OBJECTIVES : Despite recognized value in considering days at home (DAH) as a patient-centered outcome, few studies have evaluated factors associated with decreased DAH for patients with heart failure (HF). The objective of this observational, retrospective study was to develop a model for identifying modifiable factors associated with fewer DAH for patients with HF.

 

METHODS : Patients age ≥65, enrolled in a Humana Medicare Advantage Prescription Drug plan, and indexed on a claims-based HF diagnosis (first inpatient or second of two outpatient) were identified 7/2016 - 12/2017. Six months pre- and 12 months post-index continuous enrollment (or until death or hospice election) was required. The proportion of lost DAH was defined as the sum of assigned time the patient was in a non-home health care service setting over total eligible post-index days. Predictive models, three logistic regression and one random forest, including patient characteristics and modifiable factors were developed using machine learning techniques on a split sample to differentiate risk of ≥8% lost DAH post-index. This approximates to one month lost DAH per year.

 

RESULTS : A total of 205,223 patients met study criteria. At ≥8% cutpoint, lost DAH provided adequate signal (21.7%) and aligned closely with the top quintile of lost DAH for this cohort. The logistic regression main effects model with transformed covariates produced a parsimonious model demonstrating good predictive probability calibration. Modifiable factors predicting ≥8% lost DAH post-index were: lack of an outpatient visit within 10 days of an index inpatient HF discharge and fewer prescribed HF therapies within 30 days post-index. Other baseline factors included: index inpatient HF admission, higher frailty, loop diuretic use, pre-index acute lost DAH, comorbidity level and age.

 

CONCLUSIONS : Predictive modeling that identifies modifiable factors associated with fewer DAH for patients with HF could facilitate outreach and interventions to support patients with HF and increase their time at home.