Caregiver burden in a home hospital versus traditional hospital: A secondary analysis of a randomized controlled trial

Hospitals are standard of care for patients with acute illness, but hospitalization can result in heavy financial burden and iatrogenic risk, particularly in older populations. An alternative to inpatient care, “home hospital” provides acute hospital-level care in patients' homes.1 Such programs decrease cost and readmission, while maintaining or improving quality and safety.2, 3 New federal regulations have allowed for the expansion of home hospital to over 290 hospitals.4 Although the benefits of home hospital to patients and health systems appear well-defined, caregiver experience remains understudied in the United States. Prior studies established not only the essential role of caregivers in patient care but also the associated stressors.5, 6 Concern for increased caregiver burden was cited as a reason for declining participation in home hospital among 6% of eligible patients.7 We sought to identify burden experienced at home compared with the hospital. We performed a retrospective analysis of a randomized controlled trial conducted between June 12, 2017 and January 16, 2018.2 Patients were recruited in the emergency department (ED) after presenting with a qualifying illness (such as infection or heart failure), adequate functional status (able to ambulate to commode), and severity requiring admission (but at low risk of requiring intensive care). Participants were randomized to home hospital (“home”) or traditional hospital (“control”). Home patients received twice daily nurse visits, once daily physician visits, in-home diagnostics, continuous monitoring, intravenous medications, and other hospital-level care; home health aides could be deployed for 12-hour shifts to assist with caregiving. Caregivers, defined as paid or unpaid individuals aiding a person with illness, completed the Zarit Burden Interview-12 (ZBI-12) on admission and within 30 days of discharge. The ZBI-12 ranges from 0–48 with suggested scoring: ≤10, no-to-mild burden; 11–20, mild–moderate burden; >20, high burden.8, 9 To facilitate interpretability, patients without complete ZBI-12 caregiver data were excluded from analysis. We extracted caregiver demographic data from electronic health records. We compared patients across caregiver status, with and without ZBI-12, using chi-squared and t-tests. Patients with caregivers in the home and control group were compared with Fisher's exact test (for categorical patient characteristics) and Wilcoxon signed-rank test (for ZBI-12). Calculations were performed in GraphPad and RStudio. This study was approved by the Mass General Brigham IRB. Ninety-one patients were enrolled; 42 (46.2%) had a caregiver. Patients with caregivers were older (median age, 85 vs. 66 years, p < 0.001), had more prescription medications (15.0 vs. 10.0, p = 0.001), and lower baseline functional status (2.5 vs. 6 activities of daily living [ADLs]; 2 vs. 8 instrumental ADLs, both p < 0.001), but had similar comorbidity count (9 vs. 6, p = 0.07). Of patients with caregivers, 33 (79%; 22 home, 11 control) had complete ZBI-12 data. There was no difference between patients with and without complete ZBI-12 caregiver data. There was no significant difference between caregiver characteristics in each group (Table 1). Most caregivers were next of kin, legally designated healthcare proxies, and lived in the same household as patients. The most common relationship was parent–child, with daughters comprising 57.6% of all caregivers. Median burden in home was ZBI-12 of 9.5 (IQR, 4.8) on admission and 9.5 (IQR, 10.8) on discharge versus the control with a median ZBI-12 of 15.0 (IQR, 11.5) on admission and 8.0 (IQR, 10.5) on discharge (Figure 1). There was no significant difference in burden between the groups on admission (p = 0.30), discharge (p = 0.98), or burden change (difference −0.5 vs. −2.0, p = 0.33). We demonstrate caregiver burden is mild to moderate upon admission and discharge in both home and traditional hospitalization. Our study is reassuring against large differences in caregiver burden caused by home hospital, suggesting these programs are a viable alternative for acutely ill patients with caregivers insofar as it is unlikely to cause undue burden on caregivers. Prior meta-analysis similarly demonstrated insignificant effect on caregiver burden.10 Variation exists among home hospital programs regarding caregiver expectations. Notably, our program could deploy in-home aides, whereas some provide none. Considering baseline burden and patients' functional limitations, we feel this is an important capability for alleviating caregiver burden during admissions. Limitations include small sample size causing group imbalance and limited power to detect small but clinically important differences, two-site enrollment (although patients had diverse sociodemographics), approximately 20% missing data (although respondents and nonrespondents were similar), lack of detailed caregiver sociodemographics, 30-day timeframe of post-discharge survey, and exclusion of individuals with limited ambulation. To our knowledge, this represents some of the first evidence from a randomized controlled trial that caregiver burden remains constant during a home hospital admission without significant difference from those in traditional hospitalizations. Moss had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Levine, Moss, and Schnipper. Acquisition, analysis, or interpretation of data: all authors. Drafting of the manuscript: Moss. Critical revision of the manuscript for important intellectual content: all authors. Statistical analysis: Moss. Administrative, technical, or material support: Moss. Study supervision: Levine. This study and article did not receive specific funding. David Levine is supported by an investigator-initiated grant and a codevelopment with Biofourmis. David Levine also is supported by fees from the MetroHealth System. Carson Moss and Jeffrey Schnipper do not have conflicts of interest. Not applicable.