OBJECTIVES: Epidemiological research is increasingly recognizing the value of claims data as a robust and comprehensive data source. However, misclassification bias in epidemiological and health economic studies based on claims data often leads to underestimation of the ‘true’ disease burden, particularly for respiratory infectious diseases like Respiratory Syncytial Virus (RSV). The primary objective of this study is estimating misclassification of RSV in outpatient health administrative claims data. Initially, we will evaluate the validity of various diagnostic criteria and ICD-10 codes for detecting individuals with RSV in claims data by examining sensitivity (SE), specificity (SP), positive predictive values (PPV) and negative predictive values (NPV).

METHODS: Potential cases are identified through administrative claims data from a German statutory health fund for the years 2005-2020. Considering the inaccessibility of laboratory test results, ‘true’ RSV is assumed according to the following criteria: any type of RSV-specific inpatient diagnosis (J21.1, J20.5, J21.0, B97.4), any type of RSV-specific outpatient diagnosis in combination with any general outpatient general Acute Respiratory Infection (ARI)-Code within a quarter. We investigate the changes in SE, SP, PPV and NPV for different combinations of RSV-specific and non-specific ICD-10 diagnoses in claims data.

RESULTS: Preliminary findings reveal a range of SE, SP, PPV and NPV when different combinations of RSV-specific and ARI-specific codes are utilized. Without modifications and for all possible combinations of RSV and ARI specific codes, SE was 26.5%, SP was 95.1%.

CONCLUSIONS: The results indicate underestimation of the ‘true’ disease burden of RSV and demonstrate differences in predictive values, SE and SP according to the choice of ICD-10 for RSV estimation. To account for underestimation, approaches handling uncertainty in estimating RSV incidence or prevalence with claims data are needed.