Electronic databases (EMBASE and MEDLINE) were searched for papers published from 1990 to 2011. Two independent reviewers screened the articles using structured criteria for inclusion and performed full-text reviews. Pooled prevalence of diabetes was estimated using meta-analysis. Potential factors influencing the estimates were identified by meta-regression and used for sensitivity analyses.
Rural prevalence of diabetes of LMICs was 5.6%(95%CI = 4.6-6.6), and similar between men and women. This estimate remained robust in separate sensitivity analyses accounting for study quality, level of heterogeneity, age, and sex. In a multivariate meta-regression analysis, pooled prevalence varied by study year and region. Diabetes prevalence increased over time, from 1.8%(1.0-2.6) in 1985-1989, 5.0%(3.8-6.3) in 1990-1994, 5.2%(4.1-6.2) in 1995-1999, 6.4%(5.1-7.7) in 2000-2004, and to 8.6%(6.4-10.7) for 2005-2010 (p = 0.001 for secular trend).
Prevalence of diabetes in rural parts of LMICs has risen dramatically. As 55%of LMIC populations live in rural areas, this trend has enormous implications for the global burden of diabetes.