Telemedicine to deliver diabetes care in low- and middle-income countries: a systematic review and meta-analysis
Jorge César Correia, Hafsa Meraj, Soo Huat Teoh, Ahmed Waqas, Maaz Ahmad, Luis Velez Lapão, Zoltan Pataky, Alain Golay
Bulletin of the World Health Organization · 2021 · DOI: 10.2471/BLT.19.250068
Countries: China, India, Iran (Islamic Republic of), Malaysia, Turkey, South Africa, Brazil, Egypt, Mexico, Mongolia, Pakistan, Cambodia, Congo, Philippines
What Was Studied
This systematic review and meta-analysis evaluated the effectiveness of telemedicine-based interventions -- including smartphone apps, SMS, telemetry, telephone follow-up, and web-based systems -- for diabetes care (type 1, type 2, and gestational) in low- and middle-income countries. The authors searched seven databases (Web of Science, PubMed, MEDLINE, Global Health Library, Cochrane Central Register of Controlled Trials, New York Academy of Medicine, and POPLINE) through July 2020 for randomized controlled trials among adults aged 18 and older. From 932 records screened, the review included 31 telemedicine interventions from 30 RCTs conducted across 14 low- and middle-income countries between 2010 and 2020, with sample sizes ranging from 60 to 3,324 participants, assessing outcomes including HbA1c, fasting blood sugar, BMI, lipid profile, treatment adherence, diabetes knowledge, and self-efficacy.
What They Found
Across 27 studies (28 interventions, n=7,703), telemedicine produced a significant reduction in HbA1c (pooled standardized mean difference, SMD: -0.38; 95% CI: -0.52 to -0.23; I²=86.70%) and a smaller but significant reduction in fasting blood sugar (SMD: -0.20; 95% CI: -0.32 to -0.08; I²=64.28%; n=5,524). Adherence to treatment improved significantly (SMD: 0.81; 95% CI: 0.19 to 1.42; I²=93.75%; n=959), as did diabetes knowledge (SMD: 0.55; 95% CI: -0.10 to 1.20; I²=92.65%; n=1,585) and self-efficacy (SMD: 1.68; 95% CI: 1.06 to 2.30; I²=97.15%; n=866) -- the largest effect observed. No significant effects were found for BMI (SMD: -0.04; 95% CI: -0.13 to 0.05; n=5,957), total cholesterol (SMD: -0.06; 95% CI: -0.16 to 0.04; n=5,381), or triglycerides (SMD: -0.02; 95% CI: -0.12 to 0.09; n=2,360). Telephone- and SMS-based interventions consistently outperformed smartphone-app and telemetry-based delivery (subgroup difference P=0.002 for BMI; P<0.01 for self-efficacy). Of the 31 interventions, 19 were judged high risk of bias and 12 low risk of bias; significant publication bias was detected for HbA1c, fasting blood sugar, diabetes knowledge, and self-efficacy. Using GRADE, the authors rated the overall certainty of evidence for the primary HbA1c outcome as very low, due to substantial heterogeneity, risk of bias, and publication bias.
What This Means for Nepal
No Nepal-specific trial is included in this review -- Nepal currently has no published RCT evidence on telemedicine for diabetes management, a gap this global synthesis exposes rather than fills. The finding that low-tech telephone- and SMS-based interventions consistently outperformed smartphone-app and telemetry-based delivery is directly relevant to Nepal, where NCDs already cause an estimated 66-71% of deaths and diabetes/raised blood glucose affects roughly 5.8-8.5% of adults (STEPS 2019; population estimates), yet NCD care remains concentrated in urban tertiary hospitals while Karnali province has 0% of its sanctioned physician and consultant posts filled and physician density collapses from about 1:850 in Kathmandu Valley to 1:150,000 in remote districts. As MoHP advances 'telemedicine for remote access' and DHIS2 strengthening as named 2026 digital-health reform priorities, and as WHO's Package of Essential Noncommunicable Disease Interventions (PEN) expands NCD services into primary care, this evidence supports piloting low-bandwidth SMS/telephone-based diabetes follow-up -- rather than defaulting to smartphone-app models that require higher digital literacy and data access -- in remote provinces such as Karnali and high-burden Terai provinces such as Madhesh. Given the review's own 'very low' GRADE certainty rating, any Nepal pilot should be paired with locally generated RCT evidence rather than assuming the -0.38 HbA1c effect transfers directly.
Contextualisation
Nepal's NCD burden is severe -- noncommunicable diseases cause an estimated 66-71% of deaths and diabetes/raised blood glucose affects roughly 5.8-8.5% of adults (STEPS 2019; population estimates) -- yet NCD care remains concentrated in urban tertiary hospitals while Karnali province has 0% of its sanctioned physician and consultant posts filled (DoHS AHR 2080/81). With 'telemedicine for remote access' and DHIS2 strengthening named as 2026 digital-health reform priorities, this review's finding that low-tech telephone- and SMS-based interventions outperformed smartphone apps and telemetry across nearly every outcome offers directly relevant guidance for a health system where physician density collapses from roughly 1:850 in Kathmandu Valley to 1:150,000 in remote districts.