Artefactual inflation of type 2 diabetes prevalence in WHO STEP surveys

Objectives In view of erroneous type 2 diabetes prevalence reported in 3 small Pacific Island countries, a study was conducted to investigate whether this error occurred in other countries which have conducted WHO STEPS surveys associated with glucose thresholds for point-of-care (POC) measuring devices calibrated to plasma. Methods Published STEPS surveys conducted over 2001-2017 were obtained. For each survey, information was obtained on diabetes prevalence, POC glucose measuring device, blood sample tested and the fasting glucose threshold used for prevalence calculations. POC device user manuals were obtained to determine calibration. The current WHO glucose threshold for type 2 diabetes was used: >= 7.0 mmol/l for plasma glucose; >= 6.1 mmol/l for whole blood glucose. Results POC devices were used in 75 of the STEPS surveys identified to measure blood glucose. An incorrect glucose threshold was employed in 17 surveys (23%) to define diabetes. The correct threshold was applied in 20 surveys (27%). Estimates from meta-analysis and meta-regression show that diabetes prevalence in surveys using the incorrect glucose thresholds have prevalences 50% higher than surveys which use the correct glucose threshold. A definite conclusion could not be made for 38 surveys (51%) because of the absence or unclear information on the glucose metre and/or the threshold employed. Conclusion WHO STEPS surveys with likely incorrect published diabetes prevalences have been conducted across the globe, resulting in a 50% artefactual inflation. Inaccurate reporting of diabetes prevalence from widely cited STEPS surveys would have significant impacts on disease burden monitoring, policy development and resource allocation.