High early readmission rates in a digital heart failure virtual ward: real-world clinical and operational insights from a retrospective cohort study
A Sidhu, K Pattenden, M Giglia, F Damirova, Y Gao, K Mcbeath, F Windsor, C Screeche-Powell, M Sweeney, A Varnava, N Linton, C Barton, C Plymen, D Auger, S ZamanAbstract
Background
Despite advances in therapeutics and treatment delivery, heart failure (HF) imposes a substantial burden on acute secondary care services (1). In the UK, HF-related admissions account for approximately 1 million inpatient bed-days annually and prolonged admissions with a median length of stay (LoS) of around 8 days (2). Virtual wards (VWs) for HF aim to alleviate this burden by supporting early discharge or admission avoidance through specialist remote care. While early data suggest that VWs are feasible and may reduce bed-days (3), evidence for consistent reductions in admissions and readmissions remains mixed (4). We evaluated the impact of a HF VW using smartphone-based remote monitoring, staffed by non-specialist nurses with specialist support, but without the ability to deliver intravenous (IV) diuretics at home.
Purpose
To evaluate the clinical and operational impact of a HF VW by comparing interventions, LoS and readmissions between VW patients, a control group of HF allcomers receiving standard care, and national registry data.
Methods
This retrospective cohort study included 563 consecutive patients with HF referred between January and December 2024. Of these, 150 (26.6%) were enrolled across two VWs: 57 in an ‘admission avoidance’ ward and 93 in a ‘post-hospital discharge’ ward. The remaining 413 formed the ‘HF allcomers’ control group. Baseline characteristics, LoS and readmissions were compared between VW and control groups. Interventions during the VW stay were compared between the two ward types. Statistical analyses included chi-squared tests, Kaplan-Meier event analysis, and log-rank tests.
Results
Median follow-up was 8.7 months. Although the VW group was significantly younger than controls (65 vs. 77 years, p<0.01), frailty scores were similar (median 4/9 vs. 5/9, p=0.29), indicating high age-adjusted co-morbidity for VW patients. The commonest intervention was oral diuretic adjustment (54.0%). 10.7% required ambulatory IV diuretics. 40.7% had no intervention (Table 1). Median VW stay was 14 days (p=0.42 between VW type). For the ‘post-hospital discharge’ ward, the preceding inpatient stay was significantly shorter than the national average (3 vs. 8 days, p<0.01). HF readmission within 30 days occurred in 24% of VW patients: significantly higher than both allcomers and national rates (p<0.01) (Figure 2). No difference in readmission rates was seen between the two ward types (p=0.39) (Table 1).
Conclusions
VW patients were 16-fold more likely to be readmitted to hospital for HF within 30 days of VW discharge. This may be due to high age-adjusted frailty on the VW, shorter-than-average preceding hospital stays, and no home IV diuretic capability. The frequent lack of active intervention highlights the need to optimise patient selection and enrolment timing. These findings suggest that HF VWs without therapeutic capacity may have limited impact, particularly when used to expedite discharge of clinically complex patients.For image description, please refer to the figure legend and surrounding text.For image description, please refer to the figure legend and surrounding text.