Association of Costs and Days at Home With Transfer Hospital in Home

IMPORTANCE New Centers for Medicare & Medicaid Services waivers created a payment mechanism for hospital at home services. Although it is well established that direct admission to hospital at home from the community as a substitute for hospital care provides superior outcomes and lower cost, the effectiveness of transfer hospital at home—that is, completing hospitalization at home—is unclear. OBJECTIVE To evaluate the outcomes of the transfer component of a Veterans Affairs (VA) Hospital in Home program (T-HIH), taking advantage of natural geographical limitations in a program’s service area. DESIGN, SETTING, AND PARTICIPANTS In this quality improvement study, T-HIH was offered to veterans residing in Philadelphia, Pennsylvania, and their outcomes were compared with those of propensity-matched veterans residing in adjacent Camden, New Jersey, who were admitted to the VA hospital from 2012 to 2018. Data analysis was performed from October 2019 to May 2020. INTERVENTION Enrollment in the T-HIH program. MAIN OUTCOMES AND MEASURES The main outcomes were hospital length of stay, 30-day and 90-day readmissions, VA direct costs, combined VA and Medicare costs, mortality, 90-day nursing home use, and days at home after hospital discharge. An intent-to-treat analysis of cost and utilization was performed.


Introduction
Recent Centers for Medicare & Medicaid Services (CMS) waivers created during the COVID-19 public health emergency have created an opportunity for greatly expanding hospital at home (HaH) services by creating a defined payment mechanism for the program. 1 Such an expansion is a welcomed addition to proven innovations that have been rapidly scaled during the pandemic, at a time when hospital capacity has been overwhelmed in many areas. 2,3 HaH is a program that provides hospital-level care outside the hospital, usually at home. 4 In the US, rigorous observational 5 and randomized 6 studies of admission-avoidance HaH, in which patients are admitted directly to HaH, usually from a hospital emergency department, have demonstrated shorter lengths of stay, less delirium, improved function, and lower costs. Less is known about transfer HaH, 3,[6][7][8] where the program facilitates the transfer of patients being treated in the hospital, who require ongoing hospital-level care, to complete their hospital care at home. A challenge for programs is whether the additional costs of hospital-level care at home (and their duration) could outweigh an additional period in the hospital, followed by less-expensive skilled home health care.
In the Veterans Healthcare Administration (VHA), HaH programs are termed Hospital in Home (HIH). 8 HIH care aligns with high-level VHA strategies to move care to serve veterans effectively in the community and help ease capacity issues in Veterans Affairs (VA) hospitals. The Cpl Michael J Crescenz VA Medical Center (CMCVAMC; Philadelphia, Pennsylvania) has operated a HIH program since 2012, providing both admission-avoidance and transfer HaH care. The program mixes medical management and durable medical equipment (eg, oxygen) from VHA with a bundle of home infusion and home health services purchased from a home health agency, reducing the substantial overhead costs of the typical VHA HIH program, which is fully reliant on VHA personnel. As a result of interstate licensing restrictions, the home health agency providing HIH could not provide services in New Jersey, although a substantial number of veterans served at the CMCVAMC live across the Delaware River in Camden, New Jersey. We took advantage of this geographical restriction to overcome selection bias in evaluating whether the transfer component of the HIH program (T-HIH) was associated with reduced hospital length of stay, readmissions, and posthospital nursing home use, while not being associated with increased VHA or combined VHA and Medicare costs, and whether it was associated with increased days at home in the 90 days after hospital discharge, while not being associated with increased mortality.

Methods
In this quality improvement study, we identified all veterans receiving T-HIH care from the program's inception in 2012 through 2018 admitted to the CMCVAMC hospital and matched them with control patients residing in Camden, for whom HIH was unavailable, using methods described in prior admission-avoidance HIH evaluations. 9,10 Matching variables included demographics (age and age squared, marital status, and VHA priority status P1A, which is given to veterans who had Ն70% service-connected disabilities), hospitalization in the prior 30 days, intensive care unit use during the index hospitalization, number of inpatient diagnoses at admission, JEN Frailty Index score (range, 0-12, with higher scores indicating greater frailty), 11 the VHA Care Assessment of Need probability of hospitalization or death within 90 days (range, 0%-100%, with higher percentages indicating higher probability of hospitalization or death; the variable is termed pEvent), 12 diagnosis-related groups (major diagnostic category level), and admission year. We used merged VA and Medicare records for outcome assessment, as described elsewhere. 9, 10 We used the Area Deprivation Index (ADI; range, 0-100, with higher scores indicating greater neighborhood socioeconomic deprivation) from the Neighborhood Atlas to measure social determinants of health. 13 Outcome measures included costs, utilization, and mortality. We applied an intent-to-treat analysis of cost and utilization, including all veterans from admission to the hospital and followed them through 30-day and 90-day windows. For reporting posthospital utilization outcomes, we A recently adopted CMS quality measure for their Direct Contracting and Primary Care First models is days at home. 14 We used a measure of days at home, operationalized as noninstitutional days within 90 days of discharge among live discharges from hospital. Days at home for HIH patients included the days in HIH because these were days not in an institution.

Statistical Analysis
We conducted a propensity score analysis to estimate the average treatment effect on the treated

Discussion
By leveraging a geographical restriction on HIH availability, we found that in the small transfer component of a HIH program, nursing home days were decreased by 88% (6.5 days) and days at home were increased by 18% (12.6 days). Mortality and inpatient days were lower in the HIH group, but because of the small sample size, the differences were not significant. Although combined total VA and Medicare costs were not significantly lower among HIH patients than control patients at 30 and 90 days, such costs were unlikely to be increased, allaying concerns that T-HIH would add more costly posthospital days than traditional skilled home nursing would cost. By taking advantage of the geographical restriction of the HIH program, we were able to reduce the selection bias typical of T-HIH evaluations. We had intended to control for differing levels of neighborhood socioeconomic disadvantage using the ADI but were unable to include ADI in the final propensity model because of the unavailability of highly socioeconomically disadvantaged matched control patients for 23 T-HIH patients (21%) living in highly socioeconomically disadvantaged areas.
Not including ADI introduces a conservative bias on the results, given that a greater ADI has been associated with worse health outcomes, including greater readmissions and higher mortality.
Compared with less than 3% of Camden patients, more than 19% of Philadelphia patients had ADI scores greater than 85, a level associated with a 9% greater risk of 30-day readmissions. 15 The greater number of community days among T-HIH patients is an important finding for Accountable Care Organizations and Direct Contracting entities, for whom this is to be a future quality measure. The significantly reduced use of nursing homes after hospitalization 16