[Validity of case definitions for Severe Acute Respiratory Infection based on ICD-10 codes in four autonomous...

ICD-10 diagnostic codes could be useful for automated surveillance of Severe Acute Respiratory Infection (SARI). This study analyzed the validity of different SARI case definitions (CD) based on ICD-10 codes at hospital discharge in Catalonia, the Valencian Community, Galicia and the Basque Country between weeks 21/2021 and 39/2023.

Patients with respiratory system codes (J00 to J99, R06, U07) and with a laboratory test performed were included, with the gold standard being positivity to the corresponding pathogen. The validity of all possible combinations of ICD-10 codes was estimated, prioritizing the CDs according to the average sensitivity (Se) and specificity (Sp). Validity was estimated by age groups, but expert criteria were used to seek the optimal CD, unique for each pathogen, balancing validity and simplicity.

The CD selected for influenza for any age included the specific codes J09.X, J10.0, J10.1, J10.2, J10.8 and J11 (Se=95.4; Sp=99.7), and for SARS-CoV-2, codes U07.1 and J12.8 (Se=95.9; Sp=98.3). For RSV, validity was lower, with significant heterogeneity by age and inclusion of non-specific codes. The optimal CD included codes J06, J12.1, J13, J20.5, J21.0 and J45 (Se=63.9; Sp=88.4), recommending adding J98 for adults and when it is important to maximize Se (Se=88.9; Sp=65.3). The overall CD for the three viruses was consistent with the specific CDs.

The ICD-10 codes provides great validity in identifying influenza and SARS-CoV-2 hospitalizations, but shows poorer performance and greater age-group variability for RSV.