Positive predictive value of respiratory infection diagnoses in the Danish National Patient Registry: A post hoc validation analysis of the DANFLU-1 trial
Christensen J., Johansen ND., Bernholm KF., Borchsenius JIMH., Janstrup KH., Modin D., Skaarup KG., Nealon J., Samson S., Loiacono M., Huang Y., Harris R., Larsen CS., Jensen AMR., Landler NE., Claggett BL., Solomon SD., Gislason GH., Køber L., Landray MJ., Sivapalan P., Jensen JUS., Biering-Sørensen T.
Objectives To validate respiratory infection diagnosis codes in the Danish National Patient Registry (DNPR) in the DANFLU-1 feasibility trial, and to assess whether reperforming effectiveness analyses with validated events altered findings. Methods Hospitalizations for influenza or pneumonia (International Classification of Diseases, Tenth Revision codes J09-18) were validated against medical record review by two blinded physicians. Events were validated against (i) medical opinion and (ii) diagnostic criteria of symptomatology combined with laboratory or radiological verification. Positive predictive values (PPVs) were calculated along with relative vaccine effectiveness (rVE) of high-dose (HD) vs standard-dose (SD) influenza vaccines. Results Among 12,477 participants, we identified 43 events in the DNPR (influenza J09-11: n = 3, pneumonia J12-18: n = 40), which demonstrated PPVs of 86% (95% confidence interval: 72-95%) and 91% (78-97%) against the medical opinion of the respective reviewers. Against diagnostic criteria, the PPVs were 65% (49-79%) and 67% (51-81%). For influenza codes, the PPV was 100% according to reviewers and references and between 63% and 90%, depending on reviewer and reference for pneumonia. The rVE consistently favored HD vs SD against validated events. Conclusion Hospitalizations for pneumonia registered in the DNPR may benefit from subsequent validation, warranting further studies. HD influenza vaccination was associated with fewer validated events than SD. Trial Registration Clinicaltrials.gov: NCT05048589