Findings Related to Assessing Value for Private US Health Plans
A qualitative survey and quantitative database analysis of US payers indicated that determining whether they receive value for their expenditures involves substantial challenges.
Health plan formularies with tiered copayment structures or preferred drug lists (PDLs) have become increasingly popular, in part to promote competition and to create incentives for technology producers. In theory, plan decisions about medical products and formulary placement decisions about drugs should be guided by information on the overall value or cost-effectiveness of the products. Instead, observers complain that plans often base patient contributions on a drug’s acquisition cost and do not consider overall value in formulary placement and other coverage decisions. However, little empirical evidence has addressed this question. We set out to empirically explore the issues of value perception and evidence expectation with respect to medical products for US private payers. For several payer organizations, our research included a qualitative assessment of how payers describe value and what kind of evidence they want regarding all medical products, as well as a quantitative pharmaceutical analysis evaluating payers’ actual decisions. The quantitative analysis builds on previous work by Neumann et al1 by assessing formulary-positioning decisions in relation to published cost-effectiveness estimates. We present our primary qualitative and quantitative findings for these private US payers.
Qualitative Assessment of Value
The study included a qualitative assessment of health plan participants. A payer-focused data collection tool was used for structured telephone or in-person interviews with 15 study participant organizations. A discussion guide facilitated the qualitative interviews. It was designed to capture essential elements of payers’ considerations pertaining to medical product value perceptions, evidence expectations and evaluation processes, and coverage and reimbursement decision-making prioritization. The guide included the following “domains,” which directed the structured interviews:Evidence Expectations and Requirements, Payer Considerations of Constituents, Unmet Medical Need in Relation to Value Perception, and Ranking of Consideration for Coverage and Reimbursement. The aim of this study was to obtain payers’ views on value and evidence. Therefore, they were not provided with a research study definition of value. On the contrary, payer respondents were asked to provide their own definitions of value and product evidence considerations. Most of the sample included individuals from leading health plans or payer organizations. A few were with technology assessment organizations and/or were selected based on their experience with US payers. The sample consisted of a range of executives including chief executive officers, pharmacy and medical directors, actuaries, other high-level executives, and persons responsible for making recommendations related to health technology assessment, as well as health policy issues. Our sample included individuals associated with payer organizations responsible for providing health benefi ts for up to 80 million to 100 million US covered lives.2 In most cases, the payer interviews included only 1 representative employee. For 1 health plan, however, a multidisciplinary team provided organizational responses to the survey interview. It is worth noting that those working in decision making for pharmaceuticals made up the majority of the payer sample. Several individuals, however, were responsible for considering a broader set of medical products, including recommending coverage and payment policies as well as evaluating evidence and disseminating health technology assessment information for these products.
Quantitative Assessment of Value
Cost-Utility Analysis Data. This portion of the study focused entirely on pharmaceuticals. To measure a drug’s value, the assessment focused on cost-utility analysis (CUA), a specific type of cost-effectiveness analysis in which health effects are measured in terms of qualityadjusted life-years (QALYs) gained. Widely used pharmacoeconomic textbooks and guidelines and the 1995 Panel on Cost Effectiveness in Health and Medicine have recommended this outcome as a standard for cost-effectiveness research.3,4
Data on CUAs for drugs were derived from the Tufts Medical Center cost-effectiveness registry, a database of more than 1000 original CUAs published in the public health and medical literature (www.cearegistry.org). Costutility analyses are included in the registry if they were published in a MEDLINE-referenced journal and provided an original cost-utility estimate. Data on the cost-utility ratios associated with each intervention were standardized to 2002 US dollars.
To exclude potentially outmoded analyses, we restricted the sample to studies published from 1998 through 2003. Overall, the Tufts Medical Center registry contains 1408 cost-utility ratios from 1998 through 2003, from 555 separate cost analyses. For this analysis, we focused on CUAs in the registry pertaining to pharmaceutical interventions (221 CUAs reporting 571 cost-utility ratios; the number of ratios exceeds the number of studies because some analyses contained more than 1 ratio). Note that these ratios cover 144 different drugs.
Identifying Formulary Policies. We examined publicly available formularies of 9 healthcare providers: Aetna, Caremark, Express Scripts, Group Health Cooperative of Puget Sound, Humana, Kaiser Permanente, Medco, AARP, UnitedHealth Pharmaceutical Solutions, and WellPoint. These plans and pharmacy benefi t management (PBM) companies were chosen to represent formularies covering different geographic regions and types of payers. Several of the formularies represented have PDLs, whereas others have incentive-based tiered structures. The formulary data was obtained from the organizations’ publicly available Web sites in 2006.
Data Analyses. We examined the relationship between formulary placement (ie, whether on a PDL or a favorable tier) and evidence of value as reflected in published cost-utility ratios. Specifically, we investigated whether cost-utility ratios of drugs on PDLs were lower (ie, more favorable) than cost-utility ratios of drugs with nonpreferred status.
RESULTSHigh-Level Qualitative Findings
In general, there is a gap between what US healthcare purchasers and payers see as value and what the health technology and medical product industry is supplying. Medical product manufacturers (eg, pharmaceutical, diagnostic, and device) tend to focus on new technology and molecules, whereas payers and purchasers focus on solutions to medical problems and appropriate use of existing technologies. Payers and health plans reported a prominent interest in high-quality clinical safety, effi cacy, and effectiveness evidence (ie, clinical benefi t-risk considerations). Costs and cost-effectiveness were consistently identifi ed as important but secondary considerations for coverage and reimbursement decisions. However, the payers expressed an increasing interest in and willingness to evaluate information on the value of medical products and pharmaceuticals.
Payers’ demands are increasing for more studies that evaluate comparative clinical utility (ie, are new products useful for providers given the available products and do they change behavior of physicians?) and clinical performance (ie, does a new product improve patient signs, symptoms, and health outcomes?). Although they expressed an interest in having this information, they apparently are reluctant to contribute substantial payer funding and development responsibilities to such phase 4 studies. Determining the respective roles and responsibilities of payers, manufacturers, governments, and consumers in this endeavor remains a challenge.
Our research confirmed that US private payers are not uniform in their approaches to evidence evaluation. For example, they reported differences in how they weight costs and cost-effectiveness when making their decisions. Payers indicated that they have diffi culty in evaluating and utilizing cost-effectiveness evidence. Specific reasons included the following: study end points and/or study designs for competitive products are not always consistent, evidence of benefit compared with cost or cost-effectiveness claims is based on data of varying quality, and medical product manufacturers do not provide economic analysis information to health plans in a standardized and consistent format. Several payer respondents indicated that budget-impact models often are used to evaluate a product’s expected impact on a per member per month basis.
Additional responses from the interviews suggested that evaluating the direct budget impact of individual medical products, rather than assessing cost-per-outcome measures and impacts on the usage of other medical resources, is potentially related to several factors, including:
Health plans have limited resources devoted to fulltime employees or consultants who evaluate value and cost-effectiveness information. Generally, few specialists are available within health plans for assessing economic evidence, and these types of metrics have an unclear role in the development of coverage and reimbursement policies.
Health plans are subject to multiple influences: members, physicians, public/media attention, pharmacy and therapeutics committees, shareholders/investors, and others.
The US healthcare system, particularly for the well insured, includes payment structure incentives benefiting physicians who perform multiple tests and services. In general, the system does not encourage cost-conscious decision making or require that reimbursed medical decisions and product use be based on the existence of highquality evidence.
More Specific Qualitative FindingsValue Perceptions
Defining Value. The payer research confirmed that there is no agreed-on standard for defining value as it pertains to medical products. Although several payers were fairly consistent in their definitions, value is not generally a concept that many payers could easily or readily define. Several plans reported that their definition of value included assessment of the overall impact of medical products on resources (ie, total impact on pharmacy and medical expenditures), including consideration of health outcome or clinical outcome improvement versus the standard of care. Other interviewees stated that their organization assess value by evaluating clinical utility and clinical effectiveness compared with costs. Largely, payers indicated that their objective is to focus on the incremental impact of new technologies (ie, incremental net health gain vs incremental cost) compared with existing technologies.
When payers were asked about their own definitions of medical product value, their responses indicated that plans associate value with the ability of products to be “life changing” and to provide an improvement in overall net clinical benefit (ie, positive health benefit relative to safety risk). The survey respondents consistently indicated that they consider product costs along with the costs for available substitutes. Payers also said that they attach value to whether new medical products are “practical” for the average physician and patient. Examples given were whether products are easily incorporated into routine clinical practice, whether they are convenient or fairly simple to use for the average physician and patient, and whether the new innovation requires medical providers to substantially alter clinical practice. For verbatim payer responses, see the eAppendix available at www.ajpblive.com.
Defining Unmet Needs. Payers expressed frustration regarding identifiable differences in the use of terminology for “unmet medical need.” The plans suggested that high unmet clinical/medical need be defined to mean no safe options that improve health for a particular disease/condition. Several respondents indicated that the medical products industry stretches the term “unmet need” to include marginally beneficial products for non—life-threatening conditions. Pharmaceutical and surgery tool product manufacturers were specifically mentioned in this regard. It does not help payers to manage their considerable array of covered products for multiple diseases when medical technology producers focus attention on products that improve lifestyle- or health-related quality of life in non–life-threatening conditions. Payers also commented on the aggressive marketing of products that represent only slight improvement over existing products, or products that have limited differentiating clinical evidence. Several payers expressed the view that new products that work marginally better than existing standards of care, yet have substantially higher costs, do not meet the defi nition of value. Payers stated that they would prefer to see more innovative medical products being developed rather than so many targeting the same diseases with similar effi cacy/safety profiles. The marketing approaches of manufacturers were identified as contributing to perceptions of overselling the concept of unmet need. This overselling was felt to be detrimental to payers’ efforts to manage existing products and to identify truly value-adding medical innovations.
Payer Capacity Matters. Payer organizations reported varying levels of and approaches to medical product evidence evaluation. Differences in organizational size and analytic capabilities were evident in our sample. Larger payers and those with more sophisticated internal technology assessment groups reported performing their own modeling, as well as utilizing assessments from external groups and published reports. Scale affords the opportunity to devote resources to these activities. Smaller payers and those with more limited internal resources tended to utilize consultants and benefi t design advisers to recommend coverage and reimbursement strategies. Select examples of products identifi ed as receiving more restrictive coverage policies are presented in
Uneven Review Standards. There were substantive differences in the levels of health plan review and evidentiary requirements for diagnostics and devices compared with pharmaceuticals and biologics. In general, device manufacturers have faced more limited demands for evidence of value, but they can expect new devices to be scrutinized more closely in the future. Our findings suggest that manufacturers of prognostic and diagnostic tests are facing new reimbursement challenges, particularly when evidence is lacking related to the impact on clinician and patient behavior (ie, acting based on test results). Reimbursement restrictions are likely to increase for expensive new tests (ie, tests that cost thousands of dollars) aimed at diagnosing or predicting health risk. In addition, payers reported concerns about established technologies (eg, computed tomography, magnetic resonance imaging) that are being used for multiple indications with varying levels of evidence. Plans have more experience, better monitoring systems, and more elaborate infrastructure in place for evaluating drugs. There are distinct and steeper learning curves for evaluating evidence regarding devices and diagnostics. However, recently published recommendations for manufacturers of devices and diagnostic tests are intended to standardize the presentation of clinical and economic evidence on behalf of medical product manufacturers responding to payer organizations that request information about their products.5
Limitations of Available Data. In general, payers realize that data limitations exist for new products and they make the best decisions they can based on the available evidence. Partly due to US Food and Drug Administration (FDA) requirements related to needing 2 placebo-controlled studies for approval, payers must rely primarily on these phase 3 data from randomized controlled trials, particularly for drugs. They reported wanting additional real-world effectiveness data. Overwhelmingly, the lack of comparative evidence for new products was a primary concern for payers. Health plans want more information from head-to-head comparison studies so that they can improve coverage and reimbursement decisions. Therefore, companies developing, manufacturing, and marketing medical products can expect more comprehensive phase 4 commitments in the future.
Use of Clinical Guidelines and Guideline Quality Issues. Health plans were generally consistent in their use of clinical guidelines for medical decision making. Most indicated that guidelines are highly influential for setting policy, while recognizing potential biases introduced by manufacturers or specialists in guideline development. An identified challenge was that treatment alternatives usually are not fully represented in guidelines. Additionally, several payers observed that limitations are associated with inconsistent guideline quality and inconclusiveevidence. Psychiatric guidelines were identified as being based on subjective physician assessment with less evidence associated with long-term outcomes. Cholesterol guidelines based on multiple, large, long-term studies were identified as being evidence based and of high quality.6,7 Overall, payers allow for grading of the evidence base for guidelines and make a measured judgment related to the level of clinical balance and transparency present in the various guidelines.
Certification of Medical Product Value. There were mixed responses about the desire for a national public organization to certify medical product value based on technology reviews. Undoubtedly, questions remain as to funding a national technology review organization. These questions include who would be responsible for funding such an agency, what level of funding would come from the private and public sectors, what types of studies would be reviewed, who are the stakeholders of interest, and what the impact of these reviews would be on coverage and reimbursement policies. Private payers in the United States make coverage and reimbursement decisions independently, yet they evaluate national trends, observe Centers for Medicare & Medicaid Services policies, and review published professional medical technology assessments. However, a silo mentality persists with most private payers, who evaluate pharmacy and medical expenditures separately.
One reason for this divide may be that plans contract with separate PBMs and medical benefi t managers. Organizational differences in approaching pharmacy and medical benefi ts may contribute to inconsistencies in how pharmaceuticals, devices, and diagnostics are evaluated within plans. For example, when considering coverage/reimbursement for a new pharmaceutical product, pharmacy-related decision makers within plans may not fully consider medical event cost offsets such as reduced hospitalizations or physician visits.
Quantitative Results: Assessing Formulary Positioning and Cost-Effectiveness Evidence
The distributions of cost-utility ratios in the registry covering pharmaceuticals and all ratios included in the entire database are presented in
. The columns represent different thresholds for cost-utility evidence. Note that 11% of all ratios pertaining to pharmaceuticals reflected interventions that were cost saving and resulted in improvements in QALYs; 55% of all ratios pertaining to pharmaceuticals fell between $0 per QALY and $50,000 per QALY. In general, Table 2 shows that the distribution of cost-utility ratios for drugs among these 5 column categories was not substantially different from the distribution of ratios in the entire database (no statistical tests for differences were performed).
shows the distribution of cost-utility ratios by tier for each of the 9 organizations. The table shows (in boldface) that in numerous cases, drugs with favorable data from a value or cost-utility standpoint (ie, drugs associated with a cost-utility ratio under $100,000 per QALY gained) were placed on unfavorable tiers or a nonpreferred list. For example, in the case of Aetna, costutility ratios for 4 drugs on tier 3 (representing 6.5% of all cost-utility ratios corresponding to tier 3 drugs) suggested the drugs were cost saving. The rows in Table 3 associated with nonpreferred tiers show discrepancies forother providers. For example, for Caremark, there were 55 instances of cost-utility ratios indicating cost-saving and QALY-increasing drugs; however, these drugs were on the PBM’s nonpreferred drug list. For Express Scripts, there were 45 such cases.
There also were numerous drugs with unfavorable cost-effectiveness that received favorable formulary placement. For example, for Aetna there were 3 cases in which cost-utility ratios were over $100,000 per QALY gained, yet these drugs had tier 1 status. For Caremark, 18 cost-utility ratios over $100,000 per QALY were for drugs on the PDL, and there were 4 dominated cases (drugs refl ecting lower effectiveness and higher costs) in which the drugs received preferred status.
The incentive structure of the US healthcare and insurance system largely encourages utilization of services with limited regard for cost, particularly for the well insured. Ineffi ciencies associated with a nontransparent, thirdparty payer system and physician-as-agent relationships contribute to such behavior. Our payer panel suggested that limited comparative effectiveness evidence and a lack of mechanisms to control physician behavior contribute to off-label and/or inappropriate use. Our sample did not include a physician or patient panel. Thus, we are limited in discussing these perspectives. Physicians may well have different perspectives on value compared with health plan managers, and patients can have a considerable influence on prescribing and long-term adherence to guidelines. Nonetheless, there currently are limited effective mechanisms in the US public-and private-payer systems for completely denying coverage on the basis of sparse or inadequate evidence, particularly for oncology products and highly “political” cases. Payer respondents expressed concern that they have limited ability to fully refuse coverage, particularly in competitive markets. Considerations related to market share and member satisfaction were mentioned by several payers as being infl uential in establishing coverage and reimbursement policies. Copayment/ coinsurance and prior authorization mechanisms, however, are being used by plans to restrict use or to induce cost sharing for medical products and services. For example, more recent health plan policies guided by value-based insurance design principles are being evaluated for their impact on costs and clinically sensitive markers such as compliance and adherence.8
Although our payer sample included persons responsible for evaluating evidence and making decisions about the spectrum of medical products, several interview participants primarily worked in the pharmacy department of their plans. However, we believe that many of the general findings are relevant to both pharmacy and medical policies of health plans. For example, all payer respondents, regardless of position, reported that they consider multidimensional objectives. Members’ and physicians’ demands for innovative technologies conflict with the concept of rigorous cost containment (ie, “highly managed care”). Payers are increasingly considering conditional reimbursement and risk-sharing arrangements, particularly for expensive biologic products. Risk-sharing agreements usually are based on evidence and value claims. We expect conditional reimbursement to increase if federal agency efforts expand for conditional licensing and comparative effectiveness studies. The FDA’s requirements for placebo-controlled studies could be expanded to include standard-of-care comparisons, as in many oncology studies, helping to provide additional comparative evidence.
Several US health system characteristics may help toconnect our qualitative and quantitative findings. These characteristics include multidimensional considerations for private payers: business model, reputation for service, competition for members and providers, complexrisk-sharing or contracting agreements, as well as having sparse head-to-head comparative data for many marketed medical products. The payers’ stated information from thequalitative interviews and the information from the formulary- based quantitative assessment demonstrate consistency. Given that cost-effectiveness was not an identified principal criterion for payers, and that “contracted prices” are not publicly available for cost-effectiveness analyses, it follows that formulary positioning may conflict with literature-based economic value estimates. Payers are trying to keep multiple stakeholders satisfied, and they engage in multifaceted decision making. Additionally, the potential lack of understanding about cost-effectiveness analyses contributes to the limited use of these types of studies for setting medical product coverage and reimbursement policies. Accordingly, payers and employers alike are attempting to cost-shift to members/employees with consumer-directed programs9 and to identify “efficient” providers with the use of pay-for-performance initiatives and grading systems for doctors and hospitals. The health and monetary impacts of these approaches remain to be seen.
In general, our quantitative pharmaceutical analyses suggest room for improvement in health plans’ attempts to move to value-based formulary designs. However, our analyses are descriptive and exploratory. Our goal was to assess general trends in publicly available formulary positioning information versus published CUA evidence. We recommend caution in inferring causality between the cost-utility literature and formulary placement decisions, for this exercise presents inherent challenges. For most drugs listed on formularies, there are little or no corresponding data from recently published CUAs.
Despite these caveats, and other methodologic challenges associated with CUA,3 we believe that our study offers some insights. When CUA data were available, they suggested that formulary decisions often refl ect published evidence of value in that drugs with favorable formulary placement are products with “reasonable” cost-utility ratios. On the other hand, the results also suggest room for further investigation: there were numerous examples of reportedly cost-effective drugs excluded from preferred lists or favorable formulary tiers, as well as examples of drugs with unfavorable cost-effectiveness profi les that were included on PDLs and preferential formulary tiers. As Table 3 shows, there were more cases of products being positioned in unfavorable tiers with published cost-saving CUA estimates or cost-effectiveness estimates of less than $100,000 per QALY gained compared with the number of “low-value” products being positioned in favorable tiers.
We note a few other limitations. It is diffi cult to obtain a clear reading from publicly available CUAs about the actual choices facing formulary managers. A key problem in our study is that published CUAs may compare a new drug to placebo with an inferior alternative rather than to the next-best intervention, thus exaggerating the published value of the drug in question. Evaluation of specific pharmaceuticals was beyond the scope of the current study. For interested readers, Neumann et al1 presented selected drugs with potentially discordant formulary policies compared with CUA evidence. Inadequate CUA evaluations for subpopulations, variation in the quality of the economic models, and lack of data on actual drug prices paid by plans are further limitations that complicate comparison of literature-based estimates and payer decisions. In addition, CUAs generally attempt to estimate a societal perspective rather than a private-payer perspective, thus confounding comparability. On the other hand, although US health plans may not adopt a societal perspective, they do compete for enrollees, and it is not unreasonable to believe that CUAs approximate value in many cases.
Cost-effectiveness analysis contributes to the overall measurement of value for medical products and could be an important part of a sequenced assessment of value that begins with the application of evidence-based medicine. Payers in our study tended to verbalize product characteristics that represent “cost-effective” scenarios, yet they did not always use these words. Their consideration of clinically important benefi ts and harms and overall product value was influenced by evidence from subpopulations for which products have specifi c net advantages, as well as how innovations infl uence physician behavior.
Reviews of value evidence by health plans and their formulary and pharmacy and therapeutics committees are expected to evolve and expand. Our study suggests that these efforts should go further to better understand value and comparative effectiveness information for pharmaceuticals, diagnostics, and devices.10