Community groups or mobile phone messaging to prevent and control type 2 diabetes and intermediate hyperglycaemia in Bangladesh (DMagic): a cluster-randomised controlled trial

Background: Strategies are needed to prevent and control type 2 diabetes and intermediate hyperglycaemia, which together affect roughly a third of adults in Bangladesh. We aimed to assess the effects of mHealth and community mobilisation on the prevalence of intermediate hyperglycaemia and diabetes among the general adult population in rural Bangladesh, and to assess the effect of these interventions on the incidence of type 2 diabetes among people with intermediate hyperglycaemia within the study population.

Methods: DMagic was a three-arm, cluster-randomised trial of participatory community mobilisation, mHealth mobile phone messaging, and usual care (control) in 96 villages (population roughly 125 000) in Bangladesh. Community mobilisation involved 18 monthly group meetings, led by lay facilitators, applying a participatory learning and action (PLA) cycle focused on diabetes prevention and control. mHealth involved twice-weekly voice messages over 14 months promoting behaviour change to reduce diabetes risk. The primary outcomes were the combined prevalence of type 2 diabetes and intermediate hyperglycaemia in the overall population at the end of the intervention implementation period, and 2-year cumulative incidence of type 2 diabetes in a cohort with intermediate hyperglycaemia at baseline. Primary outcomes were assessed through fasting blood glucose concentrations and 2-h oral glucose tolerance tests among a cross-section of adults aged 30 years and older and a cohort of individuals identified with intermediate hyperglycaemia. Prevalence findings are based on a cross-sectional survey at the end of the study; incidence findings are based on 2-year follow-up survey of a cohort of individuals identified with intermediate hyperglycaemia through a cross-sectional survey at baseline. We also assessed the cost-effectiveness of the interventions. This trial is registered with the ISRCTN registry, number ISRCTN41083256, and is completed.

Findings: The study took place between June 27, 2015, and June 28, 2018, with the PLA intervention running in 32 villages from June, 2016, to December, 2017, and the mHealth intervention running in 32 villages from Oct 21, 2016, to Dec 24, 2017. End-of study prevalence was assessed in 11 454 individuals and incidence in 2100 individuals. There was a large reduction in the combined prevalence of type 2 diabetes and intermediate hyperglycaemia in the PLA group compared with the control group at the end of the study (adjusted [for stratification, clustering, and wealth] odds ratio [aOR] 0·36 [0·27-0·48]), with an absolute reduction of 20·7% (95% CI 14·6-26·7). Among 2470 adults with intermediate hyperglycaemia at baseline, 2100 (85%) were followed-up at 2 years. The 2-year cumulative incidence of diabetes in this cohort was significantly lower in the PLA group compared with control (aOR 0·39, 0·24-0·65), representing an absolute incidence reduction of 8·7% (3·5-14·0). There was no evidence of effect of mHealth on combined prevalence of intermediate hyperglycaemia and diabetes (aOR 0·93, 0·74-1·16) or the incidence of diabetes (1·02, 0·73-1·43). The incremental cost-effectiveness ratios for PLA were INT$316 per case of intermediate hyperglycaemia or type 2 diabetes prevented and $6518 per case of type 2 diabetes prevented among individuals with intermediate hyperglycaemia.

Interpretation: Our data provide strong evidence to support the use of community mobilisation based on PLA to prevent type 2 diabetes in this rural Bangladeshi population. Despite raising knowledge and awareness of diabetes, the mHealth intervention did not change disease outcomes in our population. Replication studies in other populations should be a priority.

Funding: UK Medical Research Council.