Association of Hospitals' Experience with Bundled Payment for Care Improvement Model with the Diffusion of Acute Hospital Care at Home

Abstract

Objective

To examine whether hospitals' experience in a prior payment model incentivizing care coordination is associated with their decision to adopt a new payment program for a care delivery innovation.

Data Sources

Data were sourced from Medicare fee‐for‐service claims in 2017, the list of participants in Bundled Payment for Care Improvement initiatives (BPCI and BPCI‐Advanced), the list of hospitals approved for Acute Hospital Care at Home (AHCaH) between November 2020 and August 2022, and the American Hospital Association Survey.

Study Design

Retrospective cohort study. Hospitals' adoption of AHCaH was measured as a function of hospitals' BPCI experiences. Hospitals' BPCI experiences were categorized into five mutually exclusive groups: (1) direct BPCI participation, (2) indirect participation through physician group practices (PGPs) after dropout, (3) indirect participation through PGPs only, (4) dropout only, and (5) no BPCI exposure.

Data Collection/Extraction Methods

All data are derived from pre‐existing sources. General acute hospitals eligible for both BPCI initiatives and AHCaH are included.

Principal Findings

Of 3248 hospitals included in the sample, 7% adopted AHCaH as of August 2022. Hospitals with direct BPCI experience had the highest adoption rate (17.7%), followed by those with indirect participation through BPCI physicians after dropout (11.8%), while those with no exposure to BPCI were least likely to participate (3.2%). Hospitals that adopted AHCaH were more likely to be located in communities where more peer hospitals participated in the program (median 10.8% vs. 0%). After controlling for covariates, the association of the adoption of AHCaH with indirect participation through physicians after dropout was as strong as with early BPCI adopter hospitals (average marginal effect: 5.9 vs. 6.2 pp, p < 0.05), but the other categories were not.

Conclusions

Hospitals that participated in the bundled payment model either directly or indirectly PGPs were more likely to adopt a care delivery innovation requiring similar competence in the next period.

What is known on this topic

• Hospitals make decisions to participate in a new program based on financial incentives and economic benefits.

What this study adds

• Hospitals that participated in a previous demonstration program, either directly or through associated physician group practices, were more likely to adopt a future innovation requiring similar competence in care coordination.

• Hospitals without exposure to the prior payment innovation were also the least responsive to the new payment program for home‐based acute care, suggesting persistent disparities in the diffusion process.

1. INTRODUCTION

Hospitals and physicians have been encouraged to participate in various payment reform initiatives run by private insurers and by the Center for Medicare and Medicaid Innovation (CMMI). ¹ Participating entities are given incentives to invest in care coordination and take broader management responsibility for healthcare spending and patients' health. ² , ³ , ⁴

Hospital‐at‐Home is a care delivery innovation with the potential to shift facility‐based services to less‐intensive alternative sites. Hospital‐at‐Home has been shown to offer highly cost‐effective hospital‐level inpatient care in a patient's home for acute illnesses. ⁵ , ⁶ , ⁷ , ⁸ In November 2020, Centers for Medicare and Medicaid Services (CMS) introduced the Acute Hospital Care at Home (AHCaH) program to pay for Hospital‐at‐Home for Medicare beneficiaries requiring inpatient care amid the COVID‐19 pandemic. ⁹ CMS grants hospital‐level approval to waive the 24/7 onsite nursing requirement for hospitals seeking to provide Hospital‐at‐Home services. ⁹ The waiver approval entails specific criteria to ensure hospitals are adequately prepared to implement the service with proper safeguards. ⁹ While the waiver program spurred renewed interest in Hospital‐at‐Home, not all hospitals were able to demonstrate their competence. This raises an important question—what are the characteristics of hospitals that respond to the invitation to develop a Hospital‐at‐Home program when the structural barrier—the lack of payment mechanism in Medicare fee‐for‐service– was eliminated and which ones were approved?

Prior empirical and theoretical research on innovation among organizations commonly viewed that the diffusion of innovation was influenced by the innovation‐specific technical and economic benefits to the organization or adopter‐specific characteristics. ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ This approach generally assumes that adoption decisions are independent of prior involvement in similar programs. ¹⁵ However, the adoption of emerging innovations in healthcare delivery and payment reforms require organizational changes that encompass broad behavioral modifications over the continuum of care, managerial disruptions, and partnership with organizations. ¹⁶ , ¹⁷ In addition, since these innovations generally target similar types of changes and actors, adoption decisions within the organization can be influenced by prior involvement. ¹⁷ , ¹⁸

The growing body of knowledge related to organizational change and innovation suggests that the adoption of change occurs over time. ¹⁵ , ¹⁹ , ²⁰ In other words, it is a continuous and interconnected process that unfolds across multiple periods. This is partly due to the fact that an organization's proficiency in the preceding period enhances its capacity for learning, ultimately leading to increased competence in the subsequent period. ¹⁶ , ¹⁹ , ²⁰ , ²¹ As seen in the waiver requirements for AHCaH, these changes require experience, infrastructure, workforce, and information systems that should already exist. ¹⁶ , ²¹ For example, these include daily in‐person and remote visit capability, advanced electronic medical record system accessible to all clinicians/ providers/staff, and contracted services, and reporting requirements. Therefore, it is reasonable to assume that these hospitals adopting AHCaH were more likely to have some experience in prior demonstrations necessitating advanced care coordination, along with the managerial competence to participate in the new program when the opportunity arose. ²⁰ In this scenario, hospitals' adoption of AHCaH is not a singular, isolated event of adoption, but rather an extension of their ongoing change process, built upon their previous experience in adopting new business models. This also implies that hospitals lacking prior experience might face being left behind.

A prior study examined early uptake of AHCaH, using some hospital‐level characteristics (e.g., size and ownership), but this study did not delve into the contextual enablers that could have increased the hospitals' readiness to adopt the new innovation. ²² Understanding these factors is critical for policymakers planning demonstrations and could help CMS and private insurers transform healthcare into a more value‐based system and increase participation rates especially for vulnerable hospitals.

1.1. Hospitals' prior change experience in payment reform

To investigate hospitals' decision to adopt the change to provide inpatient care at patients' home, this study employs hospitals' prior experience with payment innovations run by CMMI. The Bundled Payment for Care Improvement (BPCI) Initiative is CMMI's most extensive episode‐based voluntary payment program open to all eligible organizations without geographic restrictions (except for Maryland due to its all‐payer payment model). BPCI aims to incentivize participating hospitals or physician group practices (PGPs) to curtail spending over the care episode, encompassing services administered by other healthcare providers post‐hospital discharge. ³ Hospitals' indirect experience through participating PGPs has been largely underexamined in prior studies. The prior evaluations showed that hospitals generated savings primarily by reducing spending on institutional care services. ²³ , ²⁴ AHCaH is a payment program for home‐based inpatient services; thus, it requires hospitals' ability to coordinate care with a high level of care intensity for patients with acute conditions requiring hospitalization. Given the goal and design of both the episode‐based BPCI program and AHCaH, it is reasonable to hypothesize that hospitals that participated in BPCI are more ready to provide inpatient services outside hospital walls.

Using hospitals' experience in BPCI, which creates incentives for care coordination with non‐hospital healthcare care providers, this study examines whether hospitals' direct participation or indirect participation through physicians in a payment innovation is associated with their adoption of a new care delivery model during unexpected managerial challenges. It also explores whether hospitals located in the same geographic area or being part of a health system where more hospitals adopted the program increase the hospitals' own participation.

2. METHODS

2.1. Data and study sample

The list of hospitals approved for CMS AHCaH from November 2020 to August 2022 was obtained as the primary source of data to construct outcome variables. To construct explanatory variables on hospitals' experience with BPCI payment innovation, CMS participation files and Medicare claims data are used. Hospitals' indirect experience in BPCI was obtained from 20% nationally random sample of Medicare fee‐for‐service inpatient claims data from 2017, using the BPCI model codes. In Medicare fee‐for‐service claims, these codes were used to identify that a BPCI model was being used for payments of an inpatient claim.

To identify hospitals that had reported substantial BPCI volume, 50 BPCI episodes were used as a cutoff. Hospitals that had at least 50 BPCI claims in 2017 but were not included in the CMS participation files were identified as those having indirect participation through PGPs. The RAND Hospital Data derived from the CMS Hospital Cost Report was used to supplement information on utilization, cost, and other financial measures for Medicare‐certified hospitals. Other hospital characteristics were obtained from the American Hospital Association Annual Survey for the corresponding year.

Since this study examines hospitals' adoption of AHCaH as a function of prior experience in BPCI initiatives, this study focuses on general acute hospitals eligible for the BPCI initiative over the periods. Hospitals with unique payment methodologies were excluded from the BPCI program, and so were in this analysis. These included Prospective Payment System (PPS)‐Exempt Cancer Hospitals, Inpatient Psychiatric Facilities, Critical Access Hospitals, hospitals in Maryland, and hospitals participating in the Rural Community Hospital Demonstration and Participant Hospitals in the Pennsylvania Rural Health model. These selection criteria returned 3248 general acute care hospitals.

2.2. Variables

The primary outcome of interest was the hospitals' successful obtaining of the AHCaH waiver from November 2020 to August 2022.

2.2.1. Hospital's experience with care coordination under BPCI innovation

Key explanatory variable is the hospital's prior experience with the BPCI initiative as a manifestation of the hospital's interest and experience to manage a change in care delivery model. Nuanced experiences in the payment reform were defined based on three factors: Their revealed interest in payment reform incentivizing care coordination for treatment outside facility, type of change management experience (direct vs. indirect through physician agents from PGPs), and the dropout history in the program. Using the combination of these criteria, the hospitals were grouped into five mutually exclusive BPCI experience categories: (1) hospitals that directly participated in BPCI and completed the program (early adopters); (2) hospitals that initially revealed interest in BPCI but withdrew and then indirectly participated through physicians (indirect participants after dropout); (3) hospitals that did not show interest in BPCI but indirectly participated through physicians (indirect participation only); (4) hospitals that withdrew from the BPCI (dropout only); and (5) hospitals without any exposure to BPCI (no exposure). Hospitals without any exposure to BPCI were used as the primary control group. Hospitals' initial interest in BPCI was identified based on their participation history including the period before transitioning to the risk‐bearing phase. More details on the identification strategy are summarized in Appendix Methods 1 and Appendix Table 1.

2.2.2. Community's participation in AHCaH

The core‐based statistical area (CBSA) was used as the primary unit of the community. One example of CBSAs is the Washington‐Arlington‐Alexandria Metropolitan Statistical Area. The U.S. Office of Management and Budget defines CBSA as one or more adjacent counties with at least one urban core area of at least 10,000 people and adjacent territory that has high socioeconomic integration with the core by commuting. ²⁵ Considering that the adoption of CMS AHCaH necessitates substantiation of resource readiness and management experience, a higher prevalence of program adoption among hospitals within a community might indicate the greater readiness as well as peer pressure. To measure the prevalence of program adoption in the community, the percentage of other hospitals that adopted the model in the CBSA was measured. For a subgroup of hospitals that belong to health systems, the health system was used as a secondary unit of business community because hospitals in the same health system share the business leaderships and could have advanced information infrastructure and resources shared among the member hospitals in the system.

2.2.3. Other covariates

Other covariates include hospitals and communities' predisposing characteristics. In theory, adopter characteristics that influence the adoption decision can be identified into four categories: organization size, organization structure, organizational innovativeness, and slack resources. ¹⁷ , ²⁰ , ²⁶ In addition, change management theories emphasize the importance of enabling resources as critical factors determining change readiness and capabilities. ²¹ , ²⁷ , ²⁸ , ²⁹ Based on these frameworks, a set of covariates of organizational size, structure, innovativeness, and enabling resources was identified as the following: (1) size—the size of the hospital based on the number of beds; (2) structure—for‐profit hospitals, government hospitals, members of a health system (vs. independent hospital); (3) innovativeness—teaching hospitals (i.e., hospitals with major and minor teaching engagement based on the resident‐to‐bed ratio); and (4) slack resources—total profit margin. In addition, Medicare case‐mix index of the hospital was included to control for differences in severity and resource intensity of admitted patients.

While this study focuses not on population‐level CMMI models on primary care coordination or overall care management for individual patients, these programs may confound the relationship between hospitals' BPCI experience and participation in AHCaH. Thus, hospitals' participation in the Medicare Accountable Care Organization program is identified based on their inpatient claims reported with an ACO ID. Additionally, hospitals' direct participation in BPCI‐Advanced, a subsequent program of the BPCI initiative, is included as a covariate.

2.3. Analysis

First, an unadjusted analysis was conducted to assess the relationship between outcome and explanatory variables. To visually demonstrate how the adoption of AHCaH varies by the explanatory variables and if the relationship substantially changes over time, diffusion curves were constructed. These curves illustrated the unadjusted adoption rates of AHCaH for each quarter since the program's initial implementation. Subsequently, an adjusted multilevel logistic regression was performed, given that explanatory variables encompass various levels (hospital level and community level), and interactions between variables across levels are possible. To account for the variation across hospital‐community clusters in the likelihood of AHCaH adoption and to decompose the within‐community and between‐community fixed effects multilevel mixed effect regression was used. The regression results were converted to average marginal effect to estimate the variable's probability of participation in AHCaH upon a unit change in explanatory variables.

2.3.1. Subgroup analysis

First, the regression was repeated with the sample restricted to hospitals with a membership in health systems (78.6% of the sample) to assess the sensitivity of the geography‐based community definition. Second, hospitals enjoying a monopoly status might exhibit distinct behavior from those in highly competitive markets. To accommodate this distinction in hospital behavior, subgroup analyses excluding monopoly hospitals were conducted. Lastly, the regression was repeated with the subset of hospitals without any BPCI exposure, aiming to understand the factors influencing the decision of these hospitals to participate in AHCaH.

The goodness of fit of the models and relative quality of prediction were measured using log‐likelihood values and Wald tests. The variance inflation factor was calculated for each explanatory variable to examine multicollinearity among variables. Two‐tailed statistical tests were used and considered significant at α = 0.05. All analyses were performed using software package Stata version 17 (StataCorp LLC).

3. STUDY RESULTS

3.1. Hospitals' experience with BPCI at the baseline

Among eligible hospitals, about a third (31.5%) initially joined BPCI (Appendix Figure 1). Among these hospitals, about 76% withdrew, and the rest continued through the risk‐bearing period. Nearly half of non‐participant hospitals and 73% of the dropouts indirectly participated in BPCI through PGPs. As a result, 7.7% were direct BPCI participants, 17.5% were indirect BPCI participants through physicians after dropout, 34.6% were indirect participation alone, 6.3% were dropouts alone, and 34.0% did not have any exposure to BPCI.

3.2. Descriptive statistics of hospital characteristics

Figure 1 displays diffusion curves based on BPCI experience. It is evident that early BPCI adopters and hospitals that indirectly participated after dropout show a higher likelihood of adopting the program. These relationships remained consistent throughout the observed time period.

FIGURE 1.

Acute Hospital Care at Home (AHCaH) diffusion curves by hospitals' prior experience with Bundled Payment for Care Improvement (BPCI), November 2020–August 2022.

Among 3248 general acute hospitals eligible for BPCI programs, 232 (7%) hospitals adopted AHCaH (Table 1). Early BPCI adopter hospitals (17.7%) had the highest adoption rates and indirect participants through PGPs after dropout had the second highest adoption rate (11.8%). Hospitals with indirect participation only or dropout history only had 6.8% and 4.9% of adoption rates, respectively, while those without any exposure to the BPCI program had the lowest adoption rate (3.2%). The adopter hospitals were located in communities where more peer hospitals participated in the program compared to non‐adopter hospitals (median 10.8% [IQR 0–23.1] vs. 0% [IQR 0–4.0]). The relationship was even more pronounced in the adoption rate within the health system (median 27.5% [16.4–50.0] vs. 0 [0–4.7]). While the adopter hospitals exhibited a higher inpatient bed occupancy rate than non‐adopter hospitals (median 74.1% [60.3–85.4] vs. 63.3% [46.3–78.5]), there was no significant difference in the percentage of inpatient beds used for COVID‐19 cases and critical staffing shortages between the two groups. The adopter hospitals were more likely to be teaching, non‐profit hospitals with a larger bed capacity than non‐adopter hospitals. Furthermore, the adopter hospitals were affiliated with health systems (96.6% vs. 3.4%) and had a higher participation rate in the Medicare ACO (9.9 vs. 5.2) and BPCI‐Advanced programs (49.6 vs. 32.1).

TABLE 1.

Hospital and community characteristics by hospitals' implementation of Acute Hospital Care at Home (AHCaH) as of August 2022.

Variables	AHCaH adopter hospitals (N = 232)	Non‐adopter hospitals (N = 3016)	T‐test p‐value
Prior experience with Bundled Payment for Care Improvement (BPCI), %
Early adopter with full change management (n = 249)	19.0	6.8	<0.001
Indirect participation after dropout (n = 568)	28.9	16.6	<0.001
Indirect participation only (n = 1123)	32.8	34.7	0.546
Dropout only (n = 205)	4.3	6.5	0.193
No exposure (n = 1103)	15.1	35.4	<0.001
Community's adoption of AHCaH, %, median (IQR)
Share of other hospitals adopted in CBSA a	10.8 (0–23.1)	0 (0–4.0)	<0.001
Share of other hospitals adopted in the health system b	27.5 (16.4–50.0)	0 (0–4.7)	<0.001
Share of other hospitals adopted in the health system and CBSA
Hospital's inpatient bed and staffing shortages due to COVID‐19, %, median (IQR)
Percentage of inpatient bed occupancy rate c	74.1 (60.3–85.4)	63.3 (46.3–78.5)	<0.001
Percentage of inpatient beds used for COVID‐19 cases c	5.6 (3.1–8.8)	5.9 (2.5–10.1)	0.629
Percentage of hospitals with critical staffing shortages (state) b	11.9 (7.0–26.1)	18.6 (9.6–26.8)	0.336
Hospital characteristics, %
Teaching status
Major teaching hospitals c , %	19.8	10.4	<0.001
Minor teaching d , %	37.5	24.8	<0.001
No teaching	64.8	42.7	<0.001
Ownership, %
For‐profit hospitals	8.6	26.1	<0.001
Non‐profit hospitals	78.5	58.9	<0.001
Government hospitals	12.9	15.0	0.389
Metropolitan areas	88.8	75.6	<0.001
The number of beds, N median (IQR)	247 (124–450)	123 (53–248)	<0.001
Profit margin, % median (IQR)	6.5 (0.7–11.7)	4.3 (−1.8–11.5)	0.846
Medicare case‐mix index, median (IQR)	1.7 (1.5–1.9)	1.6 (1.4–1.8)	<0.001
Health system hospitals, %	96.6	3.4	<0.001
Members of accountable care organizations	9.9	5.2	0.003
Hospital's participation in BPCI‐advanced	49.6	32.1	<0.001

^a CBSA: Core‐based Statical Area.

^b This percentage was computed after excluding health system level adoption (N = 53).

^c Hospitals with the interns and residents‐to‐bed ratio >0.25 are defined as major teaching hospitals. Hospitals with the interns and residents‐to‐bed ratio < =0.25 and >0 are defined as minor teaching hospitals.

^d Quarterly mean of weekly values in the third quarter of 2020 based on the data obtained from COVID‐19 Reported Patient Impact and Hospital Capacity by Facility Timeseries, HealthData.Gov.

3.3. Adjusted association of explanatory variables with the adoption of AHCaH

In the adjusted regression, hospitals' experience with BPCI emerged as a significant predictor of AHCaH adoption (Figure 2). Direct BPCI participation was associated with a 5.9 percentage point increase in the average predicted probability of adopting AHCaH (95% CI 1.0–10.8, p = 0.018). The association was slightly greater for hospitals with the indirect participation through PGPs after the dropout group (6.2 pp, 95% CI 2.2–10.1, p = 0.002). The indirect participation alone or dropout alone was not a significant predictor (indirect participation only: 0.1 pp, 95% CI −2.8–3.0, p = 0.926; dropout only: 1.4 pp, 95% CI −2.6–5.4, p = 0.506).

FIGURE 2.

Adjusted Association of Hospitals' Prior Experience with Bundled Payment for Care Improvement (BPCI) with the Adoption of Acute Hospital Care at Home (AHCaH), November 2020–August 2022. Estimates are derived from a multilevel logistic regression model of the adoption of Acute Hospital Care at Home (AHCaH) among general acute hospitals to account for clustering by community (CBSA). Full regression results are presented in Appendix Table 2.

Hospitals were also more likely to adopt AHCaH when more peer hospitals in the CBSA adopted the program. A 10% increase in the prevalence of other hospitals that adopted AHCaH in the community was associated with a 2.9 percentage point increase in the hospital's average predicted probability of adopting the program (95% CI: 2.2–3.6, p < 0.001).

Among control variables, hospitals' affiliation with a health system (6.9 pp, p = 0.002) and hospitals' participation in BPCI‐Advanced (1.9 pp, p = 0.036) were positively associated, while for‐profit hospitals (−5.3 pp, p < 0.001) and the mean number of beds (−2.0, p = 0.010) in the community were negatively associated. Hospitals' participation in Medicare Accountable Care Organization was not significantly associated with the outcome of interest. Other variables were not statistically significant predictors of the outcome. Full regression results are presented in Appendix Table 2.

In the subgroup analysis of hospitals without BPCI exposure, hospitals were more likely to adopt AHCaH when more peer hospitals in the health system adopted the program. A 10% increase in the prevalence of other hospitals that adopted AHCaH in the system was associated with a 1.9 percentage point increase in the hospital's average predicted probability of adopting the program (Figure 3). Estimates from other subgroup analyses were consistent with estimates from the main regression (Appendix Table 3).

FIGURE 3.

Association between the Probability of Adopting Acute Hospital Care at Home (AHCaH) and the Prevalence of AHCaH adopters in the Health System Among the Subgroup of Hospitals without Prior Exposure to Bundled Payment for Care Improvement.

4. DISCUSSION

This study presents new evidence on the diffusion of care delivery innovations and its association with hospitals' prior experience in the CMMI payment model. First, this study found that hospitals' experience in payment reform is associated with the decision to implement a care delivery innovation requiring similar competence. The highest adoption rate of AHCaH was observed among hospitals that directly participated in BPCI. This finding suggests that the adoption of a care delivery innovation by hospitals is not an isolated event, but rather an extension of their ongoing change process, built upon their prior efforts.

Second, the study found that hospitals' experience with CMMI demonstrations involving physician participation can be comparably effective in motivating hospitals to adopt a change requiring similar competence in care coordination. Hospitals' indirect participation through participating PGPs in a prior CMS demonstration (BPCI) had a strong association with hospitals' participation in AHCaH if the hospital had initially demonstrated an interest in payment reform. This finding highlights the crucial role of physicians as pivotal change agents in the diffusion of payment and care delivery innovations. ¹⁵ According to the diffusion of innovation theory, change agents are personnel with a high degree of expertise in the innovation (i.e., MD or PhD) and facilitate the change for organizations when there are technical chasms between the change agents and client entities. ¹⁵ , ¹⁷ This study's finding further demonstrates that change agents are effective when the clients actively support the change.

As seen in the prior evaluations of the program, ²³ , ²⁴ , ³⁰ the effectiveness of bundled payment models is largely attributable to participants' ability to control service utilization post‐discharge. This study's findings suggest that hospitals that withdrew from the payment model, potentially because of their limited capabilities in managing extended episodes of care, may see the feasibility in coordinating care outside hospitals through PGPs making the changes for their patients.

Third, this study has identified disparities in the willingness of hospitals to engage in CMS change initiatives across different domains of healthcare. Hospitals without any exposure to BPCI are also less likely to implement the new care delivery model—AHCaH. Furthermore, these hospitals tend to be located in communities with a low prevalence of early adopters. Given that hospitals seeking the AHCaH waiver should demonstrate their competence in providing high‐quality home‐based care, the dose–response relationship between hospitals' implementation of AHCaH and the prevalence of adopters in the community suggests collective competence among hospitals that share resources and workforces. ²¹ , ²⁷ , ³¹ , ³² Consequently, hospitals without prior experience and located in communities with the low prevalence of adopters might face greater challenges to participate in future payment reforms requiring similar competence and resources.

These findings are relevant to the ongoing CMMI payment reform debates. We have observed that hospitals with substantial engagement in payment reform during prior periods are more likely to adopt care delivery innovations that have the potential to lower healthcare spending. While many hospitals aspire to participate in demonstration programs, those without prior exposure to similar change initiatives may lack the necessary expertise to perform well under the pressure of cost containment. Incorporating cost‐effective care delivery innovations into demonstration programs can empower hospitals to generate savings effectively and transform their care delivery practices. This will better equip them for the evolving landscape of value‐based payment reform.

Furthermore, this study shows that payment programs targeting a mix of providers and care coordination over the extended continuum of care may create spillovers over heterogeneous entities in the community, which contribute to building a value‐based echo system beyond the directly targeted entities and outcomes. It also implies that policy efforts will likely vary by hospitals' prior experience with CMMI programs. Given the inter‐organizational nature of the programs, current program evaluations focusing on effects within homogeneous organizations may underestimate the impact of the reform initiatives on the community. Adopting a more holistic approach to encompass the broader influence of these programs on the value‐based ecosystem beyond their immediate participants will be crucial for policymakers. This will aid in accurately assessing the return on investment of payment reform initiatives and inform the design of future programs.

Policies seeking to address disparities in participation and outcomes of the CMS reform must consider how different organizations develop their interest in the CMS reform and what changes in their care delivery adjustments are precursors for participation. ² In this context, tailored interventions and recruitment strategies based on hospitals' different baseline experiences with the prior payment reform could potentially serve as avenues to broaden program participation to hospitals without prior exposure. For example, CMMI could consider designing programs specifically targeting hospitals that have never participated in prior programs. This would involve providing more detailed training and resources to help them develop the necessary experience, partnerships, and infrastructure. As seen in the study, performance‐based financial incentives may not be enough to attract these hospitals to join. Similarly, CMMI can consider offering financial incentives to enhance the role of physicians as change agents within the hospital and community. This is especially relevant given their fluidity across multiple hospitals and their expertise in managing care coordination beyond hospital walls.

This study has several limitations that suggest directions for future research. First, this study did not examine the impact of these adoption decisions on innovation outcomes. It will be important to explore whether the outcomes associated with AHCaH and if the outcome is associated with prior hospital, physician, or community participation in payment reforms. Second, the study focuses only on the adoption decisions, which are the review outcomes for the waiver. There were two tiers of waiver requests‐expedited waivers (Tier 1) and detailed waivers (Tier 2) based on the hospitals' experience and readiness to offer inpatient services at patients' homes. If hospitals were ineligible for Tier 1, they were required to provide a more comprehensive preparedness plan to be approved by CMS. CMS did not release data on which hospitals received the waiver through the expedited track, the time spent to provide the evidence for the Tier 2 track, and hospitals that failed to get the approval despite their attempt or hospitals that were unwilling despite their competence. These data, if available, will provide insights into hospitals' nuanced readiness for the change initiative and payment model. However, even the hospitals that received the approval through the Tier 2 track could compile the evidence over a short period during the COVID‐19 pandemic, which shows these hospitals' advanced readiness or willingness to participate in the program. Third, the actual utilization of AHCaH may vary, but the data are unavailable. ³³ Fourth, this study did not examine the multifaceted mechanisms of the physician–hospital, community–hospital, and PGP hospital relationships influencing the hospitals' decision to adopt the AHCaH. However, the absence of statistical significance in the control variables suggests that prior experience has an important role. Fifth, while this study included hospitals' participation in BPCI‐Advanced and Medicare ACO, it did not investigate exposure to all other concurrent CMMI programs. This is due to this study's specific focus on the hospitals' experience with episode‐based bundled payment programs targeting hospitalization and related services, given the design and scope of AHCaH, the outcome of interest. Future research examining the exposure to the concurrent CMMI programs is warranted. Lastly, this study cannot establish causal inferences regarding the effect of the experience on the adoption of change. Instead, this study provides insights into how organizations differ in their baseline experiences, which was omitted in prior evaluations. Future research examining the effect of change agents on the diffusion of payment innovations is warranted.

5. CONCLUSIONS

Hospitals with change experience through their own participation or participating physicians in the prior period were more likely to adopt a new care delivery model. Future CMMI programs should consider hospitals' prior experience and community‐level competence in payment programs to accelerate the equitable diffusion of the programs.

FUNDING INFORMATION

This work was supported in part by the Arnold Ventures. The funders had no role in the collection of the data, analysis, interpretation, or reporting of the data or in the decision to submit the manuscript for publication.

Supporting information

Data S1. Appendix.

Kang S‐Y. Association of Hospitals' Experience with Bundled Payment for Care Improvement Model with the Diffusion of Acute Hospital Care at Home. Health Serv Res. 2024;59(4):e14302. doi: 10.1111/1475-6773.14302