A case-referent study on neuropsychiatric disorders among workers exposed to solvents.

C. A case-referent (case-control) study on neuropsychiatric disorders among workers exposed to solvents. Scand. j. work environ. & health 2 (1976) 14-20. Published reports give justification for the belief that long term exposure to solvents might induce chronic but nonspecific neuropsychiatric conditions. This case-referent study of data from a regional Swedish pension fund register indicated a risk ratio of 1.8 in regard to nonspecific neuropsychiatric disorders among workers such as painters, varnishers and carpetlayers who are exposed to solvents as compared to workers not so exposed. Moreover a dose-response rela tionship seems to exist between exposure in terms of occupational years and neuro psychiatric conditions, the result being that persons affected are considered eligible for disability pensions.

impaired performance in psychological tests in a ,study of groups of workers with different types of solvent exposure (such as xylene, toluene and "thinners") in comparison with nonexposed Icontrol,s. Recent reports on chronic poisoning by petroleum products (jet fuel and gasoline) (5,16) have mentioned TIOIlIspecific symptoms of the central nervous ,system such as headache, giddiness and forget£ulness, insomnia, etc., although to a major extent these investigations were ,concerned with neuropathies. In a review of the applications of neurophysiological methods in occupational medicine, Seppaliiinen (15) stated that at times symptoms in the nervous system are vague; consequently specific signs may be difficult to detect in purely clinical examinations. In view of these reports it seems reasonable to believe that possible mental and neurological disorders attributable to solvent exposure might be referred to different entities of neuropsychiatric disorders Olr simply regarded as "nervousness." 'I1hese general aspects, along with our recent clini:cal experience of complaints from painters and WOI1kers undergoing similar eJq)osure and our observations of the impaired pel'formance of painters in psychological tests (I, Hane et al., to 'be published), have led us to undertake a retrospective case~referent (case-control) study (7,13) on neuropsychiatric disorders in paiJnters, varnishers, and ca'11petlayers, an of whom are workers with relatively intense exposure to turpentine and mixtures of aliphatic and aromatic hydrocarbons, especially since the early 19508.

MATERIAL AND METHODS
Source of subjects. A regional pension fund register was used as the source of subjects for the study. Since the Swedish social security system provides a disabiJlity pension to all disabled persons, this regiJster constitutes a complete source of information about all the individuals for whom pensions have been suggested by reason of disease. Moreover medical data, including diagnoses specified with varying degrees of detail, are available through the register. Infol'mation is included on the social situation and, to a large extent, also data on years .of employment, etc., in various occupations.
An obvious drawback of the pension fund register for studies of mental disorders in relation to certain occupations is that those individuaLs who had never possessed the qualifications to become skilled workers such as painters, varnishers and carpetlayers would presumably be overrepresented among individuals considered for pension as a result of psychiatric disorders. Consequently in a casereferent study particular precautions must be taken in the selection of subjects to ensure comparability between cases and referents.
Subject selection. The characteristics of the pension fund register, as mentioned eaI1lier, called for the subjects to be restricted so that they included only skilled workers in reasonably well-defined occupations, chosen so that they had some general connection with the construction industry. The different occupations studied are indicated in table 1. They corre-spond to the class1fication in Sweden's official statistics. The age limits were set at 35 and 64 years so that young individuals without pertinent eXiposure would not dilute the material. The time-span covered the years 1969 through 1973. Geographically the iSbudy was confined to the province 0'£ Orebro (total population about 275,000).
Selection of cases. Individuals considered for a disability pension because of some tYipe of mental disol'der, also in combination with somatic disorders, were selected as cases, except when primary debility, schizophrenia, or marro-depressive psy.chosis of the ,circular type was the diagnosis. Mental diseases of obvious somatic origin, such as dementia from a traumatic brain injury, encephalitiJs, etc.. were not accepted either. Since the diagnoses of cerebral atrophy arrd presenile dementia might represent the same oonditton, it seemed reasonable to include cerebral atrophy among the cases. In the light of current knowledge, as earlier outlined, a diagnosis of headache of unknown origin or unspecified vertigo was also accepted as a case.
ALooholism was of particular tnterest in this context since symptoms arising from solvent exposure might easily be referred to the abuse of alcohol. Consequently alcoholism was also accepted as a "case" in this study. Moreover, if other neuropsychiatric diagnoses were oombined with alcohol1sm, alcoholism was considered as the main diagnosis. Otherwise, the neuropsychiatric diagnosLs reported ,first was talken as the main one. 'Dhe cases were classified as indicated in table 2, following the WHO classification (17), although a divergent distribution oould have been achieved, particularly as far as the nonspedfic cases of "nervositas" are concerned. However it should be stressed that few problems arose in dedding whether an individual was aocepta:ble as a ,case according to the particular neuropsychiatric diagnoses applied in the study (table 2).
Selection of referents. The referents were all skilled workers in the pension fund register who were completely free of any kind of mental disorder or social experience which might indicate mental disorder. No person with any type of brain injury was accepted as a referent.

Comparability of cases and referents.
The alPphcation of these relatively stri'Ct principles in the selection of subjects makes it reasonable to :believe in the achievement of an acceptable oomparability of cases and referents. Further confirmation of this assumption was gained with the use of a !random sample from the telephone directory (lind data frpm offidal statistics (Central Bureau of Statistics), as shown in table 1. However these data represent the cross-sectional view, and the table does not make any adjustment fO!I" age. Thus, although it is indicati'Ve, no full comparability exists in the figures listed in table 1, particularly between the longitudinal data (cases and referents) and the cross-sectional data. (A1mast everyone in Sweden has a telephone.) Assessment of exposure. Exposure was defined as year-s of painting, varnishing, or cllirpetlaying. It aveI'!aged about 30 years, the minimum ,being 6 months. The eXiposure was dichotomized at 30 years as indicated in tables 3 and 4.
When the duration of occupation as painters, varnishers, anld carpetlayers is regal1ded as exposure, age becomes a confounding factor (9). This is lhaI'!dly so if only -the occupa-tion as such is taken as the exposure since there is no reason to believe that skilled workers change jobs as they get older.
Statistical methods. 'I'he statistical analyses of the data were balSed upon the Mantel-Haenszel procedures . (8) for the calculation of p-values and for the estimation of the overall risk ratio. The prindplesapplied :Dor the deteI1mination of the standardized riSk 'ratios have been outlined by Miettinen (11) along with the method £or caloulating the confidence interval of the risk ratio (12).   Table 3 illustrates that the study included 151 ,cases and 248 referents. A total of 35 indiJviduals were found to have been exposed both among the cases and among the referents, i.e., they ha'<l a history of painting, varnishing, or carpetlaying. Thus 23 % of the cases and 14 0J0 of the referents hard a history of e~posure according to the criteria adopted. The crude risk ratio was .consequently 1.8 for neuropsychiatric disorders among the painters, etc., in ·relation to the 'lnonexposed," Le., to the other categories of skilled workers in th~study. In table 4 the cases of alcoholism have ,been excluded, but the relative frequeI).CY of ex:posure among the cases does not change. It ,seems unlik~ly that the distrrbution of exposed pel'SOns among the cases and referents is a'ttributable to 'chance, as table 3 Igives X 2 (1) = 5.24; p';'" 0.01 (one-tailed) and table 4 gives X 2 (1) = 4.28; p~0.02 {one-tailed).

RESULTS
A dose-response Telationship~tween the exposure and neuropsychiatric disorders is indicated by the elevated risk ratio 2 with increasing years of exposure. This relationshiIp is furthermore independent of the diagnosils of alcoholism.

DISCUSSION
Since mental disorders are problematic Nom the diagnostic 'standpoint, some of the diagnoses :might be inadequate. However, there iJs little reason to believe that deficiencies in the diagnoses ar~related to the partteular occupations studied. Therefore this type of observational 'bias will not invalidate the results obtained. Although classification of the different neuropsychiatric diaignoses in ter:rns of table 2 is 'somewhat arbitrary, it does indicate the types of disorders which might bear a crelaJtionship to exposure in painting and similar activities.
Although a diagnosis such as "nervositas" might be regarded as weak, such diffuse conditions cannot be exduded as long as no definition exists of dear-cut syndromes induced [by exposure to solvents. Again, there is no reason to be- lieve that these diffuse conditions (under the null-hypothesis) bear any particular relationshiJp to eXipOSure or noneX'Posure (in terms of occupation or occupational years), which is the proper concern in case-referent studies. E~posure was recorded by the same method for both cases and referents since the pension fund ,register provided notes on the numbers .of years spent by cases and referents in the different types of occupations.
Apart from observational problems, attention should be given to the question of wthether people in 'certain occupations might have been more readily considered for disability pension because 'Of nem:opsychiatric disorders or alternatively for other types of disease. Individuals suffering from, say, mental retardation and schizophrenia seemingly tend to remain unskilled, the result being a possible overrepresentation of such diagnoses among individuals in unskilled occupations. Accordingly these diagnoses and occupations were excluded from the study. Similarly the possibility exists that some occupations could demand great physical exertion, causing individuals suffering from 18 heart disease, osteoarthrosis, etc., to become ready ,candidates Eor a disability pension. If sUdh physically demanding oocupations were overrepresented among the referents, a relative underrepresentation of exposed persons would occur in the referent group, with a consequent increase in the risk ratio. Examination of table 1 indicates however that the physical strain invQlved in painting, varnishing, and carpetlaying corresponds to that eX!perienced in the other oc,cupations. Table 1 also illustrates reasonable consistency in the relative distribution of the different occupations among the cases and referents. When a comparison is made of the relative distribution o,f various occupations among both people with telephones and Sweden's ,official statistics, it is noteworthy that the painters, etc., might even be overrepresented among the referents, i.e., the actual risk ratio of neuropsychiatric disorders found in this study 'could be an underestimate of the true sHuation.
Consideration is aLso due to confounding fadors, but only age was accounted for by stratification in the analysis of the data. Tables 3 and 4 indicate that age Table 4. Duration of solvent exposure, i.e., number of years employed as a painter, varnisher or carpetlayer (average 30.5 yr), according to age and with the cases of alcoholism excluded, along with various estimates of risk ratio. only introduces weak con:founding since the standard morbidity ratio differs 'little :Erom the crude dsk ratio (10) even when the eXiposure takes the TOl"IIl of occupational years. It is apparent that when the ,occupation as such is used as a measure of exposure, age will not introduce a 'confiounding of any importance, as skilled wOI'kers do not tend to ,change professions as they age. Confounding bctolI's other than age can hardly be strong enough to cause a ,serious bias in the study. Moreover it was thought unnecessary to consider the influence of age in categories of less than 10 years. The observations initially discussed in this paper make it seem likely that the eXiposure to solvents involved in these particular occupations might have had an etiological significance. The most common solvents are turpentine and mixtures of aliphatic and aromatic hydrocarbons within the C 6 -C lO range, but little is known of the amount of exposure by inhalation or slkin absorption. Our experience has shown that amounts of iseveral hundred parts-per....million of such hydrocartbon solvents are not uncommon, and similar results have been found in an experimental study (4). The ex;posure to paint pigments is probably of minor importance, and, for Swedish painters at least, exposure to lead is rather unusual because of the special regulations for the utilization of lead pigments in paints. Other exposures, such as to components of modern plastic paints, substances for wood preservatLon, and so on, should be kept in mind, but currently too little is !known of such exposures to permit a relevant discussion.
One conclusion which might reasonably be drawn from this study is that neuropsychiatric disorders bear some relation to exposure in such occupations as painting, varnishing and carpetlaying. In the light of previous reports it is likely that exposures to different solvents play an etiological role.