Module 4: Measuring Performance Against the Target

Focus Area:
Sustainable Health Care Costs
Peterson-Milbank Program for Sustainable Health Care Costs

Once the target is set, states need to measure the change in annual per capita health care expenditures against the target. This is done using aggregate claims and non-claims spending data collected from payers, which requires developing specifications for data submission.

This section outlines considerations for how to approach the measurement of cost growth, identify the payer and provider entities whose performance will be measured, collect spending data, and analyze performance in relation to the target.

Define the Approach to Measuring Cost Growth

Define the Health Care Spending That Will Be Measured

All states calculate total health care expenditures (THCE), a measurement defined as the sum of total medical expense (TME) plus the net cost of private health insurance (NCPHI). All states define TME in terms of provider payments. TME comprises claims and non-claims payments to providers, and patient cost-sharing. States request aggregate claims data in broad categories, such as hospital inpatient, hospital outpatient, professional, pharmaceutical, and long-term care, to allow for deeper analysis. Non-claims costs include incentive program payments and prospective service payments, among others. These payments are increasingly important as more services are paid through value-based arrangements that do not flow through the claims system. To capture patient cost-sharing data, states require payers to report the “allowed amount” on a claim, which indicates what portion the patient owes the provider according to the patient’s benefit plan.

What Is Not Included in Total Health Care Expenditures?

Stakeholders in many states have expressed a desire to include spending by the uninsured in measuring cost growth. However, no state has been able to do so because there is no comprehensive source of such data.

Similarly, hospitals have noted that uncompensated care constitutes a significant medical expense that is not included in the measurement. Nationally, uncompensated care costs for uninsured individuals reached nearly $43 billion in 2020.1 These costs include charity care — free or deeply discounted services for patients who cannot afford treatment — for which hospitals must budget, and “bad debt,” or write-offs for bills that go unpaid. These are not considered payments to providers, and therefore do not represent spending as defined by states. No state has developed a provision to subtract uncompensated care from a provider’s spending performance. Because of the administrative burden of reporting charity care and bad debt consistently across all providers in a state, states have accepted these as known challenges to complete measurement for now.

NCPHI is the spending associated with administering private health insurance and is calculated as the difference between health premiums earned and benefits incurred. It includes administrative expenditures, net additions to reserves, rate credits and dividends, and profits and losses.

Define the Population Whose Spending Will be Measured

All states measure costs for the commercial, Medicare, and Medicaid populations, as they typically represent about 90% of all covered individuals in a state.2 To be more inclusive, some states have also considered incorporating spending for populations that receive health care coverage through other sources, such as veterans who typically access health care through Veterans Health Administration (VHA) facilities, incarcerated individuals for whom the state pays health care costs, the Native American population that receives care from the Indian Health Service, and employees who receive workers’ compensation health care benefits. In determining whether to include these types of health care spending, states need to account for data availability and whether the gain from including the additional spending outweighs the level of effort involved to access the data.

Currently, all states measure the health care spending of all state residents with commercial, Medicare, or Medicaid coverage, regardless of whether they seek care in or out of the state. States have also considered measuring spending of (1) state residents who seek care only from in-state providers, or (2) all individuals who seek care from in-state providers, regardless of where they live. However, no state has pursued these options due to the data collection and reporting challenges of segmenting data by provider location and/or a decision to focus only on spending associated with state residents.

Another consideration for states is what population to use as the denominator for calculating per capita spending. Reporting on a per capita basis allows states to account for migration and population changes that could significantly affect total health care spending. It also facilitates comparisons of cost growth between states that have different population sizes. States can take one of two approaches:

  • Use the state’s total population. Massachusetts calculates state performance against the target by taking the change in THCE and dividing it by the state’s entire population. Policymakers felt using the entire population was reasonable because Massachusetts has very low rates of uninsurance. However, using the total population in the denominator and using only spending reported by payers in the numerator could mask the true cost growth if there is a significant shift in the number of people who are uninsured.
  • Use membership figures reported by payers. Rhode Island uses the annualized number of member months reported in the data collection process as its denominator for calculating per capita spending and cost growth. In Rhode Island, the number of individuals for whom payers reported data was significantly smaller than the state’s population, possibly because some residents work in bordering states and are insured by out-of-state payers. When including spending from other sources such as the VHA, Department of Corrections, and workers’ compensation, states need to think carefully about how to use reported membership to avoid double-counting individuals.

Consider Strategies to Strengthen the Accuracy and Reliability of Target Performance Measurement

Because public reporting of performance against the target involves identifying specific entities’ cost growth, it is important to have confidence in the measurement. At the state and market levels, population sizes are significant enough that measurements are statistically stable and there is no need to apply additional methodologies. At the payer and provider levels, however, states should consider additional strategies to ensure the accuracy and reliability of assessments of cost growth:

  • Develop confidence intervals around an entity’s cost growth. This allows a provider entity’s performance to be reported as a point within a range of values. The state then determines performance based on whether that range intersects with the target value.
  • Truncate spending of high-cost outliers. High-cost outliers are people with extremely high levels of annual health care spending, who mostly are distributed randomly in a population. Some states mitigate their impact on payer and provider entity trends by removing per-member or per-patient expenditures above a certain threshold.
  • Decide not to apply clinical risk adjustment. Risk adjustment is a statistical process used to account for a population’s underlying health status when looking at their health care outcomes or costs. Some states risk-adjust spending data submitted by payers when assessing performance against the target. However, states’ experience and other empirical research show that clinical risk scores used for risk adjustment have increased substantially over time due to changes in how providers code patients’ conditions, and not because of actual decline in the population’s health status.3 Thus, applying clinical risk adjustment in target performance assessment could cause payer and provider organizations’ cost growth to appear lower than it actually is. Consequently, some states moved toward risk-adjusting only by age and sex to avoid overstating the population’s illness burden, and some states dropped risk adjustment altogether.
  • Establish a minimum number of members/patients for payer- and provider-level reporting. Setting a minimum threshold for the number of enrolled or attributed individuals that a payer or provider should have before performance is reported helps minimize the impact of random variation on cost trend performance. Based on analyses performed in multiple states, the recommended minimum threshold for publicly reporting performance is 5,000 members/patients at the payer and provider levels.

Identify the Entities That Will Be Held Accountable to the Target

Most states measure cost growth at the state, market, payer, and provider levels. Reporting at the state and market levels is straightforward once the state develops its measurement approach. For payer and provider entity reporting, states must first identify the payers and provider entities whose cost growth will be measured and reported against the target. Medicaid managed care states usually require all managed care contractors to report data for the target program. For the Medicare Advantage and commercial markets, states aim to include enough payers to capture approximately 85% to 90% of covered individuals in those markets. A state’s department of insurance typically collects and publishes information on payers’ market share, which states can use to identify which insurers should be required to report. However, commercial market data are usually limited to fully insured plans that are state-regulated.

In defining the list of provider entities, states typically include large provider entities that can be reasonably expected to influence total health care costs, such as medical groups, health systems, federally qualified health centers, and independent practice associations. Some states identify provider entities by whether they have a total cost of care contract. Other states include provider entities deemed large enough to have a total cost of care contract, whether or not they do so.

Once a state defines the list of provider entities, it must develop clear specifications on how to attribute member-level spending to provider entities (Exhibit 3). This requires two levels: (1) attribution of members to a clinician and (2) attribution of clinicians to a large provider entity.

EXHIBIT 3. Process for Attributing Spending to Large Provider Entities

To date, all states use a primary care–based methodology for attributing members to providers. This approach is a matter of necessity, not policy choice, as no method is available to associate per capita spending with other types of entities on a large scale. Some states leave the specific methodology to the carriers, although a few states, such as Oregon and Washington, ask insurers to follow a hierarchy that prioritizes member selection of primary care provider, followed by attribution used in value-based payment (VBP) contracts, and then utilization.

Ideally, a state will have a provider directory that maps each primary care provider to a large provider entity so that attribution is consistent across insurers. However, very few states maintain a statewide provider directory. Consequently, most states have required insurers to attribute providers to large provider entities based on their contracting arrangements. This is an imperfect approach, as different payers may contract with different configurations of provider organizations and may have different arrangements with the same provider organization by market.

Develop and Implement a Process and Timeline for Collecting, Analyzing, and Reporting Data

Target programs require significant and ongoing investment in data collection and analysis. The process typically takes approximately one year from data collection to reporting of results (Exhibit 4).

EXHIBIT 4. Typical Timeline for Collecting, Analyzing, and Reporting Target Performance Data

Because of typical delays in reporting claims and the time required to reconcile alternative models of payment, the earliest that states can require data submission is usually six months after the end of a performance period. For example, performance data for calendar year 2023 would not be available until at least summer of 2024. This determines the timing of related activities, including preparing for data collection, validating and analyzing data, and reporting results.

Document Specifications and Review Them with Data Submitters

States must develop specifications to ensure data are reported consistently. Data specifications should minimally include:

  • Description of the target policy
  • Formulae for developing the target
  • Methodology for calculating total health care spending
  • Data reporting specifications, such as population inclusions and exclusions, definition of service categories, and types of spending to include
  • Process for publicly reporting the results

States set most policies during the first year of implementation when they make key design decisions around target performance measurement. However, states should review these methodologies each year and adjust on the basis of experience with data collection and analysis, innovative practices developed by other states, and changes in the state’s health care landscape. It is also helpful to review other states’ methodologies, and, where appropriate, aim for consistency to minimize the data reporting effort for health plans that cover members in multiple states with target programs.

States should review the data submission process and specifications with data submitters to educate them and clarify the data request. This review should take place annually to accommodate new data submitters, turnover of analysts responsible for submitting data, and implementation of new methodologies.

Oregon’s Use of a Technical Advisory Group to Obtain Feedback

Before requesting data from health plans, the Oregon Health Authority (OHA) created a Technical Advisory Group (TAG) to work with OHA on its specifications and data collection process. The TAG meets monthly and is open to all payers that submit data, provider organizations, and other interested parties. OHA uses the TAG to gain insight into how best to collect data, gather feedback on implementation, and answer questions on the data collection and reporting process. Oregon payers report that this group is a positive approach to two-way communication on data reporting.

Collect, Validate, and Analyze Data from Multiple Sources

States must obtain health care spending data from multiple sources, according to the chosen methodology, including the following populations:

  • Commercial fully and self-insured: Commercial fully insured and self-insured spending data come from health insurers operating in the state. All states with target programs obtain aggregate spending data from insurers (not claim-level, member-level, or employer-level information).
  • Medicare: Medicare spending data typically come from two sources: the Centers for Medicare and Medicaid Services (CMS) and Medicare Advantage carriers. CMS offers a consistent set of data to states. While not completely aligned with state specifications, it is an excellent source for Medicare fee-for-service spending and all Part D (retail pharmacy) spending. Medicare Advantage carriers are a better source than CMS for Medicare Advantage product spending as the carriers can submit data according to the state’s specifications.
  • Medicaid: In non–managed care states, all the data will come from the state. In Medicaid managed care states, a significant portion of the data will come from the state’s contracted managed care organizations, and some will come from the state’s fee-for-service (FFS) program. States need to carefully develop a methodology to obtain nonduplicated information for the managed care and FFS populations. Duplication can occur, for example, when certain services for managed care populations are carved out for different coverage or when the state provides wraparound services through the FFS system.
  • Medicare and Medicaid dually eligible: Because of the many different combinations through which dually eligible individuals can receive Medicare and Medicaid benefits, states need to pay special attention to capture costs for this population appropriately. FFS spending information for dually eligible individuals is embedded within data supplied by CMS and, depending on the state, may be included in Medicaid FFS data supplied by the state Medicaid agency. Dually eligible individuals can also be covered through Medicare Advantage, Medicaid managed care, or, in select states, through the CMS Financial Alignment Initiative, which provides Medicare and Medicaid coverage through a unified plan. States need to tailor their data specifications and reporting processes for the dually eligible population to be clear on which entity reports what spending and to avoid omitting or duplicating any spending data. Approaches will depend on how the state provides Medicaid coverage to dually eligible individuals (e.g., through FFS, managed care, or an integrated Medicare and Medicaid product).
  • Other populations: States that choose to include spending on other sources of coverage — such as the VHA, Indian Health Service, state corrections, or workers’ compensation — need to collect data for those populations from the respective entities or agencies.

States need to validate the data received to ensure consistent reporting according to specifications, particularly in the first years of implementation. Flawed data can result in incorrect assessments of entities’ target performance. Ensuring entities are assessed correctly before performance is reported publicly is critical. Exhibit 5 depicts a process that states can implement to promote integrity and stakeholder confidence in the cost data.

Why States Can’t Use APCD Data to Measure Performance Against the Target

To minimize data collection burdens, some states with fully functioning APCDs have proposed using APCD data to measure cost growth. Yet, health insurers continue to be the most complete source of spending data for the commercial, Medicaid managed care, and Medicare Advantage populations.

APCDs lack pharmacy rebate amounts that are used to produce a net pharmacy spending calculation. In addition, APCDs typically lack payments made to providers outside of the claims system, such as incentives, shared savings, or other similar value-based payments. Finally, APCDs do not include self-insured groups, which typically represent well over half of the commercially insured population in a state.

EXHIBIT 5. Process for Collecting, Validating, Analyzing, and Reviewing Cost Data

The data validation process can be lengthy, and payers may need to resubmit data multiple times, particularly when they are new to reporting target performance data. Providing comprehensive upfront assistance and tools for data submitters will reduce the need for resubmission later in the process. For example, some states’ data submission templates include validation steps that allow data submitters to review trends before submission. States should conduct two types of validation checks:

  • Completeness checks ensure there are no obvious errors or omissions. For example, states should check each submission to ensure it has all the required data elements and includes the expected lines of business for a particular payer.
  • Reasonableness checks ensure the data are appropriate at face value and when compared with other sources. These subjective assessments can point to potential errors. For example, high per member per month spending on long-term care for a commercial insurer may point to an error since commercial plans typically do not cover many long-term care services. States can also compare Medicare member months submitted by a payer to Medicare Advantage enrollment data published by CMS to confirm that the payer included the appropriate population. Analysts can also look at year-over-year changes in populations and per capita costs, and probe areas that show significant increases or decreases.

Once a state is confident in the quality of the data, it can move on to analysis. The primary analyses consist of calculating performance at four levels:

  1. Overall state performance: The growth in per capita spending, as measured by THCE, in the state compared with the target.
  2. By market: The growth in per capita spending, as measured by THCE or TME, in each of the Medicare, Medicaid, and commercial markets compared with the target.
  3. By payer, by market: A single carrier’s THCE or TME performance for each of the markets in which it operates and for which the carrier has sufficient members.
  4. By provider entity, by market: A single provider entity’s TME performance for each of its markets, so long as the number of attributed patients meets a predetermined threshold.

States can also conduct additional analyses, such as aggregate spending at the state and market levels, costs and cost growth by service categories (e.g., hospital inpatient, hospital outpatient), and how much growth in spending in a service category contributed to overall cost growth. These reviews provide important clues about where to conduct more in-depth analyses of claims databases.

Data Validation Tips

  • Ensure that individuals conducting the validation have knowledge of market trends when determining the reasonableness of data.
  • Create a validation checklist to ensure consistency when reviewing multiple submissions.
  • Start with the largest payers, whose data will have the greatest impact on overall results.
  • Document every observation, conversation, and decision, and circulate notes to ensure all parties agree on the next steps.
  • Re-review everything in a submission, since new issues could arise as a result of resubmission.

Review Results with Payers and Providers and Publicly Report Performance

States should confidentially review the results with payers and providers whose performance is measured against the target before formally reporting results. This review provides another quality control check, gives entities the opportunity to understand and identify reasons for their performance, and helps foster goodwill between the state and those entities.

In reviewing results, provider entities may compare their target performance with their performance on total cost of care contracts, if they contract on that basis.

Variation in findings can occur for several reasons. TME and total cost of care contracts may define services differently. For example, some total cost of care contracts may not hold a provider responsible for certain services, like pharmacy or long-term care expenditures, while those are included in target policies. They may also apply risk adjustment and deal with high-cost outliers differently.

States should disseminate the results for state, market, payer, and provider performance against the target via several mediums, such as a presentation to the program’s governing body, a public forum focused on affordability, an issue brief on the findings, and other strategies outlined in the stakeholder engagement activities described in this playbook. In addition to reporting cost growth, states should consider presenting employer and consumer perspectives on affordability to reinforce the importance of controlling cost trends. For example, at Rhode Island’s Health Care Cost Trends Public Forum in April 2022, a small employer described the financial squeeze experienced by employees. This employer described the limited ability to raise employee wages because of high benefit costs and employees’ limited ability to afford high-deductible health plans. These types of stories provide human interest, context, and further justification for the target policy.

Reporting Total Health Care Expenditures (THCE) or Total Medical Expense (TME) at the Market and Payer Levels

Some states have elected not to report THCE at the market and payer levels because of the year-to-year volatility of the net cost of private health insurance (NCPHI), a component of THCE. NCPHI can vary significantly from one year to the next as payers post profits or losses on certain products, premium rates change, or federal tax and refund policies change. Additionally, these data can be hard to validate. Measuring NCPHI is important, but some states prefer to focus on TME, which accounts for the vast majority of health care spending.


Health Care Cost Growth Target Data Specification Manuals

Data specification manuals provide instructions to payers for how to submit data the state needs to calculate state- and market-level cost growth and payer and provider performance against the target.

Health Care Cost Growth Target Data Submission Templates

These data submission templates are used to collect TME data from payers.

Technical Implementation Webinar Materials and Recordings

Data Sources for Calculating the Net Cost of Private Health Insurance

  • CMS publishes Medical Loss Ratio data that health insurers are required to disclose under the Affordable Care Act.
  • The National Association of Insurance Commissioners makes available for purchase data from Supplemental Health Care Exhibits that insurers submit to states.
  • The Securities and Exchange Commission publishes Company Filings, which can be used to estimate commercial self-insured NCPHI if information on income from fees of uninsured plans is not available.

Health Care Cost Growth Target Performance Reports


  1. American Hospital Association. Fact Sheet: Uncompensated Hospital Care Cost. February 2022.
  2. Kaiser Family Foundation. State Health Facts: Health Insurance Coverage of the Total Population. 2019
  3. Massachusetts Health Policy Commission. 2019 Annual Health Care Cost Trends Report. February 2020.