shift information. Detailed current job information was pro- vided by the plant personnel office for all workers, and detailed employment histories were provided ...
Olfactory Function in Chemical Workers Exposed to Acrylate and Methacrylate Vapors BRIAN S. SCHWARTZ, MD, RICHARD L. DoTY, PHD, CARL MONROE, MD, MPH, RICHARD FRYE, BA, AND SUE BARKER, RN Abstract: An investigation of the olfactory function of 731 workers at a chemical facility which manufactures acrylates and methacrylates was undertaken using a standardized quantitative test. In a cross-sectional analysis of the data, no associations of chemical exposure with olfactory test scores were observed. A nested casecontrol study designed to evaluate the cumulative effects of exposure on olfactory function, however, revealed elevated crude exposure odds ratios (95% confidence interval) of 2.0 (1.1, 3.8) for all workers and 6.0 (1.7, 21.5) for workers who never smoked cigarettes. Logistic
regression analysis, adjusting for multiple confounders, revealed exposure odds ratios of 2.8 (1.1, 7.0) and 13.5 (2.1, 87.6) in these same groups, respectively, and a dose-response relationship between olfactory dysfunction and cumulative exposure scores-semi-quantitative indices of lifetime exposure to the acrylates. The data also revealed decreasing exposure odds ratios with increasing duration since last exposure to these chemicals, suggesting that the effects may be reversible. (Am J Public Health 1989; 79:613-618.)
Introduction A large body of animal toxicologic data indicates that many airborne chemicals can selectively damage cells within the olfactory neuroepithelium and bulb, often in a specific and dose-related manner.'19 Depending on the chemical agent and the concentrations and durations of exposure, these changes may be either reversible or irreversible and may occur after either acute or chronic exposure.6'9 Miller and others, for example, have reported that rats and mice exposed to ethyl acrylate, acrylic acid, and methyl methacrylate (5-300 ppm) evidenced, within the olfactory epithelium, respiratory metaplasia, squamous metaplasia, loss of olfactory neurons, hyperplasia of submucosal glandular elements, inflammation, degeneration, and focal necrosis.""'3 These changes were dose-related to the severity and distribution of the damage and generally spared non-olfactory epithelia. A lower level of exposure below which damage consistently did not occur was not identified. Despite these findings, the olfactory function of industrial workers exposed to such airborne chemicals has not been assessed, largely because an easy-to-administer and valid quantitative test of olfactory function has not, until recently, been available.9"4" 5 There are many cogent reasons why smell function should be assessed in the industrial setting: 0 Normal olfactory function is important for worksite safety in a number of workplaces, e.g., during accidental leakage (odor detection thresholds can be lower than the level at which poisoning or explosion occurs). 916.17
* Early detection of job-related changes in olfaction may allow reversal of detrimental effects (e.g., by decreasing exposure to the implicated agents). 823 * Olfactory testing may be a useful addition to the neurotoxicologic testing armamentarium.24'25 Quantitative tests of sensory and motor function have been employed but many of these tests require skilled technicians and are expensive, time-consuming, and invasive. Thus, the testing of olfactory function, as used in this investigation, has potential value in evaluating neurotoxicologic effects of chemicals that have detrimental effects on olfactory neurons and peripheral nerves; this testing is easy to perform, reliable, rapid, quantitative, and comparatively inexpensive. 15,26,21 Normative data are now available from thousands of subjects.26 The present study was undertaken to test the hypothesis that airborne ethyl acrylate, acrylic acid, and methyl methacrylate (all chemicals implicated to cause olfactory epithelial destruction in animals) cause a decrement in olfactory function in chemical workers after acute or chronic exposure. Both cross-sectional and nested case-control studies were performed.
Address reprint requests to Dr. Richard L. Doty, Ph.D., Smell and Taste Center, Hospital of the University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 19104. Dr. Doty is also affiliated with the Department of Otorhinolaryngology and Human Communication, and the Department of Physiology, School of Medicine, University of Pennsylvania. Dr. Schwartz is a Fellow with the Clinical Epidemiology Unit and the Section of General Internal Medicine, Department of Medicine, Hospital of the University of Pennsylvania. Dr. Monroe and Ms. Barker are with the Rohm and Haas Company of Philadelphia. Mr. Frye is with the Smell and Taste Center and the Department of Otorhinolaryngology and Human Communication, University of Pennsylvania. This paper, submitted to the Journal June 21, 1988, was revised and accepted for publication October 4, 1988. Carl Monroe died in December 1988. The authors gratefully acknowledge his primary role in the initiation and facilitation of this study. Dr. Monroe was an outstanding colleague who will be sorely missed. © 1989 American Journal of Public Health 0090-0036/89$1.50
AJPH May 1989, Vol. 79, No. 5
Methods Study Population
The study population consisted of all employees at a large manufacturing facility of the Rohm and Haas Company. The plant employs 909 workers in a variety of job activities and manufactures a number of chemicals, including acrylic acid and a variety of acrylates and methacrylates (especially ethyl acrylate, butyl acrylate, and methyl methacrylate). The facility operates 24 hours a day on three shifts. Participation in the study was voluntary, but encouraged through plant publicity from December 1986 to February 1987. Testing was conducted during two weeks in February 1987. Approximately 850 of the 909 plant employees were available for testing during the two-week study period. The 59 workers who were unavailable for testing included workers absent for long-term disability, illness, and vacation. Of these 850 persons, 731 were tested; 710 (83 per cent of 613
SCHWARTZ, ET AL.
available employees, 78 per cent of total employees) were full-time Rohm and Haas employees, and 21 were contractors or consultants who happened to be on location during the testing period and desired to participate. The participants included 618 males and 113 females, ranging in age from 1769 years (mean 42.9 years, SD = 11.3 years). The test group included 596 Whites, 98 Blacks, and 37 persons of Asian or Hispanic ethnicity. The subjects were from all three work shifts: 526 day shift, 131 evening shift, and 74 night shift. Data Collection
The testing consisted of administering the University of Pennsylvania Smell Identification Test,* a questionnaire contained within the UPSIT, and a brief questionnaire to obtain shift and job information. The UPSIT consists of four booklets, each containing 10 microencapsulated odorants positioned on brown strips on the bottom of each test page. 14'26 The test is self-administered by scratching a pencil tip over the odorant label strip on the bottom of the page to release the fragrance, smelling the label, and choosing the perceived odor from four multiple choice answers. The test is forced-choice; the subject is required to choose one of the four answers even if no smell is perceived. The test is highly reliable and correlates with more lengthy test procedures.'4'27 Development of the UPSIT, as well as its relation to age, gender, and other factors, is described elsewhere. 14,2629 The test questionnaire obtained information on these study variables: age, gender, ethnic group, smoking history and dose, educational level, history of medical problems (self-reported problems which required treatment), history of smell or taste dysfunction, current medications, and work shift information. Detailed current job information was provided by the plant personnel office for all workers, and detailed employment histories were provided for the 154 workers in the nested case-control study. Study Design A cross-sectional (prevalence) study of olfactory function in chemical workers was undertaken which considered only current jobs and exposures. Next, a nested case-control study was performed to assess the cumulative effects of the chemicals ofinterest (see next section) on olfactory function. To be eligible for the case-control study, all cases and controls had to be full-time employees of Rohm and Haas for at least the past six months. Cases (N = 77) were defined as subjects scoring at or below the tenth percentile (for their age) on the UPSIT in 10 five-year age strata: ages 21-25, 26-30, 31-35, 36-40, 41-45, 46-50, 51-55, 56-60, 61-65, and 66-70 years. These cases were matched 1:1 to a random sample of controls defined as subjects scoring at or above the 50th percentile, for their age, in these same age strata. Cases and controls were matched on gender, ethnic group, and age. It should be noted that the atypical selection of controls in this study (restricting eligible controls to the top half of their distribution) was designed to increase our ability to detect an association between chemical exposure and olfactory dysfunction if one existed. The design would tend to exaggerate the association (compared to a design using all non-cases as eligible controls, as is typically done) only if it existed; if no association between exposure and olfactory *UPSIT; commercially available as the Smell Identification Test"', Sensonics, Inc, Haddonfield, NJ.
614
function existed, there would not be an increased risk of detecting one (i.e., no increased risk of a type I error). Definition of Exposure
In the cross-sectional study, workers were classified into
one of four exposure categories based on current job only: * no significant chemical exposures (N = 319), * exposure to chemicals other than acrylic acid, meth-
acrylic acid, acrylates, or methacrylates (N = 193), acid, acrylates, or methacrylates (N = 164), * exposure to higher levels of acrylic acid, methacrylic acid, acrylates, or methacrylates (N = 55).
* exposure to lower levels of acrylic acid, methacrylic The foundation for the classification of workers into the lower and higher acrylate/methacrylate categories relied on the distinction between an Operator I and an Operator II in this plant. Those in the Operator I category spend most of their time in a ventilated control room monitoring the chemical process with little time spent outdoors on the chemical columns supervising workers or taking samples. An Operator II employee works primarily outdoors on the chemical columns taking samples and performing other duties. Industrial hygiene data collected in the plant from 1978-85 confirmed that Operator II employees and pumpers had the highest chemical exposure among workers in the plant. Pumpers work mainly to pump materials into and out of containers for shipping or receiving; during the opening of valves and lines for this purpose, high local concentrations ensue. When the distinction was unclear between Operator I and II in certain areas of the plant, Job Exposure Profiles and the aforementioned industrial hygiene data were used to classify workers into the lower and higher exposure categories. Job Exposure Profiles, developed by plant industrial hygienists, detail the chemicals worked with and the concentrations of exposure for many jobs in the plant, and thus were useful in this semi-quantitative assignment of exposure status. Finally, plant and corporate industrial hygienists helped to classify jobs whose categorization was unclear after the above steps were taken. These jobs were primarily ones at other Rohm and Haas plants and were approximately evenly distributed among the four exposure categories. In the case-control study, exposure was studied in three ways: * Exposure expressed as "never" or "ever" employed in job category 3 or 4 above for at least six weeks; * the total duration of employment at the plant analyzed as a continuous variable; and * a cumulative exposure score for each worker, analyzed as a continuous variable. The cumulative score was calculated by classifying each job title the worker ever held into one of the four exposure categories outlined above, multiplying the duration of time employed in each job title by the score for that job title, and adding these for alljob titles ever held at the plant. The first two categories-no exposure and other chemical exposure-were assigned a score of zero, the lower acrylate/methacrylate category a score of one, and the higher acrylate/methacrylate category a score of two. Current ethyl acrylate and acrylic acid levels varied from 0.01 ppm to 56 ppm; most areas of the plant have levels (as eight hour time-weighted averages) well below the threshold limit values (TLV) recommended by the American Conference of Governmental Industrial Hygienists of 10 parts per million (ppm) for acrylic acid and 5 ppm for ethyl acrylate,
A6JPH May 1989, Vol. 79, No. 5
OLFACTORY FUNCTION IN CHEMICAL WORKERS
both as eight hour time-weighted averages. The highest levels were found when chemicals were pumped into containers for shipping, or when lines were opened for new connections or to obtain samples. Much of the available industrial hygiene data were for these specific high-exposure tasks (5-40 minute samples) and revealed levels at the high end of the aforementioned range. Current levels were used in determining exposure scores. Analysis
Data analysis was performed on a microcomputer with the BMDP Statistical Software program. After detailed descriptive analyses, including frequencies and cross-tabulations, a three-way analysis of covariance with two covariates was used as the main model for the cross-sectional study. In the case-control study, matched crude exposure odds ratios were calculated by discordant pairs,30'3 and then conditional logistic regression was performed to calculate an adjusted exposure odds ratio. These two results were very similar to those calculated in unmatched analyses, so the matches were broken for the remainder of the data analysis. The only matched analysis results presented are the crude exposure odds ratio and a discordant pair table. In the unmatched analyses, crude odds ratios were calculated for all study variables, and then the Mantel-Haenszel procedure was used to calculate summary statistics after stratification by important confounders.30,31 Logistic regression was then employed to adjust for other study variables in the calculation of the adjusted exposure odds ratio. The nested case-control study examined three groups of workers: the entire group of cases and controls; only cases and controls who never smoked cigarettes; and cases redefined as those with UPSIT scores below 30 compared to all controls. The definition of cases below the tenth percentile by age allowed some cases with UPSIT scores as high as 35 out of 40; therefore, it was important to evaluate the worstperforming cases separately. A score of 30 or less on the UPSIT represents significant diminution of olfactory ability;'4 for example, based on normative data from thousands of subjects,26 the approximate tenth percentile of performance on the UPSIT in the age range 16-50 years old is 34 for females and 33 for males.
Results The cross-sectional analysis revealed mean (95% confidence interval) UPSIT scores of 37.8 (37.5, 38.0), 37.4 (37.0, 37.7), 37.0 (36.6, 37.5), and 37.6 (36.9, 38.1) in the none, other, lower acrylate/methacrylate, and higher acrylate/methacrylate exposure groups, respectively. The main model (three-way analysis ofvariance with two covariates) included ethnic group (White, Black, other), smoking status (never, previous, current), and exposure category (none, other chemicals, lower acrylates, higher acrylates) as grouping factors, and age and age-squared as covariates. This analysis revealed no differences in UPSIT scores in the four exposure groups (F = 2.19, df = 3/698, p = 0.09) after control of confounding by these variables. Characteristics of cases and controls can be found in Table 1. There were apparent differences in the smoking history and ethnic categories. Results of the crude matched analysis are presented in Table 2. In the unmatched analysis, the case-control study revealed elevated crude exposure odds ratios for all workers and for never smokers (Table 3). A crude exposure odds ratio (95% CI) of 2.6 (1.2, 6.0) was found for cases with UPSIT scores below 30 (worst-perAJPH May 1989, Vol. 79, No. 5
TABLE 1-Characteristics of Cams and Controls
Characterstics
Cases (%)
Controls (%)
Number Age (mean, SD) Gender Male Level of Education (mean, SD) (high school graduate = 12, college graduate = 16) UPSIT score (mean, SD) Medical History On medications Medical problems Smell problems Ethnic Group White Black Other Smoking History Never Previous Current Unknown Work Shift tested Day Evening Night Exposure Ever exposed Duration of employment, years (mean, SD) Cumulative exposure score (mean, SD)
77 42.8, 1.3
77 43.0, 1.2
64 (83)
64(83)
13.2, 0.3 28.7, 0.8
14.0, 0.3 38.9, 0.1
19 (25) 16 (21) 12 (16)
17 (22) 13 (17) 5 (6)
52 (68) 20 (26) 5 (6)
58 (76) 14 (18) 5 (6)
29 (38) 16 (21) 28 (36) 4 (5)
30 (39) 28 (36) 17 (22) 2 (3)
54 (70) 13 (17) 10 (13)
61 (79) 13 (17) 3 (4)
47 (61)
34 (44)
15.5, 1.2
14.6, 1.2
7.8, 1.2
6.6, 1.4
Percentages are shown in parentheses
forming cases (N = 34) compared to all controls, N = 111). Table 3 also shows the 2x 2 contingency tables after stratification by smoking status. Effect modification appears to be present, with a markedly elevated exposure odds ratio in the never smoking group. The logistic regression model used to adjust for confounding included all the measured independent variables and the matching variables: chemical exposure, smoking status (three categories), ethnic group (three categories), medications, age, history of smell dysfunction, history of medical problems, level of education, gender, and work shift tested. This model was then applied to the three groups of workers outlined above. There were no significant interactions with the exposure variable in this model for any of the three groups of workers. Control of all confounders with logistic regression revealed adjusted odds ratios of 2.8 (1.1, TABLE 2-Crude Matched Exposure Odds Ratio
Controls
Exposed
Unexposed
Totals
Exposed
25
22
CASES Unexposed Totals
47
9 34
21 43
30 77
Crude matched exposure odds ratio = 2.4 (22/9)30.31 95% Confidence Interval = (1.2, 6.7)31 NOTES: Table contains 77 case-control pairs, comparing the exposure status of each pair. Odds raio calculated by ratio of discordant pairs.
615
SCHWARTZ, ET AL. TABLE 3-Contingency Tables Stratified by Smoking Status
Acrylate Exposed A: Before Stratification Cases Controls Totals Acrylate exposure odds
Unexposed
Totals*
30 43
77 77 154
47 34 81 ratio (95% Cl): 2.0 (1.1,
73
3.8)
B: After Stratification Never Smokers Cases Controls Totals Acrylate exposure odds
18 ratio (95% Cl): 6.0 (1.7, 21.5)
Previous Smokers Cases Controls Totals Acrylate exposure odds
10 18 28 ratio (95% Cl): 0.9 (0.3,
14 4
15 26 41
Current Smokers Cases 23 11 Controls Totals 34 Acrylate exposure odds ratio (95% Cl): 2.5 (0.6,
6 10 16
3.2) 5 6
28 17 45
cumulative exposure score increases, then a decrease in this odds ratio in the highest category of exposure. Many employees worked in different areas of the plant where other chemicals were manufactured and acrylates and methacrylates were not used. In an attempt to control for the effects of other chemical exposures, a surrogate for exposure to these chemicals was calculated as years ever employed in other sections of the plant for all workers. The results of logistic regression analyses controlling for years employed in other areas of the plant, as well as all other measured independent variables, are presented in Table 5. There appears to be no confounding by these variables. Finally, a summary of the data examining the reversibility of the olfactory dysfunction associated with these chemicals can be found in Table 6. The decreasing exposure odds ratios as duration since exposure increases suggests that the olfactory dysfunction associated with chemical exposure may be reversible. This finding is not entirely consistent, however, as can be seen by the adjusted odds ratios in Table 6. It should be noted that the categories in Table 6 were chosen for illustrative purposes; the results did not change when other cutpoints were chosen or when the year since last exposed variable was included in the models as a continuous variable.
TABLE 4-Evidence of a DoseResponse Relation with Exposure
Discussion This study evaluated exposure to acrylate and methacrylate vapors in prevalent cases of olfactory dysfunction and controls without olfactory dysfunction. There are many potential problems in such use of prevalence data in epidemiologic research.3' Speaking in general terms, it is well known that numerous factors can influence the prevalence of disease (a generic term for the effect under study), such as disease duration and severity, that do not influence other effect measures such as risk or incidence. The association between exposure and disease can also be artifactual if disease duration is not identical in exposed and unexposed cases or if disease affects exposure status. The bias introduced by these factors can be in either direction (increasing or decreasing the real association) and is not predictable in this study. Review of employment records revealed evidence that persons employed in this plant enter one area (cyanhydrin, acetylene, acrylate, etc., processes with different physical areas in the plant) and stay in that area in various jobs for their entire career. This would argue against olfactory dysfunction being a prime determinant of exposure status. The finding of the highest relative risk of olfactory dysfunction in the never smoking group is consistent with many plausible biological hypotheses. These subjects were
Cumulative Exposure Score
No. Cases (%)
No. Controls (%)
Crude OR
Adjusted* OR
TABLE 5-Exposure Odds Ratios Controlling for Exposure to Other Chemicals
0 0.1-4.9 5-13 >13.1
30 (41) 14 (54) 16 (64) 17 (57)
43 (59) 12 (46) 9 (36) 13 (43)
1.0 1.7 2.6 1.9
1.0 1.4 4.2 2.4
(C)oefficient
SE*
C/SE
0.3836
0.1863
11
10.0)
Test for homogeneity for three stratum-specific odds ratios: p = 0.12. *Some subjects were missing smoking information.
7.0) for all workers (N = 141), 13.5 (2.1, 87.6) for never smokers (N = 57), and 2.9 (0.8, 11.3) for cases with UPSIT scores below 30 (N = 101). The most important confounders were smoking status, self-reported smell dysfunction, and educational level. The adjusted estimate of the overall exposure odds ratio was insensitive to changes in the method of controlling for smoking in the logistic regression model, e.g., when smoking was changed from a categorical variable to a continuous variable (in pack-years). A summary of the data suggesting a dose-response relationship between olfactory dysfunction and the cumulative exposure score can be found in Table 4. Examination of these data reveals increasing exposure odds ratios as the
Logistic regression test for linear trend, adjusted for confounders:
'Controlling for all potential confounders with logistic regression (see text). tStandard error of beta coefficient 616
2.059
Controlling for Years Employed in: Acetylene area Cyanhydrin area Laboratories Mechanical area Other areas classified as "other" chemical exposure, not listed above Total duration employed in mechanical area or any area classified as "other" chemical exposure
Acrylate/Methacrylate Exposure Odds Ratio (95% Cl) 3.1 (1.2, 7.7) 2.8 (1.1, 7.0) 2.8 (1.1, 7.0) 2.8 (1.1, 7.1)
2.9 (1.1, 7.6) 3.0 (1.1, 7.7)
AJPH May 1989, Vol. 79, No. 5
OLFACTORY FUNCTION IN CHEMICAL WORKERS TABLE S-Evidence of Trend for Reversibility
Years since Last Exposure to Acrylates
0-6.5 6.6-35 Never exposed
No. Cases (%)
No. Controls (%)
Crude OR
Adjusted*
32 (60) 15 (54) 30 (41)
21 (40) 13 (46) 43 (59)
1.0 0.8 0.5
1.0 1.0 0.3
(C)oefflclent
SE*
C/SE
-0.5452
0.2485
-2.194
Logistic regression test for linear trend, adjusted for confounders:
OR
*Controlling for all potential confounders with logistic regression (see text). tStandard error of beat coefficient
not exposed to the insult of cigarette smoking on olfactory function, and thus would be expected to be at the highest risk for olfactory dysfunction from chemical exposure if this exposure caused olfactory dysfunction. This relative risk is substantial, with an odds ratio of 6.0 in the crude analysis and 13.5 in the adjusted analysis. The cross-sectional study is best suited to evaluate the effects of recent exposure to these chemicals, and did not suggest any association between current job and olfactory dysfunction. The mean UPSIT score in the highest exposure group was virtually identical to the mean score in the non-exposed group. However, the nested case-control study, designed to evaluate the cumulative effects of lifetime expo-
sure, revealed consistently elevated exposure odds ratios in many different groups even after control for confounding.
The dichotomized exposure variable (ever or never exposed) and the cumulative exposure score can be criticized as being crude measures of exposure; however, these categorizations are likely to cause a nondifferential misclassification of exposure, which would tend to bias our findings toward the null.31 Also, the assumption that specific jobs in the plant during the 1950s and 1960s had the same exposure levels as current jobs is unlikely to be true; again, this is a nondifferential misclassification of exposure. Finally, the definition of cases in this study as below the tenth percentile for their age in the cohort can be criticized as an arbitrary cut-off. This cut-off is also more likely to bias our results toward the null. In fact, when the analyses were repeated including just the worst-performing cases (those with UPSIT scores below 30), the crude exposure odds ratio was higher than for the overall group (the crude exposure odds ratio was 2.0 overall and 2.6 for the worst-performing cases); however, the adjusted estimate was unchanged. The association between exposure and relative risk of disease is supported by the apparent finding of the doseresponse nature of this relationship. The reduction in risk in the highest category of exposure remains unexplained. It may reflect a leveling-off rather than a reduction of risk, but further studies are necessary to clarify this finding. This dose-response relationship was not observed when duration of employment was substituted for the cumulative score in the logistic regression model. This important finding supports the contention that the cumulative score variable was not just a surrogate for duration of employment, but rather reflected a weighted score of exposure to the chemicals of interest. The finding of decreasing odds ratios with increasing duration since exposure is also consistent with many biological hypotheses; however, caution should be used in interpreting the meaning of this finding. Evidence obtained from prospective AJPH
May 1989, Vol. 79, No. 5
studies of olfactory function after exposure to these chemicals would be needed to conclude that the effects are reversible. It should be noted that the plant manufactures a number of chemicals other than the acrylates/methacrylates to which workers are exposed. These other exposures were not measured but were controlled for in the analyses by controlling for years of employment in other areas of the plant. This method of controlling for other chemical exposures is acknowledged to be crude; it was the best method with the available data. The fact that the adjusted exposure odds ratio did not change after including these variables in the model reveals the absence of confounding by these "other exposures" variables. This was not surprising in view of the fact that workers in this plant tend to work in one area during their career; thus, work in other areas cannot be associated with exposure to acrylates/methacrylates, a necessary condition for confounding. It is possible, however, given the crude method of controlling for them, that other chemicals in the plant may be responsible for, or contribute to, the observed decrement in olfactory ability, alone or in combination with the acrylates/methacrylates. As participation was voluntary, the 80 per cent participation could lead to another source of error in estimating the relation between exposure and olfactory dysfunction. The direction of this bias may be either to increase or decrease the association between exposure and olfactory dysfunction. This bias was evaluated by examining the jobs of the nonparticipants; with one exception, these jobs seemed to be a random distribution of nonparticipation by job title. The exception was that many workers employed in the mechanical group (electricians, pipefitters, carpenters, painters, etc.), approximately 40 per cent, chose not to participate. The mechanical group is a large, diverse group of workers, including those exposed and unexposed to acrylates/methacrylates. Although not obviously present, self-selection bias is a well-recognized occurrence in studies of this type.3' There is a large body of literature on olfactory dysfunction in humans secondary to chemical exposure; approximately 120 chemicals have been implicated to cause such dysfunction.923 Most of the published studies are case reports or case series, few measure exposure quantitatively, and none employ rigorous epidemiologic methods. The studies that do measure exposure quantitatively suggest that hyposmia or anosmia only result from exposure to concentrations above the threshold limit value (TLV).9 17 In the present study, the uncertainty in the exposure levels before monitoring began in 1978 causes difficulty in attributing the observed changes in olfactory function to a given exposure limit. This study cannot evaluate the effects of exposures orders of magnitude below the TLV (as may be encountered with environmental exposure). What is known about the physiology of the olfactory epithelium and its mucosal barrier would suggest that a threshold to the effects of chemicals may exist, in that this mucosal barrier absorbs and possibly degrades low levels of chemicals before they reach the olfactory neuroepithelium itself. There are many points in the olfactory pathway that may be affected by exposure to these chemicals, and many pathophysiologic mechanisms may be postulated. The animal data revealed that exposure to acrylate and methacrylate vapors caused loss of olfactory neurons (neurotoxicity) as well as changes in the olfactory mucosa. The present study cannot determine the exact site of action of this exposure in humans; it may be caused by changes in the receptor site or 617
SCHWARTZ, ET AL.
mucosa, neurotoxic effects, or central nervous system effects (olfactory bulb or pyriform, prepyriform, and entorhinal cortices). The animal data suggest that the most likely site is the olfactory mucosa or primary olfactory neuron. Although this study suggested that exposure to acrylates and methacrylates is associated with a decrement in olfactory ability, as measured by a standard test of olfaction, it should be noted that this impairment is physiologic, but often not clinical. It is known that UPSIT scores in the range observed in the cases in this study are associated with a diminished ability to smell a number of environmental chemicals. The degree to which these changes effect the quality of life, lead to a decreased ability to perform job duties, or increase the risk of injury in areas where olfaction may be an early warning system for escape from hazardous airborne chemicals is unknown. The results of this work suggest that medical surveillance should be instituted to: clarify the effects of low-level exposure; determine the adequacy of the current exposure limits; determine whether the effects are reversible; and determine to what degree decreased olfactory ability affects quality of life and job safety. ACKNOWLEDGMENTS The authors would like to thank Greg Maislin, MS, MA, for statistical guidance; Paul D. Stolley, MD, MPH, for his assistance in many phases of the study; Harold I. Feldman, MD, for many helpful discussions; and Donald Martin, MPH, for technical assistance and facilitation of many phases of the project. This research was supported by the Rohm and Haas Company, Philadelphia, PA, by NINCDS Grant NS16365 (Dr. Doty), and by the Andrew W. Mellon Foundation (Dr. Schwartz).
REFERENCES 1. Chang LW, Dyer RS: Trimethyltin induced pathology in sensory neurons. Neurobehav Toxicol Teratol 1983; 5:673-6%. 2. Buckley LA, et al: The toxicity of dimethylamine in F-344 rats and B6C3F1 mice following a 1-year exposure. Fund Appl Toxicol 1985; 5:341-352. 3. Rehn B, et al: Chemical blockade of olfactory perception by N-methylformimino-methylester in albino mice. II. Light microscopical investigations. Chem Senses 1981; 6:317-328. 4. Appleman LM, Woutersen RA, Feron VJ: Inhalation toxicity of acetaldehyde in rats: I. Acute and subacute studies. Toxicology 1982; 23:293307. 5. Haschek WM, et al: Pathology of acute inhalation exposure to 3methylfuran in the rat and hamster. Exp Mol Pathol 1983; 39:342-354. 6. Jiang XZ, Buckley LA, Morgan KT: Pathology of toxic responses to the RD50 concentration of chlorine gas in the nasal passages of rats and mice. Toxicol Appl Pharmacol 1974; 71:225-236, 1974. 7. Morse CC, Boyd MR, Witschi H: The effect of 3-methylfuran inhalation exposure on the rat nasal cavity. Toxicology 1984; 30:195-204.
8. Rand GM, et al: Effects of inhalation exposure to hexachlorocyclopentadiene on rats and monkeys. J Toxicol Environ Health 1982; 9:743-760. 9. Amoore JE: Effects of chemical exposure on olfaction in humans. In: Barrow CS (ed): Toxicology of the nasal passages. Washington, DC: Hemisphere Publishing, 1986; 155-190. 10. Stott WT, McKenna MJ: Hydrolysis of several glycol ether acetates and acrylate esters by nasal mucosal carboxyesterase in vitro. Fund Appl Toxicol 1985; 5:399-404. 11. Miller RR, et al: Inhalation toxicity of acrylic acid. Fund Appl Toxicol 1981; 1:271-277. 12. Miller RR, et al: Propylene glycol monomethyl ether acetate (PGMEA) metabolism, disposition, and short-term vapor inhalation toxicity studies. Toxicol Appl Pharmacol 1984; 75:521-530. 13. Miller RR, et al: Chronic toxicity and oncogenicity bioassay of inhaled ethyl acrylate in Fischer 344 rats and B6C3F1 mice. Drug Chem Toxicol 1985; 8:1-42. 14. Doty RL, Shaman P, Dann M: Development of the University of Pennsylvania Smell Identification Test: A standardized microencapsulated test of olfactory function. Physiology and Behavior (Monogr), 1984; 32: 489-502. 15. Doty RL, Gregor T, Monroe CB: Quantitative assessment of olfactory function in an industrial setting. JOM 1986; 28:457-460. 16. Nathanson G: Des processus pathologiques occupant l'appareil olfactif chez les personnes ayant subi une intoxication aigne, involuntaire, par l'oxyde de carbonne. Acta Otolaryngol 1929; 13:409-418. 17. Amoore JE, Hautala E: Odor as an aid to chemical safety: odor detection thresholds compared with TLVs and volatilities for 214 industrial chemicals in air and water dilution. J Appl Toxicol 1983; 3:272-290. 18. Ahlstrom R, et al: Impaired odor perception in tank cleaners. Scan J Work Environ Health 1986; 12:574-581. 19. Harada N, et al: Oflactory disorder in chemical plant workers exposed to S02 and/or NH3. Jpn J Ind Health 1983; 59:17-23. 20. Emmett EA: Parosmia and hyposmia induced by solvent exposure. Br J Ind Med 1976; 33:196-198. 21. Adams RG, Crabtree N: Anosmia in alkaline battery workers. Br J Ind Med 1961; 18:216-222. 22. Ahlborg G: Hydrogen sulfide poisoning in shale oil industry. Arch Ind Hyg 1951; 3:247-266. 23. Doty RL: A review of olfactory dysfunctions in man. Am J Otolarngol 1979; 1:57-79, 23. 24. Benignus VA, Prah JD: Olfaction: Anatomy, physiology and behavior. Environ Health Perspect 1982; 44:15-21. 25. Baker EL (Guest Ed): Occupational nervous system disease. Semin Occup Med 1986; 1:153-219. 26. Doty RL: The Smell Identification Test' Administration Manual, 2nd Ed. Haddonfield, NJ: Sensonics Inc, 1988. 27. Doty RL, Newhouse MG, Azzalina JD: Internal consistency and shortterm test reliability of the University of Pennsylvania Smell Identification Test. Chem Senses 1985; 10:297-300. 28. Doty RL, et al: A cross-cultural study of sex differences in odor identification ability. Neuropsychologia 1985; 23:667-672. 29. Doty RL, et al: Smell identification ability: changes with age. Science 1984; 226:1441-1443. 30. Schlesselman JJ: Case-Control Studies: Design, Conduct, Analysis. New York, NY: Oxford University Press, 1982. 31. Rothman KJ: Modem Epidemiology. Boston, MA: Little, Brown and Company, 1986.
I
I
National Stroke Association Announces Fellowship Awards
I
Allied-Signal/National Stroke Association Fellowships are being offered to young investigators interested in making a career of research into the causes, mechanisms and treatment of stroke. Stipends of $30,000 per year are available for one to three years. Applicants must have a doctoral degree and be affiliated with an institution in the United States. Application and supporting materials are due by June 1, 1989. For further information or application, contact:
Carolyn Burkhardt, MD Research Program Coordinator National Stroke Association 3000 E. Hampden Avenue, #240 Englewood, CO 80110-2662 Tel: (303) 762-9922 618
AJPH May 1989, Vol. 79, No. 5
