Robert W. Dubois, MD, PhD

Concerns about rising healthcare spending and persistent low-value care are neither new nor unique to the United States. In the United Kingdom, Canada, and Australia, the government (or agencies deputized by it) routinely assesses the cost-effectiveness of new services, with a particular focus on medications.

Cost-effectiveness reflects the incremental cost of a service compared with its incremental benefit, often termed the “incremental cost-effectiveness ratio.” In the United Kingdom, the National Institute for Health and Care Excellence (NICE) commissions these cost-effectiveness assessments and makes recommendations to the National Health Service regarding what to fund. In the United States, this role is filled—albeit less formally&mdash;by the Institute for Clinical and Economic Review (ICER).

The approach to value assessment in the United States and the United Kingdom raises 2 concerns. First, an analysis—such as those from NICE and ICER&mdash;producing a single incremental cost-effectiveness ratio assumes that all patients are the same. Patients differ in clinical severity, level of functional impairment, the value they place on length and quality of life, ability to work, and preferences toward the complexity of a treatment regimen or the side effects that may ensue.

Given this myriad of differences, a single incremental cost-effectiveness ratio, or even a few subgroup analyses, cannot adequately reflect what is important to a diverse group of patients. To reflect those important differences, a range of values or multiple incremental cost-effectiveness ratios are needed.

Of equal concern, only a single organization effectively coordinates cost-effectiveness assessments in each country: NICE, by law in the UK; and ICER, by practice in the United States, where reviews are used by commercial and government payers. Having a single entity assumes that the incremental cost-effectiveness ratios they produce are both reliable and valid. But if 2 analyses were performed at the same time by different groups, would the results be the same? Close? Or quite different? Several recent analyses suggest that conclusions may differ.

Perhaps most notable is a new comparison of cost-effectiveness assessments produced for NICE with those produced by ICER. Comparing the results of each effort, substantial differences were noted, with analyses of the same drug producing incremental cost-effectiveness ratios that varied 3- to 6-fold. Some of the difference could be explained by differing prices in the 2 countries, but choices made during the modeling efforts are also likely to have caused markedly varying conclusions.

Similarly, the Asthma and Allergy Foundation of America reanalyzed ICER assessments for several new asthma therapies, slightly varying the assumptions used. The AAFA found that by incorporating the effect of the disease on productivity, what was deemed overpriced by ICER’s assessments was now a good value using ICER’s threshold.

One size can’t fit all: A single incremental cost-effectiveness ratio cannot reflect the value of new treatments for a diversity of patients, and one group performing those assessments will not suffice. In brief, it would be far NICE(r) to have multiple ICERs, or have 

 groups. Patients, providers, and payers would then review those multiple assessments and make more educated decisions—and more targeted decisions&mdash;based upon them.

1. Cockerill KE, Ward K, Higgins A, Gaebler JA. A comparison between ICER and NICE cost-effectiveness models. Poster presented at: ISPOR 24th Annual International Meeting; May 21, 2019; New Orleans, LA.

2. Mendez K. ICER can do better for patients. 

. 2019;25(5):514-516. doi: 10.18553/jmcp.2019.25.5.514.

Articles

From the Editorial Board: Robert W. Dubois, MD, PhD

 To update an ongoing assessment of care pathway development, implementation, and evaluation, and to evaluate the emerging relationship between care pathways and other components of value-based care.

 Targeted literature review followed by an online survey and in-depth interviews.

 The PubMed/Cochrane databases and gray literature were searched for publications on care pathways (January 1, 2014, to March 3, 2017); a supplemental targeted search was completed in October 2017. Qualitative data were collected via an online survey and semistructured, in-depth interviews with payers, providers, pathway vendors, and opinion leaders.

 A total of 112 articles or posters were identified in recently published research. The survey and interviews included 32 and 19 respondents, respectively. Care pathways are increasingly driven by providers and provider networks. Overall, we found increased awareness of and adherence to codified best practices or standards, and prioritization of high-quality evidence during development. Research findings suggest stronger links between outcomes-based measures and both physician reimbursement and care pathway evaluation. Integration with other value-based care initiatives, including alternative payment models, is also gradually emerging.

 This study identified growing use of high standards of evidence and adoption of other best practices in the development, implementation, and evaluation of care pathways. As the influence of care pathways on patient care continues to expand, additional efforts are needed to increase transparency, disclose conflicts of interest, engage with patients, effectively align care pathways with improvements in patient outcomes, and integrate efficiently with other value-based care initiatives.

As the use of care pathways has expanded in the United States, development, implementation, and evaluation processes have evolved.

Research findings support notable increases in process transparency, codification of standards, and prioritization of high-quality evidence.

Provider groups emerged as the driving force behind development and implementation, although physician resistance and administrative burden remain barriers to adoption.

Despite efforts, patient education and engagement remain inadequate.

There are indications of stronger links between outcomes-based measures, rather than traditional compliance measures, and both physician reimbursement and care pathway evaluation.

The movement toward value-based care may significantly affect the development and expansion of care pathways.

In the last decade, the use of care pathways—designed to provide evidence-based treatment options for disease management, improve quality of care, reduce unnecessary variation, and lower costs&mdash;has expanded rapidly in the United States.

 Although pathway use is increasingly prevalent for chronic diseases,

 use in oncology is well known, with 58% of oncology practices reporting use of pathways in 2016,

 Best practices to guide pathway development, implementation, and evaluation have been established in oncology, and the American Society of Clinical Oncology (ASCO) has issued a policy statement and a set of detailed criteria to ensure the integrity and quality of oncology pathways.

At the same time, other value-based healthcare initiatives and alternative payment models (APMs) are being adopted. In 2016, approximately 31% and 28% of physicians across specialties participated in bundled payment programs and medical homes, respectively.

 Overall, 43% of physicians received some form of value-based reimbursement. With the goals of improving quality of care while reducing cost, care pathways not only have objectives similar to those of other value-based initiatives but also may be critical in the transition to value-based care.

A January 2016 landscape assessment of care pathways revealed a wide range of development, implementation, and evaluation methods but a critical absence of transparency.

 In this follow-up study, we assessed changes in development, implementation, and evaluation of pathways and reviewed the latest evidence on integration of pathways with value-based care initiatives in the United States.

Similar to the 2016 study, a targeted review of published literature was followed by an online survey and in-depth interviews. Results were compared with those of the prior study, providing a perspective on the state of care pathways today and direction for the future.

Our search covered the PubMed and Cochrane Reviews databases and selected conferences’ websites. Publications, posters, and abstracts discussing care pathways in US healthcare settings that were in English and published between January 1, 2014, and March 3, 2017, were reviewed. We further searched trade journals and websites of professional organizations, pathway vendors, payers, and major provider groups. A supplemental search was completed in October 2017.

In screening materials identified in our search, we again required that a care pathway have stated objectives defining goals or key elements of care and was limited to specific disease states in defined patient populations, used evidence-based medicine or clinical guidelines to develop treatment recommendations, included a monitoring and evaluation mechanism, and included criteria for the use of pharmacologic therapies.

 Between June and September 2017, we conducted online surveys (n = 32) followed by in-depth interviews (n = 19). Respondents were selected to represent a mix of payers, providers from a range of specialties and practices, commercial care pathway vendors, and opinion leaders. Although all respondents were screened for involvement in pathway development, implementation, evaluation, or use, questions were administered only to those with experience and knowledge relevant to given sections (rated ≥3 on a 5-point scale). Respondents were offered honoraria for survey and interview participation.

 The survey was structurally similar to the 2016 study, with new sections on value frameworks and APMs. Descriptive analyses were conducted in Microsoft Excel.

 Interviewers administered an hour-long, semistructured discussion guide with open-ended questions. Interviews were designed to supplement information from the survey with additional detail on key areas of interest.

We identified 750 unique citations. Full-text review was conducted for 138 publications; 68 full-length articles and 44 abstracts and posters matched the selection criteria. Key themes and gaps in the literature informed the design of the survey and interviews. We found more transparency and codification of the development process, particularly in oncology, where pathway use has matured and more research has been published.

 More studies documented pathway adherence, resource use, and patient outcomes,

 but implementation and evaluation are still inconsistently documented or communicated. Engagement with stakeholders, particularly patients, remains inadequate.

 Efficacy and toxicity still supersede cost as key considerations during pathway development, yet pathways are viewed as a means to provide downward pressure on rising drug prices.

 Nevertheless, closer linkages with features of value-based care have emerged. As payers experiment with linking pathways and APMs,

 barriers to such integration need to be further understood. Interest is also growing in incorporating value frameworks in pathway development.

Twenty-five providers and 7 payers completed the survey. Eleven providers, 4 opinion leaders, 3 payers, and 1 pathway vendor were interviewed (

 Among all respondents with knowledge of pathway development, cost of care was the most frequently identified consideration for therapeutic area selection when care pathways are developed (77% [20/26]), followed by clinical outcomes, variation in treatment patterns, and prevalence rates (

). In 2016, the internal selection process was most cited, followed by cost of care. Pathways for relatively prevalent cancers, such as breast, colon, and lung, remain common; however, interviewees noted that rare cancers with high treatment costs are being identified for pathway development, citing recent FDA approvals of chimeric antigen receptor T-cell therapies and advances in precision medicine. Over the next 5 years, most survey respondents expect pathway use to expand in both oncology (78% [25/32]) and nononcology (84% [27/32]) areas. Pathways have also broadened to cover the care continuum, including symptom management and palliative care. Further, respondents had experience with pathways for common or high-cost chronic diseases, particularly in the areas of rheumatology, diabetes, asthma, and cardiology.

 Nearly all provider respondents (96% [24/25]) had a role in developing, implementing, evaluating, or using 2 or more pathways. Among respondents who addressed pathway development, 92% (24/26) identified clinicians as stakeholders, with fewer indicating a role for health plans (46% [12/26]) or pathway vendors (12% [3/26]). Most providers with development experience (74% [14/19]) indicated that their practice develops or implements pathways separate from those developed by payers or vendors.

Evidence considerations and transparency in care pathway development.

 Treatment guidelines (88% [23/26]) and published systematic literature reviews (73% [19/26]) were identified as key data sources for pathway development (

). Fewer identified retrospective or observational studies than in 2016. Providers were less likely than payers to identify prospective studies (53% [10/19] vs 100% [7/7]) or randomized controlled trials (58% [11/19] vs 86% [6/7]) as influential.

Interviewees identified experts and published or validated data from existing pathways as data sources for pathway development; survey results also confirmed that data from existing pathways informed development (77% [20/26]). Grading evidence quality appears to be less influential than in 2016 (58% [15/26] vs 81% [17/21]).

Efficacy (96% [25/26]), safety/toxicity (88% [23/26]), pharmaceutical costs (88% [23/26]), and medical resource use and costs (85% [22/26]) were the main factors considered when selecting treatments for the clinical algorithm. All (7/7) payers indicated considering medical resource use and costs versus 79% (15/19) of providers. In 2016, respondents identified safety (95% [20/21]), efficacy (86% [18/21]), and medical costs (81% [17/21]) as main factors. In both studies, efficacy was ranked as the most important. Although all interviewees indicated considering costs, particularly pharmaceutical costs, and clinical outcomes during development, less than half described cost as an initial consideration, and many indicated that clinical benefit precedes cost.

More than half (62% [16/26]) reported that their organizations followed best practices as detailed by ASCO’s policy statement; of these, 81% (13/16) noted that best practices, standards, or methodology were available to key stakeholders, including patients and payers. However, access may be restricted through means such as an intranet, patient portal, or membership. Further, awareness of processes for disclosing and managing potential conflicts of interest varied.

 Most (90% [28/31] with knowledge of pathway implementation) reported that pathways typically provide physicians with multiple treatment options for a given diagnosis, and 94% (29/31) reported that physicians have a choice about placing patients on a pathway. Compared with 68% (13/19) in the 2016 study, only 39% (12/31) reported a relationship between pathway compliance and reimbursement (

]). All of these respondents noted financial penalties or disincentives for noncompliance; 83% (10/12) indicated significant or extremely significant penalties. Interviewees expected incentives for performance, patient satisfaction, and other quality metrics captured in value-based care to receive traction in the near future. Process-oriented indicators, like compliance, were viewed as intermediate, surrogate measures. All payers (7/7) indicated that prior approval is required compared with 58% (14/24) of providers.

 Quality metrics (78% [18/23] with knowledge of pathway evaluation) were most commonly used to evaluate pathway performance, followed closely by readmission rates, compliance, and hospitalization rates (

). Quality metrics, readmission rates, and adverse event rates were each identified by 43% (10/23) as top 3 metrics. In the 2016 study, compliance was top-ranked and identified by most as a top 3 metric.

Interviewees observed that compliance remains an important measure. Target compliance rates of 75% to 80% were reported in 2016; however, only 43% (10/23) of respondents in the current research experienced 70% or greater compliance. Respondents reported challenges collecting or analyzing pathway data, including 75% (9/12) who encountered data collection issues.

 Some respondents (35% [11/31] with knowledge of pathway implementation) recognized changes in patient engagement over the last 3 years, among whom 91% (10/11) indicated that patients increasingly sought information on care pathways and 64% (7/11) observed increasing involvement of patient advocacy groups. Less than half of respondents (42% [13/31]) reported that on-pathway patients are provided disclosure or education about pathways; this included 46% (11/24) of providers, compared with 20% (1/5) of providers in 2016 (

). Interviewees also suggested that treatments may not be described as being part of a pathway. Financial incentives may not be transparent; 38% (5/13) of respondents who reported patient disclosure believed that patients were aware of financial incentives provided to physicians. Moreover, 26% (8/31) indicated that details on financial incentives are publicly available to all stakeholders, including patients.

Patient feedback on pathways is not generally mandated, with 30% (7/23), 43% (10/23), and 26% (6/23) reporting always, sometimes requested, and not formally engaged, respectively, among those with knowledge of pathway evaluation. Increased patient involvement is expected (78% [25/32]) in the next 5 years, particularly in evaluation (50% [16/32]) compared with development and implementation.

 The majority of survey respondents (93% [26/28] with knowledge of APMs [accountable care organizations, medical homes, and bundled payments]) and interviewees noted that care pathways are very or somewhat likely to be used in APMs; in APMs using pathways, 43% (12/28) estimated that greater than 50% of patients are on-pathway. Some interviewees noted that pathways could be key to value-based contracting and outcomes-based payments but also cautioned that costs could drive clinical care as a result. Even so, 57% (16/28) reported barriers to pathway use in APMs, including administrative burden (81% [13/16]) and lack of technological integration (75% [12/16]).

 Most survey respondents (74% [14/19] with knowledge of prominent US value frameworks) reported the use of value assessment frameworks in pathway development. Sixty-four percent (9/14), 57% (8/14), 50% (7/14), and 43% (6/14) noted value frameworks developed by the National Comprehensive Cancer Network (NCCN), American College of Cardiology and American Heart Association, ASCO, and Institute for Clinical and Economic Review (ICER), respectively. Interviewees often mentioned NCCN, ASCO, and ICER and generally advised considering value frameworks during development but not in isolation.

 Increased cost containment efforts (75% [24/32]), use of bundled payments (66% [21/32]), and applicability beyond oncology (47% [15/32]) may advance care pathway expansion and uptake. Other catalysts included the popularity of value frameworks (44% [14/32]), integration of clinical trials (34% [11/32]) and biomarker testing (25% [8/32]), and proliferation of medical homes (25% [8/32]). Interviewees additionally highlighted the value of standardization. In general, more providers reported the use of value frameworks in development; however, more payers felt that the popularity of value frameworks may expand the use of care pathways.

Physician resistance (pushback) remains the largest potential barrier to expansion (84% [27/32]), followed by administrative burden (

). Payers were more likely than providers to believe that administrative burden (71% [5/7] vs 56% [14/25]) and perceived implementation costs (71% [5/7] vs 32% [8/25]) are greater potential barriers, but more providers than payers found that the failure to demonstrate cost savings was a greater potential barrier (40% [10/25] vs 29% [2/7]). Reports of insufficient information technology (IT)/tracking systems and failure to demonstrate patient outcomes and cost savings have declined.

Care pathway development, implementation, and evaluation continue to play a significant role in utilization management in the United States, with cost considerations serving as a core motivation for these activities, collectively. During the development process, however, efficacy and safety continue to be prioritized overall. In large part, this trend continues to reflect efforts to manage care in the treatment of patients with cancer. Although pathways continue to be used in prevalent cancer types, primarily, we found a wider range of therapy areas for which pathways have been implemented, from rare cancers to chronic diseases.

We observed a higher bar for evidence during development, including greater emphasis on prospective studies and clinical guidelines, compared with the lack of codified standards and transparency in development we noted in our 2016 research. Although transparency has improved, access to specific development methodology remains limited to internal stakeholders, and disclosure of potential conflicts of interest is inconsistent. To some extent, challenges with transparency contribute to the difficulty in presenting a more definitive picture of conflicts or tension between pathways that guide medical care versus initiatives motivated by the imperative to manage expenditures, particularly in oncology.

Efforts to improve patient engagement with care pathways were among the most notable recent trends. About one-third of respondents observed changes over the past 3 years, noting increases in patient requests and patient advocacy group involvement, and more providers reported informing patients about pathway use. Most respondents anticipated future increases, with respondents also suggesting that patient engagement is becoming a feature of pathway evaluation, although this was not formally assessed in the prior study. Despite these efforts, a fundamental lack of patient awareness remains regarding the use of, and financial incentives associated with, pathways.

 We observed trends toward provider-driven pathways, the use of value frameworks during pathway development, and the growing relationship between pathways and APMs. Providers are increasingly central to the development and implementation of care pathways.

 In the 2016 study, providers were consistently identified as critical stakeholders, but the role of payers was often dominant. Pathway vendors still play an important role—particularly in serving payers that use pathways to manage plan member benefits&mdash;but providers are becoming the key drivers. Although we observed some trends distinguishing payer and provider behavior (data sources for pathway development, consideration of medical resource use and costs, requirement of prior approval, views on value frameworks, and barriers to expansion), further research in this area must be conducted.

Continued pressure to reduce costs has the potential to further expand the scope and development of care pathways. Value frameworks are often integral in the development process, ranging from being the primary approach for pathway construction to a reference for a component of treatment selection. The influence of value-based care was also noted in pathway implementation, with some respondents pointing to a shift in incentive structures. Now, fewer respondents indicated physician reimbursement being tied to pathway compliance; instead, reimbursement may become more directly linked to patient outcomes. Additionally, respondents believed that patient outcomes and other quality measures may eventually become standard pathway evaluation metrics. Pathways can be an effective tool in the delivery of value-based care, particularly in oncology.

 There is potential for more integration between pathways and APMs, which have similar objectives.

 Moreover, catalysts for pathway uptake include use of bundled payments and medical homes.

However, care pathways continue to face barriers to adoption. Although prior concerns with IT/tracking systems and the demonstration of patient outcomes and cost savings have decreased, concerns with physician pushback and administrative burden persist. Such concerns could be mitigated with provider-driven pathways and availability of evidence on patient outcomes, long-term efficiency, and reduced practice costs.

We utilized a hybrid approach with a targeted literature review followed by qualitative primary research, entailing survey and interview components, similar to the 2016 study. The survey and interview findings aligned with those of the literature review; however, the literature review findings are not exhaustive and are subject to the limitations of keyword-based searches of publication databases and ad hoc searches of the gray literature. Although survey and interview respondents had oncology and nononcology experience, the samples are not necessarily representative, and findings may not be generalizable to the wide diversity of care pathways. Several factors also limited our ability to cross-validate responses between the current and prior studies and perform statistical tests. The samples’ size and composition differed, and both the survey structure and responses to questions were updated to adapt to changes in the pathway landscape.

The authors examined the latest trends in development, implementation, and evaluation of care pathways and the impact of the movement toward value-based care.

US Care Pathways: Continued Focus on Oncology and Outstanding Challenges

 To better understand the prevalence of US value-based payment arrangements (VBAs), their characteristics, and the factors that facilitate their success or act as barriers to their implementation.

 Surveys were administered to a convenience sample of subject matter experts who were senior representatives from payer organizations and biopharmaceutical manufacturers. These data were supplemented with qualitative interviews in a subsample of survey respondents.

 Descriptive statistics, including percentages for categorical values and mean (SD) and median (interquartile range) for continuous variables, were assessed for quantitative questions. Trained reviewers collated responses to free-text survey questions and the qualitative interviews to identify themes.

Of the 25 respondents, 1 manufacturer and 4 payers reported not having explored or negotiated any VBAs. Subsequently, questionnaire results from 11 biopharmaceutical manufacturers and 9 payers who had experience with VBAs were analyzed. More than 70% of VBAs implemented between 2014 and 2017 were not publicly disclosed. Furthermore, although consideration of VBAs as a coverage and payment tool is increasing, VBA implementation is relatively low, with manufacturers and payers reporting that approximately 33% and 60% of early dialogues translate into signed VBA contracts, respectively. Respondents’ reasoning for VBA negotiation process breakdowns generally differed by sector and reflected each sector’s respective priorities.

 This study reveals that the majority of VBAs are not publicly disclosed, which could underestimate their true prevalence and impact. Given the effort required to implement a VBA, future arrangements would likely benefit from a framework or other evaluative tool to help assess VBA pursuit desirability and guide the negotiation and implementation process.

Little information is publicly available about the number and characteristics of value-based payment arrangements (VBAs) between biopharmaceutical manufacturers and payers, as well as the underlying reasons organizations do or do not enter these arrangements.

Previous estimates of VBAs, using only publicly available data, likely underestimate how many VBAs are currently being implemented.

Given the effort required to implement a VBA, future arrangements would likely benefit from a framework or other evaluative tool to help assess VBA pursuit desirability and guide the negotiation and implementation process.

The US health system faces increasing pressure to improve patient access to highly effective, yet costly, therapies. In line with the broader policy shift toward alternative payment models for healthcare delivery, there is ongoing interest amongpayers and manufacturers in exploring value-based payment arrangements (VBAs) for pharmaceuticals and medical devices. VBAs, also referred to as “outcomes-based contracts,” “performance-based contracts,” or “risk-sharing agreements,” link coverage, reimbursement, or payment for a product to a prespecified clinical or financial/utilization outcome or set of outcomes. VBAs can include contractual arrangements among manufacturers, payers, and providers to achieve these aims. For example, such arrangements could entail a rebate provided by the manufacturer to the payer when a drug or device does not perform as expected.

 VBAs have been discussed in the media and in the academic literature for drugs that treat hyperlipidemia, chronic heart failure, rheumatoid arthritis, and osteoporosis, among others.

Despite growing interest in and discussion of VBAs, there is relatively limited information currently available on the extent to which these arrangements are being pursued. A lack of public knowledge of these arrangements could (1) underestimate their use in the market and (2) limit what could potentially be learned from current experience that would advance the design and implementation of successful VBAs in the future. Recent research based on publicly available information suggests that a relatively small number of VBAs have been implemented in the United States (approximately 26), pointing to a range of contractual, procedural, and regulatory barriers that could impede their pursuit and implementation.

 Another study found that among 35 payers and 30 manufacturers, the average numbers of total pending/executed VBAs were 5.1 and 3.5, respectively, between 2012 and 2017.

 This study aimed to better understand VBAs in the United States, including both public and nonpublic arrangements, with a particular focus on gaining insights into their negotiation, operationalization, and impeding or success factors from the perspectives of manufacturers and payers.

This study used 2 data sources: an online survey and in-depth qualitative interviews. Target survey participants were subject matter experts from a cross-section of payer and biopharmaceutical companies and were identified through a snowball sampling technique based on existing working relationships, recommendations from others within the field, and internet searches. Senior representatives from 38 payer organizations and 37 biopharmaceutical manufactures were invited to participate between September 2017 and January 2018. Of those invited to participate, 26 representatives (13 each from biopharmaceutical and payer organizations) agreed to participate in the survey (response rate, 35%). Respondents were contacted for clarification when necessary to ensure validity of responses. Subsequently, 1 biopharmaceutical company’s responses were excluded due to a lack of response consistency, resulting in 25 analyzed respondents. Qualitative interviews, conducted with a subsample of survey participants, included 3 payers and 5 manufacturers. Study participants were not compensated for their participation.

First, we conducted an online survey using Qualtrics (

]). This survey, informed by a literature review and expert opinion, was designed to assess the key characteristics and proportion of public and nonpublic VBAs pursued between 2014 and 2017, as well as the factors that led to successful or, alternatively, challenged arrangement implementation. To guide exploration of VBA pursuit, we developed a process map describing 4 phases whereby manufacturers and payers (or providers) negotiate the terms of the arrangement, including (1) internal assessment and information gathering, (2) early dialogue, (3) formal negotiation, and (4) contract implementation (

). The process map and the survey were piloted and refined based on feedback from 5 experts representing both payer and biopharmaceutical organizations. Second, we conducted in-depth qualitative interviews with a subsample of survey respondents to expound upon key survey findings and address any gaps in information collected in the first phase of the study.

Descriptive statistics were assessed for each survey question by sector, including percentages for categorical values and mean (SD) and median (interquartile range) for continuous variables. A trained senior researcher (E.R.) collated responses to free-text survey questions and the qualitative interviews to identify key themes and substantiate main findings from the quantitative analysis, which were then vetted by the research team.

Of the 25 respondents, 1 manufacturer and 4 payers reported not having explored or negotiated any VBAs. Subsequently, questionnaire results from 11 biopharmaceutical manufacturers and 9 payers who had experience with VBAs were analyzed. Manufacturer respondents included 9 large firms (≥10,000 employees and annual revenues higher than $15 billion) and 2 medium-sized firms (5000-10,000 employees and revenues between $2 billion and $10 billion). Payer respondents included 1 single-state payer, 2 regional payers, and 4 national payers covering approximately 76 million lives. Of these payers, 6 offered both commercial and Medicare Advantage plans, and 1 was a public insurance contractor. There were also 2 pharmacy benefit managers covering an estimated 103 million lives.

Manufacturer and payer respondents reported implementing (phase 4) a total of 88 and 122 VBAs since 2014, respectively. However, there was considerable within-group variation in terms of the number of VBAs per firm, with manufacturers ranging from 1 to 15 implemented contracts (median of 9 contracts) and payers ranging from 1 to 40 implemented VBAs (median of 11 contracts) (

). Manufacturers and payers also reported that a majority of their company’s VBAs are not publicly disclosed (74% among manufacturers and 71% among payers). Additionally, 5 manufacturer and 2 payer respondents reported that none of their company’s VBAs are publicly disclosed (

). Respondents reported that laboratory value, medical encounter (eg, hospitalization rate/duration), financial, and drug utilization measures are the most common outcomes used in VBAs, whereas the most common payment mechanisms involved the manufacturer paying some portion of supportive product costs or providing a larger rebate or full refund to the payer if target outcomes were not achieved (

Both payer and manufacturer respondents also highlighted that there were no preferred therapeutic areas for VBAs, but rather that several factors make certain therapeutic areas (eg, asthma, cardiovascular disease, multiple sclerosis, hepatitis C, and diabetes) prime for VBAs, including validated and easily measured outcomes, uncertain value for high-cost products, limited competition, and availability of a diagnostic tool (eg, genetic testing) (results not displayed).

Direct provider engagement in VBAs is limited to date; only 3 manufacturers and 1 payer reported having engaged providers directly (eAppendix B). However, 91% of manufacturers and 78% of payers believe that incorporating providers is important for the future success of VBAs. Only 1 payer and 1 manufacturer reported that they have implemented a VBA in which a patient is reimbursed out-of-pocket costs for outcome failure. In contrast, patient adherence was a VBA component used by all payers and all but 1 manufacturer. Patient adherence was most commonly incorporated into VBAs by manufacturers and payers through patient selection (VBA only includes patients who have met a certain level of adherence) (55% and 63%, respectively), outcome terms (payment is explicitly tied to improvements in adherence [eg, proportion of days covered, medication possession ratio, and discontinuation]) (27% and 25%, respectively), and payment trigger (outcomes-based component does not take effect unless a certain adherence rate is achieved in the patient population) (18% and 38%, respectively).

Characterizing the VBA Negotiation and Implementation Process

The number of VBAs by sector in each stage of the negotiation and implementation process is detailed in the Table. The attrition rate across stages was high for both sectors. Manufacturers reported that only about one-third of the early dialogues conducted (ie, where 2 parties come together to informally discuss a potential arrangement) ultimately resulted in contract implementation. Payers reported that about 60% of the early dialogues with manufacturers resulted in VBA implementation.

Understanding VBA Negotiation Breakdown and Success

Respondents’ reasoning for VBA negotiation process breakdowns generally differed by sector and reflected each sector’s respective priorities (

). For example, manufacturers cited challenges related to data collection and evidence development (73%), the availability of appropriate outcome measures (64%), and implementation costs (64%) as major reasons for negotiation failure, whereas payers cited disagreement over incentive mechanisms tying payment to outcome (56%) and financial terms associated with the arrangement (67%). Both groups (64% of manufacturers and 78% of payers) cited Medicaid Best Price (generally, that a manufacturer must offer Medicaid programs the lowest or “best price” based on what is available to other purchasers) as a major reason for negotiation failure.

 In their top 5 negotiation success factors, both manufacturers and payers identified (1) the availability of measurable outcomes clearly tied to product use (82% and 100%, respectively), (2) a target patient population that is easily identified in claims (73% and 63%, respectively), and (3) a reasonable administrative burden (45% and 38%, respectively) (

). Manufacturers also prioritized having partners with the necessary data collection and analysis capabilities (91%), whereas payers emphasized the potential for high budget impact (75%) and having a reasonable data collection and analysis time frame (63%).

This study aimed to gain a better understanding of the prevalence and characteristics of public and nonpublic VBAs and to assess where in the negotiation and/or implementation process these arrangements fail and why. Importantly, we found that VBAs are far more prevalent in the United States than previously estimated, based on our sample of respondents. Manufacturer and payer respondents in our convenience sample reported having implemented (phase 4) a total of 88 and 122 VBAs, respectively, between 2014 and 2017. Significant variation in the number of contracts reported by each company was observed, suggesting that some companies may have more success than others in navigating the complexities of these arrangements. Both payer and manufacturer respondents reported that a majority of these contracts are not publicly known. This demonstrates that a substantial proportion of the activity related to tying payment to outcome measures is happening behind the scenes, which should be considered when evaluating the uptake and success of the broader movement to tie payment to value. However, when contemplating policy decisions associated with value-based payment, it is also important to contextualize the magnitude and potential of VBAs compared with more traditional payer—manufacturer contracts such as rebates (eg, success of VBAs in addressing health spending if they represent only a small proportion of overall contracts).

Follow-up interviews with payers and manufacturers suggested that publicizing VBAs can be beneficial, allowing a company to signal to customers and the public at large that value is an important issue for them, that they are at the forefront of the drug pricing debate, and that they have credibility and enough experience to advance value-based payment discussion and policies. However, the high proportion of nonpublic agreements indicates that many companies believe that the risks associated with publicly announcing a VBA, such as exposure to public scrutiny or yielding competitive advantage, outweigh the benefits. Additionally, contracts between manufacturers and payers are not typically publicly announced, and companies may be unwilling to challenge this norm.

Based on our analysis, successful VBAs—in which initial discussions between interested parties resulted in implementation of an arrangement&mdash;possess a range of similar characteristics, namely the availability and measurability of outcomes, uncertain value for high-cost products, and clear patient identification (eg, diagnostic tools), which have been corroborated in previous studies.

 The considerable discrepancy in the reported proportion of early dialogues that ultimately led to contract implementation between the 2 sectors was surprising, and it is unclear why payers report a higher rate of negotiation success. This could be due to certain limitations of our study, detailed later, or differences in interpreting negotiation success, but it could also reflect the relative importance of payer buy-in to achieve VBA implementation.

Although many contracts have indeed been implemented to date, this study underscores the significant effort required to successfully progress through all 4 VBA process phases. Manufacturers and payers reported that approximately 67% and 40% of early dialogues, respectively, do not reach implementation. Next, we discuss a number of major barriers that were highlighted that impede successful negotiation and implementation of VBAs, supporting previous findings.

First, both payers and manufacturers mentioned the need for clarification, and potential reform, of certain legislation and regulation, most notably the Medicaid Drug Rebate Program’s Best Price calculation requirement.

 Second, both payers and manufacturers indicated that data collection and analysis issues pose a significant challenge to VBA implementation due to a lack of data infrastructure, which could be addressed through common data standard principles to allow for not only interoperability across platforms for data of the same type (eg, electronic health record [EHR] data) for aggregation, but also linking across data types (eg, EHR, laboratory, and claims data).

 Notably, in the follow-up interviews, all respondents predicted that enhanced data sharing/analytic capabilities would result in not only more VBAs, but also more sophisticated arrangements. In the interim, and perhaps even as more advanced evidence-generation capabilities exist, respondents emphasized that contracts should be designed as simply as possible, including the data collected, outcomes measured, time horizon, and contract terms (eg, payment structure). Third, aligning with prior studies, follow-up interviews revealed that a lack of trust and understanding between payers and manufacturers can impede the negotiation process, particularly disagreement resolution.

 As a result, stakeholders often partner with organizations with which they have a history of positive, successful relationships to mitigate this risk. In new partnerships, it is essential to establish mutual trust by being transparent with regard to respective goals, notably in the early negotiation stage, which may be facilitated by using frequent communications.

 Additionally, to enhance trust, partners may consider either working with a third party to research, design, implement, and validate the VBA, or involving a greater range of stakeholders.

 For example, we found that most VBAs are between a payer and a manufacturer, with limited provider and patient partnerships. Increased provider engagement may enhance shared accountability and data measurement, but it could also introduce conflicts between new and existing provider-risk agreements, including alternative payment models, as well as add complexities to the contract negotiation process due to lack of medical benefit/pharmacy benefit integration.

 Although patient adherence was frequently used as a VBA component (eg, for patient selection), follow-up interviews suggest that there is uncertainty about whether VBAs improve patient outcomes or benefit patients more broadly, at least in terms of how they are currently designed and executed. For instance, existing VBAs appear to rarely integrate patient-reported outcomes and/or incorporate avenues for patients to share in any savings associated with the arrangement. Greater patient involvement could perhaps address this concern and align with broader efforts to mitigate high patient out-of-pocket costs. As an example, United Healthcare/OptumRx, CVS Caremark, and Express Scripts publicly announced recently that rebates would be directly shared with certain beneficiaries at the point of sale.

Overall, this study suggests that the VBA process is complex, requiring different points of alignment to successfully progress throughout all 4 of the suggested phases. As investment in these contracts by payers and manufacturers continues to increase, further research is needed to understand the scope, processes, and impact of VBAs. The study also highlights additional areas for future research, including the development and collection of VBA outcome measures, incentive mechanisms for participation, and the potential feasibility and impact of provider and patient incorporation into VBAs. However, focusing solely on publicly known VBAs greatly underestimates their prevalence and potential impact in the market, highlighting the need for research on all VBAs in a manner such that organizations can be transparent without disclosing confidential information. More transparent information exchange may also enhance trust among involved stakeholders and broaden participation in VBAs to providers and patients, where warranted. Additionally, more open discourse could help navigate the potential for overlapping contracts to enhance potential collaboration and efficiency of value-based payments.

Considering the various barriers along the path to implementation, it may also be beneficial to develop a tool or guidance document that can help manufacturers and payers assess the desirability of pursuing a VBA and optimize the difficult negotiation process. In particular, such a planning or evaluation tool could outline key components of a VBA (eg, technology characteristics, data collection and outcome measurement plan, level of risk) to consider at different phases of the process, with a view to highlight those features known to facilitate success. Furthermore, given the current diversity in VBAs, including how they are defined and conceptualized, a guidance document or the like could support greater consistency and quality across arrangements, reduce administrative burden, and increase adoption of VBAs throughout the healthcare system. As more information on VBAs becomes publicly available, such a tool or guidance can be refined and/or expanded to accommodate different types of VBAs.

Although there are several strengths to this study—including that it is the first study, to our knowledge, that focuses on understanding the extent to which these arrangements are being fully disclosed to the public and made transparent &mdash;there are limitations. Generalizability of these results may be limited because the questionnaire was administered to a small convenience sample with a low response rate, and the follow-up interviews were conducted on a selection of this sample. Next, although we provided a definition of VBAs and a framework for the negotiation and implementation process, there may be heterogeneity in respondents’ interpretations. Additionally, the questionnaire collected information only on newer VBAs administered from 2014 to 2017 to provide a feasible and realistic range for respondent recall. However, given that we collected information only from 2014 to 2017, we likely have not captured the full trajectory of previous or impending VBAs. Finally, the extent to which there was overlap between the contracts reported by manufacturers and the contracts reported by payers is unknown. To address this limitation, we surveyed a sample of both payers and manufacturers independently to cross-check validity and corroborate responses.

As VBAs are still broadly in the experimental stage, their ability to deliver on their promise (ie, increased value and/or improved patient outcomes) is yet to be determined. However, it is important to recognize the progress made in transitioning from volume- to value-based payment. This study reveals that the majority of VBAs are not publicly disclosed. Focusing solely on publicly known VBAs greatly underestimates the prevalence and potential impact of these arrangements in the market, highlighting the need for research on the proportion of VBAs among more traditional rebate contracts in a manner such that organizations can be transparent without disclosing confidential information. Although many negotiations never lead to a signed agreement, pursuit of these arrangements increasingly occurs between payers and manufacturers, highlighting that value considerations are being incorporated into stakeholder decision making. Our study identified several opportunities to address outstanding barriers to VBAs and further accelerate the transition toward value-based payment for biopharmaceuticals. Given the amount of work required, future negotiations would likely benefit from a framework or other evaluative tool that can help manufacturers and payers to assess the desirability and feasibility of pursuing a VBA for a product.

Duke-Margolis Center for Health Policy, Washington, DC (NM, ER, GWD, JQ), and Durham, NC (CS, HC, MM); UNC Eshelman School of Pharmacy (NM), Chapel Hill, NC; National Pharmaceutical Council (LB, KW, RWD), Washington, DC.

 This work was funded by the National Pharmaceutical Council.

 Dr Mahendraratnam was a University of North Carolina—Bristol Myers Squibb Worldwide Health Economics and Outcomes Predoctoral Fellow during the completion of this work. Dr Daniel receives consulting fees from AbbVie. Ms Buelt, Ms Westrich, and Dr Dubois are employees of the National Pharmaceutical Council, a policy research organization supported by the major research-based pharmaceutical companies in the United States. Dr McClellan is a board member of Johnson & Johnson, Cigna, and Alignment Health Care, LLC. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

 Concept and design (NM, CS, GWD, LB, KW, JQ, HC, MM, RWD); acquisition of data (ER, GWD, JQ, HC, MM); analysis and interpretation of data (NM, CS, ER, GWD, LB, KW, JQ, MM); drafting of the manuscript (NM, ER, GWD, LB, KW, JQ, MM, RWD); critical revision of the manuscript for important intellectual content (CS, GWD, LB, KW, JQ, MM, RWD); statistical analysis (NM, JQ); provision of patients or study materials (MM); obtaining funding (GWD, MM); administrative, technical, or logistic support (JQ, HC, MM); and supervision (GWD, KW, MM).

 Mark McClellan, MD, PhD, Duke-Margolis Center for Health Policy, 1201 Pennsylvania Ave, Ste 500, Washington, DC 20004. Email: mark.mcclellan@duke.edu.

1. Barlas S. Health plans and drug companies dip their toes into value-based pricing: the pressure is on P&T committees to monitor utilization. 

2. Silverman E. Amgen’s money-back guarantee for its pricey cholesterol drug may not deliver. STAT website. statnews.com/pharmalot/2018/04/02/amgen-money-back-guarantee-cholesterol-price. Published April 2, 2018. Accessed August 20, 2018.

3. Yu JS, Chin L, Oh J, Farias J. Performance-based risk-sharing arrangements for pharmaceutical products in the United States: a systematic review. 

. 2017;23(10):1028-1040. doi: 10.18553/jmcp.2017.23.10.1028.

4. Seeley E, Kesselheim AS. Outcomes-based pharmaceutical contracts: an answer to high U.S. drug spending? 

5. Duhig AM, Saha S, Smith S, Kaufman S, Hughes J. The current status of outcomes-based contracting for manufacturers and payers: an AMCP membership survey. 

. 2018;24(5):410-415. doi: 10.18553/jmcp.2017.16326.

6. Daniel GW, Barker T, Rai A, et al. Overcoming the legal and regulatory hurdles to value-based payment arrangements for medical products. Duke-Margolis Center for Health Policy website. healthpolicy.duke.edu/sites/default/files/atoms/files/overcoming_legal_and_regulatory_hurdles_to_value-based_payment_arrangements_for_medical_products.pdf. Published December 2017. Accessed August 27, 2018.

We surveyed biopharmaceutical manufacturers and payers to understand the prevalence and characteristics of value-based payment arrangements, as well as their implementation obstacles and success factors.

Value-Based Arrangements May Be More Prevalent Than Assumed

A primary objective of accountable care is to support providers in reforming care to improve outcomes and lower costs. Gaps in accountable care measure sets may cause missed opportunities for improvement and missed signals of problems in care. Measures to balance financial incentives may be particularly important for high-cost conditions or specialty treatments. This study explored gaps in measure sets for specific conditions and offers strategies for more comprehensive measurement that do not necessarily require more measures.

A descriptive analysis of measure gaps in accountable care programs and proposed solutions for filling the gaps. 

We analyzed gaps in 2 accountable care organization measure sets for 20 high-priority clinical conditions by comparing the measures in those sets with clinical guidelines and assessing the use of outcome measures. Where we identified gaps, we looked for existing measures to address the gaps. Gaps not addressed by existing measures were considered areas for measure development or measurement strategy refinement.

We found measure gaps across all 20 conditions, including those conditions that are commonly addressed in current measure sets. In addition, we found many gaps that could not be filled by existing measures. Results across all 20 conditions informed recommendations for measure set improvement.

Addressing all gaps in accountable care measure sets with more of the same types of measures and approaches to measurement would require an impractical number of measures and would miss the opportunity to use better measures and innovative approaches. Strategies for effectively filling measure gaps include using preferred measure types such as cross-cutting, outcome, and patient-reported measures. Program implementers should also apply new approaches to measurement, including layered and modular models.

	Measurement in accountable care programs is essential for promoting quality improvement and balancing financial incentives. This study examines gaps in current accountable care measure sets and proposes solutions to fill those gaps. 

		Gaps exist in accountable care measure sets for every condition, even for conditions that are partially addressed in a measure set. 

		Gaps cannot be completely addressed with measure types and strategies currently in use. 

		Effective means of covering gaps include increased use of cross-cutting, outcome, and patient-reported measures. 

		Layered and modular approaches to implementing measure sets would allow for more comprehensive and flexible measurement.

Accountable care systems, including accountable care organizations (ACOs), along with other value-based approaches to delivery and payment, have proliferated in the public and private sectors.

 Accountable care is focused on increasing the value of care; that is, improving quality while decreasing costs.

 Accountable care systems are implementing innovations in care delivery, including better monitoring systems, decision support tools, care coordination capabilities, and team-based approaches to care that are enabled under flexible new payment models. One of the tools available to promote higher value care is measurement.

Accountable care measure sets are tied to performance-based payment arrangements that reward providers for improving quality and avoiding waste. Waste includes underuse, which could lead to avoidable complications and costlier care overall, as well as overuse and misuse of resources. Measures are also important for balancing financial incentives. Program implementers can use measures to gauge the impact of accountable care reforms, which may be particularly important for high-cost conditions and treatments.

Outcome measures are preferred for assessing accountable care systems because they provide information about the results of care that is meaningful to patients, payers, purchasers, and policy makers.

 Cross-cutting measures that address multiple conditions and composite measures that aggregate multiple processes and/or outcomes offer the advantage of assessing many aspects of care simultaneously, thereby improving measurement efficiency. Process measures, which evaluate compliance with care guidelines, provide actionable information to support provider improvement.

Gaps in measure sets represent missed opportunities for monitoring system performance, providing transparency to consumers and purchasers, and encouraging improvement in quality and cost of care. Ideally, meaningful measures would be available for all conditions and dimensions of care, but that would subsequently increase the burden of data collection for providers and could distract them from quality improvement efforts. Understanding the best approach to more efficient and effective measurement—including new measure types and innovative approaches to measurement&mdash;would benefit from a comprehensive view of measurement gaps.

This study explored the breadth and depth of gaps in accountable care measure sets and identified methods for improving such measure sets by using preferred measure types and by adopting novel models for applying measures.

Our study included an analysis of measure gaps for specific conditions and a 1-day, multi-stakeholder roundtable discussion of national thought leaders to review the analysis and inform the conclusions. To explore gaps in accountable care measure sets, we selected 20 high-priority conditions. We then conducted a literature search for lists of high-impact conditions from authoritative sources, such as the National Quality Forum and the CDC. These lists included conditions that are either highly prevalent, leading causes of death, costly for patients, or financially and administratively burdensome to the healthcare system. Based on the search, we compiled a list of conditions which represents a diverse range of patient demographics (eg, age, gender, acute and chronic conditions, primary and specialty care) and anticipated cost drivers (eg, specialty drugs, surgery, imaging, hospitalization).

The list includes conditions that have been the historical focus of performance measurement (eg, asthma, chronic obstructive pulmonary disease, diabetes, hypertension, ischemic heart disease, influenza); that primarily affect the elderly (eg, osteoarthritis, osteoporosis, glaucoma, stroke); that primarily affect children (eg, attention-deficit/hyperactivity disorder); that affect each gender (eg, breast cancer, prostate cancer); that are related to behavioral and mental health (eg, major depression); and that require specialty pharmaceuticals or advanced imaging (eg, chronic kidney disease, hepatitis C, HIV, multiple sclerosis [MS], rheumatoid arthritis, low back pain).

We applied a 6-step analysis to each condition to identify key gaps between clinical practice guidelines and accountable care measure sets. For this comparison, we selected the Medicare Shared Savings Program (MSSP)

 and National Committee for Quality Assurance (NCQA) ACO Accreditation

 measure sets. In step 1, we identified diagnostic and treatment goals for each condition using evidence-based clinical practice guidelines from nationally-recognized sources, such as medical specialty societies and patient advocacy groups. In step 2, we compared the care goals for each condition with the measures in the ACO sets and identified measures that either directly or indirectly addressed the care goals. In step 3, we identified care goals that were not covered by the ACO sets as measure gaps. In step 4, we scanned various databases for measures that would cover the gaps identified in the ACO sets. In step 5, we identified opportunities for measure development to fill gaps that were not addressed by available measures. In step 6, we reviewed the gap assessment results across all 20 conditions to identify cross-cutting gaps and inform new measurement solutions.

For each condition, we identified a number of care goals defined by clinical guidelines that were not assessed by measures currently available in either the MSSP or NCQA ACO measure sets.

 Although treatment for some chronic diseases (eg, diabetes, hypertension) were assessed in both sets by several process and fewer outcome measures, other conditions (eg, MS, HIV) had no measures relevant to treatment or health outcomes. Eighteen of the 33 measures in the 2015 MSSP set are outcomes—including patient-reported Consumer Assessment of Healthcare Providers and Systems (CAHPS) survey measures and admission/readmission measures&mdash;and 13 of the MSSP measures directly apply to 10 of the conditions on our list. Thirteen of 40 measures in the NCQA ACO set are outcomes&mdash;including resource use measures&mdash;and 23 of the NCQA ACO measures directly apply to 12 of the conditions on our list.

We found cross-cutting measures in the ACO sets that indirectly covered important aspects of care for a range of conditions simultaneously. These include the CAHPS survey measures (eg, functional status, communication, access to care) and non—survey-based wellness and prevention measures (eg, body mass index [BMI], tobacco screening and follow-up, immunizations).

We identified measures available outside the ACO sets that could be used to fill gaps; however, the majority are process, not outcome, measures (

). We also found aspects of care that are not assessed by any available measures, which represent opportunities for measure development. Gaps requiring development included measures of health risk assessment, monitoring for disease progression, and referrals to nonphysician services such as physical or behavioral therapy. For MS, we found no measures.

Our findings illustrate the scope of quality measurement gaps in accountable care for many high-priority conditions. Whereas our analysis does not evaluate the impact of these gaps, the importance of measurement in accountable care arrangements implies missed opportunities and possible risks that could be addressed through better quality measurement systems. Extrapolating our findings to the universe of conditions—or even just high-priority conditions&mdash;illustrates the difficulty of attempting to address every important measure gap with the same types of measures and current approaches; it would require hundreds of measures and would be impractical, costly, and burdensome. The results demonstrate the need to use priority measure types and novel strategies to efficiently address these important gaps in measurement.

Leveraging use of current measures and developing better measures would help to fill gaps related to the most important opportunities for improvement. As Figure 1 depicts, relatively few outcome measures are available; however there are outcome measures that have been tested and endorsed but are not being used, including measures of functional status or rates of avoidable adverse events. The recent inclusion of the depression remission measure to the MSSP ACO set is a positive step by CMS toward the increased use of outcome measures. Additional patient-reported outcome measures that are available, such as health status and symptom control measures, should be included in measure sets to assess what matters most to patients. Where development of new measures is required to fill gaps, resources should be directed toward development of outcome measures, including patient-reported outcomes.

Cross-cutting measures enable efficient assessment of quality of care across multiple conditions as well as to evaluate important aspects of care for conditions that are not directly addressed. Cross-cutting measures include patient engagement (eg, shared decision making, education); population health (eg, BMI, smoking cessation); and care coordination/safety (eg, readmissions, medication management). However, large gaps in available cross-cutting measures remain, including patient self-management capability, activity level, and assessment of environmental factors that affect health.

Measurement Strategies to Fill Critical Gaps

Layered and modular approaches offer alternate strategies for targeting measures to address gaps, while minimizing the number of measures that must be reported for external accountability. The layered approach reflects that accountable care systems use measures at multiple levels: for external accountability reporting at the population level, for internal management of performance at the system level, and for internal improvement at the provider level. The key to the layered approach is to ensure that the measures the system chooses for internal management and improvement map to the external accountability measure set. In this way, the layered approach allows for coordinated, focused, and flexible, yet comprehensive, measurement.

The layered approach assumes that certain measure types are preferred at the various levels. Outcome measures—including patient-reported measures&mdash;and cross-cutting measures are best suited for external accountability because they allow for efficient reporting of the information that is most meaningful to program implementers and patients. Composite measures, along with indicator data such as utilization rates, are ideal for internal management dashboards to identify the largest improvement opportunities and monitor system-level improvement activities. Process measures, which are the most actionable, are valuable for supporting quality improvement initiatives on the front lines of care.

Cross-cutting measures, which assess care across multiple conditions, can also cross multiple layers of accountability, contributing to cohesive measurement through use of the same or similar measures in different layers. Where the same cross-cutting measure is applied in multiple layers, the results can be aggregated from the individual level to the population level.

Whereas relatively few outcome and cross-cutting measures would be the focus for external accountability, under the layered approach, provider organizations gain the flexibility to choose the types and numbers of measures they need to support management and improvement efforts. Providers could focus on the measures they deem essential for improving performance on outcome-oriented external accountability measures through their own quality improvement activities.

 illustrates the layered measurement approach with the following layers: 1) externally reported population-level outcome measures, such as functional status improvement or complication rates (eg, preventable admissions due to diabetes complications); 2) internally reported system-level measures to monitor outcomes, such as composite measures (eg, optimal diabetes composite of outcome [glycated hemoglobin control] and process measures [eye exam]; mortality composite for selected conditions [acute myocardial infarction, congestive heart failure, stroke, gastrointestinal hemorrhage, hip fracture, pneumonia]); and 3) internal process improvement measures, such as ordering appropriate tests and treatments that the providers deem important for improving outcomes.

 illustrates the modular measurement approach, which is another strategy that allows for more targeted and flexible use of measures to address gaps in accountable care measure sets. In the modular approach, a more granular measure set could be used to focus attention on improving quality and cost for a specific subpopulation within an accountable care program. Modules could be implemented when monitoring indicators show problems with care for a subpopulation. This model could also be used for Centers of Excellence, implementation of bundled payment arrangements, or for developing and testing new measures for widespread use.

As an example of the modular measurement approach, a specific measure set for oncology patients could be implemented under the broader accountable care measure set. Although the broader measure set, including the cross-cutting measures, would apply to the entire population, the additional measures of the oncology module would apply only to cancer patients. Measures in the module would include oncology-specific outcomes such as survival and complications, and processes such as appropriate treatment and use of palliative care, along with the cross-cutting measures from the broader accountable care measures set.

Using preferred measure types and the layered and modular measurement approaches would be particularly advantageous for assessing care for individuals with multiple chronic conditions (MCCs), whether in Medicare, Medicaid, or commercial accountable care programs. Measures of disease-specific processes—even those that address common comorbidities&mdash;often are not applicable to individuals with MCCs; moreover, it would be impractical to measure every aspect of care for each individual with MCCs.

Accountable care depends on having meaningful measures and efficient approaches to measurement. Although data systems are improving, the cost of developing and using measures effectively remains significant. To achieve success with new models of accountable care delivery and financing, the measurement strategies we have presented can help prioritize efforts to increase the benefit of measurement. To address widespread gaps in accountable care measure sets while limiting the burden of measurement, program implementers should use outcome, patient-reported, and cross-cutting measures and should promote development of those types of measures to fill gaps. In addition, program implementers should apply layered and modular approaches to broaden the reach of measurement in a targeted manner.

The authors acknowledge the valuable contributions of David Blaisdell, David Sloan, Avis Hixon, and Guy D’Andrea of Discern Health, and Adam Lustig, Andrea Hofelich, and Kathryn Gleason of the National Pharmaceutical Council in the preparation of this manuscript. The authors also acknowledge the insightful contributions of the participants in the multi-stakeholder roundtable, co-chaired by Drs Mark McClellan and Jerry Penso.

 Discern Health (TV, DD), Baltimore, MD; National Pharmaceutical Council (RWD, KW), Washington, DC; American Medical Group Association (JP), Alexandria, VA; The Brookings Institution (MM), Washington, DC.

 Funding for this project was provided by the National Pharmaceutical Council.

 Ms Westrich and Dr Dubois are employees of the National Pharmaceutical Council, an industry-funded health policy research group that is not involved in lobbying or advocacy. Dr McClellan is a member of the board for Johnson & Johnson. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

 Concept and design (TV, KW, MM, JP, RWD, DD); acquisition of data (TV, DD); analysis and interpretation of data (TV, KW, MM, RWD, DD); drafting of the manuscript (TV, KW, MM, JP, RWD, DD); critical revision of the manuscript for important intellectual content (TV, KW, MM, JP, RWD, DD); obtaining funding (TV, RWD); administrative, technical, or logistic support (DD); and supervision (TV, MM, RWD, DD).

 Tom Valuck, MD, JD, MHSA, 1120 North Charles St, Baltimore, MD, 21201. E-mail: tvaluck@discernhealth.com.

1. Barnes AJ, Unruh L, Chukmaitov A, van Ginneken E. Accountable care organizations in the USA: types, developments and challenges. 

2. McClellan M, Kent J, Beales SJ, et al. Accountable care around the world: a framework to guide reform strategies.

3. Weissman JS, Baillit M, D’Andrea G, Rosenthal MB. The design and application of shared savings programs: lessons from early adopters. 

Gaps in accountable care measure sets can be addressed efficiently using priority measure types and innovative approaches to measurement.


Robert W. Dubois, MD, PhD

Articles

From the Editorial Board: Robert W. Dubois, MD, PhD

US Care Pathways: Continued Focus on Oncology and Outstanding Challenges

Value-Based Arrangements May Be More Prevalent Than Assumed

Solutions for Filling Gaps in Accountable Care Measure Sets