Reply to Nagano et al

To the editor—We thank Nagano and colleagues [1] for their interest in and careful evaluation of our work. We performed a retrospective observational study comparing clinical outcomes in patients receiving a reduced versus standard dose of trimethoprim-sulfamethoxazole (TMP-SMX) for the treatment of Pneumocystis jirovecii pneumonia (PJP) [2]. We found no statistically significant difference in the primary or secondary clinical outcomes between the 2 treatment groups. We appreciate that Nagano and colleagues found that our study “partially validates the hypothesis that the reduced dose TMP-SMX may be similarly effective as the standard dose regimen.” However, they also state that we have “inferred non-inferiority of reduced dose TMP-SMX compared to standard dose” and that such a conclusion cannot be drawn without “a sample size calculation based on a clinically acceptable non-inferiority margin.”

We fully agree with the need for an adequate sample size in a study aimed at establishing non-inferiority between 2 treatment doses. However, due to the low incidence of PJP in patients with hematologic malignancies in Sweden, our study was not designed to establish non-inferiority. In retrospective studies, noninferiority should be calculated after propensity score matching in order to minimize the influence of hidden bias [3], and such a study would have called for a multinational effort. As Nagano and colleagues correctly point out, well-designed, randomized controlled non-inferiority trials are essential to establish the effectiveness of reduced-dose TMP-SMX for the treatment of PJP. While waiting for such studies to be performed, retrospective cohort studies may add important information and can help assess the feasibility and design of prospective studies. We believe that our study adds data to the growing body of evidence that doses below the stipulated 15–20 mg/kg per day of the trimethoprim component may be used to treat mild to moderate PJP, based on the fact that none of the 44 patients with baseline arterial-to-inspired oxygen ratio (PaO₂/FiO₂) ≥ 200 who received treatment with low-dose TMP-SMX died. This led us to conclude that “…in patients with mild to moderate PJP, the mortality was low, and treatment with reduced dose of TMP-SMX was effective.” However, as is stated in the article, the risk of residual confounding in retrospective cohort studies is evident and prospective controlled studies are warranted to confirm our findings.