Olfactory function in painters exposed to organic solvents

OHLSON C-G. Olfactory function in painters exposed to organic solvents. Scand J Work Environ Health 1989;15:60-63. The olfactory receptor cellsare in direct contact with the exterior environment, and somechemicalagents can impair olfactory function. The ol factoryfunctionof 54paintersexposed to organicsolvents wascomparedwiththat of 42unexposed referents. A new clinicaltest validated for the senseof smellwas used, the Universityof Pennsylvania Smell Iden tification Test. Age, smoking habits, exposureto organic solvents, and medical disorders of importance for the sense of smell were recorded. The painters had a somewhat lower test score than the referents. However, the influence of the exposure variable was not statisticallysignificant in a multiple regression analysis including age and smoking habits. The exposure to organic solvents was low, and therefore an effect of highexposure on olfactory function cannot be ruled out. Sincesome of the painters had earlier been highly exposed, the effects of high exposure are likely to be reversible.

The olfactory epithelium is situated in the superior region of the nasal cavity, and the area is estimated to be I cm' on each side (I). The olfactory mucosa is exposed to all pollutants in the ambient air. A description of the human olfactory organ, clinical problems, and olfactory function testing has been given elsewhere (2)(3)(4)(5)(6). The olfactory receptor cells are the only neurons which have their cell bodies in direct contact with the exterior environment, and it is therefore not surprising that some chemical agents can impair olfactory function (7).
Organic solvents can impair the function of the central nervous system (8), and some chemical compounds can cause neuritis (9). Information on the influence of organic solvents on olfactory function is scarce (10, II). Ekblom et al (12) studied the olfactory epithelium of the frog after exposure to styrene and toluene and observed morphological and electrophysiological changes. In a questionnaire study Danish electronic workers exposed to chlorinated solvents were asked about their sense of smell. Forty-nine percent reported a decreased sense of smell compared to 19 % of the unexposed workers (13). Ahlstrom et al (14) examined the odor perception of 20 tank cleaners exposed to petroleum products. The odor thresholds were increased for n-butanol and oil vapor but unaffected for pyridine and dimethyldisulfide I d after exposure. No changes in the odor thresholds were observed when the exposure stopped two or more days before the test.
In the present study it was hypothesized that, due to its direct contact with the exterior environment, the olfactory organ is the part of the central nervous system most vulnerable to organic solvents. The objective of this study was to examine the olfactory function of men occupationally exposed to organic solvents.

Methods
The University of Pennsylvania Smell Identification Test (UPSIT), a clinical test of olfactory function, was adopted (IS, 16). It incorporates 40 diverse odorants in an easily presented form. The test consists of four envelope-sized booklets, each containing 10 odorants. The stimuli are embedded in 10-to 50-Jlm microcapsules fixed in a proprietary binder and positioned on brown strips at the bottom of the pages of the booklets.
Above each odorant strip is a multiple choice question with four alternative responses for each item, and the smell is released by scratching the strip with a pen. In an investigation of its usefulness, the UPSIT test appeared to be a highly reliable and internally consistent measure of the ability to identify odors. The correlation coefficient for the test-retest reliability was determined to be about 0.95 (17).
In the United States subjects with a normal sense of smell and between the ages of 20 and 50 years were found to have a mean test score of 36-38 correct answers. Subjects with total anosmia have only about 10 correct responses, and this accomplishment is to be expected by chance alone. The test score decreases with age, but the decrease seems not to be apparent before 50 to 60 years of age (18).
Most of the smell items are well known in the Swedish culture. Some of the items are however "typically American " and therefore not qu ite suita ble in the Swedish version of the test but have similarities to other " Swedish odo rs." Therefore, eight of the 40 items were pro vided with other alternative respon ses. Th e test has been appli ed in Japan , and it appear ed that cultural factors pla y a role, since some of the odors were less familiar to the Japanese than to the Ameri can s (19).
Some days before the test the subjects received a questionnaire on present and past work conditions and smoking habit s, categorized as never smokers, curr ent smo kers (cigarettes/day) and ex-smokers (having given up smoking mo re than half a year ago). Furthermor e, present and past disorders of importance for the sense of smell were inqu ired about, eg, head tr auma , diabetes mellitus, neurolo gical disorders, and allergies. The subjects were also asked about how they perceived thei r sense of smell, in terms of "good," "imparied," or "completely lost." An otorhinolar yngeal examin ation was perf ormed immediately aft er the test so that the subjects with obstruc tive disord ers like nasal polyps and sinusitis, as well as common colds, could be excluded from the calculations . The examinations were performed earl y in the morning, when the subjects had not been exposed to organic solvents for at least 12 h , dur ing three weeks in late spring and one week in the summer.

Subjects
Male painte rs and spray painters younger than 60 years of age and employed at least one year were selected for the stud y. Painters who worked for three paint ing companies and were regularly expos ed to organic solvents were invited to participate. According to information from a questionnaire survey three years earlier, they were exposed to organic solvents at least 2 hid. As a consequen ce of the general awareness nowadays of the potential hazards of or gani c solvents, most paintin g work is done with wat er-based paints. The painters' current exposure to organi c solvents was therefore estima ted to be 20 070 o f their worktime.
All spray painters emplo yed in a department of a mechanical work shop were selected . The spray painters were expo sed every day to or gani c solvents, and their mean expo sure levels were estimated to be mod erat e. This estimate was suppor ted by a higher than expected pre valence o f neuropsychiatr ic symp toms reported in a health examination carri ed ou t by the occupationa l health service. Every eligible subject selected for the study agreed to part icipate an d received the questionna ire.
Referent s, unexposed to organic solvents and younger than 60 years of age, were chosen from the sta ff of a military regiment. All eligible subjects participated.
Two exposed subjects were excluded fro m the calculations due to nasal polyps and chronic rhiniti s. Among the referents one subject with kno wn anosmia caused by cerebral traum a and four subjects who were disturbed du ring the test procedure were excluded from the calculations . Th e exposed subjects had a mean age of 39 years, and 35 070 were cur rent smokers. Th e cor responding figur es for the referents were 36 years and 31 % . Th e study groups consisted o f 54 subj ects expo sed to organi c solvents (39 pa inter s and IS spray painters) and 42 unexposed referents.

Statistical methods
Differences between the mean scores were anal yzed with Student 's t-test. The influence of the exposure was anal yzed with a linear multipl e regression anal ysis under the assumption o f a normal distri bution of the depend ent variable.

Results
The exposed subjects and the referent s showed a difference of 1.4 in the crude mean test scores (table I). This difference was almost statistically significant (P = 0.05).
The distribution by age and the proportion of smokers differed between the study gro ups, and ther efore a multiple regression analysis was carried out in ord er to account for the possible confounding effect of these varia bles. With the test score as th e dependent variable, exposure status (dichotomously as exposed/unexposed) was not sta tistically significant (P = 0.16, SE (b) = 0.70) nor was smoking habits (P =0.08, SE (b) = 0.04) (ex-smokers = never smokers). Age, however, was statistically significant (P < 0.01, SE (b) = 0.03). The F-test value of the analysis was 6.98 (P< O.OI).
Th e subject' s percept ion of his sense o f smell (in terms of " good, " " impaired, " and " completely lost") correlate d well with the test score. A regression analysis with the test score as the dependent variable showed a significant influence on the perception of smell (P <O.Ol, SE (b) = 0.87), corre sponding to an F-test value of 24.5 (P<O.OOI).
The influence of age and smoking habit s on the test performance of the referents was furth er analyzed with a multi ple regression analysis. The result was similar to that obtained from the study groups combined. The age effect was sta tistically significant (P = 0.02), and the smo king variable was a lmost statistically significant (P = 0.05). The following equation was obtained for th e referents: test score = 39.8 -0.08 x age -0.08 x cigar ettes/day.

Discussion
The exposed subjects had a somewhat lower mean test score than the referents , but the difference was not statistically significant after adjustment for age and smoking habits. Th e test scores were no t normall y distributed, and therefo re th e use of a linear regression model may be disput ed . H owever, the distribution was not considered so skewed as to in validate the regression an alysis. One exposed subject had a test score of 12, and this extremely low value had a marked influen ce on the mean va lue for the gro up. However , it was not considered appropriate to exclude thi s value because it may be the str ongest indica tion of an effe ct of solvent exposure. Participation in th is study was voluntary, and , even though all eligible subjects accepted , the study gro ups canno t be regarded as repr esentative of expo sed and un expo sed populations in general. Th e differences in relevant determinants or confounding variates , eg, age and smo king habits, were however con sider ed in th e multiple regression ana lysis. The exclusion of three subjec ts with organic diso rders likely to a ffect th e o lfactory function and four subj ects distu rb ed during th e test was done without knowledge of th e individual test scores and should not have introduced any major bias .
Th e current expo sure to o rganic solvents was low or moderat e, a nd no one seemed to have been recently exposed to high levels, ie, the occupational exposure limit or higher. Thi s lack of variability in th e exposure variable restricted the statis tical an alysis, which had to be done with the expos ure varia ble dichotomized. A cumu lative measure of exposure (ie, intensity x time) was not considered proper, as exposure time is stro ngly co rr elated with the age of the painters. Mo st painters enter the ir profession after school and seldom change jobs. Fu rth ermore, the painters and th e spray painters had different types of exposure, but the numb er of subjects was too low to j ustify sepa rat e regression an alyses.
Th e reliability of the test is con sider ed goo d (17) and is refl ected in the relati vely small standa rd deviati on shown in tabl e I (correspo nding to a coefficient of variati o n o f about 10 0/0). As a conseq uence, a study of thi s size ha d a po wer (alpha=0.05, I -beta =0.90) sufficient to detect a difference of at least two test score units with the assumptions of unbiased comparison and a normal distribution of the variable. The painters and the referent s were alt ern ately examined on the same days during two weeks but in different offices. The spray painters were exam ined thr ee mon ths later. However, no major variatio ns in the test procedures are believed to have influenced th e comparability between the study groups.

62
The test measu res the ability to recognize various odors , and the test score can therefore be expected to correlate well with the su bjects' perce ption of smell, which in fact was the case in th is st udy . The test measuresneither thresholds nor odor intensity. Othermodalities of olfactory function may have been more relevant to the issue of olfactory dysfunction caus ed by organic sol vents. Odor threshold tests are commonly used to test the sense of smell. Th erefore we a lso examined all the su bjects except the spray pa int ers with a commercia l pyridine thre shold test (20). T he result of this test did not howe ver differ between the exposed painters and the un expo sed referents, and there was a statistica lly non significant corre lation with the subjec tive percept ion of smell.
In concl usion, thi s cro ss-section al study of paint ers, spray painters, and military sta ff did not demonstrate any statistically significant decrea se in olfactory function after exposure to organic solvents. The expo sures to organic solvents were low to moderate, but some of the painters had , without doubt, been exposed previously to high levels of organ ic solvents. I f high exposur e ha s an ef fect on olfacto ry function, it is likel y to be reversible.