For women of reproductive age, excessive gestational weight gain and/or postpartum weight retention can increase the risk of obesity. This systematic review evaluates the effectiveness of lifestyle modification control trials that utilize exercise interventions, with or without dietary intervention, on weight loss among postpartum women. A search of randomized clinical trials (RCT) was performed using the follow databases and the bibliography of candidate studies: MEDLINE, Web of Science, EMBASE, CENTRAL/Cochrane and Physiotherapy Evidence Database. English language RCT papers published up to 31 October 2012, which present changes on maternal body weight from baseline to the end of exercise intervention were included. The primary meta-analysis examined the effects of exercise interventions, with or without complementary dietary intervention, on weight loss during the postpartum period compared with usual standard of care. Five subgroup analyses were performed to examine differences in study interventions and exercise modalities: duration of intervention, quality of study methodology, supervision of exercise intervention, exercise intervention goals used and the type of dietary intervention. In total 11 studies met eligibility criteria with 769 participants, 409 under intervention and 360 in the control group. The primary meta-analysis included all 11 studies and found a mean difference (MD) on weight loss of −2.57 kg (95% CI −3.66 to −1.47). The subgroup analysis demonstrated that the most effective interventions in reducing weight in postpartum women were exercise programs with objectively defined goals, such as the use of heart rate monitors or pedometer (MD of −4.09 kg—95% CI −4.94 to −3.25, I2=0%) and exercise combined with intensive dietary intervention (MD of −4.34 kg—95% CI −5.15 to −3.52, I2=0%). Thus, there is benefit from overall lifestyle interventions on weight loss in postpartum women and exercise plus intensive diet and objective targets are the most effective intervention strategies.
The prevalence of overweight and obesity has reached drastic proportions worldwide and has a significant impact on health care costs.1 In 2008, an estimated 1.46 billion adults worldwide had a body mass index (BMI) of 25 kg m−2 or greater, of these 205 million men and 297 million women were obese with a BMI greater than 30 kg/m2. Globally, the average female BMI has increased by 0.5 kg/m2 per decade between 1980 and 2008.2
For women of reproductive age, excessive gestational weight gain and/or postpartum weight retention can significantly increase the risk of obesity.3 The risk of becoming overweight or obese is even greater when excessive gestational weight gain and/or postpartum weight retention occurs in successive pregnancies.3 During pregnancy, obesity and excessive weight gain are associated with adverse maternal and fetal outcomes.4, 5 Furthermore, an increase in BMI of just 3 kg m−2 between subsequent pregnancies increases the risk of pre-eclampsia, gestational hypertension, C-section delivery, stillbirth and delivering a large for gestational age neonate even if a woman has a ‘normal’ BMI for both pregnancies.6
By one year postpartum 14 to 20% of women retain 5 kg or more of weight gained during pregnancy, which elevates the risk of developing health problems, such as diabetes, metabolic syndrome and cardiovascular disease.7 For most women, pregnancy and the postpartum period are generally associated with a reduction in the level of physical activity.8, 9 It is assumed that in combination with poor dietary choices/habits, a lack of physical activity and an increase in sedentary behavior have an important role in the development of overweight and obesity.1 Although, not all weight gain in women at reproductive age can be related to pregnancy, an average weight gain of 0.5 to 3.0 kg can be attributed to postpartum weight retention.10
Previous reviews have addressed multiple aspects of weight loss after birth, such as the role of postpartum weight retention in obesity,10 lactation,11 smoking cessation and weight loss12 and finally lifestyle interventions.13, 14 These previous analyses suggest that physical activity coupled with a calorie limited diet in the postpartum period is an effective intervention to help the mothers to lose weight.14 However, due to the variety of exercise interventions there is a need to identify which intervention strategies and exercise modalities are most effective in order to provide clinical recommendations and avoid unnecessary and/or ineffective interventions in clinical trials.12, 15
In the last years, a number of different intervention strategies have been developed to help women lose the excess weight gained during pregnancy. The primary objective of this systematic review is to evaluate the effectiveness of lifestyle modification control trials that utilize exercise interventions, with or without complementary dietary intervention, on weight loss among postpartum women. Furthermore, we aimed to investigate different intervention strategies, including length of intervention, use of dietary intervention, study goals used and supervision of exercise intervention.
A systematic search of the following databases was conducted in order to identify relevant studies: PubMed, MEDLINE, Web of Science, EMBASE, CENTRAL/Cochrane and Physiotherapy Evidence Database. The MeSH search terms used included: (‘exercise’ or ‘physical activity’) and (‘Postpartum Period’ or ‘puerperium’). The search strategy was designed for the PubMed database and altered as needed for use in other databases. The search terms were combined with specific terms such as: ‘weight loss’, ‘weight retention’, ‘BMI’. The search was limited to papers published in English. All papers published up to 31 October 2012 were considered.
Inclusion and exclusion criteria
The inclusion criteria for studies were defined based on the participants involved, type of intervention, type of comparison group, outcome of interest and study design (PICOS). The study had to include female participants during the postpartum period (P). Randomization and study initiation had to have occurred postpartum and at any point up to and including 18 months after delivery. No restrictions were placed on participant BMI, co-morbidities, parity or breastfeeding status. The intervention (I) had to have provided a program that included supervised physical exercise or physical exercise guidance/counseling with a minimum follow-up of 10 weeks. No restrictions were placed on whether a complementary dietary intervention was utilized. The comparator (C) group had to have been a control group with no intervention, minimal intervention or usual care for the given study setting. The study had to have presented the outcome (O) of mean weight loss from study initiation to study completion. Permissible study designs (S) included randomized clinical trials and pilot randomized clinical trials. Exclusion criteria were: other study designs, diet interventions without a physical activity component, interventions with pelvic floor exercises only or exercise interventions designed to treat musculoskeletal disorders.
The study search and screening were completed independently by two reviewers (SLN and JP). At each step in the screening process studies were assessed based on the inclusion and exclusion criteria. Studies were first excluded based on an assessment of only titles and abstracts. Studies were only excluded at this stage if both reviewers were certain that it did not meet the criteria. The full-text article was then obtained and assessed for all remaining studies. The reference lists of all these remaining studies and related previously completed systematic reviews were searched manually; any relevant studies not previously found were obtained and included in this stage of the screening process.
Study quality assessment
Two independent reviewers (SLN and JP) assessed the risk of bias of the included studies using the ‘The Cochrane risk of bias assessment tool’.16 When there was disagreement in the assessment, a third reviewer (GNS) was consulted and consensus was reached through discussion.
The risks of bias were assessed according to six criteria: random sequence generation, allocation sequence concealment, blinding of participants, blinding of outcomes, incomplete outcome data and selective outcome reporting
Data extracted from each of the studies for the meta-analysis included the number of participants per group, the mean weight change in kilograms from randomization to study completion of the exercise intervention for each group, and the standard deviation for each mean. If a study used more than one relevant intervention or control group the results were combined using the formulae and methods presented in section 220.127.116.11 of the Cochrane Handbook.17 A single intervention and control group was used for each study in the meta-analysis.
The primary meta-analysis combined the results of all the included papers and examined the effect of exercise based interventions on weight loss during the postpartum period compared with usual standard of care. A total of five subgroup analyses were performed with the goal of further examining differences in study interventions and exercise modalities. The first subgroup analysis classified studies based on the length of intervention. Short term interventions were defined as durations of less than 24 weeks and long term interventions were defined as durations of 24 weeks or greater. These categories were chosen based on definitions used in a previous systematic review.14 The second subgroup analysis compared studies based on the quality of study methodology. Studies were classified as poor quality if they failed to meet two or more criteria in the Cochrane risk of bias assessment tool, and as high quality if they failed to meet one or less of the criteria. The third subgroup analysis compared studies based on the level of supervision used in the exercise intervention. Studies were classified as supervised if the participants were monitored by study personnel during one or more bouts of exercise per week. Studies were classified as unsupervised if participants were monitored less frequently during exercise or if they were not supervised at all during exercise. The fourth subgroup analysis compared studies based on the type of goals used in the exercise intervention. For our analysis only goals that were specifically defined, purposefully prescribed to the patients and evaluated during follow-up visits were considered. The first category consisted of studies that used goals based on pedometer and/or heart rate targets. The second category consisted of studies that used goals based on time, such as number of days and duration of exercise, or distance goals, such as walking 5 km per day. The third category consisted of studies that did not specify whether any type of goal was used. The fifth subgroup analysis compared studies based on the type of dietary intervention used. Studies were classified as having an intensive dietary intervention component when the participants were prescribed a calorie restricted diet and adherence to the diet was assessed at follow-up visits. Studies were classified as counseling only when basic information was given to participants (e.g., healthy food choices, avoiding sugar sweetened beverages and high calorie or energy dense snacks etc.), but no specific diet was prescribed and/or adherence to the diet was not assessed at follow-up visits. Studies were classified as exercise only if no dietary information was given and no dietary restrictions were specified.
A random effect model and inverse variance method was used due to the heterogeneity observed in study population and design. For each analysis a Forrest plot was generated and used to assess the relative strength of the intervention effect. For the primary analysis a Funnel plot was used to assess the risk of publication bias. All analyses were performed using Review Manager version 5.2.18
Some of the included studies also evaluated the impact of intervention on different variables related to weight, such as the absolute weight, BMI, waist circumference at baseline and endpoint, weight retention and body composition (fat and lean mass). These outcomes were not included in the meta-analyses as they were not consistently presented as mean change from baseline.
This systematic review was registered with PROSPERO—International register of systematic reviews (http://www.crd.york.ac.uk/prospero/): CRD42012002978. The protocol and analysis adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA Statement) standards (http://www.prisma-statement.org/).19
The flow chart in Figure 1 describes the search history and screening process. The literature search identified 903 unique journal articles. Studies were first excluded based on an assessment of their titles and abstracts and then the full-text of the remaining 36 articles was assessed, and it was determined that 11 studies met all of the inclusion and exclusion criteria to be included in the meta-analysis. The summaries of these 11 studies can be found in Table 1. Overall the meta-analysis included a total of 769 participants, 409 women in the intervention group and 360 women in the control group.20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 The reasons for study exclusion from the meta-analysis are also outlined in Figure 1.31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54
All of the included studies (Table 1) were conducted in developed countries, seven of them in the United States of America and one in each of Canada, Sweden, Australia and the United Kingdom. The inclusion criteria applied in each of the studies varied, with no two studies using the same set of criterion. The majority of the studies included only healthy postpartum women, those who had experienced pregnancy complications or had other comorbidities deemed relevant by the study personnel were ineligible for recruitment.23, 24, 26, 28, 29, 30 However, one trial specifically addressed women who had experienced gestational diabetes in their pregnancy.27 Eight of the studies recruited only overweight (BMI25.0 kg m−2) or obese (BMI30.0 kg m−2) women20, 21, 22, 23, 26, 28, 29, 30, either prepregnancy or at enrollment, whereas three studies placed no restrictions on BMI.24, 25, 27 Five of the studies recruited only women who were nonsmokers.20, 21, 23, 24, 26 Another difference observed among the studies was the breastfeeding status of the participants. Four studies recruited only breastfeeding women,20, 21, 24, 26 four studies recruited women with no restrictions placed on breastfeeding status,23, 27, 28, 29 one study recruited only women who were not breastfeeding25 and two studies did not report information about breastfeeding status.22, 30 The studies also differed in terms of the timing of the intervention. The earliest onset of intervention was four weeks postpartum,21, 26 although the latest allowed recruitment and initiation between 6 and 18 months postpartum.22 However, majority of the studies began the intervention between 4 and 14 weeks postpartum. It is important to note the variety observed in terms of exercise intervention strategies. In four studies participants were supervised by study personnel during at least one exercise session per week.21, 23, 24, 26 In all other studies the exercise intervention was unsupervised.20, 22, 25, 27, 28, 29, 30 Intervention strategies used included heart rate monitors, pedometers, personalized exercise counseling, correspondence programs, text messages and telephone calls. Walking was the most common modality of exercise recommended20, 21, 22, 23, 25, 29 followed by general aerobic exercise.24, 26, 27, 28 Supervised resistance training combined with walking was applied in one unique study.21 One trial did not report which modality of physical exercise was recommended.30
Methodology quality assessment
A summary of the risk of bias assessment can be found in Figure 2. It was determined that four studies failed to describe the random sequence generation methods and seven studies did not clarify the allocation concealment, all of these were classified as unclear risk of bias. None of the studies could blind participants leading to high risk of bias in this domain, which is expected for exercise based interventions. In most of the studies, outcome data were not collected by blinded investigators; however, as weight is an objective measure, all studies were classified as low risk in the blinding of outcome assessment. Attrition bias is common in exercise interventions; we classified studies as having a high risk of bias if they reported different rates of drop out between groups, if they lacked proper methods for controlling for attrition or used inappropriate methods for imputing missing data. All of the trials were scored as low risk of bias for selective reporting of outcomes as all predetermined outcomes were reported. Other sources of bias identified in the studies included the absence of an a priori sample size calculation and insufficient power.
Primary analysis: effect of the intervention on weight loss
The primary meta-analysis included all 11 studies and examined the effect of exercise interventions, with or without dietary intervention, on weight loss during the postpartum period compared with usual standard of care (Figure 3). Overall, a significant effect was found with a mean difference (MD) of −2.57 kg (95% CI −3.66 to −1.47). Substantial heterogeneity was observed with an I2 value of 66% and a significant chi-square test (P-value<0.01). The Funnel plot displayed symmetry, suggesting an absence of publication bias (Figure 4).
Duration of intervention
The duration of intervention ranged from 10 weeks to 52 weeks. The effect observed among the eight studies that utilized short term interventions, ranging from 10 to 16 weeks in duration, was found to be significant with a MD of −2.78 kg (95% CI −4.00 to −1.56, I2=64%).20, 21, 22, 23, 24, 26, 27, 30 The effect observed among the three studies that utilized long term interventions (24, 36 and 52 weeks duration) was not found to be significant with a MD of −1.95 kg (95% CI −4.19 to 0.28, I2=50%).25, 28, 29 A graphical representation of this comparison can be found in Figure 5a.
Quality of study methodology
Based on the risk of bias assessments seven studies were classified as poor quality22, 23, 24, 25, 26, 27, 30 and four studies were classified as high quality.20, 21, 28, 29 The effect observed among the poor quality studies was found to be significant with a MD of −2.48 kg (95% CI −3.79 to −1.17, I2=66%). The effect observed among the high quality studies was also found to be significant with a MD of −2.98 kg (95% CI −5.39 to −0.57, I2=70%). No significant difference in MD was observed between the two subgroups (Figure 5b).
Supervision of exercise intervention
In the four studies where participants were supervised by study personnel during at least one exercise session per week the effect observed was significant with a MD of −4.10 kg (95% CI −5.04 to −3.16, I2=0%).21, 23, 24, 26 Whereas, in the remaining seven unsupervised exercise studies the effect observed was weaker and the confidence interval of the estimate included zero, with a MD of −1.24 kg (95% CI −3.76 to 1.29, I2=91%)20, 22, 25, 27, 28, 29, 30 (Figure 5c).
Exercise intervention goals used
The exercise interventions in six studies used heart rate monitor or pedometer goals.20, 21, 23, 24, 26, 28 The effect observed among these studies was significant with a MD of −4.09 kg (95% CI −4.94 to −3.25, I2=0%). The exercise interventions in four studies used distance or time based goals was weaker, however, the confidence interval of the estimate did not include zero, with a MD of −1.30 kg (95% CI −2.41 to −0.20, I2=37%).22, 25, 27, 29 One study failed to identify any specific exercise intervention goals used and nonsignificant results were observed with an MD of −1.30 kg (95% CI −3.36 to 0.76, I2=Not Applicable).30 No significant difference in MD was observed between the subgroup that used distance or time based goals and the subgroup that did not use specific goals (Figure 5d).
Type of complementary dietary intervention used
An intensive dietary intervention was used in six studies with a significant MD of −4.34 kg (95% CI −5.15 to −3.52, I2=0%).20, 21, 23, 25, 26, 28 A dietary intervention consisting only of counseling was used in three studies.22, 29, 30 The effect observed among these studies was weaker; however, the confidence interval of the estimate did not include zero, with a MD of −1.31 kg (95% CI −2.11 to −0.52, I2=0%). The effect observed among three studies with no dietary intervention was not significant with a MD of −0.79 kg (95% CI −2.54 to 0.97, I2=0%).20, 24, 27 No significant difference in MD was observed between the subgroup that used dietary counseling only and the subgroup that did not use a dietary intervention (Figure 5e).
One study originally consisted of four intervention groups, an exercise only, a diet only, a diet and exercise and a control group.20 For the primary analysis the exercise and diet and exercise groups were combined. For this subgroup analysis the two groups were considered independently, one in the intensive dietary intervention group and one in the exercise only group. For both, the same control group was used.
Based on our meta-analysis results, lifestyle modification randomized control trials that utilize exercise interventions, with or without dietary intervention, showed a significant effect on weight loss among postpartum women.
Interventions that utilized heart rate or pedometer targets for goal setting and whose exercise regimen was coupled with intensive dietary interventions were found to be significantly more effective than those, which used other goal setting methods and dietary interventions. There was a trend towards supervised exercise interventions being more effective than unsupervised interventions, however, significance was not observed. No differences were observed when comparing studies based on duration of intervention and quality of study methodology.
The majority of the studies in this review focused on high-risk groups as defined by a prepregnancy or postpartum overweight/obese BMI or excessive gestational weight gain.20, 21, 22, 23, 25, 26, 28, 29, 30 Just one study included in the review focused on high-risk ethnic minority groups.30 Previous studies that have included low-income and ethnic-minority women, such as African American and Hispanic, have demonstrated that this group may be more adversely affected by pregnancy weight gain and weight retention than more affluent and predominantly Caucasian populations.55 Future studies should aim to develop and evaluate intervention strategies that address the unique needs within these higher risk populations. The effect of weight loss interventions among women who are breastfeeding is a topic of interest. However, it was not possible to perform a subgroup analysis based on breastfeeding status, as some researchers allowed both lactating and nonlactating women to participate in their trial,23, 27, 28, 29 and others did not report information about breastfeeding status.22, 30 Two of the studies included in the review did focus on this issue.24, 26 Dewey (1994)24 found that aerobic exercise performed four to five times per week for 12 weeks had no adverse effect on the volume or composition of the breast milk, infant’s weight gain or maternal prolactin levels. However, the intervention did not have a significant effect on weight loss, potentially because the intervention group was assigned a greater caloric intake than the control group. Lovelady et al.26 did not find a significant difference between the weight and height of infants whose mothers participated in a dietary intervention and exercised aerobically for 45 min per day for 4 days per week, even with a significant weight loss of approximately 0.5 kg per week, compared with the control group. A previous review that examined exercise and lactation found that they are compatible. They found that moderately intensive exercise that is supported by adequate nutritional intake is associated with a more positive sense of well-being, improved cardiovascular fitness, and had no detrimental effect on milk composition, infant milk acceptance or infant growth.56
Timing and duration of intervention
The period of intervention initiation of studies assessed in this review varied widely from 4 weeks26 up to 18 months after birth.22 The duration of intervention ranged from 1026 to 52 weeks.28 Statistically we found no difference in our comparison of studies with interventions 24 weeks to those >24 weeks. However, qualitatively we found that studies whose intervention length was 24 weeks tended to be more intensive and supervised with a greater adherence to recommendations. Based on this analysis we are unable to state the optimal time in the postpartum period to introduce lifestyle modifications. However, it is clear regardless of the timing and duration, that lifestyle modification interventions are effective in promoting weight loss among postpartum women.
Primary care providers should educate women about the importance of a healthy lifestyle, including the increased risks experienced during pregnancy when it is associated with overweight or obesity, exercise, nutrition and loss of weight gained during pregnancy during the first year postpartum, instead of the traditional 6 weeks, due to persistent physiological and psychosocial changes during this period.9
Quality of study methodology
In clinical trials, performing and properly reporting randomization techniques, allocation concealment and blinding methods are key to reducing bias and ensuring the validity of study results.57 Among the studies included in our analysis these three aspects of study methodology were the most problematic. Many of the studies failed to properly report randomization techniques, including how sequence generation and concealment methods were used for participant allocation; therefore, selection bias among participants cannot be ruled out. Given the nature of the interventions none of the studies could blind participants and the majority of the studies chose not to blind investigators to group allocation. As a result performance bias among participants could not be avoided. However, assessment bias among investigators could have been avoided if the study outcome data had been collected by investigators who were blinded to group allocation. However, weight is an objective measure; minimal bias should have been introduced by the lack of investigator blinding.
Attrition bias is common in exercise interventions.13, 58 Of those studies that properly reported adherence and/or dropout rates many were high ranging from 17% up to 40%, which while not unexpected, is of concern.23, 25, 28, 29, 30 Leemarkers et al.25 and Walker et al.30 reported that heavier women are likely to drop out of the studies. However, of more concern is the number of studies who failed to properly report adherence and/or dropout rates. Improper reporting limits the reviewers ability to properly assess the effectiveness and utility of an intervention.
Supervision of intervention
Overall, there was a trend towards supervised exercise intervention strategies being more effective and resulting in greater weight losses than unsupervised intervention strategies.21, 23, 24, 26 These supervised programs were intensive, 4 to 5 days per week, and staff were able to ensure that women achieved their intensity goals through heart rate monitors (>60 and <80% heart rate reserve)24, 26 and/or pedometer measurements (10 000 walking steps or 3000 aerobic steps). However, these kind of strategies require financial and personal resources, as women have to change their routines to attend the classes and must have childcare support available.59 Furthermore, the studies that involved supervised exercise interventions reported high refusal and attrition rates. This can generate bias, as women who chose to stay in the programs are more likely to be committed to exercise and thus have better results. Many of the studies that did not use a supervision strategy had a smaller, but still significant effect on weight loss.25 These kind of interventions likely better reflect the kind of resources available and time commitment that is reasonable for most women during the postpartum period.
A variety of intervention strategies were used, including use of support groups,23, 25, 28, 29, 30 face-to-face consultations,22, 27 telephone calls,21, 25, 27, 29 text messages20 or correspondence support.25 The utility and effectiveness of many of these technology-based strategies have previously been studied.60 In a previous review researchers indentified five key points in technology-based weight loss interventions that were associated with a successful intervention: self monitoring, counselor feedback and communication, social support, use of a structured program and use of an individually tailored program.61 Many of the studies included in our review utilized these strategies. Although a subgroup analysis based on these aspects of the interventions was not completed, we did find that studies that utilized these strategies seemed to be more effective in terms of weight loss and adherence rates.
In our review, we found that intervention strategies that used either a heart rate monitor or pedometer to objectively measure targets were more effective and resulted in greater weight losses than those, which did not use these strategies.20, 21, 23, 24, 26, 28 Heart rate monitors are an important resource and valuable component of exercise programs as they provide accurate, real-time measures of intensity. A pedometer is also a valuable tool as it provides an objective and motivational measure and may also provide a baseline measure of overall physical activity.62
Overall interventions that combined exercise with intensive dietary changes and monitoring resulted in greater postpartum weight loss than interventions that coupled exercise with dietary counseling or used exercise intervention only. These findings agree with a previous meta-analysis, which found that diet plus exercise when compared with usual care was effective on postpartum weight loss, although exercise alone was not effective.14 A possible explanation for these findings, proposed by Dewey et al.,24 is that in the absence of proper diet intervention, women tend to increase their caloric intake as their energy expenditure through exercise increases. This increased intake compensates for increased expenditure and the calorie deficit required for weight loss is not produced, resulting in an ineffective intervention.
Even though walking was the most frequent exercise in the included studies, most of the studies allowed different types of exercise and did not specify that the women must only do one type, and this is a reason why we did not perform a specific analysis to address this issue. Walking is the most commonly chosen physical exercise activity for pregnant and postpartum women63 and was strongly recommended in six of the studies.20, 21, 22, 23, 25, 29 It is a functional, easy, low cost and low risk activity. However, the intensity at which individuals walk can be highly variable and at a light intensity the caloric expenditure can be very low. As such, for pregnant and postpartum women a brisk walking pace of about 5 km h−1 or 12 min per kilometer is recommended to reach a moderate intensity, whereas for a fitter person the pace may need to be increased to 6.5 km/h in order to reach the desired intensity.64
Resistance training was used in just one study whose primary goal was to determine the role of exercise and weight loss on bone mineral density in overweight exclusively breastfeeding women.21 Their hypothesis was that resistance training attenuates the bone mineral loss related with lactation and energy restriction.65, 66, 67 Their results suggest that moderate energy restriction, walking and resistance training are safe methods for weight loss (−5.8±3.5 kg in the intervention vs −1.6±5.4 kg in the minimal care group, P=0.02) with no adverse effects on bone mineral density and infant growth.21 However, this was a small pilot study and the topic needs to be further investigated.
Some of the studies that were excluded from the meta-analyses are worth mentioning; though they did not meet the criteria for inclusion in this analysis, many did show a positive effect of intervention on weight loss. Using a convenience cluster design Kinunem et al. found that their intervention increased the proportion of primiparas returning to prepregnancy weight.49 Huang et al.46 showed that a diet and physical activity intervention initiated during pregnancy is effective for reducing postpregnancy weight retention. Finally, Maturi et al. demonstrated that a physical activity intervention based on pedometer targets is an effective means to increase overall physical activity; reducing retention of weight gained during pregnancy and can improve anthropometric measures in postpartum women.52 Studies that did not examine weight as a primary outcome call attention to other positive effects of exercise programs in this period. For example, in the study by Taveras et al. they found that a program of brief focused negotiation by pediatricians, individual coaching by health educators using motivational interviewing and group parenting workshops tended to improve infant feeding, sleep and media exposure, but had less impact on mothers’ own health related behaviors.50 Lovelady et al. suggested that resistance and aerobic exercise may slow bone loss during lactation with no difference on weight loss and Zourladani et al. found that exercise training programs appear to improve the psychosocial well-being of postpartum women, but did not find a significant difference in terms of weight.51, 53
Strengths and limitations of the report
In this analyses, we compared studies on the basis of the intervention strategies and determined which were associated with greater postpartum weight loss. This identifies important factors for investigators to consider in future studies and highlights factors that may improve outcomes in a clinical setting. Furthermore, areas that require more research were identified. Specifically, the optimal timing, duration and level of supervision in interventions need to be investigated.
This report, as with previous reviews in this field, is limited by the heterogeneity observed in intervention design. As previously described, each of the studies utilized a unique design in terms of population, timing of study initiation, intervention strategy and outcome measures. Because of the heterogeneity, a random effects model was used in the analysis, resulting in more conservative estimates and wider confidence intervals. Differences between the fixed effect and random effect estimates were compared to ensure that small studies were not inappropriately affecting the results. On sensitivity analysis no significant small study effects were observed (Data not shown). To further address the issue of heterogeneity we defined subanalyses based on common characteristics among the studies. This permits the major conclusion that interventions that utilized heart rate or pedometer goals and whose exercise regimen was coupled with intensive dietary interventions were most effective.
Future reviews would be enriched by the use of patient level data as it would allow for even further confidence in the results. In addition, it would allow for more studies to be included in the analysis as comparable outcome variables could be calculated from the original study data.
Conclusion and future directions
The postpartum period provides a window of opportunity to initiate lifestyle modification interventions that encourage healthy behaviors, such as physical activity and nutritious eating. These healthy behaviors adopted by the mother may contribute to a healthier lifestyle for the family as a whole; hence it should be encouraged by health care providers.
Some clear conclusions may be drawn from the current analysis. Specifically, there is benefit from lifestyle interventions; exercise programs with objectively defined goals that are combined with intensive dietary intervention are effective in reducing weight in postpartum women.
Further evaluation into the sustainability and long term effects of the interventions on the mother and child are needed. The cost-effectiveness of the interventions and their feasibility in terms of incorporation into clinical settings may be assessed. Finally, strategies may be developed to improve the adherence and compliance to these kinds of lifestyle interventions.
The authors would like to thank CAPES-Brazil/DFAIT-Canada agreement facilitating the collaboration between researchers from the University of Campinas (Unicamp) (Brazil) and Queen's University (Canada). A special thanks for Professors BA Croy, AT Yamada and JG Cecatti for the opportunity and encouragement to SLN to develop this work at Queeńs University, Canada.
About this article
Measuring Adherence to a Nutrition and Exercise Lifestyle Intervention: Is Program Adherence Related to Excessive Gestational Weight Gain?
Behavior Analysis in Practice (2017)