The aim of the study was to examine the correlates of exercise among obese women in early and late pregnancy. Although relevant sociodemographic, health behaviour variables and pregnancy symptoms were associated with exercise status in this study, the only obstetric variable to show an association with exercise status was previous miscarriages. Women who had a history of miscarriage were more likely to be Exercisers very early in pregnancy. This is likely to reflect the fact that many pregnant women who have a history of pregnancy loss are anxious about future loss and may subsequently adopt healthier lifestyles in an attempt to prevent future miscarriages [24, 25].
Sociodemographic variables, including the number of children living at home and education were associated with exercise status in this study. We found that women with at least one child living at home were more likely to be classified as Exercisers, which is contrary to other evidence showing that women with children at home are less likely to be physically active [6, 9, 10]. It is possible that the women in our study with children were also not currently working, or had children who were older and in school. This may have allowed the women to have more time to exercise accounting for our findings. We found that education was a predictor of exercise status in late pregnancy. Consistent with previous findings [6, 8, 9, 26], women classified as Exercisers were more likely to have completed tertiary education than Non-exercisers. Our findings are not unexpected considering that the link between obesity and low socio-economic status has been previously established . Social disadvantage is associated with a range of poor health behaviours, and thus obese pregnant women who come from disadvantaged backgrounds may benefit from greater intervention. Pre-conception counselling may be particularly important for this group of women. However, low socio-economic status is generally associated with poor access to health services and this, in combination with the low rates of planned pregnancies , suggests that community-based interventions for these women may need to be considered as a feasible alternative. The fact that women assigned to the intervention group in the RCT were more likely to be classified as Exercisers in late pregnancy suggests the importance of providing support for obese pregnant women to facilitate long-term health behaviour change.
Pregnancy symptoms were also associated with exercise status during pregnancy. Women who reported lower back pain at 12 and 20 weeks and nausea or vomiting at 20 and 28 weeks were less likely to be classified as Exercisers. These findings are consistent with other evidence suggesting that exercise in early pregnancy is related to decreased reporting of nausea and vomiting in late pregnancy . However, the direction of the relationship between these pregnancy symptoms and exercise in our study was not clear. The findings may suggest that exercise helps alleviate nausea and vomiting in pregnancy. Alternatively, women may have chosen to exercise because they experienced less nausea or vomiting during their pregnancy. In this study we found physical health differences between the Exercisers and Non-Exercisers. Specifically, pre-pregnancy weight in the Exercisers was lower at 12 weeks gestation than in the Non-exercisers. The differences between the groups for back pain found in early pregnancy also support this, and are consistent with evidence suggesting that pre-pregnancy physical activity is associated with a reduced risk of back pain during pregnancy . Further work is required to determine if this is a causal relationship. In our study, it is unclear whether women who have less back pain are more likely to exercise, or whether women who do exercise benefit from a reduction in back pain.
A major issue with lifestyle interventions is the assessment of physical activity. Both subjective and objective measures of physical activity have well known limitations (27). The PPAQ was very useful measure in our study - easy to complete and tailored to measure physical activity among women during pregnancy. The self-report nature of the questionnaire meant that we relied on the women to accurately recall their activity, and this may have led to an overestimation of exercise hours. Thus, our results may not generalise to other studies, particularly those relying on objective measures of exercise.
In the RCT, a number of women withdrew from the trial at different stages, limiting the data that was available for analysis over the course of the trial. A total of five women (n = 3, control; n = 2, intervention) dropped out soon after their baseline visit. Three women withdrew from the trial when they discovered they had been randomised to the control group and were disappointed with this outcome. Two women who were randomised to the exercise arm also withdrew at 12 weeks when they were diagnosed with gestational diabetes based on their baseline blood tests. Among the remaining women, those that withdrew from the study (n = 4, control; n = 5, intervention) did so because of medical or obstetric complications (e.g. miscarriage, intrauterine fetal death, sacroiliac joint instability, gestational diabetes); five women delivered before 36 weeks and thus did not have data collected at 36 weeks. The retention rates in the RCT and the reasons for non-completion have been published elsewhere .
The small sample size in our study limited our ability to adjust for other variables, including age and pre-pregnancy BMI. The women who remained in the study may represent a highly motivated group, which may limit the extent to which these results generalise. The results of this small pilot study suggest that it may be important to adjust for sociodemographic variables (e.g. age, education), as well as pre-pregnancy BMI in future analyses examining the correlates of exercise in pregnancy.