Introduction
We compared data in two national electronic health registries to evaluate and further optimise the surveillance of patients hospitalised with COVID-19 in Norway.
Method
We included COVID-19 patients registered in the Norwegian Patient Registry (NPR) or Norwegian Pandemic Registry (NoPaR) with admission dates between 17 February 2020 and 1 May 2022. We linked patients, identified overlapping hospitalisation periods and described the overlap between the registries. We calculated the proportion of ICD-10 diagnosis codes (from NPR) and their combinations by main cause of admission (clinically assessed as COVID-19 or other, from NoPaR), age and time. We also calculated the sensitivity and specificity of selected code combinations for representing U07.1 patients’ main cause.
Results
In the study period, 20,115 admissions with laboratory-confirmed COVID-19 were registered in NoPaR, and 21,030 with the ICD-10 code U07.1 in NPR. Up to late-2021 there was a 90 – 100% overlap between the registries, which thereafter gradually decreased to below 75%. The distribution of ICD-10 codes by main cause and sensitivity and specificity of selected code combinations for representing COVID-19 as main cause varied by age and time.
Conclusion
Changes in patient cohorts, virus characteristics and the management of COVID-19 patients in hospitals from late-2021 impacted the registration of patients in NoPaR and ICD-10 codes in NPR. Using ICD-10 codes for the surveillance of patients hospitalised due to COVID-19 requires age- and time-specific definitions, and ongoing validation to consider temporal changes in patient cohorts and virus characteristics.