Introduction

Gestational diabetes mellitus (GDM) is a common complication of pregnancy, with a rising incidence, affecting around 1 in 6 births globally, with prevalence varying across different regions and populations [1]. GDM impacts maternal and offspring health both in short- and long-term [2], the latter including increased risk for maternal type 2 diabetes mellitus (T2DM) and cardiovascular disease (CVD) [3]. Short-term consequences include increased risk for preeclampsia, which in itself is an established risk factor for CVD [4]. Women who have had GDM in their pregnancy have an increased risk of type 2 diabetes and cardiovascular disease later in life compared to those with normoglycemic pregnancies [3]. Through meta-analyses, the size of this risk association has been estimated to be a relative risk of almost 10 for T2DM and almost 2 for cardiovascular disease [5,6,7]. Although the relatively higher progression rate to T2DM in these women partly accounts for the CVD risk increase, meta-analyses have shown that GDM per se carries a residual risk [5, 7].

Epidemiological studies indicate that a healthy diet and increased physical activity can reduce the risk of developing T2DM [8, 9]. The period after a pregnancy with GDM has been referred to as window of opportunity for prophylactic interventions that can reduce the risk of T2DM and related comorbidities [10]; however, adherence to recommended postpartum screening for DM2 appears to be low [11, 12]. eHealth (electronic health) is defined as the use of information and communication technology for health. It is emerging as a tool with the potential of transforming facets of our health care systems, including perinatal care, but their practical utility and advantages over standard care remains to be determined [13, 14]. Our aim was to systematically review the existing literature on follow-up regarding cardiovascular disease after gestational diabetes, the utility of eHealth technology for this purpose, and to identify research gaps.

Methods

Protocol and registration

Following the Joanna Briggs methodology [15], a review protocol was developed and published at Open Science Framework (osf.io/p5hw6) before initiating the literature search [16]. There were no major deviances from the published review protocol. We used the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) Checklist; see Additional file 1.

Literature search and eligibility criteria

A literature search was performed by one of the authors (BSF) with the help of a librarian at the University of Oslo Library in May 2022. Four databases were searched: Ovid MEDLINE, Embase, Maternity and Infant Care, and Cochrane Database of Systematic Reviews. Some database-specific adaptions were made to the search strategy for the different databases. Detailed information on the literature search is provided in Additional file 2. We included original research articles and systematic reviews with a population of nulli- and multiparous women with GDM in a previous pregnancy, where the concept involved follow-up regarding long-term cardiovascular risk after such a pregnancy, as well as the use of eHealth technologies as a tool in such follow-up. The context was health care settings in which women receive care after a GDM pregnancy from skilled health care workers. Additionally, we included guidelines from the International Federation of Gynecology and Obstetrics (FIGO) as well as national guidelines from the UK, Canada, Australia/New Zealand, Sweden, Denmark, and Norway. These guidelines were chosen due to having a comparable population and system of ante- and postnatal care as the Norwegian health care system, a rationale that is consistent with other reviews [17]. We limited the search to publications in languages mastered to fluency by the review authors (English, German, Norwegian, Swedish, or Danish), without any date of publication restriction.

Screening, data charting process, and synthesis of results

The results were downloaded to the EndNote reference management software (version 20; Clarivate Analytics, USA) and transferred to Covidence, a web-based collaboration software platform that streamlines the production of systematic and other literature reviews (Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia, available at www.covidence.org).

Titles and abstracts were reviewed by two of the authors (BSF and MS) for relevance. All articles deemed to be relevant or of uncertain relevance underwent full-text review. In both stages of the process, if consensus was not reached, a third reviewer (ASDP) also reviewed and cast the deciding vote. The reference lists of the selected publications were manually searched for additional relevant articles.

A data charting form was developed (by all the authors) and completed for each study by BSF and MS independently. Data retrieved included information such as country of origin, methods, population, intervention, and outcomes. Certain information was extracted for guidelines that was not extracted for the primary studies and reviews and vice versa. The data charting form is provided in Additional file 3. The final version of the data charting form was expanded compared to the original one published with the protocol, but this was done in accordance with the planned testing and alignment of the data charting form early in the process. As outlined in the protocol, prior to starting the data charting stage of the review, two of the researchers (BSF and MS) tested and validated the form by independently screening three articles, comparing the results and adjusted the form to incorporate relevant findings. The final version of the data charting form was uploaded to the Covidence software platform, where both researchers doing the data charting (BSF and MS) used this independent of each other. In cases where the software flagged discrepancies in the data charted, the two researchers assessed the conflict and reached a consensus. The final version of the data charting form for each article or review then provided the basis when one researcher (BSF) created the first draft of the different tables in summarizing the results, which were then assessed by all the authors. In accordance with the JBI framework, no formal quality assessment of the scientific articles was performed; inclusion depended solely on the eligibility criteria.

Results

The search of the databases identified 2772 references. A further nine studies and systematic reviews were added after manual search of the reference list of other included articles. Ten guidelines also needed to be imported manually to obtain the full version in the original language. We removed 22 duplicates. Hence, in total, 2769 articles were screened for titles and abstract. This process excluded a further 2672 due to lack of relevance to the topic of interest. In total, 97 papers then underwent full-text review, in which 62 were excluded. Finally, 15 articles and 12 systematic reviews were included in the review. Data from eight gestational diabetes guidelines are presented in Additional file 4. A PRISMA flowchart of the process is shown in Fig. 1. An overview of the various definitions of GDM used in the different studies can be found in Table 1.

Fig. 1
figure 1

PRISMA flowchart

Full size image
Table 1 The criteria for GDM used in the various studies
Full size table

No studies or systematic reviews were identified assessing long-term CVD risk per se; rather, the outcomes assessed were either incidence of T2DM or other markers of impaired glucose homeostasis or various CVD risk factors such weight and physical activity.

Trials

Follow-up studies regarding cardiovascular risk

Eleven trials assessing follow-up regarding cardiovascular risk (excluding those with a primarily eHealth technology-based intervention, see the “The use of eHealth technologies” section) were deemed to meet the criteria of the review [24,25,26,27,28,29,30,31,32,33,34], of which seven were RCTs [26,27,28, 30,31,32,33], two cluster RCTs [24, 25], one randomized clinical with two interventions and no control group [34], and one an interventional cohort trial [29]. Details on the different trials are shown in Table 2. In four of the included trials, the primary outcomes were partly or exclusively related to postpartum weight change [24, 25, 27, 28], in three studies incidence of T2DM [26, 30, 32], and three studies reported other measures of glycemic status [28, 29, 31, 33, 34]. Glucose-related outcomes were additionally included among the secondary outcomes in two of the articles [25, 27]. The interventions were all different types of lifestyle interventions, focusing on diet and/or physical activity, delivered as either individual or group sessions, with different tools utilized as part of the follow-up (e.g., reminder systems using telephone or e-mail). Follow-up varied between six and 36 months for all the studies except Aroda et al., where follow-up was 10 years [32]. These authors found that compared to the placebo/standard care group the intensive lifestyle intervention reduced progression to diabetes by 35%. It should be noted that the setting of this trial was somewhat differing from the other studies, given that mean time since index GDM pregnancy was 12 years at the time of recruitment, a longer interval than in any of the other studies. In a smaller study, involving 180 participants, with two years follow-up, Zilberman-Kravits and co-workers found that a culturally tailored lifestyle intervention significantly reduced insulin resistance [33]. None of the other studies with incidence of T2DM or other glucose homeostasis-related outcomes found a significant effect. Three of the studies found significant effect of intervention on weight outcome [24, 25, 28]. Neither was there any consistent, significant effect on blood pressure, lipid profiles, or development of metabolic syndrome [24, 26, 29,30,31]

Table 2 Findings from the follow-up studies regarding cardiovascular risk
Full size table

The use of eHealth technologies

Four studies on the use of eHealth measures met the criteria for this review [35,36,37,38], three utilizing smartphones [35, 37, 38] and one with the additional use of a virtual reality (VR) headset [37]. The fourth study tested the efficacy of a pedometer program linked with a web-based module, in addition to a nutrition coaching workshop [36]. Detailed information can be found in Table 3. The two trials with strictly smartphone-based interventions [35, 38] did not show significant effect on their primary outcomes of weight [38] and proportion of participants achieving a certain level of Diabetes Prevention Program goals [35] or secondary outcomes related to glucose levels or lipid profiles. Both applications were found to be acceptable by participants, as assessed by data on actual use of the apps [35, 38] and a scoring system where the users rated the app’s quality and perceived impact [35] Examining the efficacy of a mobile VR program [37], Kim and co-workers in a study from South Korea found that it significantly improved body weight and fat, fasting blood glucose, and HbA1c compared to control group after a 12-week follow-up. This was a quasi-experimental study, where 64 women with recent diabetes were included, and the control group of 64 women were selected to the intervention group by matching for age, birth experience, type of birth, family history of T2DM, and breastfeeding status. In a small trial with 31 women and 3 months follow-up, Peacock et al. [36] demonstrated a significant difference in weight loss in pedometer program intervention group compared to the control group (− 2.5 kg (SD ± 1.4) vs 0.0 kg (SD ± 2.3), p = 0.002).

Table 3 Findings from the eHealth studies
Full size table

Reviews

Twelve reviews in total were included [39,40,41,42,43,44,45,46,47,48,49,50], of which one scoping review [39], one overview of other reviews [44], five systematic reviews [42, 45, 46, 48, 49], and five systematic reviews with meta-analyses [40, 41, 43, 47, 50]. Two of the reviews focused primarily on mHealth (mobile Health)/eHealth [39, 43], while the others mainly assessed lifestyle interventions. Of the two mHealth/eHealth reviews, the scoping review merely presented the existing literature and noted good engagement for app usage, but also a lack of studies where mHealth was the primary mode of intervention postpartum [39]. In their systematic review and meta-analysis, Halligan et al. [43] found that the results of the meta-analysis favored intervention compared to standard care for the outcomes of weight and BMI, but the results were not statistically significant. The meta-analyses of the lifestyle interventions showed somewhat mixed results. Li et al. [47] found that lifestyle interventions commenced within 3 years postpartum showed a 43% risk reduction for incidence of T2DM compared to standard care (RR 0.57, 95% CI 0.42–0.78), whereas the other reviews examining this outcome found statistically non-significant trend towards risk reduction [41, 50] or no effect for glucose related outcomes [40]. Hedeager Momsen and at al. [44] found in their overview of the reviews that lifestyle interventions appeared to decrease the incidence of diabetes postpartum and that the effects were larger the earlier after labor the intervention was implemented and the longer it lasted. The two meta-analyses for weight-related outcomes both showed small but statistically significant effects [40, 41]. In a review by Jones et al. [46], recruitment rates of participants to the various trials were assessed and found to be low even for primarily home-based interventions. In the systematic reviews overall, there were mixed results, but with most concluding that for outcomes such as weight/BMI, physical activity, and diet, lifestyle interventions may be beneficial. Details on the reviews is shown in Table 4.

Table 4 Findings from the reviews
Full size table

Discussion

Summary of evidence

The studies assessed in the present scoping review do not offer any clear evidence for how best to follow-up women after gestational diabetes regarding their increased long-term risk of cardiovascular disease. Various lifestyle interventions have been tested for outcomes such as diabetes incidence, weight-related outcomes, and other cardiovascular risk factors, most offering some version of patient education combined with individual or group sessions with health care professionals. The results from both primary studies and reviews indicate that such follow-up may be beneficial but differ between the various studies and reviews for the different outcomes to such a degree that it is not possible to conclude that any of them provide a clear template for how follow-up should be carried out.

The use of eHealth is increasing in health care systems across the world. In a recent WHO guideline [13] regarding the implementation of such measures, a degree of caution was advised, emphasizing the importance of rigorously evaluating their utility, to ensure that they do not divert resources from non-digital interventions if they are not superior. The eHealth interventions assessed in this review have not shown any clear and consistent advantages compared to standard care. However, it is possible that the lack of statistically significant results in the smartphone app trials [35, 38] was at least partly related to a relatively short follow-up period of 4 and 6 months, respectively. On the other hand, Kim et al. [37] found a statistically significant result after only a 12-week follow-up, but given the quasi-experimental design, a degree of caution is necessary when interpreting the results.

Another obstacle that needs to be overcome to improve the follow-up of this group of women after GDM is the low rate of adherence to existing follow-up recommendations such as postpartum glucose tolerance tests, which is attended by less than one in five [52]. Use of proactive reminder systems and mobile health technology have been suggested as possible remedies for this, and the latter highlighted as an area that warrants further research [14, 53].

Although it is established that women with a GDM pregnancy have a significantly increased risk of CVD later in life, even when adjusting for the risk of T2DM [5, 7], no studies were identified that assessed strategies for reducing overall cardiovascular risk. All the studies were focusing on either persistent hyperglycemia or other individual risk factors of CVD such as weight, diet, and physical activity. Taking into account that the risk of CVD to a certain degree is independent of the considerably increased incidence of T2DM in women with a previous GDM pregnancy compared to those without GDM, this is an area that warrants further research.

Given that women on average are relatively young and healthy at the time of reproduction, it might be that the next contact for a woman with prior GDM and a normal HbA1c or OGTT postpartum—if tested—might be the first trimester of a next pregnancy or even years later if she does not have any more children. This makes the pregnancy and peripartum period a missed window of opportunity for optimizing any modifiable cardiovascular risk factor. As presented in Additional file 4, guidelines generally suggest some type of postnatal testing for persistent hyperglycemia; however, few explicitly address the risk of CVD apart from T2DM, and the Norwegian guideline stands alone in offering a clear template for how this could be followed up. The Norwegian guidelines link obstetric outcomes with a follow-up by general practitioners in the public health system, which may assist in bridging the gap in the health follow-up of postpartum women. In patients with previous hypertensive disorders of pregnancy, guidelines such as NICE [54] and ACOG [55] recommend that women are followed by their primary care provider to manage risk factors for cardiovascular disease. The Norwegian guidelines [56] for hypertensive disorders of pregnancy offer a clear algorithm suggesting how this could be carried out from delivery to middle age and beyond, a flow-chart which has since been included also in the guidelines for gestational diabetes [57]. As both spectrums of obstetric disease confer an increased risk of CVD later in life, similar recommendations for follow-up might be a sensible approach. As the development of CVD is an insidious process developing over years [58], studies with a longer follow-up would be a welcome addition to the literature.

Strengths and limitations

To the best of our knowledge, this is the most comprehensive review that has been performed for this topic, encompassing both original research studies and systematic reviews and including both more conventional lifestyle interventions and also with a separate assessment of the utility of eHealth technologies The scoping review methodology does not entail quality assessment; hence, our review has not analyzed the quality of the included studies. Our review was not able to generate evidence that supports any specific follow-up regime to lower the cardiovascular risk in women with previous gestational diabetes, whether using conventional methods or eHealth measures. Some limitations should be noted. First of all, the review protocol was not peer reviewed. Another limitation to our study was that the search did not yield the full, original language version of any of the included guidelines; hence, these had to be inserted manually. We also acknowledge that more databases could have been searched, including those cataloging grey literature. Another limitation is the language restrictions. In itself, the inability to review articles written in other languages than the ones the authors of this paper are fluent in could be a source of bias. In retrospect, we also acknowledge that it would be preferable to have included language among the eligibility criteria rather than as restrictions in the search strategy.

Conclusions

Our scoping review has shown that although GDM is an established risk factor for CVD later in life, it is not possible to ascertain from the existing literature how women with a history of gestational diabetes mellitus should be followed up in this regard. The studies and reviews assessed in this scoping review suggest that lifestyle interventions may be beneficial for certain individual risk factors, such as weight-related outcomes and risk of T2DM, but no studies assessing the overall long-term risk has been performed. eHealth technology is not an established feature in the follow-up of women after GDM, and although such measures appear to be acceptable to participants, they have yet to prove their utility for improving follow-up and lowering cardiovascular risk.

There is need for further research on how best to follow-up concerning long-term risk of overall CVD after a GDM pregnancy. Studies with longer follow-up assessing how to utilize eHealth technologies would also be a welcome addition to the literature. The increased risk of CVD and parallel recommendations for hypertensive disorders of pregnancy suggest that similar approach regarding follow-up for optimizing cardiovascular risk factors could be reasonable. However, the existing literature does not offer any clear advice on how this should be carried out.