Document ExO4E37Ev9zaoVmKo12k5BDOV

Thomas G. Grumbles VISTA To V^oVia-al Voc^cj Do,. Ljfx | A i; a a- ---------- JH qfa :jX !\\\acWo ib Uf.^Grtt.cl_^v VW ex'o^- 3 ^possive <Qos j \o/v^Ve^s aCCw<`a ea - e s (aa.z V- 'V'C R. v\ to _,A0 - Wo.-r^ V-IH^ -- O r">\ ^ o n Vv\< . **y rv^ 5 550JZSS> toa^c^ Gn 0 or\V^ Oj itVs Q> Ci - C C.T^ 0.ta al^ $ Jb * l^r- VVV 000006373 iM s; nvQ. ASSOC J 52.4, lS!-'57 Field Validation of Three Passive Dosimeters for Excursion Limit Monitoring of Ethylene Oxide Frank G. Szopinski* Mark A. Puskart Lawrence H. Hecker Corporate Industrial Hygiene Laboratory, Abbott Laboratories, D-38A. North Chicago. IL 60064 The Occupational Safety and Health Administration (OSHA) set a 5-pptn excursion limit (EL) for ethylene oxide (EtO) in April 1988. Both active and passive sampling methods have been proposed for monitoring workers against this new standard. Passive dosimetry has considerable advantages over active sam pling for monitoring short-term exposures to EtO, including reduced sampling and analysis complexity, increased chemical stability, and reduced cost. The major disadvantage of these passive methods is their questionable ability to meet the OSHA requirementfor the test result to fall within 35% ofthe "true" result with 95% confidence at the EL over a 15-min sampling period. A field validation study was performed to estimate the accuracy ofthree EtO EL passive dosimeters: 3M 3550/3551TM. Crystal DiagnosticsAirScan*, andAssay Technology EO CHEM CHIPTM. Area samples were taken atfour unique concentration areas within a hospital products sterilizationfacility. A specially designedfield exposure chamber was used to expose 12 dosim eters ofeach type concurrently at each sampling location while concurrently collecting six Tedlar9 bag samplesfrom locations surrounding the dosimeter array. The Tedlar bag samples were analyzed on-site by gas chromatography with flame ionization detection (GC-FID). To enhance the strength of this validation study, manufacturers of the dosimeters were requested to take part in the investigation. Their input was used during the design ofthe exposure chamber and study protocol and in the interpre tation of the results. Two of the three dosimeter types were analyzed by the investigators. Dosimeters of the third type, 3M 3550/3551. were dividedbetween an outsideAmericanIndustrial HygieneAssociation-accreditedlaboratory and3M's laboratory for analysis. Using the Tedlar bag/GC-FID sample data to cal culate the "actual concentration'' during the exposure of the monitors, the precision, bias, and accuracy were calculatedfor each method at exposures bracketing the OSHA EL (ranging from 2.12 ppm to 15.8 ppm). The Assay Technology method did not meet the 35% accuracy requirement at any concentration Currently employed with The Raterman Croup, Ltd., 154 West Hubbard, Suite 602, Chicago, IL 60610. tAuthor to whom all correspondence should be addressed. Copyriglt 1991. American Industrial Hygisna Association level studied. The Crystal Diagnostics method did not meet the 35% accuracy requirement at EtO concentrations above 2.12 ppm. The 3M samples analyzed by the manufacturer met the 35% accuracy requirement with a pooled accuracy of 25%. The 3M samples analyzed by the outside laboratory did not meet the 35% requirement at concentrations below 9.41 ppm. n the hospital products manufacturing industry, ethyl Iene oxide (EtO) is used for the sterilization of materials that cannot be subjected to high-temperature autoclaving. In hospitals, EtO is used for the sterilization r surgical instruments and supplies. Although the vast majority of EtO produced is used in the chemical production of glycols and related chemicals, it is estimated that the greatest number of occupational exposures to EtO occur during its use as a sterilization agent.(>> The acute toxic effects of EtO, including skin, eye, and respiratory tract irritation and central nervous system depression, have long been recognized. Recently, major concern has focused on the carcinogenic and mutagenic effects of Et0.a3> For these reasons, in 1984, the Occupational Safety and Health Adminis tration (OSHA) lowered the permissible exposure limit (PEL) to 1 ppm 8-hr time-weighted average (TWA) with a 0.5-ppm TWA action level. A short-term excursion limit (EL) of 5 ppm was added by OSHA in April 1988. Methods for the determination of occupational exposures to BO include the use of real-time area monitors, impingers, active adsorbents, and passive dosimeters. The advantages and disad vantages ofeach type have been discussed in detail elsewhere/41 Basically, passive dosimetry has considerable advantages over active sampling for monitoring exposures to EtO, including reduced sampling and analysis complexity, increased chemical stability, and reduced cost The major disadvantage of these passive methods is their questionable ability to meet the OSHA 35% accuracy requirement at the 5-ppm excursion limit over a 15-min sampling period. In this study, the method accuracy of three types of EtC passive dosimeters (3M 3550/3551TM, Crystal Diagnostics AirScan, and Assay Technology EO CHEM CHIPTM) were estimated in a hospital products sterilization facility at con 151 VVV 0000G6374 AM. MO. HYa ASSOC. J. (K) / April 1991 centrations bracketing the 5-ppm EL. A specially designed field exposure chamber was developed to allow all monitors to be exposed for the exact same time in the field. The validation of this exposure chamber and sampling protocol is detailed elsewhere.l5) The 3M 3550/3551 dosimeter uses a derivati2ation technique that converts the EtO to 2-bromoethanoI on the surface of the dosimeter during collection. The manufacturer recommends analysis by gas chromatography with electron capture detection (GC-ECD). Others have had better accuracy with analysis by gas chromatography with flame ionization detector (GC-FID)/61 The major disadvantage of this technique is that the monitors must be sent to an industrial hygiene laboratory for analysis. The EO CHEM CHIP passive monitor utilizes a colorimetric reagent system immobilized on an adsorbent strip, which is encased within an EtO-selective membrane. EtO is chemically bonded as it enters the system and is held for later development. Concentration is proportional to surface reflectance of the ab sorbed strip/71 On-site analysis of these monitors is possible immediately after sample collection. This is not, however, a direct-reading method. Specific corrections, including time, temperature, and blank, must be made to the raw data to generate accurate results. A description of this technique for 8-hr PEL monitoring, including a detailed discussion of the necessary corrections, is avaitable.t8> The Crystal Diagnostics AirScan utilizes the principle of nucleation and crystal growth to determine EtO concentrations. Basically, EtO causes a chemical reaction within die monitor forming an array of crystal seeds on sensitized film; a develop ment solution, applied later to the dosimeter, causes rapid crystal growth directly propoitional to the exposure level191 A major advantage of this method is that the sample can be developed and directly read on-site. One disadvantage is that a 2-hr waiting period is required following sampling and prior to development. 25% of the true value in at least 95% of the samples analyzed at 1.0 ppm and 2.0 ppm and that the total error should also be less than 35% in at least 95% of the samples analyzed at the action level (0.5 ppm). With EtO EL monitoring, however, accuracv requirements have not been set by OSHA at levels above an below the 5-ppm standard. The authors believe that to be useful in assessing worker exposure, any air-sampling method for short-term EtO monitoring must have at least 35% accuracy over the range of 2.5 ppm to 20 ppm. Even if a sample's result is clearly over the standard (e.g.. 15 ppm), if the result is known to be accurate, useful exposure information has been generated. The magnitude of the overexposure may help pinpoint the contaminant's release point or help in the identification of ap propriate control strategies. From an OSHA compliance stand point, there is little need for method accuracy requirements at 4x the standard, but for the practicing industrial hygienist who may want to use the method to collect area samples inside known high-concentration areas, die known accuracy of a method at concentrations above the standard is important. The accuracy of an industrial hygiene sampling and analyt ical method is a function of both the precision and the bias.02*13* Precision can be estimated by the total coefficient of variation (CVt) for replicate determinations including both sampling and analysis steps. The bias (Bi) is estimated by using Equation 1, where the amount found is averaged over replicate determina tions or the averaged bias is determined over a range of known amounts with single determinations. ,, Amount Found - Amount Known Amount Known Once both the precision and the bias of a method have be estimated, the accuracy can be estimated by using Equation 2. Accuracy (+%) * [2(CV,) +18)1] x 100 (2) FIELD VALIDATION Field validation of a new industrial hygiene method is necessary after laboratory validation of the method has been completed. The major problem with field validation studies is that the "actual concentration present" is generally notknown in the field atmosphere; therefore, method accuracy cannot be determined. Tedlar bag sampling, followed by gas chromatography analysis has been used for monitoring the concentration ofEtO in ambient air.(,0> The problem with this method has been the lack ofstability ofthe samples between collection in the field and analysis in the laboratory. The advent of the portable gas chromatograph offers the ability to take the instrument into the plant and concurrently generate near-real-time area data during field validation of EtO passive methods/4-61 By estimating the actual concentration pres ent during field validation, the actual method accuracy can be calculated from field validation data. The OSHA standard for EtO EL method accuracy states that the absolute total error (sampling plus analysis) should be less than 35% in at least 95% of the samples analyzed at 5.0 ppm.'ln With EtO PELmonitoring, OSHA accuracy requirements are also set at levels below and above the PEL of 1.0 ppm 8-hr TWA.01 The standard states that method results should fall within METHODOLOGY Passive dosimeters developed specifically for monitoring occu pational exposure to EtO are marketed by numerous manufac turers. During this field validation study, 12 monitors from three manufacturers were simultaneously exposed at four different EtO concentrations bracketing the 5-ppm EL. Tedlar bag sam ples were also filled during the course ofthe sampling period and analyzed by portable GC-FID. Sample location parameters in cluding temperature, air velocity, and humidity were recorded at each sampling location during sample collection. The Crystal Diagnostics and Assay Technology samples were analyzed by the investigators on-site. The 3M dosimeters, which must be sent to a laboratory for analysis, were divided between an outside American Industrial Hygiene Association (AIHA)-accredited lab and 3M for analysis. 3M analyzed the monitors they received by using GC-ECD, and the outside laboratory analyzed the 3M 3550/3551 they received by using a GC-FID method. The precision, bias, and accuracy were calculated for each method at EtO-exposurc levels bracketing the 5-ppm EL by using the Tedlar bag/GC-FID data as the actual concentration present To enhance the strength of this study, the three mai facturers of the EtO passive dosimeters were requested to take AM. MO. HYO. ASSOC. J. (52) / April 1991 152 000006375 VVV - in this investigation. Their input was requested during the design of the validation protocol, during the interpretation of the results, and in the preparation of this manuscript. The major role 0f each manufacturer during the study, however, was to advise the investigators on proper sampling, handling, preservation, analysis' and shipping procedures to ensure that the sampling and analysis techniques were not compromised. Four sampling locations were selected inside a hospital products manufacturing sterilization facility, which was cho sen because it had the highest temperatures and relative hu midities of any of the authors' North American facilities. The assumption was made that if the dosimeters could operate in this environment, they would function at any facility. The goal of the sampling location selection process, inside the facility, was to find sampling locations that would average the follow ing concentrations: 2.5 ppm, 5.0 ppm, 10 ppm, and 20 ppm, for 15 min. In actuality, the concentrations averaged 2.12 ppm. 5-28 ppm, 9.41 ppm, and 15.8 ppm as determined by the Tedlar bag/GC-FID data. To ensure that each set of dosimeters was exposed to a homogeneous test atmosphere for exactly 15 min, a field expo sure chamber and sampling protocol were developed. The field use of this protocol, along with the validation studies that were performed on the exposure chamber, sampling system, and cal ibration systems, are described elsewhere.021 Briefly, the chamber was designed as a 50- by 55-cm aluminum base with a 36- by 31- by 30-cm removable Plexi glas9 cover. An adjustable aluminum grid allowed for all the dosimeter sampling ports to be held at the same level above the base. External sampling ports were installed at six loca tions across the base. Each of these ports was linked to an external personal sampling pump. The pumps were used to fill six separate Tedlar air-sampling bags during the course of the sampting period. Prior to and after the sample exposure peri od, the cover was placed on the base and the chamber was purged with EtO-free air. The purged chamber was removed from the contaminated area and the dosimeters were prepared for analysis or shipping. Blanks of each type of monitor and Tedlar bag/GC-FID samples were collected during the setup, exposure, and removal process inside the EtO-free setup lo cation, where the monitors were handled, to document that no additional EtO contamination occurred during the preexpo sure and postexposure handling of the monitors. CALIBRATION OF EQUIPMENT Model HFS-513A Gilian high-flow sampling pumps (Gilian Instrument Corp., Wayne, NJ.) were used to draw air into the 4-mil thickness, 33-cm by 51-cm (10-L) Tedlar air-sampling bags (SKC-234-10) at a rate of 0.5 L/min. The pumps were calibrated prior to and after the 4 days of sampling. The Tedlar bag samples were analyzed with an AID Model 511 portable gas chromatograph with a flame ionization detector (Analytical Instrument Development, Avondale, Pa.) equipped with a 3-mL gas-sampling loop. Air was continuously drawn through the loop at a rate of 50 mL/min with a Model XX55 vacuum/pressure pump (Millipore, Bedford, Mass.). Air samples were injected on-column by turning a two-way valve, which allowed the carrier (helium) to flow through the sampling loop and onto the column. The chromatograph was fitted with a 4-ft stainless steel column filled with 20% dinonyl phthalate on Chromosorb W (Supelco, Bcllefonte, Pa.), The GC-FID w: interfaced with a Hewlett Packard Model 3393A integrator equipped with a Model 9114B 3.5-in. disk drive for data storage. The detector's response was linear in the concentration range of interest. A linear least-squares fit typically yielded a correlation coefficient greater than 0.999. The GC-FID was calibrated over a range of standards bracketing the EtO concentrations of inter est, prior to and after sampling at each location. Standards with concentrations at the suspected EtO level were also analyzed throughout the Tedlar bag analysis process. Known concentra tions of EtO in air were generated by using a Model 450-57-D dynamic gas analyzer calibrator (Vici Metronics, Santa Clara, Calif.), as described elsewhere/5* Air dilution flow rates were measured with a Model PN D-800268 Gilibrator (Gilian Instru ment Coip., Wayne, N.J.) primary flow calibrator. With primary dilution, EtO standards were prepared in the range of 0.9 ppm to 32 ppm. With secondary dilution, standards were produced down to 0.1 ppm. The instrument's actual limit of detection is below 0.1 ppm but was not determined. DETERMINATION OF ETO CONCENTRATION USING GC-FID Each of the six bag samples collected during the 15-min sam pling period were analyzed in quadruplicate. All bags were analyzed within 1 hr of collection. The mean concentration an standard deviation (SD) were calculated for each ofthe six bags. The mean value and SD were used to calculate a 95% confidence interval for the concentration found for each field evaluation performed/14* This confidence interval, instead of the mean concentration value, was used to represent the actual concentra tion present for the calculation of the bias for each passive method. As an example, the mean concentration determined for Location 2 was 9.41 ppm. The 95% confidence interval was determined to be between 9.14 ppm and 9.68 ppm. Therefore, any passive method's mean concentration determined to be between those two values was stated as having no bias, and methods with means either below 9.14 or above 9.68 were stated to have bias. The bias was then calculated by using Equation 1. ANALYSIS OF SAMPLES Special precautions were taken to ensure accurate analysis with the 3M 3550/3551 device. To estimate analysis bias caused by laboratory error and to determine if accuracy was a function of the gas chromatography detector used to analyze the samples, the dosimeters were split and analyzed by two separate labora tories. Therefore, each laboratory received a total of 28 samples (6 samples plus 1 blank from each of the 4 sampling locations). 3M's Monitor Analysis Service performed the analysis on half of the 3550/3551 dosimeter samples collected by using the GC-ECD technique. An independent AIHA-accredited labora tory (National Loss Control [NATLSCO], Long Grove, D.) ana lyzed the second half of the samples by using a commonly used GC-FID technique. NATLSCO does not recommend using 3M VVV 000006376 'M-,NDHY<i-AssoaA(S2)/A|irtti 153 3550/3551 dosimeters with GC-FID analysis for EtO EL moni sampling location if the bias as calculated by the mean GC-FID toring. The passive dosimeters were shipped to the designated laboratories for analysis following the manufacturer's recom mendations on sample handling and preservation. result was found to be statistically significant. The Assay Technology EO CHEM CHIP dosimeters, ana lyzed the same day as they were collected, did not meet the 3 e The two remaining types of dosimeters were analyzed by the investigators ou-site. Half of each type were analyzed the day of collection, and the second half were stored and analyzed the next accuracy requirement at any of the four concentrations studio The 35% accuracy was. however, met by dosimeters at the 5.28-ppm exposure level for the monitors that were stored and day. This was done to determine what effect letting exposed dosimeters sit overnight, prior to analysis, would have on the analyzed the next day. The authors feel that the problem with the EO CHEM CHIP monitors may be directly related to the vari monitor's accuracy. Assay Technology's dosimeters were refrig erated according to the manufacturer's directions, and the Crys tal Diagnostic devices were stored at room temperature as recommended by the manufacturer. The techniques and proce dures recommended by the manufacturers were followed for developing and reading the dosimeters. ability in the blank's determination. When a series of blank monitors was analyzed, a CV of 0.15 was found. This variation affected the calculated precision and bias of the method when only one blank value was used to adjust the results for a whole box of samplers, as recommended by the manufacturer. In this study, accuracy was improved by using a mean blank correction derived from all blank values to adjust all sample values. Al though this technique reduced the total error of all pooled sam RESULTS AND DISCUSSION All the blank monitors and Tedlar bag/GC-FID samples collect ed in the study setup location were below each method's limit of detection except for one AirScan monitor, which displayed fullscale readings (>22 ppm). The problem with this monitor is discussed later. Table I lists the Tedlar bag/GC-FID sampling results and the ples by 50%, the total error of all pooled samples still did not meet the 35% accuracy requirement. In response to these data, the manufacturer is revising instructions for use and examining unit-to-unit variation as a function of shelf life. The Crystal Diagnostic AirScan dosimeters met the 35% accuracy requirement at the 2.12-ppm concentration level for monitors analyzed both the same day and the next day. However, the dosimeters did not meet the 35% accuracy requirement at sampling parameters obtained from the four sampling locations. The mean EtO concentration is presented along with the calcu lated 95% confidence intervals. Average location temperatures ranged from 28.5*C to 32.0*C, while average relative humidities ranged from 66% to 76%. The array of passive monitors was angled perpendicular to the room's airflow so that the room air could move freely between the array rows. Average location air velocities moving through the array of dosimeters were found to range from 8 to 28 fr/min when a hot wire anemometer was used. any higher concentration level for either same-day or next-day analysis. This is a very unique situation; generally, methods are more accurate as concentration increases. In the case of the AirScan monitors, the most accurate data were obtained at the lowest concentration studied. A problem occurred with the ' of these dosimeters. In approximately 10% of the AirScan simeters handled, the capacity of the monitor was exceeded (>22 ppm for 15 min). This occurred even with a blank, as was discussed earlier. It is not known if this problem was caused by A summary of the results of the three different passive a manufacturing problem, repeatedly shipping the dosimeters, or dosimeters at the four sampling locations is presented as Table improper activation of the monitors by the personnel performing II. The first column lists the manufacturer and device name. The the study. The possibility of EtO contamination during handling second column lists the analysis time or type of analysis. The and/or storage can be ruled out based on all the other blank data third column lists the mean EtO concentration determined by the available (other dosimeters and Tedlar bag/GC-FID data). Tedlar bag/GC-FID technique at each of the four sampling The 3M 3550/3551 dosimeters analyzedby 3M (GC-ECD) met locations. The fourth column lists the mean result obtained with the35% accuracy requirement at all fourconcentration levels. The each type ofsampling device (n=6) at each of the four sampling 3M dosimeters analyzed by NAILSCO (GC-FED) met the 35% locations. The fifth column lists the calculated coefficient of accuracy requirements at and above 9.41 ppm but failed to meet variation. Column 6 lists the calculated bias. The final column 35% below 9.41 ppm. This accuracy difference between labora lists the calculated method accuracy of each method at each tories was attributed to the sensitivity difference between die FID and TABLE I. Fifteen-Minute Average EtO Concentration, Temperature, Relative Humidity, and Air Velocity at the Four Sampling Locations ECD. An FID responds to carbonbased compounds and is less sensitive but more stable than an Location Number i 2 3 4 EtO by GC-FID (ppm) Mean* 95% Ct 2.12 5.28 9.41 1S.8 0.12 0.6 0.27 0.5 Temperature (C) 30 32 31 29 Relative Humidity 66 69 69 76 Air Velocity Mean6 SD 8.0 16.5 NAC 10.5 2.7 6.7 2.6 ECD detector. It is generally the detectorofchoice unless ultratrace analysis is required. The ECD re sponds specifically to compounds containing electronegative atoms, such as oxygen, sulfur, and halo gen atoms. The detector can be *Mean of six Tedlar* bag samples, each analyzed In quadruplicate. 16 measurements were collected and averaged. COatt unavailable because ot equipment malfunction. VVV G000G6377 overloaded very easily and m* be recalibrated often. When 1. died properly, however, it offers AM. NO. HYG. ASSOC. J. (52) / April 1991 154 taBUE II. Summary of Field Determination of Method Accuracy at (he Four Sampling Locations_______________________________________ Manufacturer/ Device Analysis Time or Type EtO (ppm) GC MeanA Method8 CV 4*cay Technology 0 CHEM CHIPTM Same day 2.12 5 28 9 41 15.8 0.83 36 3.1 8.9 0.87 0.27 0.36 0.11 io t. h> Bias -0.59 -0.23 -0 66 Method Accuracy + 233% 77 138 64 The results of this studs sug gest a significant bias error can be expected with both the EO CHEM CHIP and AirScan mon itors. EO CHEM CHIP results were, on the average, approxi mately 38% low; AirScan results were. or. the average, >36% high. On the average, a -8% bias error Next day 2.12 1.0 0.96 -0.50 246 was determined for the 3M 3550/ 5.28 5.3 0.09 0 18 3551 monitors when analyzed by 9.41 6.6 0.07 -0.27 41 either GC-ECD or GC-FID. Crystal Diagnostics AirScan Same day 15.8 2.12 528 9.41 10 1.9 7.5 >17.8C 0.05 0.09 0.16 0.14 -0.35 -0.05 + 0.28 > +0.84 45 23 60 >112 CONCLUSIONS The resuits of this field valida tion study suggest that the Assay 15.8 >20.7C Next day 2.12 5.28 9.41 15.8 1.9 7.2 >19.5C >21.4 3M 3560/3551TM 3M: GC-ECD 2.12 5.28 9.41 15.8 1.97 4.93 7.73 13.1 NATLSCO: GC-FID 2.12 5.2B 9.41 15.8 AMean of six Tedlar* bag samples, each analyzed in quadruplicate. BMean of six monitors. Cihe capacity of one or more of the dosimeters was exceeded, ^Coefficient of variation calculated from less than six values. 2.9 2.9 8.7 12.8 0.07 0.06 0.15 0.13d 0.06 0.16 0.06 0.04 0.06 0.59 0.55 0.10 0.03 > +0.27 -0.05 + 0.22 > +1.01 > +0.32 -0.02 0 -0.15 -0.15 4-0.29 -0.38 -0.05 -0.17 >41 17 52 >127 >44 34 12 23 27 147 148 25 23 Technology EO CHEM CHIP and the Crystal Diagnostics AirScan EtO dosimeters will not meet 35% accuracy require ments over the concentration range of 2.12 ppm to 15.8 ppm over the specific temperature, humidity, and location parame ters studied. The results also sug gest that the 3M 3550/3551 monitors analyzed by GC-ECD will meet the accuracy require ment over the 2.12 ppm to 15.8 ppm concentration range. The average bias over this range for the 3M 3550/3551 GC-ECD method was -8% with a pooled CV, of 9%, yielding an accuracy of 25%. The 3M 3550/3551 greater sensitivity than does die FID for compounds containing electronegative atoms. For example, when analyzing 2bromoethanol, the ECD is 55 times more sensitive than the FID.(151 When the 3M 3550/3551 dosimeters were field validated for PEL monitoring in a previous study* significantly different results were again generated for GC-FID and GC-ECD meth ods/61 A method accuracy of 30% was observed when the GC-ECD method was used; 7% method accuracy was esti mated when the GC-FID method was used. A comparison of data from the two studies shows that the GC-ECD technique has similar accuracy for both PEL and EL monitoring; the GC-FID method is superior for PEL monitoring but is inap propriate for EL monitoring. Additional information can be gained from the data in Ta ble II. The data suggest that both the precision and bias of the EO CHEM CHIP monitors improve when the monitors are stored for 1 day prior to analysis. One-day storage appears to have slight to little effect on the overall accuracy of the AirScan monitors. The 3M 3550/3551 data suggest that the accuracy difference between the two methods used to analyze the monitors was more ofa precision difference and not a bias difference. This again points to the sensitivity of the detectors used. monitors analyzed by GC-FID will not meet the accuracy re quirement over the 2.12 ppm to 15.8 ppm concentration range. The reason for this difference was attributed to the sensitivity difference between the two different detectors. It is strongly recommended that field validation be per formed by using at least two independent methods prior to selecting a method for routine monitoring of employee expo sures to EtO. Also, annual dual monitoring that uses two inde pendent methods to document that the selected monitoring method is working properly is recommended. The strength of this validation study and subsequent conclu sions may have been enhanced by the inclusion of a previously validated active pump method/4> Unfortunately, the complexity of the study, demands of many carefully timed steps, and size limitations of the field exposure chamber made the inclusion of the active pump method infeasible. VVV 000006378 ACKNOWLEDGMENT The authors acknowledge and thank Marilyn Alicea and Jose Vadal Ramos for helping during die sample collection portion of this study. J55 AM. MD. HYG. ASSOC. J. (52) / Aprl 1991 REFERENCES 1 Occupational Safety and Health Administration: "Occupational Exposure to Ethylene Oxide" {29 CFR 1910). Federal Register 49:122 (22 June 1984). pp. 25734--25809. Washington, D.C.: L'.S. Government Printing Otflce, 1984. 2. Golberg, I,.: Ha?.ard Assessment rf Ethylene Oxide. Boca Raton. Fla.: CRC Press Inc., 1986. 3. National Institute for Occupational Safety and Health: Current Intelligence Bulletin 35. Ethylene Oxide (EtO) (DHHS/NIOSH Publication No. 81*130). Cincinnati, Ohio: U.S. Government Print ing Office, May 22,1981. 4. Puskar, M.A., J.L. Nowak, and L.H. Hecker: Laboratory and Field Validation of a JXC Charcoal Sampling and Analytical Meth od for Monitoring Short-Term Exposures to Ethylene Oxide. Am. Ind. Hyg. Assoc. J. 49(5):237-243 (1988). 5. Puskar, M.A., F.G. Szopinski, and L.H. Hecker: Development and Validation of a Protocol for Field Validation ofPassive Dosim eters for Ethylene Oxide Excursion Limit Monitoring. Am. Ind. Hyg. Assoc. J. 52(4): 145-150 (1991). 6. Puskar, M.A. and L.H. Hecker: Field Validation of Passive Do simeters for the Determination ofEmployee Exposures to Ethylene Oxide in Hospital Product Sterilization Facilities. Am. Ind. Hyg. Assoc. J. 50(/):3O-36 (1989). 7. Medical Products: Technical Report MPD-TR-1984-9: EO Self Scan Personnel Monitoring System: Precision andAccuracy. Erie, Pa.: Medical Products, 1984. 8. Puskar, MX JX. Nowak, and L.H. Hecker: Generation of Ethylene Oxide Permissible Exposure Limit Data with Onsite Sample Analysis Using the EO-SELF-SCAN Passive Monitor. Am. Ind. Hyg. Assoc. J. 5i{5):273-279 (1990). 9. TVemblay, JuUe: "Principle of Operation: Crystal Diagnostics AirScan." May 1989. [Personal Communication). Julie TVemblay, Crystal Diagnostics, Inc.. 30 Commerce Way, Woburn, MAO 1801. 10. Association for the Advancement of Medical Instrumentation: Selecting Airborne Ethylene Oxide Monitoring Equipment of Ser vices for an EO Gas Sterilization Facility. AAMI Technology Information, Report No. 1. Arlington, Va.: Association 1984. 11. "Occupational Exposure to Ethylene Oxide" (29 CFR 1910). Fed eral Register 53:66 (6 April 1988). pp. 11414--11438. 12. Busch, K.A. and D.G. Taylor: Statistical Protocolfor the NIOSH Validation Tests. Chapter 31, ACS Symposium Series, No. 149, Chemical Hazards in the Workplace--Measurement and Control, GangadharChoudhary, Ed. Washington, D.C.: American Chemical Society, 1980. pp. 504-517. 13. National Institute for Occupational Safety and Health, Division ofPhysical Sciences and Engineering: Development and Valida tion of Methods for Sampling and Analysis of Workplace Toxic Substances. (DHHS/NIOSH Pub. No. 80-133). Cincinnati. Ohio: National Institute for Occupational Safety and Health, 1980. 14. Miller, J.C. and J.N. Miller: Statisticsfor Analytical Chemistry. 2d ed. Chichester, Great Britain: Ellis Horwood Limited, 1988. 15. Wronski, Joan: "Sensitivity Difference between FID and ECD for the Analysis ofEtO." November 1989. [Personal Communication]. Joan Wronski, NATLSCO, Route 22, Long Grove, IL 60049-0075. Ed. Note: A prepublication copy ofthis manuscript. "Field Validation ofThree Passive Dosimetersfor Excursion Limit Monitoring ofEthyl ene Oxide," was forwarded to the manufacturers for review and comment. Their responsesfollow. Dear Sir: Thank you for the opportunity to comment on "Field Valida tion of Three Passive Dosimeters for Excursion Limit Mon itoring of Ethylene Oxide" by F. Szopinski, M. Puskar, and L. Hecker. The study was well-designed, well-executed, and well, presented. The authors" conclusions appear valid for this study. Sincerely, Donald J. Larsen, CIH 3M Occupational Health and Environmental Safety Division Dear Sir: Re: Comment on "Field Evaluation ofThree Passive Dosim eters for Excursion Limit Monitoring of Ethylene Oxide" The authors did an excellent job in controlling many variables in their extensive field validation. They even ac knowledged that the erroneous off-scale AirScan readings may have been related to circumstances unknown to the manufacturer. However, I am not entirely in agreement with their treatment of the data; absolute biases for test methods should not be estimated when the "true" concentration, as determined by only one reference method, is not verified. According to Ms. Mary Ann Cassinelli et al., authors of the National Institute for Occupational Safety and Health (NIOSH) "Protocol for the Evaluation of Passive Monitors," and Dr. Richard Brown et al,, authors ofthe Health and Safety Executive "A Proposal for a Harmonized Approach to the Evaluation of Diffusive Samplers for Occupational Hygiene Measurement," at least two independent sampling/analytica. methods should be used in a field method validation to accurately determine true concentrations. A match of 10% or less between the results of two independent methods indi cates that the two methods are accurately measuring the true concentration. Calculation of test method biases assume that the true concentration is known. As described in the paper, the con fidence intervals for the gas chromatography estimates of true concentrations take into account random sampling/analytical error. Since the authors did not use a second indepen dent method, such as a validated active pump method, to verify that systematic error was not a factor, there is a possi bility that the confidence intervals do not include the true concentration. Ifso, some or all test method biases estimated by the authors may be artifacts of the study design. The contamination-related off-scale readings, noted by the authors, were causedby the hydrating solution contacting the lower pad in the diffusion port of the AirScan Monitor. As the authors suggest, the contact was probably caused by vigorous shaking during hydration and shipping the AirScan from place to place. Shortly after the authors reported the erroneous off-scale readings to the manufacturer. Crystal Diagnostics redesigned the diffusion port to prevent errone ous off-scale readings. I appreciate the authors' acknowledgment that a second independent method may have improved the study design. However, I am surprised that they evaluated the data as ifthe AM. MO. HY& ASSOC. J. (52) I Aprt* 1991 156 VVV 000006379