Secondary Logo

Journal Logo

Original Articles

Physical Exertion at Work and the Risk of Preterm Delivery and Small-for-Gestational- Age Birth

Pompeii, Lisa A. PhD, COHN-S1,2; Savitz, David A. PhD1; Evenson, Kelly R. PhD1; Rogers, Bonnie DrPH2; McMahon, Michael MD, MPH3†

Author Information
doi: 10.1097/01.AOG.0000189080.76998.f8
  • Free

Among first-time pregnant women in the United States, paid employment outside the home has increased dramatically during the past 3 decades, from 44% in 1961–1965 to 67% in 1991–1995.1 Women are also working later in pregnancy, with 73% of women working up to at least 1 month before delivery in 1991–1995 compared with 35% in 1961–1965.1 Accompanying this increase in workforce participation is the concern that women may be at risk for adverse pregnancy outcomes if they engage in jobs that are physically demanding. During the past decade several studies have examined the association between various types of occupational physical exertion and preterm delivery2–14,16 or small-for-gestational-age (SGA) birth,7,11,13–16 and the findings have been mixed.

Variations in study findings may be due to differences in how exertion has been measured and defined, with measures of prolonged standing ranging from a dichotomous measure of yes or no8 to standing at least 3 hours per day,2–4 at least 5 hours per day,9–11 or at least 8 hours per day,5 for example. Moreover, several studies have defined standing and lifting within a given workday without considering the number of days or hours worked per week.2–4,11,12 In addition, the timing of exposure to exertion during different periods of fetal development has only been examined in a few studies.10,15–16 Some methodologic issues seem pervasive in this body of literature, including inadequate sample size,2,3,6,7 failure to consider important known and suspected risk factors for adverse pregnancy outcomes,2–4,7,8,14 and the collection of work-related exertion information after the end of pregnancy,2–5,8,11,13–15 introducing possible bias related to the pregnancy outcome.

The purpose of this study was to assess a possible association between exposure to standing, lifting, night work, or long work hours and preterm delivery or SGA birth, focusing on specific periods during pregnancy. We hypothesize that exposure to physical exertion early in pregnancy could possibly affect the formation of the placenta, resulting in an increased risk of SGA birth, whereas exposure later in pregnancy could trigger preterm labor, especially if the exposure occurs in the weeks immediately preceding when most preterm deliveries occur (weeks 34–36). We examined exposure to exertion among a large cohort of women during the first trimester (weeks 1 to 12) and second trimester (weeks 13 to 27) in relation to preterm delivery and SGA birth. A third period of exposure was considered during the seventh month of pregnancy (weeks 28 to 31) in relation to preterm delivery only. We gathered the majority of exposure information from women before delivery, in addition to collecting information about a broad range of known and suspected risk factors for both pregnancy outcomes.


The Pregnancy, Infection and Nutrition study is a prospective cohort study, with a nested case–control component, of women who were recruited during prenatal visits through hospital clinics at the University of North Carolina, Wake County Human Services, and the Wake Area Health Education Center at the Wake Medical Center in Raleigh, North Carolina. Women were recruited into the study between 24 and 29 weeks of gestation and contacted within 2 weeks to participate in a telephone interview, and for women included in the nested case–control study, an additional face-to-face interview was conducted after delivery, before discharge from the hospital. The Pregnancy, Infection and Nutrition study received approval by the University of North Carolina School of Medicine and Wake Medical Center Institutional Review Boards.

Recruitment began in August 1995, and these analyses include women whose last menstrual period (LMP) fell between January 1995 and April 2000. Details of the Pregnancy, Infection and Nutrition study recruitment are described elsewhere.17 A total of 5,142 women met the Pregnancy, Infection and Nutrition eligibility criteria of being at least 16 years of age, English speaking, pregnant with a singleton gestation, having access to a telephone, and planning to begin prenatal care before the 29th week of gestation and deliver at one of the study sites. Of those eligible, 3,121 (61%) agreed to participate, and 87% (n = 2,706) of these women completed the telephone interview and had birth outcome information available. Because prior studies have found that women who work outside the home for pay are at a decreased risk for preterm delivery and SGA birth compared with women who do not work,18–19 we further excluded 737 non-working women and those who worked less than 28 days during the first 2 trimesters. Sixty-one second pregnancies to the same women in the Pregnancy, Infection and Nutrition study were omitted, retaining only their first Pregnancy, Infection and Nutrition pregnancy, resulting in a total of 1,908 women included in the preterm delivery analyses. We omitted an additional 196 women who delivered infants of other races for the SGA birth analyses, because we used fetal growth standards by Zhang and Bowes20 that include growth estimates for African-American and white infants only. This resulted in 1,712 women in the SGA birth analyses. The nested case–control sample (which is used for the seventh month of pregnancy analyses) includes all cases of preterm birth (n = 234) (not SGA) and a random selection of controls (n = 413). Women who delivered before the 32nd week of pregnancy (< 218 days of gestation), those missing interview data, or those who did not work during this period were omitted (n = 203), resulting in 135 cases and 309 controls.

During the telephone interview each woman provided information about the 2 longest-held jobs from 3 months before pregnancy up to the time of the interview and the start and stop dates for each job. Women selected a descriptor (never, rarely, sometimes, usually, almost always) that best reflected the amount of time during which they stood each day and the number of times they lifted 25 pounds or more each day. They also provided the number of hours and days worked per week, and whether they “regularly worked at night, from 10:00 PM to 7:00 AM” for either job. Nested case–control participants, who were interviewed again after delivery, were asked about any changes in employment since the telephone interview.

To better quantify the activities of standing and lifting at work, additional questions were added to the telephone interview during the last 6 months of study recruitment. Women who reported any standing (n = 272) or lifting (n = 77) were asked to provide the number of hours spent standing per day, the number of times they lifted per day, or both so that the descriptive terms could be quantified. Additional standing data were used to calculate a summary measure of the median percentage of time spent standing per workweek for each descriptor (never = 0%, rarely = 12.9%, sometimes = 26.6%, usually = 55.5%, and almost always = 91.7%). For lifting we calculated the median number of times one lifted per day for each descriptor (rarely = 1, sometimes = 2, usually and almost always = 4). This allowed us to estimate the number of hours spent standing (the median percent time standing multiplied by the total number of hours worked per week) and the number of times one lifted per week (the median number of times lifted per day multiplied by the total number of days worked per week).

The prolonged standing variable consisted of 4 levels of standing (0–5, 6 to 15, 16 to 29, or ≥ 30 hours per week), and lifting at least 25 pounds consisted of 3 levels, including ≥ 13 times per week, 1 to 12 times per week, and none. Women were considered to have worked at night if they reported working regularly during the hours of 10:00 PM to 7:00 AM for either job within the defined period. Working an extended workweek was defined as those who worked, on average, 46 or more hours per week, with others categorized as working either full-time (35 to 45 hours per week) or part-time (≪ 34 hours per week).21–22

Women provided information about cigarette smoking, vaginal bleeding or spotting, and vigorous leisure activity during the first and second trimester and anytime between the telephone interview and the end of pregnancy for those in the nested case–control. Vigorous leisure activity was defined as a dichotomous variable (yes or no) for women who engaged in vigorous exercise at least twice per week. Information about race, education, age, marital status, parity, household income, and prior preterm birth was also collected. Self-reported prepregnancy weight and measured height was gathered from the medical record for the purpose of determining body mass index (BMI), which was obtained for 87% of women included in these analyses.

Start and stop dates for each job were used to assign occupational exertion measurements for each week of pregnancy, which were used to calculate averages of exposure over each period of interest including the first trimester (weeks 1 to 12), the second trimester (weeks 13 to 27), and the seventh month (weeks 28 to 31) of pregnancy. Start and stop dates allowed us to sum exposures for women who worked 2 jobs concurrently within a period, and consider variations in exposure for women who stopped their first job and started a second.

Preterm birth was defined as delivering before 37 completed weeks of gestation, with gestational age determined by LMP and an ultrasound assessment. If the 2 measures agreed within 14 days, the LMP date was assigned (n = 1,326), otherwise, the ultrasound dating was assigned (n = 214). If a reliable LMP date was not available, the earliest ultrasound date was used (n = 210), and for missing ultrasound information, the LMP date was used (n = 158). Most (89%) of the ultrasonograms were taken before the 20th week of gestation. Fetal growth restriction was defined as infants whose birth weight fell below the 10th percentile.20

Data from the study cohort were used to examine the relationship between physical exertion and birth outcome during the first 2 periods (weeks 1 to 12 and weeks 13 to 27). Each type of exertion was considered independently of the other types. Log-linear modeling was used by means of SAS GENMOD procedure (SAS/STAT software, SAS Institute Inc., Cary, NC), to construct relative risk (RR) estimates and 95% confidence intervals (CIs) of preterm delivery and prevalence ratios (PRs) of SGA for each type of physical exertion within each period. For the third period (weeks 28 to 31), which only considered the pregnancy outcome of preterm delivery, nested case–control data were used to calculate odds ratios (ORs) through unconditional logistic regression modeling. Each period of analysis was considered separately from the other periods. Mother’s race, years of age, years of education, marital status, parity, prior preterm birth, and BMI were assessed as possible confounders. Covariates that could change over time, including vigorous leisure activity, vaginal bleeding, and smoking status, were constructed separately for each period of analysis. General associations between each covariate and occupational exposure variable, and pregnancy outcome, were examined using the χ2 statistical test. Each covariate was included in the full multivariate model, and if it changed the relationship between physical exertion and birth outcome by at least 10% when removed, it was considered a confounder and retained in the final model.

Secondary analyses were conducted to consider exposure to exertion among women who remained at work longer during pregnancy, including those who worked at some point during the first 2 trimesters (n = 1,599), and those who worked during the entire 27 weeks of the first 2 trimesters (n = 1,063). Additional analyses were also conducted to assess whether associations exist between household income and work-related exertion or pregnancy outcomes.


Approximately 12.3% (n = 234) of women delivered preterm, and 8.3% (n = 142) delivered an infant who was SGA (Table 1). Within racial groups, 11.9% of white women, 13.4% African-American women, and 8.6% of other races delivered preterm. The proportion of those who delivered an SGA infant was similar for African-American and white women (7.9% and 8.6%, respectively). Almost 25% of women who delivered preterm had experienced a prior preterm delivery. No significant findings were noted for comparisons of age, years of education, or marital status with either pregnancy outcome. A lower proportion of preterm delivery was observed among women who engaged in vigorous leisure activity during the second trimester compared with those who did not, whereas a higher proportion was observed among women with higher parity (≥ 2), or who reported vaginal bleeding (regardless of trimester). Smoking during the second trimester and low BMI (≪ 19.8) were associated with a higher prevalence of SGA birth.

Table 1
Table 1:
Frequencies of Demographic, Lifestyle, and Pregnancy Factors by Pregnancy Outcomes Among Working Women

Overall, the prevalence of working at least 1 week during the first or second trimester of pregnancy was high and varied little (94% and 90%, respectively), whereas fewer women (73%) in the nested case–control study worked at least 1 week during weeks 28–31. Women were more likely to report exposure to prolonged standing (> 30 hours per week) compared with the other types of exertion considered, with 25.2% and 19.9% standing at least 30 hours per week in the first and second trimesters, respectively (Table 2). A small proportion (9.7%) of women reported repeated lifting (≥ 13 times per week) during the first trimester, which declined to 5.8% in the second trimester. The prevalence of regular night work was low before the onset pregnancy (8.9%) (data not shown), but changed little over the first 2 trimesters, and working an extended workweek changed very little over these same periods. The prevalence of prolonged standing, repeated lifting, and long work hours during weeks 28–31 among women in the nested case–control study was slightly higher, and the prevalence of regular night work was lower compared with those reported by the full cohort during the second trimester.

Table 2
Table 2:
Frequencies of Occupational Physical Exertion During 3 Periods of Pregnancy Among Working Women

Jobs that required standing 30 hours or more per week, lifting repeatedly, and working at night were more often held by women between the ages of 20 and 24 years, unmarried women, and those with 12 years of education or less (Table 3 and Table 4). Women who were older (≥ 30 years of age) and more educated (≥ 16 years) were less likely to engage in jobs that were physically demanding; however, they were more likely to work an extended workweek (≥ 46 hours) compared with women who were younger with fewer years of education (Table 4). The distribution of physical exertion by demographic characteristics varied little between periods, so only frequencies for the first trimester of pregnancy have been provided.

Table 3
Table 3:
Distribution of Study Covariates by Occupational Standing and Lifting Among Women Working During the First Trimester of Pregnancy
Table 4
Table 4:
Distribution of Study Covariates by Night Work and Long Work Hours Among Women Working During the First Trimester of Pregnancy

Regardless of the period of exposure during pregnancy, no significant associations were observed for preterm delivery among women who stood at least 30 hours per week or lifted repeatedly compared with those women who stood 6 to 15 hours per week or those who did not lift, respectively (Table 5). An elevation in preterm delivery was observed among women who worked at night during the first (RR 1.5, 95% CI 1.0–2.0) or second (RR 1.6, 95% CI 1.0–2.3) trimesters compared with those who did not work at night. Working at night during weeks 28–31 of pregnancy was associated with almost a 2-fold increase of preterm delivery (OR 1.8, 95% CI 0.8–3.9), but this estimate was imprecise. Women who worked an extended workweek during the first trimester were found to be at a somewhat lower risk of preterm birth (RR 0.6, 95% CI 0.4–0.9) compared with women who worked an average of 35 to 45 hours per week during this same period, which persisted through the second trimester (RR 0.4, 95% CI 0.2–0.8) and the seventh month (OR 0.3, 95% CI 0.1–0.7) periods. Regardless of the period of exposure, BMI confounded the relationship between long work hours and preterm delivery, decreasing the risk estimates (indicating a more protective effect) by about 12%.

Table 5
Table 5:
Multivariate Analysis of Occupation Physical Exertion and Preterm and Small-for-Gestational- Age Birth Among Working Women During Different Periods of Pregnancy

No elevation in SGA birth was observed for women who reported prolonged standing or repeated lifting, regardless of period, and working an extended workweek was not associated with an increase or decrease in SGA birth. A modest elevation in SGA birth was observed for women who worked at night during the first (PR 1.3, 95% CI 0.8–2.2) and second trimesters (PR 1.4, 95% CI 0.9–2.4), but our estimates were imprecise.

Although differences in the distribution of pregnancy outcome and physical exertion were present with respect to some demographic, reproductive history, and lifestyle variables (Tables 3 and 4), with the exception of BMI, none were found to confound the relationship between occupational exertion and pregnancy outcomes. As a result, these variables were not retained in the final models.

In the secondary analyses that were restricted to include only a subset of women who worked longer during pregnancy, no differences from the RR estimates presented in Table 5 were observed. When we considered household income in our analyses, a large proportion (66%; n = 183) of women who worked an extended workweek during first trimester fell into the highest household income category compared with women who worked part-time (29%; n = 220) or full-time (50%; n = 380); however, household income did not confound or modify the relationship between working an extended workweek and preterm or SGA birth (data not shown).


Our study found that standing at least 30 hours per week during pregnancy, regardless of the period of exposure, was not associated with preterm or SGA birth. Risk estimates from prior studies range from 0.9 to 1.2 for women who stand at least 55,9 or at least 8 hours per day.11 Two previous studies11,13 reported risk estimates of 1.4 for SGA birth for women who stood 3–5 hours per day and more than 7 hours per day. Lack of a positive association between standing and preterm delivery during gestational weeks 28 to 31 failed to support our hypothesis that activity has an immediate effect on labor. These findings may reflect a healthy worker effect,23 where women who are more capable of performing physically demanding tasks select themselves into these jobs, while possibly benefiting from this type of work.

No significant increase in the risk of preterm or SGA birth was noted among women in the highest lifting group. These findings are consistent with an earlier prospective cohort study,12 where a modest but nonsignificant increase in preterm delivery (OR 1.3, 95% CI 0.7–2.4) was found among women who lifted 25 or more pounds at least 50 times per week. Similar to this study,12 the number of women who lifted at work during pregnancy decreased as pregnancy progressed, hindering our ability to assess adequately whether repeated lifting is associated with adverse pregnancy outcomes. To our knowledge, only 1 prior study has examined lifting and SGA birth, and no elevated risk was observed for women who lifted at least 10 kg per day.11

Women in the Pregnancy, Infection and Nutrition study who were exposed to night work were at an increased risk of preterm delivery. Because of the small sample size, it is necessary to interpret these findings with caution. We do not have external data to compare the prevalence of night work among pregnant women; however, postrecruitment power calculations indicated that we have 0.65 power (α = 0.05) to detect the differences we reported.24 An elevated risk of SGA birth was also suggested, but again, small sample sizes of both night workers and SGA births included in this analysis resulted in imprecise estimates across periods. A prior study11 reported an increase in preterm delivery (OR 1.5, 95% CI 0.8–2.5) among women who worked evenings and nights only, but no increase in SGA birth was observed. No increase in either birth outcome was observed in this same study among women who worked rotating shifts. Xu et al (1994) reported a 2-fold increase (RR 2.0, 95% CI 1.1–3.4) among women working in the textile industry exposed to rotating shift work,10 although McDonald et al5 observed almost the same level of risk (OR 1.9, P < .01) among services workers (Industrial Sector 5: food and beverage servers, housekeeper, hairdressers, laundry worker, agriculture workers) exposed to shift work.

In contrast to some studies,2,5,11,14 working an extended workweek resulted in a decreased risk of preterm delivery. We found that women who worked an extended workweek were older, had more years of education, and were married. An extended workweek has been associated with higher monetary income22 among U.S. workers, which is reflected in our study findings. Although income did not confound or modify the relationship between long work hours and preterm delivery, perhaps the decreased risk in preterm delivery among these women reflects financial or social benefits of employment that were not captured in the socioeconomic status variables used in our analyses.

The rate of preterm delivery for white women was higher, and the rate for African-American women was lower than national preterm birth rates,25 limiting the generalizability of our study findings. These findings were unlikely due to nonparticipation or inclusion criteria used for these particular analyses. The rate of preterm delivery among African-American (12.0%) and white (10.5%) women who chose not participate was similar compared with those who did,17 and although the rates of preterm delivery (13.7%) and SGA birth (10.3%) were higher among women who did not work, the overall rate among our study sample, and within racial groups, changed minimally when nonworking women were excluded from the analysis. Differences in birth outcome rates may reflect study inclusion criteria that prenatal care had to begin by 29 weeks of gestation. This resulted in excluding high-risk women who did not receive prenatal care or those who received it very late in pregnancy.26

The methods that we used to quantify lifting and standing at work is a limitation of this study. Because we did not initially ask women about the number of hours spent standing or the number of times they lifted per day, we collected additional measures from a subset of interviews to help us develop a range of exposure measures for each descriptor. We chose to do this because the use of just the descriptor information would not have allowed us to factor the number of hours a woman works per week, or the number of weeks worked per period into the overall measure of exposure within each period. We found little variability in the frequency with which women in the substudy indicated that they stood or lifted between different descriptors. Although it would have been optimal to ask all women about the number of hours or times they were exposed, the alternative of using only descriptor information would have most likely resulted in misclassification of exposure. The measure of night work could have been improved by considering the number of nights worked per week and whether rotating shifts were worked. We did not ask women about domestic responsibilities that could require significant physical and emotional demands. We considered changes in exposure to physical exertion at work if women changed jobs or ceased employment. However, we did not ask women who remained in the same job(s) about changes in physical activity within those jobs over the pregnancy period. We also did not inquire about whether their employer provided accommodations that allowed them to reduce the amount of exertion during pregnancy.

This study’s prospective design is a strength in that women were asked about exertion at work while they were still pregnant, minimizing the potential for recall bias and differential misclassification of exposure based on pregnancy outcome. There was a potential for recall bias among women included in the nested case–control study, who were interviewed after delivery; however, only 12 women in the case–control sample had changed jobs from the date of the initial interview and were required to recall information about a new job during weeks 28 to 31. In addition, we were able to examine exposure to physical exertion during different periods of pregnancy as well as changes in exposure among women who changed jobs during pregnancy. We were also able to gathered information about several known and suspected risk factors for both preterm delivery and SGA birth, allowing us to control adequately for possible confounding in the analyses.

Our study findings suggest that women who work at night during pregnancy may be at an elevated risk for preterm birth; however, further studies need to be done to explore whether or how shift work influences uterine activity during pregnancy. Earlier studies have found that rhesus monkeys have a circadian rhythm in their uterine activity during late gestation, with peak contractile activity at night, which gradually evolves into labor and delivery.27–28 In addition, researchers observed that women have a normal pattern of nocturnal surges in uterine activity that continues up to the time of term deliveries;29 however, they observed a disruption in these surges among a small cohort of women whose pregnancies ended in preterm delivery. Other studies suggest that the secretion of melatonin, which plays a role in the regulation of biological circadian rhythms,30 is temporarily suppressed among individuals when working at night.31 Further studies are needed to determine whether uterine activity during pregnancy is influenced by melatonin, and more specifically if changes in melatonin due to shift work adversely affects pregnancy outcomes.


1. Smith K, Downs B, O’Connell M. Maternity leave and employment patterns: 1961-1995, Household Economics Study. Curr Popul Rep, P70–79. Washington, DC: U.S. Census Bureau; 2001.
2. Mamelle N, Laumon B, Lazar P. Prematurity and occupational activity during pregnancy. Am J Epidemiol 1984;119:309–22.
3. Saurel-Cubizolles MJ, Kaminski M. Pregnant women’s working conditions and their changes during pregnancy: a national study in France. Br J Ind Med 1987;44:236–43.
4. Mamelle N, Munoz F. Occupational working conditions and preterm birth: a reliable scoring system. Am J Epidemiol 1987;126:150–2.
5. McDonald AD, McDonald JC, Armstrong B, Cherry NM, Nolin AD, Robert D. Prematurity and work in pregnancy. Br J Ind Med 1988;45:56–62.
6. Teitelman AM, Welch LS, Hellenbrand KG, Bracken MB. Effect of maternal work activity on preterm birth and low birth weight. Am J Epidemiol 1990;131:104–13.
7. Launer LJ, Villar J, Kestler E, de Onis M. The effect of maternal work on fetal growth and duration of pregnancy: a prospective study. Br J Obstet Gynaecol 1990;97:62–70.
8. Saurel-Cubizolles MJ, Subtil D, Kaminski M. Is preterm delivery still related to physical working conditions in pregnancy? J Epidemiol Community Health 1991;45:29–34.
9. Henriksen TB, Hedegaard M, Secher NJ, Wilcox AJ. Standing at work and preterm delivery. Br J Obstet Gynaecol 1995;102:198–206.
10. Xu X, Ding M, Li B, Christiani DC. Association of rotating shiftwork with preterm births and low birth weight among never smoking women textile workers in China. Occup Environ Med 1994;51:470–4.
11. Fortier I, Marcoux S, Brisson J. Maternal work during pregnancy and the risks of delivering a small-for-gestational-age or preterm infant. Scand J Work Environ Health 1995;21:412–8.
12. Ahlborg G Jr, Bodin L, Hogstedt C. Heavy lifting during pregnancy—a hazard to the fetus? A prospective study. Int J Epidemiol 1990;19:90–7.
13. Ceron-Mireles P, Harlow SD, Sanchez-Carrillo CI. The risk of prematurity and small-for-gestational-age birth in Mexico City: the effects of working conditions and antenatal leave. Am J Public Health 1996;86:825–31.
14. Luke B, Mamelle N, Keith L, Munoz F, Minogue J, Papiernik E, et al. The association between occupational factors and preterm birth: a United States nurses’ study. Research Committee of the Association of Women’s Health, Obstetric, and Neonatal Nurses. Am J Obstet Gynecol 1995;173:849–62.
15. Klebanoff MA, Shiono PH, Rhoads GG. Outcomes of pregnancy in a national sample of resident physicians. N Engl J Med 1990;323:1040–5.
16. Zhu JL, Hjollund NH, Olsen J, National Birth Cohort in Denmark. Shift work, duration of pregnancy, and birth weight: the National Birth Cohort in Denmark. Am J Obstet Gynecol 2004;191:285–91.
17. Savitz DA, Dole N, Williams J, Thorp JM, McDonald T, Carter AC, et al. Determinants of participation in an epidemiological study of preterm delivery. Paediatr Perinat Epidemiol 1999;13:114–25.
18. Murphy JF, Dauncey M, Newcombe R, Garcia J, Elbourne D. Employment in pregnancy: prevalence, maternal characteristics, perinatal outcome. Lancet 1984;1:1163–6.
19. Savitz DA, Olshan AF, Gallagher K. Maternal occupation and pregnancy outcome. Epidemiology 1996;7:269–74.
20. Zhang J, Bowes WA Jr. Birth-weight-for-gestational-age patterns by race, sex, and parity in the United States population. Obstet Gynecol 1995;86:200–8.
21. Nardone TJ. Part-time workers: who are they? Mon Labor Rev 1986;109(2):13–9.
22. Hecker D. How hours of work affect occupational earnings. Mon Labor Rev 1998;121(10):8–18.
23. Wilcosky T, Wing S. The healthy worker effect: selection of workers and work forces. Scand J Work Environ Health 1987;13:70–2.
24. Dupont WD, Plummer WD Jr. PS power and sample size program available for free on the Internet. Control Clin Trials 1997;18: 274.
25. Martin JA, Hamilton BE, Ventura SJ, Menacker F, Park MM. Births: final data for 2000. National vital statistics report. Vol. 50 (5). Hyattsville, MD: National Center for Health Statistics; 2002.
26. Savitz DA, Dole N, Kaczor D, Herring AH, Siega-Riz AM, Kaufman J, et al. Probability samples of area births versus clinic populations for reproductive epidemiology studies. Paediatr Perinat Epidemiol 2005;19:315-22.
27. Harbert Gm Jr, Spisso K. Biorhythms of primate uterus (Macaca mulatta) during labor and delivery. Am J Obstet Gynecol 1980;138:686–96.
28. Taylor NF, Martin MC, Nathanielsz PW, Seron-Ferre M. The fetus determines circadian oscillation of myometrial electromyographic activity in the pregnant rhesus monkey. Am J Obstet Gynecol 1983;146:557–67.
29. Germain AM, Valenzuela GJ, Ivankovic M, Ducsay CA, Gabella C, Seron-Ferre M. Relationship of circadian rhythms of uterine activity with term and preterm delivery. Am J Obstet Gynecol 1993;168:1271–7.
30. Brzezinski A. Melatonin in humans. N Engl J Med 1997;336 (3):186–95.
31. Yamauchi H, Iwamoto M, Harada N. Physiological effects of shift work on hospital nurses. J Hum Ergol (Tokyo) 2001 Dec; 30:251–4.
© 2005 by The American College of Obstetricians and Gynecologists.