Abstract Rationale Hospitalized patients with pulmonary hypertension (PH) present with a high comorbidity burden that is not captured by existing PH-specific risk scores. This can foster a disconnect between PH-specific severity and total-patient risk, resulting in inaccurate risk stratification in the inpatient setting. We hypothesized that a multi-layered phenotyping approach that integrates comorbidity patterns and functional lab traits would more accurately stratify these complex patients for inpatient outcomes. Methods We conducted under IRB exemption a retrospective analysis of 2,519 unique adult patients with PH hospitalized between August 2020 and July 2025. Using sequenced cluster analysis (JMP Pro 19, SAS Inc., Cary NC), we identified distinct patient clusters by integrating 38 Elixhauser comorbidity categories (v2025.1) with key standardized laboratory traits reflecting the functional status of those comorbidities (e.g., Creatinine, Pro-Brain Natriuretic Peptide, Hemoglobin). Cluster association with outcomes, including ICU admission, hospital length of stay (LOS), and 180-day readmission, was analyzed. Results Cluster analyses identified 6 distinct clusters. These phenotypes were driven by specific comorbidity-biomarker ensembles and were associated with distinct clinical course metrics and outcomes. Cluster 2 (n = 179, 7%), defined by severe renal failure (94%) and anemia (92%), exhibited the highest median Pro-BNP (14,695 pg/mL), highest ICU admission rate (88%), and highest 180-day readmission rate (74%). Cluster 3 (n = 447, 18%) was defined by severe heart failure (98%) and moderate renal failure (88%) and had a high ICU admission rate (91%). In contrast, Cluster 5 (n = 378, 15%) demonstrated high uncontrolled HTN (90%) but low baseline HF/renal disease yet harbored a 92% ICU admission rate. Cluster 4 (n = 205, 8%) represented a low-risk group with the lowest ICU admission (7%) and shortest LOS (4.3d). The clusters also evinced statistically significant separation in the 180-day hospital-free survival (Wilcoxon p = 0.0050) (Image 1). Conclusion Integrating comorbidity burden with attendant laboratory traits reflecting disease severity and functional status identified 6 distinct phenotypes within a heterogeneous hospitalized PH cohort. These phenotypes are associated with inpatient resource utilization and post-discharge outcomes. This data-driven approach provides a more granular framework for risk stratification beyond PH-specific scores and may help guide personalized interventions. This abstract is funded by: None
Palma et al. (Fri,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: