Using Analytics to Design Provider Networks for Value-Based Contracts

March 1, 2018 11:16 am

To build a successful provider network in a value-based world, healthcare organizations should collect and analyze several key pieces of data.

Finance leaders within provider organizations are faced with increasing pressure to evaluate and inform organizational decisions about how and when to shift into value-based contracts. These decisions highlight the need for tight alignment between a health system’s contracting strategy with health plans and government payers and its strategy for building and maintaining a high-quality provider network that will be responsible for driving performance under these arrangements. Although most health system finance executives have experience leading and managing such a contracting strategy, they also are assuming a more active role in working with their leadership colleagues to inform network design decisions.

At its core, any network development effort is aimed at closely aligning the number, type, and location of providers with the demands of the population to be served and with the organization’s overall strategic goals. Conceptually simple, network design decisions are subject to a multitude of strategic, clinical, economic, and operational factors. Meanwhile, emerging dynamics associated with value-based contracts and care models introduce new considerations. Network development should be an iterative process that reflects these myriad issues. Although some elements are hard to quantify, leading provider organizations increasingly are relying on new healthcare data sources and sophisticated analyses to apply an informed approach to their network planning.

Organizations should evaluate four key factors to inform their efforts to configure their physician and advanced practice provider (APP) networks and align supply and demand in the context of value-based care:

  • Contract parameters—the value-based contract provisions and role within the insurer portfolio strategy
  • Population needs—the characteristics of the population to be served
  • Care model—the potential impact of clinical care model developments on demand and/or supply
  • Physician/APP landscape—the market provider landscape and the organization’s current network

Understanding Value-Based Contract Parameters

For value-based care, it is essential that contracting strategy and provider network design be closely linked and iterative. The requirements for success in specific value-based contracts will shape the organization’s network development priorities, while the composition of the network will materially influence which insurer contracting and product strategies the organization pursues. Therefore, to anchor the network discussion and assessment specific to value-based care, organizations should clarify the parameters of the value-based contracts they intend to pursue and of any they have already signed.

Several contract dimensions are critical to inform network conversations and related analysis because they affect the scale, scope, and prioritization of any network design effort. Healthcare organizations should ask the following fundamental questions as they explore network design plans.

With which payer segment(s) and plan(s) is the organization pursuing value-based arrangements, and what is the size, geographic distribution, and demographic profile of the target or contracted population(s)? Before signing the contract, the organization should analyze enrollment data (typically at the county level) combined with demographic data to segment and size the target geography by insurance type and population dynamics. Once the contract is signed, analysis of encounter and claims data can help the organization to understand the population more closely, including the degree to which the organization already serves the population.

What type of value-based payment model is the organization pursuing? Possible models include shared savings with upside-only or two-sided risk, accountable care organization (ACO) models, bundled payment arrangements, and partial or full capitation arrangements. Many ACO models that are anchored in a primary care attribution model have immediate implications for growing primary care capacity to serve a requisite scale of attributed lives and/or to address geographic gaps or access for specific populations. Bundled payment arrangements may focus attention on a select set of specialists and related providers, such as post-acute care providers.

What is the basis of attribution for the model? The organization should understand how lives will be assigned to to the model, whether retrospectively or prospectively. That information will shape how the organization should evaluate the scale of the attributable population and how actively and immediately it can affect that scale of population through targeted network expansion.

Is the arrangement tied to a narrow or tiered network or an open-access network? The network coverage standards will be highest for a narrow-network product strategy and lowest in a broad-network, value-based product.

Does the model necessitate provider exclusivity? Many ACO models with primary-care-based attribution require primary care exclusivity but allow for specialists to be in multiple arrangements. In some cases, providers may already be committed through a competing ACO or clinically integrated network (CIN), which may pose a challenge of deciding which entity gets credit or bears risk for that provider’s performance.

What is included in the arrangement? As contract terms are defined, the provider organization should thoroughly understand its responsibilities under the agreement. This includes identifying the network components that will be most essential to driving value rewarded under the contract. For example, contract terms such as service exclusions, age restrictions, and the selection of specific disease or treatment-based performance incentives can inform elements of the provider network configuration, such as the proper of balance of primary care versus specialist physicians/APPs, inclusion or exclusion of specific types of specialists, and mix of pediatric versus adult providers.

How much financial risk and/or upside potential is associated with this arrangement? The type and amount of risk may inform considerations of how tightly aligned, through clinical and/or financial integration, the network should be to be able to assume shared accountability for a population. Similarly, the magnitude of both downside and/or upside risk may inform the organization’s willingness to invest in reconfiguring or expanding its network to position itself more effectively for the contract.

With a target population in mind and clarity around the value-based contract parameters, an organization’s analysis should focus next on determining the sizing and geographic-coverage requirements for the provider network. In particular, the organization should understand its overall financial exposure within a given model and project reasonable expectations around the network composition required to influence key goals related to total cost of care and quality.

Often, contract terms and the potential impact are directly influenced by scale. Therefore, before an organization can determine the exposure or upside potential of the contract, it must thoroughly understand the population likely to be attributed and the episodes to be managed. For example, in some models, the minimum-savings and minimum-loss ratios change based on the size of the attributed population.

Understanding Population Needs

Once the target population(s) for value-based care has been defined, the provider organization can employ a range of analytic approaches to enhance its understanding of that population’s specific needs. At a minimum, the network must meet standards for access and service mix for both capacity and geographic reach. A provider’s network requirements also are influenced by the population’s utilization and health needs as represented by its risk profile and demographics as well as factors such as consumer preferences.

The analytic approach to assessing population needs will depend on how much a health system knows about the population it will be responsible for managing. In some contexts, an insurer can provide enrollment details and past claims history to support a precise analysis of the covered population’s dynamics. In other instances, historical data on the contract’s covered lives will be limited or, in the case of new or prospective products, may not be available at all. In such situations, publicly or commercially available data sets can be used to calculate the approximate size and geographic distribution of the target population. For example, analysis of business locations and employee counts for major employers in the region can serve as a proxy for determining the geographic distribution of the potential covered population under commercial employer-sponsored plans.

Health systems should pursue analytics to answer questions such as the following.

What is the health status and risk level of the population? Segmentation of the population into subgroups based on risk profiles (e.g., high-cost or high-utilization members) provides insights into provider network requirements for each population segment. In the absence of claims data on the covered population, health status, access indicators, and trends can be extracted from the system’s own community health needs assessment work or public health sources, such as Robert Wood Johnson County Health Data.

How much health care does the population utilize? Projections of provider demand by specialty are available from a variety of sources. These projections can be based on physician-to-population benchmark ratios, panel sizes, claims data, or specialty need estimates. Some utilization estimates and demand ratios can be obtained from publicly available sources (e.g., the Centers for Medicare & Medicaid Services [CMS] and state all-payer utilization summaries where available) and others may require purchase of a proprietary data set. As more specific claims data are available, historical utilization can be compared with adjusted benchmarks to assess over/under utilization of various types of services.

How might the covered population grow or decline? Data from the U.S. Census or commercial consumer data vendors can be used to project changes in demographic mix (e.g., population growth or decline, changes in age distribution, shifts in socioeconomic mix) and determine how population-based factors will impact provider demand within the network over time.

Where does the population live and receive care? Optimizing the provider network requires an understanding of the healthcare migration patterns of the population in the value-based contract. This understanding can be obtained by overlaying population data, patient encounter data—either state inpatient, emergency department, and hospital outpatient data or claims data sets if available—and provider locations onto a map for analysis. Geo-spatial mapping tools, which enable statistical analysis and map-based visualizations of geographical data sets, can be used in analyzing these data sources to identify utilization “hotspots,” uncover patient preferred migration patterns, and calculate expected travel times, thereby making it possible to determine the number and location of providers by type required in each service area.

How do consumers prefer to access health care, and how do they make healthcare decisions? As patients assume responsibility for a greater portion of the cost of their healthcare services, so too will they want more say as consumers. Consumer segmentation frameworks that incorporate demographic and behavioral characteristics along with healthcare utilization data can inform how a health system’s provider network composition and access approach can be customized to meet consumer needs.

What other social determinants of health should be considered that influence population health outcomes and cost of care? Organizations may find themselves considering their network of social services and other support organizations as “second-order” network requirements beyond clinical services if they are to take accountability for a given population.

Understanding Care Model Impacts

Any population’s needs are fluid, so the next step for provider organizations is to investigate and understand how those needs will change and how changes to models of care delivery may alter the demand for providers within the network. A provider’s network-planning efforts should account not only for demographic dynamics, such as population growth and aging, but also for expected advancements that are likely to influence how care is delivered (e.g., telehealth adoption) and the impact that successful value-based care models could have on materially altering both the demand for and supply of services required (e.g., increasing preventive care and lowering demand for high-acuity services).

Organizations can use scenario-based modeling with a defined range around key assumptions as informed by organizational and broader industry experience. Some assumptions may be “directional” at best, but sensitivity analyses can enable an organization to test the impact these dynamics could have on future network demand or supply considerations.

In creating assumptions specific to value-based models of care, organizations should ask the following questions.

How do population health management care models affect the needs of a population and the capacity of a network to meet those needs? Core to primary care medical home (PCMH) and other value-based models is an increased role for primary care and an emphasis on team-based care. Organizations should consider not only the impact of increased demand for primary care services under these contracts, but also how these specific care models may affect primary care supply because the patient capacity (panel size) served by primary care physicians/APPs working in these models often can be quite different than that under traditional fee-for-service. For example, individual providers working within a team-based structure may manage a panel of more than 2,500 patients, while others working within an “intensivist” model for high-risk populations may care for under 500 active patients.

What is the potential impact of value-based models on utilization of unnecessary clinical services, and how will use of those models shift care delivery to the lower acuity/cost settings? Organizations may model shifting utilization patterns for specialty services that see an immediate impact from value-oriented efforts. For example, within a musculoskeletal service line, an organization may need to increase its utilization of medical treatments over surgical interventions, accelerate its migration of care from the inpatient to the outpatient setting, and/or increase the use of home-based care.

How will virtual care and other new access modalities affect the type and mix of services sought and delivered? Organizations should understand how virtual care models may change the demand for specific care and the overall capacity of provider networks within population health models. Utilization-based scenarios may test dynamics that are demand enhancements (e.g., net new services) versus replacements (e.g., increases in virtual care leading to decreased demand for traditional care) as well as those that increase provider capacity (e.g., tools to access and manage more lives, ability to utilize providers out of market) versus shift capacity (e.g., increased time spent on virtual care and less on traditional care encounters). Although not all digital health innovation is tied directly to population health models, closer alignment of payment to value may create new payment mechanisms to accelerate adoption of such services.

As organizations build more experience in value-based care, such analyses can migrate from a more conceptually informed approach to one reflecting actual changes experienced by the organization in its population health efforts.

Understanding the Physician/APP Landscape

This final dimension of provider network planning focuses on developing a comprehensive understanding of the provider supply in the market—including the health system’s employed and aligned providers and the broader market’s provider landscape. By assessing factors such as hospital/health system alignment and affiliation patterns, capacity to accept incremental patients, anticipated retirement and attrition levels, and consumers’ perception of “must have” or “must avoid” providers, it is possible to gain important insights for tailoring the provider network configuration to best serve a given population. Analytic methods utilizing internal, public and/or commercial data sets can be applied to inform elements of this dimension of the assessment.

A critical first step to applying this approach is to construct a comprehensive database of active physicians and APPs practicing within the market. At a minimum, the database should include provider name, national provider identifier (NPI), specialty, age, practice address, medical group, and known health system affiliations. Both public (e.g., NPI, Physician Compare) and commercial physician data sets can provide the foundation of the market provider database. Ongoing validation, augmentation, and updating of the database typically are required.

By integrating the market provider data with encounter/claims and other data sources, organizations can gain powerful insights to answer several strategic questions.

How many active physicians and APPs (by typeand specialty) are within the health system’s own employed/aligned group and in the market in the targeted geographies, and how is the supply of active providers expected to change over time? Using the market provider database described above, organizations can identify the active physicians/APP with practice locations proximate to the targeted population(s) for both their own employed/aligned providers and the broader market. Provider supply can be trended forward by calculating the expected level of attrition due to retirement or other factors and the projected number of new provider recruitments based on historical market levels along with the health system’s own long-range provider workforce development plan.

Is there sufficient capacity to meet the population’s expected demand for care under the contract—now and in the future—and to what extent is the organization’s current provider network aligned with the targeted population(s)? A key consideration here is whether there are any apparent gaps that may inhibit the organization’s ability to assume accountability for a defined population.The current and future supply should be analyzed against the demand projections calculated previously to determine whether enough providers exist by specialty within the health system’s own platform and the market overall to meet the targeted population’s needs. Health systems will want to hone these estimates further by considering available provider capacity. For employed providers, organizations can compare provider productivity levels or panel size to benchmarks or defined expectations. For other providers, an estimated panel range may suffice to estimate available capacity or, if available, analysis of market claims data can be used to calculate imputed panel sizes (e.g., number of unique patients seen over a defined period).

How are the physicians and APPs in the market employed or aligned (including participation in CINs or ACOs) with the health system or its competitors, and which providers/groups are most important to include in the network? Health systems should look beyond their employed medical group providers to build comprehensive networks that are attractive to consumers and employers and meet contractual access requirements. “Leakage” reports using actual claims experience from the value-based contract and shared patient analytics using publicly available data from CMS or commercially available data from claims vendors can help health systems identify providers to include who are not currently employed members.

In addition, publicly available quality and service data from CMS, online review sites, provider oversight bodies, commercial health plans, and other sources can provide additional insights into how specific providers/groups are performing or are perceived in the market. For example, in a market with significant numbers of providers who are members of multiple medical staffs and/or multiple CINs or ACOs, it is important to factor in considerations of exclusivity or non-exclusivity network or contract dimensions to understand the organization’s relative position to serve any given population or contract.

Using Analytical Insights to Advance the Population Health Strategy

An analytically informed approach for network optimization will generate insights into how to design provider networks to meet value-based care goals. Identification of both network strengths as well as critical gaps will help inform the organization’s physician/APP recruitment and alignment priorities. Such analysis also will enable finance leaders and their executive counterparts to better determine the appropriate future timing and pace of migration into value-based contracts. This network assessment work and the more robust understanding of the physician/APP supply in the market also can be useful for identifying options for filling existing network gaps (e.g., identifying provider partnership or acquisition targets, prioritizing areas ripe for organic physician growth).

Over time, an organization can advance its analytic capabilities and data sources as it moves further into value, so that analyses reflect the organization’s actual experience with defined populations and their associated needs and utilization. For example, as networks participate in value-based contracts and begin to receive reports and data from government payers and contracted health plans, claims-based analytics can drive insight into in- and out-of-network utilization and referral patterns that, when paired with relative cost and quality performance, can inform whether the organization may consider adding or removing specific clinicians or groups from its contracting network.

Finance leaders and teams will play an integral role in connecting the requirements for and performance on specific contracts to the network design and development process. Such network performance analytics will enable organizations to answer not only whether the network has been planned appropriately to serve a given target population, but also whether it can effectively create value for the populations it serves.

Eric Mayeda is director and head of analytics, iVantage Health Analytics and The Chartis Group, Chicago.

Anneliese Gerland is principal, strategy and value-based care, The Chartis Group, Boston.


googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text1' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text2' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text3' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text4' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text5' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text6' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text7' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-leaderboard' ); } );