Validity of Algorithm for Classification of In- and Outpatient Hospital Contacts in the Danish National Patient Registry

Abstract

Purpose: Following the implementation of the 3rd version of the Danish National Patient Register (DNPR-3), information on whether hospitalizations were inpatient, outpatient, or emergency room (ER) contacts was no longer readily available. This study examined the positive predictive values (PPV) of a common algorithm to characterize hospitalizations as inpatient, outpatient, or emergency room (ER) contacts in both DNPR-2 and DNPR-3.

Patients and methods: All hospital contacts in North Denmark Region were identified in the DNPR within a 1-year window of the implementation of DNPR-3 in early 2019. An algorithm based upon proportion of overnight (±50%) and elective (±50%) contacts for each hospital department was developed. Next, PPVs of these categorizations were computed using manual characterization of all departments and clinics by two experienced clinicians as reference. Second, the reliability of various time intervals to join department contacts and subsequent categorization of overnight hospital stays as proxies for inpatient contacts was explored.

Results: The algorithm yielded PPVs of 91% and 89% for hospital units and related contacts categorized as inpatient in DNPR-2 and 100% for both parameters in DNPR-3. In outpatient units, the PPVs were 99% in both DNPR-2 and DNPR-3, whereas the corresponding PPVs were 99.6% and 99% on the contact level. In contrast, the PPV for ERs was 33% in DNPR-2 and 56% in DNPR-3, primarily due to misclassification of outpatient clinics. Still, the proportion of correctly categorized ER contacts was 87% in DNPR-2 and 85% in DNPR-3. Using time intervals from 0 to 12 hours to join department contacts showed that overnight hospitalizations comprised inpatient contacts in 97% in DNPR-2 and 98% in DNPR-3. However, the sensitivity was moderate at 76-78% for all inpatient hospitalizations in DNPR-2 and DNPR-3.

Conclusion: This algorithm accurately categorized hospitalizations as inpatient, outpatient, or ER contacts in both DNPR-2 and DNPR-3.

Keywords: DNPR; Danish National Patient Registry; LPR-3; Landspatientregisteret; epidemiological methods; medical record linkage; registries; research design; validation studies.