We investigated the potential to reduce the rate of excessive weight gain during pregnancy by lifestyle counseling given to pregnant women. The intervention resulted in a lower proportion of women exceeding IOM recommendations compared with the control group without increasing inadequate GWG. Our findings are in line with earlier reports showing reductions in GWG by lifestyle intervention in pregnant women [35, 40, 48, 51, 52]. To date, diverse intervention strategies and intensities, and differences in study populations and design complicate comparisons between trials. In a meta-analysis, Streuling et al.  combined data from four randomized controlled trials and five non-randomized trials with a total of 1,549 women. They found a lower GWG in the intervention groups, with a mean difference of 1.2 kg. While they only selected studies with interventions combining diet and physical activity, the recent meta-analysis of Thangaratinam et al.  included all randomized controlled trials that evaluated any dietary or lifestyle intervention, and found a 1.4 kg reduction in GWG with any intervention compared with controls. These observed differences in weight gain are similar to those observed in our study.
However, little is known about the risk of increasing the proportion of women with suboptimal weight gain by lifestyle intervention, especially if delivered globally to all pregnant women . Our results show that by using our intervention scheme reducing excessive weight gain without increasing inadequate weight gain is possible. The use of weight gain charts that mark upper and lower GWG limits, as well as individual recommendations based on nutrition and physical activity questionnaires, may be key components for effective and safe interventions that are beneficial for all pregnant women. Moreover, we evaluated the effects of our intervention on weight retention at four months pp; the lifestyle counseling significantly decreased the proportion of women retaining a substantial amount of more than 5 kg weight. These data are consistent with results obtained in other intervention trials [48, 52]. A recent meta-analysis of nine observational studies concluded that gaining weight according to IOM recommendations could avoid long-term high pp weight retention .
Our feasibility study failed in demonstrating any statistically significant differences between the intervention and control group regarding pregnancy complications, as well as obstetric and neonatal outcomes. Also most other randomized controlled trials in this field failed in identifying any differences in such outcomes, and most including our study were inadequately powered to address these issues [38, 41–43, 50, 51]. However, the FeLIPO study detected some favorable trends concerning the outcomes: gestational diabetes and cesarean section. These trends seem to be in line with our expectations and fit with observational data [10, 11, 13, 15, 77, 78] and recent results from randomized controlled trials [40, 47, 49] and meta-analyses [60, 62]. Nevertheless, the FeLIPO study was not designed to assess these outcomes. Further studies, adequately powered for such outcomes, are needed for thoroughly testing the effect of lifestyle counseling on pregnancy complications, and on obstetric and neonatal outcomes.
A lower energy intake may have contributed to optimizing gestational weight gain in our intervention group. In 2011, Streuling et al. performed a systematic review of observational studies with the aim of associating weight gain with dietary intake. They suggested gestational weight gain might be reduced by lower energy intake during pregnancy as supported by our data .
Women in both groups decreased their physical activity as pregnancy progressed. Less physical activity is common during pregnancy  and is mostly caused by pregnancy-related health problems like sickness; lack of energy; feeling uncomfortable due to enlarged body size; and lack of time (due to work or childcare) []. This decrease could not be prevented by our intervention. Also, most of the lifestyle intervention trials reporting data on physical activity did not observe an effect of their intervention program [34, 35, 37, 42]. Nevertheless, women in our intervention group showed a smaller decrease in physical activity when compared to controls, which may have contributed to the effects of the intervention. Further analyses of the dietary records and physical activity questionnaires are ongoing and might provide additional insights about causes for the observed effects.
A strength of the FeLIPO study was the use of an intervention program with practical relevance that could be implemented in the health-care system for pregnant women. As intended, the intervention could be scheduled in combination with prenatal visits resulting in both high participation and low dropout rates. The possibility for a spillover effect between groups was minimized by our cluster-randomized design. However, there were some limitations in the study, for example: gestational weight gain was analyzed based on self-reported prepregnancy weight, which may have been underestimated by (especially) overweight and obese women, leading to an overestimation of total weight gain . However, comparing the first measured weights recorded in maternity cards (booking) with self-reported prepregnancy weights, the latter were about 2 kg lower in both groups and the two parameters were highly correlated. This approach is widely used in this type of study where there is a lack of data concerning measured weight , and furthermore has yielded valid estimates [83, 84]. Women in the control group were aware of participating in a trial aiming at promoting a healthy lifestyle and optimizing gestational weight gain, which may have influenced their behavior, resulting in an underestimation of the intervention effect. Significant baseline differences in prepregnancy BMI and age were identified between study groups. Although these variables were included as adjustment variables in our analyses, baseline differences between the groups contributing to the efficacy of the intervention cannot be excluded. Although the counseling sessions followed a pre-defined curriculum differences between counselors are possible. We did not account for clustering in the sample size calculations. As a further limitation, the number of women approached in the control practices was lower than in the intervention practices, which we speculate may be related to unmotivated gynecologists and practice staff recruiting participants, or to lower numbers of pregnant patients during the recruitment phase among practices randomized as the control. As practice staff and participants were not blinded to the study purpose and group allocation, referring to the control group might have influenced recruitment and participation rates, which raises the possibility of post-randomization selection. However, both groups were comparable with regard to most sociodemographic parameters. Nevertheless, larger studies are needed to confirm these results.