Group-based activities with on-site childcare and online support improve glucose tolerance in women within 5 years of gestational diabetes pregnancy

Women with a gestational diabetes (GDM) history have a greater than seven-fold risk increase for type 2 diabetes [1, 2] and increased cardiovascular disease risk [3, 4], compared to women without glucose elevations in pregnancy. The American Diabetes Prevention Program trial (DPP) [5] demonstrated that in prediabetes, an individualised eating and physical activity behaviour change program led to a greater than 50% reduction in type 2 diabetes incidence. Further, impact was specifically confirmed in women within an average of 12 years of first GDM pregnancy. However, given that the period of highest diabetes incidence is within 5 years of a GDM pregnancy [6], the DPP intervention was arguably “late.” Engaging women earlier could not only prevent more cases of type 2 diabetes but also reduce the risk of recurrent GDM [7]. Notably, subsequent pregnancies without GDM are an indicator of decreased risk of type 2 diabetes development later in life [8].

Unfortunately, trials within 5 years of a GDM pregnancy have met with limited success. Interventions have mainly been at the individual level, including dietary and physical activity counselling and support. These have been delivered through face-to-face consultations with additional phone/text/email support [9‐12] or without face-to-face contact, relying exclusively on phone calls and written material [13] or web-based programs [14]. A single study examined a group-based approach consisting of weekly physical activity sessions [15]. In all cases, impact on weight was limited [10‐14]. When assessed, no improvements in fasting [10, 12, 14, 15] or 2-hour post challenge glucose [12, 14, 15] have been observed. Further, participation appears to be challenging [15, 16], largely because of work and family-related responsibilities. There is a need for the development and testing of alternate strategies, as well as assessment of impact on a greater variety of cardiometabolic risk factors, to comprehensively capture possible benefits.

As previously reported, we conducted focus group discussions with women within 5 years of a GDM pregnancy, seeking their input on the design of a realistic intervention strategy. Our analyses signalled a need for group-based face-to-face interactions with peers and professionals [17]. However, to facilitate attendance, on-site childcare was viewed to be important. Phone, web-based, and text support were considered adjunctive. Spousal involvement was deemed critical to allow health behavioural change at home. We designed an intervention that incorporated these elements. We assessed its impact through the MoMM pilot study, evaluating not only changes in body weight but in insulin resistance, blood pressure, lipid profile, eating behaviours and daily step counts, as reported herein.

Thirty-six women were recruited. Among these, 27 (75.0%) completed baseline and final assessments. Among the 9 women that withdrew, 5 did not attend any group-sessions (new pregnancy, 2; divorce or death in family, 2; unspecified, 1) and 4 attended at least one (time pressure, 2; feeling depressed, 1; divorce, 1). Those who completed both assessments were, on average, approaching their middle years (mean age 40 years, SD 5) and educated beyond high school (88.9%). More than half (63.0%) were Europid, most lived with their partner (92.6%), had an average of two children, and worked outside the home (74.1%). Questionnaire data indicated that 5 (18.5%) had elevated anxiety levels and one had depressed mood (3.7%). Roughly half reported limited social media use (never, 18.5%; monthly, 33.3%). At baseline, most were in an action (44.4%) or maintenance (18.5%) stage for weight change.

Four women (14.8%) were included in the study although their BMIs were < 24 kg/m², (i.e., did not meet eligibility criteria). We opted to include them because of recruitment challenges. Three (11.1%) more had a BMI between 24 and 25 kg/m², 10 (37.0%) were overweight (25 ≤ BMI < 30 kg/m²) and 10 (37.0%) were obese (BMI ≥ 30 kg/m²). Eleven (37.0%) met criteria for dysglycemia (3 impaired fasting glucose alone, 1 impaired glucose tolerance alone, 6 impaired fasting glucose and impaired glucose tolerance, 1 type 2 diabetes). We opted to retain the individual with type 2 diabetes but performed a sensitivity analysis excluding this individual.

Roughly one third (33.3%, Table 1) who completed baseline and final assessments attended all in-person sessions. Approximately half (48.1%) missed a single session. Most used the free on-site childcare (88.9%) at all sessions that they attended. Among the 25 (92.6%) who lived with a partner, 10 (40.0%) had a partner who attended at least one session. Almost all participants (85.2%) visited the MoMM website. Half (51.9%) used the on-line step count-tracking tool and logged onto the eaTracker tool (59.3%). Participants emailed the staff-contact a median of 5.5 times (IQR 5.5-15.5) over the course of the study.

Among women within 5 years of a GDM pregnancy who participated in a 13-week tailored intervention (i.e., 4 group sessions, partners invited to two, on-site childcare, meal preparation, pedometers and floor exercises), attendance at sessions was high, the childcare service and website were used, and there was some spousal participation. Self-reported eating control was not enhanced and there were no changes in BMI or body composition. However, there were important increases in fruit and vegetable intake and consumption of both convenience meals and eating out declined. Step counts also increased. There were reductions in both fasting and 2-hour postchallenge glucose and improvements in all measures of insulin resistance and sensitivity. There was also lowering of both systolic and diastolic blood pressure and triglycerides. On balance, the tailored intervention demonstrated both behavioural and biological impact and thus merits further study and development.

Less than a fifth of the participants missed 2 or more of the 4 sessions. On-site childcare may have been a key facilitating factor as evidenced by high use, with almost 90% of the women using it at every session they attended. Further, monthly sessions may have also enhanced attendance, being realistically aligned with time availability. Indeed, in a comprehensive lifestyle change study post pregnancy in a non GDM population, participants were able to attend only 4 of 10 scheduled sessions [16]. Higher attendance had been observed in studies with phone-based or at home sessions, and if sessions are scheduled at the same time of routine medical follow-up [33, 34]. For example, among women with GDM who were recruited during pregnancy, 79% completed, during pregnancy, ≥ 2 out of 3 sessions (i.e., 1 in-person and 2 telephone counselling calls) and, in the postpartum period, they participated in an average of 9.4 sessions out of a maximum of 15 (i.e., a maximum of 2 in-persons and a maximum of 13 telephone counselling calls) [9]. We would note, however, that our participants expressed strong endorsement of in-person group sessions and had high levels of attendance.

In the focus group discussions that helped us to design the intervention [17], women expressed a need for spousal involvement. A need for spousal involvement to achieve changes in health behaviour has also emerged in other qualitative studies conducted among women with a GDM history [35, 36]. Spousal participation was not a component of previous intervention studies [9‐13, 15]. The attempt to involve spouses is a novel aspect of our intervention. While we did invite spouses to two of the four intervention sessions and there was some participation, a large proportion of spouses did not attend any sessions. Better engagement may further enhance health behaviour change. While time constraints are a real challenge, lack of interest could potentially be addressed with better knowledge about personal and familial diabetes risks related to eating and physical activity habits of the family. Our recent systematic review and meta-analysis [37] estimated that type 2 diabetes in one spouse is associated with a 26% increased risk for type 2 diabetes in the other; further, studies that performed blood tests systematically indicated a doubling of diabetes risk. The concept of shared diabetes risk may potentially be leveraged to increase engagement of spouses in diabetes prevention efforts.

Our post-intervention questionnaire related to impressions/opinions on the intervention strategy suggests a high level of endorsement for in-person, hands-on components, and pedometer use. The availability of on-site childcare was a clear facilitator. While a substantial proportion did use the on-line tools and website, this was much less than in-person session participation. This may not be surprising given that half of participants reported not using web-based media frequently (i.e., less than monthly) at baseline. Web-based tools appear to be underutilised by this group of adults. In a web-based pedometer intervention in women with a GDM history, Kim and colleagues [14] noted that only 3 out of 21 participants used the web forum and questions were only directed to the study staff. This is unfortunate given the low cost and convenience of web and text-based communications. Future studies may need to incorporate strategies to facilitate web-based communication and engagement. For example, study personnel may need to proactively encourage web-based discussion among participants.

The 1.5 servings/day increase in fruit and vegetable intake that we observed likely contributed to the cardiometabolic benefits realized (i.e., glucose, insulin resistance and sensitivity, triglycerides, blood pressure). In a cross-sectional study on 2,115 adults at risk for diabetes, a two-serving increase in daily fruit and vegetable consumption was associated with a 0.08 mmol/L reduction in FPG and 0.2 mmol/L reduction in 2hPG [38]. In a pooled analysis of three prospective cohort studies, fruit consumption was associated with a lower risk of type 2 diabetes (Hazard Ratio 0.98, 95% CI 0.97, 0.99 for three servings/week increment of total whole fruit consumption) [39]. Consistent with the protective effects of fruits and vegetables, diabetes risk calculators, such as FINDRISC, include daily vegetables, fruits and berries consumption as a factor that protects against diabetes development [40]. Moreover, specifically among women, the European Prospective Investigation into Cancer and Nutrition study demonstrated that a 0.5 serving increase in daily intake of vegetables to be associated with 0.20 and 0.09 mm Hg reductions in SBP and DBP, respectively [41]. As reviewed in a meta-analysis of randomised controlled trials, a 10 mm Hg reduction in SBP or 5 mm Hg reduction in DBP leads to a 22% reduction in coronary heart disease events and a 41% reduction in stroke [42]. Our participants achieved mean reductions of more than 4 mm Hg in SBP and 3 mm Hg in DBP, clinically-important changes.

We observed a 733 steps/day increase, greater than that observed in a 12-week web-based pedometer program (mean 543, SD 2074) in women with a GDM history [14]. Among NAVIGATOR trial participants (Nateglinide and Valsartan in Impaired Glucose Tolerance Outcomes Research), both step counts at baseline and mean 6-year increases were associated with reductions in occurrence of cardiovascular events [43]. Our previous studies have demonstrated inverse relationships between blood pressure and step counts, particularly in women [44]. The increase in step counts thus likely also contributed to the cardiometabolic improvements we observed in the present study.

The mean improvements that we observed in FPG and 2hPG are at least as great as those observed in the Finnish Diabetes Prevention trial (−0.3 mmol/L FPG; −0.8 mmol/L 2hPG) [45]. Given that the Finnish Diabetes Prevention Trial, like the DPP, achieved major reductions in diabetes incidence through a lifestyle intervention, the comparability of the improvement in 2-hour post 75 g glucose challenge that we observed appears promising. Further, computations of insulin resistance and sensitivity derived from glucose and insulin measurements consistently demonstrated improvements (HOMA-IR, ISI_0,120, Matsuda index). Notably, higher insulin sensitivity estimated with the Matsuda index is associated with a decreased incidence of diabetes in high-risk populations [46].

We acknowledge several limitations. First, the primary outcome of our study was a change in weight but no important weight reduction was achieved. More emphasis on energy expenditure reduction may have led to greater weight changes, although adherence may have been challenging. Although our target population was overweight women within 5 years of a GDM pregnancy, some of the women enrolled had a normal BMI. Nonetheless, the BMI range of participants enrolled was more representative of women with GDM in general [47]. Weight reduction observed in overweight/obese women alone was also modest −0.5% (95% CI: −1.7, 0.8), and other changes observed in normal-weight weight women were similar to those observed in the rest of the participants (Additional file 1: Table S1). Second, recruitment was challenging, as anticipated. We endeavoured to recruit women within 5 years of a GDM diagnosis. This is a period of high risk for conversion to type 2 diabetes [6]. While interventions even closer to the time of GDM diagnosis may have had even greater potential impact [6], it would have rendered recruitment more challenging. As reported in our previous focus group study, more than 1,000 invitation letters were sent to women with a prior GDM history. From this pool, we were able to enroll 29 in our focus group study and 28 in the intervention study, with the remainder recruited through telephone contact. The women enrolled may thus not be representative of all women within 5 years of a GDM pregnancy, although they may be representative of those willing to engage in prevention efforts. For capturing dietary intake, the use of a single 24-hour dietary intake recall limits the analysis of macronutrients and micro-nutrients intake (e.g., sodium) that may have impacted cardiometabolic risk factors but does provide an overall view of dietary intake. Accelerometry-based measurements were limited by low wear-time in this cohort. Finally, the lack of a control group is a limitation for this study. A large randomized controlled trial design would be better able to confirm that the cardiometabolic improvements and behavioural changes observed were attributable to the intervention. We plan to conduct such a study.

In summary, the MoMM pilot study indicates that, among women who enroll and participate, a group-based multimodal intervention with childcare support may be effective in lowering diabetes and vascular disease risk in women within 5 years of a GDM pregnancy. Building on such an approach has the potential to reduce diabetes risk and vascular complications- in mothers, fathers, and children.

Anne-Sophie Brazeau and Aaron Leong are Shared 1^st author.