Document Z8N4Z2r7YNXa024B1R5b8VaJV

FILE NAME: Doubt Science (DBTS) DATE: 2008 DOC#: DBTS018 DOCUMENT DESCRIPTION: Journal Article - Airborne Concentrations of Asbestos Onboard Maritime Shipping Vessels (1978-1992) Ann Occup H \ g , Vol 52, No 4, pp 2 6 7-27^2008 The Authoi 2008 Published by Oxford University Press on behalf o f the British Occupational Hygiene Society doi 10 l093/annhyg/menOI5 Airborne Concentrations of Asbestos Onboard Maritime Shipping Vessels (1978-1992) DANA M. MURBACH1*, AMY K. MADL1, KEN M. UNICE2, JEFFREY S. KNUTSEN3, PAMELA S. CHAPMAN1, JAY L. BROWN4 and DENNIS J. PAUSTENBACH1 l Chem Risk, 25 Jessie Street, Suite 1800, San Francisco, CA 94105, USA; 2ChemRisk, 401 Liberty Avenue, Suite 1652, Pittsburgh, PA 15222, USA; 3ChemRisk, 4990 Pearl East Circle, Suite 300, Boulder, CO 80301, USA 4Present address: SafeX, Inc., 140 N Otterbein Avenue Westerville, OH 43081, USA * Received 14 August 2007; in final form 9 March 2008; published online 8 May 2008 The exposure of shipyard workers to asbestos' has been frequently investigated during the in . stallation, repair or removal of asbestos insulation. The same level of attention, however, has not been directed to asbestos exposure of maritime seamen or sailors. In this paper, we assem ble and analyze historical industrial hygiene (IH) data quantifying airborne asbestos concen trations onboard maritime shipping vessels between 1978 and 1992. Air monitoring and bulk sampling data were compiled from 52 IH surveys conducted on 84 different vessels, including oil tankers and cargo vessels, that were docked and/or at sea, but these were not collected dur- ing times when there was interaction with asbestos-containing materials (ACMs). One thou sand and eighteen area air samples, 20 personal air samples and 24 air samples of unknown origin were analyzed by phase contrast microscopy (PCM); 19 area samples and six samples of unknown origin were analyzed by transmission electron microscopy (TEM) and 13 area air samples were analyzed by scanning electron microscopy (SEM). In addition, 482 bulk sam ples were collected from suspected ACMs, including insulation, ceiling panels, floor tiles, valve packing and gaskets. Fifty-three percent of all PCM and 4% of all TEM samples were above their respective detection limits. The average airborne concentration for the PCM area samples (n = 1018) was 0.008 fibers per cubic centimeter (f cc-1 ) (95th percentile of 0.040 f cc '). Air concentrations in the living and recreational areas of the vessels (e.g. crew quarters, common rooms) averaged 0.004 f cc-1 (95th percentile of 0.014 f cc-1), while air concentrations in the engine rooms and machine shops averaged 0.010 f cc-1 (95th percentile of 0.068 f cc b. Air borne asbestos concentrations were also classified by vessel type (cargo, tanker or Great Lakes), transport status (docked or underway on active voyage) and confirmed presence of ACM. Approximately 1.3 and 0% of the 1018 area samples analyzed by PCM exceeded 0.1 and 1 f e e -1, respectively. This data set indicates that historic airborne asbestos concentrations on these maritime shipping vessels, when insulation-handling activities were not actively being performed, were consistently below contemporaneous US occupational standards from 1978 until 1992, and nearly always below the current permissible exposure limit of 0.1 f cc . Keywords: asbestos; industrial hygiene; maritime; occupational'exposure; retrospective exposure assessment; seamen INTRODUCTION Due to its widespread use in maritime applications, asbestos has warranted study in terms of occupa tional ex p o su r e and the p o ss ib le c o n n e c tio n to r e sp i *Author to whom correspondence should be addressed! Tel: +415-896-2400; tax- +415-896-2444; e-mail: dmurbach@chemrisk.com ratory disease in sailors and shipyard workers iMarr, 1964; Nicholson et a t, 1971; Sheers and Coles, 1980; Hodgson and Darnton, 2000). From the early 1930s to the mid to late 1970s, naval and commercial shipyards used hundreds of tons of asbestos, primar ily in the form of chrysotile and amosite asbestoscontaining insulation, to build and repair maritime vessels. War ships, for example, contained roughly 267 268 D. M. Muibach et al. between 30 and 500 tons of asbestos insulation on pipes and machinery (United States Navy, -1945, 1947; Fleischer et a i, 1946; Mangold et al., 1970; C. A. Mangold, personal communication). This min eral was an effective insulator that prevented conciensation, reduced the amount of ventilation needed to cool compartment spaces, allowed machinery to op erate economically with minimal heat loss and pre sented injury to personnel from contact with hot shipboard components (United States Navy, 1939, 1941, 1945, 1947, 1960). While there are a relatively large number of studies that have assessed airborne asbestos exposure among shipyard workers involved in installing, repairing or removing asbestos insulation, our review of the pub lished literature found no studies that have character ized the exposure of maritime or merchant seamen to airborne asbestos (Balzer and Cooper, 1968; Cooper and Balzer, 1968; Harries, 1968, 1971a,h; Ferris et al., 1971; Bell, 1976). A few studies, however, have reported an increased incidence of asbestosrelated disease among this seafaring population (Jones etal., 1984; Selilcoff et al., 1990; Varouchakis et al., 1991; Pukkala and Saarni, 1996; Saarni et al., 2002; Rafnsson and Sulem, 2003). It has been postu lated that the increased incidence of asbestos-related disease among this population was, in part, due to the release of fibers from asbestos insulation occurring as a result of the natural movement and vibrations of a ship. No published studies, to the authors' knowl edge, however, have measured such asbestos ex posures to seamen onboard different maritime shipping vessels. Our primary objective, then, was to analyze the in dustrial hygiene (IH) data for airborne asbestos con centrations, collected on maritime vessels but not during the direct handling of insulation and other asbestos-containing materials (ACMs). The purpose of this analysis was to characterize the `background' exposure to asbestos of seamen from 1978 until 1992 and to evaluate whether the mere presence of asbestos on these ships constituted a health hazard. To that end, IH measurements generally associated with normal transport and docking scenarios of cargo vessels, based on 52 IH surveys (encompassing 1100 air and 482 bulk samples collected from the late 1970s through the early 1990s), are presented. Fac tors such as year sampled, vessel type, compartment and transport status were also considered. The air borne asbestos concentrations measured were com pared to contemporaneous and current occupational permissible exposure limits (PELs). BACKGROUND , Since the 1930s, asbestos insulation products have been used and applied according to naval and mari time commission specifications (Rushworth, 2005). These specifications vary considerably, depending on anticipated temperature ranges for machinery, as well as the temperature ranges and thickness of pip ing being insulated. The selection of asbestos for specific applications also depended upon other con siderations, including water resistance, availability and cost. Although the application processes for in sulation products have changed little since the 1930s, their composition has changed over time (United States Navy, 1939, 1941, 1945, 1947, 1960). The ma jority of asbestos-containing products, such as ther mal insulation, gaskets and packing, for example, were beginning to* be replaced in the mid to late 1970s with asbestos-free materials exhibiting similar heat-resistant properties. Studies conducted both in the USA and in Europe during the 1960s and 1970s reported that shipyard workers were exposed to elevated concentiations of airborne asbestos (Harries, 1968; Mangold et al., 1970; Ferris et a l, 1971; Nicholson et al., 1971; NIOSH 1972). These studies described the work of shipyard insulators during construction and repair of marine vessels, and reported airborne fiber con centrations ranging from 1.1 to 100 f cc~' for cutting and applying insulation and 2.4 to 132 f cc 1for re moving insulation. Short-term peak exposures during some insulating activities have reportedly exceeded 100 f cc_ l, usually during more intense activities, such as mixing or spraying (Balzer and Cooper, 1968; Harries, 1968; Mangold et a l, 1968, 1970; Harries, 197la,b; Nicholson et al., 1971; Selilcoff and Hammond, 1978; Selilcoff et a l, 1979a,b; EPA 1986). Installing and removing asbestos insulation was the source of the high concentrations of airborne asbestos found during major ship overhauls or ma chinery ripouts. To a much lesser degree, there can be exposure duringmccidental ruptures of pipes, fail ure of flange systems or pin holes in steam lines, be cause the insulation can degrade and fall to the floor but because, at that point, it is wet, the airborne con centrations are rarely appreciable. Studies of merchant seamen have reported increased incidence and mortality rates due to lung cancer, meso thelioma, and other pleural abnormalities especially among deck and engine crew and officers (Jones et al, 1984; Rafnsson et al, 1988; Velonakis et a l, 1989; Kelman and Kavaler, 1990; Selikoff et al, 1990; Greenberg, 1991; Rapid et al, 1992; Brandt et al., 1994; Rafnsson and Gunnarsdottir, 1995; Pukkala and Saarni, 1996; Saarni et a l, 2002; Karlev et al, 2005). Many ofthese studies suggest that such high cancer rates were possibly due to previous asbestos exposure and/or constant exposure to low levels of airborne asbestos fi bers from damaged or deteriorated asbestos containing insulation released due to the natural vibrations of the ship or from insulation repairs conducted during travel at sea, including ruptures, failures or blowouts on steam Airborne asbestos concentrations 269 piping. These assumptions, however, have not been quantitatively evaluated (Greenberg, 1991). Although it is clear that absent proper engineering controls asbestos exposure while installing and re pairing asbestos-containing insulation can be signif icant, one would expect that the concentration of airborne asbestos fibers released from undisturbed insulation would be very small; however, informa tion on these exposures has not heretofore been avail able. Based on the dozens of studies of school buildings which took place in the USA in the 1970s and 1980s, it was determined that airborne concentrations of asbestos from undisturbed insula tion are quite low (Mossman et al., 1990). Many in sulated surfaces, in addition to the lagging, have had. coatings (mastics) and paint layers applied to the sur face and this further reduces the possibility that as bestos fibers will be released from intact insulation.* Thus, intuitively, one would not expect appreciable airborne concentrations on marine vessels when insu lation is not being disturbed or removed, but because of speculation that ship movement might produce some stresses on the insulation, a number of studies have been conducted over the years. Our analysis of these previously unpublished studies presents the only data, to our knowledge, of the concentra tions of airborne asbestos not associated with the di rect handling of ACMs onboard maritime shipping vessels. ' , MATERIALS AND METHODS The data presented in this analysis were compiled from 52 IH surveys, collected by at least 15 different consulting firms hired by ship vessel owners. Based on the available information, it appears that these data were collected during those time periods when OSHA proposed to significantly lower the occupa tional exposure limit for asbestos. The data were ob- tained from studies or surveys conducted onboard 42 vessels in which only air samples were collected, 40 vessels in which both air and bulk samples were . collected and 2 vessels in which only bulk samples were collected. These surveys provided phase con trast microscopy (PCM) results for 63 vessels, trans mission electron microscopy (TEM) air results for 6 vessels, both PCM and TEM air results for 7 ves sels and both PCM and scanning electron microscopy (SEM) air results for 6 vessels. Of the 52 IH surveys, 26 were conducted by or under the supervision of certified industrial hygienists (CIHs), 4 were noted as being conducted by industrial hygienists and it was unclear whether the remaining 22 were ^onducted by CIHs. It was noted in 27 of the 52 surveys that the air samples were sent to an AIHA-, NIOSHor NVLAP-accredited or -certified laboratory for analysis. Since these surveys were intended to satisfy regulatory expectations, we believe it is reasonable to infer that all samples were assayed by a qualified, if not certified, laboratory. The air and bulk monitoring surveys were con ducted on a variety of merchant vessels, including self-unloaders, dry and liquid cargo vessels, con tainer ships, roll-on/roll-off vessels and a cement barge. Based on structural and functional similarities, these vessels were classified into four categories: dry cargo, tanker, Great Lakes and other types of vessels. Vessels that carried dry cargo, including roll-on/rolloff ships (those carrying wheeled cargo), and con tainers were classified as dry cargo vessels. Vessels that transported liquid cargo were classified as tank ers. The cement barge and self-unloaders were gener ally older vessels, primarily operating in the Great Lakes, and thus were classified as Great Lakes ves sels. Vessels classified in the other category included a vessel that installed marine cables and two vessels that were only identified as US flag ships. Air sampling data, including airborne fiber con centration, duration, analytical method, date, loca tion, type (personal versus area) and sample number, were entered in a database. Accompanying information, such as vessel name, status (dockside versus underway), owner, type, year built, modifica tion date, propulsion type and surveyor information, was also included. For buik sample data, similar in formation was recorded, including the type, condi tion and, in some cases, the percentage of asbestos in the bulk material. The bulk and air samples were collected from various areas of the ships, including the boiler and engine rooms, recreation areas and sleeping quarters. Bulk sample data A total of 482 bulk samples were collected from suspected asbestos-containing products and insula tion, including lagging, valve packing, gaskets, boiler and turbine covering materials, ceiling panels and floor tiles on 42 merchant vessels. The results were cited in the IH reports either quantitatively as a percent of asbestos fiber concentration in the bulk material or qualitatively by confirming the presence or absence of asbestos in the material sampled. Of the 482 bulk samples collected, asbestos fiber con centrations were reported for 287 and the remaining 195 sample results were reported qualitatively. It was generally noted that bulk samples were analyzed by polarized light microscopy (PLM), according to ei ther EPA or NIOSH bulk sampling methodology (EPA, 1982; NIOSH, 1994a). The IH reports noted that based on visual observations and the experience of the industrial hygienist, bulk samples were col lected in areas where materials were expected to be comprised of asbestos, including areas where dam age to the insulating material was visible or in areas of the ship occupied by personnel for work, eating, 270 D. M. Murbach et al. recreating and sleeping. It was also noted that insula tion and other materials containing asbestos were reasonably maintained and in good condition, and re moval or similar tasks were generally infrequent and limited in duration while at sea. Thus, the samples collected were considered to represent the back ground concentration of airborne asbestos on these ships. Air sample data The air samples evaluated in this analysis were not associated with particular work activities onboard ships and were not collected when seamen were ac tively disturbing, repairing or handling ACMs. Of the 1062 PCM samples, 20 were classified as per sonal samples, 1018 were classified as area samples and 24 were of unknown type. The influences of sampling-specific variables such as type of ship (cargo, tanker or Great Lakes), sampling location (crew or engine spaces) and transport status of the ship (docked or underway) on the airborne concen trations of asbestos were evaluated. . Sampling times, when reported, varied between 4 and 971 min. While PCM was reportedly used for the analysis of most of the air samples, the analytical and sampling procedures changed over time. From 1978 until 1986, air samples were analyzed using NIOSH P&CAM Method 239 (NIOSH, 1977). fee- tween 1985 and 1992, samples were analyzed ac cording to NIOSH Method 7400 (NIOSH, 1994b). Between 1990 and 1992, 25 samples were analyzed by TEM, according to NIOSH Method 7402 and be tween 1985 and 1987, 16 samples were subsequently analyzed by SEM (NIOSH, 1994c). During the initial review of the IH reports, 28 air samples analyzed by PCM were excluded from the analysis due to dust in terference and the failure to meet NIOSH analytical requirements regarding loading (i.e. too much nui sance dust was in the air in that specific environment for the air sampling rate). In total, 499 PCM samples were reported as below the analytical detection limit, non-detect (ND) or as 0 f cc 1 (almost 50% of all samples collected). Twenty-three samples were excluded because of in sufficient sample-specific information on volume, duration and analytical method to accurately calcu late or estimate a detection limit. When this sample information was available and detection limits were not reported, they were ad hoc calculated by the`au thors, according to the P&CAM 239 method, for 20 samples (collected in 1984) and ranged from 0.014 to 0.194 f cc 1 (NIOSH, 1977). Detection limits for the remaining 456 ND samples were identified as less than a value (e.g. <0.001 f cc" ') in the sample result data or identified in the text of the IH report and ranged from 0.0003 to 0.1 f cc" 1. It should be noted that the NIOSH P&CAM 239 method only re ported a lower limit of quantitation (LOQ) of 0.1 f cc 1 (at 2.5 l.p.m., for 2 h, for a field area of 0.003 mm2), while NIOSH Method 7400 reported both LOQs and limits of detection (LODs) for vari ous values of filter fiber density and total volume of air sampled (NIOSH, 1977, 1994b). Air data analysis The air sampling data were highly censored, with >45% of the 1062 PCM samples below their respec tive detection limits, and the results were not found to be normally or log-normally distributed using the Lilliefors test (P > 0.05). Due to the relatively high percentage of ND samples, non-parametric methods accommodating left-censored data were selected for this analysis. Except where otherwise noted, the Kaplan-Meier (KM) estimation method (also known as the product limit estimate) was used to estimate the mean concentrations. This method is useful for analyzing left-censored data sets with multiple detec tion limits and is not based on an underlying distribu tion of the data set (Helsel, 2005). The 95% upper confidence limits of the mean reported in this analy sis were based on the Chebyshev inequality using the KM-estimated mean and standard deviation (SD). Additionally, the Gehan test was used to assess whether concentrations were significantly different be tween specific sample groups as it is a non-p;irametric test for use on data sets with multiple detection limits (EPA, 2007a). ProUCL 4.0, statistical software de veloped and distributed by the US EPA, includes built-in algorithms for the above methods, and was thus used for these analyses (EPA, 2007b). The impact of having a high fraction of ND sam ples on th mean and SD results was also evaluated. To this end, sample means were calculated using var ious substitution methods. Using these approaches, surrogate concentrations were substituted for sam ples with concentrations below their respective de tection limits. A number of different substituted concentrations were assessed in this analysis, includ ing zero, LOD/2, LODA/2 and the LOD (Perkins et a l, 1990; Finkelstein and Verma, 2001). As noted by Rao et al. (1991), the sample mean is bounded by the value calculated after substituting ND concentra tions of zero, and by the value calculated after substituting ND concentrations of the full LOD. Spe cifically, the latter method (substituting the full LOD) results in a sample mean that is biased high, and will thus be larger than the true sample mean. Using this approach, the impact of the degree of censoring on statistics, such as the mean and SD, can be assessed. Airborne fiber concentrations were compared to contemporaneous PELs (8-h time weighted average) for asbestos. The fractions of samples exceeding spe cific concentrations, including 0.1 f cc" 1 ie.g. the current OSHA PEL), were also calculated using Airborne asbestos concentrations 271 Table 1. Quantitative bulk sample results by source and percent concentration of asbestos type from 34 vessels Source . n d % chrysotile % amosite % crocidolite Min Max Min Max Min Max Ceiling panels 9 8 30 75 3 8 Debris" 8 0 Floor tile 5 1 3 8 Gasket1* 2 2 90 95 Insulation" 245 131 1 100 1 61 3 10 Valve packing 1 1 70 . 75 Unspecified 13 10 12 60 Total 287d 153 -1--------- n, number of samples; d, number of samples with detectable concentrations. "Samples collected in the vicinity o f an area air sampling location or underneath machinery. ^Samples collected from an unspecified location. "Samples collected from various pipe, equipment and machinefy insulation materials. ... . dFour additional bulk samples were obtained from a pump room brake lining, an etecttical outlet wrap, toam ceiling insulation and a refractory material and were negative for asbestos content. a non-parametric method (Mulhausen and Damiano, 1998). Although not as statistically powerful as para metric methods, this method is useful because it is not affected by detection limits and is based only on the total sample size and the number of data ex ceeding the specified concentration. RESULTS A total of 1100 air and 482 bulk samples collected from 84 merchant shipping vessels from 1978 until 1992 were considered in this analysis. Approxi mately 43% of the area air samples analyzed by PCM (n = 1018) and 96% of the air samples (n = 25) analyzed by TEM were below their respective LODs. * Bulk samples Twenty-eight of the 42 ships sampled were found to contain asbestos-containing insulation, while the ` remaining 14 ships had bulk insulation sample results either below the LOD for PLM or did not contain asbestos. The asbestos bulk sample results were re ported in the IH surveys either quantitatively (287 bulk samples collected from 34 vessels from 1981 until 1992) or qualitatively (195 samples collected from eight vessels from 1984 until 1989). Of the total 482 bulk samples, 392,were collected from insulation materials, including pipe and machinery lagging, cloth and blankets, while the remaining 90 samples were collected from other potential ACMs, such as ceiling and floor tiles, brakes, clutches, valve packing and gaskets. Asbestos was detected in 232 of the 392 bulk samples collected from pipe and equipment insulation materials. Of the 287 quantitative bulk sample results, 153 samples contained detectable levels of one or more of the three types of asbestos (amosite, chrysotile or crocidolite) (Table 1). One hundred and forty sam ples contained detectable levels of chrysotile at con centrations ranging from 1 to 100%, which were found in ceiling panels, floor tiles, gaskets, insula tion, packing and samples of unknown origin. Two hundred and forty-five samples were collected from insulation materials and 72 of these samples con tained detectable levels of amosite ranging from 1 to 61%. Crocidolite was detected in three samples of insulation (one insulation jacket on a DC heater and two on an unspecified insulation material in the engine control area) and one ceiling panel at concen trations of 3-10%; amosite and/or chrysotile were also present in these samples (Table 1). Crocidolite sometimes appears as a contaminant in amosite used in the older asbestos ship insulations. In addition to asbestos, other types of materials were identified in the bulk material, such as fiberglass, glass wool, min eral wool, calcite, gypsum, quartz, diatoms, ceramic fiber, magnesium silicate and pumice. Air samples Area air samples (n = 1018) collected onboard 74 different merchant vessels and analyzed by PCM are summarized in Table 2. The average concentration for airborne fibers for all merchant vessels, calcu lated using the KM method, was 0.008 f cc" 1 (SD = 0.039 f cc~1; 95th percentile of the data set = 0.040 f cc" 1). When concentrations of zero or the LOD were substituted for ND sample results, average air concentrations were 0.007 and 0.011 f cc , re spectively; that is, there was virtually no difference in average air concentrations, irrespective of the value substituted for ND sample results. Figure 1 presents the average air concentrations over time, as calculated by various substitution methods, in cluding the KM method. The KM average for each year falls between the lower (ND = 0) and upper (ND = LOD) estimates of the sample mean using Table 2. Summary o f airborne fiber concentrations for the area (rather than personal) samples collected onboard 74 shipping vessels from 1978 to 1992 analyzed by PCM (f cc ) D. M. Murbach et al. din All vessels (1978-1992) 1978 1980 1981 1983 1984 1985 1986 1987 1989 1991 1992 Dry cargo Great Lakes * Other Crew area Engine area Unknown Docked and underway Loading and/or unloading NIS or on standby Underway Unknown Ships confirmed to contain ACMse Ships not confirmed to contain ACMse Ships not tested 578/1018 53/53 36/88 11/75 6/7 21/51 33/48 48/85 46/68 14/105 245/339 65/99 152/327 217/373 155/2.60 54/58 232/295 281/598 55/104 10/21 102/166 1/6 19/35 13/29 374/609 69/173 206/385 100/203 272/430 (ND = 0) 0.007 0.049 0.028 0.003 0.014 0.002 0.003 0.006 0.003 0.001 0.003 0.002 0.007 0.004 0.004 0.048 0.004 0.010 0.004 0.007 0.009 0.0002 0.006 0.007 0.004 0.017 0.002 0.003 0.014 Mean (ND = LOD/V2) 0.010 0.049 ' 0.030 0,(X)5 0.015 0.029 0.003 0.007 0.005 0.007 0.003 0.002 0.008 - 0.006 0.010 0.048 0.004 0.013 0.008 0.010 0.012 0.002 0.007 0 007 0.006 0.025 0.007 0.004 0.015 (ND = LOD) 0.011 0.049 0.030 D.005 0.016 0.040 0.003 0.008 0.005 0.009 0.003 0.002 0.008 0.006 0.012 0.048 0.004 0.014 0.010 0.012 0.013 0.002 0.007 0.007 0.006 ' 0.029 0.009 0.004 0.016 Meand 0.008 0.049 0.031 0.006 0.016 0.003 0.003 0.006 0.004 0.002 0.003 0.002 0.008 0.004 0.004 0.048 0.004 0.010 0.005 0.008 0 010 d 0.006 0.007 0 005 0.018 0.003 0 003 0.015 SD" 0.039 0.069 0.114 0.008 0.005 0.004 0.005 0.008 0.003 0.003 0.005 0.002 0.031 0.010 (.006 0.139 0.009 0.051 0.010 0.010 0.034 (J 0.017 0.010 0.015 0.084 0.004 0.006 0.060 U C Ljs" 0.013 0.090 0.085 0.008 0.005 0.006 0.005 0.008 0.005 0.003 0.004 0.003 0.015 0.007 0J106 0.128 0.006 0.019 0.010 0.019 0.021 0.019 0 016 0.008 0.046 0.004 ___________ Percentile___________ 25th 50th 75th 95th 0.001 0.010 0.002 0.002 0.010 0.001 0.001 0.001 0.002 0.002 0.001 0.001 0.002 0.019 0.010 0.004 0.020 0.005 0.002 0.007 0.004 0.003 0.002 0.001 0.007 0.060 0.020 0.004 0.020 0.093 0.003 0.010 0.010 0.006 0.003 0.002 0.040 0.190 0.130 0.022 0.0(X) 0.194 0.013 0.028 0.010 0.040 0.010 0.005 0.001 . 0.001 0.001 0.008 0.002 0.002 0.003 0.013 0.004 0.005 0.010 0.030 0.030 0.024 0.091 0.410 0.001 0.001 0.001 0.002 0.002 0.002 0.004 0.008 0.005 0.009 0.010 0.022 0.014 0.068 0.050 0.040 0.002 0.001 0.001 0.001 0.001 0.002 0.003 0.002 0.002 0.001 0.002 0.004 0.008 0.007 0.006 0.019 0.005 0.020 0.058 0.0(X) 0.100 0.032 0.018 0.150 0.001 0.002 0.004 0.040 0 005 0 002 0.003 0.004 0.013 0.027 0.002 0.003 0.010 0.059 Max 1.000 0.390 1.000 0.059 0.020 0.137 0.030 0.032 0.020 0.071 0.068 0.016 0.390 0.117 0.137 1.000 0.124 1.000 0.086 0.037 0.390 0.005 0.100 0.032 0.190 1.000 0.137 0.068 1.000 >PEL (%)c 1.4 17.0 5.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2.1 0.5 0.0 8.6 0.3 2.2 0.0 0.0 1.8 0.0 2.9 0.0 0.8 2.9 0.0 No. Df studies 50 3 4 4 i 5 1 4 1 6 17 6 14 23 .1 0 3 33 48 *28 7 24 2 7 2 40 13 23 0.0 18 3.3 11 d number of samples with detectable concentrations, n, number of samples; UCL95, 95% upper confidence limit of the population mean, NIS, not in service. . ,. 'Based on the Chebyshev inequality using KM estimates for the mean and SD except 1978 (based on the Chebyshev inequality using log-normal estimates) and 1985 (based on the Chebyshev inequality using non-parametric estimates). . bPercentile calculations performed using the empirical CDF method, using SYSTAT statistical software version 11. `'Fraction o f samples exceeding the current OSHA PEL of 0.1 1 cc . `'Single detect sample with concentration o f 0.001 f cc 1. KM cannot be used to estimate mean, SD oi UCL. "Confirmed presence o f asbestos-containing insulation materials by bulk fiber analyses. Airborne asbestos concentrations 273 10 q O SHA P E L (8-hrTWA) - 1976 (2 f/cc) 1 r Maximum 4-,-*--Average (ND = LOD) Average (ND = LOD/\2) Average (KM) Average (ND = 0) (Number of detects/number of samples) 5oz DC 0 1 z I- Z LU o (53/53) oz o 0 01 LzU- occ CQ CC < 0 001 - * (36/88) (11/75) 1986 (0 2 f/CC) 1994 (0.1 f/cc) & (6/7) A 6 (48/85) 0 (33/48) (46/68) (21/51) (14/105) (245/339) A (65/99) 0 0001 -- i------------1--------------- 1------------- 1------------- 1------------- 1------------- 1------------- 1------------- 1------------- 1------------- 1------------- 1------------- 1------------- 1------------- 1------------- 1 1978 1979' 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 '1991 1992 1993 YEAR i 1994 Fig. 1. Average airborne fiber concentrations for the area samples estimated using various substitution methods and the KM method. Vertical error bars represent the maximum detected sample concentration for each year. substitution methods. As also shown in Fig. 1, com parisons between these air sample results and the oc cupational standards indicate that no samples were above the contemporaneous OSHA PELs. Moreover, upper-bound estimates of the sample mean (ND = LOD) were all well below the contemporaneous PEL for all years. , As shown in Table 3, 98.7% of the results were <0.1 f cc" 1. In addition, a non-parametric upper tol erance limit estimate suggested that, based on th is, data set, there was 95% confidence that ~ 2 .1% of rel evant historical asbestos air concentrations exceeded 0.1 f cc-1. For comparative purposes, in Table 3, ex ceedance fractions and upper tolerance limits on these fractions are also shown for the concentration of 1 f cc" 1. Exceedance fractions generally de creased with time. When the 877 samples from 1981 and later are considered, for example, there is 95% confidence that >99% of the asbestos air con centrations were <0.1 f cc" 1. Based on our experi ence and published studies, in most industries and settings, asbestos air concentrations fell dramatically after 1981 because of stricter regulatory limits and increased awareness about the precautions that were necessary to decrease exposures and not because of the absence of ACMs since 275 bulk samples col lected on ships after 1981 were positive for asbestos. The average concentration of the 20 personal sam ples was 0.021 f cc" 1 (SD = 0.024 f cc" 1; 95th per centile of the data set = 0.11 f cc" 1), which was significantly higher than average area concentrations t Table 3. Non-parametric exceedance fraction estimates and upper tolerance limits for various data sets Concentration ( fe e -1) 0.1 1 n 1018 1018 ma EFb (%) 13 1.3 0 0.0 UTLC (%) 2.1 0.7 "The number o f samples with concentrations exceeding the specified concentration. TiF, exceedance fraction (fraction of samples exceeding the specified concentration). CUTL, upper tolerance limit-- there is 95% confidence that this fraction or fewer exposures exceeded the specified concentration. (P = 0.03). For approximately half of these samples, the IH reports noted that the seamen were either standing watch in a control room or involved in load ing and/or unloading activities; this task information was not available for the remaining personal samples. Mean area concentrations on dry cargo, tanker and Great Lakes vessels were 0.008, 0.004 and 0.004 f cc" 1, respectively. The mean area concentration on Great Lakes vessels was significantly higher (P < 0.001) than that measured for tanker and dry cargo ves sels, while mean tanker and dry cargo area concentra tions were not significantly different (P = 0.9). Of the 58 samples of unknown type, 54 samples were above their LODs, with a mean concentration of 0.048 f cc" 1; the LOD for the 4 ND samples was 0.001 f cc" 1. The mean air concentration of samples collected in crew areas of the ships (crew mess, recreation, 274 D. M. Murbach et al. .lounge areas and sleeping quarters) was 0.004 f cc" 1, whereas the mean air concentration in the engine compartments (engine and boiler rooms and machine shops) was 0.010 f cc" 1. Air samples collected in areas of the ships such as hallways, stairwells and loading areas were classified as `other', and had an average fiber concentration of 0.005 f cc" 1. The av erage concentration of those samples with unspeci fied sampling locations in the IH survey was 0.008 f cc The mean air concentration of samples col lected in the' crew areas was not significantly differ ent than the mean concentration of samples collected in the engine compartment (P > 0.05) or other areas of the ship (P > 0.05). Of those ships confirmed to have asbestos-containing insulation, av erage air concentrations in the engine spaces and crew areas were 0.005 and 0.004 f cc" 1, respectively. Air samples that were collected onboard ships at sea had similar fiber concentrations compared to docked ships. Of the 1018 area PCM air samples con sidered in this analysis, 166 were collected from docked ships, 35 were collected from ships docked and performing loading/unloading of cargo, 29 were collected from ships not currently in service or*on .standby and 609 samples were collected from ships at sea. In addition, six samples were collected from ships both docked and at sea for some portion of the sampling. In these instances, the sampling pumps were turned on while the ship was at dock, prior to departure. The sampling pumps continued to run af ter departure, and, as such, samples collected under these conditions were classified as average concen trations from the ship while docked and at sea. The ship's transport status, though, was not specified for 173 samples! Average fiber concentration of air sam ples collected while the ships were docked and at sea were 0.010 and 0.005 f cc-1, respectively, a differ ence that was not statistically significant (P > 0.05). The average fiber concentration of samples collected onboard ships with unknown transport sta tus was 0.018 f cc "1. Only one sample collected while ships were both docked and at sea was above its LOD, with a concentration of 0.001 f cc" 1. Comparisons were also made between air sam pling data from ships with positive bulk samples for asbestos insulation and those ships in which the presence of asbestos was not confirmed, as well as to those ships in which bulk samples were not eol,lected. The average airborne concentrations onboard the 28 ships with confirmed asbestos-containing insulation and the 13 ships unconfirmed to contain asbestos-containing insulation were both ~().0()3 f cc" 1 (rounded to one significant figure); mean air concentrations aboard ships confirmed to contain ACMs were not significantly higher than those col lected aboard ships unconfirmed to contain ACMs (P > 0.05). The average airborne fiber concentration onboard ships where no bulk samples were collected was 0.015 f cc" 1, a value that is significantly higher than that from the ships in which bulk samples were collected (P < 0.001). Twenty-five TEM air samples collected between 1990 and 1992 and 13 SEM air samples collected be tween 1985 and 1987 were included in this analysis. It is noteworthy that only one sample, with a concen tration of 0.124 f cc" was above the TEM analytical detection limit. Detection limits for the remaining 24 samples were 0.005 f cc" 1 and smaller. Of the 25 TEM samples collected, only 16 were from the four ships in which bulk sampling was conducted and only two of these ships had bulk samples that tested positive for asbestos. Six TEM samples were collected onboard these two ships, all with concen trations below their respective LODs. Nine air sam ples were analyzed using both TEM and PCM. These samples were collected onboard six different vessels in 1991 and 1992. Interestingly, although all nine samples were above the LOD for PCM, with an average concentration of 0.015 f cc" 1 (range of 0.01-0.02 f cc" 1), all nine of these samples were be low their respective detection limits for TEM and ranged from 0.00017 to 0.001 f cc" 1. Additionally, it was noted in the IH surveys that 13 air samples, collected between 1985 and 1987, ap peared to contain asbestos under PCM and were fur ther examined by SEM in order to possibly define the substrate as asbestos. Concentrations analyzed by SEM ranged from ND to 0.01 f cc" 1, with a mean concentration of 0.004 f cc" 1, ~4-fold lower than the mean PCM concentration for the same samples. DISCUSSION Although many studies have been conducted re garding asbestos workers (e.g. miners and those in the manufacturing environment) and professionals who work with ACMs (e.g. insulators, pipefitters and other craftsman), it has often been a matter of speculation regarding the airborne concentration of asbestos in environments containing intact insula tion. The data presented in this analysis characterize airborne fiber concentrations onboard maritime ship ping vessels not associated with insulation or other ACM work during the years 1978-1992. As de scribed previously, the airborne fiber concentration on -these ships during this period averaged 0.008 f cc" 1, with a 95th percentile of 0.040 f cc"1. More over, the data indicate that, with 95% confidence, only 2.1 and 0.7% of air concentrations may have exceeded 0.1 and 1 f cc" 1, respectively (and only a fraction of these fibers may have been asbestos since most analyses were by the PCM method).' Since the 1970s, numerous studies have character ized both indoor ambient asbestos concentrations (US schools and buildings) and outdoor ambient asbestos concentrations (Crump and Farrar, 1989; Airborne asbestos concentrations 275 Mossmati et a l, 1990; Cora et al., 1991; HEI-AR, 1991; Lee et al., 1992; Corn, 1994; Whysner et al., 1994). The two common asbestos sources which con tribute to the ambient or background airborne con centrations of these fibers include the application and removal of sprayed asbestos insulation in build ing construction and natural weathering and human activities that disturb serpentine rock formations (Crump and Farrar, 1989; Whysner et a l, 1994). Whysner et al. (1994) reported ranges of median ur ban air concentrations in US cities from 0.00002 to * 0.00075 f cc 1 and in European cities from 0.001 to 6.01 f cc-1 . Mangold (1982) also reported US urban air fiber concentrations in Portland, OR, and Seattle, WA, of 0.01 and 0.02 f cc-1 , respectively. Average airborne asbestos concentrations collected in US schools and universities, residences and public and commercial buildings from the early 1980s until the 1990s have been reported in the literature at concentrations of 0.00008-0.00018, ND-0.0025, 0.0001-0.008 and 0.00003 f cc-1 , respectively (Table 4). Additionally, the Agency for Toxic S u b stances and Disease Registry has estimated lifetime background cumulative asbestos exposure of the gen eral public, within the range of 0.002-0.4 f cc '-year (ATSDR, 2001). While results of the current analysis are generally higher than those reported for other ambient environments, they nonetheless are a fraction of the concentrations which have been reported to increase the risk of asbestos-related diseases, such as mesothelioma, asbestosis or lung cancer (Churg, 1998; Wong, 2001; Paustenbach et al., 2004; Roggli et a l, 2004; Pierce et al., 2007). Numerous studies have analyzed health data for seamen, from about 1950 until 1980, suggesting an elevated mortality and morbidity risk from lung can cer and mesothelioma (Jones et al., 1984; Rafnsson et al., 1988; Velonakis et al., 1989; Kelman and Kavaler, 1990; Moen et al., 1990; Selikoff et al., 1990; Varouchakis et al., 1991; Morabia et al., 1992; Rapiti era/., 1992; Brandt etal., 1994; Rafnsson and Gunnarsdottir, 1995; Pukkala and Saarni, 1996; Saarni -et al., 1()02; Rafnsson and Sulem, 2003; Karlev et al., 2005). Some of these studies have suggested that excess mortality and morbidity may be at least partly attributed to constant exposure to low levels of airborne asbestos generated from vi brations of the insulated shipboard equipment and the natural movements of the ship. Although vibra tions and natural ship movements could potentially Table 4. Some historical ambient airborne asbestos concentrations in buildings and outdoor environments reported in the literature Location City/area , Air concentrations (f cc 1) Reference Mean Range Outdoor Urban Five regions of USA" Portland, OR, USA Seattle, WA, USA Philadelphia, PA, USA European cities Urban and industrial areas Rural Indoor , School University Residential building Public building Commercial building USA San Francisco, CA, USA Oakland, CA, USA USA USA USA USA-- with ACM * Baltimore, MD, USA; Washington, DC, USA Five regions o f USAd-- no ACM Five regions o f USA"-- undamaged ACM Five regions of USA"-- damaged ACM USA nationwide 0.00039" 0.01b Q.02b . 0.01b 0-0.03" 0.0001-0.001 0 02b 0.02b 0-0.00065 0.00002-0.00075 0.004-0.033" 0.006-0.034" 0 .0 0 0 18b 0.00008b 0.004b 0 00099" 0.00053" 0.00073" 0.00003 b N D -0.0025 0.001-0.008b Crump and Farrar (1989) Mangold (1982) Mangold (1982) Wendlick (1984) WHO 1987, as cited in Whysner et al. 1994 NRC (1984) Mangold (1983) Mangold (1983) ` Corn (1994) Lee et al. (1992) Lee et al. (1992) US CPSC (1987) NIOSH (1976) Clump and Farrar (1989) Crump and Farrar (1989) Cramp and Farrar (1989) Lee et al. (1992) "Washington, DC, Kansas City, New York City, California (Lot; Angeles and San Francisco) and Denver. ''Fibers > 5p m in length. "Total asbestos fibers. 276 D. M. Murbach et al. increase airborne asbestos concentrations, the current data suggest that this is not the case. The observation that concentrations were slightly higher for docked ves sels than for those at sea might be explained by repairs or other ACM activities that generally occur while the ship is docked or simply because airborne concentra tions in cities are higher than those in the open waters. While the reported increased incidence and mortal ity from asbestos-related diseases most likdy result ing from handling or being in the vicinity of those who handled asbestos insulation, these occupational epidemiological studies of maritime seamen have historically not been relied upon to characterize the dose-response relationship due to insufficient expo sure information stemming from incomplete occupa tional histories (Jones et a l, 1984; Moen et al., 1990; Greenberg, 1991; Brandt et al., 1994; Rafnsson and Gunnarsdottir, 1995; Pukkala and Saarni, 1996; Saarni et a l, 2002; Karlev et al., 2005). Further, with respect to risk estimates for lung cancer, the majority of these studies did not address potential confounders, such as smoking history or previous exposure to asbestos, radiation, mineral oil mists or polycyclic aromatic hy drocarbons (PAHs) (inside or outside the navy) (Svendsen and Borresen, 1999). While these studies can be useful in identifying adverse health outcomes in seamen, many do not have sufficient information to accurately assess the associations between exposure to asbestos, confounding exposure to asbestos in other occupations, other carcino gens, smoking and the resulting health outcomes. Based on our review of dozens of case studies re garding seaman, we conclude that the bulk of their asbestos exposure resulted from insulation applica tion or removal practices (Selikoff et a l, 1990; Greenberg, 1991; Varouchakis et a l, 1991; Saarni et a l, 2002). Although it is possible that airborne as bestos concentrations on ships at sea were higher in years prior to 1978 due to a lesser concern about im portance of maintaining insulation, we would be sur prised if earlier data for these vessels (except during wartime conditions) revealed improper maintenance of insulation because of the resulting housekeeping problems. Based on the available information, then, there is no reason to believe that background expo sures to asbestos on ships would be responsible for causing adverse health effects. Although the asbestos IH air sampling data pre sented here represent 14 years of sampling informa tion and were collected from a wide variety of vessels, mechanical scenarios and areas on the ships, there were some limitations to our analysis. These are related primarily to the variability and differences in sampling methodologies and analytical measure ment techniques oyer this timeframe. This limitation, of course, is also present in virtually all historical IH data sets for nearly every setting studied. Extrapolating these results to other situations needs to account for a number of factors. First, de pending on the design of the sampling program, area samples are intended to characterize exposure levels associated with a particular production process, task or point source, and may not be representative of workers' breathing zone exposures. Although many of the IH surveys included in this analysis relied heavily on area samples (since the purpose was to identify concentrations typically found on an ocean-going vessel) rather than on samples from re pair activities, for understanding exposures on ships, area samples should be nearly as insightful as per sonal samples but this would not necessarily be true in other settings. An accurate characterization of seamen's exposures during this time period relies heavily on the results of area samples, for reasons previously discussed. Second, we recognize that it would have been ideal to have had all the samples analyzed using a single method. However, the available air sampling data in cluded in this analysis were collected during a time of considerable change in sampling or analytical methodologies, both in the USA and other countries, and the consistency and precision of the analytical techniques varied over time (i.e. sample collection and preparation, fiber counting methodology and in strumentation techniques). Changes in analytical methodology, especially the transition from P&CAM 239 to NIOSH Method 7400, almost certainly would yield some differences in results. Even with a consid erable amount of progress toward standardization of procedures and technical guidelines, we recog nize that even when the same method is used among different laboratories at different time periods, the re sults may not be equal (Beckett and Attfield, 1974; Beckett et a l, 1976; Gibbs et a l, 1977; Beckett, 1980; Ogden, 1982; Walton, 1982; Williams et a l, 2007). For example, the higher reported values in samples collected in earlier years may be a result of these differences. Of equal importance, the data set reported here resulted from the pooling of air moni toring data, taken from 52 IH surveys that were collected and analyzed by a number of IH firms andlaboratories. Combining data from these mixed sour ces means that, by definition, there is some lack of homogeneity in the data set. This is regrettable but necessary when conducting retrospective analy ses. We acknowledge that there is some lack of consistency in sampling and analysis over time and that there is a several-fold range in the LODs. Thus, while it is necessary to pool potentially non-homoge nous data sets when conducting historical exposure assessments, the issue of laboratory analytical vari ability has almost certainly resulted in some degree of uncertainty regarding the precision of results re ported here. This is not a significant concern with respect to making conclusions about the overall magnitude of exposure, but rather the precision of the statistical characterization. Airborne asbestos concentrations 277 Third, the results of the sampling program should be considered an upper-bound estimate of likely ex posure since most air samples were analyzed by PCM, an analytical technique that reports total fiber concentrations rather than total asbestos fiber concentrations. Thus, these results are expected to overestimate the actual airborne asbestos fiber con centrations, a phenomenon that was evaluated by* comparing PCM and subsequent TEM and SEM air concentration data compiled in this analysis. Addi tionally, an assessment as to whether higher detection limits might influence average concentrations, espe cially when substitution methods (replacing LOD/2 or LOD/,/2 for non-detect sample concentrations) were used for samples below the detection limit, was also warranted. To address this issue, we exam ined how detection limits changed over time in this data set. LODs in 1991 and 1992 were indeed lower than those in prior years; variance analyses suggest a slight (0.007 f/cc) but significant (P = 0.043) de crease in detection limit between 1980 and 1992. The KM method was used to accommodate this small change in detection limit over time. Fourth, one might ask whether these data are repre sentative of the diversity of ships and the various weather conditions at sea. The data reported here may represent a subset of all the asbestos measurements that have been collected on maritime shipping vessels from this time period. That is, the current analysis only in- * eludes those data that were provided to the authors by various shipping vessel firms. However, while it is pos sible that there are other studies of these ships, the data * presented here represent > 1000 samples collected over 14 years on 84 different shipping vessels, encompassing a wide range of vessel types, mechanical scenarios and areas on these ships. Since this data set is relatively large compared to most other published studies, and because the data sets are remarkably similar, the authors believe that this data set is representative of background asbes tos levels, when insulation-handling activities were not actively being performed, on shipping vessels during this time period. These limitations notwithstanding, the results of this study confirm that a large percentage of ships used between the 1970s and 1990s possessed ACMs, and that a significant fraction of the insulation on these ships contained amosite. As expected, the air borne concentrations of asbestos in those ships where the asbestos remained undisturbed were quite small. These data, then, indicate that for the past 30 years, sailors and others who worked on similar ships while docked and at sea had minimal exposure to asbestos. * CONCLUSION The IH data presented here are robust enough to conduct a retrospective exposure assessment for in dividual seamen working in the maritime shipping industry from the 1970s through the 1990s. The re sults indicate that the mere presence of undisturbed asbestos-containing insulation onboard these ship ping vessels did not significantly increase seamen's exposure to airborne asbestos. Overall, the data indi cate that background airborne asbestos concentra tions onboard these ships from at least 1978 until 1992 were very low, and in most cases, they were at least 10-fold <0.1 f cc ', the current OSHA PEL. FUNDING Various owners and/or operators of merchant vessels. ; ' Acknowledgements--Various shipping vessel owners and/or operators have been, and continue to be, involved in asbestosrelated litigation (some o f which involve seamen); however, the samples were not collected in association with any legal ac tion. One of the authors (D.J.P.) has served as an expert in litiga tion involving health effects associated with exposure to ail borne asbestos, and may serve as an expert for these companies REFERENCES ATSDR. (2001) Toxicological profile for asbestos Allanta, GA- US Department of Health and Human Services, Public Health Service. Balzer J, Cooper W. (1968) The work environment of insulat ing workers. Am Ind Hyg Assoc J; 29: 222-7. Beckett S. (1980) The effects of sampling practice on the mea sured concentration of airborne asbestos. Ann Occup Hyg; 23: 259-72. Beckett ST, Attfield M. (1974) Inter-laboratory comparisons of the counting of asbestos fibres sampled on membrane filters. Ann Occup Hyg; 17: 85-96. Beckett ST, Hey R, Hirst R et al. (1976) A comparison of air borne asbestos fibre counting with and without an eyepiece graticule. Ann Occup Hyg; 19: 69-76. Bell A. (1976) Industrial hygiene and occupational studies in Australian (New South Wales) shipyards. Environ Res; 11: 198-212. _ ` Brandt LA, Kirk NU, Jensen OC et al. (1994) Mortality among Danish merchant seamen frqm 1970-1985. Am J Ind Med; 25: 867-76. Churg A. (1998) Neoplastic asbestos-induced disease. In ch in g A and Green FHY, editors. Pathology of occupational lung disease. 2nd edn. New York' Igaku-Shoin Medical Pub lishers; pp. 339-91. Cooper W, Balzer J. (1968) Evaluation and control of asbestos exposures in the insulating trade. 2nd International Conference on the Biological Effects of Asbestos Dresden, Germany. Corn M. (1994) Airborne concentrations of asbestos in non occupational environments Ann Occup Hyg; 38- 495-502. Com M, Crump K, Farrai D et al. (1991) Airborne concentra tions o f asbestos in 71 school buildings. Regul Toxicol Phar macol; 13: 99-114. CPSC. (1987) Report on the first round of air sampling of as bestos in home study. Washington, DC1US Consumer Prod ucts Safety Comission. C rum p K S , Farrar D B . (1 9 8 9 ) S tatistica l an a ly sis ot`data on air borne asbestos levels collected in an EPA survey ot public buildings. Regul Toxicol Pharmacol; 10: 51-62. EPA. (1982) Interim method for the determination of asbestos form minerals in bulk insulation samples. EPA-600/M4-82-020. 278 D M. Murbach et al. Research Triangle Park, NC: US Environmental Protection Agency, Research and Development. EPA. (1986) Airborne asbestos health assessment update. Washington, DC: 'US Environmental Protection Agency, Office of Health and Environmental Assessment. EPA. (2007a) ProUCL version 4.0 technical guide. EPA/600/R- 07/041. Washington, DC: US Environmental .Protection Agency. EPA. (2007b) ProUCL 4.0 software. EPA/600/C-07/007. Washington, DC U S' Environmental Protection Agency. Ferris BG, Ranadive MV, Peters JM et al. (1971) Prevalence of chronic respiratory disease' asbestosis in ship repair work ers. Arch Environ Health; 23: 220-5. Finkelstein MM, Verma DK. (2001) Exposure estimation in the presence of nondetectable values' another look. Am Ind Hyg Assoc J; 6 2 1 195-8. Fleischer WE, Viles FJ, Gade RL et al. (1946) A health survey o f pipe covering operations in constructing naval vessels. J Ind Hyg Toxicol; 28: 9-16. Gibbs G, Baron S, Beckett R et al. (1977) A summary of asbes tos fibre counting experience in seven countries. Ann Occup Hyg; 20: 321-32. Greenberg M. (1991) Cancer mortality in merchant seaAen. t Ann NY Acad Sci; 643: 321-32. Harries P. (1968) Asbestos hazard in naval dockyards. Ann Occup Hyg; 11: 135^45. Harries P. (1971a) Asbestos dust concentrations in ship repair ing: a practical approach to improving asbestos hygiene in naval dockyards. Ann Occup Hyg; 14: 241-54. Harries P. (1971b) A comparison o f mass and fibre concentra tions of asbestos dust in shipyard insulation processes. Ann Occup Hyg; 14: 235--40. HEI-AR. (1991) Asbestos in public and commercial buildings: a literature review and synthesis of current knowledge. Cambridge, MA: Health Effects Institute. Helsel DR. (2005) More than obvious: better methods for inter preting nondetect data. Environ Sci Tech; 39: 4l9A ^ t23A . Hodgson JT, Darnton A. (2000) The quantitative risks of meso thelioma and lung cancer in relation to asbestos exposure. Ann Occup Hyg; 44: 565-601. Jones RN, Diem JE, Ziskand MM et al. (1984) Radiographic evidence of asbestos effects in American marine engineers. J Occup Med; 26: 281^1. Karlev L, Hansen J, Hansen HL et al. (2005) Cancer incidence among Danish seafarers: a population based cohort study. Occup Environ Med; 62' 761-5. Kelman HR, Kavaler F. (1990) Mortality patterns of American merchant seamen (1973-1978). Am J Ind Med; 17' 423-33. Lee RJ, Van Orden DR, Com M et al. (1992) Exposure to*air borne asbestos in buildings. Regul Toxicol Pharmacol; 16: 93-107. Mangold CA. (1982) The actual contribution of Garlock asbes tos gasket materials to the occupational exposure of asbestos workers. Bellevue, WA: Environmental Control Sciences, Inc. October. (Unpublished). Mangold CA. (1983) Asbestos fibers in the ambient air in the greater San Francisco area. Bellevue, WA: Environmental Control Sciences, Inc. March. (Unpublished). Mangold CA, Beckett RR, Bessmer DJ. (1968) Asbestos expo sure and pulmonary x-ray changes to pipe covers and insula tors at Puget Sound Naval Shipyard. Bremerton, WA: Industrial Hygiene Division, Puget Sound Naval Shipyard. Mangold CA, Becketi RR, Bessmer DJ. (1970) Asbestos expo sure and control at Puget Sound Naval Shipyard. Bremerton, WA: Industrial Hygiene Division Medical Department, Pu get Sound Naval Shipyard. < Mari' WT. (1964) Asbestos exposure during naval vessel over haul. Am Ind Hyg Assoc J, 25: 264-8. Moen BE, Riise T, Helseth A. (1990) Cancer among captains and mates on Norwegian tankers. Acta Path Micro Immunol- biol Scand; 98: 185-90. MorabiaA.MarkowitZiS, Garibaldi K etal. (1992) Lung cancer and occupation: results of a multicentre case-control study Br J Ind Med; 49' 721-7. Mossman B, Bignon J, Corn M et al. (1990) Asbestos' scien tific developments and implications for public policy. Sci ence; 247' 294-301. Mulhausen JR, Damiano J. (1998) A strategy for assessing and managing occupational exposures. 2nd edn. Fairfax, VA' American Industrial Hygiene Association. NIOSH. (1972) Criteria for a recommended standard, occupa tional exposure to asbestos. Washington, DC: US Depart ment of Health, Education, and Welfare. NIOSH. (1976) Revised recommended asbestos standard Washington, DC' US Department of Health, Education, and Welfare. NIOSH. (1977) Manual of analytical methods asbestos fibers in air P&CAM 239. Washington, DC: US Department of Health, Education and Welfare. NIOSH. (1994a) Asbestos (bulk) by PLM, Method 9002. Washington, DC' US Department of Health, Education and Welfare. NIOSH. (1994b) Asbestos and other fibers by PCM, Method 7400. Washington, DC. US Department of Health, Education and Welfare. NIOSH. (1994c) Asbestos and other fibers by TEM, Method 7402. Washington, DC' US Department of Health, Education and Welfare. Nicholson W, Holaday D, Heimann H. (1971) Direct and indi rect occupational exposure to insulation dusts in United States shipyards. Ann Occup Hyg; 14: 37-47. NRC. (1984) Asbestifdrm fibers, nonoccupational health risks Washington, DC' Board on Toxicology and Environmental Health Hazards, Commission on Life Sciences, National Re search Council. Ogden T. (1982) The reproducibility o f fiber counts. Health and Safety Executive Research Paper 18. Sheffield, UK Health and Safety Executive. Paustenbach DJ, Lu E, Finley BL et al. (2004) Env ironmental and occupational health hazards associated with tire presence o f asbestos in brake linings and pads (1900 to present)' a `State-of-the-Art' review. J Toxicol Environ Health B Crit Rev; 7: 33-110. Perkins JL, Cutter GN, Cleveland MS. (1990) Estimating the mean, variance, and confidence limits from censored (< limit o f detection), lognormally-distributed exposure data. Am Ind Hyg Assoc J; 51: 416-9. PierceJ, McKinley M, Paustenbach D et al. (2007) An evaluation o f reported no-effect chrysotile asbestos exposutes for lung cancer and mesothelioma. Crit Rev Toxicol; 38' 191-214. Pukkala E, Saarnt H. (1996) Cancer incident among Finnish seafarers, 1967-92. Cancer Causes Control; 7: 231-9. Rafnsson V, Gunnarsdottir H. (1995) Cancer incidence among seamen in Iceland. Am J Ind Med; 27: 187-93. Rafnsson V, Sulem P. (2003) Cancer incidence among marine engineers, a population-based study (Iceland). Cancer Causes Control; 14: 29-35. Rafnsson V, Johannesdottir SG, Oddsson H et al. (1988) Mortal ity and cancer incidence among marine engineers and machi nists in Iceland. Scand J Work Environ Health; M: 197--200. Rao ST, Ku J, Rao KS. (1991) Analysis of toxic air contaminant data containing concentrations below the limit of detection. J Air Waste Manage Assoc; 4T 442-8. Rapid E, Turi E, Forastiere F et al. (1992) A mortality cohort study o f seamen in Italy. Am Ind Hyg Assoc J; 2T 863-72. Roggli V, Oury T, Sporn T. (2004) Pathology of asbestos asso ciated diseases. 2nd edn. New York: Springer-Verlag. R ushw orth D H . (20 0 5 ) T h e N a v y and asb estos therm al in sula tion. Naval Eng J; 35-48. Spring 2005. Saarni H, Pentd J, Pukkala E. (2002) Cancer at sea: a case-control study among male Finnish seafarers. Occup Envuon Med; 59: 613-9. Airborne asbestos concentrations 279 Selikoff IJ, Hammond EC. (1978) Asbestos-associated disease in United States shipyards. Cancer J Clin; 28: 87-99. Selikoff IJ, Hammond EC, Seidman H. (1979b) Mortality experience o f insulation workers in the United States and Canada. Ann NY Acad Sci; 330: 91-116. Selikoff IJ, Lilis R, Levin G. (1990) Asbestotic radiological ab normalities among United States merchant marine seamen. * Br J Ind Med; 47: 292-7. Selikoff IJ, Lilis R, Nicholson W. (1979a) Asbestos disease in United States shipyards. Ann NY Acad Sci; 330: 295-311. Sheers G, Coles RM. (1980) Mesothelioma risks in a naval dockyard. Arch Environ Health; 35: 276-82. Svendsen K, Borresen E. (1999) Measurements of mineral oil mist, hydrocarbon vapor, and noise in engine rooms of ships. Appl Occup Environ Hyg; 14: 186-91. United States Navy. (1939) Bureau o f ships manual: instruc tions to heat insulation..Washington, DC' US Navy Depart ment, Bureau of Ships United States Navy. (1941) Naval machinery. Annapolis, MD: US Navy, The United States Naval Institute. United States Navy. (1945) Bureau of ships manual: thenhal in sulation. Chapter 39. Washington, DC: US Navy Depart ment, Bureau of Ships . United States Navy. (1947) Bureau of ships manual: thermal in sulation. Chapter 39. Washington, DC: US Navy Depart ment, Bureau of Ships . United States Navy. (1960) Bureau of ships technical manual, vol. 1: thermal insulation. Chapter 39. Washington, DC: US Navy Department, Bureau of Ships . Varouchakis G, Velonakis EG, Amfilochiou S et al. (1991) Asbestos in strange places' two case reports of mesothe lioma among merchant seamen. Am J Ind Med, 19: 6 7 3 - 6 '. Velonakis EG, Tsorva A, Tzonou A et al. (1989) Asbestosrelated chest X-ray changes among Greek merchant marine seamen. Am J Ind Med; 15: 511-6. Walton WH. (1982) The nature, hazards, and assessment of occupational exposure to airborne asbestos dust: a review. Ann Occup Hyg; 25: 117-247. Wendlick JD. (1984) Ambient asbestos fiber levels at selected sites in Philadelphia, Pennsylvania. Federal Way, WA. November. (Unpublished). Whysner J, Covello VT, Kuschner M et al. (1994) Asbestos in the air of public buildings: a public health risk? Prev Med; 23: 119-25. Williams PR, Phelka AD, Paustenbach DJ. (2007) A review of historical exposures to asbestos among skilled craftsmen (1940-2006). J Toxicol Environ Health B Crit Rev; 10: 319-77. Wong O. (2001) Malignant mesothelioma and asbestos expo sure among auto mechanics: appraisal of scientific evidence. Regul Toxicol Pharmacol; 34: 170-7.