https://www.ajmc.com/journals/supplement/2012/af4q/a413_12sep_mchugh_s156to62
Approaches to Improving Healthcare Delivery by Multi-stakeholder Alliances

Megan C. McHugh, PhD; Jillian B. Harvey, MPH; Dasha Aseyev, BS; Jeffrey A. Alexander, PhD; Jeff Beich, PhD; and Dennis P. Scanlon, PhD

Introduction  
Over the past decade, there has been much effort devoted to improving healthcare quality in response to compelling evidence that our health system does not perform as well as it could or should.1-3 Healthcare providers are increasingly engaging in a variety of quality improvement (QI) efforts,4 and have been encouraged to do so by the growing number of pay-for-performance programs and QI resources available.5,6 For example, the Institute for Healthcare Improvement, Agency for Healthcare research and Quality, and industry groups have produced guidance on various QI topics, and many public and private organizations have established regional collaboratives to help improve care for specific clinical conditions.7-9  

Although these QI efforts have resulted in many successes, quality problems persist.10-12 some have suggested that in order to stimulate meaningful and sustainable improvement that impacts the population in a given region, QI efforts need to advance from organizational-level initiatives to multi-level (individual, group, organization, and system), communitywide approaches.13-15 there is no single definition of a communitywide approach to QI, but examples include multidisciplinary teams from local hospitals working together through a learning collaborative to reduce infection rates; “horizontal” efforts among providers within a community (eg, hospitals, outpatient clinics, physician practices) to improve care transitions; and payment reforms that encourage local providers to implement patient-centered medical homes (PCMHs). A defining characteristic of communitywide QI is a coordinated, collective effort across organizations. Communitywide QI may be more effective than individual approaches if it eliminates duplication of efforts, improves information sharing across settings, and influences key factors that are out of the control of individual providers (eg, payment reforms).15-17  

The nascent literature on communitywide QI suggests that communitywide QI is possible, but implementation is challenging and requires complex interaction among organizations.15,17 there is currently little information to guide strategy and implementation of communitywide QI. The purpose of this paper is to describe how 14 alliances (multi-stakeholder partnerships in each Aligning Forces for Quality [AF4Q] community/location) participating in the AF4Q initiative approached the charge of improving healthcare delivery within their communities. We review the specific activities undertaken by the 14 AF4Q alliances (data from additional alliances that joined the initiative in 2009-2010 are not included in this analysis) and assess the extent to which they followed a common, communitywide vision for quality. This article will provide policy makers and leaders of QI efforts with an understanding of the activities that alliances can undertake to improve care delivery. It will also provide insight to policy makers and funders about the QI priorities of alliances.  

The AF4Q Initiative
The AF4Q initiative is a $300 million effort created by the Robert Wood Johnson Foundation (RWJF) under the premise that “No single person, group, or profession can improve health and healthcare throughout a community without the support of others.”18 Program funding was directed to alliances charged with improving quality at the community level by focusing on several programmatic areas including public reporting, equity, consumer engagement, and QI (ie, improving care delivery). Given their broad representation, alliances are well suited to facilitate improvement by securing and coordinating resources, promoting collaboration across providers, disseminating information, and prioritizing common goals and initiatives.19 the alliance’s governing body was ultimately responsible for providing vision, strategic direction, and operational oversight for the QI work, as well as linking QI with the other AF4Q programmatic areas. A more thorough description of the AF4Q initiative and characteristics of the alliances may be found in the article by Scanlon et al in this supplement20; here we focus on the AF4Q activities and expectations related to the QI programmatic area (Table 1).  

While guided by the AF4Q National program Office’s (NPO’s) technical assistance (TA), program goals, and measures, alliances had considerable latitude regarding how to pursue their work. For example, alliances could establish their own activities, partner with other organizations, or use a combination of the 2 approaches. They also had the flexibility to focus on any number of outcomes (eg, improving outcomes for diabetes or heart failure, minimizing use of the emergency department for potentially avoidable conditions).

Methods  
Data Sources  
Key informant interviews were conducted during 2 rounds of site visits and 1 round of follow-up telephone interviews. The first round of site visits was held from November 2006 to September 2007, shortly after the alliances were awarded AF4Q initiative funding. Although in-person interviews were conducted with a number of individuals in each community during the site visit, for this analysis, we limited our focus to the individuals most knowledgeable about the QI activities: alliance directors, who oversee the work of the alliance; and project directors, who are responsible for implementation of the AF4Q initiative. We asked respondents about the alliances’ structure, vision, goals, strategies, and decision-making.  

The second round of site visits was conducted from December 2009 to April 2010. At that time, the alliances identified specific individuals responsible for QI. Therefore, we interviewed the new QI leaders in addition to the alliance and project directors. During the interviews, we asked specific questions about the alliance’s QI activities, including strategies for leading versus partnering to conduct QI.  

Finally, we conducted follow-up telephone interviews with the alliance and project directors in the fall of 2010 to collect data on recent QI activities and why those specific activities were undertaken. Our analysis includes 16 interviews conducted during the first round of site visits, 54 during the second round, and 14 during the follow-up telephone interviews.  

Additionally, we reviewed a number of program documents and materials submitted by the alliances to the RWJF and the NPO between 2006 and 2010, including funding proposals, work plans, and progress reports. A more complete description of the data sources can be accessed in the manuscript by Scanlon et al in this supplement.21

Analysis  
All key informant interviews were recorded, transcribed, and uploaded into Atlas.ti. Our analysis included 4 steps. First, transcripts were reviewed and coded with predefined global codes that aligned with the AF4Q initiative’s initial programmatic areas (eg, QI, public reporting, consumer engagement). Second, following an inductive approach, a more finite set of codes was derived from a review of the transcripts to reflect the emerging QI themes.22,23 these codes reflected (1) alliance QI activities; (2) alliances’ QI strategy, vision, and goals; (3) the QI environment and context; and (4) TA, the NPO, and the RWJF. Third, these finite codes were applied to the transcripts by a researcher and reviewed for consistency by evaluation team investigators. Fourth, 2 investigators reviewed the passages and held weekly discussions to formulate the final themes.  

We also developed an inventory of the alliances’ QI activities. We organized the inventory following Leatherman and Sutherland’s taxonomy of quality-enhancing interventions, which was developed based on structured reviews of evidence and is applicable to organizational and community-level QI activities.14 two investigators independently categorized the alliances’ QI activities, and discrepancies were discussed. to be consistent with the time frame of our interview data, only interventions that were implemented prior to 2011 were included in the analysis. Interventions that were in the planning stages were excluded, and interventions that were implemented in multiple communities (eg, the transforming Care at the bedside collaborative) were counted multiple times. Activities that primarily focused on a different AF4Q programmatic area (eg, public reporting, consumer engagement) were excluded; however, QI activities that had a consumer engagement or public reporting component were retained. Categories were not mutually exclusive; most interventions were counted in more than 1 category. QI interventions that expanded over time were only counted once and their categorization reflects the expanded scope. We calculated the number of activities undertaken by the alliances in each category and the percent of QI activities that were led by the alliances, the NPO, or a local partner.

This research was approved by the institutional review boards of Penn State University and Northwestern University and informed consent was obtained from interviewees.

Results
Evolution of the Alliances’ Communitywide QI Activities
Many alliances were slow to establish plans for QI. From 2006 to 2007, alliances were in the early planning stages of the AF4Q initiative. Alliances commonly engaged in local convening activities to better understand the QI efforts under way in the community and to discuss with partners the activities that might make the biggest impact, given limited resources. Few of the alliances began new QI activities during this stage. even during the 2009 to 2010 site visits, many respondents noted that they continued to be in a planning stage for QI. In some communities, QI committees were just established and trying to identify a clear role; others struggled with how to use public reporting data to guide QI activities, and many respondents commented that their QI activities were just getting under way. According to a respondent, “We are really slow in QI.”

Respondents offered several reasons why they experienced more difficulty developing plans for QI compared with other AF4Q programmatic areas. First, respondents noted an initial lack of QI TA, numerous changes to the QI TA format, and poor communication of TA programs and processes. In the absence of clear guidance or expectations from the RWJF or the NPO, alliances struggled to focus their activities. According to a respondent, “the QI area is the one that I struggle with, and I feel like the Foundation struggles with it, and our TA providers struggle with it.”

Second, several alliances conducted surveys or led regional needs assessments to identify priority areas for QI, and those activities were time consuming. Finally, external factors, such as the passage of healthcare reform and new grant opportunities through regional extension centers (RECs), the beacon Community program, and others made it difficult for alliances to commit to a QI strategy and scope early on. “The overall picture isn’t quite clear. There are reasons for that. We just got the REC and we’re waiting [to hear about] the beacon [Community program]. There’s a lot of unknown and uncertainty.”  

Early QI activities built upon existing resources in the community; over time, alliances were more strategic in the selection of QI activities. Few alliances developed a clearly defined, communitywide vision for quality that was used to guide the selection of QI activities. Instead, early activities built upon ongoing interventions or took advantage of resources or expertise within the community. For example, respondents in Cleveland said that they had local experts who had previously led PCMH initiatives and improving performance in practice efforts, so they decided to build upon those activities. Other alliances delegated AF4Q’s QI programmatic area to partner organizations with a long history of QI. A respondent noted, “We already had a strong QI infrastructure [in the community], so we let others…continue to take the lead. They fulfilled the QI requirements of the [AF4Q] grant.”  

From 2009 to 2010, respondents were more strategic about their selection of QI activities; for example, they pursued activities that aligned with other large grant opportunities such as the beacon Community program and the REC. The additional funding allowed alliances to expand the reach of their QI activities. Selection of activities was also motivated by federal payment reform efforts. Respondents viewed the pilot and demonstration projects created under the Affordable Care Act (ACA) as key opportunities to support the creation of PCMHs and enter into discussion with local providers about the creation of accountable care organizations. Many of the alliances also focused on readmissions. According to a respondent, “these are the things that everyone is interested in—from the most meager federally qualified health centers to the most well resourced institutions. Everyone needs to reduce readmissions because we’re going to get hammered on payment. So those are things we are trying to help out with.”  

Despite the slow start in the QI programmatic area, by 2010, QI plans became more sophisticated. By 2010, many alliances expanded activities beyond learning collaboratives and focused on more sophisticated QI activities, such as those involving healthcare across different settings. They also began to consider strategies for spread and sustainability. For many alliances, large grants and payment reform efforts were central to the sustainability plans. Additionally, some alliances began to consider how to target hard-to-reach providers, such as smaller hospitals and independent physician practices.  

Classification of QI Activities  
The majority of QI activities focused on individual, organizational-level interventions. Using the Leatherman and Sutherland taxonomy to classify alliances’ QI activities, we found that alliances most commonly implemented activities aimed at changing organizational culture and behaviors (Table 2). Often, these activities were implemented through learning collaboratives facilitated locally and by the NPO. Some alliances also offered targeted coaching for individual practices.  

A common form of organizational change was the adoption of PCMHs. There were several reasons why this strategy was attractive to the alliances: the respondents believed there was an evidence base for PCMHs; there was a relatively high level of interest in PCMH models from providers; and development of PCMH activities positioned the alliances and their partners to take advantage of ACA programs and other large grants, such as the beacon Community Program.  

A majority of QI activities were data-driven. Almost all alliance leaders reported that QI activities incorporated performance data (Table 2). For example, practice-coaching activities for ambulatory care providers typically included a focused look at the practice’s data to direct the specific interventions. Additionally, alliances targeted low-performing practices to receive additional assistance, and high-performing practices were encouraged to share best practices. Similarly, many of the learning collaboratives sponsored by the NPO included requirements for the participants to submit data. The NPO analyzed the data and provided the participating organizations with feedback.  

Although the AF4Q initiative was initially designed for public reporting and QI activities to be aligned and carried out simultaneously, the alliances were offered more tA and guidance in the area of public reporting at the outset of the program, and many respondents described a sequential relationship between public reporting and QI, which also may explain delays in QI planning. For example, “QI is dependent upon first having good public reporting and health information technology” and “the collaborative learning group felt that they had to wait until the performance measurement people were further along.”  

Alliances infrequently pursued regulation or incentives. Across the 14 AF4Q communities, there were 13 efforts aimed at implementing financial incentives. These efforts typically involved projects to persuade local payers to provide financial incentives for establishing PCMHs. Regulatory interventions took the form of coaching to prepare practices for PCMH accreditation by the National Committee for Quality Assurance. Although regulation and incentives are powerful and wide-reaching motivators for community-level QI, alliances’ representation from payers and state agencies was rather limited, largely prohibiting these activities.

Alliances were just as likely to support the QI activities of local partners as they were to initiate their own QI activities. Many alliances noted that they did not want to duplicate the QI work already taking place within the community. Collectively, alliances supported a variety of local organizations (providers, health systems, state and local associations, government agencies, other QI organizations, and health plans) engaged in QI activities. The roles of the alliance in these activities varied from serving on an advisory board to providing data so that that the lead organization could monitor progress.

Notably, by 2010, several respondents suggested that their ability to partner with other organizations had improved due to greater local awareness of the alliance and/or an enhanced reputation as a QI leader. A respondent from Maine noted that they were approached by the recipient of the REC grant and asked to participate because of their increased credibility. A respondent from Cleveland said that the alliance was now viewed not only as a resource for QI, but also as a skilled convener in the community.

There was considerable variation in QI activities across communities. the number of QI activities adopted by alliances ranged from 5 to 21, reflecting differences in alliances’ histories, relationships with community partners, and QI planning times. there was also considerable variation across alliances regarding the decision to lead QI activities versus supporting the QI efforts of local partners. However, the alliances that were more likely to support partners’ QI efforts (rather than lead their own efforts) had well-established relationships with QI organizations, hospital associations, or others with a strong QI reputation. Still, all alliances led at least 1 QI activity.

Limitations
Our investigation examined alliances’ approaches to their QI task and the specific activities undertaken. Alliances could not consistently provide information on the scope of their activities and we caution readers against making assumptions about alliances’ success based on the number of activities initiated. Further, our analysis excluded activities that were primarily focused on a different programmatic area (eg, public reporting, consumer engagement), even though alliance and program directors may have considered those to be QI activities. Finally, our results are not necessarily generalizable, due to the funding and TA provided to the alliances by the RWJF and the NPO.  

Discussion  
Our analysis of the AF4Q alliances’ approaches to addressing QI offers several insights that may be beneficial to policy makers and program planners seeking to promote or establish communitywide QI. First, the protracted planning period for QI, in comparison with other programmatic areas, suggests that program planners should give careful thought to implementation timelines. Given the complexity of improving care delivery within a community, and the inherent difficulties of working within a multi-stakeholder alliance, it may not be realistic to have a clear plan within the first year. Communitywide QI efforts are likely to require multi-year tracking to identify significant implementation progress, and even longer to assess impact. Based on feedback from the alliance leaders, we also suggest that program planners develop clear expectations for alliances and offer TA that begins at the start of the program. We might have seen earlier progress in QI if the RWJF and the NPO had provided alliances with as much up-front TA for QI as they did for public reporting.  

Second, absent strong direction from program sponsors, our findings provide some information on the types of activities that are likely to be attractive to alliances. Initially, alliances may be inclined to build upon the resources and strengths that exist within a community. If program planners wish to direct efforts to new areas, possible strategies could include providing the groups with the evidence base for the activity, offering targeted TA, and highlighting additional funding sources. Based on our findings, these more practical and strategic factors may provide stronger motivation for selecting particular interventions than an abstract vision for QI within a community. Although it is sensible to build on the local strengths and resources, communities should be cautious that without a communitywide vision, their work might not meet the needs of the community.  

Third, alliances overwhelmingly chose to pursue strategies that focused on organizational-level change. Collectively, these changes may produce meaningful and coordinated improvements that impact the population within a community. However, one concern is that by focusing on institutions, participants may continue to narrowly focus on their own organizations, rather than attempt more complex interventions that require greater collaboration across providers, such as addressing care transitions. Program planners should consider pushing alliances to pursue more cross-organizational activities.

Fourth, our results demonstrate the important role of performance measurement in the selection and ongoing management of QI activities. Notably, several of our respondents suggested that the relationship between measurement and QI was sequential. It is not clear from our analysis whether communities with greater capacities for performance measurement and public reporting are more successful in their QI efforts than others, but this could be an area for future research. In addition to QI, program planners might find it beneficial to support performance measurement and public reporting, through funding and TA.

Finally, our results point to the critical role of partners in conducting QI activities and obtaining additional grant funding to support those activities. Collectively, alliances were just as likely to support the QI efforts of their local partners as they were to lead their own efforts. Again, our results do not suggest whether alliance-led strategies are more successful than other strategies or whether alliances with more partner-led activities are able to conduct QI activities more efficiently. However, the results do suggest that there are many models that alliances may use to approach QI—as a leader, partner, or some combination of the 2.  

Despite the inherent challenges associated with establishing communitywide QI, it appears to be gaining favor in the public and private sectors. The early experiences of the AF4Q initiative suggest that communitywide QI is possible; however, alliances are primarily focused on collectively improving quality within organizations, rather than in initiating cross-organizational interventions. More research is needed to determine whether these approaches are sufficient to address clear gaps in quality.   
Print | AJMC Printing...