Previous studies investigating the association between exercise during pregnancy and birth weight of the offspring have inconsistent results.1–13 Some studies indicate that exercise may reduce birth weight or increase the risk of low birth weight,4–8 whereas other findings indicate that exercise during pregnancy may increase birth weight.1–3,7,8 Yet other studies find no association between the two.9–13
Many studies have not taken maternal body mass index (BMI) into consideration, despite the well-known associations between maternal BMI and birth weight14–17 and between BMI and exercise.18,19 Different BMI categories are associated with differences in mean birth weight. These differences may, in addition to genes, be explained by variations in average maternal blood glucose concentrations in women with different BMIs. Such variations may result in differences in glucose transfer across the placenta, which in turn influence fetal growth.20,21 It has been speculated that exercise during pregnancy can modify the flow of nutrients and oxygen to the fetoplacental unit and influence fetal growth and birth weight.22 Further insight is needed regarding the associations among maternal BMI, exercise during pregnancy, and birth weight.
We assume that prepregnancy BMI, to a certain extent, both determines the degree of recreational physical activity during pregnancy and has a direct association with fetal growth. The aim of the present study is therefore to estimate the direct associations between exercise during pregnancy and offspring birth weight and between maternal prepregnancy BMI and birth weight, as well as the indirect association between maternal BMI and birth weight, explained by exercise during pregnancy.
MATERIALS AND METHODS
The data collection was performed as part of the Norwegian Mother and Child Cohort Study,23,24 conducted by the Norwegian Institute of Public Health. The Norwegian Mother and Child Cohort Study is approved by the Regional Committee for Ethics in Medical Research and the Data Inspectorate. In brief, the study is a nationwide pregnancy cohort that, in the period 1999 to 2008, has included more than 100,000 pregnancies, and aims to follow parents and children to learn more about the causes of diseases. The women were recruited through a postal invitation in connection with the routine ultrasound examination offered to all pregnant women in Norway at 17 to 18 weeks of gestation. Informed consent was obtained from each woman before inclusion. The participation rate of Norwegian pregnant women was approximately 44%, and the response rate of women answering the questionnaires included in this study was approximately 93%.
For the present study, we used data from two questionnaires, the first returned around week 17 of pregnancy and the second around week 30. The questionnaires covered a variety of items, including pregnancy-related topics, socioeconomic status, environmental exposures, health aspects, and lifestyle factors. Records in the Medical Birth Registry of Norway25,26 from the present pregnancy were included as part of the data set.
Our current analysis was based on version 3 of the quality-assured data files of the Norwegian Mother and Child Cohort Study, which counted 52,547 pregnancies included during the years 1999-2006. Inclusion criteria for our analyses were pregnancies with a single fetus born live at term or later, and that the women had to have returned both questionnaires from around gestational weeks 17 and 30 and also have information given about all the variables included in the statistical analyses. In accordance to these criteria, we excluded 3,687 women (7%) because of missing information on the variables included in the regression analyses. These 7% included missing information on weight or height for 1,329 women, missing information on birth weight for 18 children, missing information on all of the exercise items in one of the two questionnaires for 1,668 women, and missing information on the other included covariates for 672 women. Furthermore, we excluded 1,093 pregnancies where the reported value for one or several of the included variables was outside the realistic range of values because of clerical errors, 1,822 pregnancies with multiple gestations, 100 pregnancies that ended with a stillborn child, and 2,140 pregnancies with delivery before completing gestational week 37. As a consequence of the exclusions, our study population counted 43,705 pregnancies.
The main outcome variable was birth weight measured in grams, as registered in the Medical Birth Registry of Norway.26 Maternal BMI was calculated as self-reported maternal weight in kilograms around conception divided by maternal height in meters squared. Exercise during pregnancy was defined as participating in any recreational activity during pregnancy. The participants were asked how often they engaged in recreational activities during pregnancy in two different periods of time: before gestational week 17 and between weeks 17 and 30. The following activities were included in the analyses: walking (brisk), running (jogging or orienteering), bicycling, fitness or weight training in training centers, prenatal aerobic classes, aerobic classes (low or high impact), dancing, skiing, ball games, swimming, and other. The participants were not told that they would be asked about exercise again in a subsequent questionnaire. The questions about activities were identical for the two periods, and the exercise level of each activity was defined in the questionnaire as never, one to three times per month, once a week, twice a week, or three or more times per week. In the present analyses, exercise was defined as the sum of frequencies per month of all activities reported around weeks 17 and 30, respectively, and was used as continuous variables.
We estimated the direct association between exercise during pregnancy and birth weight, the direct association between maternal prepregnancy BMI and birth weight, and the indirect association between BMI and birth weight, explained by exercise, using linear regression models. These estimations were performed in three steps: First, we estimated the associations between exercise and birth weight, without including BMI in the model. Second, we estimated the associations between exercise and birth weight when BMI was included as a confounder. Third, we estimated the associations between BMI and birth weight, both as a direct association and as an indirect association mediated through the association with exercise. The model shown in Figure 1 illustrates the basis for the last analyses, which were broken down to three ordinary linear regressions: the direct association between BMI and exercise before gestational week 17, the direct association between BMI and exercise between weeks 17 and 30, and the direct associations between these three exposures and the birth weight of the offspring. For each regression model, we checked assumptions (linearity and constant variance) and looked for points with large influence. Linearity and constant variance were assessed by plotting residuals against predicted values, looking for deviations from linearity and for nonconstant variance. Delta beta values were plotted against identification numbers to look for points with high influence. In the analyses, we adjusted for potential confounders believed to be associated with both birth weight and maternal BMI, namely, maternal height, age, parity, cohabitant status (defined as married/cohabitant or not), education (years of education), smoking during pregnancy (smoker or nonsmoker), and exercise during the 3 months before pregnancy (recall of exercise level defined in a similar way as exercise during pregnancy). We also examined potential interactions between BMI and exercise, between smoking and exercise, and between smoking and BMI. The analyses were conducted in SPSS v14.0 (SPSS Inc., Chicago, IL) and STATA version 9.0 (StataCorp LP, College Station, TX).
The mean prepregnancy BMI among the respondents was 24 kg/m2 (standard deviation 4.3). The median exercise frequency was six times per month (range 0-98) during the first 17 weeks of gestation and four times per month between gestational weeks 17 and 30 (range 0-80). The mean birth weight was 3,677 g (median 3,660 g, range 970-6,320 g). Table 1 shows the distribution of exercise frequency during the first 17 weeks of gestation according to different maternal characteristics and the birth weight of the offspring. Maternal prepregnancy BMI, parity, and smoking were negatively associated with exercise during the first 17 weeks, whereas education and exercise during the last 3 months before pregnancy showed positive associations. Exercising three times per week or more was associated with lower birth weight. Table 2 shows the distribution of mean birth weight according to maternal prepregnancy BMI, exercise frequency during the first 17 weeks of gestation, and exercise during the following weeks until week 30. The mean birth weight of the offspring increased with increasing BMI, but birth weight did not vary much across the different categories of exercise frequency. However, the group of highly active women (exercising 12 times per month or more) differed significantly from the other activity groups (P<.001). The median exercise frequencies per month in both periods of pregnancy tended to decline with increasing BMI (data not shown).
First, we estimated the association between exercise and birth weight, without including maternal prepregnancy BMI, in a linear regression analysis. The results showed that the crude association between exercise during pregnancy and birth weight was a 3.9-g decrease in birth weight per unit increase in exercise (times per month). As an example, the offspring of women exercising 20 times per month (five times per week) would weigh approximately 78 g less than the offspring of inactive women. When both measures of exercise were included in the model, the association between exercise around week 30 and birth weight tended to be higher (2.9 g) than the association between exercise around week 17 and birth weight (1.0 g; Table 3: model 1). Second, when adjusting for maternal age, parity, smoking, education, cohabitant status, and exercise during the last 3 months before pregnancy, the association between exercise and birth weight declined to a 2.9-g decrease in birth weight per unit increase in exercise, comprising 0.9 g around week 17 and 2.0 g around week 30 (P<.001; Table 3: model 2).
However, when including maternal prepregnancy BMI as a confounder in addition to height, the covariates mentioned above, and the interaction between smoking and BMI, the direct association between exercise during pregnancy and birth weight was a 2.1-g decrease in birth weight per unit increase in exercise (times per month), comprising 0.72 g (95% confidence interval [CI] −1.3 to −0.1 g) around week 17 and 1.4 g (95% CI −2.0 to −0.8 g) around week 30 (P<.03; Table 3: model 3). The offspring of women exercising 20 times per month (five times per week) would then weigh only 42 g less than the offspring of inactive women.
The results of the model, used in our final analyses, are reported in Figure 1. In these analyses, exercise during pregnancy was regarded as an intermediate variable between maternal BMI and offspring birth weight. The adjusted indirect association between BMI and birth weight was a 0.3-g increase in birth weight per unit increase in BMI (1 kg/m2). This association was calculated from the adjusted regression estimates shown in Figure 1 (−0.10×−0.72 g+−0.17×−1.40 g). The direct association between BMI and birth weight was a 20.3-g (95% CI 19.2-21.4 g; P<.001) increase in birth weight for a one-unit increase in BMI, and the total association between BMI and birth weight was thus 20.6 g (20.3 g + 0.3 g). As an example, an increase in BMI by five units would result in an increase in birth weight of 103 g. The association between BMI and birth weight, explained by exercise during pregnancy, was only 1.5% of the total association between BMI and birth weight in this analysis.
We found significant associations (P<.001) between birth weight and the variables maternal height, parity, maternal age, education less than 10 years, exercise before pregnancy, and the interaction between smoking and BMI (data not shown).
In addition, we fitted multivariable linear regression models, where we sequentially excluded multiparous women, nulliparous women, and women with hypertensive disorders, preeclampsia, eclampsia, and all kinds of diabetes, and thereafter sequentially included preterm neonates and adjusted for maternal weight change during pregnancy. The linear regressions were also rerun when only addressing aerobic and anaerobic exercise, respectively. None of these adjustments had any significant impact on the main findings (data not shown). Furthermore, the associations were not influenced by including other potential confounding factors, such as alcohol or coffee consumption, or the interaction between exercise and BMI. Stratification on the sex of the offspring showed similar results. All regressions met the assumptions of linearity and constant variance, and no point with high influence was found, so the results were deemed robust.
This population-based study shows that exercise during pregnancy has a minor effect on the birth weight of the offspring. The association between exercise and birth weight is reduced when maternal prepregnancy BMI is taken into account, whereas the direct association between prepregnancy BMI and birth weight is strong.
This is one of few large-scale studies that have analyzed the associations between maternal exercise during pregnancy and birth weight of the offspring. When we do not include BMI in the analyses, our results support findings of other studies with a decrease in birth weight for increasing levels of maternal exercise.4–6 In the present study, exercise in the later period of pregnancy has a somewhat larger association with birth weight than exercise in early pregnancy, consistent with findings by Clapp et al.7 Contradictory to these findings, two other studies suggest that exercise during pregnancy increases birth weight.1,2
Our finding of a minor association between exercise frequency and birth weight diminishes when maternal prepregnancy BMI is included in the analyses. When BMI is included in the models, the association between exercise and birth weight is in agreement with other studies finding no significant association between exercise and birth weight,9–13 supporting that differences in birth weight related to different amounts of exercise may largely be explained by confounding variables, rather than by exercise.27
Most likely, the heterogeneity between studies in the associations found between exercise during pregnancy and birth weight may be due to methodologic weaknesses. Small sample sizes, the lack of including BMI as a covariate, or study groups with limited ranges in anthropometric characteristics may adversely influence the ability to detect effects of maternal BMI on the association between exercise and birth weight. In addition, the heterogeneity of the results between studies may be explained by different measures of exercise. Future research should take exercise intensity and duration into consideration, in addition to maternal prepregnancy BMI, to investigate the potentially upper safe limits of exercise level. Furthermore, future research should seek to increase the understanding of the underlying biologic mechanisms around the fetoplacental unit during maternal exercise and of the biologic effect of maternal prepregnancy BMI on birth weight.28
Our analyses are conducted on a large pregnancy cohort that represents all age groups of fertile women and all socioeconomic groups.23 The study indirectly investigates potential underlying biologic and physiologic mechanisms related to the associations among maternal BMI, exercise, and fetal growth, which most likely are mechanisms that are similar to all women. This assumption about similar exposure-outcome associations among different groups of women, despite significant differences in prevalence estimates between the groups, is supported by Nilsen et al.29 Therefore, a low participation rate, or a skewed selection of women into the study, with respect to, for example, higher socioeconomic status compared with the Norwegian source population of pregnant women, may not bias the association between maternal BMI or exercise and birth weight. The Norwegian Mother and Child Cohort Study and the Medical Birth Registry of Norway provide unique information on all the variables relevant for addressing our research question, and the prospective design of our study reduces the risk of recall bias. The present analyses use data from a wide range of BMI values, birth weights, and exercise frequencies during pregnancy. These strengths make it possible to detect variability in birth weight across the wide ranges of BMI values and frequencies of exercise. Comparisons between the women who responded to only the first questionnaire and the women who responded to both shows that the first group has a higher risk of delivering infants below 2,500 g and a higher proportion of inactive women. Analyses of the women who did not answer the questions about exercise show that they are similar to the inactive women in our sample, with respect to both birth weight and BMI. Thus, our results would most likely not be altered if any of these women were included.
A study that compares the self-reported exercise frequency of 112 women in the Norwegian Mother and Child Cohort Study during the first 17 weeks of gestation with their objectively measured energy expenditure around week 20 shows significant positive associations between these measurements of exercise.30 These associations indicate that the women would be given the same ranking of level of exercise, if their exercise level was estimated from information on frequency, duration, and intensity together. Moreover, the distribution of exercise by the maternal characteristics is as expected (Table 1). A limitation of the present study is the lack of information on intensity and duration of the types of activities included. However, the correlation between the exercise frequency and self-reported frequency of heavy breathing or sweating caused by physical activity (times per week) is moderate (rs=0.45). Furthermore, a previous study suggests that the duration of the exercise is not significantly associated with birth weight.8
In conclusion, the findings of this study show that exercise during pregnancy has a minor impact on birth weight, whereas maternal prepregnancy BMI has a larger influence on birth weight. Thus, we suggest that health care professionals should focus on preventing or treating overweight and obesity in fertile women, which would reduce the risk of giving birth to macrosomic offspring.
As long as exercise during pregnancy is believed to be safe in all aspects for the fetus and the pregnant woman and there are no contraindications, exercise during pregnancy should be recommended for the health of the mothers in accordance with the current guidelines.31
1. Leiferman JA, Evenson KR. The effect of regular leisure physical activity on birth outcomes. Matern Child Health J. 2003;7:59–64.
2. Nieuwenhuijsen MJ, Northstone K, Golding J; ALSPAC Study Team. Swimming and birth weight. Epidemiology 2002;13:725–8.
3. Clapp JF III, Kim H, Burciu B, Lopez B. Beginning regular exercise in early pregnancy: effect on fetoplacental growth. Am J Obstet Gynecol 2000;183:1484–8.
4. Perkins CC, Pivarnik JM, Paneth N, Stein AD. Physical activity and fetal growth during pregnancy. Obstet Gynecol 2007;109:81–7.
5. Magann EF, Evans SF, Weitz B, Newnham J. Antepartum, intrapartum, and neonatal significance of exercise on healthy low-risk pregnant working women. Obstet Gynecol 2002;99:466–72.
6. Alderman BW, Zhao H, Holt VL, Watts DH, Beresford SA. Maternal physical activity in pregnancy and infant size for gestational age. Ann Epidemiol 1998;8:513–9.
7. Clapp JF III, Kim H, Burciu B, Schmidt S, Petry K, Lopez B. Continuing regular exercise during pregnancy: effect of exercise volume on fetoplacental growth. Am J Obstet Gynecol 2002;186:142–7.
8. Campbell MK, Mottola MF. Recreational exercise and occupational activity during pregnancy and birth weight: a case-control study. Am J Obstet Gynecol 2001;184:403–8.
9. Voldner N, Frøslie KF, Bo K, Haakstad L, Hoff C, Godang K, et al. Modifiable determinants of fetal macrosomia: role of lifestyle-related factors. Acta Obstet Gynecol Scand 2008;87:423–9.
10. Duncombe D, Skouteris H, Wertheim EH, Kelly L, Fraser V, Paxton SJ. Vigorous exercise and birth outcomes in a sample of recreational exercisers: a prospective study across pregnancy. Aust N Z J Obstet Gynaecol 2006;46:288–92.
11. Bungum TJ, Peaslee DL, Jackson AW, Perez MA. Exercise during pregnancy and type of delivery in nulliparae. J Obstet Gynecol Neonatal Nurs 2000;29:258–64.
12. Kardel KR, Kase T. Training in pregnant women: effects on fetal development and birth. Am J Obstet Gynecol 1998;178:280–6.
13. Kramer MS, McDonald SW. Aerobic exercise for women during pregnancy. Cochrane Database Syst Rev 2006;19:CD000180.
14. Xue F, Willett WC, Rosner BA, Forman MR, Michels KB. Parental characteristics as predictors of birthweight. Hum Reprod 2008;23:168–77.
15. Dubois L, Girard M. Determinants of birthweight inequalities: population-based study. Pediatr Int 2006;48:470–8.
16. Voigt M, Heineck G, Hesse V. The relationship between maternal characteristics, birth weight and pre-term delivery: evidence from Germany at the end of the 20th century. Econ Hum Biol 2004;2:265–80.
17. Sebire NJ, Jolly M, Harris JP, et al. Maternal obesity and pregnancy outcome: a study of 287 213 pregnancies in London. Int J Obes 2001;25:1175–82.
18. Jeffery RW, Rick AM. Cross-sectional and longitudinal associations between body mass index and marriage-related factors. Obes Res 2002;10:809–15.
19. Sharpe PA, Granner ML, Hutto B, Ainsworth BE, Cook A. Association of body mass index to meeting physical activity recommendations. Am J Health Behav 2004;28:522–30.
20. Scholl TO, Sowers M, Chen X, Lenders C. Maternal glucose concentration influences fetal growth, gestation, and pregnancy complications. Am J Epidemiol 2001;154:514–20.
21. Knopp RH. Hormone-mediated changes in nutrient metabolism in pregnancy: a physiological basis for normal fetal development. Ann N Y Acad Sci 1997;817:251–71.
22. Clapp JF III. The effects of maternal exercise on fetal oxygenation and feto-placental growth. Eur J Obstet Gynecol Reprod Biol 2003;110:S80–5.
23. Magnus P, Irgens LM, Haug K, Nystad W, Skjaerven R, Stoltenberg C; MoBa Study Group. Cohort profile: the Norwegian Mother and Child Cohort Study (MoBa). Int J Epidemiol 2006;35:1146–50.
25. Irgens LM. The Medical Birth Registry of Norway. Epidemiological research and surveillance throughout 30 years. Acta Obstet Gynecol Scand 2000;79:435–9.
27. Magann EF, Evans SF, Newnham JP. Employment, exertion, and pregnancy outcome: assessment by kilocalories expended each day. Am J Obstet Gynecol 1996;175:182–7.
28. Catalano PM, Drago NM, Amini SB. Maternal carbohydrate metabolism and its relationship to fetal growth and body composition. Am J Obstet Gynecol 1995;172:1464–70.
29. Nilsen RM, Vollset SE, Gjessing HK, Skjaerven R, Melve KK, Schreuder P, et al. Self-selection and bias in a large prospective pregnancy cohort in Norway. Paediatr Perinat Epidemiol 2009;23:597–608.
30. Brantsæter 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 [published online ahead of print March 25, 2009]. Scand J Med Sci Sports. doi:10.1111/j.1600-0838.2009.00896.x.
© 2010 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.
31. Davies GA, Wolfe LA, Mottola MF, MacKinnon C; Society of Obstetricians and Gynecologists of Canada, SOGC Clinical Practice Obstetrics Committee. Joint SOGC/CSEP clinical practice guideline: exercise in pregnancy and the postpartum period. Can J Appl Physiol 2003;28:330–41.