The Health Insurance Exchanges required under the Patient Protection and Affordable Care Act (ACA) have the potential to create markets that will help millions of Americans get affordable health insurance coverage and access to high-quality care. The exchanges can enhance consumer and employer buying power to improve the overall quality and efficiency of the health care system.

To reach those results, Exchanges will have to be effective in various challenging functions. A key function is giving consumers and employers a health plan comparison tool to assist them in selecting the plans that best meet their needs and preferences.

Without such a tool, consumers will make plan choices that are bad for themselves and possibly bad for the health care system. Tests done by the Pacific Business Group on Health (PBGH)  have found that “participants do only slightly better than chance at selecting the ‘best’ plan, even in simplified environments.” Research by Consumer Reports has found that consumers “dread” shopping for health insurance, and that the “difficulties are so profound that the vast majority of consumers are essentially being asked to buy a very expensive product—critical to their health—while blindfolded.”

Unfortunately, most of the current “exchanges”—the Massachusetts Connector, the Federal website, the Utah Health Exchange, and others—fall far short when it comes to providing a plan comparison tool that addresses consumer needs. In hopes of avoiding such a shortcoming within Health Insurance Exchanges under the ACA, this post will recommend best-practice features to be built into any plan comparison tool.

These recommendations are based on the research, testing, evaluation, and experience of Consumers’ CHECKBOOK/Center for the Study of Services over the past 33 years as it has provided its Guide to Health Plans comparison tool for the eight million consumers who get insurance through the Federal Employees Health Benefits Program (FEHBP).  And these recommendations also draw on what CHECKBOOK/CSS has learned as its team has advised many thousands of consumers individually and in health benefits forums.

More information on these recommendations, some of the research and analysis of others who have focused on this challenge, and a demonstration model of the type of plan comparison tool CHECKBOOK/CSS recommends are available on the CHECKBOOK/CSS comparison tool webpage.

A health plan comparison tool in Exchanges should let users see:

  • The true insurance value of each plan—a single dollar-amount estimate of average total expected cost under each plan (premium after any tax and subsidy effects, plus out-of-pocket costs) for households similar to the user in age, family composition, and other characteristics—based on actuarial analysis of data showing the probability of different total amounts and types of expenses in the population. A description of deductibles, coinsurance levels, etc. is not enough; nor is a dollar-amount out-of-pocket estimate based only on expenses the user can predict.
  • Possible expenses in each plan in very good years and very bad years (including years when the user’s expenses exceed plan out-of-pocket limits) and the likelihood of having such years.
  • Likely effects on out-of-pocket costs of known expensive future procedures or treatments—for example, an expensive operation or a pregnancy.
  • An Exchange-wide provider directory so consumers can easily see which plan networks include their doctors, and can see quality measures for each available doctor and hospital.
  • How plans compare on care and service quality—plan ratings by members, quality and reach of plan-provided health improvement programs, accessibility of high-quality providers, frequency of member complaints, accreditation, etc.—allowing the user to focus on the quality dimensions of greatest personal interest.
  • Any coverage gaps and any unusual benefit strengths—and why they matter.
  • A simple comparison of all available plan choices followed by easy opportunities to filter and narrow—not encouraging users to narrow their choices with preliminary questions (Will you consider an HMO? What’s the highest deductible you will accept?) before they have seen the range of choices and what they might give up by ruling out options.
  • Clear, simple explanations and videos that will de-mystify insurance decisions even for unsophisticated users.
  • Excellent, personalized plan choices in the short time most users will allow, generally in less than five minutes—so the user doesn’t drop out and make a poor choice based on simplistic criteria (lowest deductible, lowest premium). It is not necessary to narrow the number of available plans to simplify consumer choices; a good comparison tool can get consumers to an excellent plan choice quickly and easily. But the tool must allow users to drill down for extensive detail, if they are able and so inclined.

The online version of such a tool should also be designed to help family members, counselors, Navigators, brokers, and other intermediaries give personalized advice and prepare personalized written materials.

Insurance Value of Each Plan Versus Other Cost Comparison Approaches

CHECKBOOK/CSS’s surveys and observation of use patterns, and research by Consumer Reports, PBGH, and others, have shown that consumers choosing health plans are most interested in comparing the total costs they can expect with different plans.

Several different approaches have been used in tools intended to help consumers compare plan costs.  These alternative approaches are described here, along with a description of what a best-practices model should do. This recommended model is very different from other approaches but incorporates the useful elements of each approach.

Benefit and Coverage Comparisons

Unfortunately, the cost comparisons in most plan comparison tools go no further than giving descriptions of each plan’s coverage provisions, including deductibles, co-payments, coinsurance, and out-of-pocket limits.  This is currently true of the Massachusetts Connector,, the Utah Health Exchange, Maryland’s Virtual Compare website, and many others.

Normal consumers just cannot assess the dollar consequences of the coverage differences.  Yet, to find good value, it is essential for the user to know how these different coverage provisions can be expected to impact actual out-of-pocket costs. Is a $200 deductible with a $10,000 out-of-pocket limit better for my family than a $1,000 deductible and a $4,000 out-of-pocket limit?  What about differences in coinsurance percentages, in whether the deductible does or does not count toward the out-of-pocket limit? Etc.?

The type of tool CHECKBOOK/CSS recommends has the coverage descriptions—taking advantage of, among other inputs, the information from the Summary of Benefits and Coverage for each plan as specified under the final ACA regulations. But even the best such descriptions are much less than what consumers actually need. A good tool for Exchanges must go further.

Known Usage Model

Another approach, which might be referred to as the “known-usage” model, is to have the user input all or most of the health care system uses the user expects to have in the coming year –how many of which drugs, how many doctor visits, etc. — and then have the comparison tool estimate a typical provider charge for each of these uses and calculate how much the user would have to spend out of pocket under each plan as the member’s share of those predicted expenses.  That approach has some intuitive appeal, and is the common approach for deciding how much to put into a flexible spending account, but it falls far short of being sufficient for selecting insurance plans.

The fundamental problem with such a known-usage approach is that a key reason for insurance is to protect the policyholder against the cost of what the policyholder can’t predict—a serious accident, new disease, or new treatment plan.  With the known-usage approach, the out-of-pocket cost estimates don’t reflect those unexpected costs—though reflecting them might dramatically affect the relative ranking of plans.  Another problem is that with even one moderately complex medical condition, or in a family with several members with different conditions, it is time-consuming and often impossible to estimate future usage by type of expense.

The type of tool CHECKBOOK/CSS recommends will allow users to adjust cost calculations to take into account large known upcoming expensive procedures or treatments, such as childbirth.  And it will certainly take into account known factors that contribute to risk — age, family size, and possibly factors like self-reported health status.  But a key distinguishing feature of the recommended tool model is that it will also take into account possible costs—potentially very large costs—that cannot be predicted.

The Insurance Value Model

The type of tool CHECKBOOK/CSS recommends does this by featuring an “insurance value” approach. It provides a best estimate of average expected costs for the user in the coming year (premium plus out-of-pocket costs) based on extensive data on the distribution of individual and family expenses of persons similar to the user (similar age, family size, etc.) even if nothing specific is known about future usage. For this estimate, the recommended model uses data from the Federal government’s Medical Expenditure Panel Survey (MEPS), which shows the distribution of expenses for a sample of Americans. The recommended model can also take into account other data that are becoming available from other sources, including All-Payer Databases.

CHECKBOOK/CSS has for many years used data of this kind in the analysis that is the basis for its Guide for Federal employees.  CMS’s comparison tool for Medicare Advantage plans, using data from the Medicare Current Beneficiary Survey (MCBS), has also in recent years had elements of this approach, but unfortunately appears to be the only other tool currently taking this insurance-value approach.

Based on using millions of patient expenditure records, this recommended approach can construct samples of usage/expense distributions of individuals and families for each of various age/family size/health status/and other characteristic combinations.  It can take into account the probability of each of various levels of total expenses and each of various breakdowns of these total expenses among different types of providers and services. Then, using these expense amounts and probabilities, it can calculate for each health plan’s benefit structure a best estimate of likely out-of-pocket costs for a user of given age, family size, and other characteristics.  Plans can then easily be compared based on a single dollar-figure best estimate of average total cost.

Why the Insurance Value Needs to Be Determined

Some might wonder why it will be important to have a plan comparison tool determine the insurance value of each plan.  Since the ACA requires that there be “metal” levels of plans—bronze, silver, gold, and platinum (and catastrophic)—and that each plan within a level have the same actuarial value, isn’t premium all that matters?

In fact, different plans with the same actuarial value might have very different value to a consumer with specific characteristics.  Under the ACA, the actuarial value may be determined based on the percent of expenses the plan would pay and not pay for a broad population representative of the total population, including 25 year-olds, 40 year-olds, and 55 year-olds with different family sizes and other characteristics.  Between two plans that have the same value for such a broad group, one might offer much better protection, and therefore better insurance value, for a subgroup—for example, persons 55 years old with relatively poor health status, or persons with unusually high usage of expensive name brand drugs.  Moreover, there will be substantial differences in relative insurance value to a user who, quite reasonably, wants to compare plans across levels—a bronze plan to a gold plan, for example.

To illustrate, CHECKBOOK/CSS’s analysis of plans in the Massachusetts Connector reveals that, for a given consumer, it is not uncommon for a Silver-level plan that has a premium $1,000 higher than the premium in a Bronze-level plan actually to have a total expected cost (premium plus out-of-pocket cost) $1,000 lower than the Bronze plan.

The Range of Risk

For each plan, the estimated average likely cost (premium plus out-of-pocket) for each user’s combination of age, family size, and other characteristics will be the correct focus for many or most users.  But users might reasonably want to know how plans would compare in a really bad year or a really good year—and how likely such years are to occur.  A comparison tool should enable users to see how plans would compare for a population with the user’s age, family size, and other characteristics that ends up being in, say, the top 10 percent for expenses. This analysis is done based on the same individual expense level data (from MEPS and/or other sources) used for the analysis of average likely cost.

An important part of this type of comparison is to show the user’s maximum possible cost for each plan.  Showing that figure is less straightforward than it might seem.  Even in systems (for example, the Massachusetts Connector) where basic benefit descriptions state a single out-of-pocket limit to the policyholder, there are often fine-print exceptions—for example, where the limit is different for drugs than for other types of expenses. (It is to be hoped that, when the Summary of Benefits and Coverage regulations under the ACA are finalized, there will be no significant loopholes or ambiguities in the benefit descriptions. )

Letting Known Future Expenses, or Likely Expenses, Be Reflected in the Comparisons

While it would be unwise to have a comparison tool built solely on calculating out-of-pocket costs based on known or planned specific usage, such as regular physician visits and known prescriptions, it does make sense to include in the out-of-pocket cost calculations estimates of expenses for large medical events that can be reasonably well anticipated.  A good comparison tool will do this, using expense assumptions that no ordinary consumer is likely to be able to make.

An example would be something like a pregnancy, which is planned, or a condition diagnosed before plan-choice open season if that condition is known to require large future expenses.  To enable users to reflect expenses for such conditions in the plan comparisons, it is desirable for a tool to give the user an overall estimate of out-of-pocket costs for such a condition, using data on likely numbers of visits, drug utilization, etc. that no consumer is likely to know in advance. Then the tool can add these predicted expenses to the expense estimates already included in the insurance-value model for users of the same age, family size, etc.

Plan Quality

A best-practices plan comparison tool should include extensive information on plan quality.  It should include an overall quality rating, for quick reference alongside overall cost comparison information, based on a formula that takes into account various dimensions of quality.

To the extent feasible, the tool should give the user the ability to drill down for, and sort on, information on various aspects of quality that are of most interest to the user—measures related to a specific health care circumstance such as having young children or having diabetes, for example, or measures of specific aspects of care or service quality like quick access to doctors or trouble-free claims handling.

Subject to the need to avoid making use of the tool too burdensome and time-consuming for users, the user can be given the opportunity to give weights to the different quality dimensions as a basis for the tool’s calculation of a user-specific overall quality rating. Recent versions of CHECKBOOK/CSS’s Guide for Federal employees have offered users this personalization capability through the use of sliders that let the user assign weights to measures of different dimensions of quality; often, some users give zero weight to dimensions that other users weigh most heavily.

There are compelling reasons to strive to enhance user interest in quality measures. An Exchange that provides a marketplace of consumers who are well informed on such measures may be an important force for overall improvement in the quality and efficiency of the health care system—driving plans and providers to redesign practices in ways that produce better outcomes, safer care, and reduced costs. The availability of information on quality of plans will depend on the commitment, resourcefulness, and legal leverage an Exchange and collaborating government and non-governmental entities have over plans, as well as their ability to get information from other sources.

Some quality measures are required to be developed under the ACA.  These include measures from surveys of plan members and information on plan accreditation status.  They also include information from reports plans must submit on coverage benefits and health care provider reimbursement structures that improve health outcomes through implementation of quality reporting, effective case management, care coordination, chronic disease management, and medication and care compliance initiatives.  It is very important that the Federal regulations be written and enforced to produce comparable, meaningful, verified information across all plans.  For example, words like “accredited” should not have highly variable meanings, as is now the case in some existing plan comparison tools.

The eValue8 program of the National Business Coalition of Health (NBCH) is a good example of what might be possible in providing important information on key dimensions of plan quality. NBCH works with purchasers in specific markets to get plans to answer, and document responses to, a wide range of questions about programs the plans have and results the plans achieve related to various quality dimensions, such as:

  • The quality and visibility of tools and other resources to help members make treatment decisions,
  • Personal health record availability to members,
  • Disease management programs, and
  • Plan effectiveness in helping coordinate care for patients with multiple chronic conditions.

An important attribute of many of the plan performance elements eValue8 examines is that consumers will easily understand that plans can differ on these dimensions and that these are dimensions that are in the direct control of the plans.  In contrast, CHECKBOOK/CSS has found that many consumers believe that the performance they experience on some quality measures, such as measures of whether members get certain recommended tests and treatments, is determined by choice of doctor rather than choice of plan.

Exchange-Wide Provider Directory

After cost, the information of second greatest interest to consumers choosing among health plans is whether plans have the consumers’ desired doctors as participating providers.  CHECKBOOK/CSS has found that the all-plan provider directory it has provided for users of some recent versions of its Guide for Federal employees is highly valued. The user simply types in the names of doctors he or she wants to use and the tool’s summary comparison of cost and quality of available plans shows which plans have the desired doctors in their networks.

In contrast, the way plan comparison tools generally answer this question is by referring users to each plan’s online provider directory, where the user can look up doctors one plan at a time.  That is a cumbersome solution.

To create an Exchange-wide provider directory as inexpensively and accurately as possible, it will be most efficient for each Exchange to have and exercise the authority to require plans to provide the entity implementing the plan comparison tool with electronic files listing all providers whenever they update their provider directories.  Once such a reporting system is in place, doing such regular reporting should not be burdensome on plans.

Provider Quality Information

A plan comparison tool for Exchanges can be enhanced by providing information on the quality and availability of participating providers. Many users coming to an Exchange may need to choose providers; this is an opportunity to help consumers choose high-quality, efficient providers, in the process motivating and guiding providers to improve.

The range of available measurement results at the physician or practice site level is still quite limited. But there are some measures available. For example, in some versions of its Guide for Federal employees, CHECKBOOK/CSS has provided a list of physicians recognized for quality, including physicians who practice in NCQA-recognized Patient Centered Medical Homes and Bridges to Excellence-recognized practices.

If CMS’s PhysicianCompare website becomes, as it should, the compiler of extensive physician quality information, plan comparison tools should include information from PhysicianCompare for every doctor affiliated with every plan.