Cardiovascular diseases and risk factors in a population-based study in The Netherlands: agreement between questionnaire information and medical records

BACKGROUND: Since 1987 the Monitoring Project on Cardiovascular Risk Factors has been carried out in The Netherlands. The purpose of the present study was to assess the agreement between the self-reported questionnaire information that was collected in this survey on cardiovascular diseases and risk factors and information from medical records.

METHODS: From 1987 to 1991, over 36,000 men and women aged 20-59 years have participated in the Monitoring Project on Cardiovascular Disease Risk Factors, a cross-sectional population-based study. We compared self-reported information on cardiovascular diseases and risk factors with the general practitioners medical records from a sample of 899 hypertensive subjects.

RESULTS: The highest proportion of positive self-reports could be confirmed by the medical records for diabetes conditional on self-report of medication use for this condition (100%), and a history of hypertension conditional on the presence of blood pressure recordings in the medical records (91%). The lowest percentage of positive self-reports confirmed by medical records was a family history of myocardial infarction (3-5%). More than 80% of the negative self-reports were confirmed by the medical records for most conditions, except for history of hypertension, conditional on the presence of blood pressure recordings in the medical records (40%). The overall agreement above chance was substantial for cerebrovascular disease (kappa = 0.78) and diabetes (kappa = 0.75), moderate for myocardial infarction (kappa = 0.55), poor for myocardial infarction of the respondents' father (kappa = 0.07) and mother (kappa = 0.05), and fair to moderate for all other self-reported conditions.

CONCLUSION: Agreement between self-reported information and medical records was variable. For further studies, self-reports of cardiovascular diseases and risk factors should preferably be complemented with information from other sources such as medical records.