To date, several studies document various health benefits of regular exercise performed during pregnancy, such as reduced risk of developing gestational diabetes mellitus, pregnancy-induced hypertension and preeclampsia, urinary incontinence, and reduced postpartum depression (26,36). Consequently, current guidelines for exercise during pregnancy are now proactive and recommend aerobic exercise of moderate intensity on most, if not all, days of the week for women with normal pregnancies (1,39), in addition to strength-conditioning exercise (10,29,35).
However, a question still to be answered is the possibility of whether exercise, especially high levels of exercise in the third trimester of pregnancy, might affect gestational age (GA) negatively and thereby influence the risk of preterm delivery (<37 wk of gestation), which is the leading cause of neonatal morbidity and mortality worldwide (13). Optimally, exercise during pregnancy, independent of the trimester in which exercise is performed, would have no effect on GA and the risk of preterm or postterm delivery (>42 wk of gestation). On the contrary, exercising women may have either shorter or prolonged gestations and hence an increased risk of preterm or postterm delivery. Furthermore, exercise patterns seem to change as pregnancy progresses (34), and this change may also affect GA differently. Therefore, when one studies the potential effect of exercise on GA, it is important to consider the timing of exercise.
Kramer and McDonald (19) concluded in a recently published Cochrane review that increasing exercise during pregnancy for previously sedentary women does not result in a clinically important shortening of gestation. However, these conclusions are based on few and small studies. Previous observational studies have primarily focused on preterm birth as an outcome (12,15,18,25) and have found physical activity (PA) or exercise to be associated with a decreased risk of preterm birth regardless of the study design and definition of PA used. In a large cohort including more than 90,000 pregnant women enrolled in the Danish National Birth Cohort, Madsen et al. (22) reported an increased risk of miscarriage by amount of exercise before week 18. On the basis of the same cohort, Juhl et al. (18) reported a decreased risk of preterm birth. Nonetheless, existing studies have not assessed the possible influence of exercise performed at different time points during pregnancy across the entire distribution of GA including preterm and postterm births in addition to mean GA.
The Norwegian Mother and Child Cohort Study (MoBa) is a large prospective study in which data on various health issues and exposures are collected twice during pregnancy via questionnaires (23). Additional information on pregnancy and birth outcomes from the Medical Birth Registry of Norway (MBRN) provides the opportunity to investigate how exercise performed during pregnancy affects GA. The aim of the present study was to examine the association between exercise performed at different time points during pregnancy and GA at birth.
MATERIALS AND METHODS
This study is based on the MoBa study conducted by the Norwegian Institute of Public Health (23). MoBa is a population-based prospective study that has included more than 100,000 pregnancies between 1999 and 2008. The study’s primary aim is to identifxy environmental and genetic factors or interactions of these for prevention of diseases in pregnancy and childhood and is described elsewhere (23). The fourth version of the quality-ensured data file released for research in January 2009 provides data for the present study.
The majority of all pregnant women in Norway were invited to participate in MoBa, and the participation rate is around 44%. In pregnancy week 30, the follow-up rate was 93.6%. Two weeks before the routine ultrasound examination offered to all pregnant women in Norway in gestational weeks 17–18, women are recruited through a mailed invitation. Participants complete two questionnaires during weeks 15–17 (Q1) and 30 of gestation (Q3). Q1 includes items of maternal health, demographics, lifestyle behaviors, and medical history. Q3 focuses mainly on health outcomes during pregnancy and contains some of the same questions from Q1 for follow-up. All questionnaires are available at www.fhi.no/morogbarn.
We first included all singleton pregnancies enrolled between 2000 and 2006 for which questionnaire 1 (Q1) was obtained (n = 63,681). Pregnancies with missing information on both GA from the MBRN (n = 230, 0.4%) and exercise in week 17 (n = 2153, 3.4%) were excluded. We then omitted pregnancies ended before 22 completed weeks (n = 107) and after 44 completed weeks (n = 18). Also, those with implausible birth weight–by-GA combinations (i.e., z-score for GA >4 or <−4) (n = 70) and weekly exercise frequencies above 25 (n = 5) were excluded. Thus, the sample size comprises 61,098 pregnancies. In pregnancy week 30, 93.1% (56,853) of the pregnancies had information on exercise level. Through the personal identification number, the record from the MBRN was linked to the MoBa data set. Since 1967, all live and stillbirths from 16 wk of gestation in Norway have been compulsorily registered in the MBRN (17).
Informed consent was obtained from each participant before inclusion. The study has received approval from the Regional Committees for Medical Research Ethics (S-95113) and the Norwegian Social Science Data Services (01/4325-6).
Main outcome
GA was determined on the basis of the predicted date of delivery according to ultrasound or the date of the last menstrual period in cases where ultrasound data were missing (21), as registered in the MBRN. Delivery (both live and stillbirth) was defined as a terminated pregnancy after 22 completed weeks. Preliminary analysis of the birth weight distribution by each gestational week showed extremely low GA for normal birth weights (e.g., 18 completed weeks and 3320 g) for some cases, whereas other records contained high GA for low birth weights (e.g., 47 completed weeks and 2150 g). Furthermore, to exclude extreme outliers and potential errors in GA, we screened for implausible birth weight–by-GA combinations by sex (38).
Main exposure variable
The main exposure was exercise before and during pregnancy weeks 17 and 30, defined in terms of frequency. In both questionnaires Q1 and Q3, the participants were asked how often they performed the following exercises: strolling, brisk walking, running (jogging or orienteering), bicycling, fitness training in training centers, swimming, aerobic classes (low or high impact), prenatal aerobic classes, dancing (swing, rock, folk dance), skiing (cross-country skiing), ball games, horseback riding, and other exercises. For all exercises, the frequency of exercise was categorized as “never,” “one to three times per month,” “once a week,” “twice a week,” and “three or more times a week.” To capture the highly active women, we divided the responses in the latter category into two exclusive categories by summing up the number of exercises performed per week: “three to five times a week” (one or two exercises three times a week or more often) and “six or more times a week” (three or more exercises at least three times a week). We also merged “once a week” and “twice a week” into “one to two times a week.” “Nonexercisers” were defined as those who responded “never” to all exercises. We then grouped exercises on the basis of type: nonexercisers (strolling and never), brisk walking, non–weight bearing (cycling and swimming), low-impact exercises (prenatal aerobics, low-impact aerobics, fitness training, dancing, cross-country skiing), high-impact exercises (HIE) (running, jogging, orienteering, ball games), and horseback riding (horseback riding and other). A mixed-exercise group included those who did not have a single dominant exercise mode (e.g., one session of jogging and one session of swimming per week). According to the definition of exercise by Caspersen et al. (6), strolling was categorized as nonexercise. Exercising before pregnancy was collected retrospectively in pregnancy week 17 (Q1). The respondents were asked to recall the type and frequency of exercises performed during the last 3 months before the present pregnancy. The questions used on recreational exercise have shown positive correlations with motion sensor measurements (ActiReg®, PreMed AS, Oslo, Norway) in a subsample within the MoBa study (5). The questions are available in English at the journal’s Website (Table, Supplemental Digital Content 1, https://links.lww.com/MSS/A139, Questions used to assess recreational exercise in MoBa).
Covariates
We assessed the following covariates from the MoBa questionnaires (Q1 and Q3): prepregnancy body mass index (BMI) (Q1), educational level (Q1), marital status (Q1), smoking habits (Q1), working hours (i.e., shift work, permanent or nonpermanent work) (Q1), predominantly standing or walking at work (Q1), physical exertion at work (Q1), high blood pressure (Q1), pregnancy-induced hypertension (Q3), time to pregnancy (Q1), vaginal bleedings in pregnancy (Q1 and Q3), and uterine contractions (Q3). In addition, maternal age, parity, spontaneous abortions, assisted reproduction (present pregnancy), cesarean section, and preeclampsia were obtained from the MBRN.
Statistical analysis
Data were analyzed using the PASW Statistics version 18.0 for Windows (SPSS, Inc., Chicago, IL). The distribution of maternal characteristics by exercise level at enrollment was examined by cross-tabulation (Table, Supplemental Digital Content 2, Maternal characteristics by exercise level in week 17; http//links.lww.com/MSS/A140). To compare mean GA by exercise levels during pregnancy, a one-way ANOVA was conducted separately for pregnancy weeks 17 and 30. For multiple comparisons, the Bonferroni test was used to determine which means differed significantly from each other (Tables 1 and 3). We estimated the adjusted association between exercise (frequency and types) and mean GA at birth using a general linear model for pregnancy weeks 17 and 30 (Tables 1 and 3, respectively). The following covariates were added into the final models on the basis of a review of previous studies and directed acyclic graphs (37): maternal age, educational level, prepregnancy BMI, smoking, and parity. To estimate the risks of preterm and postterm births by exercise levels, we used logistic regression analysis (Table 2). First, we adjusted for maternal age, educational level, prepregnancy BMI, smoking, and parity. Then, we included variables such as working hours (unemployed, evening/night, day, nonpermanent, rotating shifts, night shifts), spontaneous abortions (both dichotomized and with cutoff at or above two previous spontaneous abortions), vaginal bleedings (before or after week 20), assisted reproduction (present pregnancy), and high blood pressure. In week 30, predominantly standing/walking at work and vaginal bleedings after week 20 were added as covariates.
;)
TABLE 1: Binary and multivariate associations between exercise during pregnancy and GA at birth (UL-based GA).
TABLE 2: Crude and adjusted risk of preterm birth by level of exercise during pregnancy.
TABLE 3: Binary and multivariate associations between different types of exercise performed during pregnancy and GA at birth (UL-based GA).
To fully understand the effect of exercise at different time points during pregnancy, a model was then fitted by adding interaction terms combining all values of exercise in gestational weeks 17 and 30. We also combined all values of exercise 3 months before pregnancy with exercise in pregnancy week 17 to investigate if the association between exercise in week 17 and GA was independent of prepregnancy exercise level.
In an additional set of analyses, we excluded pregnancies complicated by preeclampsia, pregnancy-induced hypertension, persistent vaginal bleedings, at least two previous spontaneous abortions, assisted reproduction (present pregnancy), and those terminated by a cesarean section (n = 17,572). These analyses were performed to adjust for confounding by indication (33). The effect of such an exclusion will be strong for complications with a high recurrence risk and a high risk of preterm delivery, i.e., complications that are strongly associated with both the exposure (exercise) and the outcome (GA at delivery). Preeclampsia, pregnancy-induced hypertension, persistent vaginal bleedings, and having at least two previous spontaneous abortions are all contraindications for participating in regular exercise during pregnancy (1).
RESULTS
Mean ± SD GA at birth was 39.45 ± 1.94 completed weeks. Among the 61,098 pregnancies in this cohort, 5.2% (n = 3181) ended before 37 completed weeks and 7.9% (n = 4842) ended at or beyond gestational week 42, indicating a skewed distribution. The median was 40 wk of gestation ranging from 22 to 44 completed weeks. Thirteen percent (n = 7578) of the pregnant women with information from both questionnaires did not participate in any kind of exercise during pregnancy weeks 17 and 30, whereas 12.6% (n = 7168) were exercising regularly at least three times a week at both weeks 17 and 30.
The distribution of maternal characteristics by exercise level showed that women exercising at least six times a week at enrollment were more likely to be nonsmokers and nulliparous and had a higher educational level compared with those who were less physically active (Table, Supplemental Digital Content 2, https://links.lww.com/MSS/A140, Maternal characteristics by exercise level in week 17). They also reported a history of fewer spontaneous abortions. Among women exercising less than once a week, 37.2% were overweight or obese according to their prepregnancy BMI (≥25 kg·m−2), compared with 20.5% of those exercising at least six times a week. Higher proportions of women reporting predominantly standing or walking at work, working evening/nights, and with nonpermanent work were observed among those exercising at least six times per week. There were no differences across levels of exercise in the reporting of maternal age or vaginal bleedings after week 20 (Table, Supplemental Digital Content 2, https://links.lww.com/MSS/A140, Maternal characteristics by exercise level in week 17).
Women never exercising in week 17 had a significantly shorter mean GA (39.34 wk, P < 0.0001) compared with women exercising one to three times per month (39.45 wk), one to two times per week (39.48 wk), and three to five times per week (39.51 wk) (Table 1, model 1). In contrast, the mean GA for women who exercised at least six times a week did not differ from that for the nonexercisers. In the adjusted model, mean differences in GA remained for all categories of exercise with the greatest mean difference between women exercising three to five times per week and the nonexercisers (equals 1 d) (Table 1, model 1). When we excluded pregnancies with obstetrical or medical complications (n = 17,572), mean differences in GA were slightly reduced (Table 1).
Table 2 shows the crude and adjusted risk of preterm birth by exercise level. Women exercising one to two or three to five times per week in week 17 or 30 were associated with a reduced risk of preterm delivery, although the confidence intervals (CI) for all exercise categories overlapped. Neither adjusting for maternal age, prepregnancy BMI, parity, education, and smoking nor adjusting for reproductive history and work-related factors (Table 2) attenuated these associations. In addition, women exercising one to two or three to five times a week in week 15 had a slightly increased risk of postterm birth (aOR = 1.14, 95% CI = 1.04–1.24 and aOR = 1.15, 95% CI = 1.04–1.26, respectively). In week 30, exercising one to two times per week was also associated with an increased risk of postterm birth (aOR = 1.11, 95% CI = 1.02–1.20) (data not shown).
Combining prepregnancy and pregnancy exercise levels in week 17 showed that 1.4% of the women started to exercise in gestational week 17 (n = 857), whereas 13.4% previously exercising women stopped before reaching week 17 (n = 8215). Compared with women who did not exercise before pregnancy or during week 17 (n = 5185), women who increased their exercise frequency from one to three times per month before pregnancy to three to five times per week in week 17 (n = 252) had, on average, gestations longer by 2.45 d (P < 0.0001) after adjusting for maternal age, prepregnancy BMI, education, smoking, and parity. Smaller mean differences in GA, both crude and adjusted, were revealed for other combinations of exercise. Women who started exercising in week 17 and reported at least six sessions per week (n = 14) had an increased risk of preterm birth, but adjusting for prepregnancy BMI, maternal age, parity, and education diluted the association (data not shown).
We also compared maternal and social characteristics and obstetric history for women with and without information on exercise at enrollment in week 17 (data not shown). Among women who did not answer the questions on exercise in week 17 (n = 2153), it was more common to be single (P < 0.001), to be multiparous (P < 0.001), to be a smoker (P < 0.001), to be without a permanent job (P < 0.001), to work evening/night shifts (P < 0.001), to have primary school education only (P < 0.001), and to have a prepregnancy BMI between 30 and 34 kg·m−2 (P < 0.001). The mean GA for these pregnancies was slightly shorter (39.37 completed weeks) compared with the study population (P = 0.09), but the proportions of pre- and postterm births were equal. Hence, selected characteristics in these women were similar to those observed in the nonexercisers. We then repeated the regression analysis including the nonresponders in the nonexercise group. However, the mean differences in GA by level of exercise did not change (data not shown).
In pregnancy week 30, we observed mean differences in the range of 0.42–1.05 d between exercisers and nonexercisers (Table 1, model 2), with the smallest mean difference for exercising at least six times per week. Adjusting for confounding factors did not change the mean differences of GA (Table 1, model 2). In pregnancies without obstetrical or medical complications (n = 40,424), small mean differences in GA, equal to half a day, were observed between exercisers and nonexercisers (Table 1).
Twenty percent (11,703 of 56,853) of the women stopped exercising after pregnancy week 17, whereas 8% (n = 4648) exercised only in week 30. Women exercising three to five times a week in both pregnancy weeks 17 and 30 (n = 4816) had, on average, gestations longer by 1.61 d (0.23 wk) compared with women exercising neither in pregnancy week 17 nor in pregnancy week 30 (P < 0.001). Adjusting for possible confounding factors did not change the estimates substantially. Smaller mean differences in GA were observed for other combinations of exercise in weeks 17 and 30 (data not shown).
Table 3 displays the associations between different types of exercise performed during pregnancy and GA at birth. In week 17, women participating in HIE had, on average, gestations longer by 1.33 d compared with the nonexercisers (P < 0.0001) but did not differ from the other exercisers (Table 3). The adjusted model did not change the estimates substantially. Likewise, in week 30, women exercising had, on average, gestations longer by 0.77–1.19 d compared with the nonexercisers. Mean GA did not differ between the different types of exercise, and adjusting for maternal age, prepregnancy BMI, education, smoking, and parity did not influence the associations.
Given that records from the MBRN include both ultrasound-based and menstrual-based dating of GA, we repeated the analysis using the last menstrual period (LMP) method as well. Using the LMP method, the GA distribution shifted slightly to the right (mean GA changed from 39.45 to 39.71 completed weeks), and it yielded a higher number of both preterm (5.2% vs 4.7%) and postterm deliveries (13% vs 7.9%) compared with ultrasound dating (UL)–based GA. In addition, more pregnancies were excluded according to the selection criteria using the LMP method, and 2476 pregnancies did not have their LMP recorded in the MBRN. Nonetheless, the effect estimates of exercise did not differ substantially between the two methods.
DISCUSSION
This study indicates that mean GA among women exercising during pregnancy was longer compared with nonexercising women, but the difference equals 1 d and must be considered of very limited clinical importance. The protective effect of exercise on preterm and the slightly increased risk of postterm birth add to the same conclusion, namely, that engaging in regular exercise during pregnancy shifts the GA distribution slightly upward resulting in a moderately reduced risk of preterm births and a few more postterm births. This finding is consistent with other smaller observational studies that have assessed mean GA (4,12,20,41) as well as two randomized controlled trials (2,24). A reduced risk of preterm birth has also been reported by others (3,12,14,15,18,25,31), whereas few have assessed postterm birth in relation to exercise (12,14,20). In contrast to these studies, we observed an increased likelihood of postterm birth for exercising women. What this study adds is that we estimated the possible influence of exercise frequencies and types performed at different time points during pregnancy across the entire GA distribution, including both preterm and postterm births.
The strengths of our study are that it is population based and includes a large number of pregnancies, with a prospective data collection and a high follow-up rate. Exercise is assessed twice during pregnancy and includes information on both the frequency and the type of exercise performed. We also have retrospective information on exercise level during the last 3 months before pregnancy. In addition, linkage to the MBRN, from where the outcome variable was obtained, and two questionnaires in pregnancy provide information on possible confounding variables. We attempted to control for identifiable confounders such as differences in maternal demographics, obstetric and medical history, and lifestyle factors. However, regardless of these attempts to address confounding, we cannot ignore possible bias from unmeasured confounding factors such as history of preterm birth.
In this study, GA was estimated on the basis of UL. All methods of GA assessment have strengths and weaknesses, and the primary limitation of this method is that GA estimates of symmetrically large or small fetuses will be biased. Furthermore, ultrasound references were developed using pregnancies that were dated according to reliable LMP dates. Hence, UL-based dating is potentially biased in the same direction as dates estimated on the basis of LMP (21). Other studies on maternal exercise and gestational length have often used a combination of both UL- and LMP-based GA, and there is no consensus on which method to use. Finally, a higher incidence of menstrual irregularities such as secondary amenorrhea and shortened luteal phases has frequently been reported among exercising women (11). Although menstrual irregularities are not caused by exercise alone, it does influence the regularity of the menstrual cycle (40) and most likely the LMP-based GA. Hence, we decided to estimate GA on the basis of UL.
There are several methodological challenges when studying how exercising at different time points during pregnancy may affect GA. First, the measurement of exercise and PA needs to be accurate to minimize the possibility that an effect will not be detected because of measurement error. This is crucial when estimating the association between exercise and GA because the association is likely to be modest, as for other birth outcomes (8). The self-reported assessment of exercise and the fact that we did not assess all four dimensions (i.e., type, frequency, duration, and intensity) or domains (i.e., exercise, transportation, occupation, gardening, and care giving) of exercise may have influenced the results and made us unable to estimate a dose–response relationship if there is one. We defined strolling as nonexercise, and the true differences in GA between exercising and nonexercising women may therefore be larger than the difference reported here. Adjusting for working hours, physical exertion at work, or predominantly standing or walking at work did not, however, influence the GA distribution by exercise level in our study. This is in contrast to other observational studies reporting an increased risk of preterm delivery or miscarriage among shift workers and women doing physically demanding work (9,16,32,42). However, our results are in line with Zhu et al. (43), who did not find any significant difference in GA between any type of shift work and daytime work in the Danish National Birth Cohort. This may be explained by the high levels of education in the Norwegian and Danish population and that shift work is common among highly educated women (e.g., medical doctors working a shift at hospitals). In MoBa, a high proportion of shift workers were exercising during pregnancy. In contrast to other observational studies on the association between recreational exercise and GA, the questions used to assess exercise in our study have been validated against a position motion sensor (ActiReg®) in a subsample within the MoBa study (5). Significant positive associations between self-reported exercise and the motion sensor were observed, indicating their usefulness as an assessment of recreational exercise in MoBa. Besides, the present study used information from a prospective cohort of more than 60,000 pregnancies, which produced small statistically significant estimated effects. Likewise, the fact that the adjusted distribution of GA did not differ substantially from the crude distribution further strengthens our results.
Second, pregnant women who previously experienced persistent vaginal bleedings and developed pregnancy-induced hypertension or preeclampsia or who had more than two spontaneous abortions may choose not to exercise during this pregnancy or may be advised not to do so by their midwife or general practitioner. We assumed that confounding by the indication for not exercising could have influenced our results. Hence, we restricted the analysis to a subsample of normal pregnancies showing that exercise in weeks 17 and/or 30 did not influence the GA distribution.
Third, a low response rate makes it difficult to generalize the results from this study to the target population. Women participating in the MoBa differ from the target population in relation to premature delivery rates, and the proportion of newborns with low birth weight is lower among MoBa participants compared with the target population in the MBRN (23). Given that MoBa participants also have a slightly higher educational level, we may assume that they are more physically active compared with the target population. Consequently, the differences in GA between exercisers and nonexercisers may be larger than reported in the present study. Furthermore, women who were excluded from the study population because of missing information on exercise at enrollment also differed from the study population regarding educational level, smoking, marital status, prepregnancy BMI, parity, shift work, and exertion at work, although mean GA was not significantly shorter compared with women who had answered the questions on recreational exercise, and the proportions of both preterm and postterm deliveries were equal. Including these pregnancies in the analysis, assuming that they were equal to the nonexercising group, did not change the estimates. Nevertheless, we do not believe that it is likely that selection into the study is caused by exercise, and the low response rate will therefore have little or no influence on the associations estimated in our study (27).
Physically active women who become pregnant are more likely to continue their exercise routines during pregnancy (28). Furthermore, women who exercise in late pregnancy are most likely to be different from women not exercising and women quitting exercise in the latter part of pregnancy. If pregnancy is normal without complications, women are also more likely to exercise during pregnancy until childbirth (30). Although the mechanisms that initiate spontaneous delivery are far from understood (7), it seems that factors other than recreational exercise affect mean GA. Despite the limitations of our study, we believe that exercise, whether performed in the second and/or third trimester of pregnancy, shifts the distribution of GA slightly upward resulting in a reduced proportion of preterm births and a slightly increased proportion of postterm births.
The authors thank all the participating families in Norway who take part in this ongoing cohort study and acknowledge the services of the MBRN.
The MoBa is supported by the Norwegian Ministry of Health and the Ministry of Education and Research, National Institutes of Health/National Institute of Environmental Health Sciences (contract no. NO-ES-75558), National Institutes of Health/National Institute of Neurological Disorders and Stroke (grant 1 UO1 NS 047537-01), and the Norwegian Research Council/FUGE (grant 151918/S10).
The authors have no conflict of interest to declare.
The results of the present study do not constitute endorsement by the American College of Sports Medicine.
REFERENCES
1. American College of Obstetricians and Gynecologists Committee on Obstetric Practice. ACOG Committee opinion. Number 267, January 2002: exercise during pregnancy and the postpartum period. Obstet Gynecol. 2002; 99: 171–3.
2. Barakat R, Stirling JR, Lucia A. Does exercise training during pregnancy affect gestational age? A randomised controlled trial. Br J Sports Med. 2008; 42: 674–8.
3. Berkowitz GS, Kelsey JL, Holford TR, Berkowitz RL.
Physical activity and the risk of spontaneous preterm delivery. J Reprod Med. 1983; 28: 581–8.
4. Both MI, Overvest MA, Wildhagen MF, Golding J, Wildschut HI. The association of daily
physical activity and
birth outcome: a population-based cohort study. Eur J Epidemiol. 2010; 25: 421–9.
5. Brantsaeter AL, Owe KM, Haugen M, Alexander J, Meltzer HM, Longnecker MP. Validation of self-reported recreational exercise in pregnant women in the Norwegian Mother and Child Cohort Study. Scand J Med Sci Sports. 2010; 20: e48–e55.
6. Caspersen CJ, Powell K, Christenson G.
Physical activity, exercise and physical fitness: definitions and distinctions for health-related research. Public Health Rep. 1985; 100: 126–31.
7. Challis JRG. Mechanism of parturition and preterm labor. Obstet Gynecol Surv. 2000; 55: 650–60.
8. Chasan-Taber L, Evenson KR, Sternfeld B, Kengeri S. Assessment of recreational
physical activity during pregnancy in epidemiologic studies of birthweight and length of
gestation: methodologic aspects. Women Health. 2007; 45: 85–107.
9. Croteau A, Marcoux S, Brisson C. Work activity in pregnancy, preventive measures, and the risk of preterm delivery. Am J Epidemiol. 2007; 166: 951–65.
10. Davies GA, Wolfe LA, Mottola MF, MacKinnon C. Joint SOGC/CSEP clinical practice guideline: exercise in pregnancy and the postpartum period. Can J Appl Physiol. 2003; 28: 330–41.
11. De Souza MJ, Toombs RJ, Scheid JL, O’Donnell E, West SL, Williams NI. High prevalence of subtle and severe menstrual disturbances in exercising women: confirmation using daily hormone measures. Hum Reprod. 2010; 25: 491–503.
12. Evenson KR, Siega-Riz AM, Savitz DA, Leiferman JA, Thorp JM Jr. Vigorous leisure activity and
pregnancy outcome. Epidemiology. 2002; 13: 653–9.
13. Goldenberg RL, Culhane JF, Iams JD, Romero R. Epidemiology and causes of preterm
birth. Lancet. 2008; 371: 75–84.
14. Hatch M, Levin B, Shu XO, Susser M. Maternal leisure-time exercise and timely delivery. Am J Public Health. 1998; 88: 1528–33.
15. Hegaard HK, Hedegaard M, Damm P, Ottesen B, Petersson K, Henriksen TB. Leisure time
physical activity is associated with a reduced risk of preterm delivery. Am J Obstet Gynecol. 2008; 198: 180.e1–5.
16. Henriksen TB, Hedegaard M, Secher NJ, Wilcox AJ. Standing at work and preterm delivery. Br J Obstet Gynaecol. 1995; 102: 198–206.
17. Irgens LM. The Medical
Birth Registry of Norway. Epidemiological research and surveillance throughout 30 years. Acta Obstet Gynecol Scand. 2000; 79: 435–9.
18. Juhl M, Andersen PK, Olsen J, et al.. Physical exercise during pregnancy and the risk of preterm
birth: a study within the Danish National
Birth Cohort. Am J Epidemiol. 2008; 167: 859–66.
19. Kramer MS, McDonald SW. Aerobic exercise for women during pregnancy [review]. Cochrane Database Syst Rev. 2006; 3: CD000180.
20. Leiferman JA, Evenson KR. The effect of regular leisure
physical activity on
birth outcomes. Matern Child Health J. 2003; 7: 59–64.
21. Lynch CD, Zhang J. The research implications of the selection of a gestational age estimation method. Paediatr Perinat Epidemiol. 2007; 21 (2 suppl): 86–96.
22. Madsen M, Jørgensen T, Jensen M, et al.. Leisure time physical exercise during pregnancy and the risk of miscarriage: a study within the Danish National
Birth Cohort. BJOG. 2007; 114: 1419–26.
23. Magnus P, Irgens LM, Haug K, Nystad W, Skjaerven R, Stoltenberg C. Cohort profile: the Norwegian Mother and Child Cohort Study (MoBa). Int J Epidemiol. 2006; 35: 1146–50.
24. Marquez-Sterling S, Perry AC, Kaplan TA, Halberstein RA, Signorile JF. Physical and psychological changes with vigorous exercise in sedentary primigravidae. Med Sci Sports Exerc. 2000; 32 (1): 58–62.
25. Misra DP, Strobino DM, Stashinko EE, Nagey DA, Nanda J. Effects of
physical activity on preterm
birth. Am J Epidemiol. 1998; 147: 628–35.
26. Morkved S, Bo K, Schei B, Salvesen KA. Pelvic floor muscle training during pregnancy to prevent urinary incontinence: a single-blind randomized controlled trial. Obstet Gynecol. 2003; 101: 313–9.
27. Nilsen RM, Vollset SE, Gjessing HK, et al.. Self-selection and bias in a large prospective pregnancy cohort in Norway. Paediatr Perinat Epidemiol. 2009; 23: 597–608.
28. Ning Y, Williams MA, Dempsey JC, Sorensen TK, Frederick IO, Luthy DA. Correlates of recreational
physical activity in early pregnancy. J Matern Fetal Neonatal Med. 2003; 13: 385–93.
29. Norwegian Directorate for Health and Social Affairs. Report No.:2/2000.
Physical Activity and Health: Guidelines [Fysisk aktivitet og helse: Anbefalinger]. Oslo, Norway. Directorate for Health and Social Affairs. 2000; 86[in Norwegian]; 2000. pp. 1–86. Report No.: 2/2000.
30. Owe KM, Nystad W, Bo K. Correlates of regular exercise during pregnancy: the Norwegian Mother and Child Cohort Study. Scand J Med Sci Sports. 2009; 19: 637–45.
31. Petridou E, Salvanos H, Skalkidou A, Dessypris N, Moustaki M, Trichopoulos D. Are there common triggers of preterm deliveries? BJOG. 2001; 108: 598–604.
32. 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: 1279–88.
33. Rothman KJ. Confounding by indication. In: Rothman KJ. Epidemiology: An Introduction. New York (NY): Oxford University Press; 2002. p. 103.
34. Rousham EK, Clarke PE, Gross H. Significant changes in
physical activity among pregnant women in the UK as assessed by accelerometry and self-reported activity. Eur J Clin Nutr. 2006; 60: 393–400.
35. Royal College of Obstetrics and Gynaecology; Guidelines and audit committee of the Royal College of Obstetricans and Gynaecologists, Dorset, UK. Statement No. 4. Exercise in Pregnancy, 2006; 1–7.
36. Schlussel MM, Souza EB, Reichenheim ME, Kac G.
Physical activity during pregnancy and maternal-child health outcomes: a systematic literature review. Cad Saude Publica. 2008; 24 (4 suppl): s531–44.
37. Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Med Res Methodol [Internet]. 2008 [cited 2011 Jan 20]; 8: 70. Available from:
http://www.biomedcentral.com/1471-2288/8/70.
38. Skjaerven R, Gjessing HK, Bakketeig LS. Birthweight by gestational age in Norway. Acta Obstet Gynecol Scand. 2000; 79: 440–9.
39. US Department of Health and Human Services. 2008
Physical Activity Guidelines for Americans. Department of Health and Human Services; 2008. p. 76.
40. Warren MP, Goodman LR. Exercise-induced endocrine pathologies. J Endocrinol Invest. 2003; 26: 873–8.
41. Watson PE, McDonald BW. Activity levels in pregnant New Zealand women: relationship with socioeconomic factors, well-being, anthropometric measures, and
birth outcome. Appl Physiol Nutr Metab. 2007; 32: 733–42.
42. Wong EY, Ray R, Gao DL, et al..
Physical activity, physical exertion, and miscarriage risk in women textile workers in Shanghai, China. Am J Ind Med. 2010; 53: 497–505.
43. Zhu JL, Hjollund NH, Olsen J. Shift work, duration of pregnancy, and
birth weight: the National
Birth Cohort in Denmark. Am J Obstet Gynecol. 2004; 191: 285–91.
