Individualized treatment rule for early steroid use in hospitalized patients with community acquired pneumonia: a cohort study

Current evidence on an optimal patient selection strategy for adjunctive steroids to curb excessive inflammation in community acquired pneumonia (CAP) is limited. An individualized treatment rule (ITR) customizes treatment recommendations based on individual patient characteristics. The objective of this study was to develop an ITR for early steroid use in hospitalized patients with CAP. Using a single center cohort of hospitalized patients with CAP from 2009 to 2019, we developed a single decision ITR to initiate or not initiate steroids early (within 24 h) after admission. The primary outcome of interest was hospital-free days measured at 28 days. Regression-based learning with LASSO selected a model estimating expected outcomes of potential intervention with steroids individualized to predictors. The optimal ITR was compared to other treatment rules including as observed in the data. A total of 4379 patients were included in this cohort with 1412 (32%) patients receiving steroids within 24 h of hospital admission. Compared to observed practice, an optimal ITR was associated with increased hospital-free days (mean rate [95% confidence interval (CI)]: 21.74 [21.52, 21.95] versus 22.31 [22.11, 22.51]). The optimal ITR for the secondary outcomes including ventilator-free days, mortality and need for advanced respiratory failure support and/or mortality revealed better estimates compared to what was observed in the data. However, there was a lack of consistent performance when applying the advanced respiratory failure and/or mortality (a secondary outcome) ITR to other outcomes (inconsistency of results across models). An ITR for early steroid use in hospitalized patients with CAP did not consistently improve clinical outcomes.