How to improve mobile apps for diabetes

Diabetes poses significant public health concerns, affecting approximately 537 million adults between 20 and 79 years of age, with projections exceeding 1.31 billion by 2050 worldwide. Its economic burden was estimated at US $1.31 trillion and 1.8% of global gross domestic product (GDP) in 2015 and is projected to substantially rise to $2.2 trillion (2.2% of global GDP) by 2030. 

Effective self-management is essential in handling noncommunicable diseases like diabetes, yet its implementation encounters significant challenges. First, it entails actively engaging in diverse behavioral activities throughout the lifelong course of the illness. These include consistently adapting diet, maintaining a regular exercise schedule, monitoring evolving health status, and following complex medication regimens. Furthermore, diabetes management is largely independent of direct medical oversight, requiring patients to rely on their own resources to withstand the disease burden, and not easy to standardize, varying on a day-to-day basis in response to blood glucose volatility. 

Advances in digital technologies provide new opportunities to modify health behaviors and enhance self-management, to the extent that the digital diabetes care market was projected to be worth US $1.5 billion in 2024. 

While experimental evidence consistently shows that mobile apps contribute to improved glycaemic control, little is known about 

• the temporal contours of this relationship; 
• the specific components of app interventions associated with improved health outcomes;  
• their connection with behavior change theories. 

To contribute to this discussion, we conducted a meta-analysis, including a thorough review of the extant literature. We searched on PubMed, Embase, the Cochrane Central Register of Controlled Trials, and Scopus for randomized controlled trials published from 2011 to November 2023 that employed app-based interventions to support diabetes self-management and assessed their impact on glycaemic control.  

Glycaemic control was measured by glycated hemoglobin (HbA1c) levels. According to Diabetes UK, glycated hemoglobin “is your average blood glucose (sugar) levels for the last two to three months. If you have diabetes, an ideal HbA1c level is 48mmol/mol (6.5%) or below. If you're at risk of developing type 2 diabetes, your target HbA1c level should be below 42mmol/mol (6%).” 

Starting with 7,054 articles, 58 articles from 57 studies were eligible for analysis, for a total of 7,365 participants, with an average age of 50.5 and an average glycated hemoglobin equal to 8.5%. 

Overall, our findings confirmed that apps are effective in improving glycaemic control in patients with diabetes, with an average reduction of 0.36 percentage points in glycated hemoglobin (over the 0.30 threshold that defines an intervention effective in reducing diabetes complications). 

• Our data show a reversed bell-shaped curve, with a gradual improvement in app performance peaking around the nine-month time point, followed by a decrease in effectiveness thereafter. 
• “Self-monitoring of behavior” proved to be the most effective behavioral change technique, among those deployed by the apps. The related subgroup analysis showed that the 17 studies not adopting this technique produced an average effect of -0.15 percentage points, while the 37 employing it generated a more significant reduction (-0·46).  
• Another variable associated with a statistically significant impact on effect size was “taking medication” as a target behavior. 
• Only a restricted subset of the analyzed trials integrated behavior change theories, with even fewer offering a comprehensive account of the role of theories in the actual development of apps. 

Our analysis highlighted some room for improvement: current studies rarely include personalization features, as they adhere to traditional randomized controlled trial designs that offer standardized interventions to all participants, irrespective of individual characteristics, preferences, and life context. 

Finally, we also analyzed the apps’ development features and found that the involvement of users in their design significantly improves their medical outcomes. 

Rosanna Tarricone, Francesco Petracca, Liv Svae, Maria Cucciniello, Oriana Ciani. “Which behaviour change techniques work best for diabetes self-management mobile apps? Results from a systematic review and meta-analysis of randomised controlled trials.” DOI: https://doi.org/10.1016/j.ebiom.2024.105091.