Is questionnaire information valid in the study of a chronic disease such as diabetes? The Nord-Trondelag diabetes study.

Abstract

STUDY OBJECTIVE--The aim was to validate information about diabetes mellitus collected by questionnaire in a large epidemiological survey. DESIGN--Questions on diabetes diagnosis, medical treatment for diabetes, diabetes duration, and hypertension treatment were selected from the Nord-Trøndelag health survey questionnaires. One of the municipalities was selected for the validation study. SETTING--The health survey 1984-86 addressed all inhabitants > or = 20 years of age in Nord-Trøndelag county, Norway; 76,885 (90.3%) of the eligible population participated in answering the question on diabetes. PARTICIPANTS--All inhabitants in the municipality answering "yes" to the question on diabetes (n = 169) and the persons with the same sex born closest before and after each diabetic patient and answering "no" to the diabetes question (n = 338) were included. MEASUREMENTS AND MAIN RESULTS--A very thorough search was made in the medical files of the general practitioners in the municipality for corresponding information. Compared to the files, diabetes was verified in 163 out of the 169. The commonest cause of discrepancy was renal glycosuria. One out of the 338 registered non-diabetic persons was found to have diabetes. Diabetic patients tended to overestimate diabetes duration significantly. Insulin treatment was verified in 19/20 (95%) and treatment with oral hypoglycaemic agents in all 44 with an affirmative questionnaire answer. A negative answer on insulin and oral hypoglycaemic agents was verified in 100% and 99% respectively. CONCLUSIONS--The concordance was considerably higher than in a comparable Norwegian study performed 10 years earlier. Patient administered questionnaires may be a very reliable source of information for epidemiological purposes in a well defined chronic disease such as diabetes mellitus.