Positive Prediction Value of Retinal Artery Occlusion Diagnoses in the Danish National Patient Registry: A Validation Study

Abstract

Purpose: The hospital registration of retinal artery occlusions in the Danish National Patient Registry has not previously been validated. In this study, the diagnosis codes were validated to ensure the diagnoses had an acceptable validity for research. The validation was performed both for the overall diagnosis population and at the subtype diagnosis level. Methods: The medical records for all patients with retinal artery occlusion with an incident hospital record in the years 2017–2019 in Northern Jutland (Denmark) were assessed in this population-based validation study. Furthermore, fundus images and two-person verification were assessed for the included patients when available. The positive prediction values for the overall diagnosis of retinal artery occlusion, as well as for the central or branch subtypes, were calculated. Results: A total of 102 medical records were available for review. The overall positive prediction value for a retinal artery occlusion diagnosis was 79.4% (95% CI: 70.6–86.1%), while the overall positive prediction value at the subtype diagnosis level was 69.6% (95% CI: 60.1–77.7%), with 73.3% (95% CI: 58.1–85.4%) for branch retinal artery occlusion and 71.2% (95% CI: 56.9–82.9%) for central retinal artery occlusion. For the stratified analyses at the subtype diagnosis, age, sex, diagnosis year, and primary or secondary diagnosis, the positive prediction values ranged from 73.5 to 91.7%. In the stratified analyses at the subtype level, the positive prediction values ranged from 63.3 to 83.3%. The differences among the positive prediction values of the individual strata of both analyses were not statistically significant. Conclusions: the validities of the retinal artery occlusion and subtype level diagnoses are comparable to other validated diagnoses and considered acceptable for use in research.