As researchers involved in developing the CANRISK risk scoring questionnaire, we feel compelled to respond to the proposed guidelines.1 Unlike simple lists of diabetes risk factors, risk scoring questionnaires (i.e., Framingham for cardiovascular disease) can help physicians quantify the patient’s personal risk of diabetes based on statistical coefficients derived from scientific cohort studies. An organized triaged approach involving CANRISK for initial risk assessment would likely increase both the efficiency and effectiveness of diabetes screening efforts. The authors do recommend a sensible triaged approach to diabetes screening using risk scoring questionnaires.
However, we disagree that Finland’s FINDRISC should be preferred over CANRISK as the risk-scoring tool of choice for Canada. Last year’s peer-reviewed validation article2 showed that CANRISK is significantly more accurate. The ROC analysis found an area under curve for CANRISK of 0.75 compared with 0.66 for FINDRISC — where 0.5 indicates no discrimination, like a random coin toss. This reflects that CANRISK includes certain key variables that were excluded from FINDRISC such as ethnicity, gender and markers of previous gestational diabetes. CANRISK was tailored to address Canada’s multi-ethnic population. The Task Force is confusing FINDRISC’s broader international usage with validation in the intended screening target groups (e.g., First Nations).
The Task Force also recommends A1C as the first-line blood test for diagnosing diabetes and it argues that the 3 possible diagnostic tests (A1C, FPG and 2hrGTT) are essentially interchangeable. Although the equivalence of prognosis is debatable, there are clear differences regarding case detection. Each test has its own set of unique diabetes cases. The CANRISK study2 found that FPG would miss over 50% of diabetes otherwise detected by 2hrGTT. Similarly, A1C would detect only about 40% of diabetes detected by either FPG and/or 2hrGTT. These discordant case detection findings have also been found in other international studies.3,4
Because A1C does not have a prediabetes target group (unlike FPG or 2hrGTT), these guidelines also amount to an important missed opportunity for diabetes prevention. Major studies5 have shown that lifestyle interventions among prediabetics can reduce diabetes by more than 50%. The guidelines1 therefore effectively undermine any new provincial diabetes prevention programs targeting prediabetes through lifestyle intervention. This “disease-focused” A1C screening is analogous to screening for lung cancer using chest x-rays, while ignoring cigarette smoking.
We believe the Task Force should consult other organizations, and revise its current guidelines toward a more effective integrated approach to diabetes screening and prevention, involving CANRISK and organized teams of health professionals.