abortions and congenital malformations among the exposed solvents.

Spontaneous abor tions and congenital malform ations among the wives of men occupationally exposed to organi c solvents. Scand J Work Environ Health 1989;15:345-352. A case-referent study nested in a cohort monitored bio logically for exposure to six organic solvents (styrene, toluene , xylene, tetrachloroethylene, trichloroethy lene, and 1,1, I-trichloroethane) was conducted to investigate the effects of paternal exposure on pregnan cy outcome. The pregnancies were identified from medical registers. The exposures of the men during the spermatogenesis preceding the pregnancies and of the women during the first trimester of the preg nancies were obtained with questionnaires, and the available biological monitoring measurements were used in the exposure assessment. Factors which significantly increased the odds ratio of spontaneous abortion were paternal exposure to organic solvents in general, high/frequent exposure to toluene or miscellaneous organic solvents (including thinners), and maternal heavy lifting. No significant association between paternal or maternal exposure and congenital malformations was found, but because of the few cases no firm con clusions can be drawn.

Or gan ic solvents are commonly used in various indu stries, but information on the reproductive effe cts of paternal expo sure is limited.In fertility (1), low birthweight (2), and congenital malformations of the offspring (3,4) have been reported in connection with paternal exposure to organic solvents.Methodological difficulties in reliably mea suring the outcome of the pregnancy and the occupational expo sure s have complicated such studies.
Paternal effects on a pregnancy and the fetu s are believed to be caused by mutagenic or toxic mechanisms.Of the studied solvents, tetr achloroethylene and the metabolite of styrene, styrene oxide, have been classified as anim al carcinogens (5).Mutagenicit y tests have been negative or inconclusive for ethylene glycol derivatives, but testicular atrophy and pathological changes in the semen of exposed male animals have been found (6).In addition, in human studies, chan ges in semen ha ve been found (7).Ch ange s ind icat ing muta genic effects in cultured lymphocytes and in sperm cells ha ve been found among met al workers exposed to trichloroethylene (8).Xylene and toluene have lacked a demonstr able mutagenic!genotoxic pot ential in vario us tests (9,10).Incre ases and decrea ses in the plasma levels of the follicle-stimulating hormone of toluene -expo sed men have been reported, and the y indicate some effect on male reproductive,health (II ,12).
In this case-referent study we investigated the effects of paternal expo sure to organic solvents on pregnan-I Institute of Occup ationa l Health , Helsinki , Finlan d.
Reprint requests to: Dr H Ta skinen , Institut e of Occupational Health, Top eliuksenkatu 41 a A, SF-00250 Helsinki, Finland.cy outcome.We also tried to determine whether any single organic solvent or a group of such solvents could be identified as harm ful.The register of biological monitoring measurements was applied for six commonly used organic solvents (styrene, toluene, xylene, tetrachloroethylene, trichloroethylene , and I, I, l-trichloroethane) to ensure reliable information on exposure .Information on pregnancy outcome was obtained from medical registers.The results suggest that preconceptional paternal expo sure to organic solvent s may increa se the risk of spontaneous abortion .

Subjects and methods
The sour ce of the study population was workers ever monitored for organic solvent exposure by the Finnish Institute of Occupational Health during 1965 through 1983.The expo sure measurements included were the mandelic acid con centration in the urine for styrene , trichloroacetic acid in the urine for trichloroeth ylene , meth ylhippuric acid in the urine for xylene, and th e concentration of tetrachloroeth ylene, toluene and I,I ,I -tri chloroethane in the blood.There were 13 132 exposure mea surements cond ucted on about 6000 men.
In format ion about the marriages and the wives of the monitored men was obtained from the Finnish Popul ation Register Centre with the use of the personal identification numbers of the men in question.The information also yielded the personal identification number of th eir wives, and this number was used to gain data from the pregnancy outcome registers.(See the follo wing text.)Onl y men in their first marriage dur-ing the year of inquiry (1985) and with wives 18-40 years of age at the end of the first trimester of pregnancy were included.Only pregnancies which had begun during the marriages or no more than nine months before them were accepted.About 19070 (774 of 4035) of the pregnancies of the wives were excluded because they had begun more than nine months before the marriages or after a divorce.Among these were 77 spontaneous abortions.
The data on the medically diagnosed pregnancies of the wives were retrieved from the Hospital Discharge Register maintained by the National Board of Health.Data on spontaneous abortions treated in the hospital polyclinics were collected separately from the hospitals.In this data base the rate of spontaneous abortion among all recognized pregnancies ( = spontaneous abortions, induced abortions, and births) varied from 7.8 % in 1973 to 10.2 % in 1983 (13).For the pregnancies of this study the rate of spontaneous abortion was 8.8 %.The following codes of the eighth revision of the International Classification of Diseases (lCD) were used for pregnancies during 1973-1983 in the record linkage: 650-662 for delivery and 643 and 645 for spontaneous abortion.
The data on congenital malformations during 1973-I982 were taken from the Finnish Register of Congenital Malformations, to which it has been compulsory to notify congenital malformations identified during the first year of life.Studies comparing the registered data with hospital information have revealed a reporting and detection failure rate of about 30 % (14).The failure rate varies by the type of the defect, minor malformations being particularly underreported.
The descriptions of the contents, coverage, and reliability of both of the aforementioned registers have been published earlier (13,14).
This study had a case-referent design.The wives with a spontaneous abortion or with a congenitally malformed child were defined as cases.If the wife had two or more spontaneous abortions, only one was randomly selected.Three referents were selected for each spontaneous abortion and five for each congenital malformation from the wives who had given birth.Only those women who did not have a registered spontaneous abortion or a registered malformed child during the study period (1973)(1974)(1975)(1976)(1977)(1978)(1979)(1980)(1981)(1982)(1983) were eligible as referents.The referents were selected by computer so that the ages of the case mother and the referent at the time of conception were as similar as possible, at least within 30 months.Only one pregnancy per woman was included.
Separate questionnaires were mailed (January 1986) in the same envelope to both spouses of the case and reference families to obtain data on occupational exposures related to the study pregnancy.The men were requested to give information on employment, occupation, and workplace during the year of conception.They were requested to describe their work tasks and 346 potential changes in them in detail and the months of changes during that year.This procedure was necessary to certify the exposure situation for the spermatogenesis period.A new questionnaire was mailed if sufficient information was not obtained in the first one.The men were also asked to report whether they had handled any of the monitored solvents or other solvents (eg, white spirit, petroleum benzine, thinner, acetone).Almost 300 tradenames of different solvents were reported by the study subjects.The chemical composition of the commercial products was checked from the safety sheets or the Register of Safety Information on Chemical Products.The frequency of the usage (daily, 1-4 d a week, less than once a week) and exposure to gases, vapors, dusts, fumes, other chemicals, or radiation in the work was requested.Questions concerning earlier employment, chronic diseases, smoking, and alcohol consumption were also included.
The information on occupational and life-style exposures during the first trimester of pregnancy was requested from the wife.Questions were asked about heavy lifting and acute febrile diseases.The pregnancy history was also recorded.Only those couples who reported the pregnancy found in the registers were included in the analysis.
The man's exposure was assessed for the time of the calculated spermatogenesis (80 calendar days) preceding the study pregnancy without the case-referent status being known.The assessment was based on the occupation, job description, reported solvent or other chemical usage, and biological monitoring data.The likelihood of exposure was categorized as follows: unexposed, when the worker reported no exposure and he was not monitored in the same job; potentially exposed, if, according to the job description, solvent usage was possible (eg, gluing), but no clear solvent exposure was reported and no exposure measurement had been conducted; likely exposed, when the job implied solvent exposure and there were exposure measurements from the same job or solvent exposure was reported.
The quantity of the exposure was estimated on the basis of the reported frequency of the use of solvents and on the biological monitoring data, when available.Only a few measurements had been conducted during the time of spermatogenesis (17 measurements) or during the preceding year (44 measurements).However, biological measurements of solvents illustrate the exposure of very short time (the same or the previous day).Thus exposure classification was mainly based on the job description, and monitoring results supported the information.The exposure was defined as "high/frequent" if the worker handled solvents daily or if the biological measurements indicated clear occupational exposure (above the reference values for the general population).The exposure was defined as "intermediate" when the solvent was used 1-4 d a week and the biological measurements indicated intermediate/low exposure.If solvent handling occurred more rarely, the exposure was defined as "low/rare."Foremen who did not take part in the manufacturing process but were prese nt in the work area fell into the category below the one assigned to them on the basis of the frequency of expos ure .If the worker's exposure to solvents in general was likely, but no specific solvent could be identified, he was p laced into the categor y "exposed to solvents in general" in the final analysis.Ho wever, as regards the speci fic solvent var iable, such a worker fell into the unexposed category.
Many workers reported exposure to a "thinner" without specifying the chemical composition of the product.According to the safety sheets, thinners are organic solvent mixtures which may contain xylene and /or toluene and other organic solvents.Therefore, reported thinner exposure was considered as likely in the variable of miscellaneous solvents and potential in the xylene and to luene variables.Exposure to other toxic chemicals an d carcinogens was assessed blind ly on the basis of the questionnaire information.A chemical was defined as a carcinogen if classified as a human carcinogen (group I), a probable human carcinogen (group 2A), or a potential human carcinogen (group 2B) by the International Agency for Research on Ca ncer (5).
The woman's expo sure was assessed similarly, but on the basis of the questionnaire information only.The lifting of heavy burdens during the first trimester of the pregnancy was classified according to a score variab le over daily averages as follows: 5-9 kg = I point per lift, 10-1 9 kg =3 points per lift, and <:: 20 kg =5 po ints per lift.Th e total score for the variable was obtained by summing the scores of lifts at work and at home.
With respect to spontaneous abortion the response rate of the cases was higher than th at of the referents (136 of 172 or 79.1 % and 370 of 505 or 73.3 %, respectively) .Some var iation was seen in the response rates in the subpopu lations monitored for specific so lvents .Among the cases the respon se rate was 95-74 010, and among the referents it was 85-70 % .The final analysis included the matched case-referent sets confirming the study pregnancy (120 cases and 251 referents) (ta ble I) .The decrease in the num bers of the final material was mostly due to nonresponse.If a case did not respond, the three matched referents were also excluded .If non e of the refe rents for a case responded, the case was excluded.Only one referent was excluded because it was not poss ible to estimate the exposure status due to insufficient information.
In the fina l material 61 % of the men held (during the spermatogenesis per iod of the study) the same job in which they had been biologically monitored for solvent exposure.Some had cha nged jobs between the measurement and conception, and some had started to work with organic solvents after the study pregnancy.The inform atio n used in the class ification is presented in detail in tab le 2. Exp osure to org anic solvents was con sidered likely for 85 % of the cases a nd 73 % of the referents; 60 % of the cases and 45 %  (11 of 48) of the xylene-exposed men, and 34 lifo (23 of 68) of the workers exposed to miscellaneous solvents were exposed to these solvents alone.Other workers were exposed to combinations of these three solvents, 28 % (27 of 97) being exposed to all three solvents.Actually, the workers were also exposed to solvents other than the ones mentioned, and therefore a complex multiexposure situation was indicated for the workplaces.
The response rate in the congenital malformation part of the study was similar for the cases (25 of 33 or 75.8 %) and the referents (126 of 168 or 75.0 %).All the respondents confirmed the study pregnancy.After the matching procedures, 25 cases and 96 referellis remained in the final analysis.Paternal exposure to organic solvents was present in 72 % of the cases and 73 % of the referents.No case mother reported exposure to organic solvents.
The odds ratios of the medically recognized spontaneous abortions for the exposures of the male workers and their wives were estimated with the logistic regression model for individually matched data based on the conditional likelihood function.The statistical significance for separate variables was evaluated by comparing the respective standardized regression coefficients with a normal distribution (15).As an illustration of exposure to multiple solvents, three solvents have been presented in figure 1.In the high/frequent exposure category there were 97 men exposed to toluene, xylene, or miscellaneous solvents.Only 25 <t/o (14 of 57) of the toluene-exposed men, 23 <t/ o

Spo nt aneous abortion
In the analysis, in which co nfounders were not controlled, likely paternal exposure to organic solvents significa ntly increase d the odds ratio (OR) of spontaneous abortion (OR 2.7) (tab le 3).For aromatic hydrocarbons the odds ratio was increas ed at a borderline level of significance (O R 1.6).The odds rat ios were signif icantly increased for xylene (O R 1.8), for the miscellaneou s organic solvents , which included thin ners (OR 1.7), and for exposure to dusts (OR 2.3).Pat ern al smoking and alco ho l consumption had odds rat ios below unity.
Of the maternal variables only heavy lifting (at home and at work combine d) (OR 1.7) and the histor y of prev ious spo nta neo us a borti on (OR 2.0) gave significantly increased odds rati os.T he mat ernal expo sure to organic solvents did not incre ase the odds ra tio (OR 1.4) of sponta neous abort ion significantly, neither did ma tern al smoking or alcohol consumption.
The odds ratio of paternal solvent exposure in general, adj usted for potential co nfo unders (paternal ex-posure to dusts, maternal exposure to organic solvents, heavy lifting, and previous spontaneous abortion) was significantly increased (OR 2.3) (table 4).Mat ernal lifti ng also increased the odds ratio significantly (O R 1.7).The odds ratio for dust exposure was not significantly increased when the confounders were controlled .a Odd s rat ios for individual organic so lvents, con trolled for po ten tial pat ernal expos ure to the same so lvent , li kely pat ernal exposure to other organic so lvents and dus ts, ma terna l exposure to solvent s, maternal heavy lifting (all), and hi story of previous spo ntaneous abor tio n. " P< 0.05.Tab le 6. Odd s ratio (OR) of spo ntaneo us abort ion in rel at ion to paternal sol vent exp os ure, mat ernal lift in g of heavy burd ens, and hi story of previo us spontan eous abortion by the len gt h of the case pregna ncy -univariate analy sis .( 95 -----_. a The model con tained the prev ious spon taneou s aborti on and maternal heavy lift ing (at work and at home) as pot ent ial confounding fact ors., P <0.05, '" P< 0.001.Low/ rare and high/frequ ent expos ure to organic solvents and high/ frequ ent expos ure to to luene or miscellaneou s org anic solvents had significantl y increa sed odds ratios (table 5).The odd s ratios for styrene, aceton e, and halogenated hydrocarbon s were below or at unit y.For xylene and aliphatic hydr ocarb ons the od ds ratio of the intermed iate a nd high/ freq uent expos ure classes were above un ity, but the res ults did not have statistical significance.In the grou p of ar omatic hyd roca rbon s the odds rat io was non signi ficantl y in-350 creased in the lo w/rare an d high/frequent expos ure classes.Th e age of the hu sband or the wife did not relate to th e risk of paternal solvent exposu re in th e interaction model.Th e proportion of solvent-exposed workers remained unchanged over the study per iod o f 1973-1 983.Th e od ds rat io of spo nta neou s a bo rtio n for so lvent exposure was slightly greater for the latter half of th e study per iod than for the earlier half, but th e difference was not statistically significan t.

Aro MIse All
Lifting hea vy burde ns increased the odd s ratio of early spo ntane ous a bortio n ( ::510 weeks) mor e than th at of the later spo nta neous abortions (2:: II weeks) (tab le 6).In the intera ction mo del the difference was stati sticall y significant (P = 0.03).In the pregnancies lasting ::510 weeks no do se-response effect was seen for heavy lifting when the lifti ng score varia ble was used .Althoug h th e history of previou s spo nta neo us abortion was associated with ear ly a bo rtio n (::5 10 weeks), th e differ ence was not significant in the interaction mod el (P = 0.07).Th e length of th e pregnancy did not infl uence th e odds ratio for pa ternal solve nt expo sure.
In the analysis by occu pati o n the wives of the painters and wood workers (eg, carpenters in co nstruction, the furniture industry, and the boa t industry) had signifi cantly incre ased od ds rati os for spo ntaneous abortion (table 7).Mu ltiexposure to man y solvents or solvent mixtures was commo n to the stud y su bjects .Ma ny combi nations of orga nic solvents were repo rted by the paint ers (figure 2); aroma tic hydrocarbons in combination with miscell aneous organic solvents seemed to be respo nsible for the increased risk of spo nta neou s ab ortion.

Con geni tal malformatio ns
In the mult ivariate analysis, in which paterna l exposure to du sts an d mate rn al febril e diseases in th e first trimester of the pregnancy were considered pote ntial confo un ders, pate rna l expos ure to organic solve nts was not associated with con gen ital ma lformations o f the chi ld, either when the who le material was include d [OR 0.6, 95 a;o con fidence interval (95 CI) 0.2-2.0]or when luxation s of the hip (N =7) were excluded (OR 1.0, 95 % Cl 0.2-4.5).For the restricted mat erial (luxation s o f the hip exclud ed) the odds rat io was slightly increased for tolu ene (OR 1.5, 95 % CI 0.4-5.4),xylene (OR 1.6, 95 % CI 0.4-5.7),and miscellan eous organic solven ts (OR 2.0 , 95 % C I 0.6-6.1),but none of the se findi ngs reached statistical significance.

Discussion
In thi s study paternal expo sure to or ganic solvent s in genera l, and to toluene and miscellane ou s organic solvents in particular , increased the odds ra tio for spo ntan eo us abortion.For xylene the increase was not significan t.However, man y workers were exposed to multiple so lvents (see figure I ) , and it ma y be oversimplified to ascribe risk to a single solvent.For the gro ups of halo genated and aliphat ic hydrocarbons, and for indi vidual solvents like acetone and styrene, no increased risk of spo ntaneous abo rtio n was found .No significa nt association with co ngenita l malformations was fou nd , but , becau se of the small number of cases, no firm co nclusio ns can be drawn .
Expos ure to to luene and miscellaneou s solvents had a significant effect on spontaneous abort ion only when the exposure was high/frequent.The dose-response relationship increased the credibility of the result s.However, th e odds ratio fo r aromatic hydroca rb ons did not follow th e exposur e levels in a do se-related mann er.Styrene may partially explain this finding since the odds rat ios for it were low and th ere were severa l styrene-exposed men in the material.
In this study, information on the pregnancy out come can be considered reasonabl y reliable since registerbased data on spo nta neous abo rtions and co ngenita l mal formations were used (13,14).Th e pr egnan cies were also confi rmed by the stud y perso ns.Th e co nfirm ati on rate for spontaneous abortion (89.7 %) was fairly similar to tha t found for nu rses (90.6 %) a nd bett er than that of laundry workers and solvent-exposed female workers (78.4 and 74.1 , respectively).(Reviewed in reference 13.)The cases of our study respo nded slightly more actively than the referents, but th e differen ce was rather small.
Response bias in investigation s using the present design to study pat ern al ef fects is not very likely.The decrease in th e numbers of cases and referents in the final material fo r analysis was due to the mat ching procedures, ie, onl y unit s fo rmed by a case and one to five referent s were accepted .If in on e un it the case did not parti cipat e, all of its referents were exclud ed , and vice versa .If no refer en t of a case responded , the case was excluded .Onl y one referent was excluded becau se exposure estimation was not po ssible on the basis of the available information.
In onl y a few earlier studies on male reproduction hav e biologic al monitoring results been used in ex-posure assessment.More often occupation al codes from census dat a, company records, o r birth certi ficates ha ve been used .(Reviewed in refer ence 16.)In our study the qualitative assessment of expo sure to organic solvents was, in our opinion reliable, since biologica l mon itor ing result s were available for 75 070 of bo th th e cases and the referents assessed as exposed to organic solvents in the spontaneous abortion part of the study.Th e employer for th e time of the studied sper matogenesis was listed in man y cases in the laborator y log book s.Thus, it was po ssible to con firm the reported data on the employment through the use of the independent laboratory data.The blind expos ur e assessment o f the unmonitored exposures also reduc ed the possibility of selective misclassification .A nonselective misclassificat ion cannot be tot ally excluded ho wever; if present , it would shift the results toward s unity.
Because of the high prevalen ce of solvent expo sure among the study population, the sta tistical power to discover risk associated with expo sure was rather high in the final study material for spo nta neo us abort ion .T he probability of det ecting a twofold risk was 73 %, an d that of detecting a 1.5-fold risk was 34 % with alpha equal to 0.05 .Yet the sma ll number of unexposed cases was disad vant ageous for the study a nd hindered man y of the relevan t comparisons.In epidemiologic studi es in which multiple comp arisons are made, so me statistically significant po sitive findin gs ar e expected purely due to chance.Several co mpar isons were mad e also in th is study , and some o f the significant associations might be du e to chan ce.
An associa tion between pat ern al expos ures and spontaneous abo rtio n has been suggested for exposure to anesthetic gases, chlorophrene, vinyl chloride mon omer, dibromochloropropane, lead, and metallurgical factory exposures.(Reviewed in reference 17.)So me stud ies have shown co ntradictory results , and non e have been conclusive.Although expo sure to orga nic solvents in the wor k enviro nment is common, we have not fo und ear lier reports on the po tential ef fects of pa tern al solvent expo sure on spo nta neous abortion.Increased risk of fathering a child with congenit al malformation ha s been connected with paternal occupations in which organic so lvent exposure is likely (painters and wo rkers in the printing industry), but neither of these studies were con clusive.(Reviewed in refe rence 18.)In a Dani sh study in which th e occupat ional particulars ca me from birth cert ificat es, painters had a 4.9 tim es higher risk of fathering a child with a malformation of the central nervous system than oth er men (4).In our study there were no malformation s o f the central nervou s system among the cases.
Exposure to organ ic solvents in industr ial wor k is common .In Finland ab out 40 % of th e men wor king in indu str y are exposed (19).Th ere fore , measure s to decrease the exposure sho uld be considered already on th e basis of existing knowledge.Th ese results remind us of the need to protect men, as well as women , from hazardous occupational exposures .

Figure 1 .
Figure 1.Schematic diagram of exposure to three different solvents at the high/frequent level.
fell into the high/frequent exposure category.

Table 2 .
Info rmatio n used for classifying the solvent exposure .
Tab le 1. Fo rmat io n of t he final material in the study of sponta neo us abortion.

Table 3 .
Odds ratio (OR) of spontaneous abortion for individual risk indicators -conditional logistic regression models, confounders not controlled.(95 % CI =95 % confidence interval) a Each of these models included a variable for potential exposure of the same solvent., P<0.05, *' P<0.01.

Table 5 .
Odds rati o (OR) of spo ntaneo us abo rtion for likely paternal expos ure to spec if ic solv ent s acco rdin g to exp osure class -matched log is tic regressi on analys is .(95 % CI = 95 % co nf idence inte rval)

Table 7 .
% CI =95 % co nf i-See the Subjects and Meth ods section for an exp lanati on of th e scoring sys tem.••P < 0.01.Adjust ed odds ratio (OR) of spontaneous abortion by paternal occupation -match ed multivariate analysis."(95 % CI =95 % co nfi dence in te rval) a