OBJECTIVES: To investigate the diagnostic accuracy of random blood glucose (RBG) on good glycaemic control among patients with diabetes mellitus (DM) in a rural African setting.

METHODS: Cross-sectional study at St. Francis' Hospital in eastern Zambia. RBG and HbA1c were measured during one clinical review only. Other information obtained was age, sex, body mass index, waist circumference, blood pressure, urine albumin-creatinine ratio, duration since diagnosis and medication.

RESULTS: One hundred and one patients with DM (type 1 DM = 23, type 2 DM = 78) were included. Spearman's rank correlation coefficient revealed a significant correlation between RBG and HbA1c among the patients with type 2 DM (r = 0.73, P < 0.001) but not patients with type 1 DM (r = 0.17, P = 0.44). Furthermore, in a multivariate linear regression model (R(2) = 0.71) RBG (per mmol/l increment) (B = 0.28, 95% CI:0.24-0.32, P < 0.001) was significantly associated with HbA1c among the patients with type 2 DM. Based on ROC analysis (AUC = 0.80, SE = 0.05), RBG ≤7.5 mmol/l was determined as the optimal cut-off value for good glycaemic control (HbA1c <7.0% [53 mmol/mol]) among patients with type 2 DM (sensitivity = 76.7%; specificity = 70.8%; positive predictive value = 62.2%; negative predictive value = 82.9%).

CONCLUSIONS: Random blood glucose could possibly be used to assess glycaemic control among patients with type 2 DM in rural settings of sub-Saharan Africa.