Noise exposure during pregnancy and selected structural malformations in infants.

Noise exposure during preg and selected structural malformations in infants. Scand J Work Environ Health J 989;15:111-6. The study tested the hypothesis of exposure to noise during pregnancy being teratogenic. It included I 475 Finnish mothers who had given birth to a malformed child (orofacial cleft or structural defect of the cen tral nervous system, skeleton, or heart and great vessels) and I 475 reference mothers. A special interview soon after delivery yielded the primary information on exposure. Of the 783 mothers who reported noise exposure in the first trimester, 370 were case mothers and 413 were referents. Hygienic assessment indi cated that 102 case mothers and 103 referents had been exposed in the first trimester to a sound level of around 80 dB L Aeq(8h) or higher, the overall odds ratio being 1.0 (95 % confidence interval 0.7- 1.3). Adjustment for potential confounders by logistic regression methods gave similar results. There was no obvious trend suggesting a hazard when different exposure categories were considered.

Th e study tested the hypothe sis o f expos ure to noise durin g pregnanc y being teratogenic. It included I 475 Finni sh mothers who had given birth to a mal form ed child (orofacial cleft or structural defect o f the central nervou s system , skeleto n, or heart and grea t vessels) and I 475 reference mothers. A special inter view soo n aft er delivery yielded the primar y informati on on expo sur e. Of the 783 mothers who reported no ise exposure in the first trimester, 370 were case mo thers and 413 were referents. Hygienic assess ment indicated that 102 case mothers and 103 refer ent s had been exposed in the first trimester to a so und level o f around 80 dB L A eq (8 h) or high er, the over all odds ratio being 1.0 (95 % confidence interval 0. 7-1.3). Adjustment for potential confounders by logistic regress ion methods gave similar results. There was no o bvious tr end suggesting a ha zard when different expo sure categories were considered.
K ey terms: birth defects, occupati o n, work.
Noise is a commo n environm ent al pollut ant experienced as a stressor by one-th ird of the active fema le work force in Finland (1). Although epidemiologic studies on the effe cts of environmental , including industrial, factors up on reproductive outcomes have rapidly expanded during recent yea rs, littl e att en tion ha s been focu sed on noise exposur e. According to a recent study, occupational noi se exert s dire ct effects on the fetus, and the effects are dete ctable in childhood as auditory damage (2). Embryo-o r fetotoxic effects via maternal stress resp onse also seem plau sible because laboratory animals have shown decre ased pregnancy ma intena nce and redu ced fetal weight, and occasionally teratogenic effects, when exposed to noise. (See references 3 and 4.) In the Uni ted States one epidemiologic st udy has reported a high ra te of neural tube de fects am on g infant s o f mothers residing in noisy areas near th e Los Angeles airport (5). Near the Atlanta airport, no marked overall connection was found between environment al noise and the occur ren ce of birth defects, but the exa mination of rate s o f severa l categ ories of co ngenita l malformations showed a statistically significant excess of spina bifida witho ut hyd roceph alu s (6). T he case -referent study that was undertaken to study the asso ciation more thoroughl y did not rule out a slight assoc iati on betwe en neural tube defects and residen ce in the high-noi se area. Several other kinds of untoward repro ductive outcome ha ve also been associated with exposure to noise in hum an population s (7)(8)(9)(10)(11)(12)(13)(14). Reprint  In an attempt to screen for terato genic effects of variou s exposures in Finl and , we have retrospect ively quantified the individual noise exposure of I 475 mothers of infants with selected structural birth defects and an equal number of their time-and area-mat ched refer ent s. Th e ana lyses show ed no increased teratogenic risk for infa nts of mo ther s who either reported exposure to noi se in the first trim ester or who were con sider ed exposed according to hygienic criteria.

Subjects and methods
In Finland , population five millio n, nearl y all del iveries take place in ho spital s with both ob stetric and pediatric war ds. Birth defects found during the first year of life or in stillbo rn infants weighin g o ver 600 g must be notified to the Register of Congenital Malf ormations , which is a national surveillance system th at has been in op eration since 1963.
Detailed information on prenatal histor y has been collected fo r infant s with selected struc tura l malforma tions. Virtually all the se defects have been detected at the delivery hospital s. Special studies based on the Register have been nati on wide in co verag e and ha ve applied a case-referent design . The infant who se birth immediately preceded that of the case infant in the same matern ity care distr ict was taken as a referent. Validity poin ts of th e approac h have been discussed elsewhere (15).
According to the Register routine, the mothers of case infants and their referents are interviewed by midwives at their maternity care center usually at the time of the first po stpartum visit. The interviews con sist of some 80 item s, including information on the family, previous pregnancies, and details of the latest pr egnan- Table 1. Composition of the study population that completed the special interviews on occupational and leisure-time exposures.
cy. The completed interview forms and photocopies of antenatal records are returned to the Register. The retrieval rate of the forms has been better than 99 l170 (15). The data collection of the present study started in 1976. All consecutively born infants with central nervous system defects were included. The study was gradually extended to infants with oro facial clefts and structural malformations of the skeleton (excluding club foot and dislocation of the hip), heart, and great vessels. (See reference 16.) A trained pathologist checked the notifications and autopsy reports. Infants with known chromosome anomalies were excluded. A total of I 538 consecutive pairs of infants whose mothers had been interviewed according to the Register routine were eligible for our study. Sixty-three mothers (2.0 l17o) of the total of 3 076 could not be contacted, declined special interviews, or did not show up for the interview. Valid interviews were completed for I 475 case-referent pairs, ie, for 96 lifo of the pairs of mothers with eligible infants (table 1).
Two trained interviewers collected data on the mothers' exposures during work and leisure time. The interviews took place at the mother's maternity care center, usually about three months after the delivery. Information on exposures was obtained from the mother's workday description, which was recorded as such, and from fixed questions regarding specific exposures.
The interview had fixed questions on noise exposure and use of hearing protectors during pregnancy. A total of 783 (27 l17o) of all the mothers reported noise exposure in the first trimester, ie, 774 (32 lifo) of the mothers who had worked and 9 (2 lifo) of those who had not worked. Of the mothers who had reported noise in the first trimester, 100 had used hearing protectors.
Throughout the study the persons making the exposure categorizations were unaware of the case-referent status of the infants. All of the 2 950 interview forms were provisionally assessed at exposure meetings, attended by an industrial hygienist and two experts in occupational health, which convened regularly Malformation group during the 6.5 years of data collection. Workday description, the fixed question on noise exposure, hours of exposure per day, use of hearing protectors, and the results of possible noise measurements were noted. The group used the 80 dB 8-h equivalent continuous A-weighted sound level (L Aeq (8 h) as the approximate cut -off level for reasonable noise exposure (17). Exposures to several other types of occupational and leisure-time factors were classified at the same time (16). The aforementioned exposure meetings had considered 220 mothers exposed to noise in the first trimester. In 66 instances (30 lifo) information on actual noise measurement at the mother's workplace was available, and in 145 instances (66 lifo) information on measurements at closely comparable workplaces could be utilized for exposure categorization by the analogy principle. For nine mothers (4 (10) the assessment was solely based on the hygienist's judgment.
Finally, two hygienists, one a member of the study team and the other an expert on noise measurements, independently perused the interview forms of the 220 exposed mothers. Their forms were mixed with interview forms of an additional 100 mothers who had not been considered noteworthily exposed to noise. These 100 mothers were made up of two random samples, 50 mothers who had reported noise in response to the interview question and 50 who had not. The hygienists were unaware of which form belonged to which of the groups. Preselected exposure categories were the objective, and the selected noise levels were considered to represent ordinal-scaled exposure categories of (i) "no" appreciable exposure (clearly less than 80 dB L Aeq (8 h»' (ii) exposure of "low" intensity (around 80 dB L Ac q (8 h»' (iii) exposure of "moderate" intensity (around 85 dB L Acq (8 h»' and (iv) exposure of "high" intensity (around 90 dB L A cq (8 h) or higher). These categories were chosen because the hygienists felt reasonably comfortable as to their ability (i) to differentiate most of the "unexposed" mothers from those exposed to noise of around 80 dB L A cq (8 h) or above and (ii) to discern exposure to less intense noise from that of more intense noise. Finally, the information on each mother who did not receive the same classification was considered by both hygienists jointly, and the category best agreeable was chosen.
The hygienists' assessments for the 220 mothers who had been provisionally considered exposed are crosstabulated in table 2. The hygienists independently categorized 182 of them as exposed and nine as unexposed. After jointly examining the exposure information of the remaining 29 mothers, the hygienists regarded 26 of them as exposed and three as unexposed. Thus, of the 220 mothers that had been considered exposed in the exposure meetings, 208 (94.5 lifo)were also classed as such in the final scrutiny. Exposure included the first trimester for 205 of the mothers. The remaining 2 745 mothers were considered unexposed in the first trimester.
In the exposure meetings 2 730 moth ers ha d been assessed as un exposed . In the sam ple of 50 mo ther s of the 2 107 who had not reported noise exposure , the hygienists agr eed o n nonexposure in all insta nces . Of the sa mp le of 50 mothers who had reported exposure (623 mothers), and yet had been con sidered unexposed, th e hygienists ag reed that five shou ld be rega rded as exposed . According to the se sample findings, some 2 % of the une xpo sed mothers had been misclassified in the expo sur e meetings . Th e error was considered tol erable , and no fu rther effort was made to identify the misclassified mothers ind ividu ally.
Th e matching procedure had not correlated the case and reference series with respect to no ise exposure, and th erefor e the dat a were analyzed unm at ched to gain more ef ficiency. [See refer ences 18 and 19 (p 280).] Confidence intervals for the crud e odds ratio s were calculated with the modified Cornfield meth od (20,21).
Pow er calculations for single 2 x 2 tables were carried out with methods derived by Miettinen (22) with a signific anc e level of a = 0.025 and a one-sided test.
In the adjustment , unconditional logistic regression ' mod eling , mathematically identical to follow-up studies, was used (23). The variables th at were controlled in the comparisons were mother' s employment, exposure to so lvents in the first tr imester , age, parit y, previous miscarriages, ind uced abortio ns, stillbirt hs, previou s child with malformation , co mmo n cold or fever in the first tr imester , intak e o f an algetics or antip yretics, intake of sedatives or so porifics in the first trime ster , alcoho l intake, and smok ing du ring pr egnancy.
Of the moth ers who were emplo yed during pregnancy and had been expo sed to noise, 95 % belon ged to the category of manual workers acco rdi ng to the socioeco nomic classification (24), whereas 27 % of the unexposed mothers were manual wor kers. All exposed women had been employed in nonagricultural work. Control of po ssible confounding was done by a separate analysis which was restricted to the category of nonagricultural manual workers. Table 3 shows tha t 402 case mothers and 440 reference moth ers had report ed noise exposure du ring pregnan cy, 370 of the former and 413 of the latter in th e first tr imester . Th e cr ude odds ratios were belo w un ity or near it fo r all the malformation groups under study. Man y of the reported exposures had occur red in bank s, coffee bar s, shops , or schools. Street noi se and children yelling also often appeared in th e int erview form s.

Results
According to the hygienist s 205 mothers had been expo sed to no ise of at leas t low intensity in the first trim ester. Table 4 shows the distribution of these mothers by the indu stries in which the noise expo sure had occurred . Of the mothers, 44.4 % had worked in the textile or clothing industry, and the rest in severa l other econ omi c fields.
When the mothers were grouped according to the hygienists' judgment of exposure, it appeared that 102 case mothers and 103 reference mothers had been exposed to no ise of at least low intensity in th e first trimester (table 5). Th e cru de odds ratios sho wed no statistically significant indications of a teratogenic risk in an y ma lfo rma tion gro up. Table 2. Co rres pondence of the two in du stri al hyg ie nists ' independent asse ssment s of the no ise exposure' o f the 220 mo the rs con sid ered exp osed in th e exposure meet ings. The approved final numbers fo r t he exposure cat egories are g iven In parenth eses .    Table 6 shows the crude and adjusted odds ratio point estimates and their 95 0,10 confidence intervals for an overall teratogenic risk according to exposure intensity . There was no obv ious trend suggesting a hazard. For the low and moderate intensit y levels the results did not suggest a teratogenic effect for noise exposure. For exposure to a high inten sity of noise the adjusted odds ratio was 1.7, but the lower confidence limit was below unity. In the pooled material, the adjusted odds ratio was 1.0 [95 % confidence inter val (95 % CI) 0. 8-1.4] for exposure to noise of at least low intensity in the first trime ster . When the analysis Table 4. Industr ies in which mothers (both case and reference) exposed to no ise in th e f irst trimester worked .
Noise intensity -Industry was restricted to the socioeconomic category of manual workers in nonagricultural jobs, the adju sted overall odd s ratio was 1.1 (95 % CI 0.7 -1.5) Forty-two out of 102 case mothers and 36 out of 103 referen ce mothers who were considered exposed to noise in the first trimester had worn hearing protector s during pregnancy. Table 7 shows distributions of case and reference mothers according to reported use of hearing protectors without suggesting differences between the gro ups. The exposed mothers, all combined, who had not used protectors showed an adjusted overall odds ratio of 0.9 (95 % C I 0.6-1.3), and tho se who had used protectors 1.2 (95 % CI 0.7-1.9).

Discussion
Some anima l studies have show n tha t noise expos ure may produce reproductive disturb ances, including teratogenicity, the suggested mechanisms being the stressinduced increase of catecholamine levels and vasoconstriction in placental vasculature. (See references 3 and 4.) Perception of noise seems more decisive for the induction of human stress than the actual noise intens ity is (25). Con sequent ly, we analyzed the data by using both an objective and a subjective noise categorization . The resu lts of neither approach suggested that Table 5. Case and refe rence mo thers accord in g t o the hy gienist s' ass essment of t he S·h equi valent con t inuou s A-weighted soun d level (LAeq (B hi) in t he f irs t t rime ster.  Table 7. Distri buti ons of case and refe renc e mothers by use of hearin g protecto rs during pregn ancy and cr ude odds ratio s. noi se might be a ma jor teratogen a t levels commo nly detectabl e in Finl and .
It is unlikely th at the matern al recall or hygienic estimat ion o f noi se exposur e in o ur study was biased . The prim ary information on several facto rs at wo rksites and during leisur e-tim e activit ies was gathered by train ed interviewe rs who were unaware o f any specific hypotheses that might be tested. The hygienists made their expo sure classification blind as to the case-referent status, and therefor e any bia s cau sed by a system atic misclassificat ion of exposure sho uld ha ve been prevented . In th e situation o f ra re expo su re, such as noise as assessed by hygienic stand ard s, even a sma ll deviati on fro m full specif icity cou ld bias the estimation . A fter the hygien ic check-ups it is reasonable to suppo se that very few mothers were falsely categor ized as exposed. Allowing for th e estim ate d misclassification rate o f nonexposur e, approxi mately 2 0/ 0, did not in fluence the results.
Ac cording to the results o f a Finnish question naire study , 29-3 5 010 of the wor k fo rce find occ upa tio na l noise a nno ying (I ) . In o ur study 32 % of the wo rking women reported noise expo sure during pregnan cy, case and reference mo thers equa lly often . For the great maj orit y of th e mothers who had reported noi se exposu re (75 %), the level of expo sure was assessed to be below the selected cut -o ff level o f around 80 dB L Aeq (8 h i' Th e porti on of mothers co nsidered expos ed by the hygienists was 7 % of the to tal , including mo thers who wo rke d regul arl y and mothers who did no t. Some 3 070 o f the mo th ers had been mor e mar kedly (arou nd 85 dB L Acq (8 h) or higher ) exposed , and on ly 0.7 % of all the mothers had been heavily exposed (around 90 dB L Acq (8 hi) ' Thus, we co uld not eva luate malformat ions in relation to high levels o f noise expo sure.
Environmental noise is a potent ia l problem becau se of ub iquitou s background expo sure . H owever , our criterion of at least aro und 80 dB L Aeq (8 h) for exposure is mu ch higher th an typical en viro nme nta l expo sur e. Road traffic is th e most im po rta nt sour ce of noise in urban ar eas. Yet, merely an estimated 12.5 % of th e population in European OE CD countries are All birth def ects pool ed 1.00 0.86 exposed to dayti me road tra ffic no ise high er th an 65 d B L Aeq , and only 0.3 l1 10 to da ytim e aircraft noise exceed ing 65 dB L Aeq (26). Exposure can vary considera bly du e to tr a ffic den sity and type of conurbation, but Finland has a relatively low de nsit y of po pulation. When compa red to typical noise exposur e at indu stri al wo rk places, the co ntribution du e to tr an sport noi se in th e to tal noise imm ission level is small. As regar ds the mo th ers objectively expo sed (hygienic esti ma te), th e study had sufficient power to detect a 1.5-fold over all risk with a 86 % chan ce (table 8). For th e studied mal formation subgro ups, th e pow er of the study was su fficient for a reasona ble cha nce o f det ecting a 2.0-fo ld risk. W hen th e scru ti ny was lim ited to the category of high noise exposure, assessed as aro und 90 dB L Aeq (8 h) ' the power o f the study was poor (table 6).
In summary , the results of the pr esent study did not ind icate a ter at ogenic risk for infants of mo th ers exposed to noise in the first trimester o f pr egnan cy. Nu merica l limitation s pre vent detailed generalizatio ns as in most ot her epidemiolog ic stu dies on teratogenicity. Th e resul ts do no t ap ply directly to very intense noise exposure nor to na rr owly defined malformation cat ego ries . Yet, we are inclined to co nclude that no ise expo sure in a society such as Finland's today is not likely to be an impo rtant teratogenic risk factor.