Jennifer L. Zutz, MHSA

To identify quality measures that health plans can reliably report during the early years of health insurance exchanges and over time, and to suggest strategies to increase the availability of quality results to use in rating and monitoring plans.

 Projection of a set of measures available for public reporting based on prevalence and experience with health plans’ reporting of the quality measures.

For the quality measures included in the federal Initial Core Set of Adult Health Care Quality Measures for Medicaid-Eligible Adults, we looked at the proportion of people in the United States who would be eligible for each measure, and if available, the number of plans that in the past were not able to report reliable Healthcare Effectiveness Data and Information Set results to the National Committee for Quality Assurance because of low membership. We developed estimates of each state’s exchange enrollment, and used the number of plans currently accredited in the state to estimate how many plans will offer coverage to determine the average exchange plan membership per state in 2014 and 2018.

 In the early years exchange plans should be able to report a set of 14 preventive, chronic care, and access-to-service measures. As health plan membership grows through the years, more measures, including behavioral health, can be phased in.

 In 2015 and 2016, all exchanges should require plans to report the 14 measures and if needed, use suggested strategies to build the results for public reporting.

Beginning in 2015, health plans are required to report quality results to health insurance exchanges. As exchange enrollment grows, more quality results will be available for public reporting to consumers and policy makers.

		Exchanges should require health plans to report a starter set of 14 preventive, chronic care, and access-to-service measures for which the majority of plans are likely to have valid and reliable results.

		There are strategies to increase the availability of valid and reliable results for public reporting.

		Exchanges should phase in additional measures that focus on less prevalent areas of clinical care (eg, behavioral health).

The Affordable Care Act requires health plans offering coverage through Health Benefit Exchanges (exchanges) to report quality measure results for use in calculating quality ratings for consumers and for monitoring quality improvement.

 The requirement will make available information from plans that previously did not publicly report performance measure results and information on populations that have been uninsured. Especially if states and the federal government require the same measures, this requirement will support transparency, value-based purchasing, alignment among payers, quality improvement, and the national quality strategy.

The first few years of the exchanges will be challenging on a number of fronts, including establishing of new health plan oversight, reaching out to potential enrollees, and setting up systems that properly administer subsidies. If consumers are to pick plans with a good track record on clinical care and patient experience, then implementing the quality reporting and rating requirements needs to be a top priority.

As required by the Affordable Care Act, the Department of Health and Human Services will develop a quality rating system for exchanges.

 States can choose whether to use the federal rating system or develop their own that meets federally specified criteria. Because states are focused primarily on building the computer systems to support exchanges (enrollment, query of income information, administering subsidies), many are likely to use this federal approach rather than devote the resources to building their own systems, although they may turn to their own system building later on. The Department of Health and Human Services has not released any specific rules or guidance on the federal rating system, but has suggested the quality rating methodology will be implemented after the 2014 start of the exchanges.

 The federal government is not required to define quality measures for state exchanges to report beyond those underlying the quality rating system, although it can require uniform reporting in the states where the federal government runs the exchange. The federal government can also recommend a core measure set to state-run exchanges, which will promote comparability across states and assist states in their exchange development.

This article suggests a set of clinical quality measures for which health plans are likely to have reportable results in the very early years of exchange operation. We used estimates of exchange and health plan enrollment, market conditions, and prevalence of clinical conditions to determine the quality measures for which most plans offered by exchanges will have sufficient enrollment and statistically reliable results in the first years of the exchanges. We then recommend a phase-in approach to expand measure reporting over time. To construct such a measure set, we began with the already established core set of quality measures for the Medicaid population.3

To have reportable results for clinical quality measures for individual health plans, health plans must have sufficient enrollment to support statistically reliable results. This is a function of the population addressed by an individual measure (eg, adults with asthma) and the prevalence of the measure population (all enrollees with asthma). We began with the Medicaid core set as the measure set most applicable to the uninsured population to get coverage through the exchanges.

 We then estimated enrollment in plans in the early years of exchanges and therefore the ability of exchanges to have reportable results from plans. Thirty enrollees is the commonly used sample minimum for statistically reliable quality results.

In this article we use the term “reportable” to refer to plan results with at least 30 observations in the denominator, or appropriate population, for a measure. Throughout this article, we define “health plan” as the entity that is licensed in the state and is issuing products to market through the exchange.

To determine whether exchange plans were likely to have reportable results for specific quality measures, we used readily available data to identify the proportion of people in the United States who would be eligible for each of the 25 core Medicaid measures (ie, the prevalence of the measure population). For example, there are 25.8 million diabetic patients within a population of 216 million. Thus, about 12% of the exchange population could be included in the measure that assessed the quality of diabetes care.

We supplemented information about disease prevalence with actual rates of successful reporting of measures in cases where we had that information. For 12 of the 25 Medicaid measures that are National Committee for Quality Assurance (NCQA) Healthcare Effectiveness Data and Information Set (HEDIS) measures, we had publicly available data on the number of enrollees in commercial and Medicaid plans that were able to report HEDIS results in 2011. (In 2011, NCQA collected HEDIS results from 286 commercial plans, 127 Medicaid plans, and 368 Medicare plans.) Based on experience collecting clinical quality measure results from health plans, NCQA found that a reasonable estimate of the number of individuals who need to be enrolled in a health plan to support reporting of a number of reliable clinical quality measure results is 15,000.

We looked at HEDIS results for plans with fewer than 15,000 enrollees and plans with more than 15,000 enrollees. (HEDIS is a registered trademark of NCQA.) We then analyzed how many plans in each of those 2 groups had reportable results (at least 30 observations in the denominator population) and how many did not. Even though many policy makers are interested in health plan performance for individual benefit packages (platinum, gold, silver, bronze), many more plans would be unable to report quality results for a number of years if they had to report at the benefit package level.

To estimate average plan enrollment and understand the implications of enrollment for having reportable quality measure results, we used available estimates of how many people would enroll in each state’s exchange. To develop state-by-state exchange population estimates, we used data available from the Kaiser Family Foundation (statehealthfacts.org) to identify how many uninsured people per state could enroll through exchanges and how many people in each state are covered by individual insurance.

 We used national Congressional Budget Office projections of the percentage of currently uninsured and individually insured populations who would be enrolled in exchanges in 2014 and 2018.

Using Hawaii exchange enrollment in 2014 as an example, the Congressional Budget Office projects that 33% of the nation’s uninsured will move into the exchange (12,200 of the 37,100 uninsured in the state). Additionally, the Congressional Budget Office projects that 7% of the nation’s individually insured will move into the exchange (2000 of the 28,600 individually insured in the state). In total, Hawaii’s exchange can expect to enroll about 14,200 in the exchange in 2014 (

We developed an estimate of the number of health plans that will participate in each exchange based on the requirement that plans must be accredited by NCQA or the URAC to participate in an exchange.

 The estimate was based on the current number of NCQA-accredited commercial and Medicaid plans in each state, information that is publicly available on NCQA’s website. To test this estimate, we looked at how many health plans participate in the Massachusetts Connector, the health insurance exchange established in 2007, and compared that number with the number of NCQA-accredited commercial and Medicaid plans in Massachusetts.

Because fewer plans than are currently NCQA accredited may participate in the exchanges, we used a range for the number of health plans that will participate in each exchange. The high end of the range is the number of NCQA-accredited plans in the state and the low end of the range is half of the number.

We did not attempt to model disease prevalence in the exchange population. Instead, we used readily available national statistics. That probably resulted in an underestimate of disease burden. Four years after Massachusetts established its state exchange, the newly insured population reported sustained gains in healthcare access and use compared with before the exchange.

 So the exchange population is less likely to have received preventive care and more likely to need and seek care.

We assumed that the number of plans that will participate in the exchange will not change from 2014 to 2018. This is consistent with the Massachusetts Health Connector experience, in that only 1 new plan entered the market 4 years after the exchange was established. We also assumed that each plan will have an equal share of the exchange enrollment in the state (even though we know that the distribution of enrollment will vary).

We focused on measures of clinical care for adults. Measure results for pediatric care are also required of health plans in exchanges.

 Policy makers could use the same general approach to determine which measures could be included in a starter set of exchange pediatric quality measures.

Our discussion applies only to clinical quality measure results. It does not apply to Consumer Assessment of Healthcare Providers and Systems survey results, also required by the Affordable Care Act to be reported by plans offered by exchanges. These survey results should be available and reportable by virtually all plans after the first year of operation, regardless of enrollment size. These results will provide additional important information on plan quality.

Based on the population prevalence and experience with plan reporting, health plans, regardless of their enrollment, are likely to have reportable results in 2015 for a starter set of 13 of the 25 measures in the Medicaid core set (

). An additional measure, breast cancer screening, will be reportable in 2016 since clinical guidelines recommend screening every 2 years. These measures had a relatively high calculated prevalence (

12%) and most of the plans regardless of membership size were able to report it. This starter set of 13 measures covers a variety of types of clinical care including 5 preventive, 1 medication management, 2 access and availability of services, 2 cardiovascular, and 3 diabetes measures. Of the 13 measures, 7 can be reported using only administrative data, which are less resource intensive for plans than accessing medical record data. Two of the measures are collected through member survey results from the Consumer Assessment of Healthcare Providers and Systems survey.

As mentioned earlier, NCQA has found that 15,000 enrollees is a reasonable estimate of the enrollment in a health plan needed to support reporting of a number of reliable clinical quality measure results. In 2014, we estimate that between 14 and 32 exchanges (depending on assumptions) will have plans with more than 15,000 enrollees (

). Between 19 and 37 exchanges will have plans with fewer than 15,000 enrollees. In these states, plans may not be able to publicly report some of the measures in the starter set. This will be particularly true for the 5 to 14 exchanges that we estimate will have plans with fewer than 5000 enrollees. For example, in Delaware an exchange plan is likely to have about 1900 to 3800 enrollees in 2014. The predicted 4 to 9 health plans in the state may not be able to report all 14 measures, especially the ones with lower prevalence (eg, readmissions, diabetes results).

As more people enroll in exchanges in future years, enrollment per plan will increase and plans will be more able to report the full starter set. In 2018 we estimate that in most exchanges (37-46 exchange health plans), all offered plans will have more than 15,000 enrollees. By then, reportable results are likely to be available for all the measures in the starter set.

Taking into account the prevalence of disease in the measured population and the size of plans, we find 11 measures from the Medicaid core set that many of the small plans are unlikely to be able to report in 2014. These include 4 behavioral health, 5 access and availability of service, and 2 respiratory measures. By 2019, however, many plans offered by exchanges will be able to report these 11 measures. Large states in particular are the most likely to have reportable results for these measures by 2019. Below we discuss ideas on how to increase the availability of measure results in these important clinical areas for plans with lower enrollment.

Plan enrollment in the early years of the exchanges is likely to support public reporting on a starter set of 14 measures. These measures include those for prevention, chronic conditions, and access to services. To have a full understanding of the quality of care provided through the exchanges, the federal government can encourage exchanges to collect this set of 13 measures from plans starting in 2015 (plus breast cancer screening in 2016). Exchanges with larger populations could add more measures to this initial list, such as measures of respiratory and cardiovascular conditions (eg, cholesterol management for patients with cardiovascular conditions).

We recommend that states and the federal government analyze the results of the first year of reporting to determine reliability and variation in the results. First-year measure data in 2015 will help policy makers better understand their exchange population and health plans. The data will also confirm what measure results are publicly reportable. Policy makers should plan to perform basic analytic tests on the first-year data to determine reliability (how many plans have results on individual measures for at least 30 enrollees) and understand variation across plans. If the first-year data meet these tests, exchanges can use those data as the basis for a quality rating for consumers or other public reporting. If after the first year or two of data collection, many plans do not have reliable results, then exchanges can consider other strategies (described below) to increase sample size to support public reporting of results. States can also use these strategies to increase the availability of reportable results for the 11 measures not included in the starter set.

Strategies for Increasing Availability of Reportable Measure Results

Articles

This article recommends quality measures for exchange health plans and strategies to increase the availability of quality results for public reporting.

Jennifer L. Zutz, MHSA

Articles

Measuring Quality in the Early Years of Health Insurance Exchanges