Skip to main content
American Journal of Public Health logoLink to American Journal of Public Health
. 2009 Aug;99(8):1409–1416. doi: 10.2105/AJPH.2008.138412

First-Trimester Working Conditions and Birthweight: A Prospective Cohort Study

Tanja G M Vrijkotte 1,, Marcel F van der Wal 1, Manon van Eijsden 1, Gouke J Bonsel 1
PMCID: PMC2707468  PMID: 19542045

Abstract

Objectives. We investigated the relationship between women's first-trimester working conditions and infant birthweight.

Methods. Pregnant women (N = 8266) participating in the Amsterdam Born Children and Their Development study completed a questionnaire gathering information on employment and working conditions. After exclusions, 7135 women remained in our analyses. Low birthweight and delivery of a small-for-gestational-age (SGA) infant were the main outcome measures.

Results. After adjustment, a workweek of 32 hours or more (mean birthweight decrease of 43 g) and high job strain (mean birthweight decrease of 72 g) were significantly associated with birthweight. Only high job strain increased the risk of delivering an SGA infant (odds ratio [OR] = 1.5; 95% confidence interval [CI] = 1.1, 2.2). After adjustment, the combination of high job strain and a long workweek resulted in the largest birthweight reduction (150 g) and the highest risk of delivering an SGA infant (OR = 2.0; 95% CI = 1.2, 3.2).

Conclusions. High levels of job strain during early pregnancy are associated with reduced birthweight and an increased risk of delivering an SGA infant, particularly if mothers work 32 or more hours per week.


Delivery of a low-birthweight or small-for-gestational-age (SGA) infant as a result of fetal growth restriction is one of the principal adverse pregnancy outcomes. In the short term, low birthweight and small size for gestational age are major determinants of infant mortality and morbidity1 and impaired neonatal development.2 In the long term, they increase metabolic and cardiovascular disease risk.35 Prevention of fetal growth restriction is therefore of undisputed clinical and economic importance.

Maternal factors, obstetric factors (e.g., placental dynamics), and social factors,5 including employment-related factors, can all play a role in fetal growth impairment.624 Although employment in general is associated with enhanced outcomes,6,20,21 certain working conditions represent potential risk factors for the mother and child. Increased levels of risk resulting from long working hours,12,13,17,18,24 high physical workloads,1316 prolonged standing,13,18 and psychosocial job strain7,9,10,24 have been suggested, but the findings in this area are not unequivocal.8,11,22,23 So far, 2 reviews have been conducted that focused on physical workload and delivery of an SGA infant. Mozurkewich et al.16 concluded from their review of 29 studies that physically demanding work is associated with SGA births (pooled odds ratio [OR] = 1.37; 95% confidence interval [CI] = 1.30, 1.44). Bonzini et al.19 reached the same conclusion in their study. To our knowledge, job strain has not been considered in any published review.

Limitations in research designs,6,8,1921 variability in definitions and measurement of work-related factors,6,1820 and true variability across countries and cultures may account for the inconsistent results observed to date. Another important limitation of occupational hazard research is the focus on third-trimester exposures.11,13 Experimental data and emerging theory point to the first rather than the second or third trimester as a crucial period for regulating the relevant fetal hormonal set points, in particular the hypothalamic pituitary axis (HPA).2527 Stress-dependent dysregulation of the HPA affects birthweight and a child's subsequent growth and development.2531 From this perspective, employment during pregnancy is perhaps the most prevalent potential stress factor, given that few working women quit their jobs early in pregnancy.

In an effort to overcome the limitations of previous studies, we explored the association between infant birthweight and employment-related conditions (e.g., hours worked per week, hours standing or walking, physical demands of work, and job strain) in an unselected urban cohort of pregnant women. We hypothesized that after adjustment for all known major cofactors, first-trimester work-related effects on birthweight would exceed the third-trimester effects reported in previous research.

METHODS

We used prospective data derived from the Amsterdam Born Children and Their Development (ABCD) study, a community cohort investigation of pregnant women (N = 8266) residing in Amsterdam, Netherlands. The ABCD study (see http://www.abcd-study.nl) investigated differences in pregnancy outcomes with a focus on maternal lifestyle (including nutritional status) and specific psychosocial conditions (including work).3234 Between January 2003 and March 2004, all pregnant women in Amsterdam were invited to participate at their first antenatal visit (median = 13 weeks; interquartile range = 3 weeks) with their obstetric caregiver.

In addition to the routine prenatal care provided, a blood sample was taken at the women's first visit (to assess nutrient profile and thyroid function), and a modular pregnancy questionnaire was distributed via mail 2 weeks later, to be returned in a prepaid envelope. This validated questionnaire, administered in 4 languages via a translator's service, gathered information on women's sociodemographic characteristics, lifestyle, psychosocial stress levels, and obstetric history. A reminder was sent 2 weeks later to women who had not returned the questionnaire.

In total, 12 373 women were invited to participate, and 8266 women returned the questionnaire, for a response rate of 67%. Of this group, 7730 gave birth to a viable singleton infant for whom information on birthweight and pregnancy duration was available. Women who delivered a premature infant (i.e., pregnancy duration of less than 37 weeks; n = 415) and those who were younger than 20 years (n = 180) were excluded from our study. The rate of preterm delivery in our cohort was low (5.3%) but comparable to that of the overall ABCD study cohort (7.5%) and that of the Netherlands as a whole (7.1%).35 The final sample size for our analyses was 7135 women.

Measures

Employment-related factors.

Employment was defined as paid work for at least 8 hours per week during women's first trimester. Women in all other categories were classified as unemployed. Number of hours worked per week was analyzed as a categorical variable (less than 24, 24–31, 32 or more) according to the conventional classifications used in the Netherlands. The self-administered, validated Dutch version of the Job Content Questionnaire was used to measure job strain.36,37 This instrument consists of 2 subscales: job demands and job control. All items involve a 4-point response scale. The job demands scale includes 25 items focusing on work pace (11 items; e.g., time pressure, amount of work), mental workload (7 items; e.g., requirement to perform several tasks simultaneously), and physical workload (7 items; e.g., strenuous posture, load carrying). The job control scale comprises 11 items addressing such areas as perceived control of one's own work pace and perceived influence on the planning of one's tasks.

In this study, the reliabilities (Cronbach α) for the job demands and job control scales were 0.82 and 0.91, respectively. In our analyses, we divided job demand summary scores into 3 levels: low (below the 50th percentile), moderate (between the 50th and 90th percentiles), and high (above the 90th percentile). Similarly, job control summary scores were divided into high (above the 50th percentile), moderate (between the 10th and 50th percentiles), and low (below the 10th percentile) levels. Women at high job demand levels and low or moderate job control levels were rated as having high job strain, those at low job demand levels and moderate or high job control levels were rated as having low job strain, and women in all other category combinations were rated as having moderate job strain. Self-reported number of work-related hours of standing or walking per week and the Job Content Questionnaire job demands scale were used in assessing physical workload.

Covariates and confounders.

Data on infant gender, gestational age (based on ultrasound or, if unavailable, timing of most recent menstrual period), and birthweight were obtained from the Youth Health Care Registration of Amsterdam's Municipal Health Service. Information on all other maternal explanatory variables was derived from women's questionnaire responses: age, parity, ethnicity (Dutch, Surinamese, Turkish, Moroccan, other non-Western ethnicity, other Western ethnicity), educational level (measured as years of education completed after primary school: less than 5, 5–10, more than 10), smoking during pregnancy (number of cigarettes smoked per day: 0, less than 1, 1–5, or more than 5), alcohol use during pregnancy (yes or no), marital status (married or cohabiting vs single), prepregnancy body mass index (i.e., weight in kilograms divided by height in meters squared), height, high blood pressure (yes or no), and diabetes (yes or no). Given that birthweight has been shown to decrease with such characteristics as older age, lower parity, and lower socioeconomic status (SES), some of these factors were treated as both confounders and independent variables.6,810,19,20

As a means of accounting for the potential combined risk of work- and parenting-related stress among employed women with children, parenting stress was measured separately from work stress. The validated frequency subscale of the Parenting Daily Hassles Scale,38 which consists of 20 items focusing on stressful events related to everyday parenting, was used in assessing parental stress. Responses are made on a 4-point scale, and total stress scores were divided into 4 levels: no children, low stress (below the 50th percentile), moderate stress (between the 50th and 90th percentiles), and high stress (above the 90th percentile).

Outcomes.

The primary outcome variables were low birthweight (continuous in grams) and delivery of an SGA infant. Small size for gestational age was defined as a birthweight below the 10th percentile for gestational age on the basis of gender- and parity-specific standards.39

Statistical Analysis

We used predefined explorative linear regression models, with increasing levels of adjustment, to explore the association of working conditions with birthweight. Unadjusted effects were derived from the initial univariate analysis of each working condition in isolation. Multivariate model 1 controlled for maternal physiological characteristics (age, height, parity, preexisting hypertension and diabetes), pregnancy duration, and infant gender. In multivariate model 2, all other lifestyle and sociodemographic factors (education, marital status, ethnicity, smoking, alcohol use, body mass index, and parenting stress) were added to determine independent associations between working conditions and birthweight. Correlations between the covariates were all below 0.23, indicating no collinearity. In each model, employed women with the least unfavorable working conditions were designated as the reference group; unemployed women were grouped into a separate category.

We conducted a subsequent logistic regression analysis, which followed the same procedure as the linear regression models, to explore the association between working conditions and delivery of an SGA infant. The combined effects of job strain and number of working hours and of physical workload and number of working hours were tested. Therefore, these variables were redefined into 6 categories, and the reference category was designated as a combination of the least unfavorable working conditions and a workweek of less than 32 hours. The interaction between job strain and physical workload could not be tested (because of collinearity). The residuals of the linear regression models were visually checked to assess their distribution, and homogeneity of variance was assessed with the Levene test. The Hosmer–Lemeshow test was used to assess the goodness of fit of the logistic regression models.

We used SPSS version 13.0 (SPSS Inc, Chicago, IL) to analyze the data; α was set at .05. Missing values were imputed only in the case of prepregnancy body mass index (5% of body mass index values were missing; less than 1% of other data were missing).

RESULTS

Sample sociodemographic characteristics, stratified according to employment status, are shown in Table 1. In comparison with women who did not return the ABCD study questionnaire (n = 4107), those who did (n = 8266) were older (mean age: 31.7 ±5.2 vs 30.2 ±5.8), more often primiparas (55.7% vs 40.1%), and more often of Dutch ethnicity (62.6% vs 35.3%); differences between respondents and nonrespondents were largely explained by differences in employment status between the study ethnic groups. No differences were found with respect to birthweight or pregnancy duration.

TABLE 1.

Maternal and Infant Characteristics, by Maternal Employment Status: Amsterdam Born Children and their Development Study, Amsterdam, Netherlands, 2003–2004

Employed Women Unemployed Women
Maternal age, y, mean (SD) 31.8 (4.2) 29.7 (5.5)
Maternal age, y, %
    20–29 25.7 48.1
    30–39 71.2 48.4
    ≥ 40 3.1 3.5
Maternal height, cm, mean (SD) 170.0 (6.8) 166.0 (7.1)
Maternal height, cm, %
    < 160 6.0 15.4
    160–169 40.9 54.1
    170–179 45.2 26.8
    ≥ 180 7.9 3.7
Prepregnancy BMI, kg/m2, mean (SD) 22.7 (3.5) 23.8 (3.9)
Prepregnancy BMI, kg/m2, %
    < 20 18.1 14.5
    20–24 63.7 52.6
    25–29 14.1 25.9
    ≥ 30 4.1 7.0
Preexisting hypertension, % 2.2 3.1
Preexisting diabetes, % 0.4 0.9
Single, % 9.5 17.8
Cigarettes smoked/d during pregnancy, %
    0 91.8 90.0
    < 1 2.7 2.1
    1–5 3.3 3.8
    > 5 2.2 4.1
Alcohol use during pregnancy, % 27.0 12.8
Education, y, %
    < 5 10.6 39.9
    5–10 38.7 39.1
    > 10 50.7 21.0
Parenting stress level,a %
    No children 61.8 45.7
    Low 21.1 24.3
    Moderate 14.1 21.3
    High 3.0 8.7
Ethnicity, %
    Dutch 78.2 37.5
    Other Western 8.3 8.5
    Surinamese 4.2 6.9
    Turkish 1.3 8.8
    Moroccan 2.5 16.0
    Other non-Western 5.4 22.3
Parity, %
    0 61.3 40.2
    1 30.8 36.3
    ≥ 2 7.9 23.5
Gestational age, wk, mean (SD) 40.1 (1.2) 40.0 (1.2)
Birthweight, g, mean (SD) 3510 (483) 3490 (484)
Male infant gender, % 50.0 50.5

Note. BMI = body mass index. For employed women, n = 4496; for unemployed women, n = 2639.

a

Parenting stress was measured using the Parenting Daily Hassles Scale,38 which consists of 20 items focusing on stressful events related to everyday parenting, was used in assessing parental stress. Responses are made on a 4-point scale, and total stress scores were divided into 4 levels: no children, low stress (below the 50th percentile), moderate stress (between the 50th and 90th percentiles), and high stress (above the 90th percentile).

In an effort to determine whether selective participation resulted in response bias, we conducted an extensive nonresponse analysis in which our data set was linked to data from the Dutch Perinatal Registry. Results showed similar response and nonresponse group associations between risk factors and several adverse outcome indicators, suggesting that no selection bias occurred (data available on request from the authors). We checked our results by repeating the analysis with weighted factors for parity, ethnicity, and maternal age (based on prevalence in the initial cohort and the response cohort). The associations between working conditions and birthweight did not change, again indicating no selection bias (data not shown).

Most (64%) of the pregnant women in the sample worked at least 8 hours a week during their first trimester. Employed women were older and better educated than were unemployed women; also, they smoked less, consumed alcohol more frequently during pregnancy, had lower body mass indexes, were taller, and were more often primiparas and married. Unexpectedly, high parenting stress was more frequent in the unemployed group. Infant gender and mean gestational age were similar in the employed and unemployed groups.

Low Birthweight

Univariate analyses showed that all of the working conditions assessed, with the exception of 10 to 19 hours standing or walking per week, were associated with decreased birthweight (i.e., relative to the different reference categories; Table 2 ). High job strain was related to the largest reduction in birthweight, followed by a workweek of 32 or more hours, a high physical workload, and standing or walking 20 or more hours per week (all in comparison with the relevant reference groups). Although adjustment for physiological factors (model 1) indicated attenuation of the associations for weekly working hours, physical workload, and job strain, effect estimates remained significant.

TABLE 2.

Univariate and Multivariate Associations Between Working Conditions and Birthweight: Amsterdam Born Children and their Development Study, Amsterdam, Netherlands, 2003–2004

Univariate
Model 1a
Model 2b
No. b (SE) P Adjusted R2 b (SE) P Adjusted R2 b (SE) P Adjusted R2
Weekly working hours 0.004 0.242 0.282
    0 (unemployed) 2639 −72 (21) <.001 −5 (19) .79 −14 (19) .47
    8–23 (Ref) 588 0 0 0
    24–31 905 −7 (25) .75 −8 (22) .65 −10 (22) .65
    ≥ 32 3003 −87 (21) <.001 −44 (19) .02 −43 (19) .02
Weekly standing/walking hours 0.001 0.250 0.281
    0 (unemployed) 2639 −19 (13) .15 2 (12) .86 9 (13) .47
    < 10 (Ref) 2607 0 0 0
    10–19 948 17 (18) .33 19 (16) .21 19 (15) .21
    ≥ 20 861 −40 (23) .07 −12 (20) .57 −18 (19) .69
Physical workload 0.002 0.265 0.282
    None (unemployed) 2639 −39 (14) <.01 −3 (13) .81 −6 (13) .69
    Low (Ref) 2455 0 0 0
    Moderate 1554 −41 (16) <.01 −20 (14) .15 −16 (14) .24
    High 455 −75 (24) <.01 −40 (21) .07 −21 (22) .32
Job strainc 0.002 0.242 0.280
    None (unemployed) 2639 −34 (14) .02 −13 (13) .38 −2 (14) .93
    Low (Ref) 2147 0 0 0
    Moderate 2014 −21 (15) .17 −17 (13) .21 −14 (13) .24
    High 296 −115 (30) <.001 −70 (26) .01 −72 (26) <.01

Note. Sample sizes differ slightly because of missing values. The residuals from all models were normally distributed, and variances were homogeneous. The unstandardized parameter estimate (b) represents the change in birthweight relative to the reference group.

a

Adjusted for parity, infant gender, pregnancy duration (linear and quadratic), and for maternal age, height, and preexisting hypertension and diabetes.

b

Adjusted for the same variables as model 1 as well as for marital status, educational level, ethnicity, smoking, alcohol use, prepregnancy body mass index (linear and quadratic), and parenting stress.

c

The self-administered, validated Dutch version of the Job Content Questionnaire was used to measure job strain.

The addition of lifestyle and sociodemographic factors (model 2) did not alter the association of birthweight with job strain (mean birthweight decrease of 72 g) or with weekly working hours (mean birthweight decrease of 43 g). However, the association between physical workload and birthweight disappeared. Stepwise analyses in which the lifestyle and sociodemographic factors were added to model 2 indicated that smoking, which was highly related to physical workload (17.1% of women in the high-workload group were smokers vs 5.8% in the low-workload group), caused the association's disappearance. Birthweight among infants of unemployed women was, on average, 40 g lower than that among employed women; however, after the addition of physiological factors to model 1, the association became nonsignificant.

Delivery of a Small-for-Gestational-Age Infant

Results of analyses in which delivery of an SGA infant was the outcome variable closely resembled the results of analyses involving birthweight (Table 3 ). High and moderate job strain and high and moderate physical workload were associated with an elevated risk of delivering an SGA infant. The addition of physiological factors (model 1) decreased the work-related risk of delivering an SGA infant, but the association remained significant for high job strain and for high physical workload.

TABLE 3.

Univariate and Multivariate Associations Between Working Conditions and Small-for-Gestational-Age (SGA) Births: Amsterdam Born Children and their Development Study, Amsterdam, Netherlands, 2003–2004

No. SGA Births, % Univariate, OR (95% CI) Model 1,a OR (95% CI) Model 2,b OR (95% CI)
Weekly working hours
    0 (unemployed) 2639 14.7 1.4 (1.0, 1.9) 1.2 (0.9, 1.6) 1.2 (0.8, 1.6)
    8–23 (Ref) 588 10.7 1.0 1.0 1.0
    24–31 905 10.7 1.0 (0.7, 1.4) 1.0 (0.7, 1.5) 1.1 (0.8, 1.5)
    ≥ 32 3003 10.9 1.0 (0.7, 1.4) 1.1 (0.8, 1.4) 1.1 (0.8, 1.5)
Weekly standing/walking hours
    0 (unemployed) 2639 14.7 1.5 (1.3, 1.8) 1.2 (1.0, 1.3) 1.1 (0.9, 1.3)
    < 10 (Ref) 2607 10.2 1.0 1.0 1.0
    10–19 948 10.9 1.1 (0.9, 1.4) 1.1 (0.8, 1.3) 1.0 (0.8, 1.3)
    ≥ 20 861 12.9 1.3 (1.0, 1.6) 1.2 (0.9, 1.5) 1.0 (0.8, 1.4)
Physical workload
    None (unemployed) 2639 14.7 1.8 (1.4, 2.1) 1.4 (1.1, 1.6) 1.2 (1.0, 1.5)
    Low (Ref) 2455 9.0 1.0 1.0 1.0
    Moderate 1554 12.5 1.5 (1.2, 1.8) 1.3 (1.1, 1.6) 1.3 (0.9, 1.6)
    High 455 14.5 1.7 (1.3, 2.3) 1.4 (1.1, 2.0) 1.2 (0.9, 1.7)
Job strainc
    None (unemployed) 2639 14.7 1.6 (1.4, 2.0) 1.3 (1.1, 1.5) 1.2 (1.0, 1.5)
    Low (Ref) 2147 9.6 1.0 1.0 1.0
    Moderate 2014 11.4 1.2 (1.0, 1.5) 1.2 (1.0, 1.5) 1.2 (1.0, 1.5)
    High 296 16.0 1.9 (1.3, 2.6) 1.7 (1.2, 2.3) 1.5 (1.1, 2.1)

Note. OR = odds ratio; CI = confidence interval. Sample sizes differ slightly as a result of missing values. CIs in which the lower value is greater than 1 are significant. Goodness-of-fit tests showed no evidence of lack of fit (P > .15 for all logistic models). SGA analyses did not include parity, pregnancy duration, or infant gender given that, by definition, small size for gestational age incorporates this information.

a

Adjusted for maternal age, height, and preexisting hypertension and diabetes.

b

Adjusted for the same variables as model 1 in addition to marital status, educational level, ethnicity, smoking, alcohol use, prepregnancy body mass index (linear and quadratic), and parenting stress.

c

The self-administered, validated Dutch version of the Job Content Questionnaire was used to measure job strain.

Additional adjustment for lifestyle and sociodemographic factors (model 2) did not affect the risk of delivering an SGA infant in the high job strain group, but the association with high physical workload weakened and was again fully explained by smoking (data not shown). As a separate category, unemployment was associated with at least a 50% higher risk of delivering an SGA infant (OR = 1.5; 95% CI = 1.2, 1.9). After full adjustment, this risk increased only slightly, by 20% on average.

Combined Findings

The largest reduction in birthweight was observed among women with high job strain who worked 32 or more hours per week (interaction P = .04; Table 4 ). In comparison with women with low levels of job strain who worked less than 32 hours per week (i.e., the reference group), these women showed a reduction in birthweight of 150 g after full adjustment (model 2). The combination of a high physical workload and a workweek of 32 or more hours was also associated with reduced birthweight of 116 g after full adjustment (interaction P = .09). Similar to the birthweight results, after full adjustment, the risk of delivering an SGA infant was highest among women who worked 32 or more hours per week and were in the high job strain category (OR = 2.0; 95% CI = 1.2, 3.2) or the high physical workload category (OR = 1.5; 95% CI = 1.0, 2.3).

TABLE 4.

Effects of Weekly Working Hours in Combination With Job Strain and Physical Workload on Birthweight and Small-for-Gestational-Age (SGA) Births: Amsterdam Born Children and their Development Study, Amsterdam, Netherlands, 2003–2004

Birthweight
SGA Birth
No. Univariate, b (SE) Model 1,a b (SE) Model 2,b b (SE) Univariate, OR (95% CI) Model 1,c OR (95% CI) Model 2,d OR (95% CI)
8–31 h/wk
Job strain level
    Low (Ref) 730 0 0 0 1.0 1.0 1.0
    Moderate 635 −29 (26) −31 (22) −33 (22) 1.3 (0.9, 1.9) 1.3 (0.9, 1.9) 1.3 (0.9, 1.9)
    High 109 −16 (49) −15 (42) −16 (43) 1.1 (0.6, 2.2) 1.0 (0.5, 1.9) 1.0 (0.5, 2.0)
≥ 32 h/wk
Job strain level
    Low 1417 −78*** (22) −27* (20) −32* (20) 1.0 (0.7, 1.3) 1.0 (0.8, 1.4) 1.1 (0.8, 1.4)
    Moderate 1379 −96*** (22) −40** (20) −42* (20) 1.2 (0.9, 1.6) 1.2 (0.9, 1.7) 1.2 (0.9, 1.7)
    High 187 −239*** (39) −159*** (34) −150*** (35) 2.3 (1.5, 3.6) 2.1 (1.3, 3.2) 2.0 (1.2, 3.2)
8–31 h/wk
Physical workload
    Low (Ref) 662 0 0 0 1.0 1.0 1.0
    Moderate 609 −64* (27) −26 (23) −25 (23) 1.3 (0.9, 1.9) 1.2 (0.8, 1.7) 1.1 (0.8, 1.6)
    High 205 −20 (39) −20 (33) −20 (34) 1.0 (0.6, 1.7) 0.9 (0.5, 1.5) 0.8 (0.4, 1.4)
≥ 32 h/wk
Physical workload
    Low 1793 −84*** (22) −24 (19) −29 (20) 0.9 (0.7, 1.2) 0.9 (0.7, 1.2) 0.9 (0.7, 1.3)
    Moderate 945 −135*** (24) −64** (22) −51* (22) 1.4 (1.0, 1.9) 1.3 (0.9, 1.9) 1.2 (0.9, 1.7)
    High 250 −231*** (35) −127*** (31) −116*** (31) 2.1 (1.4, 3.2) 1.8 (1.1, 2.7) 1.5 (1.0, 2.3)

Note. OR = odds ratio; CI = confidence interval. Results are for employed women only. Sample sizes differ slightly as a result of missing values. The unstandardized parameter estimate (b) represents the change in birthweight relative to the reference group. CIs in which the lower value is greater than 1 are significant. The residuals of the final linear regression models were normally distributed, and variances were homogeneous. Goodness-of-fit tests showed no evidence of lack of fit (P > .17, for all logistic models). SGA analyses did not include parity, pregnancy duration, or infant gender given that, by definition, small size for gestational age incorporates this information.

a

Adjusted for parity, infant gender, pregnancy duration (linear and quadratic), and maternal age, height, and preexisting hypertension and diabetes.

b

Adjusted for the same variables as the birthweight model 1 as well as for marital status, educational level, ethnicity, smoking, alcohol use, prepregnancy body mass index (linear and quadratic), and parenting stress.

c

Adjusted for maternal age, height, and preexisting hypertension and diabetes.

d

Adjusted for the same variables as the SGA birth model 1 as well as for marital status, educational level, ethnicity, smoking, alcohol use, prepregnancy body mass index (linear and quadratic), and parenting stress.

*P < .05; **P < .01; ***P < .001.

Stratified analyses comparing women in the smoking and nonsmoking groups revealed larger combined effects of high physical workload and a long workweek on birthweight in the former group than in the latter group. However, the effect was still significant (P = .021) in the nonsmoking group after full adjustment (relative to the reference group, mean birthweight decreases of 80 g among nonsmokers and 275 g among smokers). This difference between smokers and nonsmokers was not observed in the high job strain category (mean birthweight decreases of 151 g among nonsmokers and 140 g among smokers).

DISCUSSION

In this community cohort of pregnant women, long workweeks and high job strain, measured during women's first trimester, were independently related to reductions in birthweight. The combination of high job strain and a long workweek (the situation experienced by 4.2% of the working women in our study) showed the highest impact, with (after full adjustment) an average birthweight reduction of 150 g and a 100% increase in the risk of delivering an SGA infant. This independent association between job strain and birthweight equaled that shown in previous research on smoking during pregnancy.8,23

The most likely biological explanation is that high job strain leads to the release of stress hormones, such as norepinephrine and cortisol, that impair fetal growth as a result of HPA dysregulation, which is particularly detrimental during the first trimester.25,26,31 This effect during the first trimester also explains why large, well-designed studies of stressful working conditions during the third trimester of pregnancy have not revealed birthweight effects and, moreover, why we found a strong combined effect of job strain and working hours.

The high job strain effect observed here closely matches the results of Oths et al.,9 who reported a birthweight reduction of 190 g among women at high job strain levels, and Brandt and Nielsen,10 who found a 1.46 higher risk of low birthweight among such women. In addition, Wergeland et al.24 found a protective effect of perceived influence on one's work pace, an important component of job strain, on birthweight. The association of high job strain with long working hours and reduced birthweight, however, has no precedent, although Hatch et al.12 suggested an additive effect of long workweeks and high physical workloads on birthweight (Hatch et al. did not report SGA effects). Also, independent associations between working hours and birthweight have been reported in some studies12,13,17,18 but not others.6,22,24 In the absence of a common reference group in these studies, however, the effects of working hours cannot easily be established.

The association between high physical workload and delivery of an SGA infant was convincingly shown in the meta-analysis conducted by Mozurkewich et al.16; regrettably, however, other working conditions were not evaluated with respect to delivery of an SGA infant. In their systematic review, Bonzini et al.19 considered other working conditions (prolonged working hours, shift work, lifting, standing, and heavy physical workload); however, they noted that firm conclusions could not be drawn because of design differences and small sample sizes in the studies reviewed. Our study showed effect sizes similar to those found by Bonzini et al., but, as a result of our large sample and sufficiently sized risk groups, we were able to estimate effects at a higher level of precision.

In one important study that offered dissenting results, Klebanoff et al.22 did not observe work-related birthweight or SGA effects. In that study, the high-risk group consisted of medical residents who were working, on average, more than 70 hours per week but had an otherwise healthy profile (as indirectly reflected by their low 5% rate of SGA births). Two interpretational difficulties arise from a comparison of that population and our sample. First, the “healthy worker effect” could apply, suggesting that the severe job demands assumed (albeit not verified by self-reported data) in the Klebanoff et al. study neither were experienced as extreme nor evoked detrimental stress responses. Second, the extremely high SES and health status profile of that population may have compensated for work-related adverse birthweight effects. Our study, which included women from diverse ethnic and SES groups, is more likely to represent typical work-related effects in an average urban population.

We focused on birthweight and delivery of an SGA infant in parallel, whereas most studies have focused on one or the other of these outcomes.79,12,14,17,22 In terms of clinical relevance, delivery of an SGA infant is a more important outcome; however, in addition to providing increased statistical power, our use of birthweight as a continuous measure enabled us to capture more physiological evidence on overall weight effects.40 In this cohort of term infants, reduced birthweight linearly translated to a higher risk of an SGA birth. (We included only term infants to rule out possible effects of heavy working conditions on preterm birthweights.16,21 Including preterm deliveries did not alter birthweight effect sizes but slightly increased the risk of delivering an SGA infant; see the appendix available as an online supplement to this article at http://ajph.org.)

Limitations

Our study involved several limitations. First, we measured working conditions only during the first trimester. Whether working conditions changed during pregnancy is unknown; it is therefore possible that first-trimester working conditions continued until the third trimester. Changes during pregnancy were most likely in the highest work exposure groups8 (e.g., women with highly physical workloads may have moved to a desk job). Such attenuations in exposure would imply that our estimates were conservative. Some studies have included multiple measurements during pregnancy but have restricted analyses to women who worked throughout their pregnancy.21 This approach leads to underestimates of early-pregnancy workload effects and may even result in favorable rather than adverse work effects among those who work to term, especially if early quitting is associated with work-related pregnancy complications such as suspected intrauterine growth restriction.

Second, the percentage of unemployment was high in our cohort (36%). This can be explained in part by our definition of employment as working at least 8 hours per week during the first trimester. Given that most studies include only working women,7,8,11,14,17,18,21,22 comparisons between previous studies and our investigation are difficult. However, the unemployment rate in the Netherlands among women in the 25- to 34-year age group is 24.7%,41 which is high relative to other Western countries. In our cohort, the percentage was higher than the norm as a result of the comparatively large group of women of non-Dutch origin, among whom, according to national statistics, rates of unemployment are often high.41 We believe that our employment rate was representative of large cities in the Netherlands and that selective participation among women who were unemployed did not occur.

Third, it is possible that adverse working conditions are indicative of lower SES, which in itself is associated with lower birthweights and delivery of SGA infants.6,20,21 The adverse unemployment effect observed here supports this relationship. Although education, profession, and income are all components of SES, many studies focusing on community populations indicate that SES affects health status primarily through employment (in addition to smoking) and, to a lesser extent, education. Indeed, more women in the low education group than in the high education group reported high physical workloads (28.8% vs 4.2%) and high job strain (15.2% vs 3.4%). However, an independent educational pathway apart from work can be ruled out, given that the associations between birthweight and high physical workload and high job strain did not change when educational level was removed from model 2.

Incomplete adjustment for alcohol use or smoking also must be taken into consideration. However, alcohol use had no effect on birthweight even when measured in 3 categories. As shown, smoking was highly related to physical workload, which led to the physical workload effect on birthweight being nonsignificant after full adjustment. However, repeating the combined analysis in the nonsmoking group showed that high physical workloads and job strain in combination with a long workweek were still significantly associated with reduced birthweight after full adjustment. The effect estimates were reduced only with respect to high physical workloads. This means that residual confounding from smoking could have played a role with respect to physical workload.

Fourth, it is possible that women at high levels of job strain or physical workload were more likely than were their counterparts to be exposed to chemical and biological contaminants such as solvents, pesticides, and metals. Such exposures are known to affect birthweight and the risk of delivering an SGA infant.7,42

Study Implications

We found that in general there is no reason to assume that working during pregnancy has a negative influence on birthweight. However, the combination of high job strain and a long workweek seems to result in lower birthweights.

We believe that optimizing the work environment during pregnancy is important as the participation of women of reproductive age in the work force continues to increase. Although only small percentages of the working women in our cohort were in the highest job strain and longest workweek categories, women facing such conditions should not be ignored given that these percentages will be higher in other countries in which part-time employment is less common.

We are aware of the need for our results to be confirmed by future research. However, the considerable effects observed here, in combination with epidemiological data,9,10,26,28,29,31 animal studies,2527 and sound biological theory,31 justify a renewed focus on the first trimester of pregnancy. Although pregnant women typically reduce their working hours or workloads at the end of their pregnancy, our results suggest that reducing job strain and working hours in the initial stages of pregnancy may be beneficial among women with stressful full-time jobs.

Acknowledgments

This study was supported by a research grant from the Netherlands Organisation for Health Research and Development (2100.0076).

We are grateful to all of the participating hospitals, obstetric clinics, and general practitioners for their assistance in the implementation of the Amsterdam Born Children and their Development study. Also, we thank the participating pregnant women for their cooperation. Finally, we are indebted to the LinKID research team, which provided national reference data on birthweight distributions according to gestational age, infant gender, and parity.

Human Participant Protection

This study was approved by the medical ethics committees of all Amsterdam hospitals and the Registration Committee of Amsterdam. All of the participants provided written informed consent.

References

  • 1.McIntire DD, Bloom SL, Casey BM, Leveno K. Birth weight in relation to morbidity and mortality among newborn infants. N Engl J Med 1999;340:1234–1238 [DOI] [PubMed] [Google Scholar]
  • 2.Yanney M, Marlow N. Paediatric consequences of fetal growth restriction. Semin Fetal Neonatal Med 2004;9:411–418 [DOI] [PubMed] [Google Scholar]
  • 3.Barker DJP. Fetal origins of coronary heart disease. BMJ 1995;311:171–174 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Osmond C, Barker DJ. Fetal, infant, and childhood growth are predictors of coronary heart disease, diabetes, and hypertension in adult men and women. Environ Health Perspect 2000;108(suppl 3):545–553 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Ergaz Z, Avgil M, Ornoy A. Intrauterine growth restriction—etiology and consequences: what do we know about the human situation and experimental animal models? Reprod Toxicol 2005;20:301–322 [DOI] [PubMed] [Google Scholar]
  • 6.Savitz DA, Olshan AF, Gallagher K. Maternal occupation and pregnancy outcome. Epidemiology 1996;7:269–274 [DOI] [PubMed] [Google Scholar]
  • 7.Chen D, Cho SI, Chen C, et al. Exposure to benzene, occupational stress, and reduced birth weight. Occup Environ Med 2000;57:661–667 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Croteau A, Marcoux S, Brisson C. Work activity in pregnancy, preventive measures, and the risk of delivering a small for gestational age infant. Am J Public Health 2006;96:846–855 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Oths KS, Dunn LL, Palmer NS. A prospective study of psychosocial job strain and birth outcomes. Epidemiology 2001;12:744–746 [DOI] [PubMed] [Google Scholar]
  • 10.Brandt LPA, Nielsen CV. Job stress and adverse outcome of pregnancy: a causal link or recall bias? Am J Epidemiol 1992;135:302–311 [DOI] [PubMed] [Google Scholar]
  • 11.Henriksen TB, Hedegaard M, Secher NJ. The relation between psychosocial job strain, and preterm delivery and low birthweight for gestational age. Int J Epidemiol 1994;23:764–774 [DOI] [PubMed] [Google Scholar]
  • 12.Hatch M, Ji BT, Shu XO, Susser M. Do standing, lifting, climbing, or long hours of work during pregnancy have an effect on fetal growth? Epidemiology 1997;8:530–536 [DOI] [PubMed] [Google Scholar]
  • 13.Tuntiseranee P, Geater A, Chongsuvivatwong V, Kor-anantakul O. The effect of heavy maternal workload on fetal growth retardation and preterm delivery: a study among southern Thai women. J Occup Environ Med 1998;40:1013–1021 [DOI] [PubMed] [Google Scholar]
  • 14.Hanke W, Kalinka J, Makowiec-Dabrowska T, Sobala W. Heavy physical work during pregnancy—a risk factor for small-for-gestational-age babies in Poland. Am J Ind Med 1999;36:200–205 [DOI] [PubMed] [Google Scholar]
  • 15.Homer CJ, James SA, Sieger E. Work-related psychosocial stress and risk of preterm, low birthweight delivery. Am J Public Health 1990;80:173–177 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Mozurkewich EL, Luke B, Avni M, Wolf FM. Working conditions and adverse pregnancy outcome: a meta-analysis. Obstet Gynecol 2000;95:623–635 [DOI] [PubMed] [Google Scholar]
  • 17.Peoples-Sheps MD, Siegel E, Suchindran CM, Origasa H, Ware A, Barakat A. Characteristics of maternal employment during pregnancy: effects on low birthweight. Am J Public Health 1991;81:1007–1012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Ceron-Mireles P, Harlow S. Sanchez-Carrillo. The risk of prematurity and small-for-gestational-age birth in Mexico: the effects of working conditions and antenatal leave. Am J Public Health 1996;86:825–831 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Bonzini M, Coggon D, Palmer KT. Risk of prematurity, low birth weight and pre-eclampsia in relation to working hours and physical activities: a systematic review. Occup Environ Med 2007;64:228–243 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Woo GM. Daily demands during pregnancy, gestational age, and birthweight: reviewing physical and psychological demands in employment and non-employment contexts. Ann Behav Med 1997;19:385–398 [DOI] [PubMed] [Google Scholar]
  • 21.Marbury MC. Relationship of ergonomic stressors to birthweight and gestational age. Scand J Work Environ Health 1992;18:73–83 [DOI] [PubMed] [Google Scholar]
  • 22.Klebanoff MA, Shiono PH, Rhoads GG. Outcomes of pregnancy in a national sample of resident physicians. N Engl J Med 1990;323:1040–1045 [DOI] [PubMed] [Google Scholar]
  • 23.Pompeii LA, Savitz DA, Evenson KR, Rogers B, McMahon M. Physical exertion at work and the risk of preterm delivery and small-for-gestational-age birth. Obstet Gynecol 2005;106:1279–1288 [DOI] [PubMed] [Google Scholar]
  • 24.Wergeland E, Strand K, Bordahl PE. Strenuous working conditions and birthweight, Norway 1989. Acta Obstet Gynecol Scand 1998;77:263–271 [PubMed] [Google Scholar]
  • 25.Schneider ML, Roughton EC, Koehler AJ, Lubach GR. Growth and development following prenatal stress exposure in primates: an examination of ontogenetic vulnerability. Child Dev 1999;70:263–274 [DOI] [PubMed] [Google Scholar]
  • 26.van den Bergh BRH, Mulder EJH, Mennes M, Glover V. Antenatal maternal anxiety and stress and the neurobehavioural development of the fetus and child: links and possible mechanisms. A review. Neurosci Biobehav Rev 2005;29:237–258 [DOI] [PubMed] [Google Scholar]
  • 27.Weinstock M. Alterations induced by gestational stress in brain morphology and behaviour of the offspring. Prog Neurobiol 2001;65:427–451 [DOI] [PubMed] [Google Scholar]
  • 28.Wadhwa PD, Dunkel-Schetter C, Chicz-DeMet A, Porto M, Sandman CU. Prenatal psychosocial factors and the neuroendocrine axis in human pregnancy. Psychosom Med 1996;58:432–446 [DOI] [PubMed] [Google Scholar]
  • 29.Austin MP, Leader L. Maternal stress and obstetric and infant outcome: epidemiological findings and neuroendocrine mechanisms. Aust N Z J Obstet Gynaecol 2000;40:331–337 [DOI] [PubMed] [Google Scholar]
  • 30.Paarlberg KM, Vingerhoets AJJM, Passchier J, Dekker GA, Heinen AGJJ, van Geijn HP. Psychosocial predictors of low birthweight: a prospective study. BJOG 1999;106:834–841 [DOI] [PubMed] [Google Scholar]
  • 31.Gluckman P, Hanson M. The Fetal Matrix: Evolution, Development and Disease. Cambridge, England: Cambridge University Press; 2005 [Google Scholar]
  • 32.van Eijsden M, van der Wal MF, Bonsel GJ. Folic acid knowledge and use in a multi-ethnic pregnancy cohort: the role of language proficiency. BJOG 2006;113:1446–1451 [DOI] [PubMed] [Google Scholar]
  • 33.Benhadi N, Wiersinga WM, Reitsma JB, Vrijkotte T, van der Wal M, Bonsel GJ. Ethnic differences in TSH but not in free T4 concentrations or TPO- antibodies during pregnancy. Clin Endocrin (Oxf) 2007;66:765–770 [DOI] [PubMed] [Google Scholar]
  • 34.Alderliesten ME, Vrijkotte TGM, van der Wal MF, Bonsel GJ. Late start of antenatal care among ethnic minorities in a large cohort of pregnant women. BJOG 2007;114:1232–1239 [DOI] [PubMed] [Google Scholar]
  • 35.Evers IM, de Valk HW, Visser GHA. Risk of complications of pregnancy in women with type 1 diabetes: nationwide prospective study in the Netherlands. BMJ 2004;328:915–919 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Karasek R, Brisson C, Kawakami N, Houtman I, Bongers P, Amick B. The Job Content Questionnaire (JCQ): an instrument for internationally comparative assessments of psychosocial job characteristics. J Occup Health Psychol 1998;3:322–355 [DOI] [PubMed] [Google Scholar]
  • 37.Houtman IL, Goudswaard A, Dhondt S, van der Grinten MP, Hildebrandt VH, van der Poel EGT. Dutch monitor on stress and physical load: risk factors, consequences, and preventive action. Occup Environ Med 1998;55:73–83 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Crnic KA, Greenberg MT. Minor parenting stresses with young children. Child Dev 1990;61:1628–1637 [DOI] [PubMed] [Google Scholar]
  • 39.Perinatal Care in the Netherlands 2001. Utrecht, Netherlands: Netherlands Perinatal Registry; 2001 [Google Scholar]
  • 40.Savitz DA, Hertz-Picciotto I, Poole C, Olshan AF. Epidemiologic measures of the course and outcome of pregnancy. Epidemiol Rev 2002;24:91–101 [DOI] [PubMed] [Google Scholar]
  • 41.Statistics Netherlands StatLine. Available at: http://statline.cbs.nl. Accessed September 28, 2008
  • 42.Windham G, Fenster L. Environmental contaminants and pregnancy outcomes. Fertil Steril 2008;89(suppl 1):e111–e116 [DOI] [PubMed] [Google Scholar]

Articles from American Journal of Public Health are provided here courtesy of American Public Health Association

RESOURCES