Occupational noise exposure and course of pregnancy

NURMINEN T, KURPPA K. Occupational noise exposure and course of pregnancy. Scand J Work En viron Health 1989;15:117-24. The study examined the possible relation of occupational noise exposure to adverse pregnancy outc omes. The experience of I 190 reference mothers from a case-referent study based on the Finnish Register of Congenital Malformat ions was scrutinized. Exposure to noise was blind Iy assessed from a description of the mother's workday by two industrial hygienists. Women with an esti mated level of noise of around 80 dB LAc"I" h) or higher were considered exposed. Threatened abortion was not associated with noise exposure alone, hut, when it was combined with shift work, the adjusted risk was over twofold . The adjusted risk of pregnancy-induced hypertension was twice as high among the mothers exposed to noise in shift work, and the duration of their pregnancy was shorter. The analyses produced indications of a relat ion between noise and growth retardation which was not connected with shift work. There were significantly ascending trends in the proportions of these outcomes according to increasing exposure intensity.

Expo sure to environmental aircr aft noise has been associat ed with low birthweight (1-3) , reduc ed length of pregnancy (4), teratogenic risk (5), toxemia of pregnancy (see reference 2), and a decline of human placental lactogen (2). Simila r results relat ed to noise have been reported for pregnant laboratory animals, and some possible biological mechanisms ha ve been suggested . (See references 6 and 7.) Reproductive and obstetrical effects of occupational noise exposure have not been studied to a great extent. In one study , however, an excess of hormonal disturbances, delayed conception , and infertility was reported amo ng female workers expo sed to noise (8).
McDonald et al (9) reported an association between expo sure to noise and spontaneous abortion in their Montreal study, but this relation was not confirmed in their seco nd analysis of the same dat a . Heidam (10) nominated noise and shift work as possible explanation s for the increased risk of spontaneous abortion among Dani sh factory workers . In a French study noise was an element of a composite score of fatigue which correlated with preterm delivery (I I). In the Montreal study noise was associated with low birthweight in the health and manufacturing sectors (12). In a Finnish study occupational noise exposure was not a risk factor of preterm birth or low birthweight, but , as the author s stated , the study had a low po wer to det ect mod erate risks (13).
To study the effects of occupation al expo sure to noise on threatened abortion , pregnanc y-indu ced Reprint requests to: Ms T Nurminen, Departm ent of Epidemiology a nd Biostati stics, Institute o f Occup at ional Health , Topeliuk senk atu 41 a A, SF-00250 Helsinki, Finland .
3 hyperten sion, length of gestation, and birthweight, we have used the information available on the evaluated noise exposures of the reference mothers from a Finnish case-referent study on birth defe cts and exam ined the cou rse of the mothers' pregnancies.

Subjects and methods
Our general approach to studying the effe cts of exposure to noise during pregnancy has been described in a communication that addressed teratogenic risk (14). The study included I 475 Finn ish mothers of infants with structural birth defects and the ir time-and area-matched reference mothers obtained from the Finnish Register of Congenital Malformations between 1976 and 1982. In the present study of pregnancy outcomes other than malformations we used inform ation on the noncase mothers' experience. This procedure has been described in more detail elsewher e (15).
The mothers were interviewed, according to standard Register practices, about the particulars of th eir latest and previous pregnan cies, consumption of dru gs and alcohol, smoking habit s, etc (16). We obt ained information pert ainin g to the course and outcome of the pregnan cy under study from both the antenatal record of the maternity health care center and the Register que stionnaire.
The qu estion o f whether the mother had had a threatened abortion during her pregnancy was asked in the Register interview by a midwife. We considered bleeding with or without pain as an indica tion of a threatened ab ortion but not lower abdomin al pain only. For 73 % o f the mothers who had experi enced a threatened abort io n , th e sympto ms had occurr ed in the first tr imester.
An increase of at least 20 mm Hg (3 kPa) in the mean arterial blood pressure between the mother's first visit to the maternity health care center and her last visit before delivery was considered to indicate pregnancyinduced hypertension (17). The mean arterial blood pressure was calculated as one-third of the pulse pressure plus the diastolic pre ssure .
The length of gestation was calculated from the first da y of the last normal menstrual period and was expressed in completed days. The birth was considered preterm when the duration of gestation was less than 37 completed weeks , and a prolonged pregnancy was one lasting 42 weeks or longer. For the definition of a small baby for his or her gestational age, we used as a reference value the tenth percentile birthweight of the babies of the same gender born to mothers who were not exposed to noise in the same gestational age group. The applied gestational age groups were < 37 weeks, 37-39 weeks, 40-41 weeks, and~42 weeks.
The Register data were supplemented with detailed interviews on the mothers' work conditions and various exposures related to occupational or leisure-time acti vities (18,19). Among the I 475 noncase mothers, there were 270 who had not worked during their pregnancy. We excluded these mothers from the study. Fourteen of the mothers who had worked had twin births, and in the case of one woman information on the number of fetu ses was missing; these 15 women were also excluded. Of the remaining mothers, I 044 had worked throughout most of th eir pregnancy, and 146 had worked only temporarily. Of the latter group 87 % had worked in the first trimester and the rest only later during their pregnancy. The reason for stopping work was the termination of employment for 51 070 of these mothers, maternity leave or vacation for 10 % , and sick leave or overstrain for 19 %; the rest had other reasons or the reason was unknown . When we studied the occurrence of threatened abortion, all the mothers who had worked were part of the analysis. When other aspects were under consideration , we included only mothers who had worked throughout most of their pregnancy.
Exposure to noise was inquired about in a fixed question in the special interview, and additional information was obtained from the description of the mother's workday, which had been recorded in detail. The noise exposures were assessed by two industrial hygienists. (See reference 14.) In all, 429 mothers (36 % of those who had worked during pregnancy) reported exposure to noise . According to the hygienists' assessments 102 moth ers (9 % of those who had worked) had been exposed to an 8-h equivalent continuous A-weighted sound level (L AeQ (8 h» of around 80 dB or higher. Ninety-five (9 %) of the mothers who had worked throughout most of their pregnancy were assessed as exposed. The exposure of two of these mothers had taken place only in the first trimester, while all of the other mothers had been exposed during their entire pregnancy. Seven (5 %) of the mothers whose work was temporary were considered exposed, and , of these seven, six had been exposed in the first trimester and one later during pregnancy. The hygienists used preselected exposure categories in their assessment. The selected no ise levels were considered to represent ordinally scaled exposure categories of (i) " no " appreciable expo sure (clearly less than 80 dB L AeQ (8 h»' (ii) exposure of "low" intensity (around 80 dB L AeQ (8 h» ' (iii) exposure of " moderate" inten sity (around 85 dB L A eQ (8 h»' and (iv) exposure of "high " intensity (around 90 dB L AeQ (8 h) or higher). According to the hygieni sts 59 mothers had been exposed to low, 35 to moderate, and 8 to high intensity noise. The reported use of hearing protectors during pre gnancy was 15 % among the mothers with low, 60 % among those with moderate, and 87 % among the mothers with high exposure. The category of high exposure intensity was pooled with the moderate group because of insufficient numbers for a separate statistical analysis.
Nearly all (97 %) of the mothers who were exposed to noise belonged to the socioeconomic category of manual workers (according to reference 20), whereas only 26 % of the unexposed belonged to the same category. Moreover, almost 90 % of the exposed mothers were employed in manufacturing and related occupations, and none of them had been in agricultural work. Control of confounding was done by additional analyses in which the subjects were restricted, first, to nonagricultural manual workers and, second , to mothers in manufacturing and related jobs.
The physical strain of the occupational acti vities of the mothers was evaluated with a standardized method (21) reflecting energy expenditure. When we studied the associations between noise and pregnancy outcome s, we looked at the possible modifying or confounding effects of shift work, physical work load, and exposure to solvents on these relations. For threatened abortion we controlled also for temporariness of employment.
The mothers who were exposed to noise were younger, had experienced more adverse pregnancy outcome s, more often had a low prepregnancy weight, were more often regular smokers, but con sumed less alcohol during pregnancy than the unexposed mothers who had worked during pregnancy (table 1). For threatened abortion we considered the possible confounding effe cts of maternal age , parity , outcome of pre vious pregnancies, history of menstrual irregularity, intake of drugs, alcohol consumption, and smoking. For pregnancy-induced hypertension we adjusted for maternal age, parity , outcome of previous pregnancies, alcohol intake, and smoking. The model s for gestational age included the covariates maternal age , parity , outcome of previous pregnancy, history of menstrual irregularity, alcohol intake, and smoking. When studying the occurrence of babies small for their gestation al age, we controlled for maternal age, parity, outcom e of previous pregnancies, maternal prepregnan- Table 1. Background characteristics of all the employed mothers with and without exposure to no ise . cy weight, alcohol intake, and smoking. In the preliminary scrutiny of the data, other factors, such as mother's residential area and obesity, were also considered , but they were not included in the final analyses.
Estimates and confidence limits were calculated for the risk ratios according to the chi-square function procedure presented by Miettinen & Nurminen (22). To test for trend in the proportions according to the ordinal level of noise exposure, we used the chi-square statistic of Armitage (23). Power calculations for single 2 X 2 tables were carried out with methods deri ved by Miettinen (24,25) with a significance level of a =0.025 and a one-sided test.
We calculated the estimates of the adjusted risk ratios from the results of binomial regression analy ses executed with the generalized linear interactive modeling (GUM) program and the macros written by Wacholder (26). To relate the numeric variable gestational age to no ise exposure, we fitted normal linear regression models. The distribution of the outcome variate was very skewed toward shorter gestational age and had a high peak around the modal va lue. To fulfill the assumption of normality, the outcome var iate was subjected to a normal sco ring procedure given by Blom (27). The standardized normal scores were calculated, and the linear regres sion models were fit with the SAS software system (28,29).

Trend in proportions
a The ref erence cat egory was no noise expo sure in shif t work " The reference categ ory was no noise exposur e in nonsh ift work.
e Ther e were four mothers in the unexposed group with rnlssing data on symptoms. Table 3. Adjusted ris k ratio estimates and associated 95 % confidence intervals for threatened abo rtion among all the women with exposure to noise , as assessed by the hygienists, in sh ift and nonshi ft work during pregnancy and among the correspondingly exposed nonagricultural manu al workers.
posed in shift work (

Threatened abortion
Twel ve percent of the 429 mothers who had reported exposure to noi se had experienced vaginal bleeding, whereas the corresponding value was 8 l1Jo for the 761 mothers not reporting such exposure. The crude risk ratio was 1.4 with a 95 % confidence interval (95 % CI) of 1.0-2.0. The crude risk rat io for vaginal bleeding in the first trimester was 1.6 (95 % CI 1.1-2.5) for noise exposure in the first trimester, as reported by the mother. Fifteen percent of the 102mothers who were exposed to noise according to the hygienist s had suffered from vaginal bleeding during their pregnancy, as against 9 % of the 1 088 unexposed mothers. The crude risk ratio was 1.6 (95 % CI 1.0-2.6). Table 2 shows the proportion s of mothers with vaginal bleeding according to the level of exposure intensity . There was a significant trend in the proportions, and the risk ratio for the mothers with moderate or high exposure was 2.0 (95 % CI 1.0-3 .7) when they were contrasted to the unexposed mothers. The crude risk ratio was 2.2 (95 % CI 1.3-3.6) for noise exposure and threatened abortion in the first trimester.
Upon adjustment for work conditions and maternal background characteristics, the mothers exposed to noise in shift work had an elevated risk of threatened abortion as compared with the mothers une x-When the analysis was restricted to the socioeconomic category of manual workers and to nonagricultural work, the adjusted risk ratio for noise exposure in shift work was even more elevated (  (8 of 27 women) among the mothers who worked in manufacturing and related occupations and were exposed to noise in shift work, while none (of 6) of the unexposed mothers in shift work in manufacturing jobs had experienced the symptoms.  a There were two mothers among the exposed and 15 mothers among the unexposed with missing data on the length of gestation. 120

Pregnancy-induced hypertension
The mean blood pressure of 10 070 of the mothers who reported exposure to noise and 6 % of the mothers who did not increased during pregnancy to a level suggesting pregnancy-induced hypertension (risk ratio 1.6, 95 % CI 1.0-2.4). The occurrence of pregnancy-induced hypertension was 13 % for the mothers assessed as exposed by the hygienists and 7 % for those assessed as unexposed (risk ratio 1.8, 95 % CI 1.0-3.0). The relation between noise and elevated blood pressure could be seen in the shift work group, in which 23 % of the mothers with exposure of moderate or high intensity had had pregnancy-induced hypertension (table 4). In shift work, 17 % of the exposed mothers who had worked without hearing protectors had experienced pregnancy-induced hypertension (crude risk ratio 2.5, 95 % CI 0.6-8.1), and for the mothers who had used protective equipment the corresponding percentage was 24 (crude risk ratio 3.5, 95 (110 CI 1.2-8.9). When adjustment was made for possible confounding factors, the risk ratio for noise exposure in shift work, as contrasted to noiseless shift work, was 2.5 (95 % CI 1.0-6.6). In the restricted group of nonagricultural manual workers the adjusted risk ratio for noise exposure in shift work was 1.7 (95 % CI 0.5-5.8), and for manufacturing work the corresponding adjusted risk ratio was 2.4 (95 % CI 0.3-23.1).

Length of gestation and birthweight
The median length of gestation was 281 d for the mothers who reported noise and 280 d for those who did not, with interquartile ranges of 12 and II d, respectively. The occurrence of preterm birth was 2.3 % in the former group and 2.8 (I/o in the latter. The median birthweight of the babies born to mothers who reported noise exposure was 3 520 g, and the tenth percentile birthweight was 2 960 g. For the babies of the subjectively unexposed mothers the corresponding birthweights were 3 570 and 3 000 g, respectively. The exposed and unexposed mothers, as assessed by the hygienists, had similar proportions of preterm, term, and prolonged pregnancies (table 5), but the power of the study to detect a twofold risk for preterm birth was only 35 %. However, the mothers exposed to noise in shift work had a shorter gestation (median 278 d) than the unexposed mothers in shift work (median 283 d). The adjusted difference in the means of the standardized normal scores of gestational age was -0.34 with a standard error (SE) of 0.21 for any exposure against no exposure to noise in shift work. In the restricted group of nonagricultural manual workers in shift work the corresponding medians were 277 and 284 d, respectively; and the adjusted difference of the normal scores was -0.50 (SE 0.25), indicating a significant deviation. In manufacturing Table 6. Occurrence of babies who were small for their gestational age among the mothers in nonshift work with and without exposure to noise in the groups of all mothers who wor ked throughout most of their pregnancy, nonagricultural manual workers, and mothers in manufacturing and related occupations. a 95 % confidence interval in parentheses. b There was one mother among the exposed and 18 mothers among the une xposed for whom the size for gestational age could not be defined . C There were four mothers among the une xposed for whom the si ze for gestational age could not be defined. d There were two mothers among the exposed and one mother among the unexposed for whom the size for gestational age could not be defined .
shift work the median length of gestation was 1 d shorter for the expo sed than for the unexposed . The median birthweight of the babies born to all exposed mothers was 3 470 g, and for the babies of the unexposed mothers it was 3 550 g (table 5). In nonshift work, the crude risk ratio for the exposed mothers giving birth to babies that were small for their gestational age was 1.4 (95 010 CI 0.8-2.6) (table 6). Of the exposed mothers in nonshift work who had not used hearing protectors, 12 010 had given birth to a small baby (crude risk ratio 1.2,95 010 CI 0.5-2.5), and this percentage was 23 for the exposed women who had worn protectors (crude risk ratio 2.4, 95 010 CI 0. 8-5 .5). The restriction of the study group to the socioeco nomic class of manual workers and to nonagricultural work yielded the crude risk ratio of 3.3 (95 010 CI 1.7-6.4) for the po ssibility of the exposed mothers in nonshift work giving birth to babies small for their gestational age (table 6), and there was a significant trend in the pro portions according to exposure intensity (table 7). The relation between noise and small size for gestational age was repeated in the group of manufacturing nonshift workers (tables 6 and 7).

Placental weight
There were no stati stically significant differences between the placental weights of the stud y gro ups. The mean placental weight was 596 (SD 119) g for the mothers who had worked throughout mo st of their pregnancy and had been exposed to noise according to the hygieni sts' assessments, and 617 (SD 127) g for the corresponding group of unexposed mothers. In both groups 4 010 of the mothers had placentas weighing 400 g or less. Table 7. Oc currence of babi es who were small for thei r gestatio nal age according to t he intensity of exposure to noise in nonsh ift work in the groups of all mothers who wor ked throughout most of their pregnancy, nonagricultural manual workers, and mothers in manufacturing and related occupations .

Discus sion
In our study, expos ure to noise of a level of around 80 dB Li\cq(s h ) or higher appeared only at work . The hygienists did not consider any of the unemployed moth ers who reported noise exposure as exposed. Furthermore, mo st of the expo sures had taken place in manufacturing and related work, and the expo sed mothers belonged to the socioeconomic category of manual workers. Our concern was to choose noi seless work that would be otherwise comparable to nois y work in term s of potential risk factors of unfavorable pregnancy outcomes. This requirement of comparability of effects led us to analyses in which the reference to noisy work was a subdomain of nonexposure.
(See reference 30, p 30.) Besides, noisy work could be selective. Through the restrictions we also aimed at contrasts in which, in the absence of the exposure, the compared mothers could be expected to have similar pregnancy outcomes after control for the possible confounders in the analyses. (See reference 30, p 31.) First, we restricted our study to mothers who had worked during pregnancy. In addition to the broadest reference category of noiseless work in all occupations, we referred to noiseless nonagricultural work when the mothers belonged to the socioeconomic category of manual workers. The largest groups excluded from this comparison were clerical workers in offices and mothers in nursing and health care occupations. After this restriction over 60 0,70 of the unexposed mothers were engaged in service work, mainly in housekeeping and cleaning jobs, and some 30 % did manufacturing and related work. On the other hand, almost 90 % of the exposed mothers were in manufacturing jobs, and the rest did service work. Thus it seemed that the most valid contrast would have been that between noisy and noiseless work in manufacturing occupations. However, our study population was limited so that, in the sharpest contrast, the unexposed reference group became very small and was not ideal in terms of the efficiency of the statistical evaluation.
Because it has been suggested (31) that for the indication of human stress the perception of noise would be more important than the actual noise intensity, we analyzed also subjectively reported noise. Mothers in all socioeconomic categories and in all major occupational groups reported exposure to noise. The best reference for this exposure seemed to be all subjectively noiseless work.
The risk of threatened abortion was slightly elevated among the mothers who reported exposure to noise in response to the fixed question in the questionnaire, and the group that was assessed as exposed by the hygienists had similarly experienced more symptoms. In addition, there was an ascending trend in the risks according to increasing intensity of exposure. When work conditions were taken into account, it was revealed that, in our data, threatened abortion was related to noise only in shift work. The mothers in shift work had an evidently elevated risk when exposed to noise clearly over the level of 80 dB L Aeq (8 h). The restricted analyses also suggested an elevated risk of threatened abortion involved in exposure to noise in shift work.
The proportion of women with changes in blood pressure suggesting pregnancy-induced hypertension was slightly higher among the mothers who reported exposure to noise, and it was clearly higher among the mothers who were exposed to noise in shift work, as judged by the hygienists, than among the corresponding unexposed groups. There was a trend in the propor-122 tions according to exposure intensity. In addition, the restricted analyses yielded elevated risks in shift work. However, in nonshift work, our data showed no relation between noise and pregnancy-induced hypertension.
The Finnish Maternity Act grants a benefit to mothers visiting the maternity center before the end of the fourth month of pregnancy. The typical time of the first visit is around the 12th week of pregnancy; over one-half of the mothers usually register before the end of the first trimester and most before the 20th week of pregnancy (16). Thus there was variability in the time of the first blood pressure measurements, but a connection with noise exposure that could have biased the observed results does not seem plausible because of the comprehensive maternity care system.
In the aggregate, the duration of pregnancy did not differ between the mothers exposed to noise and those unexposed, and this result applied to both subjectively and objectively assessed exposures. However, in shift work the mothers exposed to noise had a shorter gestation than the unexposed.
The birthweights of the babies born to mothers who reported exposure to noise and those who did not were similar. However, the mothers who were assessed as exposed by the hygienists had more infants that were small for their gestational age when their babies' birthweights were compared with the birthweights of the babies born to the unexposed mothers. The slight excess of small babies was not statistically significant when noisy work was compared with all noiseless work, but in the restricted comparisons the exposed mothers had statistically significantly more small babies. This relation was not connected with shift work, like the previous ones, and the association could be seen among mothers in nonshift work.
The use of hearing protectors did not seem to safeguard against the untoward phenomena under study. The wearing of protective equipment was clearly related to the level of noise and, perhaps also, to the mother's perception of noise. On the other hand, we did not know if the mothers had used protectors all the time while at work.
The exposed group in shift work included relatively more mothers with a moderate or high exposure level. However, this finding did not explain the modifying effect of shift work on the relation of noise to threatened abortion or pregnancy-induced hypertension, because mothers exposed to moderate or high levels of noise also belonged to the exposed group in nonshift work. (See table 4.) Neither could exposure to impulse noise explain the modification because there were only four mothers in shift work and two mothers in nonshift work who had been exposed to this type of noise. There were no remarkable differences in the physical strain of the occupational activities of the mothers exposed to noise and those unexposed, but noise exposure correlated positively with exposure to solvents. The obtained results concerning noise we re not a lter ed whe n we controlled for these po ssible co nfo u nders .
The background char acteristics differed between the mothers who we re exposed to noise and those who were not. (S ee table I.) These facto rs were allo wed for in the a na lyses and , in the r estricted co m pariso ns , the mothers were more similar. Still, as always in a nonexper imental st udy, ther e re m a ine d the possibilit y that the findin gs we re attributable to s o m e unrecognized co nfo u nd ers .
M aternal memory bias is a possibility when inform ation is acq uired re t rospectively . T he ex posure interviews in our st udy wer e done after th e d el ive ry. However, the a ctua l as sess m ent of noise exposure was done b y industrial h ygienist s, and re sults o f noise measurem en ts at the mother ' s wo rkplace, or a t comparabl e wo r k pla ces, were available in almost all cas es. (See reference 14.) There fore the presence of th is bia s was unlik el y .
Thus, in our data , threa tened abortion a nd pregnancy-i nd uced h ypertension were not re lated to no ise alone, but, with the additional strain caused b y shift wo rk, t he moth er s ex p os ed to noise had elev a ted ri sk s o f these outcomes and, also , a sho rter gestatio n. Moreo ver, our re sults ind icated a rel ation bet ween noise and reduced prenatal growth. Whether th ese findings represent true bi ological e ffects or were in trod uced b y undetected fa ctors sho uld be e xamined in future studi es .