Night Shift Work and Fecundability in Late Reproductive-Aged African American Women

Abstract

Background: We estimated the association between night shift work and fecundability among African American women.

Methods: Black Women's Health Study participants (n = 560) aged 30–45 years reported their history of night shift work in 2005. Time to pregnancy for all pregnancies resulting in a livebirth was reported in 2011. We estimated the fecundability ratio (FR) and 95% confidence interval (CI) using proportional probabilities regression, accounting for multiple observations of individual women using generalized estimating equations.

Results: We observed 4,417 months of pregnancy attempt time resulting in 390 births. After adjustment for covariates, women who reported ever working night shifts had 20% lower fecundability compared with those who never reported night shift work (FR = 0.80, 95% CI: 0.59–1.04). The FR for women reporting night shift work with a frequency of ≥1 time per month and a duration of ≥2 years was 0.65 (95% CI: 0.47–0.94) relative to women reporting no shift work. We observed a decrease in fecundability associated with ever working night shifts (FR = 0.74, 95% CI: 0.56–0.96) among women aged ≥35 years, but not among younger women (FR = 1.33, 95% CI: 0.78–2.28).

Conclusion: A history of working night shifts was associated with reduced fecundability among older reproductive-aged African American women attempting pregnancy.

Introduction

Nearly 17% of all U.S. employees engage in shift work, defined as work occurring between the hours of 6:00 PM and 6:00 AM.¹ The majority of shift work occurs in the evening (between 2 PM and 12 AM—27%), at night (between 9 PM and 8 AM—23%), or on a rotating basis (18%). Two thirds of shift workers are of reproductive ages, 20–44 years.¹ Working alternative shifts is of concern to health and reproduction as changing sleep/wake schedules and exposure to light at night may disrupt the circadian system.² The circadian clock is a key input to the hypothalamus–pituitary–gonadal (HPG) and hypothalamus–pituitary–adrenal (HPA) axes.³ The HPG axis regulates reproduction,⁴ and the HPA axis coordinates the body's homeostatic functions.⁵

Several studies have reported associations between shift work and reproductive outcomes. A recent meta-analysis including eight cross-sectional, two case–control, and five cohort studies found evidence for modest associations between shift work and menstrual cycle irregularity (odds ratio [OR] = 1.15, 95% confidence interval [CI]: 1.01–1.50, n = 4 studies, 71,681 premenopausal women) and miscarriage (OR = 1.41, 95% CI: 1.22–1.63, n = 7 studies, 23,243 women aged >18 years).⁶ Other studies have reported that shift work is associated with adverse birth outcomes such as miscarriage⁷ and low birth weight.⁸ Among studies that have investigated the association between female shift work and fertility, results have been inconsistent. Three studies (two retrospective^9,10 and one prospective¹¹) reported inverse associations of shift work with measures of fertility. However, other prospective,¹² retrospective,¹³ and cross-sectional¹⁴ studies have observed little association between shift work and female fertility.

African American women are more likely to work night shifts than white women.¹ Limited evidence suggests that infertility rates are also higher among African American women.¹⁵ However, we know of no studies of the association between night shift work and fecundability among African American women. We therefore assessed the association between night shift work and time to pregnancy (TTP) in the Black Women's Health Study (BWHS), a large cohort study of African American women.

Methods

Study population

Initiated in 1995, the BWHS is an ongoing prospective cohort study of 59,000 African American women.¹⁶ Women aged 21–69 years were enrolled through mailed questionnaire. Information on demographics, anthropometric factors, behavioral characteristics, reproductive history, contraceptive use, and medical history was collected at baseline. Participants update their health information through mailed follow-up questionnaire at 2-year intervals. Follow-up is ∼80% through 2011. Respondents live throughout the United States, with the majority living in California, Georgia, Illinois, Michigan, New Jersey, and New York. The study protocol was approved by the Institutional Review Board at Boston University Medical Center.

BWHS participants were given the opportunity to complete the follow-up questionnaire by mail or on the web at the beginning of each questionnaire cycle beginning in 2003.¹⁷ Among the 42,616 women who responded in 2011, 40% completed the follow-up questionnaire through the web. Reporting on the paper and web-based questionnaires was similar with regard to night shift work history (ever: 32.0% vs. 33.0%), frequency (< once/month: 7.0% vs. 8.0% and monthly or weekly: 23.0% vs. 22.0%) and duration (<2 years: 11.0% vs. 9.0%; ≥2 years: 17.0% vs. 19.0%), mean body mass index (BMI) (27.7 vs. 27.9 kg/m²), education (15.2 vs. 14.6 years), smoking (current: 13.0% vs. 15.0%; former: 21.0% vs. 19.0%; and never: both 66.0%), and childbearing history (parous: 61.7% vs. 65.7%; mean births: 2.1 vs. 2.2; mean age at first birth: 22.8 vs. 22.2 years) after adjustment for the difference in age between the two groups of respondents (mean 40.4 vs. 36.7 years). The women excluded from the study because they did not complete the web-based questionnaire do not appear to differ from women included in the study in ways that may have substantially influenced our results.

Assessment of TTP

Women who completed the web-based follow-up questionnaire in 2011 were invited to continue to a pregnancy questionnaire; 15,862 women completed it.¹⁸ The pregnancy questionnaire ascertained if the woman had ever given birth to a child, the calendar year of each birth, and whether the pregnancy was planned. For planned pregnancies, women reported whether they used fertility medications to conceive and the number of months it took to become pregnant. Women who never succeeded in becoming pregnant were identified by the question: “Have you ever tried for 12 or more months to become pregnant without success?” and, if yes, “How old were you at that time?” TTP was defined as the reported number of months between the beginning of the pregnancy attempt and conception.

Assessment of night shift work

Participants' history of shift work was assessed only in 2005. We asked participants, “Have you ever worked a night shift (graveyard shift, Midnight to 8 AM)?” Those who answered positively were asked how often they worked night shifts (never, once a year, a couple of times a year, once a month, or every week), and how many years they worked night shifts. Based, in part, on the distribution of the responses to the frequency of night shift work, we categorized frequency of working night shifts as never, <1/month, or ≥1/month. Duration was categorized as never, <2 years, or ≥2 years. Frequency and duration were combined into a single variable, including never, <1/month for <2 years, ≤1/month for ≥2 years, ≥1/month for <2 years, and ≥1/month for ≥2 years. Due to small numbers, we combined the <1/month for <2 years and <1/month for ≥2 years categories for some analyses.

Assessment of covariates

Covariates of interest included age and parity at pregnancy attempt, BMI, smoking status, alcoholic drinks per week, education, income, and state of residence. BWHS participants reported their height (inches) and years of education in 1995. In 2003, participants reported their household income. Other covariates were reported at baseline and updated on the follow-up questionnaires. We calculated BMI as weight (kg)/height (m)².¹⁹ Covariates other than education and income were treated as time-varying covariates and were taken from the questionnaire preceding the calendar year in which a woman reported a birth.

Study population

The study sample comprised of women who were premenopausal and below age 45 in 2005 (when shift work was assessed) and who had completed the 2011 questionnaire on pregnancies (n = 7,110) (Fig. 1). We excluded women who reported infertility before 2005 (n = 568) or had missing values for night shift work (n = 1,059). The remaining 5,483 women contributed 5,893 planned pregnancy attempts resulting in births. In addition to these successful attempts, we included 248 attempts from women who reported an attempt of >12 months without success but no pregnancies and 350 attempts of >12 months from women who reported ≥1 other pregnancies. We excluded 5,780 of these 6,491 attempts occurring before exposure was assessed, and 11 for which the mother's age at the attempt was >45 years. An additional 11 successful attempts were excluded due to missing data on TTP and 15 attempts because the couple used fertility drugs before the 12th month. Seven attempts resulting in multiple births were treated as a single pregnancy attempt. After these exclusions, 560 women contributed 4,417 months of attempt time resulting in 390 pregnancies.

FIG. 1.

Exclusions of women and pregnancy attempts from analysis, night shift work and time to pregnancy, 2005–2011.

Data analysis

We assessed the association between night shift work (history, frequency, duration, and intensity) and TTP in months. To ensure that the night shift work exposures preceded the study outcome (i.e., conduct a prospective analysis), we included only pregnancies occurring from January 1, 2005 through the date of completion of the 2011 questionnaire. We fit proportional probabilities regression models to estimate fecundability ratios (FRs) and 95% CIs,^20,21 accounting for multiple pregnancies per woman using a generalized estimating equations procedure in SAS (PROC GENMOD, link = cloglog, dist = bin).²² The FR represents the ratio of the average fecundability (i.e., the probability of conceiving in a given month of a pregnancy attempt) in the exposed and reference groups. An FR <1 indicates a longer TTP in the exposed group compared with the reference and therefore lower probability of conception. TTP was censored after 12 months if the woman had not conceived within that period. Women who reported never working night shifts served as the reference for all exposure variables.

We included known or suspected confounders of the association between night shift work and fecundability as covariates in our models: age at pregnancy attempt (continuous), calendar year of attempt (continuous), BMI (<25, 25–29, 30–34, ≥35 kg/m²), geographic region (Northeast, Midwest, West, South, and other), years of education (≤12, 13–15, 16, ≥17 years), income (<$25,000, $25,000–$49,999, $50,000–$99,999, ≥$100,000), smoking status (never, former, current), and alcohol consumption (never, <1 drink/week, 1–6 drinks, week, ≥7 drinks/week). Because fecundability is expected to be lower in later months of a given pregnancy attempt than in earlier months, month of pregnancy attempt was also included as a covariate in all models. All covariates other than years of education and income were treated as time varying.

Since older age is associated with reduced fertility and may modify the relationship between shift work and fecundability, we assessed the association between night shift work and fecundability within strata of age (30–34, 35–45 years). We tested for interaction by comparing models with cross-product terms between each exposure variable and age with those without the cross-product terms using the likelihood-ratio test. In addition, we conducted two sensitivity analyses. First, to explore whether night shift work is associated with TTP through mechanisms other than infertility, we restricted analyses to parous women. Second, we excluded the first month of attempt time for each couple. We reasoned that such restriction might reduce bias if couples desiring a child may retroactively assign intentionality to an unintentional pregnancy that happens quickly (e.g., TTP = 1 month).²³ All analyses were conducted using SAS version 9.4.²²

Results

There were few substantial differences in baseline characteristics between women who did and did not report ever working night shifts (Table 1). Women with a history of night shift work were slightly less likely to be parous at the time of the pregnancy attempt and to smoke, exercise more, and have lower income than women who reported never working a night shift. Ever and never night shift workers did not differ substantially in their alcohol consumption, former smoking, BMI, age at menarche, or education. Characteristics were similar among women who reported working less than one night shift per month compared with those who worked more, and women reporting working night shift for <2 years compared with those working the night shift longer.

Table 1.

Characteristics of Pregnancy Attempts by History of Night Shift Work, 2005

Characteristic	Night shift work		Night shift work frequency		Night shift work duration
	Never	Ever	<1/month	≥1/month	<2 years	≥2 years
	364	196	48	140	72	100
Age in 2005 (years), mean ± SD	34.9 ± 2.7	34.8 ± 2.9	34.9 ± 2.8	34.8 ± 2.9	35.2 ± 2.9	34.4 ± 2.8
Age at pregnancy attempt (years), mean ± SD	37.9 ± 3.0	38.0 ± 3.2	37.6 ± 3.0	38.2 ± 3.2	38.2 ± 3.1	37.9 ± 3.3
Parous at time of attempt, %	50.6	46.5	53.1	45.1	54.1	41.6
Current smoker, %	1.4	7.6	7.0	8.1	4.5	9.7
Former smoker, %	5.0	5.4	7.0	4.4	5.4	5.0
≥7 alcoholic drinks/week, %	3.1	1.3	2.0	1.2	1.2	1.7
BMI (kg/m2), mean	27.0 ± 6.0	29.1 ± 6.8	27.0 ± 6.0	29.5 ± 7.0	29.3 ± 6.8	28.9 ± 6.7
Age at menarche (years), mean ± SD	12.2 ± 1.5	12.1 ± 1.4	12.2 ± 1.5	12.0 ± 1.4	12.0 ± 1.3	12.1 ± 1.4
Education (years), mean ± SD	15.9 ± 1.1	15.5 ± 1.4	15.6 ± 1.4	15.5 ± 1.3	15.7 ± 1.4	15.5 ± 1.3
Residence in Northeast, %	26.3	23.2	24.6	23.7	20.9	27.0
Residence in West, %	16.0	17.4	22.6	21.2	17.1	19.4
Residence in Midwest, %	18.4	21.5	16.2	16.6	24.3	16.3
Residence in South, %	38.9	37.7	30.8	37.3	37.7	34.0
Income in 2003 (US$), %
≤50,000	3.6	7.6	9.4	6.1	5.4	8.0
50,001–100,000	41.1	41.1	34.4	42.8	41.8	38.7
>100,000	27.4	16.0	19.5	16.1	14.3	18.2

Number missing: age in 2005, 0; age at pregnancy attempt, 0; parity at time of attempt, 0; smoking status, 1; alcoholic drinks/week, 0; BMI, 4; age at menarche, 1; education, 2; geographic region of residence, 0; income in 2003, 20.

BMI, body mass index; SD, standard deviation.

In multivariable-adjusted models, women who ever worked the night shift were 20% less likely to achieve pregnancy in a given menstrual cycle compared with women who reported never working the night shift (FR = 0.80, 95% CI: 0.63–1.04) (Table 2). The FRs were similar for women reporting working night shifts <1/month and women reporting monthly or weekly night shifts (FR = 0.87, 95% CI: 0.56–1.36 and FR = 0.78, 95% CI: 0.59–1.04, respectively; reference = never worked night shifts). Compared with never working night shifts, working the night shift for ≥2 years was associated with reduced fecundability (FR = 0.72, 95% CI: 0.53–0.99). When we cross-classified the frequency and duration of night shift work variables, the reduction in fecundability was greatest among women who worked ≥1 night shift per month for ≥2 years (FR = 0.65, 95% CI: 0.47–0.94, reference = never worked night shifts).

Table 2.

Association Between Night Shift Work and Fecundability (2005–2011)

	n	Pregnancies	Months	Fecundability	Unadjusteda		Adjustedb
					FR	95% CI	FR	95% CI
Night shift work
Neverc	364	277	2,722	10.2	1.00	Reference	1.00	Reference
Ever	196	113	1,695	6.7	0.77	0.61–0.97	0.80	0.63–1.04
Night shift frequency
<Once a month	48	32	35	8.3	0.82	0.52–1.27	0.87	0.56–1.36
Monthly or weekly	140	79	1,231	6.4	0.78	0.60–1.01	0.78	0.59–1.04
Night shift duration
<2 years	72	50	605	8.3	0.92	0.65–1.28	0.93	0.65–1.33
≥2 years	100	54	887	6.1	0.71	0.52–0.97	0.72	0.53–0.99
Night shift intensity
<1/month, <2 years	20	14	157	8.9	0.83	0.45–1.51	0.83	0.43–1.60
<1/month, ≥2 years	51	35	447	7.8	0.91	0.46–1.81	0.98	0.52–1.85
≥1/month, <2 years	20	15	160	9.4	0.93	0.63–1.38	0.95	0.63–1.45
≥1/month, ≥2 years	79	39	715	5.5	0.66	0.48–0.92	0.65	0.47–0.94

^aAdjusted for age at attempt, month of attempt time.

^bAdjusted for age at attempt, month of attempt time, calendar year, BMI, geographic region, income, education, smoking status, and alcohol consumption.

^cNever shift work is the reference for all comparisons.

CI, confidence interval; FR, fecundability ratio.

In stratified analyses by age, the inverse association between history of night shift work and fecundability was primarily seen among women 35–45 years old (Table 3). The FR for older women who had ever worked night shifts was 0.74 (95% CI: 0.56–0.96), compared with 1.33 (95% CI: 0.78–2.28) among women 30–34 years old who had worked night shifts (p_interaction = 0.08). Associations between night shift work and TTP were similar among parous women (Supplementary Table S1). When we excluded the first cycle from the analysis, associations were somewhat stronger (Supplementary Table S2).

Table 3.

Age-Stratified Association Between Night Shift Work and Fecundability (2005–2011)

	Age 30–34 years					Age 35–45 years					pinteraction
	n	Pregnancies	Rate	FR	95% CI	n	Pregnancies	Rate	FR	95% CI
Night shift work
Nevera	57	50	17.8	1.00	Reference	323	227	9.3	1.00	Reference
Ever	29	26	20.2	1.33	0.78–2.28	174	87	5.6	0.74	0.56–0.96	0.08
Night shift frequency
<Once a month	10	9	17.0	1.02	0.41–2.54	41	23	9.3	0.85	0.54–1.35
Monthly or weekly	19	17	22.4	1.61	0.91–2.86	125	62	5.4	0.76	0.55–1.05	0.08
Night shift duration
<2 years	15	15	20.3	1.48	0.72–3.06	62	35	6.6	0.92	0.61–1.41
≥2 years	12	9	17.0	0.97	0.48–1.96	90	45	5.4	0.74	0.53–1.03	0.30
Night shift intensity
<1/month, any duration	9	8	15.4	0.80	0.31–2.07	34	21	7.9	0.88	0.55–1.43
≥1/month, <2 years	9	9	24.3	2.96	1.20–7.30	46	26	6.3	0.88	0.54–1.44
≥1/month, ≥2 years	9	7	18.4	0.83	0.35–1.96	70	32	4.7	0.65	0.44–0.94	0.40

Adjusted for age at attempt, month of attempt time, calendar year, BMI, geographic region, income, education, smoking status, and alcohol consumption.

^aNever shift work is the reference for all comparisons.

Discussion

In this study of African American women aged 30–45 years at the time of their pregnancy attempts, we found reduced fecundability among women planning pregnancies with a history of night shift work. The inverse association was observed primarily among women who reported working night shifts more frequently (>1/month) and for longer durations (≥2 years), and among women aged ≥35 years.

Night shift work may adversely affect reproduction by disrupting the circadian system. Rodent studies indicate that information about environmental light detected by the retina is sent to the superchiasmatic nucleus (SCN). The SCN uses this optically derived light information to regulate circadian rhythms that organize regular activity, responses to stressors, and metabolism.²⁴ Centrally, the HPG axis controls the timing of ovulation in women. Exposure to light and activity during the normal sleep period may cause the circadian clocks to become desynchronized and disrupt the hormonal regulation of reproduction.²

The circadian clocks also interact with the HPA axis.³ The HPA axis, whose primary output is the release of glucocorticoids from the adrenal gland, plays a large role in coordinating metabolism with feeding, daily activity, and response to stressors.³ Glucocorticoids can inhibit reproduction at multiple levels, by influencing gonadotropin-releasing hormone levels, ovarian hormone production, follicle development, implantation, and pregnancy maintenance.²⁵ High levels of glucocorticoids are also associated with obesity, which has itself been linked to altered hormonal levels²⁶ and ovulatory infertility.²⁷

Consistent with the potential influence of circadian disruption on reproductive function, shift work has been associated with numerous reproductive outcomes. Several studies have reported menstrual cycle irregularity among shift workers.^28–32 Noting a high potential for positive confounding in the included studies, a meta-analysis of the association between working fixed nights and miscarriage found a 50% increase in risk, but little or no association with working rotating shift schedules (including nights).⁷ Long or irregular menstrual cycles have been associated with longer TTP,³³ while loss of pregnancies before clinical recognition may create the perception of a longer TTP.^34,35

Although the results of prior studies of the association between shift work and fertility have been mixed, our results are similar to prior studies that reported ∼10% (Denmark, n = 10,886 women, 73% aged <30 years)³⁶ to 20% (Sweden, n = 972, 77% aged <35 years, and Danish National Birth Cohort, Denmark, n = 21,438, 90% aged <35 years)^10,11 lower fecundability, and a 24% increase in the risk of a TTP ≥9.5 months (European Studies on Infertility and Subfecundity, seven European countries, n = 2,515, 71% aged <30 years)⁹ among night shift workers. Others have concluded that shift work is not associated with fecundability (European Studies on Infertility and Subfecundity, Italy, n = 622, 87% aged <35 years),¹³ duration of current pregnancy attempt (Nurses' Health Study, United States, n = 1,739, 77% aged <37 years, 37 African American women),¹² or TTP >12 months (Thailand, n = 907, 65% aged <30 years).¹⁴ A 2014 meta-analysis concluded that, after adjustment for covariates, there was little association between shift work and infertility.⁶ However, the authors noted that the high degree of heterogeneity between studies in their analysis (I² = 0.61) may be due to differences in the definitions of shift work used as well as in methods and outcomes.⁶ A recent study in an online prepregnancy cohort (n = 6,873, North America, 84% white, mean age 29.9 years) reported that short sleep, but not shift work, was associated with reduced fecundity.³⁷ Shift work was only assessed within the month before the questionnaire in this study.³⁷

There was little evidence of an association between night shift work and fecundability among women <35 years in our study. Among younger women, those reporting more frequent night shifts but <2 years' duration of night shift work had nearly threefold higher fecundability compared with women who never worked night shifts. Because there is no reason to think that night shift work would increase fecundability, this seems likely to be due to random variation in estimates based upon a small number of pregnancy attempts. Our results were consistent with a stronger negative association of night shift work with fecundability among women who worked night shifts more frequently and for a longer duration. Our results in age-stratified analyses may therefore be attributable, in part, to a longer history of shift work among older women. The Nurses' Health Study did not find evidence that duration of current pregnancy attempt was related to duration of shift work.¹² Associations were somewhat stronger when we excluded the first month of the attempt from the analysis. This result is consistent with a situation where shift workers are more likely to misclassify pregnancies with respect to intent.

There were several strengths of the current analysis. To our knowledge, this is the first study to examine the association between night shift work and fecundability among African American women, who are more likely than white women to work evening and night shifts.¹ Our inclusion of data on pregnancy attempts that did not result in a pregnancy mitigated potential bias that would be induced if night shift work results in infertility in some women. In addition, the use of a cohort that was not solely identified by occupation and the inclusion of only pregnancies that occurred after the report of history of night shift work are expected to reduce the potential influence of the infertile worker effect, a selection bias caused when women who do not have children are more likely to remain in the workforce and be selected into a study.

Limitations of our study include the lack of data on the timing of the night shift work with respect to the pregnancy attempt. If any effect of night shift work on fecundability is of short duration, the resulting misclassification would be expected to bias the association with ever versus never working a night shift toward the null. Although we adjusted our models for several known or suspected risk factors for subfertility, which may plausibly be related to night shift work, we cannot exclude the possibility of residual confounding. To fully explain our results, residual confounding would have to be strongly associated with exposure and outcome and operate only among the older women in our study. Because the questions about shift work were asked only in 2005, we were unable to assess the association of interest among women in their early childbearing years. We did not collect data on coital frequency during the pregnancy attempt, which represents another pathway by which night shift work may influence fecundability but is of less etiologic interest. Prior studies that collected such data differ as to whether work schedule was¹⁴ or was not¹⁰ associated with lower coital frequency, and found that fecundability was not reduced if the coital frequency was >6 times per month.¹³ We also lacked data on exposures of the male partner. Data on TTP were collected retrospectively, which is likely to have introduced some error. Prior studies have reported reasonable accuracy for recalled TTP up to 20 years after the pregnancy.³⁸ We excluded unintentional pregnancies, for which TTP is undefined. Couples who have unplanned pregnancies may differ from those who plan their pregnancies in ways that could have influenced our results. Bias could therefore arise if shift work is associated with having an unplanned pregnancy.²⁰ Our sample of women and observed pregnancies were small, limiting our ability to detect potential differences in associations according to age and increasing the risk of chance findings. The lack of strong evidence for a dose–response relationship could be due to chance variation or nondifferential misclassification, or could be an indicator of residual confounding. Generalizability of our results beyond older reproductive-aged African American women may be limited.

Conclusion

Our results suggest that working night shifts was associated with reduced fecundability among older reproductive-aged women, particularly among those who work night shifts more frequently and for a longer duration. The relationship of shift work with fecundability is of particular concern to African American women, among whom this type of employment is more common.

Disclaimer

The content is solely the responsibility of the authors and does not necessarily represent the official view of the National Institutes of Health.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This work was supported by National Cancer Institute (U01-CA164974 UM1 CA164974) and National Institute of Health (T32 HL125232, T32 HL007224).

Supplementary Material

Supplementary Table S1

Supplementary Table S2