Australian researchers have questioned the accuracy of reported rates of type 2 diabetes in some Pacific Island nations after discovering incorrect glucose cut-off points were used in the Word Health Organisation’s STEP surveys.

In a joint report published in the Journal of Diabetes the researchers from the UNSW and the Baker IDI Heart & Diabetes Institute say the errors doubled the actual rate of type 2 diabetes and is likely to have global ramifications.

The errors occurred due to modern glucose meters producing measurements in plasma-equivalent concentrations from whole blood samples, the authors explained.

In the STEPS surveys the incorrect whole blood glucose cut-off (≥6.1mmol/L) was applied instead of the correct plasma glucose cut-off (≥7.0mmol/L).

This meant that in Fiji it appeared as though diabetes prevalence had increased from 16% in 2002 to 29.6% in 2011. The research team found the actual prevalence had decreased by 0.4%.

In Samoa it appeared that there was a 24.3% increase in diabetes, from 21.5% in 2002 to 45.8% in 2013. The actual increase was only 2.8%.

In Tonga the prevalence surveys suggested a 12% increase, from 22.4% in 2004 to 34.4% in 2012. The actual rate from corrected data decreased by 3.4% to 19.0%.

Study co-author UNSW Professor Richard Taylor, from the School of Public Health and Community Medicine, said the integrity of the WHO surveys were crucial because they give an indication of the success, or otherwise, of health intervention programs.

“These WHO surveys are widely cited and used as a reference by many organisations and individuals to make public health decisions which then influence future health program planning in the Pacific region,” Professor Taylor said.

Study co-author Professor Paul Zimmet, from the Baker IDI Heart & Diabetes Institute and Monash University, called for improved education and training in how to interpret the results for health workers who use the point of care blood glucose meters.

“A recalculation of results may also be required for other country surveys where levels and apparent changes in diabetes prevalence appear suspect,” Professor Zimmet said.