Turning Readmission Reduction Policies into Results: Some Lessons from a Multistate Initiative to Reduce Readmissions

Abstract

Efforts are under way nationally to reduce avoidable hospital readmissions by changing payments to hospitals, but it is unclear how well or how quickly these policy changes will produce widespread reductions in hospital readmissions. To examine some of the challenges to implementing such approaches, the authors analyzed the early experiences of 3 statewide programs to reduce preventable readmissions that began in 2009. Based on interviews with program participants in 2011, the authors identified 3 key obstacles to progress: the difficulty of developing collaborative relationships across care settings, gaps in evidence for effective interventions, and deficits in quality improvement capabilities among some organizations. These findings underscore the uncertainty of success of current readmissions policies and suggest that immediate improvement in readmission rates through a change in reimbursement may be unlikely unless these other obstacles are addressed expeditiously. In particular, cultivation of productive collaboration across care settings will be critical because these kinds of relationships are not well established or naturally occurring in most communities. (Population Health Management 2013;16:255–260)

Introduction

Medicare is prioritizing reductions in hospital readmissions because they are pervasive, costly, and suggest poor quality care. Annually, 1 in 5 Medicare beneficiaries is readmitted within 30 days of discharge, costing Medicare roughly $18 billion.¹ Because many hospital readmissions (hereafter “readmissions”) are thought to be preventable through better managed, coordinated care,² recent policy changes aim to reduce readmissions through payment incentives and care transitions interventions. Beginning in October 2012, hospitals with “excess” 30-day readmissions for heart attacks, congestive heart failure, or pneumonia will receive lower payments from Medicare as part of the Hospital Readmissions Reduction Program.³ How to prevent “excess” readmissions is left up to each hospital. One effort to improve coordination, the federal Community-based Care Transitions program (CCTP), requires that grantees develop partnerships between the hospital and community organizations to identify and test interventions to improve care transitions for Medicare beneficiaries.⁴

It is unclear how well or how quickly these policies can produce widespread reductions in readmissions. First, it is uncertain how many readmissions are preventable. Estimates range from 5% to 79% of all readmissions.⁵ Nonetheless, there is a growing consensus that care transitions, in particular, are suboptimal and that pursuing improved care coordination and quality of care, especially across care settings, is worthwhile.^6–12 Second, although some hospitals have succeeded by focusing on improving processes of care during and after hospital discharge,^8–12 the effectiveness of such interventions has not been tested in a generalized sample of hospitals. As a result, there is no proven “recipe” for hospitals to implement.

In this article, the authors explore the promise of current policies to produce extensive reductions in readmissions by examining the early experiences of a large-scale intervention to reduce preventable readmissions that began in 2009. The STate Action on Avoidable Rehospitalizations (STAAR) program aimed to produce statewide reductions of 20%–30% in readmissions through a combination of individual provider, community, and statewide approaches in Massachusetts, Michigan, and Washington.^13,14 STAAR is one of the first programs to target such extensive reductions. As such, the STAAR experience provides timely and unique insight into the challenges of implementing current approaches to reduce preventable readmissions and of collaborative, cross-provider quality improvement efforts more broadly.

Study setting

STAAR is a $3.6 million 4-year program to reduce 30-day all-cause preventable hospital readmission rates by 20%–30% across Massachusetts, Michigan, and Washington states through a cooperative, statewide approach. Funded by the Commonwealth Fund, the Institute for Healthcare Improvement (IHI) developed STAAR to be led by each state's hospital association, with IHI providing technical assistance supported by the grant funds; none of the states received any of the grant funds directly. The state hospital associations in Massachusetts and Michigan joined forces with the following state organizations to lead STAAR: the Massachusetts Coalition for the Prevention of Medical Errors, the Massachusetts Medical Society, and the Bureau of Health Care Safety and Quality in Massachusetts, and the Michigan Peer Review Organization in Michigan.

In each state, STAAR consists of 2 major components to reduce readmissions: (1) hospital-anchored collaboratives to improve the quality of care delivered in the hospital and across care settings, and (2) statewide stakeholder collaboration to change health care policies to facilitate reduction efforts. Through the former, IHI provides education, training, and tools to STAAR state leaders and provider participants to design and implement their readmissions reduction programs. IHI educates participants about rapid quality improvement methods and tools, facilitates collaborative learning through regularly scheduled meetings, and provides a How-to Guide containing recommended interventions based on IHI's review of the literature and IHI's prior collaborative experiences. The core domains of the How-to Guide focus on hospital-based interventions targeting: (1) enhanced admission assessment, (2) effective teaching and learning, (3) real-time patient and family-centered discharge communication, and (4) posthospital care follow-up. STAAR participants had the flexibility to choose and adapt the STAAR collaborative interventions.

STAAR is distinct from other readmissions initiatives in its deliberate effort to incorporate providers across the continuum of care (eg, nursing homes, home health agencies, other community providers) rather than focusing only on hospitals. STAAR state leadership selected 17–28 hospitals to participate in each state's first cohort (a second cohort was planned after 2 years). Each participating hospital was expected to improve care within the hospital and to convene a “cross-continuum team” consisting of a wide array of providers to work together to improve care related to readmissions across settings. IHI trained 3–5 individuals, volunteered by each state's participating organizations, to be “improvement advisors” who assist participating hospitals with quality improvement processes, including development and use of their cross-continuum teams. Other technical assistance included IHI consultations and learning sessions through regularly scheduled group calls and webinars. IHI also maintained a STAAR-specific Web site and listservs as a way for participants to communicate with each other, and facilitated an annual face-to-face meeting in each state.

The other component of STAAR consisted of IHI's staff leading policy work with key stakeholders in each state (eg, state hospital associations, state governments) and nationally to identify policy and payment barriers and potential solutions. This piece was intended to complement the collaborative component by effecting larger, supportive delivery system changes in the following areas: (1) increasing population-based readmissions data, (2) developing financial incentive models for hospitals and other providers to reduce preventable readmissions, (3) aligning payers to implement these financial incentive models, and (4) working with providers across the continuum of care to reduce readmissions.

This article focuses on lessons from the collaborative component of STAAR because it directly addressed caregiver behavior and operated independently from the policy component during STAAR's early stages. Additional details about the STAAR initiative are available in Boutwell et al.¹³

Methods

The authors used a multiple-case study design to identify key factors likely to affect the success of readmissions reduction policies. This design facilitates the identification of both cross-cutting and unique factors by analyzing each case independently before triangulating the results across cases; this process helps assure that unique case context and its influence is considered effectively.¹⁵ The cases developed and triangulated for this study were the first 3 states participating in STAAR: Massachusetts, Michigan, and Washington.

The primary source of data was 52 semi-structured key informant interviews conducted in 2011, roughly 2 years after STAAR began. The authors conducted 60- to 90-minute interviews with 3 national program leaders, all 6 state STAAR initiative directors, and all 12 state program improvement advisors; 31 STAAR hospital participants (frontline providers and administration/executive staff) across 15 hospitals; and 6 additional stakeholders, including post-acute providers, members of professional associations, and other state health care policy leaders (Table 1). Hospital interviewees were selected to assure a mix of respondents from hospitals with varying characteristics that could affect implementation including hospital size, ownership, and historical readmissions rate performance; state geography; and level of participation in STAAR.

Table 1.

Completed Interviews by Informant Type (n=52)

	Massachusetts	Michigan	Washington	IHI
STAAR hospital executives and staff	14*	17	0	–
STAAR state leaders	2	2*	2*	–
STAAR state-level improvement advisors	4*	4*	3*	–
Key policy stakeholders	2	4	0	–
IHI program staff	–	–	–	3

^*Indicates that some interviewees in this category served in more than 1 role. As a result, the total number of unique respondents (n=52) is smaller than the sum of the individual categories.

IHI, Institute for Healthcare Improvement; STAAR, STate Action on Avoidable Rehospitalizations.

Interview topics included informant and organization characteristics and history of readmissions work; motivation for addressing readmissions; the theory of change for STAAR and program implementation; STAAR utility and impact; barriers and opportunities to address readmissions; and opinions about readmissions policy. Each interview was audio recorded and transcribed. Each transcript was then synthesized by an author using a standard memo template organized by topic areas (eg, motivation to address readmissions, program logic model, participant engagement in program activities, perceptions of progress) developed by the authors after identifying and comparing themes across the same 5 interviews.¹⁶ Subsequently, the authors produced memos for the remaining transcripts. The memos were entered into Atlas.ti 5.0, version 6.2 (ATLAS.ti Scientific Software Development GmbH, Berlin, Germany) and coded to organize data by topic area, state, and participant type (eg, hospital, state leader, improvement advisor, other provider, STAAR program staff).

The authors systematically analyzed the memos for key themes related to the potential for widespread, successful reduction of preventable readmissions based on reviews of data extracted by topic area, state, and respondent type. Multiple authors reviewed the same data, triangulating interview findings with additional data sources collected over the course of the initiative, including STAAR participants' progress reports, notes from the authors' observations of STAAR meetings, webinars, and listserv discussions.¹⁴ The authors met regularly to come to a consensus on findings. This study was approved by the Penn State Institutional Review Board.

Findings from STAAR

Developing collaborative cross-setting relationships are critical for major reductions in readmissions but it is “an enormous challenge”

STAAR stakeholders emphasized that coordination across the continuum of care is paramount to attain significant and sustainable reductions in preventable readmissions. Participants acknowledged that hospitals play a central role in reducing readmissions, but emphasized the need to think and act more broadly:

My [hospital] colleagues will fail utterly [at reducing readmissions substantially] if the only thing that they look to do is improve service delivery inside the four walls [of their hospital].

For example, STAAR participants highlighted that effective medication reconciliation could have an impact on preventing readmissions by preventing adverse events. But they spoke of the likelihood that success necessitates a less hospital-centric approach: “…Medication reconciliation has been really put forward…[as] more hospital centric, and quite honestly, in my humble opinion, I think medication reconciliation is something that has to happen at a community level.”

However, STAAR's efforts to facilitate such collaboration highlighted the very difficult and time-consuming nature of this path. Potential collaborators doubted the ability to forge productive collaborative relationships, had trepidation about organizations' hidden agendas, and were skeptical about the value of participating, especially given competing priorities for individual and organizational resources.

Outpatient providers' motivation to invest resources and collaborate to reduce readmissions was often limited, but challenged the hospitals as well:

I was talking to this group and they're like, we wanna partner with [you]—I said, “No, you need to fix your own processes first before you start partnering with us.” They don't seem to think [certain things] are a problem [and so I don't want to work with them] because I'm just spittin’ in the wind.

STAAR hospitals agreed that recruitment of providers across the care spectrum was “a real struggle” because these entities' (real or perceived) opportunity costs of investing in reducing readmissions was too high, in part because of the competitive nature of health care delivery and limited-to-no previous experience working together on quality improvement. Providers' altruism was helpful but not always sufficient to recruit participants:

[Hospital A] could not get [some outpatient providers] to the table. They were not interested. They had “no skin in the game,” as they kept on saying, so why should they care about whether they send somebody back [to the hospital]?

In particular, STAAR hospitals had serious problems recruiting and engaging primary care physicians, frontline nursing home staff, and caregivers from outside of their own organizational networks or health systems. Some hospitals embraced this challenge, arguing that proceeding with a smaller group was logistically easier and that they would build on their success later to fill in these gaps; other hospitals never stopped trying to recruit representatives from all the “key players.” Regardless, getting people “at the table” did not assure progress.

Often participants did not have preexisting relationships with other organizations nor did they have a meaningful understanding of each other's roles in the care process. Participants said they had to “build relationships that weren't there before or were there but in a different [transactional] way.” Many noted that there were significant gaps in understanding the different organizations' and providers' roles and capabilities, adding to the challenge of realizing productive meetings: “Initially, it was a period of discovery where everyone is sort of trying to understand what everyone else does and see if there's opportunity to reference different skill sets to get better outcomes.”

Cross-continuum teams also were stymied by having participants without the knowledge or authority to harness resources and make change: “We had people at the meetings that could not do anything with the information.” This problem caused one hospital to disband its original cross-continuum team and repopulate it using more explicit criteria: “[When] we invited [new] people we said in the [invitation], ‘If you can impact change in your facility [and] bring back some of the things that you're learning and make a difference in your [organization], then we want you at the table.’” Across the STAAR states, organizations were wary of each other's motives for collaboration: “I kept hearing people say, ‘All that [the nursing homes] are doing is marketing [at these cross-continuum meetings].’ And so, I'm like, ‘Okay, something is wrong with this.’ ”

Despite the challenges, STAAR participants credited their cross-continuum teams with fostering new or improved lines of communication; contributing to a deeper understanding of readmissions issues and different caregiver roles; and cultivating collaboration among caregivers, administrators, and other interested parties (eg, professional associations) that should lead to better, more coordinated care:

I see we can usefully put together groups of people who need to work together who traditionally don't sit together at all, and so we have a much better common understanding of the problem than [we] used to have.

It's so easy to assign blame to somebody else [for a readmission, but] you don't know the full story. For example, a hospital has a standard discharge form it faxes to nursing homes [and] either the wrong person received the fax or the faxes weren't going through, or the nursing home didn't have access to the prescribed medication…It's [the cross-continuum team] bringing everybody together; and being able to present your own perspective; and then figure out how you can make it work more cooperatively together…It's an eye-opening experience.

STAAR participants argued that the cross-continuum work has the most potential value but that it is not easy to do: “I would say have patience, because this is going to really test it. It's not a fast project—to make the changes that you really need to [improve care]. Get your community people involved in the beginning so that they're all on the same page and understand it…and know that it's going to take a lot more work than anyone really said it would.”

Specific interventions to reduce readmissions are mostly ahead of the evidence

IHI's How-to Guide represented a compilation of the available evidence on readmissions interventions and was acknowledged by STAAR participants as a helpful resource. Still, participants lamented the overall lack of evidence for interventions, saying “…the problem is that there's really not a lot of literature to support that if you implement something…readmissions will be reduced,” and “I'm not even sure that we have the evidence behind those [How-to Guide] interventions to support performing them and showing that there will be changes in the outcomes based on these interventions.” Participants reported that evidence is even more limited for interventions involving outpatient care: “Quite frankly, the evidence around rehospitalization prevention is not great…my sense is there are probably some big gaps in the current interventions that we have, especially in post-acute care.” As one STAAR participant observed, “There is a cacophony of solutions out there that really have not been proven on a large scale, which complicates identifying solutions that can be incented [by policy].”

Another obstacle to understanding extant evidence and generating new evidence was the definition of preventable readmissions and readmission rates. As one STAAR participant opined, “So we're all kinda sitting there going ok, ‘What the hell is preventable?’ How do we know which ones are preventable?” Another respondent, who also is involved in the leadership of a national quality improvement association, confirmed that these concerns are widespread, observing, “The question [consistently] pops up, ‘What do you count as a readmission?’ People across the United States are still trying to get that basic foundation of understanding and it hasn't happened at a national level yet.”

Organizations' quality improvement infrastructure and capabilities vary

Organizations that did not have the quality improvement infrastructure and expertise to help project leadership navigate the complexities associated with reducing preventable readmissions were especially vulnerable to problems implementing and sustaining their readmissions reduction work. Although hospitals volunteered for STAAR, several in each state were totally new to rapid-cycle quality improvement methodologies and even more had limited experience with quality improvement efforts working across the continuum of care with other providers. STAAR provided some training and support for these methods, but individuals from these less-experienced organizations struggled to make progress. They reported that they were overwhelmed by trying to adopt a new conceptualization of quality improvement while designing and implementing a new readmissions reduction program, and described how their organization's lack of quality improvement capacity undermined advancement:

I know that really using some rigor about your evaluation [of changes] is phenomenally important and that part is hard to learn. The rigor of rapid-cycle improvement—it's not rocket science—[but] the problem is a lotta people get stuck.

Our hospital couldn't pull it off [sending more frontline providers to quality improvement training]—it was just a capacity issue.

The big guys…they have a sophistication of approaching a problem that we maybe can't get to. They have resources and depth that—we do eighteen things and they may have eighteen people doing that one thing.

However, larger and/or more experienced organizations were not immune to the challenges of successfully leveraging resources and sustaining progress: “I'm limited in what I can do with hours that I have, [so] I have to be very, very selective about what we participate in. I only have a small team of persons that I trust carrying [out] these programs,” and, “You lose that [experienced] person off the unit [and] then everything falls apart.” Still, the more experienced organizations seemed less defeated overall, perhaps because they appeared:

(a) more familiar with the ups and downs of the improvement process and the commitment entailed:

When you start delving into [reducing readmissions]—this is not a project, this is a journey. This is five years to literally change everybody's function and bake it into the processes.

(b) more comfortable with being proactive and innovative:

Because of limited capital, it was really difficult for us to establish [a new process in which the pharmacy technicians reconcile medications at admission]. But, we kind of scrunched and moved [people] to manage with the FTEs [full-time equivalents] we had; we shifted, thinking that if we do [thorough reconciliation] on the front end, we might not need to do as much at the end.

(c) better able to obtain broader buy-in, especially among organizational members with influence who can help manage competing priorities and sustain the initiative:

I feel bad [when I hear some of the less-experienced STAAR team leaders talk]. I'm like, you're going to get nowhere unless you have a COO [chief operating officer] or a president behind you; because you can't do this [alone]. I'm not in charge of nursing, case management, social workers, or the ED, but they all had to be part of this [STAAR]. I had to pull all of these people together…and [my] president said do it. I went to the president and said, “It's been 3 months and [the new vice president] is not coming to any of the meetings, this is bull”; man, the president walked into her office and the vice president has not missed a meeting since; but what would I have done [otherwise]?

Discussion

Currently, the Centers for Medicare and Medicaid Services' Hospital Readmissions Reduction program, which reduces payments to hospitals with “excess” readmissions, is estimated to improve quality of care and save Medicare $1.1–$1.5 billion a year starting in 2015.¹⁷ This change to Medicare payment policy has succeeded in raising attention to preventable readmissions as a quality issue among hospitals, and commercial payers and Medicaid programs are also in various stages of developing similar programs aimed at hospitals.^18–20

The CCTP mentioned earlier is part of the larger federal Partnership for Patients program that targets readmissions reductions by decreasing complications during care transitions. Because the Partnership for Patients only began in 2011, lessons from these interventions will not be available for some time. As such, the first 2 years of experiences of STAAR participants provides some early insight into the potential for hospitals to respond to current readmission reduction policies. The authors identified 3 important challenges to their success.

First, readmissions policies and interventions will need to consider explicitly how to develop and support meaningful collaboration across care settings because these relationships are not well established or naturally occurring in most communities. The STAAR experience highlighted the difficulty and time-consuming nature of forging these relationships even when providers agree that improved coordination is essential to improving quality of care. Organizations were concerned about investing in a voluntary endeavor for which the potential (business) rewards were skewed toward the hospital. This suggests that voluntary collaborative efforts to improve quality will be difficult to maintain, jeopardizing their ability to achieve significant, widespread change on their own.

Policies need to consider the economic incentives for coordinating care and how to help cultivate productive human relationships to improve quality across settings, which involves building mutual trust and respect, especially when the relationship is voluntary and benefits accrue differently and at different rates for the parties. The CCTP and the Partnership for Patients program emphasize provider partnerships,^21,22 and it is important that practical lessons for developing these partnerships are identified and built on.

Second, the STAAR experience suggests that the ability of providers and policymakers to successfully affect quality improving reductions in readmissions under current policy is constrained by a lack of generalizable evidence. This observation is confirmed by a 2011 literature review that found that no individual intervention or bundle of interventions was consistently associated with readmissions reductions on a wide scale.²³ In addition, the uncertainty about an intervention's effectiveness may delay its adoption.²⁴

Unfortunately, many current efforts also are not designed to generate the evidence needed. Readmission initiatives, including STAAR and Medicare's upcoming pay-for-performance program, have been designed to promote flexibility in intervention selection, adaptation, and implementation; this may enable providers to craft improvement programs that best meet their needs and capabilities but limits opportunities to compare the impact of standardized interventions that could be applied by other providers. An additional obstacle is that programs and their participants are using varied definitions for key outcomes: readmission rates and preventable readmissions. A common definition would help with assessment but it also must be meaningful.²⁵

Third, the findings of this study underscore that building quality improvement knowledge and infrastructure takes time, as does identifying interventions that produce results for a particular organization. As a result, hospitals with limited quality improvement capabilities are more likely to be at a disadvantage under current policy relative to more experienced, better resourced peers. Poorly performing hospitals that improve more slowly than their counterparts will incur financial penalties under Medicare. This might further diminish these organizations' ability to make quality improvement changes. Policymakers should consider additional incentives to support improvement across all hospitals, such as incentives for degree of improvement and resources for training in quality improvement methodologies.

As a result, current policies designed to produce widespread reductions in preventable readmissions will have a limited effect until the evidence base for readmissions interventions and their implementation is more developed. To support evidence building, policymakers should support efforts to generate, synthesize, and share information about readmissions initiatives and better practices across providers.

Conclusion

Reduction of preventable readmissions is seen as an opportunity to address costly and often uncoordinated care. The findings presented here caution us about the speed with which preventable readmissions can be reduced through improved care coordination under current policy, given the lack of evidence on effective strategies, the need for cultural changes, and the need for new and better coordination with community providers. The early experiences of STAAR maintain the value of pursuing this goal¹³ but also underscore many of the challenges to improvement. Although changing the underlying economic incentives could facilitate better care coordination, many processes need adjustment and it will take extensive effort to derive effective evidence-based approaches.

Author Disclosure Statement

This research was supported by The Commonwealth Fund. Drs. Mittler, Press, Volpp, and Scanlon, Ms. O'Hora, and Ms. Harvey declared no conflicts of interest with respect to the research, authorship, and/or publication of this article.