Aim  To determine the utility of finger-prick point-of-care testing (POCT) of blood glucose for the detection of dysglycaemia.

Methods  A fasting POCT and an oral glucose tolerance test (OGTT) with laboratory assays were performed as part of the baseline screening for 5309 participants enrolled in the Te Wai o Rona Diabetes Prevention Strategy. Participants were aged 46 ± 19 years with no self-reported diabetes. Dysglycaemia, including diabetes, was defined using World Health Organization criteria. Agreement between the two fasting plasma glucose measurements and their screening properties (with sensitivity and specificity for cut points) were compared using receiver operator characteristic analysis.

Results  A total of 3225 participants had both capillary and venous fasting blood sampled within 30 min and then underwent OGTT. New diabetes was found in 161 participants (5.0%) and pre-diabetes in 414 [impaired glucose tolerance 299 (9.3%), impaired fasting glucose 115 (3.6%)]. The mean difference in capillary and venous measures was 0.02 mmol/l (95% confidence interval −0.04 to +0.01; limits of agreement –1.37 to 1.33 mmol/l). Capillary POCT was a poorer predictor of dysglycaemia and impaired glucose tolerance and new diabetes (area under curve 0.76 and 0.71) than venous laboratory analysis (area under curve 0.87 and 0.81 respectively). Optimal screening criteria were best at a venous glucose of 5.4 mmol/l; 77% sensitivity/specificity.

Conclusions  POCT significantly underestimated the true blood glucose at diagnostic levels for diabetes. POCT cannot be recommended as a means of screening for or diagnosing diabetes or pre-diabetes.