Objective: To examine the feasibility of using information from the Dutch Municipal Population Register (Dutch acronym: GBA) to anonymously track groups of patients.
Method: Using a random sample from the Dutch National Medical Register (Dutch acronym: LMR), hospital admission with discharge in January 1996, it was determined to what extent the admission records could be linked to the GBA. The following variables were used in the linking process: date of birth, gender and the numerical part (4 digits) of the postal code. Once a record had been linked to a single record in the GBA, the pattern of moving and mortality over the next two years was investigated.
Results: Of the 124,598 different hospitalisation records in the LMR cohort, 84% could be linked with a single record in the GBA. In 11% of the patients no unique link was possible: one hospital record could be linked to several records in the GBA. No matching record could be found for 5% of the patients; some of these were foreign citizens resident in the Netherlands. Two years after discharge, the cumulative mortality was 44% for cancer patients, 28% for patients with acute myocardial infarction, 57% for patients with heart failure and 20% for patients with respiratory diseases. Eleven percent of the patients moved within the two-year period following discharge. This would have caused a considerable bias in the aforementioned mortality rates, if the patients' migration had meant that they could no longer be followed.
Conclusion: The introduction of the GBA has significantly increased the possibilities for following patients within and between registries. A prerequisite for this is a unique and correct identification of an individual within the GBA.