Performance of an NIDDM screening questionnaire based on symptoms and risk factors

To investigate to what extent a short questionnaire on symptoms and risk factors can be used to identify people at increased risk for undiagnosed NIDDM.

A general population sample of 2,364 Caucasian subjects, age 50-74 years, not known to have diabetes, completed a questionnaire on diabetes-related symptoms and risk factors. Subsequently, they underwent an oral glucose tolerance test. A backward stepwise multiple logistic regression was carried out with the absence or presence of newly detected diabetes as the dependent variable and the items from the questionnaire as the independent variables. The selected items were included in a new symptom risk questionnaire, which was evaluated in a different population sample of 786 subjects, age 45-74 years, not known to have diabetes and compared with existing questionnaires.

The newly developed symptom-risk questionnaire contains questions concerning the following items, which were independently and significantly (P < 0.05) associated with the presence of previously undiagnosed diabetes: pain during walking with need to slow down, shortness of breath when walking with people of the same age, frequent thirst, age, sex, obesity, parent or sibling with diabetes, use of antihypertensive drugs, and reluctance to use a bicycle for transportation. The 1993 American Diabetes Association questionnaire, the 1995 Herman et aL (17) questionnaire, and the newly developed symptom-risk questionnaire had sensitivities of 59, 72, and 72%; specificities of 57, 55, and 56%; positive predictive values of 5.6, 6.4, and 6.5%; and negative predictive values of 97, 98, and 98%, respectively.

The newly developed symptom-risk questionnaire has good performance characteristics, and the advantage of a variable cutoff makes it a useful screening tool for NIDDM in general practice.