Abstract
BACKGROUND: Postal questionnaire surveys are commonly used in general practice and often ask about self-reported activity. The validity of this approach is unknown. AIM: To explore the criterion validity of questions asking about self-reported activity in a self-completion questionnaire. METHOD: A comparison was made between (a) the self-reported actions of all general practitioner (GP) principals in 51 general practices randomly selected within the nine family health services authorities of the former northern regional health authority, and (b) the contents of the medical records (case notes and computerized records) of patients classified as hypertensive from a 1 in 7 random sample of all patients registered in these practices and aged between 65 and 80. Data were gathered from the GPs by self-completion postal questionnaires. Six comparisons were made for two groups of items: first, target and achieved blood pressure; secondly, patient's weight, smoking status, alcohol consumption, exercise and salt intake. The frequency with which the data items were recorded in patient records was compared with the GPs' self-reported frequency of performing the actions. RESULTS: No relationship was found between achieved blood pressure and stated target levels. For each of the other actions, more than half of the responders reported that they usually or always performed the activity. For four of these (smoking, weight, alcohol and exercise), a significant association was noted, but the size of this varied considerably. CONCLUSIONS: There is a variable relationship between what responders report that they do in self-completion questionnaires, and what they actually do as judged by the contents of their patients' medical records. In the absence of prior, knowledge of the validity of questions on reported activity, or of concurrent attempts to establish their validity, the questions should not be asked.