Table 1.
Study components based on hypothetical target trial, emulated trial and original cohort.
Fig 1.
Study design using target trial emulation with cloning and censoring.
A total of 3,193 eligible patients were initially assigned to one of three treatment groups: favipiravir with dexamethasone (FPV with Dexa), favipiravir (FPV) alone, or symptomatic treatment (ST). The cloning process expanded each patient across the three treatment arms (n = 9,579), ensuring comparability. Patients who deviated from their assigned protocol were censored, and follow-up was conducted for 30 days. Time-to-event analysis was performed using inverse probability of censoring weights to estimate treatment effects.
Fig 2.
From all patients with COVID-19 admitted to Bussarakham Field Hospital (n = 18,184), asymptomatic patients (n = 2,789) and those with mild severity (n = 9,747) were excluded. The remaining eligible patients were 3,193.
Table 2.
Demographic, clinical characteristics, treatment, and outcome of the original patient cohort.
Fig 3.
Standardize differences of each baseline characteristic between treatment strategies (A) unweighted standardized difference; (B) weighted standardized difference.
The reference lines at ±0.1 indicate the threshold for acceptable similarity between characteristics. Between ST and FPV with Dexa, 13 baseline characteristics differed (STD > 0.1). Comparisons between ST vs. FPV and FPV with Dexa vs. FPV showed fewer differences, with 10 and 9 out of 25 characteristics differing, respectively.
Fig 4.
Kaplan-Meier estimation of 30-day in-hospital mortality (right) and the predicted survival curve based on flexible parametric survival regression.
The Kaplan-Meier curve shows empirical survival probabilities over 30 days for patients receiving symptomatic treatment (blue), favipiravir alone (red), and favipiravir combined with dexamethasone (green). The flexible parametric model provides a smoothed survival estimate, accounting for time-dependent effects. Both models indicate that survival probabilities differ across treatment groups, with the symptomatic treatment group showing a slightly lower survival probability over time.
Table 3.
Restricted mean survival time of in-hospital mortality for each treatment strategy.