Effect of occupational noise on the course and outcome of pregnancy.

OBJECTIVES
The goal of this investigation was to examine the effects of occupational noise during pregnancy prospectively.


METHODS
The exposed group [continuous A-weighted sound level (LAeq(8 h)) > or = 78 dB] consisted of 111 pregnant women, and the reference group comprised 181 pregnant women with approximately similar work conditions but without noise exposure. The noise-exposed women had more frequently other inconveniences in their work, however, like shift work, impulse noise exposure, vibration, and a high or low temperature.


RESULTS
With the limit of 78 dB (LAeq (8 h)), the course and outcome of pregnancy did not differ between the groups. When the noise exposure was 90 dB (LAeq (8 h)) or more, a decline in birthweight, either absolute [mean 3304 (SD 585) g for the exposed versus mean (SD 548) g for the unexposed, 95% CI of mean difference -471--+15 g] or related to the gestational age (below the 10th percentile [5 of 25 (20%) versus 13 of 180 (7%)]), was seen. These findings were more pronounced if the woman was simultaneously exposed to a standing work position or shift work.


CONCLUSIONS
Working in high noise exposure can be considered a form of risk during pregnancy.

ly airpo rt noi se, with pretenn birth, low birthweight, and malformations , but the associ ation with malformati ons ha s not always been confirmed (7)(8)(9)(10)(11). The associati on of occupational no ise exp osure with low birth wei ght and preterm deli very is so mew hat contro ver sial on the ba sis of two recently published articl es (12-1 3).
Ou r prospective cohort study was undertaken to evalu ate the effect of occupational noise exposure on the co urs e and outcome of pregnancy , especially on maternal blood pressure, preterm birth, bi rthweight, and mal formations.

Subjects and methods
Th e s ubjects were enrolled from workpl ace s with no ise ex pos ure in the province s of Oulu, Lapland, and Harne in Finland. Oc cupation al health officers were info rmed about the research, and they in turn informed the women in the workp lac es. Enrollment to ok pl ace between April 1983 and December 1987. A measur ed 8-h equivalent continuou s A-weighted so und level (L Aeq (8 hi) of~78 dB was se lec te d as the criterio n for noi se exposure for a woman to be rega rded as an ex posed subject.
The wom en contacted the researc h group voluntaril y at the beginning of pregn ancy, and the group in turn co ntacted the respecti ve matern ity health ce nter.
The nurse in the matern ity health center selected one to three unexposed moth ers as referents for each exposed mother , matched by age (± 3 years) and parity (l = nullip ara , 2 =primipara-tripara, 3 =quadripara or more) according to the instructions of the research group . Th eir work conditions were also to be as similar as possible but without noise exposure.
The une xposed group was somewhat younger [mean 27.7 (SD 5.3) years] than the exposed group [mean 26.5 (SD 5. 3) year s]. Nulliparous subjects made up 45 % of the unexpo sed group, parity being, on the average, higher among the exp osed gro up. Social class, as j udged accord ing to the husband' s occupational status, and accordin g to the women's own status in the case of unmarri ed women, tended to be somewhat lower amon g the exposed women. The two groups did not differ with regard to their obstetric history (ie, spont aneous or induced abortions, earli er preterm deliveries, and malformed or stillborn infants). The prevalen ce of chronic diseases possibly influ encing the cour se of the current pregnancy was also similar. The groups did not differ as to their reported drinking and smoking habits (table I).
The final popul ation consisted of I I I exposed and 18 I unexp osed pregnant women. Fifty-thr ee of the exposed women had one referent , 46 had two referents, and 12 had three referents. There were problems in finding sufficie ntly concordant referents. All of the mother s were monit ored in the same way during pregnancy. The data on the course of pregnancy were obtained from the maternity health centers and also from the maternity outpatient clinics of the hospitals. The data on the deli veries and the neonates were co llected from the hospital records. The occupation al health officer at each work place filled out a questionnaire co ncerning the work conditi ons and measured exposures and work loads, and these data were chec ked at the Oulu Regional Institut e of Occupational Health. After delivery, the women were also asked to answer a postal questionnaire with regard to daily habits and soci al conditions. Six exposed and four unexposed mother s did not respond to the questionnaire.
The occupations of the exposed and unexposed women are shown in table 2 acco rding to the standard industrial classification of occupations (16 ). The largest occupational groups among the exposed women were pro cessing of food (N = 19) and textiles (N = 55 ) and among the unexposed women the largest corresponding groups were the processing of textiles (N =23), retail trade (N =31), restaurants and hotels (N = 17), and publi c services (N =27).
Almost all of the wome n in the expo sed group ( I 10 of I I I) were manual workers in term s of occupational statu s, the remaining one being of lower-grade staff status. In the unexposed grou p 68% were manual wor kers and 3 I% were lower-grade staff. One person in the unexposed group was selfemployed ( 17) . Although there were more lower-Scand J Work Environ Health 1994. vol 20, no 6 grade staff in the unexposed group, the actual types of work were simil ar in the two groups .
The average time elapsing before the first contact with the maternit y health center was 10.9 (SD 2.5) gestational weeks in the exposed group and 10.4 (SD 2.2) wee ks in the unexposed group. The mean numbe r of conta cts with the matern ity health center was 14.2 (SD 3.9) versus 14.6 (SD 3.1), respectively.
The mean height of the women was similar for the expo sed [163.7 (SD 5.3) em] and unexposed [163.5 (SD 5.1) cm] group s, whereas the exposed women were somewhat heavier at the beginning of pregnancy, 62.3 (SD 7.9) kg versus 61.0 (SD 9.4) kg. Th ere was a neglible difference in mean weight gain during pregnancy between the exposed and unexpo sed women [mean 12.3 (SD 4.3) kg versus 12.8 (SD 4.3) kg], but not in the mean hemoglobin conce ntration at the beginning of pregna ncy or at the last exami- Table 1. Background characteristics of the noise-exposed and unexposed women.  Table 3. Work load and exposures in the noise exposed and une xposed groups. The work loads and exposures were classified according to the principles used by the Oulu Regional Institute of Occupational Health (14). Impulse noise exposure was not measured, but it was classifi ed into three levels (none, moderate, considerable) acco rding to the report s of the occ upatio nal health officers.
The measured work loads and exposures are presented in table 3. The women's own opinions regarding their work loads differed from those of the occupational health officers. The physical work load, for exa mple, was j udged to be heavy by 32% of the exposed subjects and 24% of the unexposed ones, and, correspondingly, heavy men tal load was reported by 14 and 9%, respectively, these figures being substantially higher than those give n by the occupational health officers.
Blood pressure was measured at every visit to the maternity health cent er and in the hospital. The means of the systolic and diastolic blood pressures were calculated separately for eac h trimester of pregnancy.
The exposed and unexposed women were compared with respect to the outcome variables with the use of joint stratification by age and parity. The se comparisons were also performed separately for certain possible confounding or modifyi ng work conditions, such as vibration , a stand ing work position and shift work. The contrast with the unexposed women was further evaluated in subgro ups of the exposed subjects defined by noise level (low dose = noise exposure <9 0 dB L Aeq l8 hl and high dose = noise expos ure 2?: 90 dB L Aeq 18hi) and by the presence or absence of impulse noise.
The means and stan dard deviations for continuous outcomes (diastolic blood pressure, birthweight, and height ) were ca lculated in the approp riate gro ups. T he adjusted mean difference between the exposed and unexposed subjects was calc ulated as a preci-sia n-maximizing weighted average of the stra tumspecific differences, on the assumptio n of a constant error varia nce over the strata. Counts and percentages were obtained for the occ urrence of preterm birth , low birthweight for gestational age [below the l Oth percentil e ( 15)], malform ations, and care at a neonata l unit. No weight summary estimate s for the differe nces in the proportions were calculated, as the data became too sparse after stratifica tion.

Results
Thirty-two percent of the exposed women were subjected to considerable impul se noise in their work, as were two women in the unexposed group, although their 8-h equivalent continuous A-weighted sound level was still below 78 dB . The women exposed to considerable impulse noise (N =35) more ofte n worked in a standing position than the other noise-exposed women (74 versu s 56%).
Forty-eight percent of the women in the impulsenoise gro up were exposed to noise of more than 89 dB L Aeq (8 h) versus 14% of the other expose d women, and 60% of the impul se-noise gro up were also exposed to vibratio n, as co mpared with 14% of the others. Altoge ther 51% of the noise-exposed women reported using hearin g protectors for over 80% of their worktime, and 39% of them for ove r 95%. All except one woman working with noise exposure exceedi ng 89 dB reported the time spent using protectors to be over 95%. Nine percent of the exposed women stated that they did not consider their work to involve exposure to noise. Seve nty women in the exposed group had been working in noise for at least three years .
None of the women in the exposed or unexposed group had chro nic arterial hypertension, and the mean systolic and diastolic blood pressures did not differ between the group s, either in the overall comparison or when stratified separately by the presence or absence of given work cond itions (ie, vibration, standing position or shift work) (table 4). Similarly, no differences were found when the exposed women were further subdivided by the presence of impulse noise and the level of noise (table 5). Antihypertensive medication durin g pregnancy was prescribed for only two women in each gro up, and sic k leave on account of elevated blood pressure was prescribed for four in the exposed gro up and lO in the unexposed gro up (4 versus 6%).
The numbers of women admitted to a prenatal hospital ward and the main reasons for admitta nce are presented in table 6. The mean number of days spent in a prenatal ward did not differ grea tly between the groups [7.2 (SO 11.0) d versus 5.6 (SO 5. 1) d]. The mean duration of sic k leave was 3.7 (SO 4.3) weeks in the exposed gro up and 3.3 (SO 3.9) weeks in the unexposed gro up, and the mean worktime during pregnancy was 29 .6 (SD 5.8) and 29.9 (SD 5.5 ) weeks , respec tive ly.
The various outcomes of pregnancy are summarized in table 7. The mean gestational week of delivery [39.1 (SD 2.1) versus 39.2 SD 1.7)] and the number of preterm deliveries were equal in the two groups. However, four of the five preterm deliveries among the exposed mothers occurred in the highnoise group; in other words, 16% (4 of 25) of this particular exposure group had a preterm delivery. These four individuals were also exposed to vibration and a standing position in their work, and three of them also to impulse noise and shift work.
There were no differences between the groups in the prevalence of low birthweight for gestational age (below the 10th percentile), mortality (table 8), or Table 4. Diastolic blood pressure in the third trimester of the women exposed to occupational noise and of the unexposed referents. Means, standard deviations (SO), the numbers of women (N) in the groups, and the adjusted difference (stratified by age and parity) in group means with a 95% confidence interval (95% CI). Overall comparison and subdivided separately by the presence of vibration, standing position, or shift work.  Table 5. Diastollc blood pressure in the third trimester of the women exposed to occupational noise, subdivided separately by the presence of impulse noise and level of noise. Means, standard deviations (SO),the numbers of women (N) in the groups, and the adjusted difference (stratified by age and parity) of the group mean to that of the unexposed women with a 95% confidence interval (95% CI).  Table 6. Admissions of the exposed and the unexposed Table 7. Outcomes of pregnancy of the noise-exposed and unwomen to a perinatal ward. exposed women. mean birthweight of the neonates (table 9). Birthweight was not systematically related to the other work conditions (table 9) or to impulse noise (table  10), but the mean birthweight was, on the average, 0.2-0.3 kg lower in the group experiencing a high noise level than among the referents or those exposed Table 8. Data on the newborns of the noise exposed and unexposed mothers.
Newborns of Newborns of exposed unexposed mothers mothers (N = 108) ( to a lower level of noise, and this contrast became more pronounced among the exposed women in this group who also had a standing work position or were engaged in shift work. The prevalence of low birthweight for gestational age was also higher in the high-noise group (5 of 25) than in the reference group (13 of 180; difference + 13 percentage points, 95% CI -3-+29) or the low-noise group (4 of 82; difference +15 percentage points, 95% CI -1-+32). The prevalence of low birthweight was no higher in the impulse-noise group, nor was it systematically related to the other work conditions. The infants of mothers belonging to the high-noise group more commonly needed care at a neonatal unit than those in the low-noise group (5 of 25 versus I I of 83). There were nine malformed infants in the exposed group. Most of the malformations among the exposed cases (8 of 9) occurred in the low-noise group, one being an autosomally recessively inherited defect. When this case was excluded, the prevalence difference between the exposed groups was 5% (95% CI -1-+10). The only perinatal death, because of an intrauterine infection, occurred for an exposed mother (table 8). Table 9. Birthweight of the infants born to the women exposed to occupational noise and to the unexposed women. Means, standard deviations (SO), the numbers of women (N) in the groups, and the adjusted difference (stratified by age and parity) of the group means with a 95% confidence interval (95% CI), overall and subdivided separately by vibration, standing positions, and shift work.  Table 10. Birthweight of the infants born to women exposed to occupational noise subdivided separately by the presence of impulse noise and level of noise. Means, standard deviations (SO), the numbers of women in the groups, and the adjusted difference (stratified by age and parity) of the group mean to that of the unexposed women with a 95% confidence interval (95% CI).

Discussion
The occupational activity of Finnish women is high, and about 78% of women work outside the home during pregnancy (12). In an earlier report on occupational noise exposure during pregnancy (12), we ascertained that only 3.5% of the mothers were exposed to noise if the limit was set at 81 dB, and another report from Finland has set the corre sponding percentage at 2.9% if the limit is 85 dB or more (11). Data on work conditions already exi st since occupation al health care is prescribed by law in Finland, but, for more accurate data on the work conditions of the subj ects in our study , the work loads and exposures of the women were primarily obtained from the occupational health care system, which informed the women about the research. The enrollment of the subjects proceeded slowly, becau se of the obvious reluctance of some employers.
As suggested by our earlier study, it could be assumed that noise would hardly be a major hazard affecting the course and outcome of pregnancy (12). To contr ol some well-known factors int1uencing the outcome of pregnancy, such as maternal age and parity , we aimed at matching the unexposed women in this respect. In addition to other effects, women with children are more likely to terminate their employment than childless ones (18) . The matching by type of occupation also seemed to control the soci al class of the women well. The other base-line characteri stics of the exposed and unexposed women were comparable. For example, there was an exces s of women smoking in both group s (over 30%) relati ve to the figure of 15% reported in Finnish perinatal statistics (19) . Maternity care is equally available to all women in Finland, and is free of charge to ever y woman, so that 99 .8% of pregnant women use these services ( 19). The use of these serv ices was comparable to the average figures for the whole country in both groups ( 19). Statutory maternity leave begins on the 36th gestational week, and paid sick leave can be obtained earlier for medical reasons. The mean worktime during pregnancy was the same in the exposed and unexposed groups, about 29 weeks.
The matching by occupational status was not perfect in that almost all of the exposed women were manual workers, whereas one-third of the women in the unexposed group were of lower-grade staff. Still, the percentages of women with heavy physical and mental loads and a standing work position were comparable. The noise-exposed women more frequently had other inconveniences in their work , like shift work , impul se noise, vibration, and a high or low ambient temperature, and these condition s were more prevalent at higher levels of noise expo sure . On the other hand , there was a clear difference between the information given by the women them selve s and the health officer. A heavy physical load , for example , was reported by 32% of the exposed subj ects them-------------Scand J Work Environ Health 1994. vol 20. no 6 selves in contrast to the figure of 6% by the health officers.
Elevated blood pressure has been connected with noise exposure, although the result s are ambiguous (3)(4)(5)(6). In our earlier study of experimental noise exposure durin g normotensive and hypertensive pregnancy. we could not find any effect of noise on blood pressure levels (20). Nurminen & Kurppa (13) reported that pregnancy-induced hypertension was not associated with noise exposure alone but that , upon additional strain caused by shift work, the pregnant women expo sed to noise at a level of about 85 dB L Ae (8 h) or higher had a distinctly elevated risk of pregna~cy hypertension. Similarly, shift work alone was not related to this complication of pregnancy. The present survey similarly did not detect any association between occupational noise exposure and hypertension in pregnancy.
There were no differences in the number of preterm deliveries between the groups. The mean gestational week at delivery over the whole country was 39.7, and the proportion of preterm deliveries «37 weeks) was 5.2% (19) , figures which are very close to the present ones. The mean birth weights of the infants of the group s did not differ significantly, and they were only a little lower than that for the whole country [3550 (SD 582 ) g (19)]. The prevalence of low birth weight for gestational age was also similar to that for the whole country. The only difference between the exposed and unexposed group s as a whole was seen in the occurrence of congenital malformation s, but this contrast was statistically nonsignificant due to the small numbers.
When noise expo sure rose to 90 dB (L Aeq 8 h) or more, there was no difference in the systolic or diastolic blood pressure , although a lower than average birth weight , either in absolute terms or in relation to gestational age , was observed, albeit with rather wide confidence intervals. The neonate s also needed observation at the neonatal unit more often. These findings were more pronounced for women simultaneously exposed to a standing work position or shift work. Four women out of twenty-five in this exposure group (16%) had a preterm delivery, but the effect of noise on this complication was impossible to distinguish from other coincident exposures associated with preterm birth. The noise expo sure level was not associated with malformations.
In conclu sion, it can be stated that high noise levels can have an independent effect on birthweight and they may be associated with preterm delivery, although the situation may be alleviated somewhat in our country by the opportunities for obtaining sick leave. With respect to noise-induced occupational hearing loss, 39% of the women reported an adequate use of hearing protectors, and it can also be assumed that these protector s had some effect on our result s. Only a minority of women in our country are exposed to high noise levels in general or during pregnancy. On the other hand, high noise levels are of-ten associated with other untoward conditions, and therefore th ey should perhaps be co ns ide re d a form of occupationa l ri sk during pre gn an cy aft e r all.