Implementation and impact of mHealth in diabetes management in Africa: Systematic review

Methodology

This systematic review was prospectively registered with PROSPERO (CRD42021218674) and conducted following PRISMA guidelines. The review examined randomized controlled trials, non-randomized controlled trials, quasi-randomized controlled trials, and observational studies conducted in Africa between January 2010 and October 2020 that deployed mHealth strategies for diabetes management.

Comprehensive database searches were conducted in MEDLINE (PubMed), Cochrane Register of Controlled Trials (CENTRAL), Embase (Scopus), CINAHL, and the Pan African Clinical Trials Registry (PACTR). Medical Subject Headings (MeSH) terms were used to identify studies related to mHealth, diabetes mellitus, and African regions/countries. Search terms included implementation-related terms (implement, execute, administer, organize, fulfill, perform, utilize, manage, operate, act, realize, effect, impact) combined with mHealth/eHealth terminology, diabetes terminology, and comprehensive African country filters. Hand searching and snowballing supplemented database searches. The Rayyan application was used for article screening and duplicate removal.

Two reviewers independently screened titles and abstracts, retrieved full-text articles, and assessed eligibility, with disagreements resolved by a third independent reviewer through discussion. Inclusion criteria required studies using mHealth interventions (individual behavior change, chronic disease self-management, clinic appointment reminders, or clinical diagnostic aids) in diabetes patients in Africa with no age or sex restrictions. Studies used standard WHO and International Diabetes Federation diagnostic criteria for diabetes and treatment targets. Exclusion criteria included studies testing mHealth acceptability among health professionals only and duplicate publications.

The primary outcome was change in HbA1c or fasting plasma glucose comparing intervention and control groups. Secondary outcomes included reductions in diabetes complications, though most studies did not address this. Risk of bias was assessed using the Revised Cochrane Risk of Bias tool, evaluating randomization process, deviation from intended intervention, missing outcome data, measurement of outcomes, and selective reporting.

Data extraction used standardized Microsoft Excel forms capturing author names, publication year, country, sample sizes, treatments, mHealth interventions with delivery strategies and functionalities, study type and duration, methodology, outcomes and measurement times, baseline and end-of-study glycemic control, percentage HbA1c change, hypoglycemic/hyperglycemic events, quality of life, BMI changes, harms, costs, and acceptability indicators.

Narrative synthesis described interventions, mHealth categories, mobile functionalities, sample characteristics, and outcomes. Meta-analysis was performed on mean difference in HbA1c calculated from pre- and post-intervention standard deviations using Stata 18 metan command with standard methods, summarized as forest plots. Heterogeneity was assessed using I² statistics with interpretation: I²≥75% as high heterogeneity, I² 50-75% as moderate heterogeneity, and I²<50% as little to no heterogeneity.

Interventions

The systematic review examined multiple mHealth intervention modalities for type 2 diabetes management in Africa. Four of six included studies investigated SMS text messaging as the primary delivery strategy, sending educational content, reminders, and self-management prompts to patients. Messages typically included diabetes education, medication reminders, lifestyle modification encouragement, and appointment reminders. One study (Asante et al.) used nurse-led mobile phone calls providing personalized counseling and support. One study (Takenga et al.) employed a mobile application that combined remote monitoring with direct feedback to patients and providers.

All interventions shared the common goal of modifying patient behavior and attitudes toward diabetes treatment. The mHealth strategies aimed to improve diabetes self-management by reaching patients remotely, providing ongoing education and support between clinic visits, reinforcing adherence to medications and lifestyle recommendations, and maintaining engagement with the healthcare system. Intervention intensity varied, with SMS-based interventions typically delivering messages at scheduled intervals (daily to weekly), phone call interventions involving regular scheduled contacts, and the mobile app providing continuous access to diabetes management tools.

The interventions were compared to usual care or conventional diabetes management without mHealth components. All studies were hospital-based with follow-up periods ranging from 8 weeks (Takenga et al.) to 52 weeks (Farmer et al.). The mHealth interventions were provided in addition to standard diabetes care including medications (oral hypoglycemic agents and/or insulin therapy varied across studies) and routine clinic follow-up. Implementation contexts varied across Egypt, Ghana, Senegal, South Africa, and Democratic Republic of Congo, representing diverse African healthcare settings with varying resource availability and mobile phone infrastructure.

Key Findings

The meta-analysis of five studies (n=3,112 participants) demonstrated a statistically significant reduction in HbA1c levels following mHealth interventions, with an overall weighted mean difference of 0.992% (95% CI: 0.48-1.50, z=3.822, p<0.001). This represents a clinically meaningful improvement in glycemic control, as approximately 1% HbA1c reduction is associated with reduced risk of diabetes complications. Individual studies reported HbA1c percentage reductions ranging from 3.6% (Wargny et al.) to 20.53% (Takenga et al.), with three studies (Takenga, Wargny, Asante) achieving statistical significance independently while three others (Haitham, Farmer, and the study by Owolabi which measured random glucose) did not reach individual significance.

However, considerable statistical heterogeneity was present in the meta-analysis (I²=63.9%, p=0.026), indicating substantial variability in intervention effectiveness across studies. This heterogeneity appears related to differences in study characteristics, most notably sample size (ranging from 40 to 1,186 participants) and intervention duration (8 to 52 weeks). The percentage weight in the meta-analysis was predominantly attributed to the largest studies (Farmer et al. and Wargny et al.), while smaller studies showed larger individual effect sizes, raising concerns about potential publication bias. All six studies demonstrated high acceptability of mHealth interventions through administered questionnaires, with response and acceptance rates ranging from 74.4% to 100%. None of the included studies addressed secondary outcomes related to preventing or detecting diabetes complications early, representing a significant gap in the evidence base for long-term benefits of mHealth interventions in Africa.

Comparison with other Studies

The findings from this African-focused systematic review align well with global evidence on mHealth effectiveness for diabetes management. A 2011 meta-analysis by Liang et al. involving 1,657 individuals with type 1 or type 2 diabetes using SMS-based self-monitoring and education showed a 0.5% reduction in HbA1c over six months, which is comparable to the 0.992% reduction found in this African review, though the African studies showed greater variability. Similarly, a 2017 meta-analysis by Bonoto et al. of 13 studies reported a mean HbA1c reduction of 0.44% in mHealth intervention groups compared to controls, again consistent with the direction and magnitude of effect observed in Africa.

More recent global systematic reviews reinforce these findings. Eberle et al. (2021) in a scoping review found that diabetes-specific mHealth mobile applications significantly improved glycemic control and strengthened self-care perception in individuals with type 1 and type 2 diabetes. Mao et al. (2020) conducted a global systematic review of mHealth impact on diabetes and hypertension management, including 51 studies of which only two were from Africa (one being Haitham et al., included in this review). They demonstrated positive impacts across countries with different economic development levels and noted better outcomes when mHealth was combined with human intelligence. Stevens et al. (2022) examining digital health technologies for diabetes patients in 25 RCTs (3,360 patients) found overall greater improvement in HbA1c with mHealth compared to usual care, though notably none of their included studies were conducted in Africa.

The current African review differs from global reviews in several important aspects. First, the evidence base in Africa remains much smaller, with only six eligible studies identified compared to dozens in global reviews, highlighting the persistent research gap in African settings. Second, the African studies predominantly used older mHealth technologies (SMS and phone calls) rather than the smartphone applications with advanced features that dominate recent global literature, reflecting differences in mobile technology infrastructure and penetration. Third, the heterogeneity in the African review (I²=63.9%) was higher than typically reported in global meta-analyses, possibly reflecting greater variability in healthcare contexts, mobile infrastructure, and implementation approaches across African countries. Finally, the absence of studies examining mHealth's impact on diabetes complication prevention in Africa contrasts with emerging global evidence on long-term outcomes, representing a critical gap for future African research.

Journal Reference

Dike FO, Mutabazi JC, Musa E, Ubani BC, Isa AS, Ezeude CM, Iheonye H, Ainavi II. Implementation and impact of mhealth in the management of diabetes mellitus in Africa: A systematic review and meta-analysis. PLOS Digit Health. 2025;4(4):e0000776. doi:10.1371/journal.pdig.0000776