Ann. occup. Hyg., Vol. 46, No. 5, pp. 489-500, 2002
© 2002 British Occupational Hygiene Society
Published by Oxford University Press
Self-collected Breath Sampling for Monitoring Low-level Benzene Exposures among Automobile Mechanics
Department of Environmental Sciences and Engineering, School of Public Health, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-7400, USA
Received 29 June 2001; in final form 14 January 2002
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Automobile mechanics are exposed to benzene through their contact with gasoline vapor and engine exhaust. This study investigated the benzene uptake associated with these exposures. We first evaluated the reliability of self-collected breath samples among a subset of subjects and found good agreement between these samples and those collected under expert supervision (intraclass correlation coefficient 0.79, n = 69). We then used self-monitoring together with a longitudinal sampling design (with up to three measurements per worker) to measure benzene in air and benzene in end-exhaled breath among 81 workers from 12 automobile repair garages in North Carolina. A statistically significant difference (P < 0.0001, Mann–Whitney rank sum test) was observed between non-smokers and smokers for post-exposure benzene concentration in breath (median values of 18.9 and 39.1 µg/m3, respectively). Comparing pre- and post-exposure breath concentrations within these two groups, the difference was significant among non-smokers (P < 0.0001) but not significant among smokers (P > 0.05). Mixed effects regression analysis using backwards elimination yielded five significant predictors of benzene concentration in breath, namely benzene exposure (P < 0.0001), pre-exposure benzene concentration in breath (P = 0.021), smoking status (P < 0.0001), fuel system work (P = 0.0043) and carburetor cleaner use (P < 0.0001). The between-person variance component comprised only 28% of the total variance in benzene levels in breath, indicating that differences among individuals related to physiological and metabolic characteristics had little influence on benzene uptake among these workers.
Keywords: benzene; gasoline; biological monitoring; self-monitoring; exhaled air; breath analysis; mixed models; automobile mechanics
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Although benzene is clearly associated with leukemia among workers exposed to high concentrations (>10 p.p.m. = 32 mg/m3) (Infante et al., 1977; International Agency for Research on Cancer, 1982; Wong, 1995; Savitz and Andrews, 1997), its potential carcinogenicity at lower levels remains controversial (Rinsky et al., 1987; Austin et al., 1988; Schnatter et al., 1996; Hotz and Lauwerys, 1997). Automobile mechanics represent a population of workers exposed to modest levels of benzene through their contact with gasoline and engine exhausts. Although the concentration of benzene in gasoline is typically <1% (v/v) in the USA (Wallace, 1996; Menkes and Fawcett, 1997), thermolytic dealkylation of alkylbenzenes raises the level of benzene in car exhausts to
5% of total hydrocarbon emissions (Wallace, 1996). Mechanics’ benzene exposures have recently been reported to range from 0.01 to 13.6 mg/m3, with the vast majority of measurements well below the current OSHA standard of 1 p.p.m. (3.2 mg/m3) (Nordlinder and Ramnäs, 1987; Popp et al., 1994; Mannino et al., 1995; Hotz and Lauwerys, 1997; Javelaud et al., 1998). While some epidemiological investigations of automobile mechanics suggest an increased risk of leukemia (Schwartz, 1987; Hansen, 1989; Hunting et al., 1995), others find no association (Loomis and Savitz, 1991; Hotz and Lauwerys, 1997). In any case, all such studies have been hampered by poor assessments of respiratory and dermal exposures. Indeed, possible dermal exposure to benzene has been virtually ignored despite knowledge that mechanics sometimes washed their hands with gasoline and even siphoned gasoline by mouth (Hunting et al., 1995) and despite evidence that the dermal route may be the source of as much as 80% of the benzene levels measured in blood following repair work involving direct contact with gasoline (Laitinen et al., 1994).
Given the possibility that mechanics are exposed to benzene by multiple routes, biomarkers can be used to assess the total body exposure across all absorption routes. Several short-term biomarkers of benzene exposure are sufficiently specific and sensitive for routine use among low-exposed subjects, notably unmetabolized benzene in breath (Droz and Guillemin, 1986; Drummond et al., 1988; Fiserova-Bergerova et al., 1989; Pleil and Lindstrom, 1995; Yu and Weisel, 1996; Egeghy et al., 2000) and urine (Ghittori et al., 1995; Fustinoni et al., 1999; Waidyanatha et al., 2001) and the metabolite S-phenylmercapturic acid in urine (Ong, 1994; Ghittori et al., 1995; Ong et al., 1996; Medeiros et al., 1997; Dor et al., 1999). Of these, benzene in breath is arguably the most useful short-term biomarker because it is measurable at extremely low levels, is easily obtained and requires no special storage prior to analysis. Furthermore, since workers generally perceive breath collection to be less intrusive than that of urine, higher participation rates can be anticipated.
A previous study indicated that service station attendants were exposed to widely varying levels of benzene with large components of variance both within and between workers (Lagorio et al., 1997). Assuming that automobile mechanics experience similar variability of air levels, a repeated measures design is essential for proper characterization of exposure. Since such designs require many measurements to identify factors responsible for exposure, investigators have recently reported various self-monitoring strategies to reduce professional costs (Loomis et al., 1994; Saarinen et al., 1998; Tielemans et al., 1999; Egeghy et al., 2000; Liljelind et al., 2000). We previously applied self-monitoring of benzene in air and breath to investigate benzene exposure and uptake among persons refueling their automobiles (Egeghy et al., 2000). In the current study we extend these methods to evaluate benzene uptake among automobile mechanics. Another article will investigate the determinants of benzene exposure in this population.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Study design and subjects
A total of 82 employees from 12 local automobile repair garages participated in the study. Only two subjects were female, 28 smoked cigarettes and 15 were not mechanics. The garages varied in size from small facilities in gasoline stations to large service departments in automobile dealerships. This study was designed to monitor each subject’s benzene exposure and breath levels three times over the course of 1 yr. Samples were collected between May 1998 and February 1999 with at least 6 weeks between surveys. Seventy-one subjects completed the full series of three surveys.
Sample collection
A test kit for self-measurement of benzene in air and breath was adopted from our previous study (Egeghy et al., 2000). The kit consisted of a Tenax-containing passive monitor for personal exposure, two glass bulbs of 75 ml volume for breath samples, illustrated instructions and a questionnaire. The test kits were distributed to the subjects on the mornings of the surveys. Subjects were instructed to wear the passive monitor for a 4 h period in the morning and to provide end-exhaled breath samples before and after the work period. Subjects also completed a brief questionnaire on smoking habits, specific work tasks, ventilation, personal protective equipment and meteorological conditions. The samples were collected at midday and transported to the laboratory. The median length of time between sampling and analysis was 23.5 days (range 2–100 days).
Self versus expert collection of breath samples
To investigate the assumption that self-monitored breath samples were unbiased, a supplementary (post-exposure) breath sample was obtained under the guidance of a trained expert (occupational hygienist) from each subject during the first survey at each garage. First, the subject provided a self-collected breath sample using only the written instructions for guidance. Then, the sampling technique was demonstrated to the subject, who provided an additional breath sample under supervision of the expert. The time between the two measurements was minimized to reduce any effect associated with the elimination of benzene from the body.
Air and breath monitors
The passive air monitors were custom fabricated aluminum tubes (90 mm x 6.3 mm o.d. x 5.0 mm i.d.) containing 0.1 g of 20/35 mesh Tenax TA (SKC Inc., Eighty Four, PA) with an open diffusion channel of 1.5 cm x 5.0 mm i.d. A stainless steel screen, recessed 1 mm from the surface, served as a turbulence barrier. This sampler is similar to a commercial device that has undergone extensive environmental testing (Brown, 1999). Before each use, monitors were conditioned for 3 min at 250°C followed by 3 min at 225°C, using an automatic thermal desorption system (model ATD 400; Perkin Elmer Corp.) to remove traces of benzene.
The breath samplers were custom fabricated glass bulbs (75 ml volume, 13 cm length) sealed with threaded, plastic caps containing PTFE-lined septa (Chemglass, Vineland, NJ). The subject was instructed to remove the end caps and completely exhale through the bulb following a normal inhalation. Since the bulb volume was small compared to the vital capacity, only end-exhaled air was collected (based upon measurements of CO2; Egeghy et al., 2000). Losses were avoided by immediate capping of the devices. Breath samplers were thoroughly cleaned in an industrial glassware washer and the cap liners were replaced before each use. Previous recovery experiments confirmed minimal losses during storage for at least 4 weeks (Egeghy et al., 2000). Upon receipt at the laboratory, breath samplers were checked for loose or deformed caps and for the presence of condensed water vapor. (Since condensation is inevitably observed in a bulb following collection of a valid breath sample, the lack of condensation indicates loss of the sample prior to analysis.)
Analysis of monitors
Directly before gas chromatographic analysis, breath samples were transferred from the bulbs to sorbent tubes of the type described above for air monitoring. Initially, the method described by Egeghy et al. (2000) was used, in which the bulbs were purged with zero grade air. However, midway through the first round of sampling, the following passive method was substituted to reduce the time of analysis and increase the sample throughput. One cap was removed from the bulb, whereupon a sorbent tube was quickly placed inside and the cap replaced. After 24 h the sorbent tube was removed and sealed prior to analysis. Based upon replicate trials (n = 4) with bulbs containing 8.2–32 µg/m3 benzene. The new method was found to be equally precise [pooled coefficient of variation of 3.2% (new) versus 4.0% (old)], but produced measurements 4–6% lower than the old method, primarily due to air displacement by the sorbent tube (4%). A 5% loss correction was applied to all samples transferred by the passive method.
All samples were desorbed with a Perkin Elmer ATD 400 automatic thermal desorption system (Periago et al., 1993) for 2 min at 225°C to transfer analytes onto a Tenax-packed, cryogen-free focusing cold trap maintained at –30°C. The cold trap was then rapidly heated to 225°C and held at that temperature for 0.1 min to transfer the contents to the analytical column via a fused silica transfer line, maintained at 200°C. Benzene was measured with a Hewlett Packard 6890 Series II gas chromatograph (Hewlett Packard Corp., Palo Alto, CA) initially equipped with a Hewlett Packard G1562A flame ionization detector (FID), which was later replaced by a HNU PI-52-02A photoionization detector (PID) with a 9.5 eV lamp (HNU Systems Inc., Newton, MA) to focus more explicitly upon aromatic constituents. Separation was achieved with a megabore DB-1 column of 60 m x 0.53 mm i.d. dimethylpolysiloxane (1.5 µm film thickness) (J&W Scientific, Folsom, CA). Ultra-high purity helium was used as the carrier gas at a flow rate of 8 ml/min. With the FID the oven temperature was held at 30°C for 13 min, then increased at 50°C/min to a final temperature of 250°C and held for 5 min. With the PID the oven temperature was held at 40°C for 5 min, increased to 75°C at 10°C/min, then increased at 50°C/min to 260°C and held for 5 min. Chromatograms were manually integrated using Hewlett Packard GC ChemStation software. Benzene was identified by the retention times of 10.97 (FID) and 6.05 min (PID).
Samples were quantified against external benzene standards prepared in Tedlar bags (SKC Inc.) by serial dilution of liquid benzene (99.9%) (Fisher Scientific, Pittsburgh, PA) with zero grade air. Calibration curves were determined by linear least squares regression. The limits of quantitation (LOQs) were 3.2 µg/m3 for the breath monitors and 2.0 µg/m3 for the air monitors (4 h sampling duration) based on three times the average residual benzene peak from the analysis of unexposed air samplers.
Statistical analysis
A total of 226 sets of air and breath measurements were collected from 82 subjects. Some measurements were excluded from statistical analysis because they did not satisfy requirements for quality control: eight because of equipment malfunction; six (breath samples) because of loose or deformed caps and/or lack of condensed water; 22 (breath samples) because of excessive periods of storage. As a result, 41 subjects had complete sets of measurements of benzene in air and breath (n = 3), 27 subjects had n = 2 and 13 subjects had n = 1. Observations below the LOQ were assigned values of 2/3 LOQ prior to statistical analysis (18 pre-exposure breaths, one post-exposure breath and two personal exposure measurements) (Hornung and Reed, 1990).
All statistical analyses were performed using SAS Statistical Software (SAS Institute, Cary, NC). Benzene concentrations were compared between smokers and non-smokers using the Wilcoxon–Mann–Whitney rank sum test available with the NPAR1WAY procedure of SAS. All other analyses employed (natural) logarithmic transformation to remove heteroscedasticity and satisfy normality assumptions.
Agreement between self and expert breath sampling methods was evaluated according to criteria suggested by Lee et al. (1989), which included the intraclass correlation coefficient (ICC) (ratio of the between-method variance component to the total variance of the logged measurements), a paired t-test of the logged measurements and a scatter plot comparing the two methods. This analysis was restricted to the 69 pairs of initial measurements having values above the LOQ (eight data pairs were excluded).
Mixed effects regression analysis was performed using the MIXED procedure of SAS to investigate the relationship between benzene in breath and benzene in air and to obtain restricted maximum likelihood (REML) estimates of between- and within-person variance components. Manual backward stepwise regression procedures were used to build models for post-exposure breath levels using the following explanatory variables: exposure, pre-exposure breath level, smoking status, body mass index (BMI), work with fuel systems, other contact with gasoline or gasoline vapors, use of carburetor cleaner, use of local exhaust ventilation, wearing gloves, working outdoors, ambient temperature, season and job (mechanic or non-mechanic). The following two-way interactions were investigated: exposure x BMI, exposure x smoking, exposure x temperature and temperature x BMI. The least significant variable was eliminated until only those variables with a significance level of P < 0.05 remained. The Bayesian information criterion of Schwarz (BICR) was used to select among competing models. [BICR is essentially the log likelihood penalized for the number of parameters estimated. A lower BICR indicates a better fit (Wolfinger, 1997).]
The mixed-effects model is defined as follows:
for m = 1, 2,..., p covariates, for i = 1, 2,..., k individuals, for j = 1, 2,..., ni measurements of the ith individual, where Cmij represents the observed value of the mth covariate on the jth day for the ith individual, Xij represents the benzene concentration in breath on the jth day for the ith individual and Yij is the natural logarithm of the individual measurement Xij. The logged variate Yij represents the sum of the effects consisting of 
0 representing the intercept, 
representing the fixed effects of the p covariates, ßi representing the random effect for the ith individual and 
ij representing the random error for the jth observation on the ith individual. It is assumed that the ßi and ij values are normally distributed with means of 0 and variances of 
B2 and W2 (representing the between- and within-person components of variance), respectively. This mixed model is ideal for repeated measures designs because observations from the same subject are not required to be statistically independent, a balanced design is not required and the two variance components, B2 and W2, are simultaneously estimated.
Standard regression diagnostics were performed. Extreme values were investigated for data input errors. Collinearity among the explanatory variables was investigated using Pearson correlation matrices and eigenvalues. Graphical analysis of residuals was performed to evaluate assumptions of linearity and homogeneity.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Self versus expert collection of breath samples
The scatter plot in Fig. 1 illustrates the agreement between self and expert breath samples for 69 subjects. The ICC for the two methods was 0.79 with a 95% confidence interval of 0.70–0.86. No systematic bias is apparent in Fig. 1. The difference between the means of the logged values from the two methods was 0.043, which was not significant (P = 0.89). The differences between measurements by the two methods are plotted against their means in Fig. 2, as suggested by Bland and Altman (1995). Nearly all points lie between the 95% limits of agreement and the difference between the two methods is independent of the magnitude of the measurements.
|
|
Benzene in air and breath
Measurements of benzene in air and breath are summarized in Table 1. The distributions were highly skewed, as evident from the large differences between the means and medians. The interquartile range is included in Table 1 as a measure of variability. The median benzene exposure was 59.7 µg/m3, with a range of <1.9–1140 µg/m3. The median pre-exposure breath concentration of benzene was 16.7 µg/m3, with a range of <3.2–972 µg/m3. The median pre-exposure breath concentration was significantly higher (P < 0.0001) for smokers (33.2 µg/m3) than for non-smokers (12.9 µg/m3). The median post-exposure concentration among all subjects was 22.5 µg/m3, with a range of <3.2–2030 µg/m3. Again, the median concentration was significantly higher (P < 0.0001) for smokers (39.1 µg/m3) than for non-smokers (18.9 µg/m3). The scatter plots in Fig. 3 illustrate the relationship between benzene in air and benzene in breath, with data from non-smokers (Fig. 3a) and smokers (Fig. 3b) plotted separately.
|
|
Within- and between-person variation
REML estimates of between-person (
) and within-person (
) variance components (Rappaport et al., 1999) are summarized in Table 2. For each analysis the mixed model was fitted with the random effects only, excluding all fixed effects. The ICC [
] was 0.18 for pre-exposure breath measurements, 0.28 for post-exposure breath and 0.27 for exposure measurements. This indicates that inter-subject variability contributed 18–28% of the total variability in air and breath measurements, while 72–82% was due to intra-subject variability. The addition of fixed effects to the model for post-exposure breath decreased the total variance, preferentially decreasing the between-subject component.
|
Table 2 also indicates the estimated fold ranges (designated
), which would include 95% of the log normally distributed observations of the respective distributions within and between persons (Rappaport, 1991). These estimates indicate that levels of benzene in air and breath varied 25- to 50-fold from day to day but only 7- to 8-fold from subject to subject.
Model fitting
As summarized in Table 3, the final model for the post-exposure breath concentration of benzene included five main effects, namely exposure [ln(exposure), µg/m3], pre-exposure breath concentration [ln(before), µg/m3], self-reported smoking status (smoker, 0 = non-smoker, 1 = smoker), duration of work on fuel systems during shift (fuel system, h) and use of carburetor cleaner (carburetor, 0 = no, 1 = yes). A more complex model including the additional main effects of temperature (temp, °F) and season (season, 0 = winter, 1 = autumn, 2 = spring, 3 = summer) and the interactions of exposure with temperature and exposure with smoking status (all effects with a significance level of P < 0.05; except season, for which P = 0.058) did not provide a better fit to the data. All other variables, namely BMI (kg/m2), local exhaust ventilation (local, 0 = no, 1 = yes), gloves (glove, 0 = no, 1 = yes), contact with gasoline or gasoline vapors beyond fuel systems work (gasoline, 0 = no, 1 = yes), working outdoors (out, 0 = no, 1 = yes) and job (job, 0 = not mechanic, 1 = mechanic) did not contribute significantly to the linear prediction of post-exposure breath concentration and were excluded.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Self versus expert collection of breath
Recently Liljelind et al. (2001) used passive monitors to evaluate self-assessment of exposures to terpenes in sawmills and styrene in the reinforced plastics industry. In both cases they found self and expert measurements to give essentially the same results. Use of self-monitoring to obtain biological samples, however, has been limited (Saarinen et al., 1998; Tielemans et al., 1999; Egeghy et al., 2000) and possible measurement error during self-collection of breath samples has not been evaluated.
To evaluate whether self-administered breath sampling was reliable, we investigated agreement between self and expert breath sampling for 69 subjects with detectable levels of benzene collected during the first round of sampling. As shown in Fig. 1, 90% of the sample pairs lie near the line of equality. The ICC for these data pairs was 0.79 (95% CI 0.70–0.86), which indicates satisfactory agreement between the self and expert measurements. The scatter plot shows one highly influential observation (self-measurement 5.34, expert measurement 1.84) where the self-measurement was much higher than the expert measurement. If this pair were excluded, the ICC would be 0.88 (95% CI 0.82–0.92). Lee et al. (1989) advised that in addition to having a large ICC, two interchangeable methods should exhibit no marked systematic bias and should have a statistically insignificant mean difference. No systematic bias is apparent from Fig. 1. The aggregate mean (logged) difference between the two methods was 0.043, which was not significant with a two-tailed paired t-test (P = 0.89). Bland and Altman (1995) recommended plotting the differences between measurements by the two methods on the same subject against the means of the measurements. Figure 2 illustrates that nearly all of the differences of the log transformed values are within the 95% limits of agreement (mean ± 1.96 SD) and that they are independent of the magnitude of the measurements. Considering these criteria, we conclude that the agreement between the two methods was good and that subjects can obtain reliable breath samples without expert supervision.
Benzene in air and breath
The scatter plots in Fig. 3 indicate strong linear trends between the (logged) benzene concentrations in air and breath, more random variation is apparent among smokers than non-smokers. This suggests that benzene in cigarette smoke affected the breath–exposure relationship among smokers, as would be expected among mechanics exposed to modest levels of benzene.
The median concentration of benzene in breath at the beginning of the exposure period was 12.9 µg/m3 for non-smokers and 33.2 µg/m3 for smokers (Table 1), a difference that was significantly different by the Mann–Whitney test (P < 0.0001). This indicates that smokers’ breath levels of benzene prior to the work shift were dictated largely by their smoking habits. The median pre-exposure benzene concentration in end-exhaled air of non-smokers (12.9 µg/m3) was similar to the value of 10 µg/m3 reported among urban non-smoking blood donors in Italy (Brugnone et al., 1989) and significantly higher (P = 0.0017) than the value of 8.6 µg/m3 we observed among self-service gasoline customers (almost exclusively non-smokers) prior to refueling (Egeghy et al., 2000). We suspect that these pre-exposure breath levels reflect benzene exposures received during travel to work (Björkqvist et al., 1997) and at the work sites prior to sampling.
After 4 h of work benzene levels in the breath of our subjects ranged from <3.2 to 2030 µg/m3, with a median value of 22.5 µg/m3 and an interquartile range of 36.2 µg/m3. Median concentrations in the breath were significantly higher among smokers (39.1 µg/m3) than non-smokers (18.9 µg/m3) (P < 0.0001, Mann–Whitney test), despite the fact that smokers were exposed to lower benzene concentrations in air (Table 1). Among non-smokers, post-exposure breath levels were significantly higher than pre-exposure levels (P < 0.0001). In a similar study Hotz and Lauwerys (1997) collected end-of-shift breath samples (end-exhaled air) from 170 Belgian vehicle mechanics with a median benzene exposure of 32 µg/m3, however, none of the breath measurements were above their detection limit of 160 µg/m3, which was 50-fold higher than our LOQ.
Since alveolar air is in equilibrium with pulmonary blood (Wilson, 1986), breath concentrations (Ca, µg/m3) can be predicted from measurements of benzene in blood (Cb, ng/l) using the relationship observed by Perbellini et al. (1988) among 34 occupationally exposed workers: Cb = 7.4 x Ca + 104 (r2 = 0.83). This relationship was used to predict alveolar air levels of benzene from blood measurements in several recent investigations, as summarized in Table 4. Of these, only the results of Mannino et al. (1995) (median exposure 41 µg/m3; median predicted breath concentration 25 µg/m3) are consistent with the current investigation (median exposure 59.7 µg/m3; median breath concentration 22.5 µg/m3). Predicted breath concentrations in the other studies were all much higher than those in the current investigation and probably reflect higher air levels of benzene. Since these studies were conducted either in Europe (Popp et al., 1994) or in Alaska (Moolenaar et al., 1997), where the benzene content of fuel was generally between 3 and 5%, higher blood and breath levels would be expected.
|
Within- and between-person variation
The great variability in breath levels observed in our study (Fig. 3) is consistent with findings among benzene-exposed workers in other industries [i.e. chemical processing (Kivisto et al., 1997) and coke ovens (Drummond et al., 1988)]. However, the relative magnitudes of the intra-individual variance component (from factors that vary from day to day, e.g. benzene sources, ambient temperature, exposure concentration–time profiles, etc.) and the inter-individual variance component (from factors that vary from person to person but are assumed to remain constant over time, e.g. personal work habits, biological factors that modify absorption, distribution, metabolism and excretion) have rarely been investigated. Figure 4 illustrates the relative magnitudes of these variance components by plotting the changes observed in breath levels of benzene from survey to survey among the 19 subjects monitored at one of the work sites. In this garage benzene levels appeared to vary more from survey to survey than among subjects.
|
In our previous study of benzene in air and breath during self-service refueling we used the magnitudes of
and 
to make inferences regarding environmental and subject-specific sources of variability (Egeghy et al., 2000). We reasoned that whereas 
arises from environmental factors affecting all persons equally, 
points to systematic differences in exposure or breath levels among subjects. Since 
accounted for only 28% of the variation in breath benzene concentration (Table 2), we conclude that subject-specific factors had little influence on the breath levels of benzene among mechanics. Adding fixed effects to the model preferentially reduced the between-person component, thereby reinforcing this conclusion. This result is quite similar to that observed previously during self-service refueling, where 
accounted for 26% of the variation in breath benzene concentration (Egeghy et al., 2000).
Since either environmental or breath measurements can be used to evaluate benzene exposures among mechanics, the respective ratios of 
(designated 
) can be used to gauge the relative efficiency of the two types of measures for use in an investigation of health effects of benzene among mechanics (Rappaport et al., 1995), i.e. the exposure measure with the smaller 
would lead to less measurement error in the exposure–response relationship at a given sample size. As shown in Table 2, values of 
are essentially equal for benzene in air or breath, suggesting that either measure would be about equally efficient in evaluating the health effects of benzene exposure among garage mechanics. With no loss of efficiency breath measurements might actually be preferred, since they offer the added benefit of accounting for dermal as well as airborne exposures to benzene.
Model fitting
Our analysis yielded five significant predictors of post-exposure breath levels (Table 3), namely ln(exposure), ln(before), fuel system, carburetor and smoker. Residuals are plotted against predicted values in Fig. 5. The plot indicates that the residuals are evenly scattered about 0, supporting the assumptions of homogeneity and linearity. Individually excluding from the analysis each of the seven observations that form a ring around the main cluster did not meaningfully change any of the regression parameter estimates or significance levels.
|
The findings that exposure and cigarette smoking would be significant predictors of benzene in breath (P < 0.0001) were anticipated at the modest air levels of benzene expected among mechanics. The effect of smoking is readily apparent when data were aggregated by subject, as shown in Fig. 3c, where breath levels were shifted to higher values among smokers at a given exposure to benzene. Interestingly, the model fit was slightly better with smoking status classified as a dichotomous variable (yes/no) rather than a continuous variable (number of cigarettes smoked in the previous 4 h), with BICR values of 465.8 and 468.0, respectively.
Pre-exposure breath level was also a significant but somewhat weak predictor of benzene in breath at the end of the workshift (P = 0.021). Omitting this variable slightly decreased the model fit (BICR increased from 465.8 to 469.3) and inflated the regression coefficient for exposure (from 0.60 to 0.66). This suggests some utility in obtaining pre-shift breath measurements if this can be done easily.
Use of carburetor cleaner (carburetor) was associated with significantly higher benzene concentrations in breath (P = 0.0086), even while controlling for exposure. Carburetor cleaner is a pressurized, volatile solvent mixture often containing toluene (30–35%), propane (25–30%) and methanol (20–25%) (Valvoline Oil, 2001). Despite its name, carburetor cleaner is used primarily for degreasing parts. [The correlation between use of carburetor cleaner and fuel system work in this study was low (r = 0.29) and only 33% of subjects who reported using carburetor cleaner also reported working on fuel systems.] Although concomitant toluene exposure has been found to competitively inhibit benzene metabolism (Purcell et al., 1990), metabolic saturation is unlikely at the low exposure levels experienced by mechanics. While it is possible for carburetor cleaners to contain benzene as an impurity, our chemical analysis of one popular brand found no detectable benzene (<870 µg/l; data not shown). We suspect that undetermined work practices associated with the use of carburetor cleaner increase benzene levels in the breath and recommend that future studies investigate workers’ behaviors while carburetor cleaner is used.
Finally, duration of work on fuel systems (fuel system) was significantly but negatively correlated with benzene concentrations in breath (P = 0.0043). Previous studies (Nordlinder and Ramnäs, 1987; Popp et al., 1994; Javelaud et al., 1998) have shown that tasks involving fuel systems (for example replacing fuel filters, cleaning fuel injectors, draining gas tanks, etc.) produced the highest benzene exposures. Assuming that these tasks were short [e.g. changing a fuel pump was reported to require only 17 min (Nordlinder and Ramnäs, 1987)], a longer duration would indicate more tasks being performed with fuel systems, each of relatively short duration. Thus, a mechanic’s report of a long duration of work with fuel systems suggests a series of brief periods punctuated by intense (near-field) exposure. On the other hand, a mechanic’s report of either no work or a short duration with fuel systems suggests a more constant (far-field) exposure to gasoline vapor and exhaust throughout the day. Petreas et al. (1995) reported significantly higher retention of styrene during fluctuating exposures than during constant exposures, with the result that the concentrations of styrene in breath were significantly lower during fluctuating exposures. The authors attributed this difference to the difficulty in achieving equilibrium between alveolar air and blood during fluctuating exposures over time scales of a few minutes. Such a phenomenon could explain the negative correlation between benzene in breath and the duration of fuel system work (while controlling for exposure) in the current study.
Self-measurement of dermal exposure is impractical and was not attempted. The results of our statistical analysis, however, suggest that dermal uptake was unimportant in comparison with pulmonary uptake. Of the variables possibly related to dermal contact (i.e. glove use, fuel system work, other contact with gasoline or gasoline vapors), only fuel system work was a significant predictor of breath levels, but its negative correlation indicates that the tasks could not have resulted in significant dermal uptake. It appears that the fuel system work may not have produced the direct skin contact with gasoline that accompanied the significant dermal uptake previously reported by Laitinen et al. (1994).
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Our results indicate that unbiased measurements of benzene in breath can be obtained by self-administered methods. Since inter-individual variability comprised
18–28% of the total variation in breath levels of benzene, we conclude that subject-specific factors were less important determinants of benzene exposures than environmental factors among mechanics. Indeed, of the five factors that significantly affected benzene uptake in our study, three (benzene exposure, fuel system work and use of carburetor cleaner) represented environmental factors. This indicates that under conditions of rather low benzene exposure (median benzene level 59.7 µg/m3) physiological and metabolic differences among subjects had relatively little influence upon benzene uptake. Based upon considerations of measurement error, related to ratios of within- to between-subject variance components, we conclude that post-exposure breath samples would be about equally efficient as personal monitors for evaluating benzene exposures in investigations of health effects among mechanics. Acknowledgements—We thank Mr Matthew Ray (Triangle Creative) and Mrs Lisa Hauf-Cabalo for graphic design and the content of the instruction sheet, Mr Randall Goodman and Mr Cliff Burgess (University of North Carolina) for manufacture of passive monitors, Mr Calvin Ghoddoussi for assistance with chemical analysis of the samples and Mr Joachim Pleil (US EPA), Dr Rogelio Tornero-Velez and Dr Suramya Waidyanatha (University of North Carolina) for many helpful discussions. This work was supported by the National Institute for Environmental Health Sciences through grants P42ES05948 and T32ES07018.
|
FOOTNOTES |
|---|
* Author to whom correspondence should be addressed. E-mail: stephen_rappaprt@unc.edu
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Austin H, Delzell E, Cole P. (1988) Benzene and leukemia: a review of the literature and a risk assessment. Am J Epidemiol; 127: 419–39.
Björkqvist S, Spetz A, Ramnäs O, Petersson G. (1997) Isoprene from expired air inside a private car. Sci Total Environ; 207: 63–7.[Medline]
Bland JM, Altman DG. (1995) Comparing two methods of clinical measurement: a personal history. Int J Epidemiol; 24 (suppl. 1): S7–14.[Abstract]
Brown RH. (1999) Environmental use of diffusive samplers: evaluation of reliable diffusive uptake rates for benzene, toluene and xylene. J Environ Monit; 1: 115–6.[Web of Science][Medline]
Brugnone F, Perbellini L, Faccini GB et al. (1989) Benzene in the blood and breath of normal people and occupationally exposed workers. Am J Ind Med; 16: 385–99.[Web of Science][Medline]
Dor F, Dab W, Empereur-Bissonnet P, Zmirou D. (1999) Validity of biomarkers in environmental health studies: the case of PAHs and benzene. Crit Rev Toxicol; 29: 129–68.[Web of Science][Medline]
Droz PO, Guillemin MP. (1986) Occupational exposure monitoring using breath analysis. J Occup Med; 28: 593–602.[Web of Science][Medline]
Drummond L, Luck R, Afacan AS, Wilson HK. (1988) Biological monitoring of workers exposed to benzene in the coke oven industry. Br J Ind Med; 45: 256–61.[Web of Science][Medline]
Egeghy PP, Tornero-Velez R, Rappaport SM. (2000) Environmental and biological monitoring of benzene during self-service automobile refueling. Environ Health Perspect; 108: 1195–202.[Web of Science][Medline]
Fiserova-Bergerova V, Lowry LK, Rosenberg J. (1989) Biological monitoring II. Measurements in exhaled air. Appl Ind Hyg; 4: F10–3.
Fustinoni S, Giampiccolo R, Pulvirenti S, Buratti M, Colombi A. (1999) Headspace solid-phase microextraction for the determination of benzene, toluene, ethylbenzene and xylenes in urine. J Chromatogr; 723B: 105–15.
Ghittori S, Maestri L, Fiorentino ML, Imbriani M. (1995) Evaluation of occupational exposure to benzene by urinalysis. Int Arch Occup Environ Health; 67: 195–200.[Web of Science][Medline]
Hansen ES. (1989) Mortality of auto mechanics. A ten-year follow-up. Scand J Work Environ Health; 15: 43–6.[Web of Science][Medline]
Hornung RW, Reed LD. (1990) Estimation of average concentration in the presence of nondetectable values. Appl Occup Environ Hyg; 5: 46–51.
Hotz P, Lauwerys RR. (1997) Hematopoietic and lymphatic malignancies in vehicle mechanics. Crit Rev Toxicol; 27: 443–94.[Web of Science][Medline]
Hotz P, Carbonnelle P, Haufroid V et al. (1997) Biological monitoring of vehicle mechanics and other workers exposed to low concentrations of benzene. Int Arch Occup Environ Health; 70: 29–40.[Web of Science][Medline]
Hunting KL, Longbottom H, Kalavar SS, Stern F, Schwartz E, Welch LS. (1995) Haematopoietic cancer mortality among vehicle mechanics. Occup Environ Med; 52: 673–8.
Infante PF, Rinsky RA, Wagoner JK, Young RJ. (1977) Leukaemia in benzene workers. Lancet; 2: 76–8.[Web of Science][Medline]
International Agency for Research on Cancer. (1982) IARC monographs on the evaluation of carconogenic risks to humans, Vol. 29, Some industrial chemicals and dyestuffs. Lyon: IARC. pp. 95–148.
Javelaud B, Vian L, Molle R et al. (1998) Benzene exposure in car mechanics and road tanker drivers. Int Arch Occup Environ Health; 71: 277–83.[Web of Science][Medline]
Kivisto H, Pekari K, Peltonen K et al. (1997) Biological monitoring of exposure to benzene in the production of benzene and in a cokery. Sci Total Environ; 199: 49–63.[Medline]
Lagorio S, Iavarone I, Iacovella N et al. (1997) Variability of benzene exposure among filling station attendants. Occup Hyg; 4: 15–30.
Laitinen J, Kangas J, Pekari K, Liesivouri J. (1994) Short time exposure to benzene and gasoline at garages. Chemosphere; 28: 197–205.
Lee J, Koh K Ong CN. (1989) Statistical evaluation of agreement between two methods for measuring a quantitative variable. Comput Biol Med; 19: 61–70.[Web of Science][Medline]
Liljelind IE, Strömbäck AE, Järvholm BG, Levin JO, Brangert BL, Sunesson AK. (2000) Self-assessment of exposure—a pilot study of assessment of exposure to benzene in tank truck drivers. Appl Occup Environ Hyg; 15: 195–202.[Medline]
Liljelind IE, Rappaport SM, Levin JO, Strömbäck AE, Sunesson AK, Järvholm BG. (2001) Comparison of self and expert assessment of occupational exposure to chemicals. Scand J Work Environ Health; 27: 311–7.[Web of Science][Medline]
Loomis DP, Savitz DA. (1991) Occupation and leukemia mortality among men in 16 states: 1985–1987. Am J Ind Med; 19: 509–21.[Web of Science][Medline]
Loomis DP, Kromhout H, Peipins LA. (1994) Sampling design and field methods of a large, randomized, multisite survey of occupational magnetic field exposure. Appl Occup Environ Hyg; 9: 49–52.
Mannino DM, Schreiber J, Aldous K, Ashley D, Moolenaar R, Almaguer D. (1995) Human exposure to volatile organic compounds: a comparison of organic vapor monitoring badge levels with blood levels. Int Arch Occup Environ Health; 67: 59–64.[Web of Science][Medline]
Medeiros AM, Bird MG, Witz G. (1997) Potential biomarkers of benzene exposure. J Toxicol Environ Health; 51: 519–39.[Web of Science][Medline]
Menkes DB, Fawcett JP. (1997) Commentary: Too easily lead? Health effects of gasoline additives. Environ Health Perspect; 105: 270–3.[Web of Science][Medline]
Moolenaar RL, Hefflin BJ, Ashley DL, Middaugh JP, Etzel RA. (1997) Blood benzene concentrations in workers exposed to oxygenated fuel in Fairbanks, Alaska. Int Arch Occup Environ Health; 69: 139–43.[Web of Science][Medline]
Nordlinder R, Ramnäs O. (1987) Exposure to benzene at different workplaces in Sweden. Ann Occup Hyg; 31: 345–55.
Ong CN, Lee BL. (1994) Determination of benzene and its metabolites: application in biological monitoring of environmental and occupational exposure to benzene. J Chromatogr; 660B: 1–22.
Ong CN, Kok PW, Ong HY et al. (1996) Biomarkers of exposure to low concentrations of benzene: a field assessment. Occup Environ Med; 53: 328–33.
Perbellini L, Faccini GB, Pasini F et al. (1988) Environmental and occupational exposure to benzene by analysis of breath and blood. Br J Ind Med; 45: 345–52.[Web of Science][Medline]
Periago JF, Prado C, Ibarra I. (1993) Application of thermal desorption to the biological monitoring of organic compounds in exhaled breath. J Chromatogr; 657A: 147–53.
Petreas MX, Woodlee J, Becker CE, Rappaport SM. (1995) Retention of styrene following controlled exposure to constant and fluctuating air concentrations. Int Arch Occup Environ Health; 67: 27–34.[Web of Science][Medline]
Pleil JD, Lindstrom AB. (1995) Collection of a single alveolar exhaled breath for volatile organic compounds analysis. Am J Ind Med; 28: 109–21.[Web of Science][Medline]
Popp W, Rauscher D, Müller G, Angerer J, Norpoth K. (1994) Concentrations of benzene in blood and S-phenylmercapturic and t,t-muconic acid in urine in car mechanics. Int Arch Occup Environ Health; 66: 1–6.[Web of Science][Medline]
Purcell KJ, Cason GH, Gargas ML, Andersen ME, Travis CC. (1990) In vivo metabolic interactions of benzene and toluene. Toxicol Lett; 52: 141–52.[Web of Science][Medline]
Rappaport SM. (1991) Assessment of long-term exposures to toxic substances in air (review). Ann Occup Hyg; 35: 61–121.
Rappaport SM, Symanski E, Yager JW, Kupper LL. (1995) The relationship between environmental monitoring and biological markers in exposure assessment. Environ Health Perspect; 103 (suppl. 3): 49–54.
Rappaport SM, Weaver M, Taylor D, Kupper L, Susi P. (1999) Application of mixed models to assess exposures monitored by construction workers during hot processes. Ann Occup Hyg; 43: 457–69.
Rinsky RA, Smith AB, Hornung R et al. (1987) Benzene and leukemia: an epidemiologic risk assessment. N Engl J Med; 316: 1044–50.[Abstract]
Saarinen L, Hakkola M, Pekari K, Lappalainen K, Aitio A. (1998) Exposure of gasoline road-tanker drivers to methyl tert-butyl ether and methyl tert-amyl ether. Int Arch Occup Environ Health; 71: 143–7.[Web of Science][Medline]
Savitz DA, Andrews KW. (1997) Review of epidemiologic evidence on benzene and lymphatic and hematopoietic cancers. Am J Ind Med; 31: 287–95.[Web of Science][Medline]
Schnatter AR, Armstrong TW, Thompson LS et al. (1996) The relationship between low-level benzene exposure and leukemia in Canadian petroleum distribution workers. Environ Health Perspect; 104: 1375–9.
Schwartz E. (1987) Proportionate mortality ratio analysis of automobile mechanics and gasoline service station workers in New Hampshire. Am J Ind Med; 12: 91–9.[Web of Science][Medline]
Tielemans E, Heederik D, Burdorf A et al. (1999) Assessment of occupational exposures in a general population: comparison of different methods. Occup Environ Med; 56: 145–51.
Valvoline Oil. (2001) Vermont SIRI material safety data sheet archive. [on-line] http://www.vetmed.ucdavis.edu/msds/h/ q497/q250.html
Waidyanatha S, Rothman N, Fustinoni S et al. (2001) Urinary benzene as a biomarker of exposure among occupationally exposed and unexposed subjects. Carcinogenesis; 22: 279–86.
Wallace L. (1996) Environmental exposure to benzene: an update. Environ Health Perspect; 104 (suppl. 6): 1129–36.
Wilson HK. (1986) Breath analysis: physiological basis and sampling techniques. Scand J Work Environ Health; 12: 174–92.[Web of Science][Medline]
Wolfinger RD. (1997) An example of using mixed models and proc mixed for longitudinal data. J Biopharm Statist; 7: 481–500.[Medline]
Wong O. (1995) Risk of acute myeloid leukaemia and multiple myeloma in workers exposed to benzene. Occup Environ Med; 52: 380–4.
Yu R, Weisel CP. (1996) Measurement of benzene in human breath associated with an environmental exposure. J Expos Anal Environ Epidemiol; 6: 261–77.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  G. Ramachandran Toward Better Exposure Assessment Strategies--The New NIOSH Initiative Ann. Hyg., July 1, 2008; 52(5): 297 - 301. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. HERTSENBERG, D. BROUWER, M. LURVINK, C. RUBINGH, E. RIJNDERS, and E. TIELEMANS Quantitative Self-Assessment of Exposure to Solvents Among Shoe Repair Men Ann. Hyg., January 1, 2007; 51(1): 45 - 51. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. SYMANSKI, S. MABERTI, and W. CHAN A Meta-Analytic Approach for Characterizing the Within-Worker and Between-Worker Sources of Variation in Occupational Exposure Ann. Hyg., June 1, 2006; 50(4): 343 - 357. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y S Lin, L L Kupper, and S M Rappaport Air samples versus biomarkers for epidemiology Occup. Environ. Med., November 1, 2005; 62(11): 750 - 760. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. LAVOUE, C. BEAUDRY, N. GOYER, G. PERRAULT, and M. GERIN Investigation of Determinants of Past and Current Exposures to Formaldehyde in the Reconstituted Wood Panel Industry in Quebec Ann. Hyg., October 1, 2005; 49(7): 587 - 602. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. F. PERIAGO and C. PRADO Evolution of Occupational Exposure to Environmental Levels of Aromatic Hydrocarbons in Service Stations Ann. Hyg., April 1, 2005; 49(3): 233 - 240. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P P Egeghy, L Hauf-Cabalo, R Gibson, and S M Rappaport Benzene and naphthalene in air and breath as indicators of exposure to jet fuel Occup. Environ. Med., December 1, 2003; 60(12): 969 - 976. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I Liljelind, S Rappaport, K Eriksson, J Andersson, I A Bergdahl, A-L Sunesson, and B Jarvholm Exposure assessment of monoterpenes and styrene: a comparison of air sampling and biomonitoring Occup. Environ. Med., August 1, 2003; 60(8): 599 - 603. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||