Reply to Lai and Wei
Journal
Clinical infectious diseases : an official publication of the Infectious Diseases Society of America
Journal Volume
75
Journal Issue
6
Pages
1113
Date Issued
2022-09-29
Author(s)
Abstract
To the Editor—We thank Lai and Wei [1] for their interest in our work [2]. First, to inform clinical practice, our clinical diagnosis–based inclusion criteria simulate the timing of antiviral treatment in real-world settings (“when indicated, should not be delayed until the results of diagnostic testing are available for persons at high risk of complications”) as recommended by the Infectious Diseases Society of America (IDSA), the US Centers for Disease Control and Prevention (CDC), and the Infectious Diseases Society of Taiwan [3]. We did not have access to laboratory data of patients (molecular assays, rapid influenza diagnostic testing, or concomitant bacterial culture). However, the laboratory diagnosis–based inclusion criteria proposed by Lai and Wei [2] had been used in a previous clinical study (N = 1113) [4]. In keeping with our findings, the study showed that inhaled zanamivir is not inferior to oral oseltamivir for shortening the duration of fever or viral shedding [4]. Second, propensity score stratification controlled the confounding from multiple comorbidities (including those covered by the Charlson comorbidity index) [5]. There is no need to further match the Charlson comorbidity index again. Third, our study was designed to detect the superiority of oral oseltamivir over inhaled zanamivir (reference). We cautiously concluded that inhaled zanamivir is not inferior to oral oseltamivir, because analysis of the main outcome was in favor of inhaled zanamivir but the difference did not reach statistical significance (adjusted hazard ratio [aHR] for influenza-related hospitalization or death: 1.01; 95% confidence interval [CI]: .96–1.16). Prespecified subgroup analyses of the high-risk groups, elderly individuals aged 65 years and older (aHR: 1.14; 95% CI: 1.05–1.25) or those with chronic lung diseases (aHR: 1.23; 95% CI: 1.08–1.41) further strengthen this conclusion. Fourth, patients with consciousness disturbance or low inspiration effort have a propensity score (probability) (of assignment to oral oseltamivir) approaching 1.0. Such patients were excluded from analyses during trimming (exclusion of the extreme 2.5% in both tails of propensity score distribution) and will not cause a selection bias. Finally, to avoid repeatedly counting the same individuals, we only include the first outpatient visit of each eligible patient with a diagnosis of influenza during the same influenza season. We considered influenza episodes (of the same individual) in different influenza seasons (with different viral strains) as independent events. Propensity score weighting balanced any duplicated counting of the baseline characteristics.
SDGs
Publisher
OXFORD UNIV PRESS INC
Type
letter