Robinson and Sohal1 criticize our guideline2 for not giving sufficient weight to the CANRISK instrument, which he and his colleagues developed, whereas Stone3 criticizes our guideline for recommending that CANRISK be used. Finally, Ball4 and colleagues prefer an alternative instrument (FINDRISC) to CANRISK, and suggest that our guideline (which clearly states that it applies to only adults) should be “challenged” — in part because it may not adequately reflect the high burden of risk for diabetes in Aboriginal children.
First, our guideline states that both FINDRISC and CANRISK are acceptable risk prediction instruments. Like Robinson and Sohal,1 we believe that using validated instruments to identify people who should be screened is superior to risk stratification based primarily on age. Not all older people are at high risk of diabetes, and conversely not all young people are at low risk. Although it is understandable that Robinson is partial to the instrument that he helped to design, we stand by our assessment of the relative merits and limitations of CANRISK, which we think will be an important aid to practitioners who are trying to decide which patients should be screened for diabetes.
Second, we agree with Ball and colleagues4 that the burden of diabetes is high in certain groups of children and young adults. We believe that validated instruments are preferable to “clinical experience” for identifying adults at high risk of diabetes. However, we agree that available instruments are imperfect. As stated in our guideline, the weak recommendation against routine screening in adults at low or moderate risk of diabetes means that screening is valid and appropriate in those who place a high value on the uncertain benefit of screening, including those from high prevalence populations. We hope that this reassures Ball and colleagues that our guideline does not preclude the use of clinical judgment.
Third, we agree with Robinson and colleagues1 that all available screening tests for diabetes have both advantages and disadvantages, and that A1C is more expensive than fasting glucose measurements. As stated in the guideline, we placed a higher value on convenience for patients than on cost when selecting A1C as the preferred screening test.
Fourth, Stone3 states that 4 questions in CANRISK require knowledge of blood glucose levels, and indicates that it is difficult to understand how this information would be available in the absence of screening. His comment may reflect a failure to differentiate between screening (measuring A1C or blood glucose in asymptomatic people) and case-finding (e.g., routine use of oral glucose tolerance tests in pregnant women) or clinically indicated testing (e.g., measuring blood glucose in an obese adult with blurred vision and polydipsia). Our guideline refers to screening only. So, that asymptomatic adults could have a history of abnormal A1C or fasting glucose values (previously obtained for clinical indications or for case-finding) is clear, even if they have never been screened for diabetes. Stone3 also states that the Task Force has “failed to acknowledge the inherent conflict of interest between guideline developers and those who pay for their time to develop guidelines.” We can reassure Stone that Task Force members are volunteers who are not paid for their service.