Computerised record linkage: Compared with traditional patient follow-up methods in clinical trials and illustrated in a prospective epidemiological study

Abstract

Computerised record linkage systems have great potential for enhancing or even replacing traditional methods of adverse event reporting based on active patient follow-up, both in clinical trials and in epidemiological studies. However, these methods must be evaluated. The West of Scotland Coronary Prevention Study (WOSCOPS) is a randomised double-blind clinical trial of pravastatin versus placebo in the primary prevention of coronary heart disease, with coronary heart disease death plus nonfatal myocardial infarction as its primary end point. Adverse event reporting is based on active patient follow-up at routine trial visits. In parallel with this approach, we have obtained computer records of all deaths, incident cancers, and hospitalisations for our subjects by linking their names, dates of birth, and postcodes of their home addresses with a Scottish national database operated by the Scottish Record Linkage system. The results of this comparative study, based on follow-up of the 6595 men ages 45–64 randomised in WOSCOPS, demonstrate minor flaws in both systems, show that follow-up based on computerised linkage alone can be as effective as reporting based on direct contact with the patients, and show that a system based on both approaches provides a direct cross-validation of the two approaches to adverse event reporting while minimising the frequency of unreported events. Preliminary results are reported for a prospective epidemiological study of 80,184 men, ages 45–64 years, who were screened for coronary heart disease risk factors as part of WOSCOPS. This study is based solely on computerised linkage reporting of events on these subjects. This provides an indication of the number of events in various categories that will be available for analysis in future reports. The associations between death rates and standard risk factors such as age, blood pressure, total cholesterol level, and smoking status mirror those reported in other studies.