Objectives. To review the availability of data sources to study health plan competition in the United States.

Data Sources. The literature on health plan competition was reviewed. Possible data sources to study health plan competition were evaluated. Experts in the field of health plan

Principal Findings. There is much more quantitative data available on HMO plans than on other types of health plans that are growing in popularity, such as PPOs. A key source for health plan data, state health insurance filings, lacks information on beneficiaries in non-HMO plans. Data on health plan quality is growing. In addition, case studies of particular markets is providing useful qualitative information on the dynamics of the health plan industry.

Conclusions. The fragmentation of the health care market and the hesitancy of governments and private organizations to provide detailed information across markets and providers creates serious obstacles to the study of health plan competition.

Key Words. Health plan competition, economic competition, antitrust, insurance

More than most countries in the world, the United States relies on markets to allocate health care services. As such, a key aspect of U. S. health care policy is the promotion of market efficiency. Concerns about health care cost inflation have served to place increasing emphasis on maintaining competitive markets in health care. While early on these markets were viewed as largely exempt from antitrust policies that attempt to ensure fair market competition, this is no longer the case. For example, physicians have been subject to antitrust regulation since the landmark case in 1975 in which the Supreme Court decided in Goldfarb v. Virginia State Bar that "learned professions" were not exempt from antitrust law. Similarly, hospitals and health plans also increasingly are being scrutinized for antitrust violations.

Despite the courts' increasing willingness to treat health care markets as they would any other market, health care markets are undeniably different from other markets in a number of important ways. Regulation is extensive in health care. For example, prices for many health care products are dictated by government insurers, such as Medicare and Medicaid. Health care products are complex and heterogeneous, information about health care is only imperfectly understood by consumers, and due to consumers' lack of information, providers typically serve as patients' agents. Further, health care markets often consist of private not-for-profit and public firms, in addition to for-profit firms, all competing in the same market. Finally, the widespread presence of insurance creates pervasive moral hazard and adverse selection problems.

The complexities inherent in health care markets complicate the application of antitrust laws and traditional economic theory of market competition. Questions that might be answered with traditional economic tools, such as whether a merger between two firms is anticompetitive, become much more difficult when the firms are health care providers. The lack of information on the functioning of health care markets also hinders the development of health policies in other areas. For example, state legislatures around the country are currently debating a variety of regulatory proposals pertaining to the structure and functioning of managed care plans, such as any willing provider laws, gag clauses, and appeals processes.

In the absence of generalizable theories, empirical research becomes critical to the development and application of effective health care policies to promote market efficiency. Although there has been a significant amount of research in health economics on market competition, a recent survey finds that in many respects this area of research is still in its infancy (Gaynor and Vogt 2000).

A prerequisite for market competition research is data that can be used to empirically examine critical questions, such as how the structure of health care markets affects the price, quantity, and quality of health care services. Concerns have been raised, however, about the data that currently exist to conduct this research. Because the health care system in the United States is fragmented and complex--with hundreds of health plans and hundreds of thousands of providers in varying geographic settings and organizational arrangements offering heterogeneous products--data for research on health care markets is difficult to assemble, incomplete, or often simply unavailable.

This article outlines how existing data sources can be exploited to answer questions about health insurance markets, highlights gaps in existing data resources that are likely to hinder needed empirical research, and suggests efforts that might be undertaken to develop better data resources. The article is focused on health plans. While competition is an important policy issue to other health care markets, such as hospitals, physicians, nursing homes, and the pharmaceutical market, in the interest of space we do not address these markets here. The article begins by presenting a selected review of health plan competition studies that highlights omissions from the existing research base. We then review the key data sources that may be used to study health plan competition focusing on their strengths and weaknesses. The third section of the article summarizes how existing data sources may or may not fill in the gaps in the research base.

SELECTED REVIEW OF STUDIES OF HEALTH PLAN COMPETITION

Research on health plan market competition is abundant in some areas--such as the impact of health maintenance organizations' (HMOs) market penetration rates on various outcomes, and sparse in other areas--such as the effect of non-HMO plan competition.

Numerous studies have examined the effect of HMO penetration on various health system outcomes, including health insurance premium growth rates (Feldstein and Wickizer 1995), employment of hospital nurses (Spetz 1999), physician earnings (Simon, Dranove, and White 1998; Hadley and Mitchell 1999), breast cancer screening and survival rates (Decker and Hempstead 1999), hospital cost inflation (Robinson 1991; Gaskin and Hadley 1997), adoption of cost-containment measures by fee-for-service (FFS) plans (Joesch, Wickizer, and Feldstein 1998), adoption of new technologies (Baker and Wheeler 1998), and health expenditures for Medicare FFS patients (Baker 1999) (Table 1). For example, to analyze the effect of HMO market penetration on premium growth rates, Feldstein and Wickizer (1995) used premium information from a selected sample of 95 insured groups over the period 1985-92. Data on HMO enrollment penetration came from InterStudy and from specific HMOs.

The vast majority of studies that examine the effect of HMO penetration use data from InterStudy. InterStudy data include HMO enrollment by market area, where market area is typically defined as a metropolitan statistical area (MSA) or county. A few studies use data from California discharge abstracts that list whether the patient was enrolled in an HMO. Studies of physicians have measured HMO penetration using survey data on physician contracts with managed care plans (e.g., Simon, Dranove, and White 1998; Hadley and Mitchell 1999).

Although most studies use HMO penetration to characterize the health insurance market, a few studies have used the number of HMOs in a market as the measure of market structure (Wholey, Feldman, and Christianson 1995; Wholey et al. 1997). Wholey, Feldman, and Christianson (1995) examined the relationship between the number of competing HMOs in a market and HMO premiums. HMO financial and utilization data came from annual statements filed by HMOs with state regulators. The dependent variable was premiums per member per month, which was measured by dividing annual premium revenue by member months of coverage. Data on the number of HMOs in a market (defined as a county) came from the HMO census conducted by InterStudy and the Group Health Association of America (GHAA) Directory of HMOs. Feldman, Wholey, and Christianson (1996) also extended their model to examine the effect of HMO mergers on premiums. The same data sources were used. Information on whether a merger occurred was obtained from InterStudy.

In a 1999 study, Feldman, Wholey, and Christianson looked at the effect of national HMO mergers on the structure of local markets. Market structure was measured using the Herfindahl index, which was calculated before and after the merger using plan enrollment data from InterStudy. Markets were defined as MSAs.

Another, more indirect, indicator of market structure used in health plan studies is the elasticity of demand for health insurance. Competitive markets require that firm-level demand curves are elastic with respect to price. A variety of studies have examined this issue by studying health plan choice among employees of a large employer that offers a variety of health plans (McGuire 1981; Welch 1986; Long, Settle, and Wrightson 1988; Feldman et al. 1989; Short and Taylor 1989; Barringer and Mitchell 1994; Dowd and Feldman 1994-95; Marquis and Long 1995; Buchmueller and Feldstein 1997; Royalty and Solomon 1999). [See Royalty and Solomon (1999) and Scanlon, Chernew, and Lave (1997) for reviews of this literature.] Although the demand elasticities calculated in these studies varied, all of the studies showed that consumers were sensitive to the price of health insurance and can and do change plans in response to price changes. Few of these studies, however, control for other characteristics of the plans, in parti cular, measures of health plan quality.

In addition to quantitative economic analyses, surveys and administrative reports also have a salient role in understanding the implications of health plan market structure. Grossman (2000) recently reported on the results of a survey by the Community Tracking Study in which interviews were conducted with health care industry informants representing health plans, providers, and purchasers. The study examined how local plans were responding to national plans and the strategies that plans were using to increase market share and market power. Baumgarten regularly publishes data describing the managed care market in seven states around the country (e.g., Baumgarten 1998). The reports describe the market shares of the top managed care plans and how they have changed over time, managed care enrollment, and the financial status of HMOs. Much of the data come from annual filings by health plans with state insurance departments.

While these studies provide essential information about health care markets, some clear gaps remain. First, key to any analysis of market structure is the definition of the relevant product market. The relevant product market should include all firms that provide the same product or close substitutes for that product, and all potential competitors or firms that could provide that product or a close substitute with relative ease (Haas-Wilson and Gaynor 1998). Previous studies of HMO competition and its effect on premiums have assumed that HMOs constitute a separate product market (e.g., Wholey, Feldman, and Christianson 1995). In contrast, one of the deciding factors in the widely publicized antitrust case Blue Cross and Blue Shield United of Wisconsin v. Marshfield Clinic (1996) was Judge Posner's conclusion that there is not a separate market for HMO services that is distinct from other types of non-HMO insurance products. Thus, there is a need for research into the nature of the product market for health in surance.

A second key element of market structure analysis is the definition of the relevant geographic market. Health plan studies to date have defined the geographic market as an MSA or county. Little justification is given for selecting this definition. One problem with conceptualizing health plan market areas is that health plans, as financial services, are not physically tied to geographic markets and seldom operate in only one geographic area. However, health providers, which form the structure of many health plans' products, are physically tied. In the hospital literature, the boundary of a hospital's market has been defined using patient flow data, often zip codes from hospital discharge abstracts, and the Elzinga-Hogarty technique. The Elzinga-Hogarty technique measures patient flow and defines a geographic area as large enough so that sales from sellers (e.g., hospitals) outside the area to buyers (e.g., patients) inside the area are small and sales from sellers inside the area to buyers outside the area are small (Gaynor and Vogt 2000). Given appropriate data such as that which would indicate the location of buyers (e.g., zip codes), one might apply techniques such as the Elzinga-Hogarty to health plans rather than relying on political boundaries such as counties.

The literature on health plan market structure is clearly dominated by InterStudy data, which are solely focused on HMOs. There is a critical need for data and research on other types of health plans such as indemnity plans, preferred provider organizations (PPOs), and Employee Retirement Income Security Act (ERISA) plans. Furthermore, the FFS indemnity plan, which has been the traditional reference for studies of effects of financing arrangements, no longer dominates the financial market of health insurance. Consideration needs to be given to the appropriate reference group: managed versus nonmanaged care; types of care management (e.g., utilization review, preauthorization for specific services, ongoing case management for select conditions); financial risk-sharing arrangements (e.g., full capitation, partial capitation); and financial incentives facing providers (maintenance of plan affiliation, number of beneficiary referrals to providers, profit sharing in the plan, none).

Another data element that would facilitate research on health plan competition across geographic markets would be uniform health plan identifiers. With such identifiers, the same plans operating across different markets could be identified. Public Law 104--191 mandates unique health plan identifiers as part of the Administrative Simplification provisions of the law. The final rule for standard electronic transactions between health plans and providers was announced August 17, 2000 by the Secretary of Health and Human Services (2000). Health plans must comply by October 16, 2003. The time line for implementing uniform health plan identifiers has not been set.

Another gap in the literature is the effect of health plan competition on health plan quality. Antitrust cases have relied on reported reputation as a proxy for quality (Sage 1997), but more empirical measures such as standardized health plan ratings [e.g., Health Plan Employer Data and Information Set (HEDIS) measures] clearly would be preferable. A final gap in the literature is research on health plan entry. Concentrated markets may provide short-term profits and elevated prices, but these should erode upon entry by competing firms. Little is known about the ease with which plans can enter markets to enhance competition.

DATA SOURCES FOR HEALTH PLANS

Filling in the gaps in the literature on health plan competition requires the use of several key data elements. These include measures of premiums, market structure, and health plan quality. In addition, to further understand the product and geographic boundaries of health plan markets, one needs information on health plan choice. In the subsequent review of existing data sets we focus on these key data elements. Table 2 summarizes the data sources discussed here and the types of the data elements that they supply.

State Insurance Department Annual Filings

State insurance departments are responsible for administering and enforcing insurance laws. All states require that insurers be licensed before providing insurance and that they regularly submit financial and operating information to state insurance departments. Several organizations sell the information collected from state insurance filings, including InterStudy, Inforum, HCIA Inc., A. M. Best, and the National Association of Insurance Commissioners (NAIC). However, only A. M. Best and NAIC sell both HMO and non-HMO data; the others sell HMO filings only. NAIC's non-HMO database is called the HMDI database (Hospital, Medical and Dental Services of Indemnity Corporations) and sells for $4,800 (LeCluyse, personal communication, 1999).

Because each health plan in a state is required to submit a report to the state insurance department, annual filings provide a census of most plans operating in the state. Plans that cross state lines and are headquartered in another state are still required to submit an annual filing in each state in which they operate. However, some HMOs are operated as lines of business of licensed insurance companies and in some states are required to file only portions of the annual statements, for example, only limited financial information (Baumgarten 1998).

Although state insurance filings will capture most insurance plans in the state, they still may leave out some types of plans. For example, an analysis by the American Accreditation Health Care Commission/Utilization Review Accreditation Commission, the main organization for accrediting PPOs, shows that only 12 states require registration, licensing, or both of PPOs--now the most common form of health coverage (Moskowitz 1999). Furthermore, third-party administrators that manage claims for ERISA plans are typically not required to report financial and operating information because they do not take on insurance risk. Finally, there is controversy in the industry as to whether providers that take on risk in the form of capitated payments, such as physician groups, should be licensed and regulated as insurance products.

Annual filings contain extensive financial information. Both HMOs and non-HMO plans report premium revenues by lines of business (e.g., Medicare, Medicaid, Federal Employee Health Benefit Plan). Typically, only HMO plans, however, are required to report enrollment data. Thus, only HMO annual filing data can be used to calculate premiums per beneficiary.

The lack of enrollment data from non-HMOs also means that non-HMO market share cannot be determined using the annual filings. Calculations for HMOs are also limited because only some states (e.g., Minnesota, Michigan, and Florida) require HMOs to submit enrollment by geographic unit, typically county.

Information on other plan characteristics, such as benefits offered, size and characteristics of the plan's network, and characteristics of enrollees, may also be available. However, these data vary by state and are limited. Some utilization data are captured for HMOs, such as total member ambulatory encounters and hospital patient days incurred. Finally, some states such as Michigan include HEDIS reports and enrollment by county (Baumgarten, personal communication, May 1, 1999).

InterStudy

InterStudy collects information on HMO or point-of-service (POS) plans through surveys and links the survey data to information from state insurance department filings. InterStudy's HMO data include information on premium revenue by product line (e.g., commercial, Medicare, Medicaid), expenditures, and net profits and losses. Enrollment by product line is also included, so that premiums per beneficiary can be calculated by product line.

The ability to use InterStudy data to calculate market shares has gotten better over time. Historically, InterStudy reported total enrollment in a state and the counties and MSA in which a plan operated. Researchers would allocate total enrollment to each county. Starting in 1993, InterStudy began reporting enrollment by MSA, and starting in 1997, they reported enrollment by county.

InterStudy also reports on other plan characteristics such as the plan's headquarters age and tax status; some information about the size of their network (number of primary care contracts, number of specialty contracts, number of hospital contracts); and types of provider reimbursement mechanisms (e.g., capitation, fee schedules, salary). Measures of health plan quality are not provided.

Economic Census

The Economic Census profiles the U. S. economy every five years, from the national to the local level, as mandated by law under Title 13 of the United States Code. The most recent Economic Census is for 1997. The law requires firms to respond and specifies penalties for firms that fail to report. Large- and medium-size firms, plus all firms known to operate more than one establishment, are sent questionnaires. Data from existing administrative records of other federal agencies are used for small firms. Insurance is one of the industries captured. For insurance companies, the 1997 questionnaire asked about net insurance premium revenues by type of insurance (e.g., health and medical insurance, life insurance) for the establishment and total net insurance premium revenues by state. The questionnaire asked about the county in which the establishment was physically located. No information is collected on total enrollment or enrollment by geographic unit.

Medicare Health Plan

While the relationship of market structure to price is not relevant to the Medicare program because its premiums are regulated, the relationship between Medicare health plan market structure and quality is an important issue. Files on Health Care Financing Administration's (HCFA's) web site indicate the name of each plan with which Medicare has a contract, the state and counties in which it operates, whether it is for-profit or not-for-profit, the types of plans it offers [e.g., independent practice association (IPA), HMO], enrollment within each county by plan, and market penetration by plan (http://www.hcfa.gov/stats/mmcc.htm).

HCFA is mandated to provide comparative plan information, based partly on consumer surveys, to Medicare beneficiaries. Effective January 1, 1997, HCFA began collecting HEDIS measures on health care quality and services provided in calendar year 1996 from all Medicare managed care plans, which include all Section 1876 risk and cost health plans, all Social Health Maintenance Organizations (S/HMOs), and all Medicare Choice demonstrations. These measures could be used to study how health plan competition affects the quality of health plans offered in the Medicare program. HCFA's Medicare Compare database compares quality measures of health plans by geographic region (http://www.medicare.gov/mphCompare). Quality measures include beneficiary satisfaction, use of preventive services, board certification rates, and length of time physicians stayed in the plan. One drawback with the HCFA Medicare data on health plans is that it does not appear to be combined into one electronic central file that would be readily usab le by researchers.

The Medical Expenditures Panel Survey

The Medical Expenditure Panel Survey (MEPS) is a nationally representative survey of health care use, expenditures, sources of payment, and insurance coverage for the U. S. civilian noninstitutionalized population, as well as a national survey of nursing homes and their residents. MEPS is cosponsored by the Agency for Healthcare Research and Quality (AHRQ) and the National Center for Health Statistics (NCHS).

MEPS comprises four component surveys: the Household Component (HC), the Medical Provider Component (MPC), the Insurance Component (IC), and the Nursing Home Component (NHC). The HC serves as the core survey from which the MPC sample and part of the IC sample are based. Data collected from the household respondents include demographic characteristics, health conditions, health status, use of medical care services, charges and payments, access to care, satisfaction with care, health insurance coverage, income, and employment. The MEPS IC collects data on health insurance plans from employers, unions, or other sources.

MEPSIC consists of two subcomponents: the household sample and the list sample. The household sample collects detailed information on the health insurance held by and offered to respondents to the MEPS HC. The number of employers and union officials interviewed varies from year to year. The second subcomponent interviewed managers at more than 20,000 business establishments and governments to obtain national and regional estimates of the availability of health insurance at the workplace. Health insurance plan questions are asked for each offered plan, up to a maximum of four plans. If an establishment indicates during the telephone prescreening interview that they offer more than four plans, interviewers ask the names of all the plans offered and send them questionnaires for the plans with the largest enrollment, making sure they get at least one plan of each type offered. The survey questions capture information on health plan premiums, enrollment, and benefits.

NCQA's Quality Compass

The National Committee for Quality Assurance (NCQA) is a private, not-for-profit organization that accredits managed care plans and assesses and reports on the quality of care received under these plans. NCQA also provides consumers information about managed care plans' performance. NCQA's 1998 Quality Compass contains HMO performance statistics based on encounter data analyses, member satisfaction, and accreditation information from 447 managed care plans throughout the United States.

One drawback to NCQA's Quality Compass is that participation is voluntary. As the health plan market becomes more competitive, health plans may be less willing to participate in or disclose the results of surveys and performance monitoring efforts. For example, of 447 plans that submitted data to be used in calculating national and regional performance averages in NCQA's 1998 Quality Report, 155 plans asked that their data not be reported publicly (Moretz 1998). On the other hand, pressure from employers and other health plan sponsors may encourage greater participation in and release of NCQA health plan quality measurement efforts.

NCQA also developed HEDIS, a set of standardized performance measures that assess dimensions of health care and services provided by managed care plans. The dimensions include access to services (e.g., immunization and screening rates for health plan members), provision of preventive services (e.g., enrollment in smoking cessation programs), and processes relating to care quality (e.g, use of beta-blockers, cesarean section rates, follow-up after hospitalization for mental illness). HEDIS measures are included in NCQA's Quality Compass report and are also collected as part of the Medicare program. HEDIS data elements rarely measure directly the outcomes of treatment for specific health problems of patients. Other NCQA measures listed in Table 2, for example practitioner turnover, may be valuable for assessing the continuity of care provided under health plans.

CAHPS[R] and Other Beneficiary Surveys

Surveys of health plan beneficiaries are increasingly being used to develop measures of health plan quality. One such effort is the Consumer Assessment of Health Plans Study (CAHPS[R]), which was developed by AHRQ. The overall goal of CAHPS [R] is to provide an integrated set of tested and standardized survey questionnaires and accompanying report formats that can be used to collect and report meaningful and reliable information from health plan enrollees about their experiences. The CAHPS [R] materials are designed for use with all types of health insurance enrollees (Medicaid, Medicare, and private insurance beneficiaries) and across a range of health care delivery systems, from FFS to managed care plans. In addition to a core set of items designed for use with all respondents, additional questions are targeted for use with certain subgroups, such as persons with chronic conditions or disabilities, Medicaid and Medicare beneficiaries, and families with children. CAHPS [R] is being demonstrated and evaluated in a variety of settings that include Medicaid programs, large employers, and health plan purchasing coalitions.

Other beneficiary survey instruments have also been created. These include:

* MEDSTAT's Quality Catalyst, a health plan enrollee survey designed by The MEDSTAT Group, J. D. Power and Associates Inc., and The New England Medical Center;

* the Minnesota State Employees Survey, a joint effort of labor and management in the state;

* the Consumer Experience Surveys, conducted by the California Public Employees' Retirement System (CALPERS), covering all plans that enroll state employees and retirees; and

* the Survey of Federal Employees Health Benefit Plan, conducted for the Federal Office of Personnel Management.

The main drawback of using CAHPS [R] or other survey instruments as a quality measure is that they typically do not apply to all health plans in a market. Furthermore, they would probably not apply to all enrollees in a health plan but rather only those of one or more employers. These are useful data sets, however, for studying how aspects of health plan quality influence consumer health plan choice.

Complaints Filed with State Departments of Insurance

Information on complaints filed about HMOs have recently become more accessible and might be used as another type of quality indicator. For example, the State of New York Insurance Department now posts complaint ratios (i.e., complaints per $1 million in premiums) on its web site for all HMOs operating in the state. Filed complaints, however, are relatively rare events. In addition, it is not clear how easily a consumer can file a complaint and therefore how accurately these complaints reflect quality. Finally, all complaints are not alike. Some complaints may be more substantive than others, yet there is no measure for weighting within state systems the nature of the complaint.

USING EXISTING DATA TO FILL IN KNOWLEDGE GAPS

In this section we turn back to our analysis of the gaps in the existing literature on health plan competition and summarize the extent to which existing data can and cannot be used to fill in some of the missing pieces in that literature.

Product Market Definition

The appropriate definition of insurance markets could be further elucidated from studies of cross-price elasticities between HMOs and other types of health plans. Some studies have been conducted to estimate the price elasticity of demand for health insurance using data on employee choice of and switching among plans offered by large employers (e.g., Long, Settle, and Wrightson 1988; Feldman et al. 1989; Short and Taylor 1989; Barringer and Mitchell 1994; Dowd and Feldman 1994-95; Buchmueller and Feldstein 1997). For example, Feldman et al. (1989) found that IPAs and FFS plans are close substitutes and that HMO group, staff, and network models are close substitutes, but that there is less substitution between these two classes of plans. Data sets used in these studies include the National Medical Care Expenditure Survey (Short and Taylor 1989), enrollment data from private employers (Feldman et al. 1989), and health plan data (Long, Settle, and Wrightson 1988). To our knowledge, studies have not specifically examined the cross-price elasticity for insurance plans among employers who sponsor plans. Because employers select health plans for employees and most employers offered only one plan, this is a critical issue (Long and Marquis 1998).

Data from MEPS might be used to further elucidate the appropriate definition and parameters of insurance product markets. As described, MEPS has information from which one could model individual health plan choice. In addition, if one knew employers' health plan choice sets as defined, for example, in the annual state insurance filings or the Economic Census, one could use MEPS to model employers' choice of health plans and determine the degree to which HMO and non-HMO plans are substitutes.

Geographic Market Definition

Defining the boundaries of health plans' geographic market areas is a complex undertaking. The relevant market area will depend on the location of numerous physicians, hospitals, and other health plan providers with whom the plan has contracted. Detailed information on a health plan's provider network (including zip code locations of providers) could be used to crudely characterize a plan's market if it were in an electronic format. Another approach might be to determine the location of enrolled beneficiaries. This would require data on beneficiary zip codes, information that is not readily available to researchers. None of the data sets reviewed are ideal for this undertaking. Antitrust agencies may have access to data that could be used to define geographic markets as part of legal investigations, but the ability of researchers to access this type of private, proprietary data would likely be limited unless health plans had a particular interest in the study results.

Non-HMO Health Plan Competition

Existing data on non-HMO plans is weak. Premium information is collected in state health insurance filings, but premiums per beneficiary cannot be easily determined, as noted above. Similarly, because the state insurance filings lack information on beneficiaries in non-HMO plans and on the geographic markets in which they operate, even crude measures of market share cannot be calculated. Finally, the correlation between HMO market share and other forms of managed care, such as PPOs, is weak; therefore, HMO data cannot be used as a proxy for managed care penetration in general (Morrisey and Jensen 1997).

To conduct research on the relationship between premiums and market structure similar to that which has been done on HMOs, one would need to conduct a special survey of non-HMO plans to, at a minimum, determine the number of beneficiaries per line of business and the geographic markets in which the plans operate. Ideally, one would also like to know beneficiaries by geographic market, such as county. By combining this information with data on HMO plans, a relatively complete view of plans operating in a geographic market might be seen.

Quality and Competition

Health plan quality measurement has made significant advances in recent years. Several of the databases that have recently become available may prove useful for studying the relationship between health plan competition and health plan quality, as well as for studying the effect of health plan quality on health plan choice. In general, the field of health plan quality measurement is still in its infancy, and we anticipate that the amount and specificity of information about health plan quality is likely to grow over the next several years.

Market Entry

InterStudy data can form the backbone for studies of health plan market entry into the HMO market, particularly data from recent years that include counts of enrollees by county. Data on other types of health plans is more limited and could currently (without additional development, noted above) only be used if a market was defined as a state.

CONCLUSIONS

In his review, Sage (1997) highlights how the Marshfield clinic antitrust case could have greatly benefited from empirical evidence on health plan competition. In this article, we attempt to make some inroads into identifying data sources that could be developed or further exploited to address this research need.

Data collected by regulatory agencies represent one source of information on health plans that has been used in market competition studies. These data might be further enhanced by adding salient data items to make it more useful for research purposes. For example, health plan annual filings with state departments of insurance contain information on premiums and financial condition. Filings for some states and types of plans (e.g., HMOs) also include information on market area, enrollment, and other operating characteristics. A work group with representation from state insurance departments and health plans could evaluate whether this additional information, such as on enrollment, could be more uniformly reported across states.

We have also tried to identify ways in which existing data could be used to answer pressing policy questions about competition. For example, new quality measures, such as those developed by AHRQ and NCQA, could be used to fill the hole in information about market structures' effect on quality.

In general, we have grounded this review on the notion that major changes in the content or amount of existing data resources would be prohibitively expensive. We have tried to err on the side of providing practical and relatively cost-effective methods for improving data resources. The review highlights two main obstacles identified to developing comprehensive data sources for studying health care competition. One is the fragmentation and complexity of the U. S. health care market; the other is the hesitation of governments and private organizations to provide detailed linked information across markets and providers due to privacy concerns and concerns about divulging trade secrets.

The authors thank Allan Baumgarten, Hal Luft, Marty Gaynor, anonymous reviewers, and the many participants of the "Data Needs for Studies of Competition in Market Areas" conference sponsored by The Agency for Health Care Policy and Research and The Robert Wood Johnson Foundation and organized by The Alpha Center.

Address correspondence to Tami L. Mark, Ph.D., The MEDSTAT Group, 4301 Connecticut Avenue, N.W., Suite 330, Washington, DC 20008. Rosanna M. Coffey, Ph.D. is from The MEDSTAT Group, Washington, DC. This article, submitted to Health Services Research on January 19, 2000, was revised and accepted for publication on November 13, 2000.

REFERENCES

Baker, L. C. 1999. "Association of Managed Care Market Share and Health Expenditures for Fee-for-Service Medicare Patients." Journal of the American Medical Association 281 (5): 432-37.

Baker, L. C., and S. K. Wheeler. 1998. "Managed Care and Technology Diffusion: The Case of MRI." Health Affairs 17 (5): 195-207.

Barringer, M. W., and O. S. Mitchell. 1994. "Workers' Preferences Among Company Provided Health Insurance Plans." Industrial and Labor Relations Review 48 (1): 141-52.

Baumgarten, A. 1998. Illinois Managed Care Review 1998.

Blue Cross and Blue Shield United of Wisconsin v. Marshfield Clinic, 65 F.3d 1406 (7th Cir. 1995), cert.denied, 116 S. Ct. 1288 1996.

Buchmueller, T. C., and P. J. Feldstein. 1997. "The Effect of Price on Switching Among Health Plans." Journal of Health Economics 16 (2): 231-48.

Chrisianson, J. B., R. D. Feldman, and D. R. Wholey. 1997. "HMO Mergers: Estimating Impact on Premiums and Costs. Health Affairs 16 (6): 133-41.

Decker, S. L., and K. Hempstead. 1999. "HMO Penetration and Quality of Care: The Case of Breast Cancer." Journal of Health Care Finance 26 (1): 18-32.

Dowd, B., and R. Feldman. 1994-95. "Premium Elasticities of Health Plan Choice." Inquiry 31 (4): 438-44.

Feldman, R., M. Finch, B. Dowd, and S. Cassou. 1989. "The Demand for Employment-Based Health Insurance Plans." Journal of Human Resources 24 (4): 115-42.

Feldman, R., D. Wholey, and J. Christianson. 1996. "Effect of Mergers on Health Maintenance Organization Premiums." Health Care Financing Review 1996 17 (3): 171-89.

_____. 1999. "HMO Consolidations: How National Mergers Affect Local Markets. Health Affairs 18 (4): 96-104.

Feldstein, P. J., and T. M. Wickizer. 1995. "Analysis of Private Health Insurance Premium Growth Rates: 1985-1992." Medical Care 33 (10): 1035-50.

Gaskin, D. J., and J. Hadley. 1997. "The Impact of HMO Penetration on the Rate of Hospital Cost Inflation, 1985-1993." Inquiry 34 (3): 205-16.

Gaynor, M., and W. B. Vogt. 2000. "Antitrust and Competition in Health Care Markets." In Handbook of Health Economics, edited by A. J. Culyer and J. P. Newhouse. Amsterdam and New York: Elsevier.

Grossman, J. M. 2000. "Health Plan Competition in Local Markets." Health Services Research 35 (I, Part I): 17-35.

Haas-Wilson, D., and M. Gaynor. 1998. "Increasing Consolidation in Healthcare Markets: What Are the Antitrust Policy Implications?" Health Services Research 33 (5): 1403-20.

Hadley, J., and J. M. Mitchell. 1999. "HMO Penetration and Physicians' Earnings." Medical Care 37 (11): 1116-27.

Joesch, J. M., T. M. Wickizer, and P. J. Feldstein. 1998. "Does Competition by Health Maintenance Organizations Affect the Adoption of Cost-Containment Measures by Fee-for-Service Plans?" American Journal of Managed Care 4 (6): 832-38.

Long, S. H., and M. S. Marquis. 1998. "Results from the Community Tracking Survey." Health System Change Data Bulletin, 12 (Summer): 1.

Long, S. H., R. F. Settle, and C. W. Wrightson. 1988. "Employee Premiums, Availability of Alternative Plans, and HMO Disenrollment." Medical Care 26 (10): 927-38.

LeCluyse, L. K. 1999. Database Product Sales. National Association of Insurance Commissioners, personal communication, May 5.

Marquis, M. S., and S. H. Long. 1995. "Worker Demand for Health Insurance in the Nongroup Market." Journal of Health Economics 14: 47-63.

McGuire, T. G. 1981. "Price and Membership in a Prepaid Group Medical Practice." Medical Care 19 (2): 172-83.

Moretz. 1998. www.ncqa.org. Downloaded May 10, 1999.

Morrisey, M. A., and G. A. Jensen. 1997. "Switching to Managed Care in the Small Employer Market." Inquiry 34: 237-48.

Moskowitz, D. B. 1999. "Who's Minding the PPOs?" Business & Health 17 (5): 47.

Robinson, J. C. 1991. "HMO Market Penetration and Hospital Cost Inflation in California." Journal of the American Medical Association 266 (19): 2719-23.

Royalty, A. B., and N. Solomon. 1999. "Health Plan Choice. Price Elasticities in a Managed Competition Setting." Journal of Human Resources 32 (1): 1-42.

Sage, W. M. 1997. "Judge Posner's RFP: Antitrust Law and Managed Care." Health Affairs 16 (6): 44-61.

Scanlon, D. P., M. Chernew, and J. R. Lave. 1997. "Consumer Health Plan Choice: Current Knowledge and Future Directions." Annual Review of Public Health 18: 507-28.

Secretary of Health and Human Services. 2000. "Health Insurance Reform: Standards for Electronic Transactions." Federal Register 65 (160): 503 12-372. http://aspe.hhs.gov/admnsimp/final/txfin00.htm (downloaded August 17, 2000).

Short, P. F., and A. K. Taylor. 1989. "Premiums, Benefits, and Employee Choice of Health Insurance Options." Journal of Health Economics 8 (3): 293-311.

Simon, C. J., D. Dranove, and W. D. White. 1998. "The Effect of Managed Care on the Incomes of Primary Care and Specialty Physicians." Health Services Research 33 (3, Part I): 549-69.

Spetz, J. 1999. "The Effects of Managed Care and Prospective Payment on the Demand for Hospital Nurses: Evidence from California." Health Services Research 34 (5, Part I): 993-1010.

Welch, W. P. 1986. "The Elasticity of Demand for Health Maintenance Organizations." Journal of Human Resources 23 (3): 16-22.

Wholey, D., J. B. Christianson, J. Engberg, and C. Bryce. 1997. "HMO Market Structure and Performance: 1985-1995." Health Affairs 16 (6): 75-84.

Wholey, D., R. Feldman, and J. B. Christianson. 1995. "The Effect of Market Structure on HMO Premiums." Journal of Health Economics 14 (1): 81-105.

Wickizer, T. M., and P. J. Feldstein. 1995. "The Impact of HMO Competition on Private Health Insurance Premiums, 1985-1992." Inquiry 32 (3): 241-51.

                 Data Sources used in Prior Research of
                    the Effect of Health Plan Market
              Structure on Health Plan Premiums or Quality
                                                  Market Structure
Reference               Outcome Measure           Measure
Feldstein and Wickizer  Private indemnity health  HMO market
(1995); Wickizer and    insurance premiums        penetration
Feldstein (1995)
Wholey, Feldman, and    HMO premiums              Number of HMOs
Christianson (1995);                              in a market
Feldman, Wholey, and
Christianson (1996)
Christianson, Feldman,  HMO premiums              Herfindahl index
and Wholey (1997)                                 using HMO
                                                  enrollment
Reference               Unit of Analysis
Feldstein and Wickizer  Market (MSA
(1995); Wickizer and    or county)
Feldstein (1995)
Wholey, Feldman, and    HMO
Christianson (1995);
Feldman, Wholey, and
Christianson (1996)
Christianson, Feldman,  HMO
and Wholey (1997)
Reference               Data Sources
Feldstein and Wickizer  Premium data from 95 insured
(1995); Wickizer and    groups that had policies with
Feldstein (1995)        a major private, commercial
                        insurance company from 1985
                        through 1992
                        * HMO penetration from
                        InterStudy
                        Issues: Only includes HMOs. Does
                        not measure market share by each
                        HMO
Wholey, Feldman, and    Financial and utilization data
Christianson (1995);    in HMO annual statements
Feldman, Wholey, and    filed with state regulators
Christianson (1996)
                        Number of HMOs from
                        InterStudy and GHAA.
                        Issues: Only includes HMOs
Christianson, Feldman,  Herfindahl index created from
and Wholey (1997)       InterStudy HMO enrollment data
                      Data Sources for Health Plan
                   Competition Studies: Descriptions,
                       Strengths, and Weaknesses
Data Source          Frame, Geography, and Time Period
State insurance      Frame: All insurance
department filings   plans licensed in the state
(repackaged and      Geography: All states
sold by A. M. Best,  Period: Varies by state but
HCIA, Healthcare     generally back to the 1970s
Databank, Inc., and
others. Note: some
companies only sell
HMO filings)
Economic Census      Frame: Universe of insurance
                     plan establishments (an
                     establishment is at a
                     single physical location)
                     Geography: United States
                     Period: Profiles the economy
                     every five years; latest
                     information is for 1997
Data Source          Key Data Elements
State insurance      Price: Premiums usually collected
department filings   by lines of business, enrollment
(repackaged and      by lines of business for HMOs
sold by A. M. Best,  No. plans in a market: By state
HCIA, Healthcare     for HMOs and indemnity plans
Databank, Inc., and  Enrollment market share:
others. Note: some   Enrollment by lines of business
companies only sell  by state for HMOs, enrollment
HMO filings)         by country in some states, no
                     or little enrollment information
                     generally for non-HMO plans
                     Quality: HEDIS indicators
                     in some states for HMOs
                     Other information: Detailed
                     financial statements; member
                     ambulatory encounters and
                     hospital patient days by line of
                     business for HMOs
Economic Census      Price: Net insurance premium
                     revenues by type of insurance
                     (e.g., health and medical
                     insurance); total net insurance
                     premium revenues by state
                     No. plans in a market: County
                     where plan located
Data Source          Strengths                  Weaknesses
State insurance      Participation              Data not always
department filings   is mandatory               reported at geographic
(repackaged and      Wide geo-                  unit (e.g., counties)
sold by A. M. Best,  graphic coverage           Not all states report
HCIA, Healthcare     Quality of information     financial data by
Databank, Inc., and  submitted is likely to be  lines of business
others. Note: some   high                       HEDIS reports (e.g,
companies only sell                             on utilization and
HMO filings)                                    effectiveness) not
                                                typically included
                                                Excludes self-insured
                                                ERISA plans and PPOs
                                                Little enrollment data
                                                on non-HMO plans
                                                Utilization data
                                                increasingly
                                                problematic, with
                                                capitation models and
                                                23-hour stays
Economic Census      Census of all insurers     Not clear what
                     in the United States       geographic unit
                     Financial information      the data cover
                                                No information on
                                                enrollment
InterStudy HMO         Frame: All "full-service"
(directory, financial  HMO/POS plans, not
data, reports, and     behavioral carve-outs or
custom data files)     TPAs; based on state
                       department of insurance
                       filings (n = 652 in 1998)
                       Geography: United States
                       Period: 1986 to present
Medicare Prepaid       Frame: All HMOs with
Health Plans           a Risk Medicare contract;
Monthly Report         varies considerably as
                       plans move in and out of
                       Medicare risk contracting
                       Geography: All markets
                       in the United States
                       where plans operate
                       Period: 1985 to present
CAHPS [R] Surveys      Frame: Not defined; specific to
                       employers of third-party payers
                       Geography: Varies
                       Period: Implementation
                       varies by program/plan, but
                       started in 1997
InterStudy HMO         Price: Premiums
(directory, financial  Market Share: Enrollment by
data, reports, and     business line (e.g., Medicare, Medi-
custom data files)     caid); enrollment by MSA (starting in 1995);
                       enrollment by county (starting in 1997);
                       counties served
                       Other information: Tax status, age of plan,
                       HMO model type, number of contracts
                       by type of
                       provider
Medicare Prepaid       Prices: Premiums by plan
Health Plans           Market share: Enrollment by
Monthly Report         county
                       Other information:
                       Nonprofit/for-profit status,
                       summary of benefits offered
CAHPS [R] Surveys      Enrollment: Identifies
                       respondent's health plan
                       and length of coverage
                       Access: Waiting times, phone
                       access, time with doctor
                       Satisfaction: Ratings of doctor,
                       care, and plan; doctor-patient com-
                       munications; plan administration
                       Utilization: Doctor visits, ER visits
                       Health status: Respon-
                       dent's overall health status
                       Demographics: Age, gender,
                       education, race/ethnicity
InterStudy HMO         Includes most HMOs in
(directory, financial  the United States
data, reports, and     Survey can be
custom data files)     modified to capture
                       additional information
                       More detail on char-
                       acteristics of products
                       than available from state
                       insurance filings
Medicare Prepaid       Captures universe
Health Plans           of Medicare plans
Monthly Report         Free and eas-
                       ily accessible
                       Has information
                       on health plan quality
CAHPS [R] Surveys      Standard question-
                       naire format for
                       beneficiary surveys
                       Useable with different
                       population groups
InterStudy HMO         Excludes in-
(directory, financial  demnity plans
data, reports, and     Excludes self-insured
custom data files)     employer ERISA plans
Medicare Prepaid       Limited to
Health Plans           Medicare plans
Monthly Report         Information not
                       provided in one data file
CAHPS [R] Surveys      Use is voluntary-would
                       not be available for
                       all plans in a market
                       Tendency for plans to
                       customize survey and
                       dilute its uniformity
                       across plans is unknown.
Medical Expenditure  Frame: MEPS comprises four
Panel Survey         component surveys: HC, MPC,
                     IC, and NHC. Nationally
                     representative survey of the
                     U. S. civilian population.
                     Frame for HC drawn from
                     the National Health Interview
                     Survey. The MEPS IC collects
                     data on health insurance plans
                     obtained through employers,
                     unions, or other private
                     health insurance sources.
                     Geography: United States
                     Period: 1996 onward
NCQA's Quality       447 managed care organizations
Compass 1998         that volunteer to participate
Medical Expenditure  Premiums: Monthly premium
Panel Survey         equivalent for one typical
                     full-time equivalent with
                     single or family coverage
                     Enrollment: Number of
                     active employees enrolled
                     during typical pay period
                     Other information: Number
                     of plans offered employees,
                     employer and employee share
                     of premiums, plans benefits,
                     type of plan (e.g., gatekeeper),
                     characteristics of workers at the
                     company/establishment (e.g., wage
                     distribution, gender, age, union
                     members)
NCQA's Quality       Quality: Clinical HEDIS
Compass 1998         measures (e.g., adolescent
                     immunizations, advising smokers
                     to quit, beta-blocker treatment,
                     breast cancer screening, cervical
                     cancer screening, cesarean section
                     rates, childhood immunizations,
                     diabetic eye exams, follow-up after
                     hospitalization for mental illness,
                     prenatal care in the first trimester);
                     HEDIS member satisfaction
                     measures; NCQA Accreditation;
Medical Expenditure  Nationally rep-
Panel Survey         resentative
                     Useful for modeling
                     health plan choice
NCQA's Quality       Detailed quality
Compass 1998         information based
                     on audited data and
                     satisfaction surveys
                     Potential for longitudinal
                     data
Medical Expenditure  Does not capture a
Panel Survey         whole market area
NCQA's Quality       Participation is voluntary
Compass 1998         Not all plans in a
                     market participate
                     Participation varies
                     from year to year
                     Only captures
                     managed care plans
                     Not all plans are willing
                     to share data publicly
Complaints and        Complaints that con-
grievances filed      sumers/providers file against
with state insurance  insurance companies operating
departments by        in the state. Only released
consumers and         publicly in some states
providers
                      financial information (e.g., cost of
                      care trends, high-cost diagnosis-
                      related groups); utilization
                      information (e.g., outpatient drug
                      use, inpatient use in general
                      hospitals); health plan stability
                      (e.g., disenrollment, practitioner
                      turnover, indicators of financial
                      stability); health plan descriptive
                      indicators (e.g, board certification
                      status of providers, total enrollment,
                      enrollment by payer)
Complaints and        Quality: Varies by state. In New
grievances filed      York, report ratio of complaints
with state insurance  and grievances filed by 1 million
departments by        premiums, ratio of complaints
consumers and         and grievances upheld, type of
providers             complaint (e.g., prompt payment,
                      grievances regarding medical
                      necessity)
Complaints and        Free and accessible     Complaints are
grievances filed      in some states          infrequent events
with state insurance  (e.g., New York)        Complaints are
departments by        Available for all       an imperfect
consumers and         licensed health plans,  proxy for quality
providers             not only HMOs           Not clear whether
                      Available over time     provider and
                                              consumer complaints
                                              can be separated
                                              Detail in the type of
                                              complaint