Changes in Diagnoses and Site of Care for Patients Receiving Hospice Care From Agencies Acquired by Private Equity Firms and Publicly Traded Companies

Key Points Question Is the acquisition of hospice agencies by private equity firms and publicly traded companies associated with changes in site of care and clinical diagnoses of Medicare patients? Findings This cohort study used a difference-in-difference analysis to examine 1967 US hospice agencies, of which 158 were acquired by private equity firms and 250 were acquired by publicly traded companies. Prior to acquisition, private equity firms and publicly traded companies targeted hospices with more patients in nursing homes; after acquisition, both agencies of private equity firms and publicly traded companies saw a rise in Medicare patients with dementia and patients who received hospice services at home, and hospices acquired by publicly traded companies were associated with lower patient Hierarchical Condition Category score after acquisition. Meaning These results suggest that after private equity firm and publicly traded company acquisitions, hospice agencies, as compared with for-profit hospices that were never acquired, saw an increase in Medicare patients receiving care at home and those diagnosed with dementia.


eFigure 1. Sample Selection
For the sample selection, we start with the base dataset consisting of variables from the CMS public use files (PUF).The dataset consists of 34,362 observations from years 2013 to 2020, with 5,708 unique CMS certification numbers (CCNs).To get to our final analytic sample, we implement the following three step process.
First, we include private equity (PE) and publicly traded company (PTC) owned facilities that have been acquired between 2014 and 2019, this brings the dataset to 17,124 and 18,185 total observations, with 3,598 and 3,713 unique CCNs, for PE and PTC samples respectively.Secondly, we implement a rule requiring for-profit (FP), PE, and PTC facilities to be present in the sample at the year of their acquisition as well as in the subsequent year, bringing the dataset to 16,984 and 18,025 total observations, with 3,567 and 3,666 unique CCNs, for PE and PTC samples respectively.Lastly, for our selected outcome variables, we implement a final criterion requiring that there be no NA/missing values present across each observation's outcomes, i.e, ensuring no NA/missing values in the entirety of the final analytic samples.This yields our final PE and PTC samples, with the FP comparison group, from 2013 to 2020.
For the final analytic PE sample, we have a total of 8,786 observations with 158 unique PE owned facilities and 1,559 unique FP facilities which make up our comparison group.The final PTC analytic sample has a total of 9,407 observations with 250 unique PTC facilities and the same FP comparison group with 1,559 unique facilities.The final analytic samples ensure that across all models, PE and PTC agencies are compared to the same for-profit agencies and the same agencies are compared across all outcomes.
eFigure 2. The Number of Transactions and Agencies Acquired by Private Equity Firms and Publicly Traded Corporations, 2013-2021 NOTE: Transactions (represented by the black lines) refer to unique acquirer acquisition events, while agencies (represented by the bars) refer to the number of hospice agencies involved in those transactions.eFigure 3. Adjusted Differences in Sites of Care between Treatment and Control Agencies Relative to the Year of Hospice Acquisition NOTES: The sample includes 158 freestanding hospices owned by PE and 250 owned by PTCs, which were acquired between 2014 and 2019.In addition, there are 1,559 for-profit hospices in the comparison group.We only included hospices without missing observations across all outcomes.Callaway and Sant'Anna (2021) difference-in-differences method was used for estimation.Models included fixed effects for agency and year.Outcomes were Winsorized at the top and bottom 1%.The unit of analysis was the hospice agency-year; standard errors were adjusted for clustering at the level of the hospice agency.
eTable 1. Adjusted Differences in Sites of Care and Patient Characteristics between Treatment and Control Agencies Relative to the Year of Hospice Acquisition, with State Fixed Effects Callaway and Sant'Anna (2021) difference-in-differences method was used for estimation.Models included fixed effects for agency and year.Outcomes were Winsorized at the top and bottom 1%.The unit of analysis was the hospice agency-year; standard errors were adjusted for clustering at the level of the hospice agency.eTable2. Adjusted Differences in Sites of Care and Patient Characteristics between Treatment and Control Agencies Relative to the Year of Hospice Acquisition, with State Fixed Effects and Controls NOTES: The sample includes 136 freestanding hospices owned by PE and 209 owned by PTCs, which were acquired between 2014 and 2019.In addition, there are 1,279 for-profit hospices in the comparison group.We only included hospices without missing observations across all outcomes.Callaway and Sant'Anna (2021) difference-in-differences method was used for estimation.Models included fixed effects for agency and year.Outcomes were Winsorized at the top and bottom 1%.The unit of analysis was the hospice agency-year; standard errors were adjusted for clustering at the level of the hospice agency.
NOTES:The sample includes 158 freestanding hospices owned by PE and 250 owned by PTCs, which were acquired between 2014 and 2019.In addition, there are 1,559 for-profit hospices in the comparison group.We only included hospices without missing observations across all outcomes.