Arrhythmia phenotype and 30-day readmission in selected cancers: Propensity-matched NRD study.

Abstract

11180 Background: Cardiac arrhythmias are common in oncology and may increase rehospitalization and resource use, but arrhythmia-type and cancer-specific readmission risk is incompletely characterized. Methods: Retrospective cohort of adult index hospitalizations for colorectal, breast, lung, prostate, kidney, or melanoma in the HCUP Nationwide Readmissions Database. Arrhythmias coded during the index stay (atrial fibrillation [AF], atrial flutter [AFL], ventricular tachycardia [VT], supraventricular tachycardia [SVT]) were identified using ICD-10-CM codes. Primary outcome was all-cause 30-day readmission. For each phenotype, propensity-score matching (caliper 0.05) created arrhythmia and no-arrhythmia cohorts balanced on demographics, payer, income quartile, and comorbidity. Multilevel logistic regression with hospital clustering estimated adjusted odds ratios (aOR) with 95% CI and tested arrhythmia-by-cancer interactions. Secondary outcomes were index length of stay (LOS) and total charges, analyzed with mixed-effects models; results are reported as adjusted mean differences (days) and incremental charges (USD). Two-sided p < 0.05 was significant. Results: Among 1,562,919 index hospitalizations, AF was present in 214,475 (13.7%), AFL 27,545 (1.8%), VT 5,184 (0.3%), and SVT 25,714 (1.6%). After matching/adjustment, AFL was associated with higher 30-day readmission (aOR 1.095, 95% CI 1.03-1.17; p = 0.005). AF (aOR 1.001, 0.98-1.03; p = 0.940), VT (aOR 1.128, 0.99-1.29; p = 0.078), and SVT (aOR 1.018, 0.95-1.09; p = 0.598) were not significant overall. AF demonstrated cancer-specific heterogeneity: lung aOR 1.092 (1.043-1.144; p < 0.001) and melanoma aOR 1.143 (1.019-1.283; p = 0.023), while breast was not increased (aOR 0.964, 0.908-1.025; p = 0.242). In the AF model, older age was inversely associated with readmission (aOR 0.974 per year, 95% CI 0.972-0.976; p < 0.001). In the AF model, highest income quartile was associated with higher readmission (aOR 1.409, 95% CI 1.275-1.557; p < 0.001). AF/VT/SVT increased LOS: AF +0.658 days, VT +1.419 days, SVT +1.146 days (p < = 0.001); charges increased: AF +$12,399 (95% CI 8,231-16,569), VT +$48,114 (36,846-59,382), SVT +$26,656 (19,822-33,489) (all p < 0.001). Conclusions: In selected cancers, AFL was independently associated with higher 30-day readmission, while AF readmission risk varied by cancer type. AF/VT/SVT were linked to increased LOS and charges, with the largest resource impact in VT. Prospective studies incorporating cancer stage and therapy exposures should test phenotype-guided cardio-oncology transitional care to reduce readmissions and costs. Limitations include ICD-coded phenotyping, inability to distinguish incident vs prevalent arrhythmias, and unavailable outpatient therapy/anticoagulation; residual confounding is possible. Misclassification may bias estimates toward null, particularly for VT.