Sensitivity analyses

To account for uncertainty, we performed four deterministic scenario sensitivity analyses by varying key model parameters in branches of the decision trees where pulse oximetry was used, with increasingly conservative assumptions regarding the performance and benefits of using the pulse oximeters. In the first, we assumed a reduced sensitivity and specificity of pulse oximetry plus clinical assessment for the diagnosis of severe LRTI of 70% and 85%, respectively and that the mortality benefits were halved. In the second, we did not vary the original sensitivity, specificity, and mortality estimates but assumed no reduction with pulse oximetry in the median hospital length of stay of admitted patients, a lower reduction in antibiotic prescriptions of 25% in each patient category, and no antibiotic resistance cost. In the third, we combined both the reduced performance characteristics and mortality benefits of pulse oximetry plus clinical assessment used in the first sensitivity analysis with the increased costs used in the second. Finally, in a worst‐case scenario we combined the values used in the third sensitivity analysis and assumed no reduction in inappropriate antibiotic prescriptions. The sensitivity analyses are shown in Supporting Information S3.

This report was prepared in accordance with the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) guideline [39]. The completed CHEERS checklist can be found in Supporting Information S4.