36 Administrator Neil J. King, Esq. 37 Biomedical and Environmental Wilmer, Cutler & Pickering 38 Special Programs 1666 K Street, N.W. 39 Washington, D.C. 20006 40 41 42 43 44 March 6, 1986 45 46 48 Chemical Manufacturers Association 49 2501 M Street, N.W. 50` Washington, D.C. 20037 51 (202) 887-1100 53 53 ,96 53 00ov 54 54 54 56 57 53 59 71 72 74 75 76 77 78 79 80 81 82 83 84 85 66 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 66 67 68 TABLE OF CONTENTS Introduction......................................................................................................................... I. Health Effect Issues................................................................................... A. Non-Malignant Health Effects....................................... 1. Central Nervous System and Hematotoxic Effects........................................................................................... 2. Teratogenic and Reproductive Effects.... B. Carcinogenicity.................................................................................... 1. The Qualitative Evidence of BenzeneRelated Leukemia Does Not Demonstrate an Increased Risk at Occupational Exposure Levels of 10 ppm and Belov...................................... a. The Rinsky Study.. b. The Wong Study.... c. The Dow Study............ 2. Epidemiological Studies Do Not Show an Association Betveen Occupational Benzene Exposures and Cancers Other Than Leukemia.......................................................................................... 3. Animal Bioassays................................................................... C. Mutagenicity, Cytogenetic Effects, and Hematological Effects at the Cellular Level.. 1* Mutagenicity.............................................................................. 2*, Cytogenetic Effects.......................................................... a. Chromosome Aberrations in Human Studies.............................................................................. b. Increased Production of SCEs and Micronuclei in Experimental Animal Test Systems............................................ -i- Page 00' 62 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 66 67 68 II. * III c. The Significance of Subclinical Cytogenetic Effects................................................. 3. Hematological Effects at the Cellular Level.......................................................................................................... Risk Assessments for Benzene..................................................................... A. Animal Versus Human Data.................................................................. B. Selection of Epidemiological Data........................................ C. The Crump and Allen Risk Assessment Should Be Preferred Over the Risk Assessments by White, et al. and IARC..................................................................................................... ET. The Crump and Allen Risk Assessment Conservatively Indicates That the Increased Risk of Leukemia From a Working Lifetime Exposure to 1 ppm Benzene Is Approximately An Order of Magnitude Lower at Exposure Levels of 1-10 ppm Than OSHA Has Assumed.................... E. Some Perspective on the Significance of the Risk........................................................................................................ Feasibility Issues................................................................................................. A. Profile of Current Petrochemical Industry Operations.......................................................................................................... B. Employee Exposure Profile............................................................... 1. Number of Employees Exposed to Benzene (and Person-Years of Benzene Exposure) in the Petrochemical Industry...................................... 2. Current Exposure Levels of Petrochemical Employees............................................................................................... C. Inventory of Benzene Emission Sources at Petrochemical Facilities and Identification of Job Assignments Where Use of Respirators May Be Needed To Comply with the Proposed Standard................................................................................................................ D. JRB Has Overestimated the Emission Reductions That Would Result from the Controls Identified in the Report................................................................................................... - ii - 00001559V MCD 61 62 167 163 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 IV. 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 V. 208 209 210 211 212 213 214 214 66 67 68 1. Baseline Case Assumptions............. .2 Control of Wastewater Emissions E. Control of Fugitive Emissions......................................... F. JRB's Conclusion. That a 1 ppm PEL Is Technologically Feasible in the Petrochemical Industry Is Unjustified....................... G. JRB Has Greatly Underestimated the Costs of Complying With a Standard Having a PEL of 1 ppm as an 8-Hour TWA.................................................................. H. Feasibility Considerations Dictate Setting the PEL Above 1 ppm, or at a Minimum, Establishing Compliance Criteria Which Account for Exposure Variability.................................................................................................. There Is No Basis for Adopting a Short-Term Exposure Limit in the Benzene Standard................................................................ A. Principles To Be Used in Determining Whether a STEL Is Needed................................................................................ B. The Best Available Evidence Does Not Support a Finding That Short-Term Benzene Exposures Consistent With the Proposed 8-Hour PEL Will Present a Significant Risk of Material Health Impairment.................................................................................................. 1. Non-Malignant Health Effects................................ 2. A Dose-Rate Effect for Benzene-Related Leukemia Has Not Been Demonstrated............... C. The Proposed Standard Would Protect Against High Short-Term Exposures Even Without the Adoption of a STEL............................................................................ Medical Surveillance.................................................................................... A. Employee Coverage.............................................................................. B. Frequency of Periodic Examinations............................. C. Required Elements of Medical Examinations.... - iii - 0o Page / w S' 9$ 61 62 215 216 217 218 219 220 221 222 223 224 225 22 6 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 247 247 247 247 247 247 247 247 247 247 247 247 247 247 247 247 247 247 247 66 67 68 1. Chest X-Rays.............................................................................. 2. Emergency Examinations................................................. D. Criteria for Referral to a Hematologist or Internist.......................................................................................... E. Medical Removal and Wage Rate Retention............... F. Comments on Appendix C - Medical Surveillance Guidelines for Benzene.................... 1. The General Provisions of Section V.A................................................................................. 2. The Hematology Guidelines of Section V.B................................................................................. a. The Hematology Guidelines Should Be Updated To Recognize the Use of Automated Blood Counters.................... b. The Hematology Guidelines Should Be Modified To Avoid Inaccurately Attributing Certain Indicia to Benzene Exposure.................................................... c. Various Other Statements in the Guidelines Should Be Altered in the Interest of Accuracy............................. Page iv 00 q0^ 249 BEFORE THE 250 UNITED STATES DEPARTMENT OF LABOR 251 OCCUPATIONAL SAFETY AND HEALTH ADMINISTRATION 252 259 259 260 261 ") 262 Occupational Exposure to Benzene: ) 263 Proposed Rule, 50 Fed. Reg. 50512 ) Docket No. H-059-C 264 (December 10, 1985) ) 265 ) 266 266 270 COMMENTS OF THE 271 CHEMICAL MANUFACTURERS ASSOCIATION 272 ON OSHA'S PROPOSED STANDARD FOR 273 OCCUPATIONAL EXPOSURE TO BENZENE 274 * 274 275 Introduction 276 276 280 These Comments are submitted by the Chemical Manufac- 281 turers Association ("CMA") in response to the Notice of Proposed 282 Rulemaking on Occupational Exposure to Benzene published in the 283 Federal Register of December 10, 1985. 50 Fed. Reg. 50512. CMA 284 is a non-profit trade association whose member companies repre- 285 sent more than 90 percent of the productive capacity for basic 286 industrial chemicals in the United States. Many of CMA's member 287 companies produce, use, or market benzene, and the petrochemical 283 facilities of CMA's members would be subject to the revised 289 benzene standard that OSHA has proposed. 290 291 Because of the substantial interest that CMA's member 292 companies have in regulatory determinations and actions relating 293 to benzene, CMA participated extensively in the 1977 rulemaking 294 proceeding, which eventuated in a standard that was set aside by 295 the courts. CMA also has participated in the activities that MCD OOJ5600 296 OSHA has undertaken or sponsored over the course of the past sev- 297 eral years to consider whether the existing benzene standard 298 should be revised. Thus, CMA provided information and views in 300 response to OSHA's Request for Information published in the 301 Federal Register of July 8, 1983 (48 Fed. Reg. 31412) and partic- 302 ipated actively in the discussions held under the auspices of the 303 Institute for Environmental Mediation. We welcome the 304 opportunity to present our views on the present proposal and hope 305 that they will be reflected in the final action that OSHA takes 306 ift this proceeding. 307 308 The rulemaking proposal raises a wide variety of 309 issues, many of which are of great concern to CMA. In these Corn- 310 ments, we will begin by addressing the critical health effect and 311 risk assessment issues, and will then proceed to deal with ques- 312 tions of feasibility and other matters raised in the rulemaking 313 notice. 314 317 I. Health Effect Issues 318 319 A. Non-Malignant Health Effects 320 322 1. Central Nervous System and Hematotoxic Effects 323 325 Tha^preamble to OSHA's rulemaking proposal quite 326 properly focuses on the cancer risk (more particularly, the 327 leukemia risk) that has been associated with exposure to benzene 328 in various occupational settings in the past. It is true, of 256 257 258 -2- 000015601 WCI> 329 course, that exposure to elevated levels of benzene also has been 330 associated with non-malignant health effects, such as central 331 nervous system effects, aplastic anemia, bone marrow depression 332 and cytopenias. However, these non-malignant health effects are 334 not ones that would be found at occupational exposure levels of 336 10 ppm and below. For that reason, they are not relevant to the 337 present proceeding. 339 340 For example, while exposures to benzene at levels of 341 50-150 ppm have been reported to produce headache, lassitude, and 342 weakness, exposures at levels of 25 ppm have no such effect.1/ 347 As EPA has observed: "Lower levels of benzene fi.e., below 50 348 ppm] do not seem to elicit these Xceivtral nervous system] 349 responses no matter how long the exposure."2/ 351 352 Similarly, long-term exposure to moderate or high 353 Revels of benzene (i.e., 40 ppm - 500 ppm) can produce signs of 354 hematotoxicity, including bone marrow depression, pancytopenia, 355 aplastic anemia and less severe blood abnormalities.3/ However, 358 the evidence does not indicate that these non-malignant 360 360 343 1/ See EPA, Draft Criteria Document for Benzene <(F*bru*rY 344 1984) at 51. Moreover, the mild central nervous system effects 345 that may occjar at benzene levels of 50-150 ppm ^appear to be rap346 idly reveraifee following cessation of expsoureT" Id. 350 2/ Id. at 52. 356 3/ See, e.g., 42 Fed. Reg. 27467 May 27, 1977); Testimony 357 of Dr. Robert Snyder in 0SHA Docket H-059, Ex. 156, Tab 2, pp. 358 1-3. 256 257 258 -3 ... oooo^60nA- 359 hematological effects will result from short- or long-term 360 exposure to benzene at levels of 10 ppm and below. Thus, in the 361 1977 benzene proceeding. Dr. Robert Snyder stated that "[i]n both 363 animals and man the lowest level [of exposure] demonstrated to 364 produce bone marrow depression was approximately 40 ppm."4/ 369 Other witnesses in the 1977 proceeding expressed the same view.5/ 379 360 These conclusions are supported by the findings of a 381 recent study conducted at a coke oven by-product recovery 382 pmlant.6/ In that study, hematological data -- including red 385 blood cell counts, white blood cell counts, and hemoglobin -- 387 from benzene exposed workers were compared to comparable data 388 from a non-exposed reference group of supervisory employees. The 389 benzene-exposed workers, who had an average exposure of 10.5 ppm, 391 were divided into three groups having cumulative exposures equiv- 392 alent to 20-year occupational exposures of less than 1 ppm, 1-10 395 395 365 4/ Testimony of Dr. Robert Snyder in 0SHA Docket H-059, 366 Ex. 156, Tab 2, p. 8; Transcript ("Tr.") in OSHA Docket H-059 at 368 3224-3225. 370 5/ See Testimony of Dr. Hervey B. Elkins, Tr. 3210 370 (threshold for injury to the blood-forming system is between 25 371 and 50 ppm); Testimony of Dr. Robert E. Olson, Tr. 2925 (most 372 people put the threshold for blood dyscrasias at 50 ppm); Testi- 373 mony of Dr. Hebert E. Eckardt, Tr. 2088 (unaware of any reports 374 of blood dymprasias below 35 ppm); Testimony of Dr. Irving R. 376 Tabershaw, 149A, p. 6 (cytopenia has not been demonstrated to 377 occur below A ppm). All citations in this footnote are to OSHA 378 Docket H-059. 382 6/ Hancock, et al., "Hematological Findings Among Workers 383 Exposed to Benzene at a Coke Oven By-Product Recovery Facility," 384 39 Archives of Environ. Health 414 (1984). 256 257 258 -4- Mod ooQ 1^603 393 ppm and more than 10 ppm, respectively. No statistically signify 395 icant differences among the various exposure and comparison 396 groups were found for the hematological parameters examined.7/ 397 The authors concluded that "the estimated cumulative exposures 398 examined . . . [appear to] represent doses which fall below the 399 threshold at which non-leukemogenic, hematotoxic effects on 400 humans are manifested . . ."8/ They noted that their findings 401 are consistent with the results of an earlier chemical industry 403 study9/ and observed that the results of their study "provide no 406 evidence that the current OSHA permissible exposure level of 10 407 ppm is inadequate to protect employees against the 408 non-leukemogenic chronic effects of benzene on the hematopoietic 409 system . . . ,"10/ 410 411 In short, as Or. Bernard Goldstein recently observed: 412 ^The available evidence suggests that the current 10 ppm TWA OSHA 413 standard is sufficient to protect against overt symptomatic 414 pancytopenic effects."11/ And, Dr. Goldstein notes, even if 418 418 397 7/ See id. at 416. 401 8/ Id. at 417. 403 9/ Id*, citing Townsend, et al., "Health Exam Findings 404 Among Individuals Occupationally Exposed to Benzene," 20 J. 405 Occup, Med. S43 (1978). 409 10/ Id. at 417. 414 11/ Goldstein, "Clinical Hematotoxicity of Benzene," in 415 Mehlman, ed., Carcinogenicity and Toxicity of Benzene, 51, 57 416 (1983). 256 257 258 5- co^60 A 418 there is "variation in the individual response to benzene . . . 419 the extent of variation in susceptibility to the pancytopenic 420 effects of benzene does not appear to be great.1112/ 421 422 2. Teratogenic and Reproductive Effects 423 424 Exposures to benzene even at levels well in excess of 425 10 ppm have not been shown to present a teratogenic or reproduc- 426 tive hazard. Thus, EFA has concluded that "it is unlikely that 427 benzene administered by inhalation during the principal period of 428 organogenesis constitutes a teratogenic hazard."13/ Other 434 experts in the field have reached a similar conclusion. Thus, 435 Dr. Bernard Schwetz has reported that 439 440 441 441 442 443 443 444 445 446 446 451 ___ 451 420 12, developmental toxicity studies conducted in three species . . . using three different routes of administration, are in agreement that teratogenic effects are not associated with exposure to levels of benzene that are not maternally toxic .... Even at levels of benzene sufficiently high to cause maternal toxicity (in the inhalation study], there was little or no evidence of a teratogenic effect.14/ Id. at 52. 429 13/ EPA, Ambient Water Quality Criteria for Benzene 430 ^October 1980) at C-42. See also EPA's Response to Public Com- 431 ments on Reflation of Benzene, 49 Fed. Reg. 23478, 23481, col. 2 432 (June 6, ("EPA agrees . . . that the available data do not 433 implicate be&zene as a potential teratogen or embryotoxin in test 434 species."). 447 448 449 450 14/ See Schwetz, "A Review of the Developmental Toxicity Benzene," in Mehlman, ed.. Carcinogenicity and Toxicity of Benzene (1983) at 17, 20. Maternal toxicity generally has not been reported to occur at concentrations below about 300 ppm. of 256 257 258 6 00001 6606 453 453 454 He concludes that based on these teratologic studies, "exposure 455 to levels of benzene which do not cause other forms of toxicity 456 would not be expected to cause adverse developmental effects."15/ 475 477 ***** 478 479 In short, non-malignant health effects of benzene are 480 not a matter of concern at exposure levels of 10 ppm and below, 481 the levels that are of interest in the present proceeding. The 482 current 10 ppm standard, as OSHA acknowledges, was designed to 483 protect against risks of "aplastic anemia and other blood 484 dyscrasias as well as acute and chronic health effects."16/ The 487 decision to propose a 1 ppm standard was not based on new 488 evidence showing a risk of aplastic anemia and other blood 490 490 457 15/ Id. at 21. Other investigators who have reviewed the 458 literature agree with this conclusion. See Lee, et al., "Assess- 459 ment of Benzene Health Effects in Ambient Water," in Mehlman, 460 ed.. Carcinogenicity and Toxicity of Benzene (1983) at 91, 102 461 ("In summary, there is general agreement among scientists that 462 teratogenic effects are not associated with exposure to levels of 463 benzene that are not toxic to the pregnant animal. Even at 464 levels of benzene sufficiently high to cause maternal toxicity, 465 there was little or no evidence of teratogenic effect."). 466 Accord, European Chemical Industry Ecology & Toxicology Center, 467 Technical Report No. 16? A Review of Recent Literature on the 468 Toxicity of Benzene (December 12, 1984) (hereinafter "ECETOC 470 Report") At i^There is no reliable evidence to suggest any 471 association Qpltveen exposure to benzene and adverse effects on 472 human reproduction. **). The ECETOC Report is submitted herewith 474 as Appendix A. 485 16/ OSHA, Preliminary Regulatory Impact and Regulatory 486 Flexibility Analysis for the Benzene Standard (December 1985) 487 ^"Regulatory Impact Analysis*) (Ex. ) at 1-1. 256 257 258 -7- 4S9 dyscrasias at 1.0 ppm but on studios published since 1974 which 490 OSHA interpreted as suggesting a need to lower the standard in 493 order to provide adequate worker protection against a perceived 494 risk of leukemia.17/ Since non-malignant hematological effects 495 such as aplastic anemia and pancytopenia will not be caused by 497 occupational exposure to benzene at levels of 10 ppm, there is no 498 basis for suggesting that lowering the permissible exposure limit 500 j/PEL") from 10 ppm to 1 ppm would reduce the incidence of 501 aplastic anemia and other blood dyscrasias.18/ 523 523 523 523 523 523 523 523 523 523 494 ' 17/ Id. at 1-1, 1-20. 501 18/ See 50 Fed. Reg. 50512, 50539, col. 2, 50541, cols, 502 1-3. Thus, even if the Vigliani study did indicate a ratio of 503 deaths from benzene-induced non-malignant blood dyscrasias to 504 deaths from benzene-induced leukemia of 0.42 (see id. at 50541, 505 col. 2), it would be entirely irrelevant to the questions at 506 issue in the present proceeding, since the benzene exposure 507 levels in the Vigliani study were far in excess of 10 ppm. By 508 contrast, the exposure levels at issue in this proceeding are 509 below the threshold for~the production of non-malignant blood 511 dyscracias. Accordingly, OSHA's "Benefits Analysis," which 512 estimates specific numbers of aplastic anemia deaths that would 513 be avoided under alternative regulatory proposals, misses the 514 mark, since^t Is "[b]ased on the assumption that the Vigliani 515 case study dp typical . . . Regulatory Impact Analysis at 516 II1-8. The Ott and Rinsky studies, to which OSHA refers for sup517 port (see id.).do not demonstrate an increased risk of aplastic 518 anemia associated with occupational exposure levels of 10 ppm and 519 below. Consequently, they do not support OSHA's unjustified use 520 of the Vigliani study to predict the avoidance of aplastic anemia 521 deaths through a reduction in the PEL from 10 ppm to 1 ppm. 256 257 258 -8- 60^ 0000^ 525 B. Carcinogenicity 526 529 530 531 532 533 533 537 1. The Qualitative Evidence of Benzene-Related Leukemia Does Not Demonstrate an Increased Risk at Occupational Exposure Levels of 10 ppm and Below. As OSHA points out in the Notice of Proposed 538 Rulemaking, there are a variety of epidemiological studies and 539 clinical reports associating occupational exposure to benzene in 540 various contexts with an increased risk of leukemia, principally 541 of the acute myeloid variety.19/ For the most part, the exposure e 542 levels involved in these studies and reports were very high -- in 544 excess of 100 ppm. For example, the shoeworkers studied by Aksoy 545 and co-workers were estimated to have average exposures of 546 150-210 ppm, with excursions between 210 and 640 ppm.20/ Simi- 548 larly, the workers studied by Vigliani were exposed to benzene 549 concentrations that were mostly around 200-500 ppm, with peaks up 550 to 1,500 ppm.21/ While certain other studies may suggest the 556 possibility of an increased risk of leukemia at benzene exposure 557 levels below 100 ppm, no study conclusively demonstrates that 559 559 541 19/ See 50 Fed. Reg. 50512, 50516-50524. 547 20/ Sm id. at 50517. 550 21/ Sf*. id. at 50512, 50518, col. 1; Brian MacMahon, 551 Epidemioloqft" Evidence on the Possibility that Risk of Leukemia 552 May Be Increased by Exposure to Benzene at Low Concentrations 553 (April 23, 1985) at 45 (hereinafter referred to as "MacMahon 554 Report"), A copy of the MacMahon Report is submitted as Appendix 555 B to these Comments. 256 257 258 -9- CD 000015608 558 occupational exposures of 10 ppm and below present a signifi- 559 eantly increased risk of leukemia. 560 561 a. The Rinskv Study 562 563 In its rulemaking proposal, OSHA relies extensively 564 upon the NIOSH study by Rinsky, et al. of rubber hydrochloride 566 (Pliofilm) workers at two manufacturing facilities in Ohio (here- 567 inafter referred to as the ^Rinsky Study") as direct evidence 568 that benzene is a human leukemogen. Whatever that study may dem- 571 onstrate about an association between benzene exposure and 572 leukemia, it certainly does not constitute direct evidence of an 573 increased risk of leukemia at levels of 10 ppm and below. As Dr. 574 Brian MacMahon points out, a "major difficulty with this study is 575 the poor information on the level of exposure with which the 576 excess of leukemia was associated -- a matter about which there 577 has been a great deal of debate."22/ Dr. MacMahon goes on to 579 observe that "every one of the 8 leukemias in the cohort, and 4 580 other cases which were not eligible for the cohort, occurred 581 after exposures substantially higher than 10 ppm."23/ Based upon 582 his extensive review of the epidemiologic literature and the 583 debate that has surrounded the interpretation of the Rinsky 584 study. Dr. MacMahon concludes that the excess leukemias found in 586 586 578 22/ MacMahon Report at 2. 581 23/ Id. at 2, 27. . 256 257 258 - 10 00001&609 585 the Rinsky study are associated with average exposure levels that 586 are five to ten times the current 10 ppm standard.24/ 588 589 OSHA's discussion of the exposure data for the Rinsky 590 cohort provides no basis for reaching a contrary conclusion. 591 OSHA appears to accept the suggestion that worker exposures in 592 the study "were generally within the recommended limits in effect 593 at the time of employment, that is, they were between ^00.ppm 594 and 10 ppm during the years 1941-1975."25/ Even if that conclu- 595 smion were correct, it would mean that the average exposures were 597 considerably in excess of 10 ppm. Exposures of the workers who 598 actually contracted leukemia would have been considerably higher, 599 since they all were employed during the 1940s and 1950s, when the 600 recommended limits (to the extent they existed at all) were 35 601 ppm-100 ppm.26/ 605 606 Moreover, there is good reason to believe that 607 exposures of the Rinsky cohort often exceeded the recommended 608 limits. OSHA itself refers to a 1955 report in which workers at 610 one of the two Pliofilm plants entered areas where benzene 611 exposures ranged from 19 - 680 ppm, at a time when the 613 613 587 24/ 1^ at 3. 594 25/ 5OSFed. Reg. 50512, 50518, col. 2. 602 26/ See JRB Associates, Technological Feasibility and 602 Economic Impact Study of Alternative Standards for Benzene (1984) 603 ("JRB Report") (Ex'. 153), Table" 1-2. 256 257 258 11 000 612 recommended exposure limit was 35 ppm.27/ Although respirator 613 usage allegedly was required in such high exposure areas, the 614 fact is that, even as late as 1973*74, workers at the same 615 facility often did not wear respirators when they entered areas 616 of high benzene concentrations that were well above the 10 ppm 617 exposure limit in effect at the time.28/ During the 1930s and 618 1940s (when many members of the Rinsky cohort were exposed to 619 benzene) industrial hygiene practices and the awareness of 620 occupational health problems were far less developed than is the 621 c^se today |or was the case in 1973-1974) -- with the result that 622 employees exposed to unacceptably high levels of benzene often 623 might not have worn respiratory protection.29/ 627 628 Moreover, exposure information was severely limited for 629 the Pliofilm plant at which most of the leukemia cases were 630 found.30/ At that facility, benzene levels ranging up to 100 ppm 636 were measured in 1957, when the recommended exposure limit was 25 637 ppm.31/ At the plant for which more exposure information is 640 640 612 27/ See 50 Fed. Reg. 50512, 50519, col. 2. 618 28/ See id. 624 29/ S-- Supplementary Statement of Dr. Hervey B. Elkins, 625 filed in OSHft Docket H-059. A copy of Dr. Elkins' Statement is 626 submitted herewith as Appendix C. ~ 630 30/ The fact that the incidence of leukemia was so much 631 higher at the facility having very little exposure data suggests 632 that exposure levels may have been significantly higher at this 633 facility than at the plant for which more exposure data are 634 available. 637 31/ See 50 Fed. Reg. 50512, 50519, col. 3. 256 257 258 - 12 - 00001&61 640 available, 73 percent of the 15 measurements taken from 1946-1950 1 641 were above the permissible exposure limit in effect at the 642 time.32/ ' 646 647 In sum, average benzene exposures of the Pliofilm 647 cohort studied by Rinsky, et al., were far in excess of the 648 exposure Revels that are of interest in the present proceeding. 650 Furthermore, all of the leukemia cases were found in workers 650 first exposed to benzene prior to 1963, when a Threshold Limit 651 Value ^"TLV") of 25 ppm was proposed by ACGIH.33/ No leukemia 653 deaths have been found among cohort members first exposed after 654 1963. [Check the accuracy of these statements.] 656 657 In addition, the numerically low incidence of leukemia 657 in the study and "the marked heterogeneity of benzene exposure in 658 the work force studied" further complicate the effort to 659 attribute the increased leukemia incidence to particular exposure 661 levels.34/ For these and other reasons. Or. Bernard Goldstein, 663 like Dr. MacMahon, has concluded that "the Rinsky et al. data, 664 and the original findings of Infante et al., do not provide sig- 665 nificant direct support for leukemogenesis occurring at levels of 675 675 643 643 644 32/ S-- Letter from Dr. H.G.S. van Raalte, et al. to editor of Rlefc Analysis, June 9, 1983. A copy of Dr. van Raalte's letter is submitted herewith as Appendix D. the 652 33/ See 50 Fed. Reg. 50512, 50514, col. 3. 661 34/ See Bernard D. Goldstein, Benzene Toxicity: Review of 662 Recent Literature (February 3, 1983) (Ex. ____) at 5. 256 257 258 13 >-\ 0o 0o^ S 667 benzene at or near the present OSHA standard."35/ 675 676 b. The Wong Study 677 678 The rulemaking notice also devotes considerable atten- 679 tion to the study by Wong, et al., of chemical industry 680 workers.36/ That study also should not be interpreted as 682 demonstrating an increased risk of leukemia or lymphopoietic can- 684 cer among workers exposed to benzene levels below 10 ppm. pie 686 standardized mortality ratio ("SMR") for all cancers among 687 benzene-exposed members of the Wong cohort was 102.3 -- barely 689 elevated and certainly not statistically significant.37/ Simi- 691 larly, although the SMR for lymphatic and hematopoietic cancer 692 was slightly elevated among the benzene-exposed workers, the 693 increase was not statistically significant.38/ This also was 694 true of the various subcategories of lymphatic and hematopoietic 695 cancer, including leukemia, and of lung cancer.39/ In short, 704 704 667 35/ Id. It also should be noted that the most recent 668 update of data on the Rinsky cohort shows that the Standardized 669 Mortality Ratio "SMR") for leukemia has declined from the level 670 of 560 reported by OSHA (see 50 Fed. Reg. 50512, 50519, col. 1) 671 to 328. See Rinsky, et al., nBenzene and Leukemia: An 673 Epidemiologic Risk Assessment," August 9, 1985 (Ex. 176A) at 12. 680 36/ Sey 50 Fed. Reg. 50512, 50522-23. This study (Ex. 681 151-A) will hereinafter be referred to as the "Wong Study.n 690 37/ Se Wong Study, Table 22. 693 38/ See .id.. Table 22. 696 698 704 5 56 .57 258 39/ See insignificant id.. Table 22. In the case of excess found among the exposed [Footnote - 14 - lung cancer, the workers was largely continued next page} O'.0 fcv 704 overall cancer mortality among the benzene-exposed workers in the 705 Wong Study was comparable to the rate in the general U.S. male 706 population, while the rate for lymphatic and hematopoietic cancer 707 was only slightly, but not significantly, higher than the 708 national norm. 710 711 A statistically significant increase in the relative 711 risk for lymphatic and hematopoietic cancer was found in the Wong 713 Study only when workers occupationally exposed to benzene were 714 compared to an internal control group of workers who did not have 715 such exposure at the plants studied ^although they may have been 716 exposed to benzene when working for other employers). However, 717 as Or. Wong and various peer reviewers and other investigators 718 point out, this increased relative risk is attributable primarily 719 to the unexplained and unusually low mortality rate experienced 720 by the internal control group. 722 723 This is particularly evident in the case of leukemia, 724 where there were seven deaths among the exposed workers and none 725 in the control group, even though 3.4 leukemia deaths would have 726 been expected among the controls based upon the age- and 728 728 5 (Footnote continued from preceding page] 5 699 attributable to a single plant which (i) was located in a county 700 having an above average lung cancer rate, and~(ii) contributed a 701 large percentage of the exposed cohort members and a dispropor702 tionately small ercentage of the unexposed cohort members to the 703 study. 256 257 258 15 - 00001* 727 sex-adjusted experience of the general population not exposed to 728 benzene.40/ As EPA's Carcinogen Assessment Group observes, the 751 carcinogenic potency of benzene estimated from the Wong Study 752 "can be attributed to the fact that the non-exposed group had no 753 cases of leukemia. As a result, the increase in risk over con- 754 trols may be due to an artifact in the data rather than to a true 755 carcinogenic response."41/ Dr. Wong himself noted that the sig- 768 nificant mortality deficit among the control group magnifies the 772 772 728 40/ See Wong Study at ii and Table 23; B. MacMahon, "Review 729 of 'An Industry-Wide Mortality Study of Chemical Workers Exposed 730 to Benzene'" at 2, 4 (October 12, 1983) (submitted herewith as 731 Appendix E) (The difference in leukemia and lymphatic cancer 732 between the exposed and unexposed workers "results primarily from 733 a deficit of both leukemia and lymphatic cancers among unexposed 734 workers, not from an excess among exposed employees .... I do 735 not think that one can conclude from these data that there is an 736 excess of leukemia in the exposed group."); P. Enterline, "Review 737 of Draft Report 'An Industry-Wide Mortality Study of Chemical 739 Workers Exposed to Benzene'" at 6 (November 1, 1983) (submitted 740 herewith as Appendix F) ("(T]he absence of JLeukemia deaths in the 741 control group is most unusual and statistically significant 742 .... If the control group is inappropriate, then comparisons 743 with this group are invalid."); Letter from Kenneth J. Rothman to 744 R. T. Richards, January 13, 1984, at 4 (submitted herewith as 745 Appendix G) _(_"most striking feature ... is the 746 lower-than-expected mortality in the occupational cohort lacking 747 benzene exposure. . . . The unusual mortality experience of the 749 comparison group inevitably raises doubt about the validity of 750 the data on the exposed cohort as well."). 756 41/ EPA Carcinogen Assessment Group, Interim Quantitative 757 Cancer Unit Risk Estimates Due to Inhalation of Benzene, February 758 15, 1985 at 23-24. The cancer mortality deficit among the 759 non-exposed workers cannot be explained by the so-called "healthy 760 worker effect." As Dr. Brian MacMahon points out: "Cancer in 761 general tends to have a less marked ^healthy worker effect1 than 762 other causes of death, and it is difficult to understand how any 763 of the known mechanisms of such an effect could apply to 764 leukemia." MacMahon Report at 57. 256 257 758 - 16 - 000015615 MCD 770 apparent relative risk of the exposed workers.42/ 712 773 In addition, the size of the cohort as a whole "was 774 small for a number of specific analyses/*43/ resulting in a high 775 degree of statistical variability. In the case of leukemia par- 776 ticularly, the number of deaths was very small, the statistical 777 variability was very large, and the dose-response relationship 778 was not monotonic -- i.e., the incidence of leukemia did not 779 increase steadily as increasing cumulative exposure levels were 780 e*xamined.44/ In a quantitative risk assessment based on the Wong 782 Study, Dr. Frank Carlborg concluded that because the data set in 785 the study is relatively meager, it fits a variety of 786 dose-response models, including "the no-effect model which 787 assumes that an exposure to benzene does not affect the incidence 788 of lymphopoietic cancer. . . ."45/ Since the no-effect model 793 "fits the data very well in the statistical sense,"46/ the data 794 from the Wong Study "are consistent with the hypothesis that the 795 incidence of lymphopoietic cancer in the exposed group is 798 798 770 42/ See Wong Study at 54. 774 43/ Id. at 64. 780 44/ S-- id. at 64 and Figure 2. 788 45/ F-- Carlborg, A Quantitative Cancer Risk Assessment for 789 Benzene Bas#dTon Data from an Industry-Wide Mortality Study of 791 Chemical Workers at Ti Dr. Carlborg*s report is submitted here- 792 with as Appendix H. 794 46/ Id. at 4. 256 257 258 - 17 - Mod Oo015616 796 age-dependent and not related to the benzene exposure."47/ 798 799 Two basic conclusions emerge from the Wong Study. 800 First, the overall cancer rate and the rates of lymphatic and 801 hematopoietic cancer (including leukemia) among the 802 benzene-exposed workers were not significantly elevated when corn- 803 pared to the general population. Second, the increase in rela- 804 tive risk for lymphatic and hematopoietic cancer among the 805 benzene-exposed workers when compared to the internal control 806 group is of questionable relevance. The increase must be conm 807 sidered in light of the low number of deaths in the cohort as a 808 whole, the unexplained and dramatic mortality deficit among the 809 internal controls, and the absence of any leukemia in the control 810 group.48/ 816 817 For these reasons. Dr. MacMahon has concluded that ^it 818 is unclear whether leukemia risk is or is not increased in the 819 [Wong] study group . . . ."49/ Indeed, because the most "strik 820 ing and statistically significant feature of [the Wong] data is 821 the deficit of leukemia in the unexposed workers," Dr. MacMahon 823 finds it "difficult to accept these data at face value."50/ He 824 824 797 47/ Id* 810 48/ It. also is important to note that none of the seven 812 leukemia deaths in the Wong cohort were of the acute myelogenous 813 cell type, the type that has generally been associated with 814 benzene related leukemia. See 50 Fed. Reg. 50512, 50523, col. 1 819 49/ MacMahon Report at 8. 824 50/ MacMahon Report at 57. 256 257 258 18 MOD 000015617 824 also notes that "the small numbers make the SMRs for leukemia 825 consistent with almost any hypothesis,"51/ and he ^decline[s] to 828 accept the estimated SMRs as reliable evidence either for or 829 against association of leukemia with the reported cumulative 830 exposures,"52/ Dr. Kenny Crump has expressed a similar view, 831 stating that "the significant findings [in the Wong study] are 832 due in .large measure to the deficits in cancer in the unexposed 833 group ^ . . . Therefore, without some explanation for the cancer 835 deficit in the unexposed group, this study by itself does not 636 provide strong evidence of a relationship between occupational 837 exposure to benzene and lymphatic and hematopoietic cancer."53/ 841 842 c. The Dow Study 843 844 OSHA preliminarily suggests that the study by Ott, et 845 al., of Dow Chemical Company workers54/ "represents direct 849 observation of a leukemogenic risk from low level benzene 850 exposure."55/ For a variety of reasons, such an interpretation 851 851 826 827 51/ at p. Id. at 38. ____, supra. Cf. the discussion of Dr. Carlborg's report 830 52/ MacMahon Report at 58. 838 53/ See Crump 6 Allen, Quantitative Estimates of Risk of 839 Leukemia From Occupational Exposure to Benzene (May 1984) ^Ex. 840 152) (hereinafter referred to as "Crump Report") at 14. 845 54/ Ott, et al.# ^Mortality Among Individuals 846 Occupationally Exposed to Benzene," Archives of Environ. Health 847 (January/February 1978) 3-10 (Ex. 128-33) (hereinafter referred 848 to as the "Dow Study"). 850 55/ 256 .57 258 50 Fed. Reg. 50512, 50520, col. 1. 19 MCD 00001561B 851 of the Dow Study is not justified. 853 854 the Dow Study as originally reported, two cases of 855 .leukemia were found in an occupational cohort where 0.8 cases 856 were expected. When a third worker (whose cause of death was 857 recorded as bronchopneumonia, with myeloblastic leukemia listed 858 under "other significant conditions") is treated as a case of 859 l.eukemia, the incidence rate is three cases of leukemia where 0.8 860 were expected.56/ Several points must be made about this find- 861 862 863 ing. * First, it is questionable whether this study, which was 864 denominated and structured as a mortality study, should be ana- 865 lyzed as an incidence study.57/ Even when it is so analyzed, the 871 finding of three leukemias where 0.8 were expected (p=0.047) is 872 only "of borderline statistical significance."58/ Moreover, the 882 882 860 56/ See Dow Study at 9. 865 57/ See MacMahon Report at 4. As EPA observes, 866 declassification generally is not appropriate in mortality 867 studies that must rely on death certificates, since the same 868 reclassification is not applied to the group from which expected 869 numbers are derived." EPA, Health Assessment Document for Nickel 870 (EPA-600/8-83-012F, September 1985) at 8-31. 873 58/ Goldstein, "Clinical Hematotoxicity of Benzene," supra, 874 n.ll at 55. As Dr. MacMahon observes: "The statistical signifi- 875 cance of the excess based on three cases is in question." 877 MacMahon Report at 22. Thus, it is a "fine judgment" as to 878 ^whether or not there is a statistically significant increase in 879 these data . . . Id. at 4. See also EPA, Ambient Water 880 Quality Criteria for Benzene (October 1980) at C-59 (the leukemia 881 incidence in the Dow Study ffis only of marginal statistical sig- 882 nifieance"). 256 257 :58 20 000015619 tfCD 882 data, as EPA has observed, "are too few to provide an independent 883 stable estimate of the relation between benzene and leukemia."59/ 886 887 Second, there was a distinct possibility of confounding 888 exposures among the Dow cohort, including confounding exposures 889 at other places of employment. For example, one of the employees 890 who developed leukemia had only 18 ppm-months of benzene exposure 891 at Dow and had earlier been employed in a saw mill which manufac 892 tured veneer, an occupation for which an increased incidence of 893 myelocytic leukemia has been reported.60/ Observing that 901 "workers in benzene-related occupations [in the Dow Study] typi 902 cally were exposed to other chemicals," the National Academy of 903 Sciences X"NAS,t) concluded that "extrapolation of benzene-induced 904 cancer risk from such data as these would be tenuous."61/ 907 908 909 Study: Third, as Drs. Crump and Allen point out, in the Dow 912 913 913 914 916 916 884 885 59/ 1984) No dose response trend is apparent for any cancer type. There is no particular indica tion from the dose response analysis that EPA, Draft Criteria Document for Benzene (February at VI-16. 894 60/ See Milham, "Neoplasia in the Wood and Pulp Industry," 895 271 Ann. N.Y. Acad, of Science 294-300 (1976); Dow Study at 8-9. 896 See also Environ Corporation,Review of Benzene Risk Assessments 897 (November 11, 1983) (Ex. ____) (hereinafter referred to as 898 "Environ Report") App. A at 23-24 ("In addition to benzene, 899 workers at the Dow plant were exposed to a large variety of chem 900 icals, some of which may have carcinogenic potential . . . ."). 905 61/ 256 257 258 NAS, Drinking Water and Health: Vol. 3 (1980) at 85. - 21 00^ 915 leukemias are related to [benzene] exposure 916 919 . . . .62/ 919 920 As noted in a review of the Dow Study by Environ Corporation, the 921 fact that the workers who developed leukemia had relatively low 922 cumulative benzene exposures compared to the large number of 922 cohort members who did not develop leukemia despite having modern- 923 ate or high benzene exposures makes it difficult to attribute the 924 leukemias to the benzene exposures.63/ As a general proposition: 926 "An increase in the amount of exposure should be associated with 928 an increased risk of developing the disease, if the exposure is 929 of causal importance."64/ 942 943 For the foregoing reasons, the authors of the Dow Study 944 observed that "a retrospective assessment of the possible rela- 945 tionship [of leukemia] to benzene exposure [of cohort members is] 946 very judgmental."65/ And they concluded: "No mortalities 947 _______________________ 947 916 62, Crump Report at 10. 925 63/ See Environ Report, App. A at 25. 930 64/ Id. at 25*26. Dr. Philip Cole also has commented on 931 this point. He states that the fact that the three leukemia 932 cases among the Dow cohort "experienced an exposure that was less 933 than the overall average is persuasive evidence of a lack of 934 'dose-response' in this data set [which] in turn, detracts 935 appreciably from the prospect that the data reflect a cause- 936 effect relationship." Cole, "A Quantitative Estimate of Leukemia 937 Mortality Associated with Occupational Exposure to Benzene: A 938 Critique," p. 3 submitted as an attachment to March 22, 1983 939 letter of R.T. Richards to Leonard Vance, Ex. 137). Hereinafter 940 this document will be referred to as the "Cole Critique." 946 65/ Dow Study at 9. 256 257 258 - 22 0 0<* 947 directly attributable to benzene exposure were observed."66/ EPA 948 has characterized the study in the same fashion, stating: "No 950 association with benzene exposure was detected,"67/ and 952 concluding that the Dow Study taken alone cannot be viewed "as 953 conclusive evidence of an association between low-level (2-9 ppm) 954 occupational exposure to benzene and leukemia . . . ."68/ The 957 United States Supreme Court has expressed a similar view.69/ 967 968 Recently, the Dow Study was updated to cover the mor- 969 tality experience of an expanded cohort through 1982.70/ The Dow 973 Study Update identified two additional cases of leukemia among 974 the expanded cohort, both of whom died at age 80. The authors 976 noted a non-significant excess of leukemia deaths which became 978 978 948 66/ Id. at 3. 950 67/ EPA, Ambient Water Quality Criteria for Benzene 951 (October 1980) at C-59. 955 68/ Response to Public Comments on the Regulation of 956 Benzene, 49 Fed. Reg. 23478, 23483, col. 2 (June 6, 1984). 958 69/ Industrial Union Department AFL-CIO v. American 959 Petroleum Institute, 448 U.S. 607, 633 (1980) ("The authors of 960 the study . . . concluded that it could not be viewed as proof of 961 a relationship between low-level benzene exposure and leukemia 962 because all three workers had probably been occupationally 963 exposed to a number of other potentially carcinogenic chemicals 964 at other points in their careers and because no leukemia deaths 965 had been uncovered among workers who had been exposed to much 966 higher levels of benzene."). 9?0 70/ Bond, et el., Executive Summary of Report "An Update of 971 Mortality Among Chemical Workers" (hereinafter referred to as 972 ^Dow Study Update"). A copy of the Dow Study Update is submitted 973 herewith as Appendix I. 256 257 258 - 23 - 9,*^ 977 statistically significant when the focus was narrowed to 978 myelogenous leukemia.71/ As in the earlier study, no cumulative 979 dose-response relationship was found.72/ 981 982 The authors explained that the "limited available 983 industrial hygiene survey data" made it difficult to estimate 984 ^chronic and acute exposure levels for individual employees 985 ^ . . . "73/ They noted, however, that most of the J.eukemia cases 986 "were likely to have had possible exposures to intermittent. 987 short-term, high levels of benzene measured in excess of 250 ppm 988 in some samples."74/ Mirroring the comments that had been made 989 with respect to the original study, the authors identified 993 994 994 995 996 997 997 998 999 999 1000 1001 1004 1004 1004 1004 1004 1004 978 21/ 980 22/ 985 21/ 988 24/ 1001 21/ 256 257 758 a number of factors [that] complicate use of these data for risk assessment. These include the small number of leukemias observed, the lack~of an apparent dose-response relationship, competing exposures to other potentially hazardous materials, and the uncertain contribution of brief high exposures (up to 250 ppm and more] which most likely occurred during the time period of these individual's sic] employ ment. 75/ See id. at 3 See id. id. at 4. Id. Id. at 4-5. 24 w0.0\ o & 1006 * * it 1007 1008 In sum, the available studies do not demonstrate a sig- 1009 nificantly increased risk of leukemia at occupational benzene 1010 exposure levels of 10 ppm and below. As Dr. Cesare Maltoni 1011 recently observed/ 1014 1015 1015 1016 1017 1018 1019 1019 1020 1027 1027 1028 the available epidemiological and experimental data at present do not provide precise information on the risk of doses around or below 10 ppm. In such a situation, any decision can only be based on social and political consid erations rather than on scientific ones.76/ Thus, in 1985, as in 1978, any decision 0SHA might make 1029 to reduce the current PEL would not rest on a demonstrated human 1030 health risk at occupational exposure levels of 10 ppm and below. 1031 Rather, it would be based on the presumption that there is no 1033 threshold for the leukemogenic effects of benzene and on the 1034 application of a highly conservative linear extrapolation model 1035 to estimate increased leukemia risks down to a benzene concentra- 1036 tion of zero. As discussed below, both the no-threshold presump- 1037 tion and the assumption of dose-response linearity down to zero 1038 are subject to serious question in the case of benzene. 1040 1040 1040 1040 1021 76/ Maltoni, et al., ^Experimental Studies on Benzene 1022 Carcinogenicity at the Bologna Institute of Oncology: Current 1023 Results and Ongoing Research," 7 Am J. Indus. Med. 415, 418 1024 (1985). 256 257 .58 - 25 - nOV O,'o 1041 For these reasons, OSHA should, at the very least. '.042 explicitly recognize the uncertainty that exists on the guestion 1044 whether there is any leukemia risk at all at the occupational 1045 exposure levels that are of interest in this proceeding. Fur 1046 thermore, it should be emphasized that the cancer risks OSHA pre 1047 dicts at exposures below 10 ppm not only are likely to be 1048 overstated (because of OSHA1s conservative approach to estimating 1049 risk), but are frankly hypothetical in nature. 1051 1054 1055 1056 1057 1057 1060 * 2. Epidemiological Studies Do Not Show an Association Between Occupational Benzene Exposures and Cancers Other Than Leukemia. As discussed in the previous section, the association 1061 between exposure to benzene in certain occupational settings and 1062 an increased risk of leukemia rests upon a variety of -063 epidemiological studies and clinical reports in which the 1064 leukemogenic response appears to be associated with benzene 1065 exposure levels well in excess of 10 ppm. There also have been 1066 ^occasional suggestions of other tumors - lymphoreticular, pros 1067 tate, stomach, lung and multiple myeloma - being associated with 1068 benzene exposure . . . ."77/ However, as Dr. Brian MacMahon 1069 observes, "the evidence (for jsuch non-leukemogenic tumors] is 1070 anecdotal or weak and, generally speaking, not replicated."78/ 1072 1072 1072 1068 77/ 1071 78/ MacMahon Report at 1. Id. 256 257 '58 - 26 O' O.'O'o^ & 1073 For example, DeCoufle, et al., suggest that the results 1074 of their study may indicate a possible association between 1075 benzene exposure and multiple myeloma.79/ But, as Dr. MacMahon 1077 points out, these suggestions "are no more than speculation."80/ 1078 Moreover, as 0SHA notes, DeCoufle et al. "did not present any 1079 information on benzene exposure levels for (the cohort stud 1080 ied.] "81/ Thus, because of "the absence of any objective 1082 benzene-in-air measurements, . . . the possible concomitant 1083 exposure of the workforce to other chemicals and the small size 1084 of the study population . . . [it is] difficult to place any 1085 reliance on . . . [the] conclusions" of the DeCoufle study.82/ 1087 1088 The most recent update of the Rinsky Study83/ does not 1090 demonstrate a causal relationship between benzene exposure and 1091 multiple myeloma either. Although an excess incidence of multi 1092 ple myeloma was found in the study, several factors call the sig 1093 nificance of the finding into question. For one thing, there was 1094 no dose-response relationship. As the authors note, "SMRs for 1095 multiple myeloma ^ . . did not increase with increasing 1097 1097 1076 79/ See 50 Fed. Reg. 50512, 50520, col. 1. 1078 80/ MacMahon Report at 18. 1080 81/ 50 Fad. Reg. 50512, 50520, col. 1. 1086 82/ ECETOC Report at 25. 1088 83/ Rinsky, et al., "Benzene and Leukemia: An Epidemiologic 1089 Risk Assessment," August 9, 1985 (Ex. 176A). 256 257 258 - 27 - 0 1096 exposure.*84/ Moreover, cumulative benzene exposures of the 1097 multiple myeloma cases were quite low, with three of the four 1093 cases having less than 40 ppm-years of exposure and one of the 1099 cases being exposed for only four days.85/ Furthermore, for two 1101 of the cases, no corroborating medical reports were available.86/ 1102 Understandably, the authors of the study concluded that the 1103 observations of multiple myeloma ''must ... be interpreted cau- 1104 tiously in the absence of further corroboration."87/ 1111 1112 In short, as one review of the benzene literature con- 1113 eludes, reports of non-leukemogenic cancers "are too few in num- 1114 ber, and the evidence quoted in their support is insufficient, to 1115 make a convincing case that cancers other than leukemia can be 1116 causally associated with excessive exposure to benzene."88/ OSHA. ' 125 itself has recognized this point, stating: 1129 1129 1129 1096 84/ Id. at 13. 1100 85/ See id. at 13 and Table 5. 1102 86/ See id., Table 5. 1105 87/ Id. at 19. Nor does the Wong Study, in the words of 1108 the author, "offer any firm data on the relationship between 1109 benzene exposure and multiple myeloma." Wong Study at 62. 1117 88/ ECZTOC Report at 21-22. Drs. Robert Snyder and Debra 1118 Laskin have expressed a similar view, stating that "the total 1119 body of evidence suggests that . . . the expression of the 1120 carcinogenic activity [of benzene] in man [takes] the form of one 1121 of several types [of] leukemia." R. Snyder & D. Laskin, A Review 1122 of Recent Developments in the Study of Benzene Toxicity ^October 1123 4, 1985) ("Snyder Review") at 28. A copy of theSnyder Review is 1124 submitted herewith as Appendix J. 256 257 is a 28 .0^ .0"' O' 1130 Although suspicion of types of cancers 1130 other than of the lympho-hematopoietic system 1131 has been raised, these have not been ade- 1132 quately evaluated fromepidemiologic cohort 1133 or case-control studies of workersexposed to 1134 benzene.89/ 1137 1137 1138 At the present time, then, there is no justification for associ- 1139 ating non-leukemogenic tumors with exposure to benzene, and spec- 1140 ulation as to any such possible association cannot serve as a 1141 basis for reducing the current standard. 1142 1142 1143 * 3. Animal Bioassavs 1144 1145 At the time of the 1977 hearings, an animal model for 1146 benzene carcinogenicity had not been demonstrated. In the inter- 1147 vening years, studies conducted at the University of Bologna, New 1148 York University, and the National Toxicology Program have shown 1149 that benzene is a carcinogen in animals through the oral and 1150 inhalation routes. In the preamble to the proposed Benzene Stan- 1151 dard, OSHA suggests that "[t]hese findings add support to 1152 evidence that benzene is a human carcinogen and suggest that can- 1153 cers other than of the lympho-hematopoietic system also may be 1154 involved in humans."90/ 1156 1156 1156 1156 1156 _ 1156 1134 89, 50 Fed. Reg. 50512, 50516, col. 3. 1154 90/ 50 Fed. Reg. 50512, 50527, col. 1. 256 257 258 - 29 - MOD 000015628 1157 pie absence of an animal model for benzene carcinogen- 1158 esis in earlier years had been a puzzling phenomenon, since in 1159 virtuallyall cases, a substance found to cause cancer in humans 1160 (as was the case for benzene) also has been shown to cause cancer 1161 in animals. As Dr. Robert Snyder observes, the fact that animal 1162 models for benzene carcinogenesis now have been demonstrated 1163 "add[s] further assurance to the conclusion derived from 1164 epidemiological and clinical studies that in a qualitative sense 1165 benzene poses the potential to produce leukemia in humans."91/ 1166 The development of an animal model for benzene-related neoplasia 1167 is important, since it "may help us to understand the molecular 1168 events underlying leukemogenesis in man."92/ 1170 1171 However, the results of the animal bioassays do not 1172 provide a basis for concluding that benzene exposure is likely to 1173 produce cancers other than of the lympho-hematopoietic system in 1174 humans. As Dr. Snyder notes, "the total body of evidence sug- 1175 gests that benzene is indeed carcinogenic, with the expression of 1176 the carcinogenic activity in man taking the form of one of sev- 1177 eral types of leukemia."93/ The fact that solid tumors have been 1178 produced in animal studies "should not be interpreted to predict 1179 that benzene will cause solid tumors in man, because this is 1181 1181 1166 91/ See Snyder Review at 28. 1169 92/ Id. 1177 93/ Id. 256 257 258 - 30 - 1X80 . . . an example of differences in response among species."94/ 1181 As Dr. Snyder explains. 1184 1185 1186 1186 1187 1188 1189 1189 1193 1193 1194 The animal studies do not provide a basis for concluding that exposure of humans to benzene either at the doses and routes to which the animals were exposed or through other routes of exposure and at other doses is likely to lead to the production of forms of cancer other than leukemia.95/ After reviewing the recent literature on benzene toxi 1195 cology, the Benzene Task Force of the European Chemical Industry 1196 Ecology & Toxicology Center concluded that there is adequate 1197 evidence to categorize benzene as an animal carcinogen.96/ How 1198 ever, the Task Force emphasized 1201 1202 1203 1203 1204 1205 1205 1206 1207 1208 1208 1209 1210 1210 1214 1214 1214 1214 1214 1181 94/ 1190 95/ that the types of neoplasm (hepatomas, oral carcinomas) reported in the recent animal studies are rare in humans. In view of the long period over which workers exposed to benzene have been under observation, the Task Force believes that if a causal association existed between human exposure to benzene and the onset of such neoplasms, it would have been detected. Benzene is carcinogenic to rats, mice and man, but the substantial dif ferences in the responses makes Isic] the qualitative and quantitative extrapolation of the animal results to man uncertain.97/ 14* Id. 1198 96/ ECETOC Report at 14. 1211 97/ Id. at 14-15. 256 257 258 31 00 1215 In sum, the recent animal studies provide qualitative 1216 confirmation of benzene's carcinogenic potential in humans, but 1217 they do not provide a basis for concluding that benzene will 1216 cause non-lymphopoietic cancers in humans. Nor, as discussed 1219 below, is it appropriate to use the animal data for purposes of 1220 quantitatively estimating the risk of leukemia that occupational 1221 exposures may present to workers.98/ 1223 1227 C. Mutagenicity, Cytogenetic Effects, and 1228 Hematological Effects at the Cellular Level 1229 1229 - 1230 1. Mutagenicity 1231 1231 1235 Until very recently, benzene had not been found to be 1236 mutagenic in short-term tests of mutagenicity.99/ Thus, in 1239 December 1984, a review of recent literature on the mutagenicity i.240 of benzene concluded: "It is well established that benzene does 1241 not induce point-mutations in chromosomes."100/ While benzene or 1243 its metabolites are known to bind to DNA, this binding "has not 1244 been linked to the transformation of the affected cell to a can- 1245 cer cell."101/ 1247 1247 1247 1221 98/ See pp. ___ - ___, infra. 1236 99/ Snyder Review at 30; International Agency for 1237 Research on Cancer "IARC") Monographs Volume 29: Benzene (1982) 1238 ("IARC Monograph") (Ex. 128-8) at 114. 1242 100/ ECETOC Report at 15. 1246 101/ Snyder Review at 30. 256 257 258 - 32 - MCD 000015631 1248 Jn the rulemaking notice, OSHA states ^that benzene was 1249 found to be mutagenic in human cells in a recently developed 1250 gene-locus mutation assay utilizing a metabolically competent 1251 human lymphoblastoid cell line."102/ It is difficult for us to 1253 comment on this report (by Crespi and Penman), since the document 1255 in the OSHA docket (Ex. 159-19A) is simply a one paragraph 1256 abstract. Nonetheless, the brief description we have seen leads 1257 us to question whether this report should be taken as 1258 establishing that benzene is mutagenic in the face of a host of 1259 negative test results over the years. .2=0 12 51 The assay employed by Crespi and Penman ^measuring the 1262 induction of mutations at the hypopxanthine guanine 1263 phosphoribosyl transferase or "HGPRT" locus in human 1264 lymphoblastoid cells) is relatively new and is not a well vali 1266 dated assay for studying point mutations at this locus.103/ 1271 Moreover, the abstract reports mutagenic activity at one dose 1273 only, and nothing is said about the percentage of cell survival 1274 at this dose. In most short-term mutation tests, a cytotoxicity 1275 screen is done, various dose levels are selected, and the assay 1276 is run at several doses corresponding to an anticipated range of 1279 1279 1252 102/ 50?Fed. Reg. 50512, 50528, col. 2. 1267 1268 1269 1270 103/ The HCPRT point mutation assay using Chinese hamster ovary cells is favored by genetic toxicologists and frequently is included in a mutagenicity test battery performed on a specific chemical. 256 257 258 - 33 0o 1277 percent cell survival, so that the existence of a dose-response .279 relationship can be investigated. The absence of dose-response 1280 information in the Crespi and Penman abstract further detracts 1281 from the significance that can be attributed to the report. 1283 1284 In sum, the brief abstract of the novel assay to which 1285 OSHA refers does not appear to provide a sufficient basis for 1286 reaching a conclusion regarding the mutagenicity of benzene that 1287 is contrary to the results of studies performed by a variety of 1288 investigators in other laboratories. 1289 1290 2. Cytogenetic Effects 1291 1292 The rulemaking notice refers to the fact that benzene 1293 has been found to induce various types of cytogenetic effects, 294 including aberrations and other chromosomal damage in humans, and 1295 increased production of sister chromotid exchanges ("SCEs1*) and 1296 micronuclei in experimental animal test systems.104/ 1298 1301 1302 1303 1303 1306 a. Chromosome Aberrations in Human Studies Most of the human studies of chromosome aberrations 1307 involved relatively high benzene exposures or limited information 1308 about benzene exposure levels. The study by Picciano (Ex. 1309 144-118) on which primary reliance is placed, found that at 1311 1311 1296 104/ See 50 Fed. Reg. 50512, 50524-25, 50527-28. 256 257 258 - 34 - MCD 000015633 1310 exposure levels estimated to average less than 10 ppm, "[cJells 1311 with marker chromosomes Xrin93' dycentrics, translocations and 1312 exchange figures) were significantly more common in benzene 1313 exposed workers compared to controls'1 and that ^[cjells with 1315 chromosome breaks were also more common among exposed workers 1316 . . . ."105/ The Picciano study, however, must be interpreted 1317 with caution. 1318 1319 Apart from the question of what clinical significance, 1320 if any, these chromosomal effects may have,106/ ^the implication 1329 of the chromosome changes,11 as Or. MacMahon notes, ^is not as 1330 certain as Picciano suggests, and these 52 workers were exposed 1331 to a variety of other aromatic hydrocarbons, including toluene, 1332 styrene, ethylene and diethylbenzene which may have played a 1333 role."107/ Dr. Robert Snyder expresses similar reservations 1334 about the Picciano study, stating that "it is unclear if smoking. 1336 recent illness, other occupational clastogens and other con- 1337 founding factors were carefully excluded."108/ Moreover, other 1338 1338 1316 105/ Snyder Review at 14. 1320 106/ See gp. ___ - _____, infra. As Or. Brian MacMahon 1321 observes: wThe clinical significance of chromosome aberrations 1322 of these types not] clearly established." MacMahon Report at 1323 24. Similarly, EPA states that "no direct evidence of a casual 1325 I.sic] linkage between chromosomal aberrations and leukemia 1326 exists." Rtiponse to Public Comments on the Regulation of 1327 Benzene, 49 Fed. Reg. 23478, 23481, col. 3 (June 6, 1984). 1333 107/ MacMahon Report at 6. 1337 108/ Snyder Review a't 14. 256 257 258 - 35 0 C)0 O.Vf-^<6 :j-A 1338 factors have been identified which ^may have affected the appar .339 ent significance of the results, particularly the unusually low 1340 overall incidence of aberrations in the controls."109/ 1342 1345 1346 1347 1350 1350 1351 b. Increased Production of SCEs and Micronuclei in Experimental Animal Test Systems The rulemaking notice points to a variety of studies in 1352 which benzene exposures (including low cumulative exposures) have 1353 resulted in a significant increase of SCEs and micronuclei in 1354 experimental animal test systems.110/ The fact that benzene can 1355 induce these cytogenetic effects even at relatively low levels of 1356 exposure seems clear. However, as Dr. Robert Snyder points out, 1357 "it is not clear what the relationship, if any, is between these 1358 effects and .leukemia at low doses of benzene."111/ The *360 metabolites responsible for the clastogenic effects of benzene 1361 may be different from the metabolites responsible for its 1362 leukemogenic effects. Thus, clastogenic changes may be "parallel 1363 but independent of leukemogenesis,"112/ in which case "the dose 1364 effect relationships would be different, and the pharmacokinetics 1365 would be different."113/ Accordingly, as Dr. Snyder observes: 1367 1367 1341 109/ Iff, at 14-15. 1354 110/ 50 Fed. Reg. 50512, 50527. 1359 Ill/ Snyder Review at 30. 1363 112/ Id. 1365 113/ Id. 256 357 258 - 36 0.OO' o' CP ~-?p 1369 1370 1370 1371 1372 1373 1373 1374 1401 1401 1402 It can be said that in given test systems, benzene can be shown to be clastogenic but not mutagenic, and it can be shown to induce some types of cancer. However, there is as yet no direct link between clastogenic effects and carcinogenic effects of benz ene."114/ In short, at the present time, as the President's 1403 Office of Science and Technology Policy ("0STP") points out, ^the 1406 association between an increase in the frequency of SCEs in cells 1407 in vitro or in vivo and heritable changes has not been clearly 1408 defined."115/ Nor is it clear "that the induction of micronuclei 1417 1417 1374 114/ Id. at 30-31. Dr. Snyder points out that 1375 benzene-exposed workers with increased chromosome aberrations 1376 have been found to have no significant increase in SCEs. See id. 1377 at 15. These discrepancies have led Tice, et al., to suggest 1378 that different molecular mechanisms and metabolites of benzene 1379 may underlie the induction of SCEs and chromosomal aberrations. 1380 See id. at 16, 30. Gebhart has demonstrated a significant lack 1383 of correspondence between SCEs and other cytogenetic changes. 1384 Gebhart, "Sister Chromatid Exchange (SCE) and Structural Chromo- 1385 some Aberration in Mutagenicity Testing," 58 Hum. Genet. 235-254 1386 (1981). The Congressional Office of Technology Assessment 1387 ("OTA") has noted that "the fundamental way in which a particular 1388 chemical interacts with the DNA to produce SCEs may be different 1389 from the mechanism that produces chromosomal aberrations." OTA, 1390 The Role of Genetic Testing in the Prevention of Occupational 1391 Disease at ___ (April 1983). Even if a mechanistic link between 1393 the clastogenic and carcinogenic effects of benzene did exist, 1394 low level benzene exposures capable of producing clastogenic 1395 effects might not result in carcinogenic effects, for "thresholds 1396 for carcinogenesis may be different from the thresholds for 1398 clastogenic effects of the same chemical." Snyder Review at 33. 1408 115/ See 0STP, "Chemical Carcinogenesis; A Review of the 1409 Science and Its Associated Principles," 50 Fed. Reg. 10372, 1410 10404, col. 3 March 14, 1985). As stated by one authority 1411 relied on by OSHA, "SCEs . . . are not, as far as we know, 1412 mutational agents in the strict sense." Bloom, et al., 1413 "Guidelines for Studies of Human Populations Exposed to Mutagenic 1414 and Reproductive Hazards," (Ex. 159-12) at 3. 256 257 258 - 37 - MOD 0000156 1417 is related to heritable changes in cellular phenotypes."116/ Our 1420 present state of knowledge simply does not allow us to conclude 1421 that there necessarily exists sequential series of events 1424 leading from one or more forms of clastogenic changes to the pro1425 duction of leukemia."117/ 1426 1429 1430 1431 1431 1434 c. The Significance of Subclinical Cytogenetic Effects The fact that relatively low cumulative exposures to 1435 benzene may induce various cytogenetic effects (including 1436 chromosomal aberrations and increased production of SCEs and 1437 micronuclei) in humans or experimental animal test systems does 1438 not demonstrate the existence of material health impairment 1439 resulting from low level benzene exposures. These subclinical 1440 cytogenetic effects have not been shown to be precursors of 1441 actual illness; nor can they appropriately be used as a basis for 1442 making clinical predictions. 1444 1445 pie fact that benzene exposure could increase the inci- 1446 dence of chromosomal aberrations was considered during the ^977 1447 Benzene Standard proceeding. At that time, OSHA, as the Supreme 1448 Court noted, "took no definitive position as to what these aber- 1449 rations mea^^in terms of demonstrable health effects 1452 1452 1418 1419 ___________________" 116/ OSTF, "Chemical 10372, 10404, col. 2. Carcinogenesis," supra, 50 Fed. Reg. 1425 117/ 256 257 258 Snyder Review at 30. - 38 - MCD 00001563' 1450 ^ . . . "118/ There is still no evidence indicating that these 1452 cytogenetic effects have any clinical importance.119/ Thus, a 1458 study on which OSHA relies in evaluating the significance of 1460 clastogenic effects acknowledges that "[w]e are aware of no 1461 health consequences of SCEs per se" and states that the presence 1463 of chromosomal aberrations and other "cytogenetic changes cannot 1464 be used to predict specific health effects in an individual."120 1467 As Bloom, et al., explain. 1471 1472 1472 1473 1474 1475 1475 1476 1477 1478 1478 L479 1482 1482 1483 In human populations, no associations have been definitively drawn between those indi viduals with induced (chromosomal] breakage and the subsequent development of cancer. This is true even in Japan, where thousands of persons have been shown to have chromosomal aberrations of radiation origin . . . . Thus . . . the presence of . . . chromosome breakage ... is not ... a har binger of cancer for the carrier individ ual.. " 121/ NIOSH has expressed a similar view, stating that it 1484 ^knows of no data that correlates [chromosomal abnormalities and 1486 1486 1450 1451 118/ Petroleum Industrial Institute, Union Department, 448 U.S. 607, 633 AFL-CIO (1980). v. American 1453 119/ In a recent decision, the Appellate Division of the 1454 Superior Court of New Jersey refused to accept subclinical cellu 1455 lar damage mm a precursor of actual illness or as a basis for 1456 recovery, gee Avers v. Township of Jackson, ___ A.2d ___ (June 4, 1458 1985). 1465 120/ Bloom, et al.. "Guidelines for Studies of Human Popula- 1466 tions Exposed to Mutagenic Reproductive Hazards," (Ex. 159-12) at 1467 3. 1479 121/ Id. at 31. 256 257 258 - 39 - V\C-P 1485 increased frequencies of sister chromatid exchanges] ... to the 1486 manifestation of cancer or adverse reproductive effects in an 1487 individual."122/ Thus, according to NIOSH "the presence of 1491 detectable chromosomal damage does not appear to provide a firm 1493 basis for predicting the likelihood of an individual 1494 demonstrating a tumorigenic response."123/ 1495 1496 Various scientific organizations and individual com- 1497 menters echoed this same theme in OSHA's Ethylene Oxide proceed- 1498 ing. Thus, the American Academy of Industrial Hygiene pointed 1500 out that "although SCEs and chromosome aberrations are indices of 1501 DNA damage, the end points are different, they are not 1502 necessarily correlated, and the relevance of one or the other in 1503 a given application is not obvious."124/ The Academy noted that 1506 there is ^no current concensus [sic] concerning the relationship 1507 between DNA damage and the liklihood [sic] of an ultimate clini- 1508 cal outcome such as cancer."125/ 1510 1510 1510 1510 1510 1510 1487 122/ NIOSH Comments to OSHA Proposed Rule on Occupational 1488 Exposure to Ethylene Oxide (Ex. 11-146 in Docket H-200) at 4 1490 IJune 22, 1983). 1494 123/ Id. 1504 124/ American Academy of Industrial Hygiene Position Paper 1505 on a STEL for Ethylene Oxide (Ex. 175 in Docket H-200) at 2 (July 1506 21, 1984). 1509 125/ Id. 256 257 258 - 40 - MCD 00001568 1511 Other experts commenting in the Ethylene Oxide proceed- 1512 ing expressed similar views. For example, in discussing the sig- 1513 nificance of an increase in SCEs following chemical exposure, 1514 Vernon L. Carter observed that "no relationship has been 1515 established between this response and effects of concern such as 1516 cancer and reproductive problems."126/ Dr. J. W. Grisham stated 1524 that neither chromosomal aberrations nor SCEs "have been corre- 1525 lated with any disease outcome and, indeed, there is evidence 1526 suggesting that SCE may not represent a pathological (toxic) cel- e 1527 lular reaction."127/ And, after describing the mechanism by 1530 which chromosome aberrations and SCEs are induced, Dr. R. Julian 1531 Preston of the Oak Ridge National Laboratory concluded that 1535 1536 there is no evidence to suggest that an *537 increase in SCE is responsible for any other .538 cellular change .... The important point 1538 is that increases in SCE are not represents- 1539 tive of chromosomal changes that themselves 1540 can cause adverse health effects.128/ 1547 1547 1547 1547 1517 126/ Letter of Vernon L. Carter, Jr., DVM to R. Leonard 1518 Vance (Ex. 178 in Docket H-200) at 3 (July 19, 1984). Dr. Betty 1519 Dabney has charactrerized the SCE assay as "highly experimental" 1520 and not now appropriate for use in setting occupational exposure 1521 standards. See Dabney, "The Role of Human Genetic Monitoring in 1522 the WorkplmciT* 23 J. Occup. Med. 626-31 (1981). 1528 127/ Letter of J. W. Grisham, M.D. to Mr. Robert C. Barnard 1529 (Ex. 49 in Docket H-200) at 3 (June 24, 1983). 1540 128/ Preston, "The Induction of Chromosome Aberrations and 1541 Sister Chromatid Exchanges in Human Peripheral Lymphocytes by 1542 Ethylene Oxide, and the Use of Such End Points for Establishing 1543 Exposure Standards" (Ex. 189-16; Appendix A in Docket H-200) at 1544 3. 256 357 258 41 - MCp 000015640 1548 Accordingly, Dr. Preston cautioned that "measurement of SCE in 1549 peripheral lymphocytes is not known to be a predictor of subse- 1550 guent adverse health effects, and data obtained cannot be used 1551 for setting exposure standards."129/ The same is true. Dr. 1552 Preston stated, of measurements of chromosome aberrations in 1553 peripheral lymphocytes.130, 1554 1555 In short, as a recent review of benzene toxicity con 1555 cluded. 1558 1559 1559 1560 1561 1561 1562 1563 1567 T567 568 In EPA*s no relationship between the types of chromo some damage observed [in connection with benzene exposure] and effects on human health can be established at present. The observations have to be taken as indicating a response to exposure to benzene of unknown biological significance.131/ words, "no direct evidence of a [causal] linkage between 1569 chromosomal aberrations and leukemia exists."132/ 1573 1574 OSHA itself has acknowledged the limited and uncertain 1575 significance of these cytogenetic effects, stating that findings 1577 of "chromosomal damage do not provide direct evidence for a 1578 genetic or carcinogenic effect"133/ and contending in court that 1580 1580 1551 129/ Id. at 4. 1553 130/ Id. 1563 131/ ECETOC Report at 20. 1570 132/ Response to Comments on the Regulation of Benzene, 49 1571 Fed. Reg. 23478,23481, col. 3 IJune 6,1984). 1578 133/ 50 Fed.Reg. 50512, 50538, col. 2. 256 257 258 MCD 000015641 42 1580 there is no evidence linking cancer or other adverse health 1581 effects to sister chromatid exchanges.134/ As pointed out by the 1583 Congressional Office of Technology Assessment, the fact is that 1584 "no occupational studies directly relate positive findings for 1585 any chromosomal endpoint with increased risk for any dis- 1586 ease."135/ Thus, the fact that relatively low cumulative 1597 exposures to benzene may increase the incidence of various 1598 cytogenetic markers does not show that a risk of material health 1599 impairment exists at those levels of exposure. Regulatory action 1600 dannot properly be predicated upon the potential occurrence of 1601 such subtle cytogenetic effects. 1602 1602 1603 3. Hematological Effects at the Cellular Level 1604 1605 In recent years, several investigators have studied the 1606 effects of benzene inhalation on bone marrow and splenic progeni- 1607 tor cells, using the spleen colony forming unit X',C^*"S") tech- 1608 nique and the erythroid progenitor cell colony forming unit 1611 1611 1581 1582 134/ January See BNA Occupational 30, 1986 at 915-16. Safety and Health Reporter, 1586 135/ OTA, The Role of Genetic Testing in the Prevention of 1587 Occupational Disease at ___ _____ (April 1983). Investigators from 1588 the Centers for Disease Control, Brookhaven National Laboratory 1589 and Oak Ridge National Laboratory have expressed a similar view. 1590 In the conclusions to a~cytogenetic study of persons living near 1591 the Love Canal, they stressed that it is still impossible to know 1592 whether findings of increased chromosome damage "might predict 1593 later clinical illness in individuals." Heath, et al., 1594 "Cytogenetic Findings in Persons Living Near the Love Canal," 251 1595 J.A.M.A. 1437, 1440 (1984). 256 257 258 43 He'D 0001,6, 1609 ("CFU-E") assay.136/ In these tests, benzene exposures as low as 1611 10 ppm have been found to produce hematologic responses at the 1612 cellular level. These hematologic responses, however, are not 1613 examples of aplastic anemia or leukemia.137/ While they may 1615 possibly indicate the early stages of a disease process, it is 1616 equally likely, as Dr. Robert Snyder points out, that they are 1618 reversible biological responses which will not .lead to 1619 hematological diseases such as aplastic anemia or leukemia 1620 "because the dose is too low and normal repair mechanisms would 1621 nbt permit progression of the disease process."138/ 1623 1624 Rather than predicting potential bone marrow disease, 1625 these subtle hematologic responses may simply reflect the sophis 1626 tication and sensitivity of the assays.139/ Accordingly, as Dr. 527 Snyder observes, the responses seen in these assays must be 1628 interpreted with caution.140/ They "do not, in themselves, pre 1629 dict a .significant adverse health effect. Nor do these data 1630 reflect a potential risk of cancer."141/ 1632 1632 1632 1609 1610 136/ 18-20. See 50 Fed. Reg. 50512, 1614 137/ Sea Snyder Review at 34. 1622 138/ Id. at 34-35. 1626 139/ See id. at 35. 1629 140/ See id. 1631 141/ Id. at 35. 256 ''57 258 - 44 mod 00001564 1634 II. Risk Assessments for Benzene 1635 1637 The rulemaking notice devotes considerable attention to 1638 quantitative risk assessments for benzene. On the basis of these 1639 assessments, OSHA preliminarily concludes that the risk of 1640 leukemia mortality "from a working lifetime exposure to benzene 1641 at 10 ppm is 44-152 cases per 1,000 exposed employees."142/ On 1642 the assumption that the dose-response relationship is linear down 1644 to zero, the risk assessment preferred by 0SHA143/ estimates that 1647 the risk of leukemia mortality from a working lifetime exposure 1648 to 1 ppm of benzene is 5-16.144/ In the pages that follow, we 1649 discuss various issues relating to the risk assessments for 1650 benzene and show that the likely risks are smaller than OSHA has 1651 assumed, particularly at exposure levels in the neighborhood of 1 .652 ppm. 1653 1655 A. Animal Versus Human Data 1656 1658 One question raised in the rulemaking notice is whether 1660 animal data should be used to derive a quantitative estimate of 1660 the cancer risk to benzene-exposed workers and, if so, how this 1662 should be done. The ishort answer to this question is that a risk 1665 1665 1642 142/ 50 Fed. Reg. 50512, 50530, col. 3. 1644 143/ White, et al., A Quantitative Estimate of Leukemia 1645 Mortality Associated With Occupational Exposure to Benzene (Ex. 1646 127) (1982). 1648 144/ 256 257 258 50 Fed. Reg. 50512, 50532, col. 3. - 45 o0oo 1664 assessment for benzene should be based on human epidemiological 1665 data rather than on animal studies. 1666 1667 In the preamble to the proposed rule, OSHA states that 1668 "[t)he underlying epidemiological studies which provide a basis 1669 for the quantification of risk are in general of reasonable 1670 quality . . . and provide a basis for risk assessment."145/ In 1671 these circumstances, the human data should be preferred over the 1672 animal data for estimating the cancer risk that may be associated 1673 with exposure to benzene in the workplace. e 1675 1676 The preference for human data has been expressed by a 1677 variety of authoritative sources and can be considered a guiding 1678 principle in quantitative risk assessment. Thus, EPA's Proposed 1679 Guidelines for Carcinogen Risk Assessment state clearly and 1680 unequivocally: "If available, estimates based upon human 1681 epidemiologic data are preferred."146/ And, in its most recent 1684 risk assessment for benzene, EPA points to "a number of factors 1685 [which] strongly suggest that animal studies are less reliable 1686 than those based upon human responses."147/ OSHA's risk 1692 1692 1670 145/ Id. at 50538, col. 2. 1681 146/ EPA, Proposed Guidelines for Carcinogen Risk Assess- 1682 ment, 49 Fed. Reg. 46294, 46298, col. 1 (November 23, 1984). 1687 147/ EPA Cancer Assessment Group, Interim Quantitative 1688 Cancer Unit Risk Estimates Due to Inhalation of Benzene, February 1689 15, 1985 at 18. See also id. at 24 (unit risks derived from ani- 1690 mal studies are "intrinsically less reliable than those based 1691 upon the human response"). 256 257 258 46 0C`JO0 1564: 1692 assessment contractors in the present proceeding have expressed 1693 the same view, emphasizing that while "comparisons between 1694 estimates-made from human and animal data are instructive, in the 1695 case of benzene, estimates made from human data should take pre 1696 cedence over those from animal data."148/ 1698 1699 The reasons for this preference for human data are easy 1700 to discern. As Sir Richard Doll recently explained, human 1701 epidemiological studies showing a relationship between disease 1701 and estimated cumulative exposure 1705 1706 1706 1707 1708 1709 1710 1710 1711 1712 1712 1713 1714 1715 1720 1720 1721 are better guides to control measures than attempts to extrapolate from the results of animal experiments. For not only are animal experiments unlikely to reflect the exact conditions of human exposure, but we also do not know how to allow for species differences in reaction. Nor, most importantly, do we have any means of predicting quantitative[ly} an effect in animals with a long life like Man from the relationship between life-time exposure to unit dose and cancer incidence in animals with lives measured in weeks rather than years.149/ Thus, while animal data "may help us to understand the 1722 molecular events underlying leukemogenesis in man," and may pro 1723 vide an interesting perspective on risk estimates derived from 1724 human epidemiological studies, the animal data 1728 1728 1697 148/ Crump Report at 33. 1715 149/ Doll, "Epidemiological Discovery of Occupational Can 1716 cers," 13 Annals of the Singapore Academy of Medicine 331, 332 1717 (April 1984). 256 257 258 - 47 \<b o'.o.' 0 1729 1730 1730 1731 1732 1732 1733 1734 1735 1738 1738 1739 are not appropriate for use in a quantitative risk assessment until such time as we can be certain that the mechanistic progression involving pharmacokinetics, metabolism, DNA, alkylation, DNA repair or the lack thereof, promotion, etc., at the exposure levels used in the animal studies are an accurate reflec tion of the course of human benzene-induced leukemia.150/ In short, risk estimates for benzene based on human 1740 data should take precedence over those based on animal studies. 1741 It should be recognized, however, that even if the animal data 1742 were used, the resulting risk estimate would not be higher than 1743 estimates based on human data. This fact has been recognized 1744 both by EPA and by OSHA's own risk assessment contractors. 1746 1747 plus, in its most recent estimate of cancer risks asso- 1748 dated with exposure to benzene, EPA concluded that risk *749 estimates for benzene based on the preferred animal studies ^are 1750 an order of magnitude lower" than risk estimates based upon human 1751 epidemiological data.151/ Drs. Crump and Allen, OSHA's own out- 1754 side experts on this issue, ^reached a similar conclusion. After 1755 performing a variety of risk assessments on both human and animal 1756 data, Drs. Crump and Allen found that their estimates of human 1757 leukemia risk derived from animal inhalation studies were "all 1760 1760 1735 150/ Snyder Review at 28-29. 1751 151/ EPA Carcinogen Assessment Group, Interim Quantitative 1752 Cancer Unit Risk Estimates Due to Inhalation of Benzene, February 1753 15, 1985, at 24-25. 256 257 258 48 ,.o 1758 considerably less than those made from human data."152/ Their 1760 estimates of human risk based upon the NT? Ravage study were 1761 higher than the risks estimated on the basis of the animal 1762 inhalation studies.153/ However, even "the largest estimates of 1764 human risk from the animal studies, derived from all squamous 1765 cell carcinomas in male mice in the NTP study," were found by 1766 Crump and Allen to be lower than those based upon human data.154, 1768 1769 In short, animal data should not be used to perform a 1770 quantitative risk assessment for benzene, but if such data were 1771 used, the result would provide no basis for concluding that risk 1772 estimates derived from human epidemiological studies are 1773 understated. 1774 1776 B. Selection of Epidemiological Data 1777 1779 Although a variety of epidemiological studies pur 1779 porting to indicate an association between benzene exposure and 1780 leukemia are available, the studies by Rinsky, Ott, and Wong 1781 appear to provide more complete exposure data than the others. 1783 Accordingly, recent risk assessment efforts for benzene have 1784 focused on those studies.155/ While the Rinsky and Ott Studies 1785 1785 1759 152/ Crump Report at 33. 1762 153/ Compare id., Table 25, with id., Table 26. 1767 154/ Id. at 33 and Tables 21 and 26. 1784 155/ See, e.g., id. at 15. 256 257 '58 - 49 mod 000015648 1785 may be appropriate for quantitative risk assessment, we believe 1787 that the Wong Study should not be used for that purpose. 1789 1790 As discussed at some length at pages ____ - ____ above, 1791 the Wong Study does not show a significantly increased incidence 1792 of lymphatic and hematopoietic cancer (including leukemia) among 1794 the benzene-exposed workers when compared to the general popula 1795 tion, and the apparent increase in relative risk among the 1796 benzene-exposed workers as compared to the internal control group 1797 is highly questionable. Because of the significant and * 1798 unexplained deficit in leukemia mortality among the internal con 1799 trols, the apparent increase in risk in the exposed workers, as 1800 EPA acknowledges, "may be due to an artifact in the data rather 1801 than to a true carcinogenic response."156/ 1805 1806 This unexplained leukemia deficit among the non-exposed 1807 workers in the Wong Study has been described as the most "strik 1808 ing and statistically significant feature of the Wong] 1809 data."157/ It has led Dr. Brian MacMahon and others to question 1810 whether the study can serve as reliable evidence of even a quali 1811 tative association between leukemia and exposure to benzene.158/ 1814 1814 1801 156/ EPA Carcinogen Assessment Group, Interim Quantitative 1802 Cancer Unit Risk Estimates Due to Inhalation of Benzene, February 1803 15, 1985 at 23-24. 1809 157/ MacMahon Report at 57. 1812 158/ See id. at 58; pp. _____ - _____, supra. 256 257 258 50 -6* 0oOV 1813 But whatever may be the case regarding possible qualitative 1814 inferences, the enormous uncertainty surrounding the study 1815 results (particularly the mortality experience of the unexposed 1816 workers) makes it inappropriate to use the Wong Study for pur- 1817 poses of quantitative risk assessment. 1819 1820 A particularly telling consideration in this regard is 1821 that the mysterious absence of any leukemia at all among the 1822 internal control group has an enormous impact on estimates of 1823 risk based on the Wong Study. Thus, Ors. Crump and Allen found e 1825 that the leukemia deficit among the internal controls more than 1826 doubles the dose-response slope that would exist if the leukemia 1829 experience of the controls in the Wong Study had been normal.159/ 1837 ''SSS A proportional mortality analysis of the Wong Study 1839 performed by Professor Robert Sielken Jr. confirms this 1840 point.160/ Prof. Sielken*s analysis shows that if the 1846 lymphopoietic mortality experience of the non-exposed workers in 1849 the Wong Study had been closer to what normally would have been 1852 1852 1829 159/ Compare Crump Report, Table 12, Data Set VII, with id., 1830 Data Set VIII. In fact, the mortality deficit among the internal 1831 control group has an even bigger impact than Table 12 suggests, 1833 since the effect of the deficit is moderated in Table 12 by the 1834 fact that the results of the Wong Study are aggregated with those 1836 of the Rinsky and Ott Studies. 1840 160/ R. Sielken Jr., A Quantitative Risk Assessment Based on 1841 an Industry-Wide Mortality Study of Chemical Workers 1843 Occupationally Exposed to Benzene. February 1985 (hereinafter 1844 referred to as the "Sielken Report"). A copy of the Sielken 1845 Report is submitted herewith as Appendix J. 256 157 258 - 51 - o' 1851 expected, the estimated cause of death probabilities associated 1852 with benzene exposure would change exponentially. According to 1853 Prof. Sielken: 1856 1857 I^f the number of non-exposed workers whose 1857 death was assumed to have been caused by 1858 lymphatic or hematopoietic cancer was changed 1859 from its observed value of 3 to either 6, or 1860 9, or 12 then the increases in the estimated 1861 probabilities due to durations at positive 1861 dose levels were divided by approximately 1.5 1862 or 3 or 50, respectively. Thus, the 1863 estimated increases (in the probability of 1863 mortality from lymphatic or hematopoietic 1864 cancer] decreased rapidly as the number of 1865 non-exposed workers whose cause of death was 1865 assumed to be lymphatic or hematopoietic can 1866 cer increased.161/ 1870 1870 1871 Based on his analysis. Prof. Sielken concluded that 1874 1875 .875 1877 1877 1878 1879 1880 1880 1881 1884 1884 1885 the estimated increases in the probability of a worker's death being caused by lymphatic or hematopoietic cancer associated with non-zero durations at non-zero doses [of benzene] could easily be several times too large if the number of deaths caused by lymphatic or hematopoietic cancer was unusually low among the non-exposed workers in the study due to chance.162/ indeed, when Prof. Sielken tested the assumption that 1885 the number of lymphatic and hematopoietic cancer deaths among the 1887 non-exposed group was normal (i.e., 13 such deaths), the 1888 best-fitting model 1891 1891 1867 161/ Sielken Report at ii. 1881 162/ 256 257 258 Id. at 27. - 52 0oo^ 1892 1892 1893 1894 1895 1896 1896 1899 1899 1900 indicated no dose-response relationship: that is, for the exposures reported in the study the robability of an exposed worker's death being caused by lymphatic or hematopoietic cancer would be estimated to be the same as it is for a non-exposed worker.163/ In sum, use of the mortality results for the unexposed 1901 workers in the Wong Study significantly distorts any effort to 1902 develop a quantitative risk assessment for benzene. Unless and 1903 until the significant and highly unusual cancer deficit among the 1904 unexposed workers is explained, the Wong Study, as Drs. Crump and 1905 Allen observe, "does not provide strong evidence of a relation- 1906 ship between occupational exposure to benzene and lymphatic and 1907 hematopoietic cancer."164/ For that reason, the Wong Study 1908 should not be used for quantitative risk assessment. If it is 1909 used, however, the data relating to the unexposed workers should 1910 be omitted from the analysis, which would produce risk estimates 1911 that "are not much different from those [that Crump and Allen 1912 derive] from the Rinsky et al. and Ott et al. combined data.11165, 1936 1936 1896 163/ Id. 1907 164/ Crump Report at 24. 1913 165/ Id. Alternatively, the unexposed workers in the Wong 1914 study should be assumed to have had an SMR of 100, which would 1915 produce essentially the same result as excluding them from the 1917 analysis. 1919 As noted above, Dr. Frank Carlborg developed a quanti 1920 tative risk assessment based on data sets for all workers and for 1921 exposed workers only in the Wong Study. He concluded that, from 1936 5 [Footnote continued next page) 256 257 258 - 53 oo(0- 015652 nci> 1940 1941 1942 1943 1947 c. The Crump and Allen Risk Assessment Should Be Preferred Over the Risk Assessments by White, et al. and IARC. -The rulemaking notice discusses three risk assessments 1948 for benzene the analysis performed by White, et al.,166/ the 1952 summary presented by IARC (Ex. 159-58) and the risk assessment 1954 performed under contract to OSHA by Drs. Crump and Allen (Ex. 1955 152). For the reasons discussed below, the risk assessment per- 1956 formed by Drs. Crump and Allen should be preferred over the IARC 1957 and White Risk Assessments. 1958 1959 To assert, as OSHA does, that IARC "conducted a guanti- 1960 tative risk assessment of workers exposed to benzene"167/ surely 1961 is a classic misnomer. What OSHA describes as a "quantitative 1962 risk assessment" is no more than one and one-half pages in an 965* 5 j.965 5 [Footnote continued from preceding page] 5 1922 a statistical standpoint, the data from the study are consistent 1923 with a "no-effect model which assumes that an exposure to benzene 1924 does not affect the incidence of lymphopoietic cancer and that 1925 all the variation among the observed response proportions can be 1926 attributed to age differences among the workers." Carlborg 1927 Report at 7. Only one model tested by Dr. Carlborg provided a 1928 statistically significant improvement in fit over the "no-effect 1929 model." Under that model, which assumes that the lymphopoietic 1930 cancer response depends upon exposure but not upon age, ^a worker 1932 having 45 ppm-years of occupational exposure to benzene would 1933 have an estimated excess risk from lymphopoietic cancer of about 1934 one to two in one thousand, regardless of his age." Id- at 7-8. 1949 1950 1951 1952 166/ White, et al., "A Quantitative Estimate of Leukemia Mortality Associated with Occupational Exposure to Benzene," 2 Risk Analysis 195 (1982) (Ex. 128-37) (hereinafter referred to the "White Risk Assessment"). as 1961 167/ 50 Fed. Reg. 50512, 50531, col. 1. 256 257 58 54 1964 Annex to the IARC Monograph on Benzene. It does not reflect an 1965 examination and analysis of exposure levels, and treats two of 1966 the studies discussed in no more than one or two sentences. 1967 Moreover, IARC does not draw all of the implications that OSHA 1968 itself has tried to draw from the page-and-a-half discussion. In 1969 short, the casual discussion of relative risk contained in the 1970 Annex to the IARC Monograph cannot seriously be treated as a 1971 quantitative risk assessment when compared to the lengthy, care 1972 ful and detailed analysis presented by Drs. Crump and Allen.168/ 1978 1979 The analysis by White, et a^L., can with more justifica 1980 tion be described as a quantitative risk assessment. However, 1981 the White Risk Assessment has been completely superseded by the 1982 much more thorough and detailed risk assessment that was per .983 formed for OSHA by Crump and Allen. 1985 1986 From the outset, the White Risk Assessment has been 1987 seriously criticized. For example. White, et al., based their 1986 risk assessment solely upon workers who had been employed for 1989 five or more years, since they found that "most of the elevated 1990 leukemia risk was observed in this group."169/ They then applied 1992 the results of that analysis to derive risk estimates for workers 1994 1994 1973 168/ Moreover, as Dr. Robert Snyder observes, the risk 1974 estimate presented by IARC appears to be radically inconsistent 1975 with practical experience with the heavy industrial use of 1976 benzene during this century. See Snyder Review at 35. 1990 169/ White Risk Assessment, 2 Risk Analysis at 198. 256 57 258 - 55 000015*54 VICE 1993 at all levels of exposure. Various peer reviewers criticized 1994 this statistical manipulation, calling it ^indefensible" (Dr. 1995 Phillip Cole)170/ and pointing out that it amounts to using "the 1996 same set of data both to formulate and test [the] 1997 hypothesis."171/ As Dr. Brian MacMahon points out, 2002 2003 2003 2004 2005 2006 2006 2007 2008 2008 2012 2012 2013 * there was no a priori hypothesis that such employees would be at greatest risk and there is no justification for selecting this par ticular group post hoc as most representative of the truth. Had the SMR been higher for persons exposed 1 or more years or 10 or more years, such subgroups could, with equal lack of justification, have been chosen for the risk assessment.172/ 0SHA now recognizes that the approach taken by White, 2014 et al., was inappropriate, but attempts to excuse it with the "014 observation that the leukemia mortality experience of workers 2015 having less than five years' exposure appears to be consistent 2016 with the risk estimate developed on the basis of workers with 2017 more than five years' exposure.173/ But such post hoc gerryman- 2019 dering does not remove the suspicion, expressed by Dr. Philip 2020 Cole, that the White Risk Assessment was a biased attempt "to 2021 make benzene appear as hazardous as possible."174/ As Dr. Cole 2026 2026 1995 170/ Cole Critique, supra, X^x* 137) at 3. 1997 171/ March 30, 1984 letter of Professor Norman Breslow to R. 1998 Leonard Vance, Ex. 137. 2009 172/ MacMahon Report at 47. MCD 00001565 2018 173/ See 50 Fed. Reg. 50512, 50533, cols. 2-3. 2022 '23 .j26 5 256 257 174/ also Cole Critique, supra. Ex. 137 at 3. Dr, Charles Brown described the exclusion of workers exposed less than five [Footnote continued next page] 253 - 56 Oo 2026 points out, the White Risk Assessment "appears to have been writ 2027 ten to advocate a particular position rather than to represent a 2028 serious and unbiased attempt to develop a meaningful risk assess 2029 ment for leukemia in relation to benzene."175/ In the words of 2034 Dr. Charles C. Brown, "it appears to be one-sided in its evalua 2035 tion. The presentation is not as balanced as it should be to 2036 serve as the basis for regulatory decisions."176/ 2055 2056 Even if the objectivity of the White Risk Assessment 2057 were not subject to question, it clearly could not be deemed as 2059 2059 5 5 2024 2025 [Footnote continued from preceding page] years as being "an incorrect, biased decision." May 16, 1983 letter of Charles C, Brown, Ph.D. to Ralph E. Yodaiken, Ex. 137. 2030 175/ Cole Critique, supra. Ex. 137 at 2. In this regard. 2031 Professor Norman Breslow stated that~^the paper cannot be 2032 regarded as good science." March 30, 1983 letter of Prof. Norman 2033 Breslow to R. Leonard Vance, Ex. 137. 2037 176/ May 16, 1983 letter of Charles C. Brown, Ph.D. to Ralph 2038 E. Yodaiken. The risk estimates that White, et ai., make for the 2039 Rinsky and Ott studies add to the suspicion of a lack of 2040 objectivity. Thus, as Dr. Brian MacMahon states: 2044 2044 2045 2046 2047 2047 2048 2049 2050 2050 2051 Given the gross assumptions regarding exposure and small numbers on which the estimates of risk are based, the similarity of the estimates of excess risk derived from the data of Rinsky, et al, and of Ott, et al, just have to be contrived or the result of the most extraordinary coincidence. Much of the exposure data in the Rinsky study are assumed and one cannot escape the suspicion that the assumptions have been made with the desired outcome in mind. 2054 MacMahon Report at 46-47. 256 257 258 - 57 - 000015657 MOD 2058 reliable as the more recent risk assessment performed by Drs. 2059 Crump and Allen. Apart from the fact that their risk assessment 2060 is far more objective, detailed, comprehensive and careful than 2061 the analysis of White, et al., the critical fact is that Drs. 2062 Crump and Allen obtained and utilized far more extensive exposure 2063 information than White, et al. 2064 2065 The White Risk Assessment rests on the assumption that 2065 all members of the Rinsky and Ott cohorts were subject to the 2066 same average benzene exposure. The authors simply calculated 2068 what they believed to be a representative average exposure for 2069 the cohort taken as a whole and drew a line from that point 2069 through zero to create a dose-response curve. As Or. Irving 2070 Kessler observes, however: 2074 2075 2075 2076 2077 2082 2082 2083 This Averaging of exposures over numbers of workers and time represents only a first -and very poor -- method of dealing with defi cient information.177/ approach is particularly susceptible to producing unreliable 2084 estimates when applied to studies, such as the Rinsky study, 2085 where the incidence of leukemia is small and the range of 2086 exposure levels within the workplace and among the work force is 2087 wide.178/ In such cases, the average benzene exposure of the 2091 2091 2077 2078 177/ March 8, Leonard Vance, Ex. 1983 _137. letter of Irving E. Kessler, M.D. to R. 2087 2088 2091 5 256 257 258 178/ See B. Goldstein, "Benzene Toxicity: Review of Recent Literature11 ("February 3, 1983) (Ex. ____) at 4-6; EFA, Ambient [Footnote continued next page) - 58 56^ 2091 cohort as a whole (even if properly determined) may not be rea- 2092 sonably representative of the exposures of those workers showing 2093 a leukemogenic response.179/ Thus, if we are to have any confi- 2095 dence at all in the reliability of the risk assessment, it is 2096 critical that an attempt be made to develop individual exposure 2097 profiles, rather than arbitrarily assigning a single average 2098 exposure value to the entire cohort. 2099 2100 That is precisely what Drs. Crump and Allen did in 2101 developing a risk assessment for OSHA. In contrast to the White 2102 Risk Assessment (not to mention the brief IARC discussion). Crump 2103 and Allen have performed a comprehensive, detailed and 2104 worker-specific analysis of the occupational benzene exposures of 2105 the Rinsky cohort. They did not simply attribute a single aver- 2106 age benzene exposure to all members of the cohort and assume that 2107 the increased leukemia risk found among members of the cohort (or 2108 among members of a subgroup within the cohort) was attributable 2109 to the average exposure value for the cohort as a whole. 2110 Instead, Crump and Allen obtained the underlying data tapes from 2111 the Rinsky study and developed a complete exposure profile for 2112 each worker in the cohort.180/ They were thus able to evaluate 2114 2114 5 [Footnote continued from preceding 5 2089 Water Quality Criteria for Benzene 2090 C-60. page] (October 1980) at C-59 and 2093 179/ 2094 at 6. See B. Goldstein, "Benzene Toxicity," supra (Ex. _____) 2113 180/ See Crump Report at 11 and Appendix B. 256 257 258 59 2114 the increased risk on the basis of a more highly refined data set 2115 with a larger number of risk-exposure points than previous inves- 2116 tigators, including White/ et al. and IARC. 2117 2113 In sum, quite apart from questions about its 2119 objectivity, the White Risk Assessment (and, a fortiori, the IARC 2120 discussion of risk) is far less reliable and less suitable for 2121 regulatory decisionmaking than the risk assessment developed by 2122 Drs. Crump and Allen because, as the authors of the Rinsky study 2123 recently observed, the IARC and White Risk Assessments are "based * 2124 on estimates of group exposure rather than on estimates of the 2125 exposure of individual workers. The resultant risk estimates 2126 were subject, therefore, to wide variances."181/ By contrast, 2129 the Crump and Allen risk assessment, as EPA points out, "contains 130 much better human exposure estimates in the epidemiologic studies 2131 than has been previously available."182/ Accordingly, OSHA's 2135 quantitative evaluation of the potential cancer risk presented by 2136 occupational exposure to benzene should be based on the Crump and 2137 Allen risk assessment rather than on previous reports. 2139 2139 2139 2139 2139 2139 2139 2127 2128 181/ Rinsky, et al., "Benzene and Epidemiologic Risk Assessment," August Leukemia: An 9, 1985 (Ex. 176A) at 4. 2132 182/ EPA Carcinogen Assessment Group, Interim Quantitative 2133 Cancer Unit Risk Estimates Due to Inhalation of Benzene, February 2134 15, 1985 at 2. 256 257 258 - 60 2143 2144 2145 2146 2147 2143 2152 D. The Crump and Allen Risk Assessment Conservatively Indicates That the Increased Risk of Leukemia From a Working Lifetime Exposure to 1 ppm Benzene Is Approximately An Order of Magnitude Lower at Exposure Levels of 1-10 ppm Than OSHA Has Assumed. In the rulemaking notice, OSHA preliminarily estimates 2153 that the increased risk of leukemia from a working lifetime 2154 exposure to 10 ppm benzene is approximately 44-152/1,000 workers 2155 and that at an exposure level of 1 ppm, the increased risk is 2156 approximately 5-16/1,000.183/ By contrast, even when various 2158 conservative assumptions are made, the Crump and Allen risk 2159 assessment indicates that the increased risk of leukemia is 2161 almost an order of magnitude lower than OSHA suggests. 2163 2164 Although Drs. Crump and Allen analyzed the 2164 epidemiological data for benzene using a variety of exposure 165 models, they prefer the weighted cumulative exposure model, since 2167 it fits the data somewhat better than the other models tested and 2168 is more consistent with the latency pattern of leukemia.184/ 2169 Crump and Allen analyzed the weighted cumulative exposure data 2170 under relative risk and absolute risk forms of the dose-response 2172 model.185/ They found that, under an absolute risk model, the 2181 2181 2156 183/ See 50 Fed. Reg. 50512, 50532. 2169 184/ See Crump Report at 23-24. 2172 2174 2175 2176 2177 2178 2179 185/ The relative risk model assumes that the increased risk of benzene-related leukemia mortality is proportional to the background leukemia mortality -- i.e., it assumes that the carcinogenic potency of benzene increases with age. The absolute risk model assumes that the increased risk of benzene-related leukemia mortality ia the same at all ages, given equal doses. See id. at 16. J56 257 258 61 - MCD 0000156 2181 increased risk of leukemia mortality after a working lifetime 2182 exposure to 10 ppm benzene is 15/1,000, while under a relative 2184 risk model, it is 29/1,000186/ The comparable risk values after 2135 a working lifetime exposure to 1 ppm benzene were found to be 2186 1.5/1,000 under the absolute risk model and 3/1,000 under the 2187 relative risk model.187/ 2188 2189 The weighted cumulative exposure models favored by 2190 Crump and Allen do not include data from the Wong Study, since 2191 weighted cumulative exposure data were not available from that 2192 study. As discussed at pages _____ - _____ above, the Wong Study is 2193 not appropriate for use in performing quantitative risk assess 2194 ments for benzene. Accordingly, the fact that the risk estimates 2196 based on the weighted cumulative exposure model do not reflect 197 data from the Wong Study makes those estimates more reliable than 2198 estimates which include the Wong data. 2199 2200 Even if the Wong Study were included, however, the risk 2201 estimates would not be significantly different, as long as the 2202 unexposed workers from the study are excluded or assumed to have 2203 a normal leukemia experience (i.a., to have an SMR for leukemia 2205 mortality of 100).188/ Moreover, as discussed above,189/ Dr. 2211 2211 2184 186/ See id., Table 21 2187 187/ See id. 2205 2206 2211 5 ``*56 57 258 188/ Compare Crump (dose-response slope of Report, Table 12, Data Set IV 0.13 for the Rinsky and Ott studies (Footnote continued next 62 page] MCI) 000015661 2211 Frank Carlborg found that a quantitative risk assessment of data 2212 from the Wong Study would produce an estimated increased leukemia 2213 risk of approximately 1-2/1,000 after 45 years of occupational 2214 exposure to 1 ppm benzene, a level of risk that is almost pre- 2215 cisely the same as the increased risk estimated by Crump and 2216 Allen under the absolute risk-weighted cumulative exposure model. 2218 2219 The increased risk of 1.5-3/1,000 after a lifetime 2220 exposure to 1 ppm benzene estimated by Crump and Allen also is in 2221 close agreement with the most recent risk estimate for benzene 2222 developed by EPA's Carcinogen Assessment Group ("CAG"). In that 2223 document, CAG developed a composite unit risk estimate of 2224 0.026/ppm (i.e., an increased risk of 26/1,000) after 70 years of 2225 continuous exposure to 1 ppm benzene.190/ This continuous 2228 70-year lifetime exposure risk estimate equates to an increased 2229 risk of slightly under 3/1,000 for 40 years' occupational 2230 exposure to 1 ppm benzene.191/ This composite CAG risk 2240 2240 5 [Footnote continued from preceding page] 5 2207 alone), with id., Data Set VIII (dose-response slope of .016 for 2208 the Rinsky, Ott and Wong studies assuming SMR=100 for unexposed 2209 workers in the Wong Study). 2210 189/ See p. ______ n._, supra. 2226 190/ See EPA Carcinogen Assessment Group, Interim 2227 Quantitative Cancer Unit Risk Estimates Due to Inhalation of 2228 Benzene, February 15, 1985 at 22-23. 2231 191/ 2232 years' 5 5 256 257 258 The increased risk of 0.026 which CAG estimates for 70 continuous exposure to 1 ppm benzene can be translated [Footnote continued next page] - 63 - 00001566 mod 2240 assessment, which is in such close agreement with the weighted 2241 cumulative exposure model estimates of Crump and Allen, gives 2242 equal weight to cumulative dose and weighted cumulative dose 2243 data, as well as to relative and absolute risk models.192/ 2248 2249 Thus, the best available evidence!93/ in this proceed- 2252 ing indicates that, taking a conservative approach to risk 2254 assessment, the increased risk of leukemia after a working J.ie- 2255 time exposure to 1 ppm benzene would be in the neighborhood of 2256 1-3/1,000. * It must be emphasized, however, that, for a variety 2257 of reasons, this estimate almost certainly overstates the true 2258 risk -- if, indeed, any risk whatsoever exists at exposure levels 2259 below 10 ppm. 2260 2261 First, as discussed above, no increased risk of 2262 benzene-related leukemia has been demonstrated at levels of 10 2263 ppm and below.194/ To the contrary, as shown in the Comments of 2264 2264 5 [Footnote continued from preceding page) 5 2233 into a risk from 40 years' occupational exposure by multiplying 2235 by a factor of 0.111 to account for the fact that a worker is 2236 exposed only 8 hours per day, approximately 210 days per year for 2237 40 years. Thus, 210/360 x 8/24 x 40/70 = .111. Multiplying 2238 0.026 x 0.111 = 0.0029 -- or an increased risk of 2.9/1,000 after 2239 40 years of occupational exposure to 1 ppm benzene. 2244 2245 2246 192/ See SPA Carcinongen Quantitative Cancer Unit Risk Benzene, February IS, 1985 at Assessment Group, Interim Estimates Due to Inhalation 22. of 2249 193/ See Section 6(b)(5) of the Occupational Safety and 2250 Health Act of 1970 XoccuPationai health standards should be set 2251 "on the basis of the best available evidence'1). 2263 194/ 256 257 258 See pp. ______ - _____, supra. - 64 - .qV 0 2264 Exxon Company, U.S.A. submitted in this proceeding, 2265 epidemiological studies of workers exposed to low levels of 2266 benzene (including a recently published study of 21,698 employees 2267 and retirees of the Exxon Company, U.S.A.) indicates that no 2268 increased risk of leukemia exists at low exposure levels. 2270 2271 The absence of a demonstrated increased risk of 2272 leukemia at exposures of 10 ppm and below is consistent with the 2273 possibility that there is a threshold for benzene-related 2274 ^leukemia above 10 ppm. Although OSHA has presumed, as a matter 2275 of policy, that no threshold exists for the carcinogenic effects 2276 of benzene, the fact is, as Dr. Bernard Goldstein points out, 2277 that a "major question" is whether the dose-response for 2278 benzene-related leukemia is "in fact, a linear relationship 2279 extrapolating back to zero ... or whether there [is] some no 2280 effect level, which is inadequate to cause leukeraogenesis."195/ 2283 EPA also has recognized that the question whether the 2284 non-threshold presumption should be applied to benzene "is not 2285 without uncertainty."196/ 2287 2288 Dr. Robert Snyder points out that ^[t]he data strongly 2289 indicate that one or more metabolites of benzene mediate benzene 2292 2292 2281 195/ 2282 Mehlman, B. Goldstein, "Clinical Hematotoxicity of Benzene," ed.. Carcinogenicity and Toxicity of Benzene at 57. in 2285 196/ Response to Comments on the Regulation of Benzene, 49 2286 Fed. Reg. 23478, 23479, col. 3 (June 6, 1984). 256 257 258 - 65 - 0000^ 4 2291 toxicity."197/ He goes on to explain: 2294 2295 2296 2296 2297 2298 2299 2299 2300 2301 2301 2302 2305 2305 2306 Dr. Because of the need for a complex metabolic intervention . . - the concentration of benzene necessary to produce these [toxic 1 changes [and binding to DNA] must be real and measurable. In other words, there must be a threshold level of benzene which can be metabolized to a toxic metabolite. Enzyme kinetics teach us that the lower the concen tration of substrate the slower the reaction and, below critical levels, reaction rates may be insufficient."198/ ' Snyder concludes that the "empirical evidence does not permit 2307 us to reach firm conclusions one way or the other" regarding the 2308 existence of threshold levels for carcinogens."199/ However, he 2309 points to a variety of biochemical and molecular factors which 2310 support the hypothesis that a carcinogenic threshold for sub 2311 stances such as benzene may indeed exist. In Dr. Snyder's words. 2315 2316 2316 2317 2318 2319 2320 2320 2321 2322 2323 2323 2327 2327 2327 2291 197/ 2302 198/ 2309 199/ it is apparent that the process of carcin ogenesis can be interrupted at several stages both during initiation and promotion. It is reasonable, therefore, to asstime that each of these stages represents a threshold level for carcinogenesis. Thus, there are many thresh old levels for carcinogenesis rather than a single threshold. Moreover, these thresholds for carcinogenesis may be different from the thresholds for clastogenic effects of the same chemical.200/ Snyder Review at 22 id. at 31. Id. 2323 200/ Id. at 33. 256 257 258 - 66 - 2328 In the case of benzene, there are some special reasons 2329 to suspect that a carcinogenic threshold may exist. For one 2330 thing, asEFA points out, "the weight of evidence as a whole 2331 [makes] it . . . evident that benzene may exert its carcinogenic 2332 effect via non-genotoxic mechanisms."201/ Moreover, as Dr. 2334 Snyder reminds us, "an underlying question which has yet to be 2335 adequately dealt with is: 2338 2339 2339 2340 2341 2342 2343 2343 2344 2345 2345 2355 2355 2356 * Must frank bone marrow damage preclude a leukemogenic response to benzene? ... If aplastic anemia or other severe bone marrow damage is a precursor to benzene-induced leukemia it would appear that reasonably high exposure to benzene, i.e., doses high enough to produce severe bone marrow depression, would be necessary to induce leukemia, even tehough some clastogenic responses could be produced at _lower doses.202/ As the preceding discussion shows, there is good reason 2357 to believe that benzene exposures at levels of 10 ppm and below 2358 will not be associated with any increased risk of leukemia. But 2359 even if they were, the risk existing at exposure levels in the 2360 neighborhood of 1 ppm is likely to be considerably lower than the 2361 value of 1.5-3/1,000 estimated by Crump and Allen. Crump and 2363 2363 2332 2333 201/ 1984) EPA, Draft Criteria Document at XI-16. for Benzene (February 2346 202/ Snyder Review at 34. In this regard. Dr. Bernard 2347 Goldstein makes the point that "[i]f leukemia is a conseqeunce 2348 solely of significant pancytopenia, then a more stringent stan- 2349 dard [than the current 10 ppm lmit] is unnecessary." B. 2350 Goldstein, "Clinical Hematotoxicity of Benzene," in Mehlmen, ed., 2351 Carcinogenicity and Toxicity of Benzene at 57. 256 257 258 - 67 A vO,ov 2362 Allen used a linear model to extrapolate risks to exposure levels 2363 of 1 ppm. However, it is far from clear that the dose-response 2364 for benzene-related leukemia is linear, particularly at dose 2365 levels in the range of 1-10 ppm. 2367 2368 Linear extrapolation models were developed to reflect 2369 the apparent dose-response relationship observed in radiation 2370 carcinogenesis. Chemical carcinogenesis, however, differs from 2371 radiation carcinogenesis in several significant respects, includ- 2372 ing the following: 2373 2375 2376 2377 2378 2379 2380 2381 '382 2383 2383 2364 2385 2386 2387 2386 2389 2390 2390 2391 2392 2394 Chemical agents are inhibited by physical trans~ port barriers, while radiation reaches cell nuclear material without such inhibitions. ^ Many chemical agents, such as benzene, require metabolic activation; radiation does not. ^ In contrast to radiation, the body has various detoxification, excretion and repair processes that operate on chemical agents. As OSHA acknowl edges, a linear model "cannot take into account repair, detoxification reactions and metabolic activation."203/ The high level of energy ion radiation can break chemical bonds; by contrast, chemical reactions are modulated by the limited molecular energies available from the reactants to overcome activation energy. plus, radiation derived extrapolation models reflecting 2395 linearity may not be appropriate for chemical carcinogenesis even 2396 if they accurately reflect the effects of radiation. However, 2398 2398 2386 203 50 Fed. Reg. 50512, 50530, col. 3. 256 257 258 - 68 - tfCD c 2397 questions have been raised as to whether linear models are appro* 2398 priate even in the case of radiation when low doses are involved. 2399 Thus it has recently been reported that at radiation doses below 2400 10 rad., the carcinogenic effect per rad. diminishes markedly, 2401 thus throwing into question assumptions of low dose linearity 2402 even in the case of radiation.204/ 2406 2407 In short, the assumption of low dose linearity for 2408 benzene*related leukemia may not be well founded. As EPA 2408 acknowledges, there are data suggesting that dose*response curves 2409 for carcinogens are nonlinear, and the data for benzene "do not 2410 conclusively support either [a linear or nonlinear) 2411 hypothesis."205/ Dr. Bernard Goldstein also points to the 2414 assumption of low*dose linearity as one of the major questions 2415 and difficulties with risk assessments for benzene206/ And Dr. 2418 Irving Kessler observes that a no-threshold linear model for 2419 benzene-related leukemia "is not very likely."207/ 2423 2423 2423 2423 2403 204/ See Kohn and Fry, "Medical Progress -- Radiation Car- 2404 cinogenesis/* 310 New Eng. J. Med. 504-511 2411 205/ Proposed Withdrawal of Proposed Benzene Standards, 49 2412 Fed. Reg. 8386, 8387, col. 3 (March 6, 1984). 2416 2417 2418 206/ See B. in Mehlman, ed.. 57. Goldstein, "Clinical Hematotoxicity of Benzene," Carcinogenicity and Toxicity of Benzene at 55, 2420 207/ March 8, 1983 Letter of Irving I. Kessler, M.D., to R. 2421 Leonard Vance, Ex. 137. See also December 13, 1982 letter of 2422 Bruce W. Karrh, M.D. to Leonard Vance, Ex. 137. 256 257 258 - 69 noo^5661 X) 2424 Further evidence for non-linearity in the dose-response 2425 curve for benzene-related leukemia at levels between 1 and 10 ppm 2426 is provided by the most recent update of the Pliofilm workers 2427 study. After analyzing additional exposure and mortality infor 2428 mation for an expanded Pliofilm cohort, the authors of that study 2429 concluded that "the association between cumulative benzene 2430 exposure and leukemia" was explained best by a log-linear model, 2431 under which the incidence of benzene-induced leukemia decreases 2432 exponentially between exposure levels of 10 ppm and 1 ppm.208/ 2435 Uhder this analysis, the increased leukemia risk would decline by 2436 a factor of 130, rather than by a factor of 10, as exposures 2437 decline from 10 ppm to 1 ppm.209/ If the dose-response curve for 2441 benzene-related leukemia does follow this particular log-linear 2442 relationship between 10 ppm and 1 ppm, the increased risk of 2443 1.5-3/1,000 at 1 ppm estimated by Crump and Allen could be high 2444 by more than an order of magnitude, and the true risk might be in 2445 the neighborhood of 0.1-0.3/1,000. Of equal importance is the 2446 fact that, based on the latest update of the Rinsky study, the 2448 additional health benefit of reducing exposures below 1 ppm would 2449 be negligible.210/ 2455 2455 2433 2434 208/ Risk Rinsky, Assessment,*1 et al., ^Benzene and Leukemia: Epidemiological August 9, 1985 (Ex. 176A) at 2, 18. 2439 209/ M. at 20. 2449 210/ Thus, while the study estimates a 13,000 percent reduc 2450 tion in risk from 10 ppm to 1 ppm (a reduction in the odds ratio 2452 from 221.4 to 1.7), it predicts only a 38% reduction in risk 2453 (from 1.7 to 1.06) as exposure are reduced form 1 ppm to 0.1 ppm. 2454 See id. at 20. 256 257 258 70 MCI) 000015668 2456 In sum. Crump and Allen's assumption of low-dose 2457 linearity was extremely conservative. They themselves recognize 2458 that their estimates based on linear models "should be regarded 2459 [as] plausible upper bounds' to the risk/' particularly at a 2460 dose level of 1 ppm.211/ EPA has noted that use of a linear 2463 model may overestimate benzene-related leukemia risks substan2464 tially -- perhaps by as much as one or two orders of magni- 2465 tude.212/ The most recent update to the Rinsky study, as dis- 2470 cussed above, suggests that this may indeed be the case with 2471 respect to benzene exposures in the neighborhood of 1 ppm. Thus, 2472 at the very least, the assumption of linearity may have had an 2473 important effect on the risk estimates for exposure to 1 ppm 2474 benzene.213/ Since ^the risks under non-linear models decrease 2476 rapidly with decreasing dose,"214/ the fact that Crump and Allen 2478 used only linear models means that the risks they have estimated 2479 at exposures of 1 ppm are on the high side. 2481 2482 Furthermore, it is important to bear in mind that the 2483 risk assessments performed by Crump and Allen, as well as the one 2485 2485 2461 211/ Crump Report at 35. 2465 212/ See Proposed Withdrawal of Proposed Benzene Standards, 2466 49 Fed. Reg. 8386, 8387, col. 3; Proposed Recommended Maximum 2467 Contaminant Levels for Volatile Synthetic Organic Chemicals in 2468 Drinking Water, 49 Fed. Reg. 24330, 24348, col. 2 (June 12, 2469 1984). 2474 213/ See Crump Report at 8. 2477 214/ Id. at 34. 256 257 258 - 71 - MCD 000015669 2484 performed by White, et al., assume that members of the cohorts 2485 that were studied did not have any other occupational exposure to 2486 benzene apart from their experience in the work operation being 2488 studied- If, for example, workers in the Pliofilm cohort in the 2489 Rinsky Study worked in tire building operations as well, they 2491 might have had significant additional benzene exposure both 2492 through inhalation and as a result of dermal absorption.215/ 2506 Needless to say, if the SMRs in the Rinsky study really were 2507 associated with significantly higher cumulative benzene exposures 2508 than Crump and Allen assumed (because of additional exposures 2509 during tire building), the dose-response relationship and result- 2510 ing risks would be much lower than Crump and Allen calculated. 2512 2513 Finally, it is important to bear in mind that the 2514 leukemia risk of 1.5-3/1,000 estimated by Crump and Allen is a 2515 risk projected to occur at age 60, after 40 years of occupational 2516 exposure to 1 ppm benzene have been accumulated. Increased risk 2517 levels would be smaller at earlier stages of the worker's career. 2519 2519 2493 215/ 0SHA estimates that approximately 32 mg of benzene per 2494 day could be absorbed through the skin as a result of building 2495 tires with a solvent containing 0.5 percent benzene. See 50 Fed. 2496 Reg. 50512, 50528, col. 3 - 50529, col. 1. OSHA also estimates 2497 that 8-hour exposure to 1 ppm benzene results in an intake 2498 through inhalation of 14 mg of benzene. See id. at 50529, col. 2499 1. On that basis, if members of the Rinsky cohort also had 2500 worked in tire building operations and if, for example, the per- 2501 centage of benzene in the solvents used during the 1940s and 2502 1950s were 5 percent, the potential daily intake of benzene 2503 through dermal absorption alone would have been 320 mg, or the 2505 equivalent of an 8-hour inhalation exposure to 22 ppm benzene 2506 J_320 mg divided by 14 mg * 22 ppm). 256 257 258 72 6-1 00^ 00 2518 Moreover, a worker who has attained 60 years of age would be sub- 2519 ject to significant risks from a large number of competing causes 2520 of death. This further emphasizes the speculative and 2521 hypothetical nature of the risks estimated by Crump and Allen, as 2522 well as in all of the other risk assessments that OSHA has con- 2523 sidered. 2524 2524 2528 E. Some Perspective on the Significance 2529 of the Risk. 2530 2534 In the preceding pages, we have shown that, using a e 2535 variety of conservative assumptions, the increased risk resulting 2536 from a working lifetime exposure to 1 ppm benzene would be no 2537 higher than 1.5-3/1,000. Moreover, because 2538 2540 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2553 the ^ the Rinsky cohort may have had additional occupational exposures to benzene that are not reflected in the underlying epidemiological studies. Q there may well be a threshold for benzene-related leukemia above 10 ppm, and in any event, there is good reason to believe that the dose-response relationship for benzene-related leukemia is non-linear in the dose range of 1-10 ppm. risks estimated by Crump and Allen on the basis of 2554 non-threshold linear models are likely to be greatly overstated. 2555 The best available evidence suggests that, if any risk of 2556 benzene-related leukemia exists at all at exposure levels in the 2557 neighborhood of 1 ppm, the increased risk after a working life- 2558 time of such exposure is likely to be well below 1/1,000. 256 257 258 - 73 - 2561 We do not believe that a hypothetical occupational risk 2562 of this size should be considered significant, particularly since 2563 the true risk, if any, is likely to be substantially lower. As 2565 OSHA recognizes, occupations such as firefighting and mining 2566 involve increased mortality rirks in the neighborhood of 2567 20-30/1,000 employees, while the lifetime mortality risk for 2568 manufacturing occupations overall is 2.7/1,000.216/ Even in ser 2569 vice employment, the increased mortality risk is 1.62/1,000.217/ 2570 An increased mortality risk of approximately 1/1,000 after a 2571 working lifetime exposure to 1 ppm benzene compares favorably to 2573 mortality risks in these other occupations. It also is instruc 2573 tive to note that the increased cancer mortality risk of 1/1,000 2574 is only 0.6 percent of the background lifetime risk of fatal can 2576 cer in the United States.218/ 2580 2581 In sum, whatever may be said about the potential 2582 increased risk of leukemia mortality after a working lifetime 2583 exposure to 10 ppm benzene, there simply is no basis for 2584 concluding that the hypothetically increased risk of 1/1,000 or 2585 less after a working lifetime exposure to 1 ppm benzene is sig 2586 nificant. Given the enormous uncertainty surrounding the 2588 2588 2568 216/ See 50 Fed. Reg. 50512, 50539, col. 1. 2570 217/ Id. 2576 218/ See Radionuclide Standards, 49 Fed. Reg. 43906, 43910, 2577 col. 3 (October 31, 1984) (background lifetime risk of cancer 2578 mortality in the U.S. is about 165/1,000). 256 257 258 - 74 2587 estimation of any increased risk of leukemia at benzene exposures 2588 in the neighborhood of 1 ppm, OSHA should studiously avoid any se ev/jT' 2589 suggestion that theflhealth risk that would remain under its pro- 2590 posed standard,wyuld sLill~Lie jjgnifleant.219/ 2599 2599 2602 III. Feasibility Issues 2605 2607 OSHA*a preliminary determination that the proposed 2608 standard would be feasible in the petrochemical industry is based 2609 on a report prepared for OSHA by JRB Associates, entitled "Tech- 2611 nological Feasibility and Economic Impact Study of Alternative 2612 Standards for Benzene" (Ex. 153) (hereinafter referred to as the 2613 "JRB Report"). The JRB Report, as supplemented by an Addendum 2614 prepared by Meridian Research, Inc. XEx* 155), provides a profile 2615 of the industry and of current employee exposures, identifies the 2617 assumed baseline controls and additional controls that JRB 2618 asserts would be sufficient to achieve compliance with a 1 ppm 2619 standard, estimates the cost of compliance, and assesses the 2620 impact that such costs would have on the industry, its customers, 2621 and employees. In the following pages, we discuss various of 2623 these points and show that JRB's analysis is flawed or incomplete 2625 2625 2591 219/ The fact that of 16 countries listed in the JRB Report, 2592 only 2 (Switzerland and the USSR) have established permissible 2593 exposure limits or guidelines below 5 ppm further supports the 2594 view that no significant health risk exists at benzene concentra- 2595 tions of 1-5 ppm. See JRB Report, Table 1-4. I_t should be noted 2596 that 8 of the 16 countries are reported by JRB to have permissi- 2597 ble exposure limits of 10 ppm or higher. Id. 256 257 258 - 75 - n q00oV 2624 in a variety of respects and that OSHA's preliminary conclusions 2625 regarding feasibility are, therefore, inadequately supported and 2626 subject to serious question. 2629 2630 A. Profile of Current Petrochemical Industry Operations 2631 2634 Table 3-3 in the JRB Report shows the number of 2636 petrochemical units using benzene as a feedstock by type of major 2637 benzene derivative or coproduct produced. According to the 2638 table, there are a total of 117 such petrochemical units, which 2639 JRB divides into seven categories. We have examined Table 3-3 2640 and the accompanying notes and find that it does not present an 2641 accurate profile of the petrochemical industry today. CMA Table 2642 3-3 below shows JRB's profile and a corrected profile prepared by 2643 CMA. 2644 2644 2644 2644 2644 2644 2644 2644 2644 2644 2644 2644 2644 2644 2644 2644 2644 2644 2644 2644 2644 2644 2644 256 257 258 - 76 - 0000 V&67 4 2646 CMA Table 3-3 2647 2651 Number of Petrochemical Units Using Benzene as a Feedstock 2652 Bv Type of Major Benzene Derivative or Coproduct Produced 2653 2655 Benzene Derivative/Coproduct Number of Petrochemical Units 2656 2657 JRB CMA 2658 2659 Ethylene.......................................... 2660 Ethylbenzene/Styrene. .46 .16 3*7 10 2661 2662 2663 Alkylbenzene............................ Nitrobenzene/Aniline. Cyclohexane............................. .5 .8 .11 5 3 8 2664 Monochlorobenzene.... 2665 Cumene/Pheno1......................... 2666 Other.................................................. .5 .26 .0 4 12 2 2667 2668 Total Number of Units 117 r/ 82 2670 2670 2673 The explanation for CMA1s corrections is as follows. 2674 2675 1. Ethylene Units -- JRB assumed that there are 35 2676 ethylene-producing facilities in the United States, with an aver- 2677 age of 1.3 ethylene units per facility.220/ CMA has gathered 2678 information on each individual ethylene facility. We find that 2679 rather than the 35 ethylene-producing facilities asssumed by JRB, 2680 there presently are only 31. The ethylene units at the Arco 2681 facility at Wilmington, California, the Cities Service facility 2682 at Lake Charles, Louisiana, the Koch facility at Corpus Christi, 2683 Texas, and the Texaco facility at Port Neches, Texas, have been 2684 shut down.221/ Seven of the remaining 31 facilities have two 2688 2688 2677 220/ See JRB Report, Table 3-3, Note a. 2685 221/ When we use the term "shut down," we mean either that 2685 the unit has been dismantled or that there are no current plans 2686 to place it back in operation. 257 258 - 77 hCD 000015b( 2688 operating units each,222/ thus bringing the total number of eth2693 ylene units to 38. 2694 2695 2. Ethvlbenzene/Styrene Units -- Benzene is present 2696 in ethylbenzene units, where benzene and ethylene are combined to 2697 make ethylbenzene. Virtually no benzene is present in styrene 2698 units, where ethylbenzene is dehydrogenated to make styrene. 2699 Therefore, in a styrene plant where ethylbenzene is supplied as a 2700 feedstock rather than being manufactured, occupational exposure 2701 to benzene will be minimal.223/ JRB assumes that benzene is used 2702 at all ethylbenzene and styrene plants and that all of the plants 2703 are in operation and have one unit per facility. These assump- 2704 tions should be corrected as follows: Five of the ethylbenzene 2705 units do not use benzene.224/ The ethylbenzene and styrene units 2712 at two of these plants (American Hoechst at Baton Rouge, 2713 Louisiana and U.S. Steel at Houston, Texas) have been shut down. 2714 The Cosmar facility at Carville, Louisiana has two units rather 2716 2716 2688 222/ These facilities are Amoco at Chocolate Bayou, Texas; 2689 Arco at Channelview, Texas; Dow at Plaquemine, Louisiana; DuPont 2690 at Chocolate Bayou, Texas; Tennessee Eastman at Longview, Texas; 2691 Phillips Petroleum at Sweeny, Texas; and Shell at Norco, 2692 Louisiana. 2701 223/ See JRB Report at D-18. 2706 224/ These are Charter at Houston, Texas; Conoco at Choco- 2707 late Bayou, Texas; and Tenneco at Chalmette, Louisiana (all of 2708 which recover "native'' ethylbenzene rather than synthesizing 2709 ethylbenzene from benzene); Arco at Beaver Valley, Pennsylvania; 2710 and Dow at Midland, Michigan (both of which are styrene units 2711 which use ethylbenzene as a feedstock.) 256 257 258 - 78 - HOP 00001* 2715 than one. The net effect of these changes is a decrease of six 2716 ethylbenzene/styrene units from the number assumed by JRB. 2717 2718 3. Nitrobenzene/Aniline -- Three of the 2719 nitrobenzene/aniline units reflected in JRB's Table 3-3 have been 2720 ^hut down.225/ A fourth, U.S. Steel at Haverhill, Ohio, uses 2724 phenol rather than nitrobenzene. A fifth, the second Mobay unit 2725 at New Martinsville, West Virginia, produces aniline as a 2726 by-product in the production of iron oxide. No benzene is pres2727 ent at that unit. This results in a net reduction of five units 2728 industry-wide. 2729 2730 4. Cyclohexane -- Three of the cyclohexane units 2731 included in JRB's calculation (American Petrofina at Big Spring, 2732 Texas; CORCO at Ponce, Puerto Rico; and Exxon at Baytown, Texas) -733 have been shut down. 2734 2735 5. Monochlorobenzene -- One of the monochlorobenzene 2736 units assumed by JRB (Dow at Midland, Michigan) has been shut 2737 down. 2738 2739 6. Cumene/Pheno1 -- Benzene is present in cumene 2740 units, where benzene and propylene are combined to make cumene. 2741 No benzene is present in phenol units, where cumene is reacted to 2743 2743 2720 225/ These are American Cyanamid at Willow Island, West Vir- 2721 ginia; DuPont at Gibbstown, New Jersey; and one of the Mobay 2722 units at New Martinsville, West Virginia. The DuPont unit may 2723 restart some time in the future. 256 257 158 - 79 - . MCD 000015677 2742 make phenol. Three of the cumene facilities included in JRB 2743 Table 3-3 have been shut down.226/ In addition, eleven of the 2745 facilities included in JRB's cumene/phenol listing are phenol 2746 plants that use cumene as a raw material. No benzene is present 2747 in these plants.227/ 2756 2757 7. Other -- In addition to the feedstock categories 2758 listed by JRB, two other plants use benzene and have been added 2759 to the CMA total.228/ 2761 2763 * B. Employee Exposure Profile 2764 2767 2768 2769 2770 2770 2774 1. Number of Employees Exposed to Benzene (and Person-Years of Benzene Exposure) in the Petrochemical Industry JRB estimates that 12,242 employees are exposed to 2775 benzene in the petrochemical industry.229/ This estimate 2777 2777 2743 2744 2745 226/ at El Rico. These are Monsanto at Chocolate Bayou, Texas; Chevron Segundo California; and Union Carbide at Penuelas, Puerto 2748 227/ These are Allied Chemical at Frankfort, Pennsylvania; 2749 Diamond Shamrock at Tuscaloosa, Alabama; Dow (Oyster Creek Divi- 2750 sion) at Freeport, Texas; Ferro Corporation at Sante Fe Springs, 2751 California; General Electric at Mt. Vernon, Indiana; Georgia 2752 Pacific at Plaguemine, Louisiana and Bound Brook, New Jersey; 2753 Koppers at Follansbee, West Virginia; Merichem Company at 2754 Houston, Texas; U.S. Steel at Haverhill, Ohio; and Stimson Lumber 2755 at Anacortes, Washington. 2759 228/ These are Koppers at Petrolia, Pennsylvania 2760 j[_resorcinol) and Monsanto at Anniston, Alabama (biphenyl). 2775 229/ See JRB Report at 3-16. 256 257 758 80 o*- 2776 (assumed to represent person-years of benzene exposure) is used Mil by JRB to calculate the alleged benefits (in terms of reduced 2778 leukemia mortality) that might be expected from establishing 2779 alternative pels, jrb's estimates of risk reduction benefits are 2780 seriously flawed for a variety of reasons discussed in these Com- 2781 ments. For the moment, we wish to focus only on one very simple 2782 but fundamental reason -- i.e., JRB has vastly overestimated the 2783 person-years of benzene exposure of employees in the 2784 petrochemical industry. 2785 2786 . It is not unreasonable to estimate that 12,242 2787 petrochemical workers are assigned for at least some portion of 2788 the year to positions in which there is exposure to benzene. The 2789 basic flaw in JRB's analysis (as it applies to the calculation of 790 risk reduction benefits) is the implicit assumption that each of 2791 these 12,242 employees is assigned to a post where there is 2792 exposure to benzene throughout the year. This assumption is 2793 incorrect. 2794 2795 The most logical way to estimate worker exposure to 2796 benzene in the petrochemical industry is to: 2797 2798 XI) identify the types of benzene-exposed job assign- 2799 ments that are found at various types of petrochemical units and 2800 the number of employees performing that job assignment in the 2801 plant on any one shift; 2802 2802 256 257 58 81 - >0 O' 2303 {2) where applicable, multiply the value developed in 2804 step (1) times the number of shifts needed to provide 2805 round-the-clock (plus vacation) coverage of the job assignment; 2806 2807 X3) sum the totals derived in step (2) for each job 2808 assignment at the particular type of petrochemical plant; 2809 2810 4) multiply the plantwide value developed in step (3) 2811 by the number of plants of that type in the industry; 2812 2813 ^5) sum the plant category totals to develop a total 2814 for the petrochemical industry as a whole; 2815 2816 X6) add the number of benzene-exposed workers in any 2817 special job category (i.e., on-site distribution) not included in 2818 the foregoing total. 2819 2819 2820 This calculation is presented in Table I below. 2821 2822 As can be seen from Table X, approximately 4,400 2823 employees would be required to provide full-time coverage for the 2824 various benzene-exposed work assignments in the petrochemical 2825 industry on any given day.230/ Consequently, there are approxi 2827 mately 4,400 annual aggregate person-years of benzene exposure 2828 (at varying concentrations) among employees in the petrochemical 2830 2830 2825 2826 230/ treats Indeed, this probably is an overestimate, since it vacation time as though it adds an additional shift. 256 257 258 - 82 oo'.eA 60 Table I Estimate of Number of Employees Potentially Exposed to Benzene in the Petrochemical Industry (Assuming the Employees Are Assigned To the Benzene-Exposed Positions Throughout the Year) Type of Unit Number of Process Positions Ethylene Units.. Benzene Derivative Plants........................ Number of Employees Assigned to Process Position*/ Total BenzeneExposed Employees Per Unit Number of (Process Support **/ Plus Employees-1 Support) Number of Plants Total BenzeneExposed Employees -> '/ o ... 2??.... ...........-766 -------25................. 1, I o Hydro dealkylation Units.......................... ................15................ ....25.................. ...16........... ............400 Pygas (Ifydrofining) Ur;ts.........................................2................................10........................................15.........................25........................13.........................325 B/T/X (Extraction) Units.........................................6...................................30......................................15........................45.........................35...................1,575 ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- --13--------------------- q tv 3 TOTAL (Vliere Applicable).....................................................................................................................................................-t50.....................4 m-- ***/ * * Additional Erployees Assigned to Distribution--'.......................................................................................................... GRAND TOTAL...................................................................................................................................................................................................4,4.>V V This assures that 5 employees are needed to provide round-the-clock coverage on eacr. process position plus vacations. **/ The breakdown of support enployees is assured to be as follows: laboratory =* 3 Wastewater Treatment * 4 Itointenance 8 Total Support * 15 ***/ The distribution estimate is limited to enplcyees working in the plant to hook up rail tank cars, and tank trucks. The estimate assumes two distribution vrorkers for eacn TT* benzene derivative unit -- i.e., 2x(35+48) * 166iy/*The distribution estimate does not wr 'xs employed on outside terminals or barges, which would add approximately 270 tht. cotal. tu MOD 000015681 2829 industry. This figure, 4,400 person-years of exposure (rather 2830 than the JL2,242 person-years of exposure assumed by JRB), would 2831 be the approriate value to use in any computation of potential 2832 risk reduction benefits resulting from application of alternative 2833 PELs to the petrochemical industry. Since JRB utilizes the fig 2834 ure of 12,242 in its calculation of risk reduction benefits,231/ 2836 its estimate of benefits is high by a factor of three for that 2837 reason alone. 2838 2839 2840 * 12,242 For purposes of computing costs, however, the figure employees is appropriate. Although there are approxi of 2641 mately 4,400 benzene-exposed job assignments to be staffed on a 2842 full-time basis throughout the year, many companies make a prac 2843 tice of rotating employees into and out of these benzene-exposed 1844 positions for portions of the year. (This contributes to job 2845 satisfaction and creates a pool of employees with a broad base of 2846 knowledge and experience who can be called upon to cover a vari 2847 ety of job assignments when employees regularly assigned to par 2848 ticular positions are absent from work.) As a result, 12,242 is 2849 a reasonable estimate of the total number of employees who have 2850 benzene-exposed job assignments for some non-trivial period 2851 (e.g., more than 30 days) during the course of a year. 2853 2853 2853 2853 2853 2853 2835 See JRB Report, Tables 3-20 through 3-23. 256 257 258 - 83 000V ris) 2854 Although these 12,242 employees would accumulate no 2855 more aggregate person-years of exposure than 4,400 employees 2856 assigned to these positions throughout the year, the cost of 2857 implementing the standard for 12,242 employees is much greater 2858 than the cost of implementing the standard for 4,400 employees. 2859 plus, engineering controls at these plants will cost the same 2860 regardless of whether one employee is assigned to the 2861 benzene-exposed position throughout the year or three employees 2862 rotate through that position during the course of the year. At 2863 the same time, since each of these employees would have to be 2864 provided with medical surveillance and training under the stan- 2865 dard, many of the annual costs of the standard will triple when 2866 applied to 12,242 rather than 4,400 employees. 2867 7868 In short, 4,400 is the proper figure to use when 2869 computing potential benefits of the standard, while 12,242 is an 2870 appropriate figure to use when computing costs. 2871 2875 2876 2877 2881 2. Current Exposure Levels of Petrochemical Employees Based on information provided by the American Petroleum 2882 Institute ("API"), JRB has calculated the number of petrochemical 2883 workers who allegedly are exposed to various 8-hour TWA benzene 2884 concentrations in six exposure intervals ranging from 0-.01 ppm 2885 at one extreme to 10+ ppm at the other extreme.232/ JRB's 2886 2886 2885 232/ 256 257 258 See JRB Report, Table 3-8. - 84 - MCD 00001568.3 2886 estimate shows 5,460 petrochemical workers exposed between 0 and 2887 .1 ppm; 4,064 workers exposed between .11 and .5 ppm; 1,224 2888 workers exposed between .51 and 1 ppm; 1,212 workers exposed 2889 between 1.1 and 5 ppm; 159 workers exposed between 5.1 and 10 2890 ppm; and 122 workers exposed at or above 10 ppm.233/ Several 2891 points must be made regarding JRB's estimate. 2892 2893 First, as discussed at pages _____-_____ above, the total 2894 number of employees exposed at the various concentration levels 2895 i.s greatly overstated if it is interpreted to s employees are 2896 exposed to benzene at these levels throughout the year. Rather, 2897 it reflects employees exposed to benzene for at least some por- 2898 tion of the year, possibly a matter of just a few months. 2900 2901 Second, the percentage distribution of benzene exposure ^902 levels utilized by JRB is derived from information provided by 2903 API rather than from information provided by CMA. This differ- 2903 ence is significant, since the information provided by CMA shows 2904 that the percentage of chemical industry workers exposed to an 2905 8-hour TWA greater than 1 ppra is more than twice as high as the 2907 comparable percentage provided by API.234/ 2909 2910 Third, the exposure information collected from CMA 2911 Xand' presumably, the exposure information collected from API as 2913 2913 2891 233/ _________ Id. 2907 234/ 256 257 258 See JRB Report,* Table 3-7. - 85 - MCD 000015684 2912 well) cannot be relied upon to establish fine distinctions among 2913 exposure levels below 1 ppm. The current OSHA Benzene Standard 2914 establishes a PEL of 10 ppm. Exposure measurements taken to 2915 ensure compliance with a PEL of 10 ppm need not be calibrated for 2916 a high degree of accuracy at an exposure level of 1 ppm or below. 2917 And, in fact, the experience of many CMA companies indicates that 2918 benzene monitoring methods (designed to ensure compliance with a 2919 PEL of 10 ppm) have not historically been calibrated to record 2920 accurately and reproducibly benzene concentrations of less than 1 2921 ppm in the workplace. Accordingly, we are extremely skeptical 2922 about supposed distinctions reported in exposure levels below 1 2923 ppm. 2924 2925 Fourth, JRB's exposure profiles are based on informa 2926 tion that was solicited from various associations and individual 2927 companies. Our understanding is that the data provided to OSHA 2928 and JRB reflect the prevailing or average 8-hour TWAs for the 2929 various operations as to which information was reported.235/ 2932 These average or prevailing exposure levels indicate only that 2933 benzene exposures in the identified operations fall, on average, 2934 within the exposure range reported to OSHA or JRB.236/ This does 2935 not mean that the average 8-hour TWA is never exceeded, or even 2937 2937 2930 2931 2932 235/ See, e.g., Declarations , and , Appendices ___-____. of submitted herewith as 2935 236/ See id. 256 257 258 66 - O',000' 2936 that such exceedances occur only infrequently. Rather, it means 2937 that some exposure measurements will exceed this value while 2938 others will fall below it, with the various measurements 2939 averaging out to the reported level. Thus, a report that the 2940 prevailing 8-hour TWA for 70 percent of a company's 2941 benzene-exposed workers is less than 1 ppm should not be inter- 2942 preted to mean that the company is presently complying with the 2943 equivalent of a 1 ppm PEL for those employees237/ While their 2944 average 8-hour TWA exposure may be less than 1 ppm, occasions on 2945 which the 8-hour TWA exceeds 1 ppm may not be infrequent. 2947 2948 in short, achieving a prevailing or average 8-hour TWA 2949 of 1 ppm is very different from complying with a mandatory PEL of 2950 1 ppm, since the latter implies that on whatever day an OSHA 2951 inspector may choose to take measurements, the 8-hour TWA would 2952 be found not to exceed 1 ppm. To ensure that exposures are below 2953 a specified PEL 95 percent of the time, a company probably would 2954 have to reduce average benzene exposure levels to less than 2955 one-half of the PEL, a point that OSHA itself appears to recog- 2956 nize.238/ Consequently, it is a grave misconception to assume, 2964 2964 2943 237/ See id. 2956 238/ See 50 Fed. Reg. at 50552, col. 3 ("where exposure mea- 2957 surements are above one-half of the permissible exposure iimit, 2958 . . . the employer cannot be reasonably confident that the 2959 employee may not be overexposed"). Examples of 8-hour TWA data 2960 sets averaging approximately 0.5 ppm which, under a statistical 2961 sampling program developed by Du Pont, would not meet a PEL of 1 2962 ppm 95 percent of the time are provided in Appendix _____, along 2963 with a paper describing the Du Pont statistical procedure. 256 257 258 Mco 0000; 87 2964 as JRB does, that an employer whose workers are exposed to 2965 benzene concentrations of 1 ppm or below on average would not 2966 have to implement any additional controls in order to comply with 2967 a PEL of 1 ppm. What average benzene measurements below 1 ppm 2968 can be taken to imply in many cases is that the employer has a 95 2970 percent assurance that a PEL of 2 ppm is not being exceeded.239/ 2990 2991 Because of this misconception regarding the meaning of 2992 exposure information collected by JRB, statements in the JRB 2993 Report to the effect that 85 percent of petrochemical units have 2994 already achieved 1 ppm as an 8-hour TWA exposure leve!240/ or 2995 that only 63 percent would incur additional compliance costs to 2996 achieve a PEL of 0.5 ppm241/ simply are not reliable or credible. 2997 2997 2970 239/ See NIOSH comments on a possible short-term exposure 2971 limit for ethylene oxide, 50 Fed. Reg. 64, 70 (January 2, 1985) 2972 ("As a practical matter, an employer would have to maintain con- 2974 centrations in the vicinity of 0.5 ppm in order to ensure that an 2975 1 exceedance [of 1 ppm] due to random variation1 had not 2976 occurred.11). See also Liedel, et al. (NIOSH), Expo sure 2977 Measurement, Action Level and Occupational Environmental 2978 Variability (1975). Cf. EPA Guideline Series, Control of 2979 Volatile Organic Compound Leaks from Synthetic Organic Chemical 2980 and Polymer Manufacturing Equipment (EPA-450/3-83-006, March 2981 1984) at 3-12 ("if a random annual inspection indicated that no 2982 more than two percent of valves are leaking, the probability is 2983 greater than ninety-five percent that an average of one percent 2984 of valves leaking is actually being achieved in practice'1). 2985 Depending on the nature of the work operation and the number and 2986 distribution of measurement data points, long-term average 2987 exposures may have to be maintained at a level considerably below 2988 one-half of the PEL in order to provide 95 percent assurance that 2989 the PEL will not be exceeded on a random inspection basis. 2994 240/ See JRB Report at 4-20. 2996 241/ See JRB Report at 5-32. 256 257 258 - 88 56^ ooooi 2997 They result in a gross underestimate of the number of 2998 petrochemical facilities that would have to implement additional 2999 engineering and work practice controls in an effort to comply 3000 with a PEL of 1 ppm or below. Moreover, this misconception has 3001 led JRB to reach facile conclusions regarding the feasibility of 3002 a 1 ppm standard that the exposure information, when properly 3003 understood, does not warrant. 3005 3009 3010 3011 3012 3013 3013 3017 * C. Inventory of Benzene Emission Sources at Petrochemical Facilities and Identification of Job Assignments Where Use of Respirators May Be Needed To Comply with the Proposed Standard JRB identifies four principal sources of benzene 3018 exposure at petrochemical facilities: 3019 3020 ^ Equipment leaks 3021 3022 ^ Process sampling 3023 3024 2. Wastewater collection systems 3025 3026 ^ Loading of railcars, tank trucks, and barges.242/ 3028 3029 2.n evaluating the feasibility of alternative PELs, JRB identifies 3030 what it believes to be appropriate controls for these benzene 3031 emission sources and calculates compliance costs based on the 3032 assumption that these are the only controls necessary to achieve 3034 _________ 3034 3026 242/ See JRB Report at 4-20. 256 257 258 - 89 - MCI) 00001568? 3033 the respective PELs. On the basis of this analysis, JRB con- 3034 eludes that a PEL of 1 ppm can feasibly be achieved through the 3035 use of engineering and work practice controls in the 3036 petrochemical industry, except for those employees exposed during 3037 barge loading or in maintenance operations. In both of these 3038 latter cases, JRB concludes that "respirators are required to 3039 achieve the 5.0 and 1.0 ppm alternative PEL'S . . . "243/ 3041 3042 While it is correct that respirators would be required 3043 to achieve a PEL of 1 ppm in barge loading and maintenance 3044 operations, JRB fails to identify several other important 3045 benzene emission sources or tasks which, either regularly or from 3046 tirae-to-time, would require the use of respirators to achieve a 3047 PEL of 1 ppm or a short-term exposure limit ("STEL") of 5 ppm 3048 averaged over 15 minutes.244/ Moreover, efforts to reduce 3052 benzene emissions from these additional sources through the use 3053 of engineering and work practice controls would significantly 3054 increase the costs of complying with the standard beyond the 3055 amounts estimated by JRB. These additional benzene emission 3056 sources are as follows: 3057 3057 3057 3057 3057 3039 243/ See JRB Report at 4-2. 3048 3049 3050 3051 244/ The need to utilize respirators might not be limited to the workers directly performing the specific activity but might encompass other employees who are present in the vicinity of the activity as well. 256 257 258 - 90 - MCI) 000015689 3058 1. Instrument Maintenance and/or Calibration -- Pro- 3059 cess instrument maintenance and/or calibration is perhaps the 3060 most difficult operation in which to control benzene exposures; 3061 at the same time, it is an operation which is critical to the 3062 groper functioning of the plant. Although JRB does not identify 3063 instrument maintenance and/or calibration as an important source 3064 of benzene exposure, the fact is that in most facilities this 3065 activity may potentially create the highest benzene exposures in 3066 the plant. And the activity will affect not only the instrument 3067 technician, but other employees who have to be present in the 3068 area as well. Moreover, in contrast to equipment maintenance 3069 operations, instrument maintenance must take place while the pro 3070 cess continues to run; consequently, clearing the benzene out of 3071 the equipment prior to opening for maintenance or calibration of 3072 the instrument (e.g., pH probe replacement) is not possible. 3073 3074 In some cases, use of pump-loops from reactors to 3075 instruments will allow isolation and clean-out prior to 3076 calibration or maintenance, but this approach is not always gos- 3077 sible. Where pump-loops and other specialized systems can be 3078 utilized, use of respirators may not be necessary; however, JRB 3079 did not evaluate or include costs for these specialized systems 3080 in its feasibility analysis. In those cases where such special 3081 ized systems cannot be used, respiratory protection probably 3082 would be required to comply with a PEL of 1 ppm or a STEL of 5 3083 ppm, just as it is in regular maintenance and loading 256 257 258 91 - 3084 operations.245/ Moreover, since instrument maintenance and 3090 calibration is an intermittent task which should not require hard 3091 physical labor, it is an assignment in which respirators can be 3092 used effectively and in which engineering controls generally 3093 would not achieve a significant reduction in overall employee 3094 exposure despite large implementation costs.246/ 3096 3097 2. Intraplant Transfer of Benzene -- Although JRB 3096 recognizes the need for respirator protection during interplant 3099 transfers of benzene in barge loading, the Report fails to recog* 3100 nize the exposures associated with intraplant transfers of 3101 benzene -- such as the filling of day tanks, the "pigging" of 3102 common transfer lines, and other intraplant distribution 3103 operations which create fugitive and secondary emissions. Respi- 3104 rators may well have to be worn if benzene exposures of distribu- 3105 tion operators and other employees involved in these intraplant 3106 transfer operations are expected to be kept below 1 ppm as an 3107 8-hour TWA and 5 ppm as a 15-minute STEL. Benzene emissions in 3108 these intraplant transfers can be controlled through the use of 3109 dedicated piping. However, this would be very expensive, and the 3110 costs have not been taken into account in the JRB Report. 3112 3112 3084 245/ At a PEL of 1 ppm, instrument technicians (and probably 3085 operating personnel in the immediate vicinity as well) might have 3086 to wear respirators to comply with the 8-hour TWA or with a 5 ppm 3087 STEL. At a higher PEL and STEL, respiratory protection probably 3088 would be required only for the instrument technician. 3094 246/ Cf. 50 Fed. Reg. at 50512, 50558, col. 3. 256 257 258 92 r, ^ 3113 3. Dewatering of Benzene Storage Tanks -- Drawing 3114 water from benzene storage tanks is a routine operation that 3115 potentially can result in significant fugitive and secondary 3116 emissions. Depending on the location of the tank, these emis- 3117 sions can affect benzene exposures of operating personnel as well 3118 as of workers actually performing the operation. Process 3119 requirements or environmental concerns require that the water, 3120 which is saturated with benzene, be removed from the tanks. 3121 Draining a saturated solution of benzene in water will result in 3122 emissions equivalent to those that arise when pure benzene is 3123 drained. To eliminate these emissions and the resulting 3124 exposures, interface controls and closed piping would be needed 3125 for each and every tank. Again, JRB has not factored this into 3126 its estimate of compliance costs. 3127 3128 4. Gauging of Tanks -- Another source of benzene 3129 emissions overlooked by JRB is the gauging of tanks, an operation 3130 that is necessary in order to document raw material usage and 3131 production quantities. The gauging is accomplished by inserting 3132 a calibrated measuring device into the tank through an open 3133 hatch, an operation which occurs approximately monthly in most 3134 facilities and creates exposures that might well exceed a 5 ppm 3135 STEL in the absence of respiratory protection. In many cases, 3136 there are no automatic guaging devices that can be depended on to 3137 provide reliable measurements. Even where such devices do exist, 3138 it would be very cost-ineffective to install them to control 256 257 256 - 93 O' 3139 exposures that occur only once a month, a point that OSHA itself 3140 has recognized.247/ 3146 3147 5. Process Upsets, Spills and Vent Failures -- The 3148 spills and emissions that occur during process upsets and vent 3149 failures give rise to benzene emissions which are unplanned and 3150 initially uncontrolled. While these events are nonroutine in 3151 nature, the fact is that they do occur from time to time, and 3152 respirators are likely to be required when they do occur. We 3153 understand subsections (b) and (g)(l)(v) of the proposed standard 3154 to treat these events as "emergencies11 under which use of respi- 3155 ratory protection to comply with the PEL is permitted. Such an 3156 emergency respirator use provision is critical and should clearly 3157 be included in the final standard regardless of what decision is 3158 made with respect to the PEL. 3159 3160 6. Process Venting -- Various process vessels and 3161 storage tanks at petrochemical plants must be vented from time to 3162 time. Process venting of a vessel or tank in which benzene is 3163 present contributes to benzene concentrations in the ambient air. 3164 Depending upon where the vessel is located, process venting may 3165 contribute to occupational exposure to benzene. JRB appears to 3166 have completely ignored such process emissions as sources of 3168 3168 3140 247/ See 50 Fed. Reg. 50512, 50558, col. 3, 50560, col. 3. 3142 In any event, JRB has not taken the costs of automatic guaging 3144 devices into account in computing the costs of the standard. 256 257 258 - 94 - hcd uoooitby- 3167 benzene within petrochemical facilities. The JRB Report does not 3168 evaluate whether these process vents would have to be controlled 3169 in order to comply with a 1 ppm standard and, if so, what the 3170 cost of such controls would be. 3171 3172 7. Regular Maintenance -- JRB acknowledges the diffi 3173 culty of controlling exposure to benzene by maintenance workers 3174 and concludes that respirators will be required to maintain 3175 benzene exposures of maintenance workers below a PEL of 1 ppm. 3176 However, JRB fails to recognize that, during some maintenance 3177 operations, operating personnel also may have to be provided with 3178 respirators to comply with a 1 ppm standard. While major mainte 3179 nance typically is performed on a campaign basis when the unit is 3180 shut down, routine maintenance occurs on an on-going basis and is 3181 performed while the unit is running. This routine maintenance 3182 will increase background concentrations of benzene and may 3183 require respirator use by the operating personnel as well as by 3184 the maintenance workers themselves. For example, the repacking 3185 or repair of spare equipment often involves both maintenance and 3186 operating personnel in the same location. OSHA should make clear 3187 that respirators may be used as a means of compliance for 3188 operators (as well as maintenance personnel) when the operator is 3189 exposed in a maintenance situation. 3190 3191 In sum, JRB has failed to address all of the benzene 3192 emission sources at petrochemical plants. Occupational exposures 256 257 258 95 Ibt) 3193 associated with some of these sources clearly are significant. 3194 The significance of exposures associated with others has yet to 3195 be determined. The consequences of JRB's failure to address all 3196 these sources are twofold: 3197 3198 First, it means that JRB has not identified all of the 3199 controls that would be required to achieve the emission reduc 3200 tions contemplated by the Report and, derivatively, by the pro 3201 posed standard. Since the feasibility and cost of these addi 3202 tional controls have not been evaluated, JRB's conclusions 3203 regarding technological and economic feasibility are suspect. 3204 3205 Second, JRB has failed to identify all of the work 3206 assignments and activities in which a PEL of 1 ppm and a 3207 1.5-minute STEL of 5 ppm could not be achieved through the use of 3208 engineering and work practice controls. For these activities -- 3209 most of which occur only intermittently -- respirators would be 3210 required to ensure that employees are not exposed above 1 ppm as 3211 an 8-hour TWA (and/or above 5 ppm as a 15-minute STEL); alterna 3212 tively, a PEL in excess of 1 ppm would be appropriate. 3214 3214 3218 3219 3220 3220 3221 3222 3222 3226 D. JRB Has Overestimated the Emission Reductions That Would Result from the Controls Identified in the Report. 1. Baseline Case Assumptions JRB makes the following assumptions regarding baseline 3227 controls at petrochemical facilities where benzene exposures 256 257 258 96 mcd ^'o0 3228 exceed 1 ppm as an 8-hour TWA. According to JRB, the baseline 3229 controls and practices at such plants are as follows: 3230 3232 3233 3234 3235 3236 3237 3238 3239 3240 3242 3242 3244 _ Uncontrolled process sampling points. * No pretreatment of benzene-containing wastewater. Absence of double mechanical seals on all pumps and compressors. ^ Open hatch top-loading of pure benzene from BTX units into railcars and tank trucks.248/ JRB's assumption regarding baseline controls in the 3245 petrochemical industry must be corrected in three principal 3246 respects. 3247 3248 First, JRB mistakenly assumes that open-hatch 3249 top-loading of benzene into railcars and tank trucks is a stan- 3250 dard industry practice. In fact, standard industry practice 3251 involves bottom loading and/or closed-loading systems with the 3252 use of dry disconnect. This is basically the same approach to 3253 controlling benzene emissions during railcar and tank truck load- 3254 ing that JRB recommends as a means of reducing benzene exposures 3255 during those operations.249/ Since they already are standard 3256 practice in the industry, it is unrealistic to assume that a fur- 3257 ther reduction in benzene exposures will result from implementing 3258 these controls. 3259 ____ 3259 3240 248, _ _ _________ See JRB Report at 4-21. 3255 249, See JRB Report at 4-28. 256 257 258 97 MCD 00001 5696 3260 Second, JRB is wrong in assuming that baseline grac- 3261 tices have not included leak detection programs. In fact, while 3262 formal leak detection programs as contemplated by JRB250/ may not 3263 have been standard in the industry in past years, less formal 3264 leak detection activities were widespread. In some cases, formal 3265 leak detection programs were implemented in order to comply with 3266 EPA's new source performance standards for volatile organic com 3267 pounds or as part of the control technique guidelines followed 3268 under State Implementation Plans in order to attain or maintain 3269 the ambient air quality standard for ozone. In other cases, less 3270 formal leak detection activities were carried on as part of pre 3271 ventative maintenance programs. In short, formal or informal 3272 leak detection procedures have been in place for some time at 3273 most petrochemical plants. Moreover, EPA's new benzene fugitive 3274 emissions standard, adopted under Section 112 of the Clean Air 3275 Act, formalizes leak detection requirements throughout the 3276 petrochemical industry.251/ Consequently, implementation of a 3279 leak detection program as contemplated by JRB cannot be expected 3280 to result in a significant reduction in benzene emissions. 3282 3283 Third, contrary to JRB's assumption, process sampling 3284 is widely controlled in the petrochemical industry today. JRB 3286 3286 3262 See JRB Report at 4-10. 3276 3277 3278 Benzene 49 Fed. National Emission Standard for Equipment Leaks of in the Petrochemical and Petroleum Refining Industries, Reg. 23498 (June 6', 1984). 256 257 258 98 MCI) 000015697 3285 proposes the use of ventilated enclosures to minimize benzene 3286 emissions during process sampling. In fact, most petrochemical 3237 companiesalready employ closed-system, minimum emissions sam- 3288 pling devices, which provide more effective control than the sam- 3289 pling approach proposed by JRB. Accordingly, there is no reason 3290 to believe that any further significant reduction in benzene 3291 emissions would be realized from further controls on process sam- 3292 pling. 3293 3293 3295 - 2. Control of Wastewater Emissions 3296 3298 JRB suggests that a significant reduction in emissions 3299 from benzene-containing wastewater at petrochemical plants can be 3300 achieved by installing an oil/water separator.252/ In fact, 3301 while installation of an oil/water separator is a relatively 3302 inexpensive undertaking, it is not likely to result in signifi- 3303 cant emission reductions from benzene-containing wastewater at 3304 the process unit itself. 3305 3306 The example cited by JRB apparently involved the use of 3307 an oil/water separator at the wastewater treatment site prior to 3308 bio-oxidation of the wastewater. Use of an end-of-pipe oil/water 3309 separator just prior to introducing the wastewater into the 3310 treatment plant is an appropriate and effective device to reduce 3312 3312 3300 252/ See JRB Report at 4-22. 256 257 258 99 69^ oOv6 00 3311 the insoluble benzene and oil phase prior to biological oxidation 3312 of the wastewater. And, in fact, most industrial waste treatment 3313 facilities already utilize oil/water separation prior to bio- 3314 oxidation. But addition of an oil/water separator back at the 3315 process operating unit would be redundant and, in any event, 3316 would have relatively little impact in reducing benzene emissions 3317 in the area of the operating unit. Indeed, it may actually 331S increase exposures at the operating unit by adding a new source 3319 of benzene emissions in an area where employees are present. 3320 3321 The real solution to the problem of emissions from 3322 benzene-containing wastewater collected at the process unit would 3323 be either (i) to steam strip the benzene in a closed system at 3324 the operating unit or (ii) to install a closed piping system to 3325 transfer the wastewater from the process unit to the treatment 3326 plant. In the absence of steam stripping or a closed pipe sys- 3327 tem, the potential for benzene concentrations to exceed 1 ppm 3323 near wastewater collection sites at the process unit is substan- 3329 tial. While steam stripping or a closed piping system can effec- 3330 tively reduce these emissions from benzene-containing wastewater, 3331 either of these approaches is very expensive. JRB has not taken 3332 the costs of these expensive control techniques into account in 3333 its feasibility analysis. 3334 3334 3334 3334 3334 3334 256 257 258 100 - 3336 E. Control of Fugitive Emissions 3337 3339 JRB assumes that baseline control practices at 3340 petrochemical plants which are not currently achieving a PEL of 1 3341 ppm do not include the use of double mechanical seals for pumps 3342 and compressors or the implementation of a leak detection pro* 3343 gram. In JRB's view, use of double mechanical seals on pumps and 3344 compressors and adoption of a monthly leak detection program 3345 would result in a substantial reduction in benzene fugitive emis- 3346 sions, making achievement of 1 ppm as a PEL feasible. 3347 3348 JRB's optimistic projection of the emission reductions 3349 achieveable through the use of double mechanical seals and 3350 monthly monitoring in the petrochemical industry is unfounded. 3351 In the first place, as JRB recognizes, the only equipment with a 3352 significant percentage of leaks at petrochemical plants are pump 3353 and compressor seals.253/ This contrasts markedly with the situ* 3354 ation at petroleum refineries, where a wider range of equipment 3355 displays a significant leak percentage.254/ Moreover, even in 3357 the case of pump and compressor seals, the leak percentage at 3358 petrochemical plants is much lower than the comparable leak per- 3359 centages at petroleum refineries.255/ This reflects the good 3361 maintenance practices that are followed at petrochemical plants. 3362 3362 3353 253/ See JRB Report at 4-24 and Table 4-2. 3356 254/ See JRB Report at Table 4-1. 3359 255/ Compare JRB Report, Table 4-1, with id.. Table 4-2. 256 257 258 101 b7 lOOO1 3363 The use of double mechanical seals on pumps and com- 3364 pressors in petrochemical plants is not likely to achieve a sub- 3365 stantial reduction in emissions even on those pumps and com- 3366 pressors that do leak. In the first place, there are very few 3367 compressors handling gases with significant benzene concentra- 3368 tions in the petrochemical industry. A typical ethylene plant 3369 will have a single compressor. The same is true of toluene 3370 dealkylation and pyrolysis gasoline hydrogenation units. Other 3371 types of petrochemical plants typically have no compressors. 3372 Consequently, further control of compressor seals is unlikely to 3373 result in significant reductions in benzene emissions at 3374 petrochemical plants. 3375 3376 As discussed above, the percentage of leaking pumps and 3377 compressors at petrochemical plants is relatively low, and very 3378 few compressors in benzene service are found in the petrochemical 3379 industry. For those reasons alone, the potential to reduce 3380 benzene fugitive emissions through the use of double mechanical 3381 seals on pumps and compressors is quite limited. It also is 3382 limited by the fact that emission rates from pumps and com- 3383 pressors in the petrochemical industry are relatively low under 3384 existing practices. 3385 3386 JRB does not present data on emission rates from pump 3387 and compressor seals at petrochemical plants, although the Report 3388 presumably relies on data derived from tests sponsored by 256 257 258 - 102 - MCD 000015701 3389 EPA.256/ Additional information available from the general 3394 literature also has been compiled by EPA.257/ An examination of 3399 these sources suggests that the use of double mechanical seals is 3400 not likely to result in as great an emission reduction as JRB 3401 apparently assumes. 3402 3403 The available information indicates that, where good 3404 maintenance practices are followed, emission rates from leaking 3405 pumps and compressors in the petrochemical industry are low even 3406 when double mechanical seals are not used. Thus, use of a single 3407 mechanical seal in light liquid service is estimated to result in 3408 volatile organic compound leakage at a rate of only 6 grams per 3409 hour when good maintenance practices are followed and process 3410 fluid is used to flush the seal.258/ Where a water flush is 3414 used, the emission rate from a single mechanical seal is 3415 estimated to be only 0.02 grams per hour.259/ Similarly, when 3416 3416 3389 256/ See Office of Air Quality Planning and Standards, U.S. 3390 Environmental Protection Agency, "Fugitive Emission Sources of 3391 Organic Compounds -- Additional Information on Emissions, Emis3392 sion Reductions, and Costs," Document No. EPA-450/3-82-010 April 3393 1982). 3394 257/ See Office of Air Quality Planning and Standards, U.S. 3395 Environmental Protection Agency, "Guideline Series -- Control of 3396 Volatile Organic Compound Leaks from Synthetic Organic Chemical 3397 and Polymer Manufacturing Equipment," Document No. 3398 EPA-450/3-83-006 (March 1984). 3410 258/ J. Schroy, "Prediction of Workplace Contaminant 3411 Levels," in NIOSH Symposium Proceedings, Control Technology in 3412 the Plastics and Resins Industry, U.S. Department of Health and 3413 Human Services (NIOSH) Publication No. 81107 (January 1981), pp. 3414 190-206. 3416 259/ Id. 256 257 258 103 MCD 000015702 3416 rod packed reciprocating seals are used on gas compessors and 3417 good maintenance practices are followed, volatile organic com3418 pound emission rates are estimated to be 16 grams per hour in the 3419 case of single rod packed seals and 13 grams per hour in the case 3420 of double rod packed seals260/ 3422 3423 As indicated by the low leak percentage rates for 3424 pumps, compressors and related equipment in the petrochemical 3425 industry, petrochemical plants do follow good maintenance prac3426 tices with respect to these potential sources of fugitive emis3427 sions. Accordingly, the low emission rates associated with the 3428 use of good maintenance practices are what one would expect to 3429 find in the petrochemical industry. For that reason, installa3430 tion of double mechanical seals would not be expected to result 3431 in any significant reduction in benzene fugitive emissions at 3432 petrochemical plants. 3433 3434 The reduction in benzene emissions resulting from a 3435 monthly leak detection program also is likely to be much smaller 3436 than JRB assumes. For one thing, as indicated above, fugitive 3437 emissions from pumps and compressors in the petrochemical indus3438 try are much lower than JRB assumes. If double mechanical seals 3439 were installed as JRB suggests, the volume of fugitive emissions 3440 presumably would decline somewhat further. In either case, the 3442 3442 3421 260/ 256 257 258 Id. - 104 - 000015703 HOD 3441 volume of fugitive emissions that could potentially be reduced as 3442 a result of monthly inspections would not be substantial. 3444 3445 Second, JRB's assumption that a monthly leak detection 3446 program would reduce "leakage from valves and pumps in light liq 3447 uid service by 77 and 80 percent, respectively" is based on test 3443 data for petroleum refineries. The percentage reduction at 3449 petrochemical plants is likely to be much smaller. This is par 3450 ticularly true, since the percentage of leaking valves and pumps 3451 iQ light liquid service at organic chemical plants is much lower 3452 than the comparable percentages at petroleum refineries.261/ 3454 pius, even if the reduction percentages were the same, the abso 3455 lute emission reduction would be smaller. 3456 3457 Finally, as indicated above, leak detection activities 3458 Xof a greater or lesser degree of formality) have been standard 3459 practice for some time in the petrochemical industry and are now 3460 formalized under EPA's National Emissions Standard for Equipment 3461 Leaks of Benzene.262/ The leak detection program proposed by JRB 3462 would represent little or no improvement over current practice 3463 and is very unlikely to result in a significant reduction in 3464 benzene exposures at the plants. 3465 3465 3465 3465 3465 3453 261/ Compare JRB Report, Table 4-1, with id., Table 4-2. 3461 262/ See pp. ___- ' , supra. 256 257 258 105 ,lCA 0 00 V* 3467 ***** 3468 3470 In short, because of (i) a misunderstanding of baseline 3471 controls and practices and (ii) unrealistic assumptions regarding 3472 current emission levels and the effectiveness of controls, JRB 3473 has greatly overestimated the potential for achieving further 3474 reductions in benzene emissions and exposure levels in the 3475 petrochemical industry. For this reason alone, the feasibility 3476 of achieving 1 ppm as a PEL at petrochemical plants is much more 3477 problematical than JRB suggests. 3478 3482 3483 3484 3485 3485 3489 F. JRB's Conclusion That a 1 ppm PEL Is Technologically Feasible in the Petrochemical Industry Is Unjustified. As indicated above, JRB concludes that, with the 3490 exception of barge loading and maintenance operations (where use 3491 of respirators would be required)/ a PEL of 1 ppm as an 8-hour 3492 TWA and a 15-minute STEL of 5 ppm would be technologically feasi- 3493 ble in the petrochemical industry. In the preceding sections of 3494 these Comments, we showed that this conclusion is unsupported for 3495 several reasons, including the following: 3496 3498 1. JRB's assumption regarding the extent to which 3499 petrochemical operations are presently meeting a PEL of 1 ppm 3500 reflects a misunderstanding of the measurement data provided by 3501 members of the industry.263/ 3502 3502 3501 263/ 256 257 258 See pp. , supra. 106 3503 2. JRB has failed to identify several important 3504 sources of benzene in the petrochemical industry, including a 3505 variety of activities in which engineering controls are not fea- 3506 sible and others in which engineering controls have not been 3507 identified or costed by JRB.264/ A variety of task-related 3503 exposures other than just barge loading and maintenance 3509 operations would require the use of respirators to ensure that a 3510 PEL of 1 ppm or a STEL of 5 ppm is not exceeded. 3511 3512 ,, 3. JRB has vastly overstated the extent to which 3513 benzene emissions can be reduced through the implementation of 3514 designated engineering controls and work practices in the case of 3515 railcar and tank truck loading, process sampling, wastewater col- 3516 lection, and fugitive emissions.265/ At the same time, JRB has 3517 failed to take account of the very large expenses that would have 3518 to be incurred in reducing benzene emissions from wastewater col- 3519 lection through the use of steam stripping or closed pipe sys- 3520 terns. 3521 3522 In short, current benzene exposures in the 3523 petrochemical industry are higher than JRB appears to assume, 3524 while the potential for reducing these exposures is much smaller 3525 than JRB assumes. The fact is that, even if all of the 3527 3527 3507 264/ See pp. - , supra. 3516 265/ See pp. - , supra. 256 257 258 107 ,lb706 3526 engineering controls and work practices identified by JRB were 3527 implemented, it is unlikely that most petrochemical plants would 3528 be able to assure compliance (at a 95 percent level of confi- 3529 dence) with a PEL of 1 ppra and a 15-minute STEL of 5 ppm without 3530 using respirators on a variety of job assignments that extend 3531 beyond barge loading and maintenance.266/ 3543 3547 3548 3549 3550 3550 3554 ,, G. JRB Has Greatly Underestimated the Costs of Complying Wtih a Standard Having a PEL of 1 ppm as an 8-Hour TWA.____________ JRB's approach to estimating compliance costs with 3555 alternative PELs is basically as follows; 3556 3557 1) A baseline of current industry practice was 3558 established on the basis of exposure data for 27 petrochemical 3559 process units, representing approximately 15 percent of total 3560 process units in the industry. 3561 3562 X2) 0n the basis of exposure data collected from these 3563 27 units, JRB projected that only 22.2 percent of petrochemical 3565 3565 3531 3532 3533 266/ In this connection, we would note that in setting a 40-hour time-weighted average benzene exposure limit of 5 ppm (with a goal of 1 ppm), the Ontario Ministry of Labour concluded 3534 that reducing the 40-hour time-weighted average benzene exposure 3536 of workers below 1 ppm "would be exceedingly difficult for the 3537 coke oven by-products plants and petrochemical companies." 3538 Ontario Ministry of Labour, Summary of Information and Data 3539 Gathered by the Ministry of Labour After the October 11, 1983 3540 Public Meeting on the Proposed Benzene Regulation (October 1984) 3541 at 28. A copy of the Summary is submitted herewith as Appendix _ 3542 . 256 257 258 - 108 - MCD 15707 3564 facilities would "incur compliance costs to achieve the 1 ppra 3565 8-hour TWA exposure level."267/ The value of 22.2 percent was 3566 then multiplied by the number of petrochemical units of each pro 3567 duction type to derive an estimate of the number of units of each 3568 type that would incur compliance costs to achieve a PEL of 1 ppm. 3570 3571 3) The various kinds of petrochemical plants were 3572 grouped into three Model categories representing large, medium 3573 and small production units.268/ 3574 3575 * 4) The values derived in steps (2)and (3) were then 3576 combined to predict how many process units in each Model category 3577 would require engineering and work practice controls (and thus 3578 ncur compliance costs) to achieve a PEL of 1 ppm.269/ 3580 ?581 5) Costs to achieve a PEL of 1 ppm at each Model unit 3582 were then developed and multiplied times the number of units 3583 affected in each Model category. 3584 3585 6) The aggregate costs for each of the three Model 3586 categories were then summed to produce an estimate of aggregate 3587 industry-wide compliance costs. 3588 3588 3588 3588 3588 3565 267/ JRB Report at 5-33. 3573 268/ See JRB Report at 15-33. 3578 269/ See JRB Report at 5-35 and 5-36. 256 257 258 - 109 - 0'.o.0V .uS> ;o 3589 In performing the foregoing calculations, JRB assumed 3590 that no engineering controls would be used to reduce exposures in 3591 barge loading.270/ Similarly, no cost estimates were developed 3592 for reducing exposures at railcar goading facilities, because JRB 3593 was unable to estimate the number of railcar loading facilities 3594 at BTX units.271/ 3595 3596 For a variety of reasons, JRB's estimate of compliance 3597 costs in the petrochemical industry is unfounded and vastly 3598 understates the true costs that would be involved in attempting 3599 to comply with a PEL of 1 ppm. 3600 3601 For one thing, JRB's assumption that only 22.2 percent 3602 of petrochemical plants would incur any compliance costs to 3603 achieve a PEL of 1 ppm is unrealistic. It is based on a sampling 3604 of only 15 percent of the industry, a sampling which may very 3605 well not be representative of the industry as a whole.272/ More- 3611 over, two of the plants included in the survey are aniline units 3612 at which almost no benzene is present,273/ and there are only 3614 3614 3591 270/ See JRB Report at 5-39. 3594 271/ See id. 3605 272/ In addition, there is no indication of how the particu- 3606 lar plants were selected. We do not know what percentage of all 3608 questionnaires sent out by JRB was returned. Moreover, it may be 3609 that plants which returned questionnaires are more likely to have 3610 ^ower benzene exposure levels than plants which did not respond. 3612 273/ The only benzene to be found at an aniline unit occurs 3613 in a small stream that is approximately 5 percent benzene. 256 257 258 - 110 - MCD 000015709 3614 three operating aniline units in the petrochemical industry.274/ 3616 At the same time, JRB's sample included only two ethylene units, 3617 even though it estimates that there are 46 ethylene units 3618 industrywide.275/ This does not appear to be an appropriate mix 3619 of units upon which to base industry-wide exposure level 3620 estimates. 3621 3622 Quite apart from the question of the representativeness 3623 of the sample is the question of how to interpret the exposure 3624 data provided to JRB. As indicated at pp. ___ - ___ above, what 3625 JRB received was information regarding the prevailing or average 3626 6-hour TWA ben2ene exposures in the job assignments or operations 3627 for which monitoring results were reported. The fact that the 3628 8-hour TWA at a particular operation is slightly below 1 ppm on 3629 the average is very different from establishing that the 3630 operation already is in compliance with a PEL of 1 ppm which can- 3631 not be exceeded on a random inspection basis. Furthermore, as 3632 discussed at page ___ above, historical measurements of benzene 3633 exposures at or below 1 ppm are of questionable precision and 3634 accuracy. Therefore, JRB's exposure profile at concentration 3635 levels below 1 ppm is not reliable. 3636 3636 3636 3636 3636 3636 3615 274/ See p. _ A supra. 3618 275/ 256 257 258 JRB Report, Table 5-8. 111 mod 000015710 3637 In addition to underestimating the number of 3638 petrochemical facilities that would incur compliance costs to 3639 achieve a PEL of 1 ppm, JRB, as discussed above, did not identify 3640 all of the benzene emission sources in the plants that would have 3641 to be controlled; nor did JRB identify all of the necessary con 3642 trols for the sources listed in the Report. For example, con 3643 trary to JRB's assumption, installation of an oil/water separa 3644 tor is not likely to be effective in reducing benzene exposures 3645 from wastewater collection streams at the process units.276/ VI 3646 Instead, expensive steam stripping or closed pipe systems would 3647 have to be installed; yet JRB makes no allowance for these costs 3648 in its analysis. Nor has JRB estimated any costs for automatic 3650 guaging devices, or for interface controls and closed piping in 3651 the dewatering of benzene storage tanks, or for dedicated piping 3652 in intraplant transfers of benzene. Yet, unless respirators are 3653 to be permitted in these tasks, these expensive engineering con 3654 trols may be required. 3655 3656 For the foregoing reasons, the JRB Report cannot be 3657 taken as presenting a realistic and supportable analysis of the 3658 costs of complying with a 1 ppm 8-hour TWA standard having a 3659 15-minute STEL of 5 ppm. Needless to say, JRB's estimate of the 3660 cost of complying with a 0.5 ppm standard is even less credi 3661 ble.277/ 3681 3661 3646 276/ See pp. ____ - ____, supra. 3661 "562 i81 5 256 257 277/ Meridian We would Research, note that in an Inc., developed 0000157 3-1 MCD Addendum to the JRB Report, a reduced cost estimate for [Footnote continued next page] 258 112 oo01 3685 3686 3687 3688 3689 3694 G. Feasibility Considerations Dictate Setting the PEL Above 1 ppm, or at a Minimum, Establishing Compliance Criteria Which Account for Exposure Variability. In the preceding sections, we have shown that the 3695 feasibility of achieving a 1 ppm 8-hour exposure limit and a 5 3696 ppm STEL on a never-to-be-exceeded basis in the petrochemical i p * - ^ ^ 3697 industry is highly questionable and +**** f1 -i 3698 pounded fnrfViei1 1 ij~ ^ n 1 ill i il i r y in measurements of exposure. We 3701 also have shown that the health risk, if any, at exposure levels 3702 in the neighborhood of .1 ppm is nonexistent or negligible.278/ 3704 In these circumstances, the best available evidence does not sup- 3705 port establishment of a 1 ppm standard which would be deemed vio- 3706 lated whenever the PEL is exceeded. / 3708 ^ 3708 3708 3708 5 [Footnote continued from preceding page] 5 3663 complying with a 1 ppm benzene standard in the petrochemical 3664 industry, based upon the assumption that the industry will incur 3665 some of the costs estimated by JRB in order to comply with EFA's 3666 National Emission Standard for Equipment Leaks of Benzene in the 3667 Petrochemical and Petroleum Refining Industries. 49 Fed. Reg. 3668 23498 (June 6, 1984). See Meridian Research, Inc., "Addendum to 3669 Technological Feasibility and Economic Impact Study of Alterna3670 tive Standards for Benzene," July 16, 1984. As noted above, the 3671 JRB Report did not provide a realistic and supportable estimate 3672 of the costs of complying with a 1 ppm benzene standard; accord3673 ingly, the revised estimate prepared by Meridian Research, which 3674 takes the JRB estimate as a starting point, is subject to the 3675 same criticisms as the JRB Report. Moreover, in evaluating the 3676 economic feasibility of an OSHA standard, it is not appropriate 3677 to ignore the cumulative impact of the costs of complying with 3678 standards issued by other agencies as well as by OSHA itself. 3679 Cf. ASARCO, Inc, v. OSHA, _______ F.2d _____, ______ (9th Cir. 1984). 3703 278/ See pp. ______ - ______. supra. 256 257 258 113 MCD 00l5?13 3709 One way of dealing with this situation would be to 3710 establish a higher PEL than OSHA has proposed -- e.g. , a PEL of 2 3711 ppm. Such a standard would respond to the difficulties of reduc- 3712 ing benzene exposures below 1 ppm through the use of engineering 3713 and work practice controls for all job assignments in the 3714 petrochemical industry and would take account of the inherent 3715 variability of exposures. In these respects, it would be consis- 3716 tent with the action taken by the Ontario Ministry of Labour, 3717 which recently established a 5 ppm 40-hour time-weighted average e 3718 exposure limit for benzene.279/ Moreover, as discussed 3719 above,280/ to comply with a 2 ppm standard, employers would have - Z' 3720 to maintain average exposures below 1 ppm, thus providing all of 3721 the health protective benefits that OSHA has calculated for a 3722 1 ppm standard. 3724 3725 At the very least, OSHA should recognize the inherent 3726 variability in measurements of benzene exposures281/ by providing 3728 some means for averaging benzene exposure measurements in order 3728 to determine whether a violation of the standard exists. OSHA 3730 has suggested that this might be accomplished by allowing an 3731 employer to rebut a presumptive exceedence of the PEL by showing 3733 that an average of at least five 8-hour time-weighted 3735 3735 371.8 279, _________ See p. _____ - ______, supra. 3719 280, 3727 281, See p. _____, supra. See pp. _____ - ______ & n., supra. 256 357 258 - 114 - MCD 000015714 3734 measurements taken in the same area (or for the same job assign 3735 ment) within a "reasonable time" of the apparent exceedence is 3736 below the PEL. If the employer presents such information, a 3737 citation for violating the standard would not be issued unless 3739 remonitoring by the OSHA inspector confirms the initial measure 3740 ment in excess of the PEL.282/ 3742 3743 CMA strongly endorses this averaging approach to de 3744 termining compliance or non-compliance and suggests that one year 3745 [J.8 months?] would be a "reasonable time" within which the rebut 3746 ting measurements should have to have been taken. 3750 3751 3752 3756 3756 3757 IV. There Is No Basis for Adopting a Short-Term Exposure Limit in the Benzene Standard. The proposed standard does not contain a short-term 3758 exposure limit ("STEL"). However, the rulemaking notice refers 3759 to a possible STEL of 5 ppm averaged over a 15-minute period283/ 3760 and requests comment on whether a STEL is needed for benzene.284/ 3762 For the reasons discussed below, we believe a STEL is not needed 3763 and should not be included in the Benzene Standard. 3768 3768 3768 3768 3768 3768 3768 3740 282/ See 50 Fed. Reg. 50512, 50515, cols. 1-2. 3760 283/ See, e.g., id. at 50555, col. 3. 3762 See id. at 50554, col. 1. 256 257 258 - 115 - 0000^ r 3769 3770 3774 3774 3775 A. Principles To Be Used in Determining Whether a STEL Is Heeded Requirements in an occupational health standard must be 3776 "reasonably necessary or appropriate to provide safe or healthful 3777 employment and places of employment."285/ This is as true for a 3781 STEL as for the 8-hour PEL itself. Thus, as a matter of law, a 3782 STEL may not be adopted unless it will serve a demonstrated 3783 health-protective function. 3784 3785 ,, As a matter of science and public health policy, a STEL 3786 is justified to supplement an 8-hour time weighted average PEL in 3788 only two sets of circumstances -- 3789 3791 3792 -*793 ^794 3796 3797 3798 3799 3799 3800 3801 3802 3803 3804 3805 3808 3808 3777 3778 3779 3780 XI) Where there are recognized acute health effects associated with short-term exposures of the type that might be expected to occur even if compliance with the 8-hour PEL is achieved;286/ 12) Where there is a demonstrated "dose-rate effect" for the chronic health effect of concern i, e. , where short-term exposures to the substance pres ent a chronic health risk above and beyond their contribution to cumulative exposures. As explained by the Environmental Mutagen Society, a STEL is appropriate where there is "a greater yield of damage X*ffct) from an acute treatment as compared to a chronic or fractionated treatment for the same total dose."287/ 28S/ Section 3(8) of the Occupational Safety and Health Act of 1970, 29 U.S.C. 652(8). See also Industrial Union Department. AFL-CIO v. American Petroleum Institute, 448 U.S. 607, 639-646 (1980). 3794 286/ (Cite to ACGIH Booklet.] 3805 3806 3807 287/ Statement in 50 Fed. Reg. 64, original). of the Environmental Mutagen 66, col. 3 (January 2, 1985) Society, quoted (emphasis in ^56 257 258 - 116 000 V0) 3810 OSHA itself has clearly recognized and endorsed these 3811 principles, stating that a demonstrated dose-rate effect is a 3312 ^critical" finding "to justify the adoption of . . . [a] STEL 3813 "288/ In this regard, OSHA has explained that "it is nec 3818 essary to have data that compares the biological outcomes that 3819 result from . . . two exposure scenarios before a conclusion can 3820 be reached that a dose-rate effect exists."289/ Under the two 3821 exposure scenarios, 3823 3826 3827 3828 3828 3829 3830 3831 3831 3834 . health effects observed in a test group receiving a given total dose over a con tinuous period of time must be compared with the health effects observed in a separate test group receiving the same total dose over a shorter period.290/ No such set of exposure scenarios exists in the case of 835 benzene. Nor, as discussed below, does the evidence which exists 3836 under other exposure scenarios justify the finding of a dose-rate 3837 effect for benzene. 3841 3842 B. The Best Available Evidence Does Not Support 3843 a Finding That Short-Term Benzene Exposures 3844 Consistent With the Proposed 8-Hour PEL Will 3845 Present a Significant Risk of Material Health 3846 Impairment. 3847 3847 _____________ 3847 3813 288/ Id. at 73, col 3. See also id. at 73, col. 3 (an 3814 observed dose-rate effect over a certain dose range would have to 3815 be demonstrated "[bjefore a STEL can be justified based on health 3816 effects from short-term exposures"). 3821 289/ Id at 75, col. 1. 3830 290/ Id at 74, col. 1. 256 257 >58 - 117 - 'X> 00 00 V? 3848 3852 3852 3853 1. Non-Maliqnant Health Effects. As discussed at pages _____ - _____ above, the non- 3854 malignant health effects of benzene appear to have a chronic 3855 exposure threshold of approximately 40-50 ppm. Exposure levels 3856 would have to be considerably higher than that in order for non- 3857 malignant effects to result from short-term exposure. Since the 3858 proposed 8-hour PEL effectively precludes ^hort-term exposures in 3859 excess of the threshold for non-malignant health effects,291/ a 3861 STEL cannot be justified as necessary to protect against risks of 3862 acute benzene toxicity. 3865 3866 3867 3870 3870 3871 2. A Dose-Rate Effect for Benzene-Related Leukemia Has Not Seen Demonstrated. As indicated above, none of the existing data sets 3872 even approach the two exposure scenarios that OSHA has described 3873 as necessary in order to reach a conclusion that a dose-rate 3874 effect exists for benzene-related leukemia.292/ Even apart from 3875 the absence of the required exposure scenarios, the fact is that 3876 the epidemiological and animal data that do exist, as well as the 3877 available biological and pharmacokinetic information, do not pro3878 vide a basis for concluding that benzene-related leukemia is 3879 dose-rate dependent. 3881 3881 3881 3860 291/ _________ See pp. ___-____, infra. See EP- supra. 0000l&7ia 256 257 258 - 118 - 3882 For the most part, the human epidemiological studies do 3883 not provide sufficient information to develop short-term exposure 3884 profiles, separate and apart from cumulative exposure estimates, 3885 for members of the cohorts that were studied. Thus, the risk 3886 assessments that have been performed all relate increased risk to 3887 total cumulative exposure (or average exposure for a specified 3888 period). 3889 3890 The limited information that is available regarding the 3891 effect of intermittent, short-term exposures is either inconclu- 3892 sive or inconsistent with a postulated dose-rate effect. Thus, 3893 in the Rinsky Study, no statistically significant excess of 3894 leukemia cases was found among numbers of the cohort who had less 3895 than five years of employment with benzene exposure.293/ As Drs. 3897 Crump and Allen observed after analyzing the Rinsky data for 3898 workers having varying degrees of cumulative and peak exposures: 3901 3903 3904 3904 3905 3906 3906 3907 3907 3908 3911 3911 3911 3911 3911 3911 3896 This analysis does not support the hypothesis that peak exposure has any [impact] upon risk over that which can be explained by the contribution of these exposures to cumulative dose. If anything, it suggests that high exposures are less effective per ppm-year in producing leukemia than Lower exposures.294/ See 50 Fed. Reg. 50512, S0519, col. 1 3908 294/ 256 257 258 Crump Report at 21. - 119 CnO 0 CV Q 3923 To tlie extent It indicates anything about the 3914 relationship between benzene and leukemia, the Wong Study sup 3915 ports a similar conclusion. Thus, as stated by Dr. Wong: "The 3916 findings in this study suggested that cumulative exposure 3917 (ppm-months), and not peak exposure, was the major parameter in 3918 quantifying mortality risk from lymphopoietic cancer."295/ OSHA 3919 has interpreted the Wong Study in the same way, stating: 3923 3924 3925 3925 3926 3927 3928 3931 3931 3932 ,, No significant peak exposure response relationship was observed. These findings suggest that a cumulative dose concept may be better than a maximum peak exposure concept when trying to determine dose-response rela tionships . 296/ plus, the epidemiological data do not show a dose-rate 3933 effect for benzene. Nor do the data from animal studies. OSHA 935 points to a study by _Irons (Ex. 159-41A) as suggesting that 3936 intermittent exposures may be more potent in producing certain 3937 bone marrow effects plan continuous exposure.297/ But, apart 3939 from the fact that these bone marrow effects are different from 3940 leukemia, the fact is that the intermittent exposures in the 3941 Irons study were at the identical level of benzene as the 3942 continuous exposures.298/ Thus, .Irons* study does not 3946 3946 3919 Wong Study at 61. 3928 3938 50 Fed. Reg. 50512, 50523, col. 2. Id. at 50554, cols. 1-2. 3942 3943 3944 See Preliminary Regulatory Flexibility Analysis for the Benzene _____) at I11-15. Impact and Regulatory Standard, December 1985 (Ex. 256 257 256 120 MOD 000015720 3946 demonstrate a dose-rate effect in which higher short-term 3947 exposures have a greater health effect impact than lower long 3948 term exposures amounting to the same cumulative dose.299/ 3953 3954 The studies of cytogenetic effects in animals to which 3955 OSHA refers300/ also fail to demonstrate a dose-rate effect for 3957 benzene-related leukemia. The cytogenetic effects involved in 3958 those tests have no known clinical significance or demonstrated 3958 causal relation to benzene-related leukemia.301/ Furthermore, 3960 there is no basis for concluding that the observed effects relate 3961 to short-term peaks rather than to overall cumulative exposure. 3962 Since the dose was administered at a constant rate, the latter 3963 hypothesis is at least as likely as the former. 3965 3966 In sum, neither the epidemiological data nor the animal 3967 studies demonstrate a dose-rate effect for benzene-related 3968 leukemia. Nor does the biological or pharmacokinetic information 3969 provide a basis for presuming that a dose-rate effect exists, 3970 for, as OSHA acknowledges: "The basic mechanism by which benzene 3971 affects bone marrow precursor cells is still unclear."302/ pius. 3974 3974 3949 3950 3951 299/ workers them to What the Irons data might suggest is that exposing to benzene continuously is more protective than exposing the same level of benzene from time to time. 3955 300/ See 50 Fed. Reg. 50512, 50554, col. 3. 3959 301/ See pp. - , supra. 3972 302/ 50 Fed. Reg. 50512, 50516, col. 3. 256 257 256 121 - 3973 there is no basis for finding dose-rate dependency in the case of 3974 benzene-related leukemia or for adopting a STEL to protect 3975 against dose-rate effects, 3979 3980 3981 3982 3986 3986 3987 C. The Proposed Standard Would Protect Against High Short-Term Exposures Even Without the Adoption of a STEL. Even without a STEL, the proposed standard would pro- 3988 tect against high short-term benzene exposures in a variety of 3989 ways. 3990 3991 First, and most fundamentally, the proposed 8-hour PEL 3992 of 1 ppm, as OSHA recognizes, automatically establishes a maximum 3993 15-minute exposure limit of 32 ppm even if no other benzene 3994 exposure occurs during the course of the day.303/ In practice, 3996 the proposed 8-hour PEL would establish a much lower 15-minute 3997 limit, since workers would be exposed to background levels of 3998 benzene during all or substantial periods of the workday. For 3999 example, if exposure for the rest of the shift is at the action 4000 level of 0.5 ppm, the maximum 15-minute exposure would be 16 ppm. 4002 4003 Furthermore, where several short-term exposure 4004 excursions occur during the day, additional limits automatically 4005 are placed on the maximum level of each excursion. Thus, as 4006 OSHA points out, where an employee is exposed to three 15-minute 4008 4008 3994 303/ See 50 Fed. Reg. 64, 76, col. 3. 256 257 258 - 122 Mco 00j5 ?22 4007 excursions per day, the maximum exposure at each excursion could 4008 be no more than approximately 10 ppm in order to comply with an 4009 8-hour PEL of ,1 ppm.304/ Since background levels of benzene will 4010 exist in combination with multiple short-term excursions in most 4012 cases, the maximum 15-minute exposure level generally would have 4013 to be kept in the range of 5-10 ppm in order to assure compliance 4014 with an 8-hour PEL of 1 ppm. 4016 4017 Finally, it is important to bear in mind that employers 4017 wpuld have to maintain average 8-hour exposures at or below 0.5 4018 ppm in order to have a high degree of assurance that the PEL of 4019 1 ppm is not exceeded.305/ Accordingly, 15-minute exposures 4021 would have to be held to even lower levels than the foregoing 4022 discussion would suggest. In all of the respects discussed 1023 above, the 8-hour PEL, as OSHA has pointed out elsewhere, places 4024 "internal limitations on the levels and durations of short-term 4025 exposures" and acts "as a check on the number and extent of 4026 short-term exposures during the day."306/ 4028 4029 In addition to the inherent mathematical constraints 4029 imposed by the 8-hour PEL, OSHA and its contractor, JRB Associ 4032 ates, have preliminarily concluded that the same engineering 4034 4034 4010 304/ See id. 4020 305/ See pp. - & n. supra. 4027 306/ See 50 Fed. Reg. 64, 73, col. 2, 76, col. 3. 256 257 258 123 0 S> 4033 controls and work practices that will be used to achieve the 4034 8-hour PEL of 1 ppm would reduce ^5-minute short-term exposures 4035 as well.307/ Moreover, short-term exposure excursions are most 4039 likely to occur in those activities -- such as maintenance and 4040 repair, vessel cleaning, and other operations in which benzene 4041 exposures are intermittent in nature and limited in duration -- 4042 where respirators would be used to comply with the 8-hour PEL in 4043 any event. As discussed above, it is precisely in these situa4045 tions that the feasibility of complying with a 5 ppm STEL would 4046 be most questionable.308/ Thus, in combination, the engineering 4047 and work practice controls and respirator requirements of the 4048 standard will effectively establish an infoiraal STEL under the 4049 Benzene Standard. 4050 1051 The proposed standard also establishes requirements for 4052 training and providing information to employees regarding health 4053 risks of benzene and methods of protecting against those risks. 4054 This training, combined with the sign and labeling requirements 4055 and the regulated area provisions of the standard, will further 4056 minimize the risk that employees may be exposed to unacceptably 4058 high short-term concentrations of benzene without wearing appro- 4060 priate respiratory protection. 4061 4061 4035 307/ See Preliminary Regulatory Impact and Regulatory 4036 Flexibility Analysis for the Benzene Standard, December 1985 4037 ___) at IV-7; 50 Fed. Reg. 50512, 50543, col. 1. (Ex. 4046 308/ See pp. ___-____, supra. 256 257 258 124 - A. ;V` oV 0,-aO > 4062 In short, the internal limitations imposed by the 4063 8-hour PEL itself, along with a variety of requirements for 4064 methods of compliance, respirator usage and employee training, 4065 will assure that 15-minute benzene exposures will be limited to 4066 jLevels that do not pose any significant risk of material health 4067 impairment to workers. 4068 4070 ***** 4071 4072 in announcing the proposed Benzene Standard, Acting 4073 Assistant Secretary of Labor, Patrick R. Tyson, stated that 4074 "right now, we don't have the scientific evidence before us to 4075 justify a STEL" for benzene.309/ Similarly, the American Confer 4078 ence of Governmental Industrial Hygienists iMACGIH") recently 4080 proposed to remove the STEL for benzene on the ground that 4081 ...........................................310/ We concur with the judgment that a STEL for 4087 benzene is not warranted. Under the principles that should gov 4088 ern in this area, there simply is no justification for adopting a 4089 15-minute STEL for benzene. 4091 4091 4091 4091 4091 4091 4091 4076 309/ The Washington Post, December 4, 1985 at ____. 4081 4082 4083 4084 4085 310/ [Cite to ACGIH Booklet.] OSHA has described ACGIH's Threshold Limit Value Committee as na respected group . . . [con sisting] of professional industrial hygienists (and toxicologists) in the employ of various governmental bodies." 50 Fed. Reg. 64, 71, col. 3. 256 257 258 - 125 - b 4093 7. Medical Surveillance 4095 4096 Section 1910.1028(i) of the proposed standard provides 4097 for the medical surveillance of benzene-exposed workers. CMA 4093 supports the inclusion of appropriate medical surveillance' 4099 requirements in the Benzene Standard. While the proposed medical 4100 surveillance provisions are well conceived overall, we believe 4101 they should be revised in a number of respects discussed below. 4103 4105 A. Employee Coverage 4107 4108 The proposed standard requires that each covered 4109 employer institute a medical surveillance program for all 4110 employees (1) who are exposed at or above the action level for 30 4111 or more days per year; or (2) who are exposed to benzene above 4112 the PEL for 10 or more days per year; or (3) who have been 4113 exposed to more than 10 ppm of benzene for 30 or more days in a 4114 past year while employed by their current employer.311/ We sup 4118 port providing medical surveillance to employees who are exposed 4119 to benzene at or above the action level for 30 or more days per 4121 year, or above the PEL for 10 or more days per year. However, we 4122 question the medical necessity and administrative feasibility of 4123 providing medical surveillance to employees who are not currently 4125 4125 4114 4115 4116 4118 311/ See proposed Section 1910.1028(i)(1)<i). In addition, medical examinations are to be provided for employees exposed in an emergency situation. See proposed Section 1910.1028(i)(4). This provision is discussed at p.___ infra. 256 257 258 126 MOD 000015726 4124 exposed to benzene simply because they were exposed to more than 4125 10 ppm benzene for more than 30 days in a prior year. 4127 4128 According to OSHA, the aims of medical surveillance are 4129 as follows: 4131 4133 1. Early detection and reversal of 4134 cytopenias and aplasias. 4135 4136 2. The prevention of some leukemias by 4137 reducing dose to the more susceptible 4137 workers. 4138 4139 3. Early recognition and treatment of 4140 those cases of leukemia which might occur and 4141 improvement in remission rate and duration. 4142 4143 4. Better evidence of the effec 4143 tiveness of the proposed standard.312/ 4148 4148 4149 Providing medical surveillance to formerly exposed workers is not 4150 likely to advance these objectives. 4151 4152 Since these workers would not currently be exposed to 4153 benzene above the action level, they clearly would not be at risk 4154 of developing cytopenias and aplasias as a result of current 4155 exposure. Furthermore, since a 10 ppm standard has been in 4155 effect for the past 15 years, any cytopenias or aplasias that 4156 might have been associated with high benzene exposures in the 4157 past would have been detected or have run their course long ago. 4158 Thus, with respect to these formerly exposed employees, medical 4161 4161 4144 312/ 50 Fed. Reg. 50512, 50563, col. 2. 256 257 258 - 127 - MCt> 000015727 4160 surveillance would not play any role in detecting and reversing 4161 cytopeniaa and aplasias. 4162 4163 Nor would medical surveillance play any role in pre- 4164 venting leukemias in these workers by reducing their benzene 4165 exposures. Under the proposed standard, any employee who is 4166 still subject to even a minor benzene exposure of 0.5 ppm for ^0 4167 days per year or 1 ppm for 10 days per year would be subject to 4168 medical surveillance. A worker whose benzene exposure is below 4169 eyen these low thresholds (of approximately 0.03 ppm-year) can 4170 hardly be said to be receiving a dose that would have to be 4171 reduced in order to prevent a significant risk of leukemia. 4173 4174 Providing medical surveillance for employees who were 4175 exposed to more than 10 ppm benzene in the past, but who are not 4176 currently exposed above the action level, will not provide any 4177 useful evidence regarding the effectiveness of the proposed stan- 4178 dard. Nor would it provide useful evidence of a comparative 4179 nature, since these employees would have had past exposures in 4130 excess of the current 10 ppm standard. 4181 4182 Thm only rationale for including these workers in the 4133 medical surveillance program that has even the semblance of 4134 plausibility is early recognition and treatment of leukemia cases 4185 which may be attributable to high exposures in past years. How- 4186 ever, since a 10 ppm standard has been in effect since 1971 (with 4187 a 10 ppm ANSI recommendation in effect since 1969), it is 256 257 258 128 MCD 00001572a 4183 unlikely that very many current employees (who are not now 4189 exposed above the action level) would fit the medical surveil- 4191 Lance criterion. Moreover if past exposures in excess of 10 ppm 4192 did create a risk of leukemia for these workers, the latency 4193 period (which OSHA estimates as 11 years313/) would have run some 4195 time ago, so that the leukemias should already have become appar- 4196 ent. Thus, medical surveillance of these employees is very 4197 unlikely to result in the early recognition and treatment of 4198 benzene-related leukemias attributable to past exposures in 4199 excess of 10 ppm. Moreover, there is likely to be very little 4200 improvement in remission rate or duration of any leukemias that 4201 are identified through medical surveillance. 4202 4203 In sum, there is little if any medical justification 4204 for requiring medical surveillance of workers who would not oth- 4205 ervise be covered as a result of their current exposures. Vet 4206 the task of identifying workers who were exposed to more than 4207 10 ppm benzene in a past year could be enormously difficult and 4208 burdensome -- so much so, that the slight possibility of some 4209 marginal health benefit is far outweighed by the administrative 4210 burdens. 4211 4212 As noted above, a 10 ppm standard has been in effect 4213 since 1971. Accordingly, very few current employees are likely 4215 4215 4194 313/ See id. at 50524, col. 3. 256 257 258 129 4214 to have been exposed to more than 10 ppm benzene without respira- 4215 tory protection during the past 15 years. Thus, identification 4216 of employees who fit the medical surveillance criterion based 4217 only on past exposures above 10 ppm would have to focus on 4220 employment and exposure records from the 1950's and 1960's. 4221 Exposure data for those years are likely to be sparse, much less 4222 reliable than recent exposure data, and difficult to tie to par- 4224 ticular current employees. These difficulties would be com- 4225 pounded by the fact that workers frequently are transferred 4226 between work sites on a temporary or more-or-less permanent 4227 basis. Further complications would arise in the case of 4228 employers which have closed some of their plants or have gained 4229 new employees through acquisitions of other companies or indus- 4230 trial units. In either event, there is a good chance that 4231 records for the past few decades may have been lost or simply not 4232 transferred. 4233 4234 In short, the task of identifying formerly exposed 4235 employees for purposes of medical surveillance would be formida- 4236 ble to say the least, even when the search is limited to workers 4237 whose exposures occurred while employed by their current 4238 employer. If benzene exposures that may have occurred while the 4239 worker was employed by a previous employer had to be identified, 4240 the task simply would become impossible. 4241 4241 4241 42 41 256 0000157 4242 Because of the enormous administrative burdens involved 4243 and the extreme unlikelihood that any significant health benefit 4244 would be realized, the requirement that medical surveillance be 4245 provided for formerly exposed employees should be deleted from 4246 the standard. If it remains, however, it must, at the very 4247 least, be clarified in several respects. 4248 4249 For one thing, as in the case of currently exposed 4250 employees, the test should be stated in terms of 8-hour time- 4251 weighted average exposure. Thus, coverage should apply only to 4252 employees who had more than an Q-hour time-waited average benzene 4253 exposure >lfor 30 or more days in a year prior to the effective 4254 date of the standard. Second, OSHA should attempt to mitigate 4255 the administrative burden by stating explicitly in the preamble 4256 to the standard that the employer need only make a "reasonable 4257 effort" to identify formerly exposed employees who meet the fore- 4258 going criterion. One approach that might be considered is 4259 requiring the employer to investigate past benzene exposures only 4260 in the case of those employees who identify themselves to the 4262 employer as likely to have had 8-hour average exposures of more 4263 than 10 ppm. In that way, the scope of the search could at least 4265 be focused on areas where the likelihood of a positive identifi- 4266 cation is highest. 4268 4268 4268 4268 4268 4268 256 257 258 131 4269 B. Frequency of Periodic Examinations 4270 4272 Section 1910.1028(i)(3 ) provides that periodic medical 4273 examinations shall be provided at least semi-annually. While 4274 semi-annual examinations might be appropriate if workers were 4275 exposed to benzene levels of 10 ppm or above, we believe that 4276 annual medical surveillance will be sufficient if the PEL .is 4277 reduced to the level OSHA has proposed. 4278 4279 OSHA's rationale for semi-annual examinations consists 4280 of the following two elements: 4283 4284 4285 4286 4286 4287 1288 4289 4290 4291 4291 4292 4293 4294 4294 4297 4297 4299 1. Early detection of marrow suppres sion before the cell counts are low enough to be life threatening, followed by removal should, in affected persons, prevent signifi cant morbidity including hemorrhage or infec tion or mortality .... 2. Employee questioning and counseling during the periodic examination to determine possible exposure to other bone marrow toxins . . . and other chemicals at work, in hobbies and in the home may enable the physician to counsel the employee as to reducing risks.314/ The second prong of OSHA's rationale is entirely 4300 unpersuasive. There is no reason why counseling employees about 4301 the risks of other bone marrow toxins cannot be done effectively 4302 on an annuel basis.315/ The first prong of the rational (early 4305 4305 4294 314/ 50 Fed. Reg. 50512, S0565, col. 1. 4302 315/ In addition, one may question whether OSHA should be 4303 concerned about counseling for non-occupational health risks %304 related to personal habits and hobbies. 256 257 258 132 - 0000 tf'7"'2 MCI) 4305 detection of marrow suppression) also fails to justify a 4306 semi-annual frequency for periodic examination. As discussed at 4307 pages ___-____ above, marrow suppression unrelated to leukemia 4308 should not be associated with benzene exposures at the levels 4309 that are of concern in this proceeding. And, if benzene 4310 exposures at ^Levels in the neighborhood of 1 ppm create any risk 4311 of leukemia at all, the risk will be very low. Thus, medical 4312 surveillance, whether provided annually or semi-annually, is not 4313 likely to result in the early detection of leukemia. Moreover, 4314 since the most likely form of leukemia to be found is acute 4315 myelogenous leukemia, the disease is likely to progress too rap 4316 idly once detected for medical surveillance to be of significant 4317 benefit. 4318 4319 OSHA correctly points out that various formed elements 4320 of the blood have relatively brief life spans, measured in days 4321 in some cases.316/ If medical surveillance is to be provided at 4322 an interval corresponding to the life span of these blood ele 4323 ments, it would have to be given on a weekly or monthly basis to 4324 ensure that the examination falls within the brief period in 4325 which the abnormality is first observable. Semi-annual 4326 examinations offer relatively little advantage over annual 4327 examinations ,in this respect. 4330 4330 4330 4330 4321 See id at 50565, col. 1-2 256 257 258 - 133 Of\v>0 VY~ 4331 4333 4334 C. Required Elements of Medical Examinations For the most part, we believe that OSHA has identified 4335 the appropriate types of laboratory tests for inclusion in medi 4336 cal examinations of benzene-exposed workers. We would, however, 4337 comment on two specific points. 4338 4339 1. Chest X-Rays 4340 4341 Proposed Sections 1910.1028(i)(2) and (3) state that 4342 workers required to wear respirators for at least 30 days a year 4343 must be provided with a pulmonary function test and a chest X-ray 4344 at the initial examination and at subsequent 3-year ^pulmonary 4345 function test) and 5-year intervals (chest X-ray). The objective 4347 of these tests is to assure that workers wearing respirators 4348 "will not be compromised by a pulmonary defect not detected by 4349 regular clinical examinations ."317/ According to OSHA, the pul 4351 monary function test "will pick up obstructive and restrictive 4352 pulmonary disease while the x-ray is designed to pick up lesions 4353 which may be clinically silent."318/ 4355 4356 CMA supports the concept of medically screening 4357 employees required to wear respirators in order to determine 4358 their pulmonry fitness. We also agree that the pulmonary 4360 4360 4350 317/ Id. at 50564, col. 3. 4354 318/ Id. 256 257 258 134 MOD 000015734 4359 function test is an appropriate procedure to employ for this gur- 4360 pose, and the 3-year repeat frequency seems appropriate. How- 4361 ever, we do not believe there is justification for requiring 4362 chest X-rays as a matter of course in order to determine pulmo- 4363 nary fitness. Instead, the question of whether a chest X-ray is 4364 provided should be left to the discretion of the examining physi- 4365 clan. We reach this conclusion for two reasons. 4367 4368 First, in contrast to a pulmonary function test, a 4369 cjiest X-ray is likely to provide little, if any, information 4370 about the employee's current pulmonary function or capacity. A 4371 chest X-ray also is likely to be of little use in predicting 4372 future diminished pulmonary capacity, given the multitude of 4373 factors that may be influential in this connection. 4374 4375 Second, and of fundamental importance, chest X-rays 4376 create an increased cancer risk that should be avoided whenever 4377 the chest X-ray is not really needed. For this reason, both the 4378 Food and Drug Administration (,rFDAn) and the American College of 4379 Radiology have strongly recommended against mandatory employment 4330 pre-placement chest X-rays of workers who have not been selected 4381 on the basis of individual history or examinations. FDA's 4332 National Center for Devices and Radiological Health has recom- 4333 mended: MOD 0000157:15 4386 4387 All mandated routine screening 4388 examinations of unselected populations should 4388 be discontinued, unless a significant yield 4389 can be shown.319/ 4393 4393 ,389 319/ FDA The Selection of Patients For X-Ray Examinations: 4390 Chest X-Ray Screening Examinations (September 1, 1983), as 4391 summarized in 13 FDA Drug Booklet 13 (August 1983). 256 257 258 - 135 ,0.' 0\ S> 4394 4394 1396 The American College of Radiology has made similar recommenda- 4397 tions in a formal Policy Statement concerning chest X-ray 4395 examination in occupational medicine. The Policy Statement, in 4399 relevent part, reads as follows: 4403 4404 4405 4406 4407 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4416 4417 418 4423 4423 4424 Preemplovment/Preplacement Examinations for Appropriate Job Placement: Preplacement chest X-ray examinations should be done selectively based on pertinent factors in the (1) occupational and medical hi tory, (2) clinical examination, and 3) proposed work assignment. Exposure Surveillance Chest X-ray surveillance of persons who work with or may be exposed to substances that adversely affect pulmonary function or cause pulmonary disease should be based on a periodicity consistent with the current understanding of the disease process.320/ Neither benzene exposure nor respirator use are sus- 4425 pected of adversely affecting pulmonary function or causing pul- 4426 monary disease. Accordingly, in light of the marginal contribu- 4427 tion that chest X-rays may make to evaluating pulmonary function 4428 and the radiation risks associated with chest X-rays, OSHA should 4429 not establish a mandatory requirement for periodic chest X-rays 4430 in the Benzene Standard. Instead, the question whether a chest 4433 4433 4418 4419 4420 320/ Referral 1982). American College of Radiologly Policy Statement, Criteria For Chest X-Ray Examinations (September 22, 256 257 258 136 MC0 000015737 4431 X-ray is to be provided should be left to the discretion, of the 4433 examining physician, at least after the initial baseline 4434 examination. 4435 4436 2. Emergency Examinations 4437 4438 Section 1910.1028(i)(4) of the proposed standard 4439 requires that a urinary phenol test be performed in the case of 4440 employees who are exposed to benzene in an emergency situation. 4441 OSHA has asked whether the provision of a urinary phenol test is 4442 appropriate and whether measurement of blood or breath benzene 4443 also should be required. 4444 4445 At the outset, we would urge OSHA to clarify the emer- 4446 gency exposure provisions of the standard, so that they more 1447 accurately reflect the exposure scenario that is of concern. 4448 Section 1910.1028(b) of the proposed standard defines ^emergency" 4449 to mean any occurrence "which may or does result in an unexpected 4450 significant release of benzene." Since the principal function of 4451 the term "emergency" is to trigger special medical examinations, 4452 the last clause of the definition should be revised to read as 4453 follows: "which results in an unexpected significant release of 4454 benzene." The fact that an occurrence "may . . . result in an 4455 unexpected significant release of benzene" would not justify an 4456 emergency medical examination if the release does not actually 4457 occur. 4458 4458 256 257 258 137 Mc- 000015738 4459 A corresponding revision should be made in Section 4460 1910.1028(i)(1)(i) of the proposed standard, which currrently 4461 requires medical surveillance, inter alia, "for employees who 4462 have been exposed to an emergency situation." We believe that 4463 this clause could be clarified by providing that medical surveil 4464 lance is required for "employees who have been exposed to an 4465 unexpected significant release of benzene in an emergency situa 4466 tion." A corresponding change should be made in Section 4467 1910.1028(i)(4)(i), so that emergency examinations would be 4468 required: "If the employee is exposed to an unexpected signifi 4469 cant release of benzene in an emergency situation . . . 4471 4472 Once the foregoing clarifications are made, we support 4473 the provision of a urinary phenol test to employees who have been A474 exposed to unexpected significant releases of benzene. However, 4475 OSHA should not mandate that the tests be given "at the end of 4476 the employee's shift." The level of phenol in the urine is 4477 likely to reach its peak three to four hours after the emergency 4478 exposure. If the urinary phenol test must invariably be given at 4479 the end of the shift, this peak would be missed in those 4480 instances in which the emergency occurs either early or late in 4481 the shift. Proposed Section 1910.1028(i)(4)(i) should, there 4482 fore, be revised to provide that the urinary phenol test be given 4483 "at the appropriate interval following exposure" rather than "at 4484 the end of the employee's shift." 4486 4486 256 257 258 138 MCD 000015739 4487 While we support the provision of a urinary phenol test 4488 in emergency exposure situations, we do not believe that measure- 4489 ment of blood or breath benzene should be required. The urinary 4490 phenol test will identify all cases in which follow-up blood 4491 tests are appropriate. (Indeed, if anything, the urinary phenol 4492 test is likely to produce false positives, since phenol Revels 4493 may be elevated for reasons other than exposure to benzene.) 4494 There simply is no need for blood or breath tests, particularly 4495 since the timing of such tests would be critical in light of the 4496 rapid metabolization of benzene.321/ 4498 4502 4503 4504 4504 4508 D. Criteria for Referral to a Hematologist or Internist Section 1910.1028(i)(5) of the proposed standard .509 establishes "normal" parameters for hemoglobin level, thrombocyte 4510 count, and leukocyte count, and provides that where blood count 4511 results fall outside of these parameters, the blood count must be 4512 repeated within two weeks. If the "abnormality" persists, the 4513 employee must be referred to a hematologist or internist for fur- 4514 ther evaluation "unless the [examining] physician has good reason 4515 to believe such referral is unnecessary." 4517 4518 CMA supports the concept of referring employees with 4519 ^abnormal" blood count results to a hematologist or internist for 4520 4520 4496 321/ See 50 Fed. Reg. 50512, 50566, col. 2. 256 257 258 - 139 - v67 40 4520 further evaluation in appropriate cases. However, in Light of 4520 the uncertainty as to whether particular blood count results are 4522 ^normal" or "abnormal" in the case of a particular individual, we 4523 believe tht medical judgment and physician discretion should play 4524 a large role in determining whether referral is indicated in 4525 individual cases. There is no question that variations from 4526 hypothetically "normal" blood count limits may be related to such 4527 factors as sex, age, race, smoking, exercise, and geography.322/ 4532 As OSHA recognizes, "all these factors need to be cosidered when 4533 defining acceptable Revels of formed blood elements for 4534 preemployment assessment as part of routine medical surveillance 4535 and for decisions to refer benzene workers to a \537 4538 With the foregoing points in mind, we would urge that 4539 the medical referral provisions of the standard be revised to 4540 provide a clearer emphasis upon the role of medical judgment and 4541 physician discretion in this area. This could be accomplished 4542 by: (1) deleting the requirement that the blood count be 4543 repeated within two weeks,324/ and (2) removing the specific 4548 4548 4528 4529 4530 4531 322/ See id. at 50565, col. 3 - 50566, col. 1. For example, blacks have been reported to have lower white cell counts than whites. See ___________________ ___________. In addition, the question of what limits are "normal" may vary from one laboratory to another. 4536 323/ 50 Fed. Reg. 50512, col. 3. 4543 324/ 4544 may be 4548 5 256 257 258 The appropriate interval for repeating a blood count greater than two weeks in some cases. This would be true. [Footnote continued next page] 140 4548 triggering values from the standard itself and including them, 4549 instead, as recommended guidelines in an appendix to the stan 4550 dard. 4552 4553 E. Medical Removal and Wage Rate Retention. 4555 4556 Section 1910.1028(i){8) of the proposed standard pro 4557 vides that an employee who is referred to a 4557 hematologist/internist for further evaluation must be temporarily 4558 removed from benzene exposure and that a decision on whether the 4559 removal should continue is to be made by the examining physician 4560 after consulting with the hematologist/internist. We believe 4561 that these provisions for medical removal are appropriate. 4563 4564 In this regard, we would emphasize the particular 4565 importance of having the examining physician, rather than the 4566 hematologist/internist, make the decision as to whether the 4567 removal should continue. While the advice of the 4568 hematologist/internist will be an important element in the ulti 4569 mate decision, the examining physician will be much more familiar 4570 than the outside hematologist/internist with the occupational 4571 environment and the nature of benzene exposures. He will, there 4572 fore, be in the best position to evaluate the need for contining 4573 medical removal. 4574 4574 5 5 4545 4546 4547 [Footnote continued from preceding page] for example, when the physician suspects that the "abnormal" count is the result of a virus of greater than two-weeks* expected duration. 256 257 258 141 MOD *>00015742 4575 As proposed, the standard does not contain a wage rate 4576 retention provision. That is, employers are not required to pro- 4577 vide an alternative job at the same rate of pay to employees who 4578 are subject to medical removal. The decision not to include a 4579 rate retention provision in the proposed standard was correct, 4580 since such a provision is neither legally supportable nor neces- 4581 sary in practice. 4532 4583 On the legal side, two considerations are relevant. 4533 First, we believe there is a substantial question as to whether 4535 OSHA has authority to require wage rate retention under any set 4586 of circumstances. A rate retention provison would appear to con- 4587 flict with Section 4(b)(4) of the Occupational Safety and Health 4583 Act of 1970, 29 U.S.C. 5 653(b)(4), since it can be said to 4592 4593 4594 4595 4595 4596 4597 4597 4598 4601 4601 4602 affect any workmen*s compensation law or to enlarge or diminish or affect in any other manner the common law or statutory rights, duties, or liabilities of employers and employees under any law with respect to injuries, diseases, or death of employees arising out of, or in the course of, employ ment. Moreover, the inference that Congress did not intend to 4603 authorize wage rate retention under the Occupational Safety and 4604 Health Act is strongly suggested by the fact that Congress passed 4605 the Act without expressly providing for rate retention only one 4606 year after considering and expressly including a rate retention 4607 provision in the Federal Coal Mine Health And Safety Act of 1969, 256 257 258 142 43 4603 30 U.S.C., 811(d), 843(b)(2)(3). For these reasons, despite 4609 the holding of the court in United Steelworkers of America 4610 AFL-CIO v. Marshall, 647 F.2d 1189 (D.C. Cir. 1980), cert, 4611 denied, 453 U.S. 913 1981), a substantial question exists 4612 regarding OSHA's authority to include a wage rate retention pro 4613 vision in any occupational health standard. 4615 4616 But even if OSHA has. authority to impose a rate reten 4617 tion requirement in some cases, it is clear, under American 4618 Textile Manufacturers Institute, Inc, v. Donovan, 452 U.S. 490, 4619 538 (1981), that a rate retention provision, if authorized at 4620 all, "must be justified on the basis of [its] relation to safety 4621 and health." As the Supreme Court emphasized in that case, OSHA 4622 does not have a roving commission to establish wage rate reten 4623 tion requirements, since "the Act in no way authorizes OSHA to 4624 repair general unfairness to employees that is unrelated to 4625 achievement of health and safety goals . . . ." Id. at 540. 4627 4628 In the case of benzene, a rate retention provision 4629 could not be justified on the basis of a relation to safety or 4630 health. In the United Steelworkers decision, there was concern 4631 that employees might take a chelating agent in order to reduce 4632 the levels of lead in their blood in order to avoid being trans 4633 ferred to a lower paying position. This, it was feared, might 4634 result in workers being exposed to concentrations of lead which, 4635 in light of their true blood lead levels, might be dangerous to 256 257 258 143 MCD 000015744 4636 their health. This potential problem does not exist in the case 4637 of benzene, since employees have no way of making an abnormal 4638 blood count appear normal, and cannot manipulate the results of 4639 pulmonary function tests in order to make their pulmonary func- 4640 tion appear better than it actually is. In short, there is no 4642 health justifiction for requiring rate retention under the 4643 Benzene Standard. 4644 4645 Nor does there appear to be any need for such a provi- 4646 sion in practice. The fact is that very few employees will be 4647 removed from a benzene-exposed assignment because of adverse 4646 blood effects attributable to benzene exposure. Even under the 4649 current 10 ppm standard, the experience of the petrochemical 4650 industry indicates that virtually no removals attributable to 651 benzene-related blood abnormalities have occured in recent 4653 years.325/ If anything, even fewer removals would be expected 4654 under the proposed standard, since it contemplates lower permis- 4655 sible levels of exposure than are presently permitted. 4657 4658 Moreover, rate retention is the subject of collective 4659 bargaining agreements and formal and informal personnel policies 4660 and practices in most affected industries. In the petrochemical 4661 industry, for example, .... [what can we say here?]. Thus, as 4662 OSHA observes "the details of [medical removal transfers] . . . 4665 ________ 4665 4653 325/ [Cite to results of CMA Survey.] 256 257 258 144 q 4663 are best left to collective bargaining and employer personnel 4665 policies."326/ 4666 4667 In sum, a rate retention provision in the Benzene Stan 4668 dard would be improper as a matter of law, inappropriate as a 4669 matter of labor-management relations, and unnecessary in prac 4670 tice. Such a provision should not be included in the final stan 4671 dard. 4672 4672 4676 F. Comments on Appendix C - Medical Surveillance 4677 Guidelines for Benzene 4678 4682 The medical surveillance guidelines of Appendix C 4683 require a few modifications in order to bring the information up 4684 to date, sharpen its accuracy, and avoid misleading implications 4685 that the described indicia are solely attributable to benzene 4686 etiology. 4687 4688 1. The General Provisions of Section V.A. 4689 4690 The first paragraph in section V.A. of Appendix C 4691 describes the principal relevant effects of benzene exposure as 4692 ^alterations of the hematopoietic system as reflected by changes 4693 in the peripheral blood and leukemia." It also states that the 4694 purpose of the medical surveillance program is to observe early 4695 signs of "these effects." Alterations of the hematopoietic 4697 4697 4665 50 Fed. Reg. 50512, 50567, col. 1. 256 257 258 - 145 - O00(- Ab 4696 system, however, are not specific to benzene exposure but may 1697 arise from other causes of bone marrow depression or from idio 4698 pathic leukemia. This paragraph therefore should be modified as 4699 follows to avoid implying that any such observed effects 4700 necessarily would have occurred as the result of benzene 4701 exposure: 4704 4705 4706 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4718 4718 '719 The medical surveillance program is designed to observe indices for early signs of alterations of the hematopoietic system, as reflected by changes in the peripheral blood and leukemia. Alterations of the hematopoietic system are the principal effects of benzene exposure that form the basis for this regulation, but they are not specific to benzene etiology. The same effects may also arise as the result of idio pathic leukemia and other causes of bone mar row depression. 2. The Hematology Guidelines of Section V.B. 4720 4721 The hematology guidelines should be brought up to date 4722 by incorporating currently accepted practice. They should also 4723 avoid wrongly implicating benzene overexposure as the exclusive 4724 causal mechanism for the observed abnormalities. 4726 4726 4728 4729 4730 4733 4734 a. The Hematology Guidelines Should Be Updated to Recognize the Use of Automated Blood Counters. pie opening paragraph of section V.B., which states 4735 that the guidelines are derived from the analysis of Dr. Jandl, 4736 should further indicate that the date of submission of that 256 257 258 - 146 - MOD 0000.15747 4737 information was 1977. It should also indicate that, in the 4738 intervening years, automated blood cell analyzers have come into 4739 common use in clinical laboratories and may properly be used to 4740 meet the requirements of Appendix C. 4741 4742 The requirement of section V.B.l. that blood counts be 4743 performed using an "automated (Coulter) counter" should be 4744 modified to delete the reference to "Coulter", a particular pro- 4745 prietary brand. That same section should also acknowledge that 4746 automated analyzers now provide an index of the distribution of 4747 red blood cell volume (RDW), which measures the heterogeneity of 4748 the red blood cell population. In fact, the RDW represents a 4749 coefficient of variation of red blood cell volume distribution. 4750 Section V.B.l. also should recognize that mean platelet volume *751 MFV) is a sensitive measure of platelets. 4753 4753 4755 4756 4757 4758 4761 4762 b. The Hematology Guidelines Should Be Modified To Avoid Inaccurately Attributing Certain Indicia to Benzene Exposure. Section V.B.3. prescribes the minimum mandatory 4763 observations to be made from the peripheral blood smear and 4764 describes their significance. Subsections a through d are not 4765 particularly useful to the practitioner and do not represent uni- 4766 versally accepted practice. They should be deleted from Appendix 4767 C. Subsection V.B.3.e., on the other hand, contains a comprehen- 4768 sive and generally helpful discussion of the observation of 256 257 258 - 147 - oooo^ 0 4769 hematologic abnormality and should be retained, with a few modi- 4770 fications. 4771 4772 pie end of the first paragraph in subsection e notes 4773 that vacuolation in erythroblasts and myelocytes is a relevant 4774 phenomenon "induced by many toxins apart from benzene, including 4775 chloramphenicol and alcohol; and by infections."327/ This char- 4777 acterization of benzene as only one among several possible causes 4778 is correct. Statements to the contrary in the second and third 4779 sentences of the same paragraph should be made to conform by 4780 deleting inaccurate and misleading references to benzene as the 4781 cause of these general changes reflected in the peripheral blood. 4783 4784 pie fourth sentence in the first paragraph of 4785 subsection V.B.3.e. states that "the findings of two or more 4786 cytopenias, or of pancytopenia, must be regarded as highly suspi- 4787 cious of more advanced although still reversible, benzene 4788 toxicity."328/ This effect is not specific to benzene exposure, 4790 but is true for all bone marrow depression etiology. Thus, the 4791 statement should be modified to delete "benzene toxicity" and 4792 replace it with "bone marrow depression." 4793 4794 pie next sentence states that when pancytopenia 4795 develops and becomes associated with the appearance of immature 4797 4797 4775 327/ 4788 328/ 256 257 258 50 Fed. Reg. 50512, 50580, col. 2 . .Id. col 1 - 148 - oooO^ 4796 cells, or with inappropriate elevations or monocytes, basophils, 4797 or eosinophils, "the findings must be regarded as evidence of 4798 benzene overexposure unless proved otherwise."329/ Although 4799 benzene exposure is one possible cause of such findings, it is 4800 not necessarily the most likely. The statement should be 4801 modified to provide that "the findings must be regarded as 4802 evidence of benzene overexposure if there are no other plausible 4803 or more reasonable explanations." 4805 4805 4807 4808 4809 4812 4813 c. Various Other Statements in the Guidelines Should Be Altered in the Interest of Accuracy. To avoid implying that a single observation is suffi 4814 cient, the third sentence of subsection V.B.l. should specify 4815 that a "persistent decline from a normal to a subnormal" red cell 4816 count is indicative of potential toxicity. Also, the succeeding 4817 statements specifying normal values should be replaced with one 4818 that reminds the physician that a normal white count varies (1) 4819 among individuals, (2) in the same individual from day to day, 4820 and (3) from one laboratory to another. The determination of 4821 "abnormal" should be made case-by-case, based on these factors 4822 rather than through the application of absolute generic values. 4824 4825 If subsection d of section V.B.3. is retained, it 4826 should at least make clear whenever it cites an indicator as 4828 4828 4798 329/ Id. 256 257 258 - 149 4827 evidence of benzene toxicity that the indicator is not 4828 pathognomonic. This fact is properly recognized in the second 4829 paragraph of subsection d, which provides that an increase in the 4830 proportion of band forms among the neutrophilic granulocytes 4831 be considered as an early warning of benzene toxicity in 4832 the absence of other causative factors (most commonly infec- 4833 tion)."330/ Similarly, the first sentence in the third paragraph 4835 of subsection d should provide that an upward trend in the number 4836 of basophils is to be regarded as "possible evidence of benzene 4837 toxicity in the absence of other causative factors." And the 4838 last sentence of the fourth paragraph should contain the same 4839 phrase with regard to monocyte counts. 4841 4842 finally, the fifth paragraph should make clear in its 1843 first sentence that acquired Pelger-Huet anomaly is an extremely 4844 uncommon indication of benzene produced injury. The subsequent 4845 sentences admit that the Pelger-Huet anomaly is sometimes 4846 hereditary and unrelated to leukemia, and that even when not 4847 hereditary, it is not invariably predictive of leukemia.331/ And 4848 OSHA acknowledges that only about two percent of leukemics 4850 exhibit the anomaly.332/ Proper emphasis would be placed on the 4852 predictive value of the Pelger-Huet anomaly if the paragraph were 4853 reworded in the following manner: 4856 4856 4833 330/ Id. at 50579, col. 3 emphasisadded). 4847 331/ Id. at 50579, col. 3 - 50580, col. 1. 4850 332/ 256 257 258 Id. at 50580, col. 1. ISO 4857 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4874 4874 4876 4876 4876 4876 4876 4876 4876 4876 4876 4876 t876 4876 4876 4876 4876 4876 4876 4876 4876 4876 4876 4876 4876 4876 4876 4876 4876 4876 4876 4876 4876 4876 4876 256 257 258 Approximately two percent of patients who ultimately develop acute myelogenous leukemia show the "pseudo" (acquired) PelgerHuet anomaly. In this anomaly many, or some times the majority, of the neutrophilic granulocytes possess two round nuclear segments--less often one or three round segments--rather than three normally elon gated segments. When this anomaly is not hereditary, it is often but not invariably predictive of subsequent leukemia. A finding in the peripheral blood of the acquired Pelger-Huet anomaly is therefore a serious indication of bone marrow injury that may result from benzene overexposure or other causes. o'.0O' ' 151