Abstract
In the Netherlands, the postal code is needed to study hospitalizations of individuals in the nationwide hospitalization register. Studying hospitalizations longitudinally becomes troublesome if individuals change address. We aimed to report on the feasibility and validity of a two-step medical record linkage approach to examine longitudinal trends in hospitalizations and mortality in a study cohort. First, we linked a study cohort of 1564 survivors of childhood cancer with the Municipal Personal Records Database (GBA) which has postal code history and mortality data available. Within GBA, we sampled a reference population matched on year of birth, gender and calendar year. Second, we extracted hospitalizations from the Hospital Discharge Register (LMR) with a date of discharge during unique follow-up (based on date of birth, gender and postal code in GBA). We calculated the agreement of death and being hospitalized in survivors according to the registers and to available cohort data. We retrieved 1477 (94%) survivors from GBA. Median percentages of unique/potential follow-up were 87% (survivors) and 83% (reference persons). Characteristics of survivors and reference persons contributing to unique follow-up were comparable. Agreement of hospitalization during unique follow-up was 94% and agreement of death was 98%. In absence of unique identifiers in the Dutch hospitalization register, it is feasible and valid to study hospitalizations and mortality of individuals longitudinally using a two-step medical record linkage approach. Cohort studies in the Netherlands have the opportunity to study mortality and hospitalization rates over time. These outcomes provide insight into the burden of clinical events and healthcare use in studies on patients at risk of long-term morbidities.
Highlights
Survivors of childhood cancer are an example of a patient group that has an increased risk of long-term morbidity and mortality [1,2,3,4,5]
It is appealing to use readily available data such as data from national administrative registers and to link a study cohort to these registers. Such medical record linkage studies allow examination of the relation between detailed information on risk factors of the cohort and the clinical events that are routinely registered in administrative registers
Dutch acronym for Municipal Personal Records Database (GBA) is an administrative database in which municipalities register demographic information, such as the Dutch citizen service number (Dutch acronym: BSN), gender, birth, address and postal code, country of birth, marital status and death of their residents
Summary
Survivors of childhood cancer are an example of a patient group that has an increased risk of long-term morbidity and mortality [1,2,3,4,5]. To study clinical events in these kinds of patient groups, it is possible to assess outcomes of interest in a study cohort periodically through clinical assessments This approach is very time-consuming and costly, due to the relatively low absolute frequency of unfavorable health conditions and the potentially long duration between risk factor (for example: cancer treatment) and clinical event (such as treatment-induced health problems). Clinical follow-up of such a reference population will generate additional costs For these reasons, it is appealing to use readily available data such as data from national administrative registers and to link a study cohort to these registers. It is appealing to use readily available data such as data from national administrative registers and to link a study cohort to these registers Such medical record linkage studies allow examination of the relation between detailed information on risk factors of the cohort and the clinical events that are routinely registered in administrative registers. This level of anonymity previously limited longitudinal identification of hospitalizations in medical record linkage studies due to moving (change in registered postal code)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
