Effects of occupational exposure to organic solvents and noise on hearing.

Effects of occupa tional exposure to organic solvents and noise on hearing. Scand J Work Environ Health 1993;19:245 54. This study explored the effects of occupational exposure to solvents and noise on hearing. Inter views and hearing tests were conducted for printing and paint manufacturing workers. The experi mental groups included unexposed (N = 50) workers and workers exposed to noise (N = 50), noise and toluene (N = 51), or an organic solvent mixture (N = 39). The risk of hearing loss was greater for the exposed groups than for the unexposed group. The adjusted relative risk estimates were four times greater [95% confidence interval (95% CI) 1.4-12.2] for the noise group, II times greater (95% CI 4.1- 28.9) for the noise and toluene group, and five times greater (95% CI 1.4-17.5) for the sol vent-mixture group. The findings suggest that exposure to the studied solvents had a toxic effect on the auditory system and that an interaction between noise and toluene took place. The audiological results of the noise and toluene group suggest a central auditory pathway involvement in the hearing losses observed. ,

Hearing loss is still one of the mo st prevalent occupational di seases in the United State s and most other industri ali zed countries (1-3). In the United State s at least one million workers in manufacturin g are estimated to have sustained job-related hearing impairment , and about a half million of the se work er s have moderate to severe he aring imp airment (I ). Occupation al hearing loss has long been reco gnized as a direct health effect of overexpo sur e to nois e, but only recently has expo sure to solv ent s been con sidered as a contributor to the development of hearing impairment (4)(5)(6)(7)(8)(9)(10). There is evidence wh ich suggests that noise interact' > synerg istica lly with various drugs and chemicals (11)(12)(13)(14)(15)(16)(17)(18). Evidence is also beginning to accumulate which suggests that an ototra umatic interaction may ex ist between noise and org an ic sol-ve nts. Data on the ototoxic effect s of solvents like tolu ene, xylene, and styr ene come mainl y fro m limited animal studies and ca se report s on subs tance abu sers (19)(20)(21)(22)(23)(24).
Such ototraumatic interaction was suggested in 1984, when its biological plau sibility was discu ssed . It has been observed that the incidence of sensorine ural he arin g lo ss was higher than expe cted in workers expo sed to solvents. Organ ic solvents are well kno wn for their neurotoxic effec ts, which, in expo sed workers, can giv e rise to both ce ntra l and peripheral injuries in the nervous system. It has been hypothesiz ed that solvents can injure the sensory cells and peripheral nerve endings of the cochlea, and, co nsidering kno wn solvent-related effects on the brain , a retrocochlear influence ca n also be ex pec ted (4) .
In a 20-year longitudinal study of the hearin g sensitivity of 319 employees, it was observed that 23 % of the wo rkers from a chemical division showed pronoun ced hearing lo ss through exposure to lower noise levels [80-90 dB (A) ] as compared with 5-8% of workers from nonchemi cal environments with exposure to higher noi se levels [95-100 dB(A)] (5) .
In the present study we evaluated , usin g pure tone audiometry, the occurrence of hearin g dis orders in gro ups of workers (i) not expose d to noise or organ ic solvents, (ii) exposed to noi se, (iii) exposed to noi se and tolue ne, and (iv) exposed to a mixture of solvents including toluene. More ov er , acoustic refl ex measurements were performed with the objective of obta ining information concerni ng the anatomic location of the obs erved hearing di sord ers.

Study design
To investigate the effe cts of occupational exposure to noise and solvents on workers' hearing, we interviewed samples of Brazilian workers from the printing and paint manufacturing indu stries, tested their hearing, and assessed their exposure to both agents. Male workers employed for a minimum of one year were included.

Study population
At the printing plant, groups of worker s from thr ee divi sions, (preparation, rotogravure print ing, and finishin g and binding) were rand oml y selected to participate. Workers were selec ted fro m the prep aration divi sion to serve as an unexposed comparison group, as they were not occupation ally expos ed to any known or suspected ototo xicant. These workers were involved in setup operations that incl uded graphic arts, composition, photocomposition, scanner operation, offset , and retou chin g. Both retro spective and current sound pressure measuremen ts in this division indicated that the noi se levels were in a range that was assumed not to pos e a hazard to employees ' hearing [below 85 dB(A)].
Work ers selected from the finishing and binding divi sion served as the noise-only expos ure group. Th e work activities included cutting, trimm ing, co llating, and binding magaz ines. Sound pressure measurements conducted during the pre sent investigation in this division were in agreement with the indu stry 's historical records. These measurements showe d continuous nois e levels in the range of 88-97 dB( A). Noise dosimetry indi cated levels that ranged from 209 to 335%. The recomm end ed limit of 85 dB(A ) and the 5-dB exchange rate were used in this evaluation. No hearing protect ors were used .
Work ers were selected from the rotogravure printing division becau se of their expo sure to both execssive levels of noise and toluene (98 % purity). Sound pressure measurements showe d noise levels in the range of 88-98 dB(A). Noise dosimetry revea led dos es ranging from 140 to 350 %. To evaluate toluene exposure , historic al data obtained by the com-pany using colo r indi cator tubes (Draege r) and data collected by the pre sent investigation (March 1990) using SKC charcoal tubes and Dupont air pump s (model P 4000) were used. Th e results are shown in table I. An assess ment of toluene expo sure was first made in Jul y 1978 . A ventil ation system was installed by Sept ember 1978. The result s of the measurements obtained in the present study, analyzed through gas chromatography, are displayed in the 1990 column of table 1.
Tho se who worked during 1978 while the ventilation system was being installed were expos ed to extremely high concentrations of toluene . A level of 1860 ppm of toluene was once registered in 1978 . It can be ob served in table I that , after the vent ilation system was installed, the concentrations of toluene in the air decreased , but they are still exce ssive when compared with the recommended threshold limit value ( 100 ppm ).
At the paint manu facturing plant , the fourth group of workers, exposed to a mixture of solvent s but not noi se, was randomly selected from the filling divisio n. Their work acti vities included placing can s on conveyor belts , contro lling their filling, and closing and labeling them. These workers were exposed to a mixture of organic solvents that contained toluene . Th e noise levels register ed in this division were below the recomm ended limit of 85 dB(A). The majo r components in the mixtu re of solvents were toluene , xylene, methyl ethyl ketone , and methyl isobut yl ketone, but their relative propo rtion varied. On three different occasi ons organic solvents were measured in the mixture throu gh passive sampling with samplers from the Minnesota , Mining and Manufacturin g Comp any, type 3500 . Th e concentration of the mixture in the air, relati ve to their respecti ve threshold limit values (TLV ) was calculated according to the formula of the American Conference of Governmental Indu stri al Hygieni sts and the Brazilian thre shold limits.
The result s displ ayed in table 2 indicate that for three of eleven samples the recommended concentration limit for mixtures was exce eded. It should be noted that for eight of the eleven samples toluene had the highest concentration. In add ition , this division had no ventilation system . Due to the lack of historical data from this division and the restricted acces s of the authors to plant operations, the result s displayed by table 2 are insuffic ient to allow an extensive evaluation of worker exposure.
To dete rmine whether underl ying differences might have existed between the studied groups, we assessed previous exposure to noi se, previous exposure to chemi cals, medical and audiological history, hobby history, and prior militar y service. The characteristics of the four study popul ations are presented in table 3.
The mean length of employment for workers from the printing plant was 8 to 13 years; for the paint manufacturing plant it was about six years (F := 11.54, P <0.0 I). Overall, the study groups comprised a fairly young sample of employ ees. The age of the workers in all of the groups was also similar to the mean age in years, ranging from about 32 to 36 years. In the analysis, age was not included as a risk factor due to the similarities of the groups and because it was correlated with the length of employment (ie, the longer one works the older he or she becomes).

Questionnaire
A standardized medical and work history questionnaire was devel oped with major portions extracted from a questionnaire of the National Institute for Occupational Safety and Health and from various clinical questionnaires. The final version of the questionnaire was translated to Portuguese. Included in the medical portion were questions concerning demographi c data , health information that focused on events that could be related to hearing status, and nonoccupational noise-e xpo sure data. A work history was collected which included j ob de scriptions and exposure to noise and chemicals. The 12-pa ge questionnaire was administered by 11 person s who were either graduate students in audiology or qualified audiologists . All of the interviewers were unaware of the indi vidual' s expo sure group. To minimize interviewer bias, all of the health questions (including tho se on hearing) wer e asked prior to the question s on work history.

Data analysis
Th e data from the que sti onnaire and the test results were ent ered into dBase III Plu s and data files using a pro gram written in Clipper. Extensive checks and rech ecks were mad e for illogical codes or consistency errors . After correction a cle an data file was read y for analys is. The data were analyzed with the use of the statistical anal ysi s sys tem (SAS, version 5,1985). The data were analyzed with multivariate and uni vari ate analy ses of variance. Analyses of associatio n bet ween the hearing sta tus and exposur e conditions were also performed. Logistic regres sion was used for estimating relative risk, adj usted for con founding variables .

Testing procedures
Pure tone audiometry. To asses s the workers' hearing status, otoscopy, pure ton e audiometry , and irnmittance audiome try were perfo rmed. Ot oscopy was performed to screen for conditions that would excl ude the per son fro m the study . All of the subjects we re inte rviewed with regard to health history, work history, and solvent and noi se e xpos ure. The tests were per formed by aud iolo gists under the prin cipal investigat or ' s supervisi on. All of the subjects und erwent pure tone audiometry at the frequencies of 0.5, I , 2, 3, 4, 6, and 8 kHz for air co ndu ction. When there was an indicatio n of conductive hearing loss, bone conduction testing was performed. The subjects were tested in a soun d-isolated chamber which met the requirements of the American National Standards Institution (ANSI S 3.1.-1991 ) for audiometric testing environment s. The audiometers used were a Medical Acoustics In struments Company MA41 and a Dan avox DB 23, and both unde rwent electroacoustic calibration a wee k befo re the data coll ection started. Biological calibration ch eck s were also performed every day im mediately before the subjec ts were tes ted.
The high-frequ ency hearing losses were class ified by sev erity, into ca teg ories I through IV according to criter ia described else where (9,25) . Th e thresholds in the frequency ranges 0.5 to 2 kHz wer e averaged . The interaural threshold avera ge of the most affected frequency in the 3 to 8 kHz frequ ency range was con sidered when a classification was assigned to the audiograms. A nono ccup ation al ca tegory was included to account for those hearing losses that could not be attributed to occupational factors. The V-C category is for co nductive hearing losses, while the V-U category is for unilate ral hearing losses. (See  table 4.) lmmittan ce audiomet ry. Immittance audiometry, another routine battery of tests at the audi olo gy clinic, was perfo rmed on all of the subjects. It consisted of tympanometry , static compli anc e, cros sed and uncrossed acoustic reflex testin g (at freq uencies 0.5, I , and 2 kHz), a reflex de cay test (at frequenc ies 0.5 , I, and 2 kHz) and a physical volume test. Th e main objective in performing immi tta nce audiome try was to obtain informatio n on lesion sites through the investigation of the acoustic refl ex findings in the studied groups . Th e aco ustic reflex tests were chos en because of their rel iabil ity and availability and beca use they can be performed in a few minu tes. Th e immittance audio meters used were Inter acou stics AZ7 and Damplex ZA28. Both und erw ent elec troacaust ic ca libration a week befor e the data collection sta rted and biol ogical calibration eve ry morning before the subjects we re tested. Table 4. Audiometric classification criteria.

Results
Eac h audiogra m was e valu ated for hea ring loss according to the crit eria described in tabl e 4. The audiograms cl assifi ed as norm al were identified by the cod e O. The presen ce of a high-frequ en cy, bil ateral hea ring loss was coded from level I (mild hearin g loss ) to IV (profo und he arin g loss). The inform ation pro vided by the bone co nduct ion audiomet ry and tympanom etr y was used to classify hear ing loss as conductive . Conductive hearing loss was identified by the V-C co de . Unil ateral hearing loss was identified as V-U . Table 5 shows the audio metric result s of all the subjects by cod e.  The prevalence of high-frequen cy hearing loss in the group exposed to noise and toluene simultan eously (53%) was higher than in the other groups (table 4): 8% in the unexposed group, 26% in the noiseexposed group, and 18% in the group exposed to a mixture of solvents. The prevalence of high-frequency hearing loss observed in the solvent-mixture group (18%) was two times higher than in the unexposed group (8%).
No statistically significant differences were observed between the group s for the classifications V-C and V-V (conductive or unilateral hearing losses). A one-way analysis of variance comparing the means of the classification 0 (normal hearing) for the studied groups indicated that they were significantly different (P < O.OOl) . The results of the analyses of variance indicated that the group exposed to noise plus toluene had significantly more workers with mild hearing losses (classified as level I) than did all of the other groups (P < O.OOI). No statistically significant differences were observed between the groups for the audiogram classifications of II, III, and IV. The percentages of several audiometric classifications [with T bars indicating upper bounds of 95% confidence intervals (95% CI)] are displayed in figure I by group. All of the subjects were included in the analysis (N =190).

Analysis of association between hearing status and expos ure conditions
The next analysis combined classifications I to IV and analyzed hearing loss as a dichotom ous variable (yes/no). The high-frequenc y hearing losses were examined with a multiple logistic regression. For this analysis, conductive and unilateral hearin g losses were entered as normal hearing, since they could not be clearly related to the occupational exposures. This test was conducted to estimate the relative risk adjusted for confounding variables . The variables considered for inclusion in the model were exposure group, length of employment, previous occupational exposure to noise or to chemicals, and exposure to nonoccupational noise. Age was not includ ed because it was highly correlated with the exposure variable length of employment, and the study population was relatively young (median age 33 years). The approaches used were the stepwise forward and backward logistic regre ssion. The only variable that met the significance level criterion for inclusion in the model, besides exposure group, was length of employment. Table 6 gives the results of the final multiple logistic regression model selected by the stepwise procedure, with the relative risk estimates for developin g a hearin g loss, as adjusted for length of employment (26). The predict ed probab ility of each    gro up for developing a hearin g loss is illustrated in figure 2.

Acoustic reflex measurements
The measure s analyzed were absen ce or elevation of the reflex (in relation to expectation on the basis of values for norm al ears or ea rs with coc hlear heari ng loss), presence of recruitme nt [ob served when the difference between the pure tone and acoustic reflex thres holds was less than 60 dB (S L»), and presence of aco ustic refl ex decay (50% reflex decay before 10 s). Initiall y, multivar iate analyses of variance were performed using the following independent variables: group, ear , frequ enc y of the stimulus, and stimulus presentation (co ntra-or ipsilateral). An overall significant differe nce was observed between the groups (P < 0.00 I). Subsequent analyses of variance (between and within subje cts, with the subjects nested in gro ups) indic ated signific ant di fferences between the grou ps regarding recru itment (P<0.05) and reflex decay (P < O.OOl), but not absenc e or eleva tion of reflex . Group contrasts regardin g recrui tmen t indicated that the mean percen tage of cases in the noise-only exposed group was sig nificantly higher than in any of the other group s. (See figure 3 for the percen tage of cases and upper limits of the 95 % CI values.) Group co ntrasts regard ing reflex decay showed that the mean percentage of cases in the noise-pl us-tolu ene gro up was sig nificantly higher than in all of the other gro ups (P <0.00 I).
An overall interaction was foun d between group, frequency of the stimulus, and stimulus presen tation (P<O. OO I). Reflex decay was the only outco me that was show n to interact significantly (P< O.OI) . (See figure 4 for the perc entages and 95% CI values .) At every test frequency the noise-plus-toluene group had a significantly higher percentage of reflex decay than the other thr ee groups; the percen tage of refle x decay was significa ntly hig he r for contralateral sti rnu- High-frequency hearing loss was also ex amined with a multipl e logistic regre ssion for the estim ation of relative risk adjusted for confounding variables. All of the exposure groups were shown to have significantly elevated risk ratios for hearing loss. These finding s strongly suggest that chronic expo sure to the studied solvents ha d a toxic effect on the auditory sys tem. Moreover, it is very likely that an otot raumatic interaction between noise and toluene took place. If a group only expos ed to toluene had been located, it would have been possibl e to inve stigate lation than for ipsilateral stimulation, and higher at 2000 Hz than at 500 or 1000 Hz.

Discussion
To assess the workers' hearin g status, pure tone audiometry and immitta nce audiometry were performed. Both noise and solve nt exposure s were mea sured. All of the subj ects were interviewed by the first author (whose native language is Portuguese) with regard to health history, which focused on hearing, work history, and solvent and noise exposure.
The prevalence of high-frequency hearing loss found in the group exposed to noi se and toluene simultan eously (53%) was considera bly higher than in the other groups (8% in the unexposed group, 26% in the noise-onl y group, and 18% in the group exposed to a solvent mixture).
Significant differenc es were ob served between the groups regarding the hearin g cla ssifications of 0 (normal hearing) and I (mild hearing loss). It was obse rved that the group expos ed simultaneously to noi se and toluene had a significantly higher percentage of cases of mild hearing loss than the other three groups. This finding raises the questi on of why the development of more serious hearin g losses was not evident in the data . One possible answe r for this que stion is that , since the workers who parti cipated in the study had relative ly short exposure times (averagi ng 5-1 3 years) , they probably were not exposed long enough for the hearing loss to progress. Another possible answer is that the natu re of the hearin g loss was such that pure tone audiometry was not the appropriate test to assess the extent of the imp airment.  the kind of interaction (additive or multiplicative) that might exist between noise and toluene . Toluene ototoxicity has been discus sed previously. Evidence indicating that toluene exposure affects hearing comes mainly from studies conducted with toluene abusers (21)(22)(23)(24) and animal s (8,(27)(28)(29)(30)(31). In rats toluene interacted synergistically with noise. Hearing losses in groups of workers exposed occupationally to toluene have been reported previously (7,32), but as a secondary finding.
Though our results of the hearing tests of the group exposed simultaneously to noise and toluene raises concern , the adjusted relative risk for hearing loss in the group exposed only to a solvent mixture has more critic al impli cation s. It was surprisin gly high (relati ve risk 5.0) and higher than the risk for the noise-only group (relative risk 4.1). The solvent-only group was exposed to a mixture of toluene , xylene, and ketone s. In rats, xylene has been reported to be a more powerful ototoxicant than toluene (33). The ketones present in the studied mixture (methyl ethyl ketone and methyl isobutyl ketone ) are presumably less toxic than toluene and xylene (34)(35)(36). This finding suggests the urgent need for research on the hearing function of workers expo sed to chemicals. The issue to be raised is whether hearing conservation regulations appropriately define who should and should not have their hearing tested.
Significant differences between the studied groups in the probable lesion site were observed through the determination of the occurrence of recruitment and acoustic reflex decay. In the present investigation the noise-exposed group had a significantly greater (P < 0.05) percentage of cases of recruitment than did the other three groups. As has been seen in other studies, this find ing suggests that the noise-only group had significantly more hearing loss of a cochlear origin than did the other three groups (37--4 1). In contra st, the noise-plu s-toluene group had a significantly greater (P < O.OOI) percentage of cases of refle x decay at every test frequency than did the other three group s. Furtherm ore, the percentage of reflex decay for the contralateral stimulation was significantly greater than the ipsilate ral stimulation. These findin gs suggest that the hearing losses in the noise-plus-toluene group had an intraaxial brainstem site of disorder. In such cases , when the uncrossed pathways are not involved on either side, the reflexes are typicall y normal to uncrossed stimulation and typically abnormal to crossed stimulation (37)(38)(39). Although such diagnostic statements cannot be made solely on the basis of these test result s, it constitutes strong evidence about the probable lesion site. This observation does not eliminate the possibilit y of the hearing loss having a peripheral component.
Groups of workers with long-term occupati onal exposure to solvent mixture s have been evaluated with extensive audiological and vestibular test batterie s (6,10). Some of the subj ects had a diagnosis of solvent-induced psychoorganic syndrome. The 252 findings of pure tone audiometry, immittance audiometry (including refle x decay), and auditory brainstem response were essentiall y normal for age and noise-exposure history. Nevertheless, a significant abnormality was found in the discrimination of interrupted speech and cortical responses to frequency glides. The outcome of the audiological test batteries suggested that long-term exposure to solvent mixture s may give rise to lesions in the auditory pathways more central than the structures generating the auditory brainstem respon ses which differ from indications of this study. This contra st may be explained by the difference in the chemicals studied, exposure conditions , and the schedul es and specific characteristic s of the studied populations.
Abnormal brainstem auditory evoked respon ses and atrophy of the brainstem , cerebellum, and cortex have all been reported for chronic toluene abusers (2 1-24, 42, 43). The regional distribution of toluene in the central nervou s system may, in part, explain these clinical observations. Levels of inhaled toluene in specific brain regions have been measured in rats (44). Toluene was detected in all brain regions with the highest concentrations at the brainstem (P < O.05). A significant correlation for regional lipid content and maximum toluen e concentration was demonstrated. The authors argued that low doses of toluene would be expected to affect lipid-rich regions of the central nervous system (medulla/pons and midbrain) preferentially and therefore lead to specific brainstem and cerebellar outflow signs. Higher doses would further suppress these areas, together with the cortex and subcort ical structure s.
In summary, the acoustic reflex findings of recruitment and refle x decay indicate that the hearing loss observed in the noise-plus-tolu ene group was not only more prevalent , but also different from the hearing loss observed in the noise-only exposed group. According to the reflex decay observations, the hearing loss observed in the group exposed simultaneously to both agents cannot be attributed merely to noise exposure. The acoustic reflex measurements strongly suggested that the site, as well as the mechanisms underlying the lesions of the group exposed to both agents, is probabl y different from those in the noise-only group. This possibility supports the idea that chronic occupational exposure to high concentrations of toluene, in combination with excessive levels of noise, might have a toxic effect on the auditory system beyond the inner ear.

Concluding remarks
In this study, simultaneous occupation al exposure to excessive levels of toluene and noise were found to increase the predi cted probability of developing a hearing Joss significantly among rotogravure printers, when compared with a group of workers exposed to matching doses of noise. Furthermore, in the paint manufactur ing industry, occupational exposure to a solvent mixture containing mainly toluene, xylene, methyl ethyl ketone, and methyl isobutyl ketone was also shown to increase the probability of hearing loss, with an adjusted relative risk greater than the one obtained in the noise-only group. The positive association between occupational exposure to solvents and hearing disorders raises serious concerns. Hitherto, occupational hearing conservation programs have not taken chemical exposures into consideration. Thus there may be numerous workers with unmet needs concerning hearing conservation.
In addition, the results from the acoustic reflex decay test suggest that there might be central auditory pathway involvement in the hearing loss observed among the rotogravure printers. This finding indicates that hearing testing not only could contribute as an early indicator of those who are more susceptible to the development of hearing loss, but also help identify those who are more susceptible to eventual neurotoxic effects of solvent exposure. The ideal procedure for studying and assessing the effects of solvents on hearing consists of a complete audiological test battery. Nevertheless, the tests used in the present study (pure tone and immittance audiometry) can be useful screening tools when hearing disorders are investigated in industrial populations in which complex exposure conditions occur.