Multisensor trends during worsening heart failure events treated with outpatient intravenous diuretics

Background Acute worsening of heart failure (HF) may be treated in outpatient settings with intravenous (IV) diuretics. However, limited data exist on the physiological changes that occur before and after these events [1]. Objective Compare the time-course of implanted multisensor measurements during HF events (HFEs) treated in outpatient versus inpatient settings. Methods Retrospective data from subjects with implanted cardioverter defibrillator and resynchronization therapy devices (ICD/CRT) were linked with US Medicare claims. Inpatient HFEs were identified using primary HF diagnosis codes (DRG 291-293 and ICD-10-CM I50.xx), while outpatient HFEs required additional IV diuretic use (J1940). The device calculates a daily HF index (0-100, with higher values indicating worse status) from sensor-derived measures of night heart rate, heart sounds, impedance, respiration, and patient activity. An alert is issued when the index increases above a threshold (nominal value = 16) [2]. Alert diagnostic performance was evaluated based on HFE detection sensitivity and false positive rate (FPR). Sensor changes were analyzed in a subset of HFEs that met criteria for follow-up duration, event separation, and sensor data availability. Sensor trends were aligned by event date (day 0), and changes were calculated between baseline (-60 to -31 day) and pre-event (-5 to -1 day) averages using generalized estimating equations. Results A total of 6612 linked subjects, with an average follow-up of 1 year, provided 1091 HFEs for performance evaluation, of which 119 (11%) were outpatient HFEs. The multisensor HF alert had a nominal sensitivity of 79% [95%CI: 75, 81%] at a FPR of 1.40 alerts/pt-year [95%CI: 1.39, 1.46]. Sensitivity was not different for outpatient events compared to inpatient events (73% vs 79%, p=0.4). A subset of 53 outpatient and 547 inpatient HFEs were available for sensor change analysis. Sensor changes from baseline to event were not different by HFE group (Table), which was reflected in comparable changes of the HF index (9.9 for outpatient and 6.7 for inpatient, p=0.23). However, in outpatient HFEs the HF index suggested a trend towards delayed but more rapid worsening closer to the event (Figure). Conclusion In a real-world US Medicare cohort of ICD/CRT patients, the magnitude of sensor changes preceding outpatient HFEs treated with IV diuretics was similar to that preceding inpatient HFEs. Further research is needed to explore differences in sensor temporal dynamics and their prognostic significance.