Abstract Background: Chimeric antigen receptor T-cell (CAR-T) therapy has transformed the treatment of relapsed and refractory hematologic malignancies. However, it is frequently accompanied by significant toxicity and a high burden of hospitalization, with nearly one in four patients requiring hospital readmission within 30 days of infusion. Identifying real-world predictors of early readmission could help tailor supportive care and guide discharge planning. However, real-world evidence remains scarce, with existing studies often constrained by small sample sizes, heterogeneous populations, and inconsistent data capture. We sought to explore the feasibility of structured data collection and identify potential clinical predictors of 30-day readmission following CAR-T therapy at our institution. Methods: We conducted a retrospective analysis of 34 adult patients who received CAR-T therapy between 2022 and 2024 at a single university-affiliated community cancer center. Patient-level data were manually abstracted from medical records and included demographics, diagnosis, type of bridging therapy, length of stay (LOS), and toxicities. The primary endpoint was 30-day hospital readmission post–CAR-T infusion. An initial logistic regression model, including five predictors (gender, age, LOS, CRS grade, and bridging therapy group), exhibited sparse data bias. A Firth-corrected model was attempted but remained unstable. We therefore fit a reduced multivariable logistic regression model using only two variables: LOS (continuous) and bridging therapy (binary: any vs none). Results: Among 34 patients, 9 (26.5%) were readmitted within 30 days. Bridging therapy was administered in 25 patients (73.5%). The final model included 33 patients with complete data. In this model, LOS showed a non-significant trend toward increased readmission risk (OR 1.08, 95% CI 0.94–1.24, p = 0.28). Receipt of bridging therapy appeared to be associated with lower odds of readmission, though this also did not reach statistical significance (OR 0.44, 95% CI 0.08–1.59, p = 0.36). The model demonstrated strong discrimination (c-statistic = 0.88), although this was likely influenced by the small sample size and potential model overfitting. No variable independently predicted readmission in this cohort. Conclusions: In this single-center CAR-T cohort, neither LOS nor receipt of bridging therapy independently predicted 30-day hospital readmission. Although the analysis was underpowered to detect statistically significant associations, the observed directional trends, particularly the potential link between longer hospitalization and higher readmission risk, provide a basis for future hypothesis generation. Our findings underscore the feasibility of capturing structured real-world data in a community-affiliated academic setting and highlight its value in informing operational improvements in CAR-T delivery and post-infusion care. Larger, multi-center studies are essential to validate these preliminary insights and guide evidence-based interventions.
Ammakola et al. (Mon,) studied this question.