Strengths and limitations
Our systematic review was comprehensive in its scope and search. We conducted the review in line with contemporary recommendations and complied with the PRISMA (preferred reporting items for systematic reviews and meta-analyses) statement.92 Our search of literature aimed to minimise the risk of selection and publication bias. Most of the published reviews on effects of dietary and lifestyle interventions on maternal and fetal outcomes were limited to specific groups of women or types of intervention. There was no formal prioritisation of the importance of the clinical outcomes, and few assessed the quality of the evidence for the important outcomes. We undertook rigorous quality assessment and formally prioritised the outcomes for clinical importance. Reliable data were identified on clinically important outcomes related to weight and pregnancy by the Delphi survey. We explored for sources of heterogeneity when required. Appropriate subgroup analyses and sensitivity analyses planned a priori were undertaken for important factors such as BMI, diabetic status, maternal weight change with intervention, and study quality that could influence outcomes. We formally rated strength of evidence for key outcomes identified through Delphi survey. This enabled our confidence in the estimates of the important effects observed. Our careful scrutiny and presentation of evidence profiles provides the much needed clarity necessary to make judgments about effects. The validity of a meta-analysis depends on the quality of the component studies, heterogeneity observed, and the risk of publication bias. The quality across various outcomes assessed by GRADE was moderate for the benefit observed with gestational weight gain but low for other important obstetric outcomes such as pre-eclampsia, gestational diabetes, gestational hypertension, and preterm delivery. This weakens the inferences for these outcomes. The reasons for low evidence rating were the significant heterogeneity observed in the effect size, deficiencies in the quality of the individual studies, and risk of publication and related biases. We observed heterogeneity for beneficial effects of interventions on maternal weight gain that persisted after accounting for the type of intervention, BMI, and diabetic status. Further information is needed on characteristics of included women—such as age, ethnicity, socioeconomic status, parity, and underlying medical conditions—and characteristics of the interventions—such as frequency, duration, and intensity—that could influence the outcomes. We were limited in our ability to identify the optimal weight change in pregnancy with interventions that would minimise maternal and fetal complications. Furthermore, constraints in the available data limited assessment of baseline prognostic factors on the effectiveness of outcomes. Such questions were difficult to answer with extracted results from trial publications because patient level information was not available and subgroup effects (“treatment-covariate interactions”) were rarely reported in sufficient detail. Although the Delphi panel of clinicians identified long term neurological sequelae and metabolic syndrome of the fetuses exposed to the intervention, they were not reported in any of the studies.