Medicare Advantage Reimbursement Structures Impact Home Health Delivery and Outcomes

Abstract

Objectives:

Medicare Advantage (MA) plans provide fewer home health (HH) services than Traditional Medicare (TM), but MA plans vary in how they reimburse HH agencies. Like TM, Episodic MA plans allow agencies to determine the number and type of visits. Alternatively, Per-Visit MA plans dictate a specific number of visits, and which disciplines provide them. This study examined differences in HH care delivery and patient outcomes between TM, Episodic MA, and Per-Visit MA plans.

Study Design:

Secondary analysis of agency data from January 2019 – December 2022.

Methods:

For 285,297 HH stays, inverse probability of treatment weighting regression compared TM to each MA plan type and Per-Visit to Episodic MA plans. We examined HH length of stay, number of visits from nursing, therapy disciplines, social work, and aides, transfer to an inpatient facility during HH, improvement in self-care and mobility function, and community discharge.

Results:

Compared to TM, both MA plans had shorter stays and fewer visits from nursing, therapy, and aides. Comparing MA plans to each other, Per-Visit MA had 2.3% shorter stays, 3.0% more physical therapy visits, and 6.8% fewer social work visits versus Episodic MA. Differences in outcomes comparing MA plans to TM were mixed. Comparing MA plans, per-Visit MA had 12% higher likelihood of inpatient transfers (95% CI 1.06,1.18) compared to Episodic MA.

Conclusions:

Compared to MA plans that dictate the amount and type of visits, Episodic MA plans that allow HH agencies flexibility in determining visit delivery may also have fewer adverse inpatient transfer outcomes.

Précis:

Medicare Advantage plans that place more restrictions on home health agency care delivery may have more adverse patient outcomes than plans that provide episodic payments.

Objective/Introduction

In 2023, Medicare Advantage (MA) enrollment surpassed Traditional Medicare (TM) enrollment for the first time.¹ MA is the private managed care alternative to TM, wherein private companies receive a capitated amount of Federal funding per enrollee to cover beneficiaries’ healthcare costs.² MA insurers are incentivized to reduce costs to improve margins, and they frequently employ cost-saving strategies that are not commonly used by TM like prior authorization, limited provider networks, cost sharing, and lower payments to providers.^2–8 Despite these strategies, MA plans are attractive to enrollees because they offer additional benefits like vision and dental coverage, often with no additional premium compared to TM.^8,9 However, MA costs the Federal government 22% more – over $2,300 – per beneficiary annually, due in part to potential upcoding by MA plans to make enrollees appear sicker and garner higher payments.^2,10–12 Additionally, there are concerns that MA insurers excessively deny care and do not adequately pass savings to enrollees through benefits.^4,13

Millions of beneficiaries receive home health (HH) services annually when a provider certifies that they are homebound and require intermittent nursing and/or rehabilitation care.² In multiple studies, beneficiaries with MA were less likely to receive HH than similar patients with TM.^4,14,15 When they do receive HH, beneficiaries with MA plans receive care from lower-quality agencies, have shorter stays, and receive fewer visits from nurses, therapists, and other staff.^4,14–19 However, despite receiving fewer services, the outcomes for MA patients are mixed. Compared to similar patients with TM, two studies found patients with MA were less likely to improve in function during HH^17,20 and another found higher risk of mortality for MA patients with stroke.¹⁶ Other studies suggest positive effects of MA, including fewer costly readmissions and more days at home after HH.^15,16

One contributor to the mixed impacts in the literature may be the heterogeneity among MA plans. There are nearly 4,000 different MA plans available nationwide, and the average beneficiary can choose between 43 separate plans.⁹ While all MA plans must include a HH benefit, how the benefit is structured varies. One study compared HH use and length of stay (LOS) between MA plans with different cost-sharing, prior authorization requirements, and plan types (i.e., health maintenance organization, preferred provider organization, or special needs plan).⁴ However, these characteristics primarily impact HH admissions through restricted provider networks and administrative burdens at the start of care, and do not account for MA plan reimbursement structure, which has substantial implications for HH agency operations during the stay. HH leaders qualitatively described challenges with MA utilization management and differing reimbursement mechanisms in HH and how they complicate care delivery,²¹ however no quantitative work has examined the different MA reimbursement structures.

This study categorizes MA plans into two groups: Episodic MA and Per-Visit MA, to acknowledge major differences in how HH agencies are reimbursed and better capture the impacts of plan reimbursement structure on agency operations and patient outcomes. Similar to TM, Episodic MA plans pay agencies a lump sum to cover all costs anticipated during an authorized 60-day episode.²² While the agency’s total HH payment is typically less from an Episodic MA insurer compared to TM,⁴ episodic payments allow the agency to determine the number of visits, distribution of visits across the stay, and which disciplines (e.g., nursing, physical therapy, occupational therapy, speech therapy, social work, and/or HH aides) are necessary for each patient’s care plan. In contrast, Per-Visit MA plans dictate the total number of visits – and the number of visits per discipline – that are covered during a specified duration of days, and the agency must seek re-authorization for additional visits or to add covered days. Per-Visit MA plans thus offer minimal flexibility for the agency to determine how many visits, and from which disciplines, each patient receives. However, because each individual HH company has separate negotiations with MA insurers, it is unknown how frequently MA contracts include episodic versus per-visit payments. Additionally, it is unknown whether disparate reimbursement structures lead to differences in HH service delivery or outcomes when comparing MA plans to TM, or when comparing MA plans to each other.

This study is the first to acknowledge the two primary ways that MA plans are structured to reimburse HH agencies when evaluating differences in HH care delivery and patient outcomes between TM and MA. By examining differences across TM, Episodic MA, and Per-Visit MA plans, we aim to inform future regulatory and policy efforts, negotiations between HH providers and MA plans, and potentially beneficiary decisions during enrollment.

Methods

Data and Patient Cohort

We partnered with a large national non-profit company that provided de-identified data on HH stays from 102 locations in 19 states. Stays represent individual plans of care for a single injury or illness. As published recently,¹⁷ data included the Outcomes and Assessment Information Set (OASIS), the number of visits by discipline, insurance plan, and occurrence of inpatient transfers during the stay. We included patients ages 65 and over that were covered by TM or MA from January 2019 through December 2022. We included stays with complete admission and discharge assessments, from which we calculated LOS. To remove stays that provided only intermittent maintenance care, we included stays with up to two certified episodes. Episodes are standard time periods for which payers certify plans of care before requiring re-certification to continue the care plan. Medicare’s new Patient-Driven Groupings Model (PDGM) implemented separate 30-day payment periods within 60-day episodes for TM in January 2020, but PDGM did not change the 60-day episode certification timeframe.

Payer Groups

We indicated whether each stay was covered by TM, an Episodic MA plan, or a Per-Visit MA plan using the specific plan entered in our partner’s billing system. For MA plans, they indicated whether their local contract with the MA insurer was structured to provide episodic versus per-visit payments.

Outcomes

For care delivery, we included LOS, defined as number of days between admission and discharge, and the number of visits from the following disciplines: nursing, physical therapy, occupational therapy, speech therapy, social work, and HH aides. Patient outcomes included transfer to an inpatient facility during the stay, discharge to the community (versus an institution), and dichotomous indicators of functional improvement on OASIS self-care and mobility scores. Function scores are calculated at admission and discharge by totaling nine validated items that have been used in multiple studies^17,23,24 to rate independence with self-care (e.g., grooming, dressing, bathing, feeding), and mobility (e.g., transfers, locomotion/ambulation).

Covariates

To account for differences between TM, Episodic MA, and Per-Visit MA patients,²⁵ we controlled for a comprehensive set of demographic, clinical, social, and environmental characteristics.¹⁷ Demographics included age, sex, and self-reported race and ethnicity. We indicated whether patients received post-acute services versus community-entry HH.²⁶

Clinical factors included admission self-care and mobility scores, level of cognitive impairment, pain that interferes with activity, history of falls, cognitive or behavioral symptoms (i.e., memory deficits, impaired decision making, verbal disruption, physical aggression, or disruptive behavior), levels of dyspnea, and incontinence.^27–29 We calculated Elixhauser comorbidity indices from active diagnoses at admission.^17,30 We indicated whether the patient had 2+ hospitalizations in the previous six months, were taking 5+ medications, or had a pressure ulcer or surgical wound at admission.

Social factors included whether HH was provided in the community (e.g., home or assisted living) versus an institution (e.g., long term care), whether the patient lives alone, and the availability of assistance at home. Environmental characteristics^27,31–33 included rurality of patients’ ZIP codes³⁴ and Social Deprivation Indices (SDI),³⁵ with higher scores indicating more socioeconomic disadvantage across categories including employment, poverty, housing, and transportation access.

Analysis

To account for differences between groups, we used inverse probability of treatment weighting (IPTW), a robust method that reduces selection bias by calculating treatment weights reflecting the propensity of being in the TM, Episodic MA, or Per-Visit MA group based on all demographic, clinical, social, and environmental factors detailed above.^36,37 After calculating weights, we verified that samples were balanced.³⁸

We estimated differences in care delivery and patient outcomes between TM and both MA plan types using linear regression for LOS, negative binomial models for visit counts, and logistic regression for dichotomous outcomes. In each model, we included IPTW weights and an indicator for if the stay spanned PDGM implementation, which changed reimbursement incentives for TM.³⁹ We included year and office location fixed effects to account for the course of the COVID-19 pandemic and declines in HH visits that occurred after PDGM.^39,40 Location fixed effects also account for varying geographical impacts of the pandemic and any differences in processes for managing each plan type between offices. Finally, we used robust standard errors to account for multiple stays for the same patient across the study.

To estimate differences between Episodic MA and Per-Visit MA plans, we used post-hoc linear hypothesis tests that accounted for the weighted model, clustering, and robust standard errors. We also conducted multiple sensitivity analyses. First, we ran all models with IPTW that excluded social, environmental, and OASIS-based clinical covariates, similar to published methods.^4,16 We also ran all models using 2022 data only to reduce COVID-19 contamination. Analyses were conducted in RStudio Version 2024.09.0 with significance at two-sided alpha <0.01. This study was exempted by the University of Washington institutional review board.

Results

We had complete data for 285,297 HH stays, of which 178,195 (62.5%) were covered by TM, 43,299 (15.2%) were Episodic MA stays, and 63,803 (22.4%) were Per-Visit MA. There were significant differences between the three groups for all demographic, clinical, and social, and environmental factors (Table 1). TM patients were the oldest and most likely to identify as Non-Hispanic White; Per-Visit MA patients were more demographically similar to TM patients than to Episodic MA patients. While differences in clinical and social factors varied, TM patients were the most complex in terms of cognitive and function impairments, dyspnea, incontinence, pressure ulcers, falls, and behavioral symptoms. Conversely, MA patients had higher rates of social risk factors such as living alone, having no assistance, and living in more socioeconomically disadvantaged communities. The IPTW approach was successful in balancing covariates between all groups (Figure 1).³⁸

Table 1.

Descriptive statistics for 285,297 home health stays between January 2019 and December 2022 in 102 locations in 19 states for three payer groups.

	Traditional Medicare	Episodic Medicare Advantage	Per-Visit Medicare Advantage
Total Stays	178,195	43,299	63,803
Care Delivery Variables
Length of Stay (Days), Mean (SD)a	46.21 (26.07)	45.03 (25.24)	41.92 (24.27)
Nursing Visits, Mean (SD)a	4.25 (5.26)	3.83 (4.91)	3.77 (4.93)
Physical Therapy Visits, Mean (SD)a	7.26 (5.35)	6.85 (5.01)	5.77 (4.79)
Occupational Therapy Visits, Mean (SD)a	3.05 (3.75)	2.74 (3.48)	2.91 (3.51)
Speech Therapy Visits, Mean (SD)a	0.57 (1.95)	0.44 (1.73)	0.51 (1.81)
Social Work Visits, Mean (SD)a	0.18 (0.54)	0.22 (0.59)	0.19 (0.54)
HH Aides, Mean (SD)a	0.43 (2.00)	0.44 (1.92)	0.41 (1.79)
Patient Outcomes
Mobility Improvement, %a	94.15%	94.77%	94.61%
Self-Care Improvement, %a	94.28%	95.01%	94.65%
Community Discharge, %a	92.64%	94.18%	93.46%
Transfer to Inpatient, %	7.44%	7.14%	7.38%
Demographic Characteristics
Age, Mean (SD)a	79.81 (10.91)	76.15 (11.96)	79.07 (10.58)
Female Sex, No. (%)a	113,587 (63.75%)	25,260 (58.34%)	41,438 (64.94%)
Admitted from Inpatient Facility, No. (%)a	108,144 (60.69%)	28,396 (65.59%)	39,890 (62.52%)
Received HH in a Community Setting (e.g., private home or assisted living), No. (%)a	177,727 (99.74%)	42,962 (99.22%)	63,675 (99.80%)
Race/Ethnicity, No. (%) a
Non-Hispanic White	154,899 (86.93%)	30,041 (69.38%)	51,077 (81.62%)
American Indian/Alaska Native	379 (0.21%)	83 (0.19%)	124 (0.19%)
Asian	4,600 (2.58%)	3,408 (7.87%)	949 (1.49%)
Black	13,591 (7.63%)	6,646 (15.35%)	8,457 (13.25%)
Native Hawaiian/Pacific Islander	798 (0.45%)	970 (2.24%)	191 (0.30%)
Hispanic	3,592 (2.02%)	1,895 (4.38%)	1,919 (3.01%)
Multiracial	336 (0.19%)	256 (0.59%)	86 (0.13%)
Function and Clinical Characteristics
Admit Mobility Function Score (0–15), Mean (SD)a	8.96 (2.33)	8.92 (2.33)	8.77 (2.32)
Admit Self-Care Function Score (0–23), Mean (SD)a	14.37 (3.21)	14.13 (3.11)	14.08 (3.27)
Cognitive Impairment, No. (%) a
None	71,142 (44.41%)	21,948 (50.69%)	31,343 (49.12%)
Mild	46,589 (26.14%)	11,252 (26.00%)	15,417 (24.16%)
Moderate to Severe	52,464 (29.44%)	10,099 (23.32%)	17,043 (26.71%)
Weighted Elixhauser Comorbidity Index (−19 to 89), Mean (SD)a	5.04 (5.44)	5.16 (5.48)	4.87 (5.37)
Pain that interferes with activity or movement, No. (%)a	146,299 (82.10%)	35,881 (82.89%)	50,221 (78.71%)
Pressure ulcer at HH admit, No. (%)a	6,996 (3.93%)	1,513 (3.49%)	2,241 (3.51%)
Surgical wound at HH admit, No. (%)a	37,144 (20.84%)	10,653 (24.60%)	12,706 (19.91%)
Dyspnea level, No. (%) a
Not short of breath	13,945 (7.83%)	3,708 (8.56%)	7,852 (12.31%)
Short of breath with walking more than 20 feet or climbing stairs	23,455 (13.16%)	5,462 (12.61%)	8,508 (13.33%)
Short of breath with moderate exertion	45,777 (25.69%)	10,062 (23.24%)	13,531 (21.21%)
Short of breath with minimal exertion	81,254 (45.60%)	20,312 (46.91%)	29,126 (45.65%)
Short of breath at rest	13,764 (7.72%)	3,755 (8.67%)	4,786 (7.5%)
Incontinence level, No. (%) a
Not incontinent	79,597 (44.67%)	22,097 (51.0%)	30,289 (47.47%)
Incontinent	93,424 (52.43%)	19,948 (46.07%)	31,839 (49.90%)
Requires a urinary catheter	5,174 (2.90%)	1,254 (2.90%)	1,675 (2.63%)
History of 2+ falls or injurious fall in last 12 months, No. (%)a	77,901 (43.72%)	18,157 (41.93%)	27,096 (42.47%)
History of 2+ hospitalizations in the past 6 months, No. (%)a	42,281 (23.73%)	10,828 (25.00%)	14,407 (22.58%)
History of 2+ Emergency Department visits in the past 6 months, No. (%)a	40,860 (22.93%)	10,591 (24.46%)	13,831 (21.68%)
Currently taking 5+ medications, No. (%)a	164,859 (92.52%)	39,887 (92.12%)	58,435 (91.59%)
Cognitive or Behavioral Symptoms Occurring at least weekly, No. (%)a	67,165 (37.69%)	13,787 (31.84%)	31,483 (33.67%)
Home and Community Environment
Availability of Assistance at Home, No. (%) a
None	2,870 (1.61%)	894 (2.06%)	1,204 (1.89%)
Some	39,865 (22.37%)	11,774 (27.19%)	16,398 (25.70%)
Around the Clock	135,460 (76.02%)	30,631 (70.74%)	46,201 (72.41%)
Lives Alone, No. (%)a	35,158 (19.73%)	9,881 (22.82%)	14,73 (22.53%)
Social Deprivation Index (1–100), Mean (SD)a	36.38 (25.63)	46.86 (26.48)	40.31 (27.59)
Rurality, No. (%) a
Urban	165,706 (92.99%)	38,806 (89.62%)	60,002 (94.04%)
Large Rural	8,432 (4.73%)	3,411 (7.88%)	2,845 (4.46%)
Small Rural	2,357 (1.32%)	666 (1.54%)	593 (0.93%)
Isolated Rural	1,700 (0.95%)	416 (0.96%)	363 (0.57%)
Episode Characteristics
Certified for two 60-day episodes, No. (%)a	21,463 (12.04%)	4,856 (11.22%)	6,105 (9.57%)
Year, No. (%) a
2019	46,513 (26.10%)	8,495 (16.62%)	16,052 (25.16%)
2020	40.759 (22.87%)	8,417 (19.44%)	13,951 (21.87%)
2021	46,890 (26.31%)	13,124 (30.31%)	16,808 (26.34%)
2022	44,033 (24.71%)	13,263 (30.63%)	16,992 (26.63%)
Stay Spanned PDGM Implementation, No. (%)a	5,117 (2.97%)	907 (2.09%)	1,614 (2.53%)

^aSignifies statistically significant difference at alpha <0.01 between TM, Episodic MA, and Per-Visit MA groups based on one-way ANOVA for continuous variables or chi-squared tests for dichotomous and categorical variables.

Abbreviations: SD - Standard Deviation, HH - Home Health, PDGM - Patient-Driven Groupings Model

Figure 1.

Standardized mean differences of covariates between Episodic Medicare Advantage, Per-Visit Medicare Advantage, and Traditional Medicare patients before and after adjustment with inverse probability of treatment weights, demonstrating adequate balancing of all covariates after weighting as evidenced by absolute standardized mean differences well below 0.1. Abbreviations: HH – home health; ED – Emergency Department; RUCA – rural urban commuting area; SDI – social deprivation index.

Results for adjusted differences in care delivery variables are in Table 2, with relative differences expressed as percentages. Compared to TM, Episodic MA stays were 0.98 days shorter (95% CI −1.29, −0.67), equating to a 2.1% difference. Per-Visit MA patients were 1.99 days shorter (95% CI −2.24, −1.75), or 4.3% shorter than TM. Both MA plans had fewer visits from nursing and all therapy disciplines compared to TM. While Episodic MA plans had 3.8% more social work visits (95% CI 1.00, 1.07) than TM, Per-Visit MA plans had 3.3% fewer social work visits than TM (95% CI 0.94, 0.99). Compared to TM, both MA plans had fewer HH aide visits (9.6% fewer for Episodic MA [95% CI 0.84, 0.97] and 8.1% fewer for Per-Visit MA [95% CI 0.87, 0.97]).

Table 2.

Results of adjusted weighted linear and negative binomial models estimating differences in care delivery between home health stays covered by Traditional Medicare (TM), Episodic Medicare Advantage (MA), and Per-Visit MA plans between 2019–2022.

	Episodic MA Versus TM		Per-Visit MA Versus TM		Per-Visit MA Versus Episodic MA
	Coefficienta (95% CI)	Relative Difference (%)	Coefficienta (95% CI)	Relative Difference (%)	Coefficienta (95% CI)	Relative Difference (%)
Length of Stay (Days)	−0.98 (−1.29, −0.67)	−2.1%	−1.99 (−2.24, −1.75)	−4.3%	−1.02 (−1.37, −0.66)	−2.3%
Nursing Visits	0.95 (0.94, 0.97)	−5.0%	0.94 (0.92,0.95)	−6.3%	0.98 (0.97, 1.00)	-
Physical Therapy Visits	0.95 (0.94, 0.96)	−5.0%	0.98 (0.97, 0.99)	−2.1%	1.03 (1.01, 1.05)	3.0%
Occupational Therapy Visits	0.96 (0.04, 0.97)	−4.3%	0.97 (0.96, 0.98)	−3.0%	1.01 (0.99, 1.03)	-
Speech Therapy Visits	0.92 (0.88, 0.97)	−7.7%	0.92 (0.88, 0.96)	−7.9%	1.00 (0.94, 1.06)	-
Social Work Visits	1.04 (1.00, 1.07)	3.8%	0.97 (0.94, 0.99)	−3.3%	0.93 (0.90, 0.97)	−6.8%
Home Health Aide Visits	0.90 (0.84, 0.97)	−9.6%	0.92 (0.87, 0.97)	−8.1%	1.02 (0.94, 1.10)	-

Models are adjusted using inverse probability of treatment weights accounting for a comprehensive set of patient demographic, clinical, social, and environmental factors as well as year and office fixed effects. Relative differences are listed as ‘-‘ if the coefficient was not statistically significant at p<0.01.

^aCoefficients for visit count outcomes are from negative binomial models so coefficients have been exponentiated to reflect the percent difference in number of visits between groups.

When comparing MA plans, Per-Visit MA patients had shorter LOS by 1.02 days compared to Episodic MA (95% CI −1.37, −0.66). Per-Visit MA plans had 3.0% more physical therapy visits than Episodic MA plans (95% CI 1.01, 1.05) but there were no differences in nursing or other therapy visits between MA plans. Per-Visit MA plans had 6.8% fewer social work visits compared to Episodic MA (95% CI 0.90, 0.97), but HH aide visits were similar.

Odds ratios for patient outcomes across plans are included in Figure 2. Compared to TM, Episodic MA patients had 8% lower odds of improving in mobility function (95% CI 0.87, 0.98), 6% lower odds of improving in self-care function (95% CI 0.88, 0.99), no difference in community discharge, and 5% lower odds of transferring to an inpatient facility during the HH stay (95% CI 0.90, 0.99). Compared to TM, Per-Visit MA patients had no difference in functional improvement but had 6% higher odds of discharging to the community (95% CI 1.02, 1.10) and 6% higher odds of transferring to an inpatient facility (95% CI 1.02, 1.10). When comparing MA plans to each other, the only statistically significant difference was 12% higher odds of inpatient transfers for Per-Visit MA versus Episodic MA (95% CI 1.06, 1.18).

Figure 2.

Estimated adjusted differences in outcomes for home health Episodic Medicare Advantage (MA) beneficiaries compared to Traditional Medicare (TM) beneficiaries (Panel A), Per-Visit MA beneficiaries compared to TM (Panel B), and Per-Visit MA beneficiaries compared to Episodic MA. Effects are expressed as odds ratios with 95% confidence intervals. Logistic regression models include inverse probability of treatment weights that include all demographic, clinical, social and environmental characteristics from Table 1 as well as location and year fixed effects, an indicator for whether the home health stay spanned the implementation of the Medicare Patient-Driven Groupings Model in January 2020, and robust standard errors.

Results of sensitivity analyses without adjustment across clinical, social, and environmental domains are in Appendix Table 1. Less adjusted models had different results for care delivery and patient outcomes compared to primary analyses. Unlike primary analyses, differences in function between Episodic MA and TM were not significant, but community discharge differences were significant. Differences in community discharge between Per-Visit MA and TM were no longer significant in less adjusted models. Sizes of other significant effects also varied. Compared to primary analyses, 2022 sensitivity analyses found similar results for care delivery except for fewer differences in social work visits. Unlike primary analyses, differences in inpatient transfers and community discharge between MA and TM plans in 2022 were not significant, but there were better functional outcomes for Per-Visit compared to Episodic MA plans.

Discussion

This is the first study to estimate differences in HH care delivery and patient outcomes between TM and MA that accounts for the different ways MA insurers reimburse HH agencies. To strengthen the analysis, we accounted for demographic, clinical, social, and community differences between patients across insurer groups. We found substantial differences between TM, Episodic MA, and Per-Visit MA patients across all domains. Consistent with prior literature,^4,17,20 TM patients were the most clinically complex, but MA patients, especially those with Episodic MA, were more likely to be from marginalized racial and ethnic groups and have higher social risk. Notably, many studies comparing care delivery and outcomes between TM and MA do not account for differences in clinical severity measures or social determinants of health.^15,16,36 Especially considering our results differed when models were not robustly adjusted, these findings highlight the importance of including social determinants and OASIS-based clinical factors in addition to standard adjustment for comorbidity indices and demographics.^4,15,16,36

Consistent with previous work, we found shorter LOS and fewer visits from nursing, therapy, and HH aides for both MA plan types compared to TM.^15,17 In adjusted analyses, Per-Visit MA patients had the shortest LOS and the fewest visits from nursing, speech therapy, and social work compared to TM patients, while Episodic MA patients had the fewest visits from physical and occupational therapy and HH aides. When comparing the two MA plan types, Per-Visit MA plans had more physical therapy visits and fewer social work visits over shorter stays for similar patients compared to Episodic MA plans. These differences are likely related to how Per-Visit MA plans dictate specific numbers of visits by discipline (e.g., via commercially available or proprietary care guidelines), while Episodic MA plans allow agencies to determine the types and numbers of visits.

Differences in patient outcomes comparing MA and TM were mixed, however; worse outcomes were often seen for the MA plan that had the fewest visits from disciplines who target that specific outcome. For example, Episodic MA plans had the fewest physical and occupational therapy visits alongside the worst functional improvement outcomes. While Episodic MA plans allow agency flexibility in determining visit mix, Per-Visit MA plans had the fewest social work and nursing visits and relatively more physical therapy visits than Episodic MA, but the most adverse inpatient transfer events. This may be related to Per-Visit MA restrictive pre-authorization processes that limit the HH agency’s ability to send the discipline that can better address the specific issue causing the inpatient transfer. The higher number of physical therapy visits for Per-Visit MA stays compared to Episodic MA was consistent in 2022 sensitivity analyses, which also found better functional outcomes for Per-Visit compared to Episodic MA. However, the optimal number of visits, and mix of visits by discipline remains unknown, and future work should examine direct relationships between care delivery variables and HH outcomes.

Limitations

While our data from a non-profit HH company provided unique information on MA plan structures, each HH company negotiates separate reimbursement contracts with MA plans. Thus, the mix of MA reimbursement structures varies, and these findings may not generalize to all agencies. We might expect larger differences in the distribution of Episodic versus Per-Visit MA plans, as well as care delivery and patient outcomes, in for-profit agencies that are more likely to prioritize cost containment.^41–43 Future work with data from additional HH providers will be essential for determining whether findings extend to all patients. We could not account for use of third-party intermediaries contracted by MA plans to manage care, so findings do not reflect heterogeneity in other MA utilization management processes.

While differences between TM and MA patients in our sample were similar to previous literature, no comprehensive description of all MA patients receiving HH is available, so we cannot assess how our sample compares to the entire HH MA population.^4,44 While SDI measures community deprivation, no data on dual Medicare/Medicaid eligibility were available to adjust for socioeconomic status at the individual level. We did not have hospital data to include measures of hospital readmissions after HH, or to account for factors related to a preceding hospital stay. Finally, while IPTW is a robust approach to reducing selection bias, we could only adjust for observed covariates, and unmeasured differences likely still exist between payor types.

Conclusions

In the first study examining Episodic versus Per-Visit payments by MA insurers, we found both MA plans had fewer nursing, therapy, and HH aide visits versus TM. Compared to Episodic MA, Per-Visit MA patients received less social work, but more physical therapy. While differences in outcomes between TM and MA varied by MA plan type, Per-Visit MA plans had more adverse inpatient transfer outcomes than Episodic MA. MA plans that allow HH agencies flexibility in determining the delivery of visits may have fewer adverse outcomes compared to MA plans that dictate the amount and type of care provided. HH agencies negotiating with MA insurers may choose to prioritize episodic payments that have reduced administrative burdens and potentially fewer adverse outcomes than Per-Visit MA contracts.

Takeaway Points.

Medicare Advantage plans that restrict home health agencies’ ability to determine the number and type of visits delivered may have more adverse patient outcomes compared to Traditional Medicare and Medicare Advantage plans that provide episodic payments.

• Medicare Advantage plans that dictate the number and types of visits for home health providers have the shortest stays, fewer social work visits, and the highest percent of transfers to inpatient facilities during the home health stay.

• Home health agencies negotiating with Medicare Advantage plans can seek episodic payments to allow flexibility in determining care delivery and minimize adverse outcomes.