Gestational weight gain and group prenatal care: a systematic review and meta-analysis

The published literature was searched using strategies created by a medical librarian (S.A.F.) for the concepts of group prenatal care and gestational weight gain.(Additional file 1) These strategies were established using a combination of standardized terms and key words, and were implemented in Medline 1946-, Embase 1947-, Scopus 1823-, Cochrane Database of Systematic Reviews, Database of Abstracts of Reviews of Effects, Cochrane Central Register of Controlled Trials, clinicaltrials.gov 2000-, and Google Scholar. All searches were completed in April 2017. No database limits such as language or years were applied. All literature database and Google Scholar results were exported to EndNote.

Studies were included if they described a group prenatal care setting (exposure) and reported gestational weight gain (outcome), either as means or by adequacy of gestational weight gain categories, per the Institute of Medicine guidelines. We included original research studies such as randomized controlled trials or observational studies (retrospective cohort, case-control) with a comparison group of traditional or individual prenatal care. We excluded case reports, case series, review articles, studies without comparison groups, and studies published in languages other than English.

The first and second authors (M.A.K., A.K.L.) screened the titles and abstracts and then retrieved the full-text articles if they appeared relevant or if there was uncertainty regarding the screening criteria. Full-text articles and abstracts (when full-text was not available) were independently reviewed for inclusion and exclusion criteria by the same authors (M.A.K., A.K.L.), who have expertise in prenatal care. Consensus was achieved between these two authors for included articles. For each article that met the inclusion criteria, study characteristics, participant demographics (age, parity, race-ethnicity, body mass index), study-specific inclusion criteria, setting, gestational weight gain definition, and outcome data relevant to gestational weight gain were abstracted and summarized. Specifications for body mass index (pre-pregnancy vs. first prenatal care visit measurement vs. not recorded) and gestational weight gain (difference between the pre-pregnancy vs. first prenatal visit and final prenatal visit vs. delivery weight vs. not recorded) were also abstracted. Categorization of gestational weight gain was defined by the 2009 Institute of Medicine guidelines per pre-pregnancy body mass index category (28–40 pounds for < 18.5 kg/m², 25–35 pounds for 18.5–24.9 kg/m², 15–25 pounds for 25.0–29.9 kg/m², and 11–20 pounds for ≥30 kg/m²), if applicable, with the terms inadequate, adequate, or excessive [16]. We also noted whether or not gestational age at delivery was accounted for in the total gestational weight gain measurement and whether or not the total number of group or traditional prenatal care visits attended was reported.

The quality of each study was assessed with the Down’s checklist which contains 27 questions pertaining to threats to validity including reporting, external validity, internal validity, confounding or selection bias, and power [17]. The Down’s checklist is validated for both randomized and observational studies. The maximum possible score was 28, indicating the highest quality study. We considered studies receiving the majority of the points available in at least four of the five categories of threats as high quality with scoring similar to O’Connor et al. (24–28 excellent, 19–23 good, 14–18 fair, and < 14 poor) [18]. Both M.A.K. and A.K.L. completed the quality rating form for each article, with E.C. resolving any discrepancies in scoring to achieve a final consensus score.

The primary outcome was the occurrence of excessive gestational weight gain as determined by body mass index and the Institute of Medicine guidelines, but we also evaluated the occurrence of adequate gestational weight gain. Data were analyzed with STATA (version 14, College Station, TX) using the METAN software package. Study heterogeneity was assessed with Cochran’s Q and Higgins I² tests. Conservative significance thresholds of P < 0.1 for the Q tests or I² > 30%, were used to test heterogeneity [19]. Relative risks (RR) were calculated with raw data from each study with a 95% confidence interval (CI) using raw data from each study. The DerSimonian-Laird random-effects models were used to pool data, regardless of whether there was evidence of statistical heterogeneity. RR for each categorical outcome were plotted graphically as forest plots. A sensitivity analysis by study quality with those studies at the top tertile of Down’s scores was also performed to assess the effect of these factors on our estimates. Publication bias was assessed graphically using funnel plots.

Institutional review board approval was not necessary for this study of de-identified data available in the public domain through prior publications. This systematic review was registered in PROSPERO on February 1, 2017 (#CRD42017056296). The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed for all aspects of reporting [20].

The initial electronic search found 631 results. One hundred twenty six duplicates were accurately identified and removed with the automatic duplicate finder in EndNote. An additional 47 duplicate citations were identified by hand and removed for a total of 458 unique citations.(Fig. 1) Titles and abstracts for each citation were reviewed and 64 full-text articles (or abstracts if full-text was not available) were screened according to the inclusion and exclusion criteria. Reference lists of excluded articles were reviewed to identify other relevant articles and no additional studies were found. A total of 49 studies were removed for no reported gestational weight gain outcomes, no group prenatal care arm, no control group, and/or review article only.

Fifteen studies met inclusion criteria, including one randomized controlled trial (RCT), one secondary analysis of an RCT, one study with “random assignment” and 12 cohort studies for a total of 13,779 subjects, 4243 (31%) in group and 9536 (69%) in traditional prenatal care. Table 1 describes the characteristics of studies meeting inclusion criteria. CenteringPregnancy™ group prenatal care was the most common prenatal care model used in included studies (n = 13). In CenteringPregnancy™, women receive prenatal visits, build relationships with other women, and gain knowledge and skills in pregnancy and childbirth in 10 sessions [21]. During the first group session, nutrition, including caloric requirements and macronutrient recommendations, is typically discussed. Women also receive a notebook that includes a food diary and a body mass index table labeled with the categories of normal, overweight, obese, and extreme obesity. Additionally, women chart their own weight over time in the notebooks. The curriculum overall encourages goal setting, including diet, exercise, and weight gain. None of the studies using the CenteringPregnancy™ model in this systematic review reported whether or not the content was specifically adapted to target weight gain. The two non-CenteringPregnancy™ models are described as follows. Harden et al. performed a mixed methods study on group-based lifestyle sessions within prenatal care with six-1 h group visits consisting of safe group exercises, nutrition education and demonstrations, and group activities such as goal setting [22]. Mazzoni et al. described a group prenatal care intervention consisting of four sessions for women with gestational diabetes [23]. Several of the other included studies describe an intervention that was targeted at specific populations such as adolescents [24, 25], African-Americans [26], Hispanics [27], active-duty military or their spouses [28, 29], women with obesity [22] and women with gestational diabetes [23, 30].

The validated Downs scale for randomized and observational studies was used to assess the quality of the studies with scores ranging from 9 to 19 with a median value of 12 (interquartile range 12–16). (Table 2) Of the 15 studies selected for this systematic review, 6 were not included in the meta-analysis because they reported only a mean gestational weight gain without regard to the participants’ pre-pregnancy body mass index [22, 23, 26, 29‐31]. Two of these studies also did not report the variance of the gestational weight gain measure [29, 31]. Without a pre-pregnancy weight value, there is limited clinical interpretation for a mean gestational weight gain. Four of the six studies also did not specify if Institute of Medicine guidelines were used to counsel participants on overall gestational weight gain goals [23, 24, 30, 31].

Table 3 describes the sample size, weight gain outcomes, how the control group was chosen if applicable, any adjustments for gestational age in the analysis, total number of prenatal visits, and provider types for all 15 studies. The six studies not included in the meta-analysis are summarized as follows [22, 23, 26, 29‐31]. Of note, only two studies had gestational weight gain as a primary outcome [22, 29] and only one study adapted the content of care to target weight gain [22]. A retrospective cohort study based in a clinic which served predominantly low-income minority women was the first to report on gestational weight gain in CenteringPregnancy™ from 2004 to 2006. This study found increased mean gestational weight gain in women in CenteringPregnancy™ (n = 61) compared to women in traditional prenatal care (n = 207) (32.2 pounds vs. 28.5 pounds p < 0.05), yet the pre-pregnancy body mass index was not reported so the proportion of women who met the gestational weight gain goals is not known [26]. Similarly, Holbrook evaluated gestational weight gain among 100 Spanish-speaking women receiving group (based on CenteringPregnancy™ model) or traditional prenatal care in 2009 in the U.S via a “pilot chart review”; however, pre-pregnancy body mass and statistical comparisons for gestational weight gain were not provided [31]. A RCT in a military setting from 2005 to 2007 found no difference in mean gestational weight gain between CenteringPregnancy™, but the authors also did not specify the pregnancy body mass index [29].

Harden et al. reported on group-based lifestyle sessions within prenatal care in two “feasibility pilot studies with random assignment” that aimed to limit excessive gestational weight gain [22]. In the first study, women with a pre-pregnancy body mass index ≥30 kg/m² were randomly assigned to either standard of care (n = 8) or the intervention (n = 8). Women in the control group gained significantly more weight at the 20th week of the program compared to those in the intervention (8.6 ± 3.9 kg vs. 5.3 ± 5.3 kg, p < 0.01). The investigators received positive feedback from both providers and patients about the program, but because the participants suggested that the medical appointments be conducted separately from the group-based lifestyle sessions, they conducted a follow-up study whereby women of all body mass index categories were randomized to either the group-based lifestyle sessions (n = 28) or the control group (n = 23) which consisted of individual goal setting, didactic nutritional education, and a 30 min standard exercise class. The group-based lifestyle sessions were similar to the initial study and also included discussion of barriers and potential solutions to reaching goals, sharing of baked snacks from the provided recipe book, exchanging of telephone numbers, and group exercises such as a walking track, dance, and prenatal yoga. Because the follow-up study occurred separately from prenatal care visits, the information was not included in the meta-analysis. Furthermore, 19 (37%) of all participants were missing gestational weight gain, so the finding that 36% in the group-based lifestyle sessions vs. 13% in the control group met the gestational weight gain goals (p = 0.06) needs to be interpreted with caution [22].

The two other studies not included in the meta-analysis specifically targeted women with gestational diabetes. Mazzoni et al. aimed to compare the progression of gestational diabetes (from diet to medically-treated) in group and traditional prenatal care in a prospective observational study in the U.S [23]. Participants started the program between 26 and 32 weeks gestation for a total of four sessions which covered topics such as meal planning, glucose log review, mindful eating, and lifelong diabetes prevention. Mean gestational weight gain was similar in both group (19.2 ± 13.0 pounds) and traditional (18.0 ± 15.0 pounds) prenatal care, p = 0.57. Lastly, Schellinger et al. developed an adapted CenteringPregnancy™ model in Spanish only for Hispanic women with gestational diabetes. The CenteringPregnancy™ model was adapted to a total of four visits in the 3rd trimester with facilitated discussions on blood sugar monitoring, nutrition, and exercise [30]. Gestational weight gain, a secondary outcome in their retrospective cohort study, was compared to women in traditional prenatal care with gestational diabetes, but only mean values were reported (9.3 ± 4.5 pounds vs. 10.2 ± 6.7 pounds, p = 0.26) [30].

Of the nine studies in the meta-analysis, all used the CenteringPregnancy™ model and three studies had both excessive and adequate gestational weight gain outcomes. Seven studies reported the primary outcome of excessive gestational weight gain (Fig. 2) [25, 27, 28, 32‐35]. Overall, there was no significant difference in the occurrence of excessive gestational weight gain [7 studies: pooled rates 47% (1806/3582) vs. 43% (3839/8521), RR 1.09, 95% CI 0.97–1.23] in group and traditional prenatal care. Given the low overall quality values (fair to poor), we separately evaluated studies in the top tertile quality scores based on the range and distribution of scores and found that excessive gestational weight gain was higher in group compared to traditional care [pooled rates 45% (1608/3207) vs. 39% (3631/8149), RR 1.15, 95% CI 1.01–1.30] (Fig. 3). There was significant heterogeneity between the studies that evaluated excessive gestational weight gain, as demonstrated by I² values of 74.2% (p = 0.001) and 68.0% (p = 0.014) (Figs. 2 and 3). Six studies reported the secondary outcome of adequate gestational weight gain (Fig. 4) [24, 27, 33‐36]. There also was no significant difference in adequate gestational weight gain [6 studies: pooled rates 31% (798/2875) vs. 30% (1410/5187), RR 0.92, 95% CI 0.79–1.08] in group and traditional prenatal care. There also was significant heterogeneity between the studies that evaluated adequate gestational weight gain, as demonstrated by an I² value of 58.6% (p = 0.03) (Fig. 4). There appeared to be symmetry with the funnel plot and therefore publication bias was minimal (Fig. 5).

This systematic review found 15 studies of poor to fair quality with 13,779 subjects meeting the criteria of comparing gestational weight gain according to a group vs. traditional prenatal care model, with the number of studies further limited for the meta-analysis of excessive (n = 7) and adequate (n = 6) gestational weight gain [22‐36]. We found that group prenatal care was not associated with excessive or adequate gestational weight gain except in the highest quality studies in the meta-analysis, which had a very modest but increased risk for excessive gestational weight gain for group prenatal care. Of the studies included in this meta-analysis, Tanner-Smith et al. and Magriples et al. both found reductions in excessive gestational weight gain in adjusted analysis (propensity score matching for 20 maternal characteristics and multi-level modeling, respectively) for CenteringPregnancy™ women, which suggests other confounding factors such as race-ethnicity, gravidity, pre-pregnancy body mass index, and gestational age at the entry to prenatal care are responsible for the relationship between prenatal care model and gestational weight gain [25, 35]. These two studies used the 2009 Institute of Medicine definitions for body mass index and gestational weight gain, directly compared CenteringPregnancy™ and traditional prenatal care, and had a greater total number of subjects (n = 569 and 984, respectively) than most studies. Conversely, Kominiarek et al. found a direct association between excessive gestational weight gain and group prenatal care, primarily among normal and overweight women, in a study of 6351 women [33].

Our study has several strengths. First, we used a predesigned protocol with a comprehensive search strategy conducted by a trained, expert research librarian, and two authors independently extracted data, which reduced bias. Second, we used a pre-specified sensitivity analyses according to the risk of bias assessment. Third, we pooled data from studies using a more conservative random-effects model to account for clinical heterogeneity between studies. Finally, because gestational weight gain is an important construct that impacts maternal and offspring outcomes and there are many variations of group prenatal care across the U.S., this systematic review and meta-analysis sheds some light on the state of the inconsistent evidence for group prenatal care as it relates to this important outcome. We acknowledge several limitations to this systematic review and meta-analysis. The majority of the studies were secondary analyses from RCT or retrospective cohort studies where the risk for selection bias and confounding is greater especially since women typically opt-in or self-select for group prenatal care. For example, women in group prenatal care are typically younger, of lower parity, and more frequently minorities of lower socioeconomic status, all of which can positively or negatively influence gestational weight gain [37]. The selection of women from traditional prenatal care varied among the studies and included matching for age, delivery date, ethnicity, and/or pre-pregnancy body mass index, matching for women who declined group prenatal care, and matching according to a propensity score. We noted a greater number of nulliparas [24, 27, 28, 36] and minority women [24] in group care compared to traditional prenatal care in some of the studies included in this systematic review. In summary, these factors, including participant demographics (age, race-ethnicity, geographical location), participant characteristics (age, body mass index, gestational diabetes), and intervention dose (length and number of visits, gestational age at entry) may explain the heterogeneity we observed in the meta-analysis.

We were not able to evaluate the content (i.e., how much emphasis placed on gestational weight gain goals) or dose (i.e., number of group sessions attended) of the group prenatal care program or the experience or effectiveness of the group prenatal care facilitator. The primary outcome, gestational weight gain, was not well defined in several studies and also varied from a difference between a first and last prenatal visit to a difference between the start and end of the intervention. Other limitations of these individual studies for a meta-analysis include lack of pre-pregnancy body mass index information, differing statistical approaches, and small cohort sizes.

With respect to weight management, highly effective components of weight management interventions include calorie and physical activity goals, meal replacements, daily self-weighing and monitoring of food intake, behavior therapy, and frequent provider-patient contact [13]. Several of these components are already integral elements of group prenatal care such as CenteringPregnancy™. In general, group health care visits have the potential to improve the effectiveness and efficiency of health care because multiple patients are seen in the same clinical setting by a multidisciplinary team consisting of physicians, nurses, dieticians, and other health educators [38, 39]. Weight management is just one example where group care models are practiced among others including diabetes, smoking cessation, geriatrics, and osteoarthritis [40‐42].

Social support in the context of group meetings with either peers or self-identified family and friends has been studied in weight management interventions that have highly effective components. A systematic review of five RCT with a group intervention found improved weight change within 1 year for the participants in the group compared to usual care [weighted mean difference − 1.4 kg (− 2.7/− 0.1 kg)] with a greater effect seen when a financial incentive was incorporated [weighted mean difference − 2.8 kg (− 5.4/− 0.2 kg)] and when the groups were led by psychologists as compared to nutritionists [9]. Furthermore, an adapted version of the Diabetes Prevention Program showed that participants in either large (≥16 members) or small groups (< 16 members), still had weight loss outcomes (5.1–5.8 kg) similar to the goals of the initial study (7% weight loss from initial weight) [43]. Commercial weight loss programs promote the concept of social support as critical to achieving goals [44, 45]. It has been proposed that support from attending meetings [e.g., Weight Watchers® and TOPS Club, Inc.® (Taking Off Pounds Sensibly)] enhances feelings of control and confidence and consequently group-based interventions result in greater weight loss compared to individual care [9, 10, 44‐46]. For example, in a prospective, 2-year clinical trial that randomly assigned participants to either Weight Watchers® meetings or the self-help method, those assigned to Weight Watchers® meetings lost and kept off significantly more weight [44]. However, findings regarding social support and weight loss are mixed as another meta-analysis of 21 studies concluded that couples programs were effective in short-term weight loss, but not in long-term weight loss maintenance [47]. Given that social support is cited as a successful element in weight management programs in non-pregnant populations and the social support element that accompanies many group prenatal care models such as CenteringPregnancy™, it is reasonable to evaluate group prenatal care models regarding their effectiveness in changing health behaviors and achieving gestational weight gain goals.