• Center on Health Equity and Access
  • Clinical
  • Health Care Cost
  • Health Care Delivery
  • Insurance
  • Policy
  • Technology
  • Value-Based Care

Value-Based Care Through Postacute Home Health Under CMS PACT Regulations

The American Journal of Managed CareFebruary 2022
Volume 28
Issue 2

Among a patient population defined by CMS postacute care transfer regulations, home health vs no postacute care was associated with reduced 30-day readmissions and costs.


Objectives: To assess in a Medicare Advantage population (1) whether discharge to home health, compared with discharge to home, following an inpatient stay subject to CMS postacute care transfer (PACT) regulations, is associated with better outcomes or lower expenditures and (2) whether the impact differs among subpopulations.

Study Design: Claims-based retrospective cohort study.

Methods: Instrumental variable (IV) analysis, with prior hospital-level probability of discharge to home health as the IV, to control for unobservable as well as observable confounders.

Results: Compared with 15,071 patients discharged to home, 4160 patients discharged to and receiving timely home health services were 60% less likely to be readmitted within 30 days and 37% less likely at 90 days. Total expenditures from time of admission to 90 days post discharge were 11% lower in the home health group. The association of discharge to home health with reduced readmission and reduced costs varied by subpopulations defined by surgical vs medical diagnosis-related group and receipt of intensive care management following discharge.

Conclusions: The PACT policy may be promoting greater value by reducing readmissions while lowering total expenditures for patients who do not require intensive postacute care. Findings were in contrast to those of previous studies, in which discharge to home health has been associated with higher rates of readmission. Earlier studies did not control for unmeasurable confounders, involved narrowly defined populations, and used older data.

Am J Manag Care. 2022;28(2):e49-e54. https://doi.org/10.37765/ajmc.2022.88827


Takeaway Points

The CMS postacute care transfer policy reduces hospital reimbursement for shorter-than-average stays for certain diagnosis-related groups if patients enter postacute care.

  • Among policy-eligible patients discharged to home health or home, home health was associated with a 60% reduction in 30-day readmissions and a 37% reduction at 90 days; a statistically significant relationship persisted in the surgical subpopulation.
  • Total expenditures (admission through 90 days post discharge) were 11% lower with home health.
  • Results should encourage clinical entities to consider home health for patients with relatively short hospital stays who do not require intensive postacute care, particularly patients with surgical admissions.


Under the CMS postacute care transfer (PACT) policy, hospitals receive reduced payment for shorter-than-average stays if the discharge is considered a postacute transfer, defined as discharge to a skilled nursing facility, a long-term acute care hospital, a distinct hospital unit (eg, a long-term care or rehabilitation unit), or hospice, or discharge with orders for home health services to be provided by a home health agency and to begin within 3 days of discharge. Annual applicable diagnosis-related groups (DRGs) are determined by recent CMS inpatient payment experience.1,2 The policy rationale is that hospitals should not be paid at the usual DRG rate for shorter stays when the total episode includes postacute care.

This study explored the PACT policy’s implications for improving the value of care for an entire episode (admission to 90 days post discharge) for patients discharged to home health instead of to home without home health services. Home health entails care delivery in the home to help prevent recurrence and complications following an acute event and is one of the fastest-growing areas of US health care expenditures.3

Evidence concerning the value of postacute home health is sparse, derived from old data, and representative of narrowly defined patient populations.3-6 PACT policy DRGs, in contrast, include both surgical and medical DRGs across a spectrum of severity and complexity. Moreover, previously published comparisons of home health with no postacute care do not address unmeasured confounding, a relevant consideration because the decision to discharge with home health can be influenced by unobservable patient needs.

The objectives of this study were to assess (1) whether discharge to home health following a PACT policy–eligible inpatient stay, compared with discharge to home, is associated with better outcomes or lower expenditures and (2) whether the impact of home health differs among subpopulations defined by type of inpatient stay and exposure to care management. To improve analytic validity through control of unmeasured confounding, an instrumental variable (IV) approach was taken.


Study Population and Data Sources

An Advarra institutional review board granted a waiver of informed consent for this retrospective cohort study. Participants were in a Humana Medicare Advantage prescription drug plan and had a hospital discharge during 2018 for 1 of 280 PACT policy DRGs (eAppendix Tables 1 and 2 [eAppendix available at ajmc.com]). Humana’s 2018 Medicare Advantage population exceeded 2.8 million, almost all with a Part D plan.7 Inclusion also required a rounded-down length of stay (LOS) less than the CMS-reported geometric mean for the assigned DRG and a discharge code for home health or home. Each patient’s first eligible discharge was the index event. These exclusion criteria were applied relative to the index date: age less than 19 or greater than 89 years, lack of continuous enrollment for the previous 12 months, end-stage renal disease or hospice care during the previous 12 months or in the following 90 days, and attribution to a primary care physician who delegated claims adjudication to a third party (an option under certain forms of value-based payment). The remaining cohort was divided into preliminary groups of home health and home based on discharge code. Individuals discharged to home health who did not receive home health services within 3 days and individuals discharged to home but receiving home health services within 3 days were then excluded from the final groups, making the 2 treatment groups equivalent for everything except the discharge code and home health services initiation within the time window dictated by CMS PACT policy rules. The 3-day requirement avoids classifying on the basis of home health unlikely to be related to the discharge.

Study data were derived from claims, CMS (mortality), enrollment files, program participation records, and primary care contracts. Humana at Home is a care management service for Medicare Advantage participants with complex medical and functional needs that might be addressed through regular contact with nurse case managers. Special Needs Plans (SNPs) are CMS-approved coordinated care plans for Medicare Advantage participants who are institutionalized, are dually eligible for Medicaid and Medicare, or have a severe or disabling condition. Two Humana at Home programs provide intensive care management: the Transitions Program for certain patients newly discharged from the hospital and the Long-Term In-Home (LTIH) Program for patients with chronic needs. Participants in Humana at Home may or may not be enrolled in SNPs. Neither Humana at Home programs nor SNPs entail delivery of care in the home.


The primary outcome measure was all-cause readmission within 30 days of discharge. Additionally, all-cause readmission at 60 and 90 days; emergency department (ED) visits within 30, 60, and 90 days (excluding those associated with admission); all-cause mortality at 30, 60, and 90 days; and costs at 90 days, with and without inclusion of admission costs, were measured. Costs included total allowed amounts for all medical and pharmacy claims. The following baseline characteristics were measured as of index date: age, sex, race, geographic region and population density of residence, low-income indicators, and surgical vs medical DRG.8 Baseline variables for inpatient admissions, ED visits, and the Charlson Comorbidity Index (Deyo version; see eAppendix) were created, using data from the 12 months preceding the index admission. In addition to these baseline variables, the following care management variables were covariates in adjusted analyses: (1) Humana at Home Transitions Program participation anytime during 90 days post discharge, (2) Humana at Home LTIH Program participation anytime during 90 days post discharge, and (3) eligibility for a CMS SNP on the index date.

To permit IV analysis, 3 versions of a variable representing the hospital preference for discharge to home health were constructed. See the eAppendix for more detail.


Conventional analyses. Descriptive statistics were used to compare baseline characteristics and postdischarge care management between the home health and home groups and to test the unadjusted association between treatment group and each outcome. The adjusted association between treatment group and outcomes was assessed through Cox proportional hazards models for risk of readmission, ED visits, and mortality and through a generalized linear model (log link, gamma distribution) for costs. Model covariates are described in the preceding section. Patients without continuous enrollment for 90 days following the index date were excluded from the cost analysis.

IV analysis. IV analysis was used to improve the validity of findings by accounting for potential sources of unmeasured confounding, also taking into account measured confounders. Important unmeasurable confounding was suspected because of potential differences between the home health and home groups in patient selection factors that would have influenced both discharge orders and outcomes. For example, a patient living alone might be more likely to be discharged to home health. Living alone could increase the risk of falls, nonadherence to medications, failure to attend follow-up physician visits, and deficiencies in self-care, in turn increasing the risk of poor outcomes and greater expenditures. Other confounding could occur because of clinical factors absent from claims data. By definition, an IV only indirectly influences outcomes through its direct association with the treatment. Unmeasurable confounders are assumed to be randomly distributed across different levels of the IV. The IV considered in the present study was hospital-level prior preference for (ie, prior probability of) discharge to home health among all patients discharged during the look-back period. The value of the IV was to be substituted for treatment group in the same models used in the conventional analysis. Although conclusions were to be based on the results of IV analysis, results from conventional analysis are reported to show how control for unmeasured confounding altered results. See the eAppendix for additional detail.

Subgroup analyses. Stratified analyses demonstrated the differential impact of home health in subpopulation pairs defined by surgical vs medical DRGs, participation vs no participation in one or the other of the 2 Humana at Home programs, and eligibility vs no eligibility for a SNP.


Inclusion and exclusion criteria yielded a cohort of 19,231 patients: 4160 in the home health group and 15,071 in the home group (see eAppendix Figure 1). Table 1 summarizes unadjusted group differences. Some differences were not only statistically significant but also large enough to be meaningful. Patients in the home health group were more likely to be female and to live in an urban area, were less likely to qualify for a low-income program, had lower levels of preadmission comorbidity and general health care utilization, were much more likely to have had an index admission for a surgical rather than a medical DRG, and were less likely to receive postdischarge Humana at Home services or be in an SNP. The index admissions had occurred in 8419 hospitals.

Testing for the association of the IV with discharge to home health resulted in defining the IV as the probability of discharge to home health in the 6 months prior to the date of a study patient’s index discharge. See the eAppendix for testing details.

Results for patient outcomes favored the home health group (Table 2). With adjustment only for measured confounders (conventional analysis), patients in the home health group had more than a 25% smaller risk of readmission at 30 days. With additional adjustment for unmeasured confounders (IV analysis), the relative risk reduction was more than 50%. The observed benefit of home health became somewhat attenuated over time, but a significant reduction in risk of 90-day readmission remained after adjustment for both measured and unmeasured confounders. No statistically significant association between home health discharge and ED visits or all-cause mortality was observed.

The benefit of home health discharge varied somewhat by subpopulation (Figure). Reduction in 30-day readmission among those receiving home health was greater in the surgical (vs medical) DRG subpopulation and was statistically significant only in the surgical DRG subpopulation. Reduced risk of 30-day readmission was observed for recipients of home health regardless of receipt of intensive Humana at Home services during 90-day follow-up but was not statistically significant among patients receiving Humana at Home services. A similar pattern was observed when the analysis was stratified by SNP and no SNP. The reduction in readmissions in the home health group remained greater at 60 and 90 days for patients with surgical DRGs, no Humana at Home, or no SNP eligibility than for patients in the alternate subpopulations, but differences grew smaller over time (eAppendix Figures 2 and 3).

Total adjusted episode costs, including admission and 90-day postdischarge care, were comparable between the home health and home groups in conventional adjusted analysis but were significantly lower in the home health group in IV analysis. When hospital admission expenditures were disregarded, no overall cost savings were detected. The actual cost savings, according to unadjusted analysis, was $239 per patient. See Table 3 and the graphic display in eAppendix Figure 4.


In an IV analysis, discharge to and timely receipt of home health services was associated with reduced readmission rates and total episode expenditures in a national patient population defined by the CMS PACT policy. The 30-day risk of readmission was 60% lower for patients discharged to home health compared with patients discharged to home. Readmission remained lower in the home health group up to 90 days. Home health was additionally associated with an 11% reduction in total 90-day allowable costs, taking into account both the reduced hospital payment and the added cost of home health services for patients in this group. Because IV analysis was originally conceived as a theoretical simulation of randomized controlled trials, it is reasonable to infer that these associations represent causal relationships.9,10 The added robustness of results, compared with results controlled only for measured confounders, demonstrates the potential utility of the IV approach.

Although the cost savings demonstrated by this study are from a payer perspective, they have implications for clinicians and health care systems, particularly those under value-based payment arrangements involving downside financial risk. Clinical entities can apply these findings to those situations in which intensive postacute care is not needed. Study findings suggest that for a wide range of DRGs, home health would improve outcomes and result in shared savings even after accounting for home health costs.

The present study is, to the authors’ knowledge, the only published investigation of the impact of the PACT policy. It also adds to the body of evidence concerning home health services through its methods and the patient population studied. Previous studies have found higher readmission rates among patients discharged to home health as opposed to home.3-6 By contrast, the present study found readmission rates to be lower for recipients of home health, perhaps because of control through IV analysis for unobservable differences. Nevertheless, the present study also demonstrated a significant though smaller reduction in readmission through conventional multivariable analysis. The present study’s unique classification requirement of receipt or nonreceipt of home health services in the first 3 days post discharge may have improved the likelihood of detecting home health benefits. Another difference from previous publications was a more broadly defined patient population who had been hospitalized for a wide range of procedures and conditions. Because of the shorter-than-average LOS that defines PACT-eligible discharges, these patients differ clinically from other patients with the same DRG in ways that may affect the risk of readmission independent of discharge orders. The PACT policy population in the present study did in fact have a relatively low overall rate of 30-day all-cause readmission: 7% compared with 17% to 30% in previously published studies making the same comparison.3,4,6 Although patients may be discharged to home health rather than home because of an elevated risk of readmission, home health services may not completely eliminate the risk difference, depending on the nature of the admission. But for the population defined by the PACT policy, whose primary common denominator is shorter lengths of stay, home health may in fact improve outcomes. Because previous studies used data sets from 2005 to 2011, it is also possible that the present study, using data from 2018, reflects improvements in home health quality.11

Stratified analysis showed that in all subpopulations, patients experienced fewer readmissions if they received home health, but statistical significance varied and differences diminished over time. Stratification by surgical vs medical DRG revealed a significant reduction in 30-day readmissions, the primary outcome, comparing home health with home discharge, in the surgical subpopulation, but the reduction was nonsignificant in the medical subpopulation. Surgical procedures often create new needs, such as wound care or occupational therapy. By contrast, care needs after medical hospitalizations may not differ from preadmission needs because the condition has been stabilized. It is noteworthy that 85% of patients in the home health group had had a surgical stay, compared with only 50% of the home group. It is also possible that medical patients have more complex needs, which would be consistent with the greater levels of preadmission comorbidity and utilization observed in the home group, and that these needs are less easily resolved with home health services. The direction of results in other stratified analyses suggested that home health may have benefited patients regardless of whether they received intensive care management in the form of a Humana at Home program or through a SNP. However, the observed benefits of home health were larger and statistically significant in the subpopulations without these additional forms of oversight. Conclusions about the differential impact of home health among subpopulations must be made with caution because the CIs of subpopulation estimates overlapped.

The CMS PACT policy was designed to distribute payments across the providers where care is received. This study’s findings suggest that, additionally, policy incentives may be promoting greater value by reducing readmissions while reducing total episode costs. Study results should encourage physicians to consider home health for a broad range of admissions, especially for surgical admissions, that might otherwise result in discharge to home.


A key study limitation is that although the IV was designed to control for unmeasured patient-level confounding, it may not capture all hospital-level confounders, such as variations in care quality, adequacy of discharge planning, hospital culture, and geographical practice patterns. The available data did not allow creation of hospital-level variables to control for these factors. This source of confounding may be mitigated by adjustment in all models for patient-level geographic location, population density, income, and race. These patient-level factors are likely to be associated with the quality of care delivered in each study patient’s index hospital. For example, it can be assumed that high-income earners are more likely to be admitted to hospitals delivering higher-quality care and that urban residents have access to hospitals with more resources. Still, hospital-level confounding remains an important limitation. A remaining potential source of confounding is differences across hospitals in the distribution of the 280 PACT-eligible DRGs.

Control for the effect of the LTIH Program was imperfect because there was no differentiation between initiation of these services before and after the outcome measurement time point. Additional limitations inherent in administrative databases, such as missing data, incorrect coding, and the unavailability of such information as planned readmission at the time of discharge, also apply.

Results may not generalize to true real-world practice in which receipt or lack of home health services immediately after discharge can be inconsistent with the actual discharge orders. In the present study, approximately 27% of otherwise study-eligible patients with a home health discharge received no home health services within 3 days and were thus excluded. Results may not generalize to traditional Medicare, other Medicare Advantage populations, or future PACT policy DRG lists.


In an IV analysis, discharge to home health was associated with improved patient outcomes (fewer readmissions) and reduced expenditures in a national patient population defined by the CMS PACT policy. Results should encourage physicians and health care systems to consider home health for patients whose hospital stays, particularly surgical stays, have been relatively short and who do not require intensive postacute care to ensure that early discharge does not jeopardize patients’ chances of avoiding a readmission. Further research is needed to more precisely define the diagnoses, procedures, hospital stay parameters, and patient characteristics that can be used to identify patients likely to benefit from home health services.


The authors appreciate the contributions of Adrianne Casebeer, PhD, whose experience in home health research guided our analytic plan, and Dana Drzayich Antol, MS, who coordinated the production of this manuscript and provided a critical review.

Author Affiliations: Humana Inc (PR, TR, BSt, MF, CD, YL, BSa, KW, AS, DA), Louisville, KY.

Source of Funding: None.

Author Disclosures: Mr Racsa, Mr Flagg, Ms Sallee, and Drs Li and Worley are employees of Humana Inc. Ms Rogstad was a full-time employee of Humana at the time of study execution and now performs medical writing services for Humana on a freelance basis. Mr Stice, Mr Dailey, and Drs Sharma and Annand were employed by Humana at the time of study design, execution, and interpretation.

Authorship Information: Concept and design (PR, TR, BSt, MF, CD, YL, BSa, KW, AS, DA); acquisition of data (PR, MF); analysis and interpretation of data (PR, TR, BSt, MF, CD, YL, BSa, KW, AS, DA); drafting of the manuscript (PR, TR, MF); critical revision of the manuscript for important intellectual content (PR, BSt, YL, BSa, KW, AS, DA); statistical analysis (PR); administrative, technical, or logistic support (PR, TR, MF); and supervision (TR, BSt, CD).

Address Correspondence to: Patrick Racsa, MPH, Humana, 500 W Main St, Louisville, KY 40202. Email: pracsa2@humana.com.


1. 42 CFR § 412.4. Discharges and transfers. Legal Information Institute. Accessed May 7, 2020. https://www.law.cornell.edu/cfr/text/42/412.4

2. Update to the hospital transfer policy for early discharges to hospice care. CMS. June 20, 2018. Accessed May 8, 2020. https://www.cms.gov/Regulations-and-Guidance/Guidance/Transmittals/2018Downloads/R2094OTN.pdf

3. Keehan SP, Cuckler GA, Poisal JA, et al. National health expenditure projections, 2019-28: expected rebound in prices drives rising spending growth. Health Aff (Millwood). 2020;39(4):704-714. doi:10.1377/hlthaff.2020.00094

4. Martin RCG, Brown R, Puffer L, et al. Readmission rates after abdominal surgery: the role of surgeon, primary caregiver, home health, and subacute rehab. Ann Surg. 2011;254(4):591-597. doi:10.1097/sla.0b013e3182300a38

5. Sanford DE, Olsen MA, Bommarito KM, et al. Association of discharge home with home health care and 30-day readmission after pancreatectomy. J Am Coll Surg. 2014;219(5):875-886.e1. doi:10.1016/j.jamcollsurg.2014.07.008

6. Al-Masrouri S, Garfinkle R, Al-Rashid F, et al. Readmission for treatment failure after nonoperative management of acute diverticulitis: a nationwide readmissions database analysis. Dis Colon Rectum. 2020;63(2):217-225. doi:10.1097/DCR.0000000000001542

7. Humana reports first quarter 2018 financial results; raises full year 2018 EPS guidance. News release. Humana; May 2, 2018. Accessed January 18, 2021. https://humana.gcs-web.com/static-files/e96955a3-da19-44a4-b652-cbc6582de399

8. Ammori JB, Navale S, Schiltz N, Koroukian SM. Predictors of 30-day readmissions after gastrectomy for malignancy. J Surg Res. 2018;224:176-184. doi:10.1016/j.jss.2017.12.004

9. Brooke BS, Goodney PP, Kraiss LW, Gottlieb DJ, Samore MH, Finlayson SRG. Readmission destination and risk of mortality after major surgery: an observational cohort study. Lancet. 2015;386(9996):884-895. doi:10.1016/S0140-6736(15)60087-3

10. Kramer DB, Normand ST, Volya R, Hatfield LA. Facility-level variation and clinical outcomes in use of cardiac resynchronization therapy with and without an implantable cardioverter-defibrillator. Circ Cardiovasc Qual Outcomes. 2018;11(12):e004763. doi:10.1161/CIRCOUTCOMES.118.004763

11. Dick AW, Murray MT, Chastain AM, et al. Measuring quality in home healthcare. J Am Geriatr Soc. 2019;67(9):1859-1865. doi:10.1111/jgs.15963

Related Videos
Julie Patterson, PharmD, PhD, National Pharmaceutical Council
James Robinson, PhD, MPH, University of California, Berkeley
James Robinson, PhD, MPH, University of California, Berkeley
James Robinson, PhD, MPH, University of California, Berkeley
Carrie Kozlowski
James Robinson, PhD, MPH, University of California, Berkeley
Carrie Kozlowski, OT, MBA
Miriam J. Atkins, MD, FACP, president of the Community Oncology Alliance (COA) and physician and partner of AO Multispecialty Clinic in Augusta, Georgia.
Carrie Kozlowski, OT, MBA
Shawn Gremminger
© 2024 MJH Life Sciences
All rights reserved.