Task-sharing interventions for improving control of diabetes in low-income and middle-income countries: a systematic review and meta-analysis

BACKGROUND: Task-sharing interventions using non-physician health-care workers might be a potential diabetes management strategy in health systems that are constrained by physician shortages, such as those in low-income and middle-income countries (LMICs). METHODS: We did a systematic review and meta-analysis of task-sharing intervention strategies for managing type 2 diabetes in LMICs. We searched PubMed, Embase, and CINAHL from database inception to Sept 25, 2019, for studies that were randomised control trials or cluster randomised trials with task-shifted or task-shared interventions delivered to adults (≥18 years) by non-physician health workers versus usual care, done in LMICs with glycated haemoglobin (HbA(1c)) or fasting blood sugar (FBS) as outcome measures. The methodological quality of included studies was assessed using the Cochrane risk of bias tool. Random-effects model meta-analysis was used to estimate the population average pooled mean difference for HbA(1c) and FBS with 95% CIs. Our study protocol was registered in the PROSPERO database (CRD42018081015). FINDINGS: We found 4213 studies from the literature search, of which 46 (1·1%) were eligible for the narrative synthesis, including a total of 16 973 participants. 16 of these studies were excluded from the meta-analysis due to high risk of bias. 24 studies with a total of 5345 participants were included in the meta-analysis of HbA(1c) and 18 studies with a total of 3287 participants for FBS. Interventions led to an average reduction in HbA(1c) when tasks were delivered by nurses (averaged pooled mean difference −0·54% [95% CI −0·89 to −0·18]; I(2)=80%) and pharmacists (−0·91% [–1·15 to −0·68]; I(2)=58%), but not when they were delivered by dietitians (−0·50% [–1·10 to 0·09]; I(2)=54%) or community health workers (0·05% [0·03 to 0·07]; I(2)=0%). A reduction in average FBS was also observed when interventions were delivered by pharmacists (average pooled mean difference −36·26 mg/dL [–52·60 to −19·92]; I(2)=78%) but not nurses (−7·46 mg/dL [–18·44 to 3·52]; I(2)=79%) or community health workers (−5·41 [–12·74 to 1·92]; I(2)=71%). Only one study reported on FBS when tasks were delivered by dietitians, with a mean difference of −35·00 mg/dL (−65·96 to −4·04). INTERPRETATION: Task sharing interventions with non-physician healthcare workers show moderate effectiveness in diabetes management in LMIC settings. Although relatively high heterogeneity limits the interpretation of the overall findings, interventions led by pharmacists and nurses in LMICs with relatively high physician density are effective strategies in the management of diabetes. FUNDING: Wellcome Trust–Department of Biotechnology India Alliance.