This Article Abstract Reprint (PDF) Submit a response to this article Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (15) Citing Articles via Google Scholar Google Scholar Articles by Born, P. H. Articles by Simon, C. J. Search for Related Content PubMed PubMed Citation Articles by Born, P. H. Articles by Simon, C. J. Related Collections Managed Care Quality Of Care

Health Tracking

MARKETWATCH

Patients And Profits: The Relationship Between HMO Financial Performance And Quality Of Care

Abstract

 
This paper matches health plans’ financial performance with information on quality ratings as measured by 1997 Health Plan Employer Data and Information Set (HEDIS) 3.0 data. We address three policy questions: (1) Is the quality of care delivered by a plan influenced by the plan’s financial performance? (2) Do for-profit plans behave differently than nonprofits do? (3) What other factors are associated with variation in plan performance? We find, first, that more profitable plans achieve higher quality scores in subsequent years. Profits may enable a plan to pursue higher quality of care and invest in better management systems. Second, there is little systematic evidence that for-profit plans have different HEDIS scores than not-for-profits have.

New focus on quality of care and the cost effectiveness of care purchasing decisions has put pressure on plans to provide data that allow comparisons on the basis of quality as well as price. Public and private organizations are actively developing systems for measuring and monitoring outcomes of care, to hold plans accountable for quality.

This paper matches health plans’ financial performance with information on their quality ratings, using data on nearly 200 of the largest health maintenance organizations (HMOs) in the nation. In doing so, we address three important policy questions: (1) Is quality of care influenced by financial performance? (2) Do for-profit plans behave differently from organizations that are organized as not-for-profit? (3) What other factors are associated with the considerable variation in plan performance that we observe?

Background. Databases have only recently become available that permit empirical research on the quality of care delivered by managed care plans. Previously, managed care "performance" was defined almost exclusively in terms of accounting and operating data.² Our empirical strategy is similar to that used in a handful of other studies linking HMOs’ quality measures to finances. Most have focused on the relationship between quality of care and profit status, with mixed results.³

This research has sparked a debate over the usefulness of HEDIS and other report card systems for comparing plans. Presumably, quality measures enable decision makers, such as employee benefits managers, to select plans on the basis of desired plan attributes and, in turn, encourage health plans to improve quality. Employer and employee use of HEDIS scores, however, has appeared to be limited.⁴ Interestingly, this has not discouraged plans from investing in quality improvement activities to improve their ratings.⁵

A plan’s decision to make quality improvements can be viewed as an investment decision. Plans with more financial resources will have an advantage over financially strapped plans in making investments and obtaining high quality scores later on. A plan might finance new projects by taking on new debt or, in the case of for-profit plans, raising funds in the equity market, but the ability to arrange external financing also depends on the plan’s performance. Thus, we expect that past financial performance will be associated with higher quality as it captures the plan’s ability to undertake quality improvement activities.

Data Sources

Top Data Sources Empirical Model And Results Discussion And Policy... NOTES

 
Our primary sources of data are the 1997 and 1998 National Committee for Quality Assurance (NCQA) Health Plan Employer Data and Information Set (HEDIS) 3.0 databases, measuring plan characteristics and performance for 1996 and 1997.We link the HEDIS data, by plan, to financial statement data obtained from the National Association of Insurance Commissioners (NAIC). These data contain measures of financial performance derived from the plans’ balance sheets and income statements, including total premium revenue, medical and administrative expenses, net income, and total assets. They also include plan characteristics, such as plan type, chain affiliation, geographic markets of operation, and enrollment, and some measures of utilization including physician visits and inpatient hospital days. HEDIS information was available for 292 plans. Our multivariate analyses are limited to somewhat smaller samples (140–240 plans) because of lower rates of reporting on specific variables.⁶ The 292 reporting plans cover just over fifty million members (more than 75 percent of the industry). Plans that submit data to HEDIS are believed to be larger and more established than plans that choose not to participate in HEDIS.

HEDIS measures of quality. Since 1991 the NCQA has been collecting information from health plans and developing a standardized system for comparing plan performance on the basis of quality: HEDIS. The specific measures included in HEDIS were chosen for their relevance, scientific soundness, and feasibility.⁷

The HEDIS measures were chosen to facilitate comparison, not for quality improvement. The determination of a measure’s relevance, however, also considers whether it encourages cost-effectiveness and stimulates self improvement, and whether the plan has control over its performance on a measure. These attributes imply an expectation that plan managers will seek ways to improve scores and thereby improve quality.⁸ To the extent that resources may be diverted to those services that are "HEDIS-measurable" and away from those that are not, the end result may not be an improvement in overall quality.

HEDIS has gained general acceptance as tool for measuring managed care plan performance. Some critics still question the reliability of the measures, though, since the data are not fully independently audited and since HEDIS measures are not risk-adjusted.⁹

We focus on the "effectiveness of care" domain in HEDIS (one of five HEDIS domains), in part because the preventive care measures it includes have been the subject of prior research. There is much debate on the cost effectiveness of prevention, but the measures chosen are generally in keeping with practice guidelines that have been promoted by research and medical professional societies.¹⁰

Empirical Model And Results

Top Data Sources Empirical Model And Results Discussion And Policy... NOTES

 
We conducted multivariate analyses to determine if and how the quality measures are related to plans’ financial performance, other plan characteristics, and market characteristics. We selected nine measures of preventive care: childhood immunizations, breast cancer screening, cervical cancer screening, prenatal care in the first trimester, checkup within fourteen days after delivery, prescription for beta-blockers following heart attack, annual eye exams for diabetics, and follow-up office visit after a mental health hospitalization (Exhibit 1). There is substantial variation in the HEDIS measures across HMOs. Simple univariate comparisons suggest that for-profit and not-for-profit plans attain different levels of preventive care services. Where differences are statistically significant, for-profit plans score lower than not for-profit plans do.

We posit that quality of care can be affected by (1) an HMO’s organizational form—through the plan’s objectives and its ability to control and document services; (2) local market conditions—competition, purchaser power, managed care penetration, and local demographics, which are proxies for demand and/or cost shifters; and (3) an HMO’s financial condition, which would affect the resources available for investments in quality and reflect consumers’ demand for quality of care.

We include three measures to capture the influence of organizational form: for-profit status, HMO model type, and whether there is a POS plan. Based on prior research, we expect for-profit HMOs to have lower HEDIS quality scores. The model type and POS variables capture plans’ control over integrated versus nonintegrated (out-of-network) activities.¹¹

Measures of local market conditions are based on the primary metropolitan statistical areas that are included in the HMO service area. Where the HMO serves multiple markets, we average the respective market measures. Our measures include per capita income; the percentage of the population that is nonwhite, resides in urban areas, is over age sixty five, is under age five, is working and employed by the federal government, and is working and employed in large firms; the average managed care penetration from 1995 and 1996, and the hospital Herfindahl index.¹²

The financial measures in our model are net operating income per member, measured over the prior two years, and the percentage change in net worth measured from the previous year. Financial measures were lagged, allowing us to examine how prior financial performance is related to subsequent plan quality. In both cases, we expect more-profitable HMOs to increase their investments in quality, while the financially strapped HMOs reduce expenditures.

We included two additional controls in our analysis. The first captures the data collection method used by the plan. We expect that plans using medical record review would have better reporting of the use of services, which may translate to higher measures. The second measure, the percentage of primary care physicians who are board certified, captures differences in quality of physician training. Except for the financial measures, all variables are contemporaneous and describe sources of variation and associations, not causality.

For each of the nine measures, we estimated two models. The first estimates 1997 HEDIS values as a function of market socioeconomic measures, a rich set of plan descriptive characteristics (including model type and for-profit status), and financial performance measures.¹³ For comparison purposes, our second model is a simpler specification that replicates the 1996 analyses conducted by David Himmelstein and colleagues and updates their results using 1997 data: that is, quality of care is a function of profit status, model type, data collection method, and dummy variables that capture average differences among large geographic regions. We focus our discussion on the contributions of profit status and past profitability.

Do for-profit HMOs provide higher quality? Exhibit 2 summarizes the estimated impact of for-profit status on each of the selected HEDIS preventive care measures. We report average differences in the percentage of patients that receive each service in for-profit plans, relative to not-for-profits. Estimates are obtained by computing regression adjusted means from the regression analyses.

Himmelstein and colleagues found significant differences in all nine prevention measures in 1996. Updating their analysis to 1997, we find a generally similar pattern, although the effect of for-profit status is smaller for most of the measures. This discrepancy suggests that either the relationship has changed over time or other omitted factors confound the effect of ownership.

Do more profitable HMOs have higher HEDIS scores? Exhibit 3 summarizes the relationship between HMO profitability and the HEDIS preventive care measures. We examine differences in quality between low- and high-profit health plans using two alternative measures of profitability: average operating profits per member for 1995–1996, and the change in net worth in the previous year. Operating profits reflect current financial condition, while the percentage change in net worth reflects operating and investment performance, which proxies longer-run profitability.¹⁵ For each measure we report (regression-adjusted) average HEDIS scores for HMOs with low levels of profits and compare this to HMOs with high levels of profits.¹⁶ Significant differences are noted on the "high-profits" columns.

Influence of other factors. Expanding the set of local market characteristics alters the estimated relationships between profit status and the quality measures, which suggests that these characteristics are important sources of variation in the HEDIS scores. Among those measures included to capture variations in the populations served, we find that markets with larger nonwhite and urban populations are associated with lower HEDIS scores. Per capita income, federal employment, and the proportions of members under age five and over age sixty-five do not play important roles.

The remaining market variables are generally important determinants of the quality measures. For example, managed care penetration is positively associated with higher HEDIS scores in eight of the nine measures and significantly so for seven of them. Increasing managed care penetration by 15 percent (approximately one standard deviation in 1995) is associated with three- to ten-point increases in HEDIS measures in our findings. This result is consistent with other findings that suggest that managed care increases competition on the buying side in health care markets and that this competition occurs on the basis of both price and quality.¹⁷

Quality is also associated with the presence of large employers. An increase in the percentage of workers employed in large firms has a significant positive effect in six of the nine measures. This is not surprising, since concentrated purchasing power in health insurance should benefit consumers. While there is much evidence that large employers can push down insurance premiums and other health care prices, we provide some of the first evidence that quality of care may benefit from employer purchasing power as well.

Competition in provider markets, as measured by the hospital Herfindahl index, is also a significant determinant of quality. We find that more competition (less concentration) is associated with higher HEDIS scores. This is consistent with a fair bit of literature that suggests that managed care works better where plans and providers are competing.¹⁸

We acknowledge the possibility that if plans mimic the predominant type of plan in a given geographic area, specification of market controls might mask some of the impact of plan type. We find no evidence, however, that plan type is determined by market characteristics in our data.

Discussion And Policy Implications

Top Data Sources Empirical Model And Results Discussion And Policy... NOTES

 
The relationship between managed care plans’ financial status and their ability to deliver appropriate quality of care is a flashpoint on the policy screen. For better or worse, the HEDIS data set is the single most comprehensive, widely used quality report card system available for gauging the nonfinancial dimensions of plan performance.

For-profit status, in our analysis, is not a major factor in observed differences in plan performance. Other economic and market factors explain, in a statistical sense, a significant amount of the variation we observe. Competition and the nature of the health care marketplace are important. Quality measures are generally higher when plans operate where there is greater managed care presence and where the structure of the hospital market gives opportunities for more competition. The presence of large employers, which may wield strong purchasing power, is also important.¹⁹

Finally, financial performance and profitability matter. Our findings suggest that plans may not be making a choice between profits and quality but, rather, that plans need to generate profits to enable quality. While this is not new to economists, it is hardly a conventional notion in health policy circles.

State regulation of HMOs has traditionally focused on insurance aspects of the business and less on the operating side in the delivery of care. Recently even large, well-established HMOs—including industry quality leaders such as Harvard Pilgrim—have had financial problems and operating losses associated with poor pricing decisions and poor cost management. If, as this paper suggests, financial resources and profitability enable HMOs to invest in higher quality of care, then it is instructive to recognize these links as part of an early warning system for monitoring quality of care.

Editor's Notes

 
Patricia Born is assistant professor of finance at the University of Connecticut in Storrs. Carol Simon is associate professor of health services and director of the Health Economics Program at Boston University School of Public Health.

This research was supported by the University of Illinois-Chicago Campus Research Board. The authors thank participants at the 2000 Annual Meeting of the American Economic Association for comments.

NOTES

Top Data Sources Empirical Model And Results Discussion And Policy... NOTES

 

1. The Bipartisan Consensus Managed Care Improvement Act of 1999, which passed 7 October 1999, allows patients the right to sue their health plan. However, if a plan complies with an external reviewer’s decision, it may not be held liable for punitive damages. See H.R. 2723, 106th Cong., 2d sess.
2. For a survey of studies of plan performance, see R.H. Miller and H.S. Luft, "Managed Care Plan Performance since 1980: A Literature Analysis," Journal of the American Medical Association 271, no. 19 (1994): 1512–1519. Studies of financial performance include R. Feldman, D. Wholey, and J. Christianson, "Effect of Mergers on Health Maintenance Organization Premiums," Health Care Financing Review 17, no. 3 (1996): 171–189; H.J. Bryce, "Profitability of HMOs: Does Non-Profit Status Make a Difference?" Health Policy 28, no. 3 (1994): 197–210; and M. Schlesinger, D. Blumenthal, and E. Schlesinger, "Profits under Pressure: The Economic Performance of Investor-Owned and Nonprofit Health Maintenance Organizations," Medical Care 24, no. 7 (1986): 615–627.
3. D.U. Himmelstein et al., "Quality of Care in Investor-Owned versus Not-for-Profit HMOs," Journal of the American Medical Association 282, no. 2 (1999): 159–163; F.A. Sloan et al., "Hospital Ownership and Cost and Quality of Care: Is There a Dime’s Worth of Difference?" Health Policy Discussion Paper (Washington: American Enterprise Institute, 11 September 1998); and P. Born and C. Geckler, "HMO Quality and Financial Performance: Is There a Connection?" Journal of Health Care Finance 24, no. 2 (1998): 65–77.
4. Prior studies suggest that the quality measures collected ignore the patient perspective, so they are not useful for consumers’ decision making. See S. Edgman-Levitan and P.D. Cleary, "What Information Do Consumers Want and Need?" Health Affairs (Winter 1996): 42–56; D.P. Scanlon and M. Chernew, "HEDIS Measures and Managed Care Enrollment," Medical Care Research and Review 56, Supp. 2 (1999): 60–84; and J.R. Gabel, K.A. Hunt, and K. Hurst, "When Employers Choose Health Plans Do NCQA Accreditation and HEDIS Data Count?" (New York: Commonwealth Fund, September 1998).
5. T. Bodenheimer and L. Casalino, "Executives with White Coats—The Work and World View of Managed-Care Medical Directors," Part 2, New England Journal of Medicine 341, no. 26 (1999): 2029–2032.
6. The data match was not always one-to-one because the units of observation do not always match. HEDIS quality measures are reported by line of business (HMO, POS, PPO), but the NAIC financial information is reported at the overall organization level. For managed care organizations that only offer an HMO or a POS product, the link is one-to-one. For organizations that offer an HMO and a POS plan, or all three types, the financial data are duplicated for each line of business. Limited financial data are available in the HEDIS database, for roughly 195 plans, depending on the variable. When the data are linked to the NAIC set, 141 matches occur.
7. National Committee for Quality Assurance, "Desirable Attributes of HEDIS Measures" (Washington: NCQA, January 1998).
8. Bodenheimer and Casalino ("Executives with White Coats") suggest that priority is given to activities that are closely related to HEDIS scores.
9. In 1997 and 1998 some plans voluntarily submitted to audits of selected measures. The NCQA intends to require audits in (undesignated) future HEDIS releases. Where we have found obvious errors or omissions in the data, we have attempted to obtain corrected values from other sources. For example, many plans did not indicate profit status; we obtained this information from the HMO/PPO Directory (Montvale, N.J.: Medical Economics, 1998).
10. The cost-effectiveness of some common preventive services is debatable. See J. Wagner, "From the Congressional Office of Technology Assessment: Costs and Effectiveness of Colorectal Cancer Screening in the Elderly," Journal of the American Medical Association 264, no. 21 (1990): 2732.
11. More integrated models may be better able to monitor and influence provider/patient behavior and suffer less from measurement and "denominator" problem (identifying the potential pool of patients eligible for services). IPA and network plans, by virtue of their more open panels of providers, typically have difficulties amassing patient records and identifying eligible enrollees. Biases run both ways—failure to document services provided outside of the network understates prevention and incidence measures; failure to count eligible patients who did not receive care biases measures upward. Finally, patients may select HMO types based on anticipated use of services and need for specialized care. Healthier persons have been more likely to enroll in more tightly controlled plans.
12. We posit that quality of care is a normal good and expect a positive association. We expect lower effective demand for high-quality care in markets with higher nonwhite populations, where financial and geographic access may be limited. We expect urbanization to be associated with greater demand for high-quality care. We expect the demand for certain preventive services (childhood immunizations) and the costs of providing certain services to be related to the number of potential users in the elderly and infant populations. We expect that the federal government exercises purchasing power in health insurance markets. Plans included in the Federal Employees Health Benefits Program (FEHBP) are required to provide, on average, more information to enrollees and the government. Finally, employment in large firms captures employer purchasing power and is expected to encourage quality. Our managed care penetration variable is measured as the share of physician revenues earned from all managed care contracts. We expect competition on HEDIS measures to be positively associated with the level of managed care penetration in the marketplace. The hospital Herfindahl index is the squared market shares of hospitals in the market, measured on number of admissions to area hospitals, and measures the degree of competition in the local market. The higher the index, the lower the potential competition.
13. In unreported results, we introduced only financial variables, and then only market variables. The results were similar. We also estimated equations including both the regional dummies and the local market variables. In all but one HEDIS measure regression, we rejected the hypothesis that the regional dummies added significant additional explanatory power when the market variables were also included. Results are available from the authors; send e-mail to Patty Born at <pborn{at}sba.uconn.edu>.
14. Moving from specification 1 to specification 2 reduces sample size because financial data are not reported by some plans. We reestimated specification 1 for the subset of plans that are included in specification 2. Results were largely the same, especially with respect to the coefficients on for-profit status. The subsample of plans that report financial data—in particular, the subset for which financial data are available for more than two years—is different from the HEDIS reporting set. In general, plans with longer time series of financial data are larger, older, and somewhat more profitable, although a simple test of differences was not significant. Further research will investigate nonreporting biases in greater detail.
15. To the extent that operating profits are the predominant source of the change in net worth, we expect the measures to yield similar results. The percentage change in net worth captures the any additional profit earned through investments and other activities.
16. Low profitability is defined as one standard deviation below the mean, and high profitability is defined as one standard deviation above.
17. R.H. Miller and H.S. Luft, "Does Managed Care Lead to Better or Worse Quality of Care?" Health Affairs (Sep/Oct 1997): 7–25; and J. Escarce et al., "Health Maintenance Organizations and Hospital Quality for Coronary Artery Bypass Surgery," Medical Care Research and Review 56, no. 3 (1999): 340–362.
18. D. Wholey, R. Feldman, and J.B. Christianson, "The Effect of Market Structure on HMO Premiums," Journal of Health Economics 14, no. 1 (1995): 81–105.
19. Our findings that market characteristics are more closely related to quality measures than to organizational form support the premise that the behavior of managers in organizations is influenced by industry wide events and ideological consensus, which may not be rational. W.W. Powell and P.J. DiMaggio, eds., The New Institutionalism in Organizational Analysis (Chicago: University of Chicago Press, 1991).

                      What's this?

This article has been cited by other articles:

				C. Kim, D. F. Williamson, C. M. Mangione, M. M. Safford, J. V. Selby, D. G. Marrero, J. D. Curb, T. J. Thompson, K.M. Venkat Narayan, and W. H. Herman Managed Care Organization and the Quality of Diabetes Care: The Translating Research Into Action for Diabetes (TRIAD) study Diabetes Care, July 1, 2004; 27(7): 1529 - 1534. [Abstract] [Full Text] [PDF]

				E. C. Schneider, A. M. Zaslavsky, and A. M. Epstein Use of High-Cost Operative Procedures by Medicare Beneficiaries Enrolled in For-Profit and Not-for-Profit Health Plans N. Engl. J. Med., January 8, 2004; 350(2): 143 - 150. [Abstract] [Full Text] [PDF]

				W. S. Brewbaker III Will Physician Unions Improve Health System Performance? Journal of Health Politics Policy and Law, August 1, 2002; 27(4): 575 - 604. [Abstract] [PDF]

				B. H. GRAY and J. P. CLEMENT Databases for Research on Nonprofit Health Care Organizations: Opportunities and Limitations American Behavioral Scientist, June 1, 2002; 45(10): 1550 - 1591. [Abstract] [PDF]

				H. T. Tu and J. D. Reschovsky Assessments of Medical Care by Enrollees in For-Profit and Nonprofit Health Maintenance Organizations N. Engl. J. Med., April 25, 2002; 346(17): 1288 - 1293. [Abstract] [Full Text] [PDF]

				J. L. Zarin Quality of Care and HMO Financial Performance AAP Grand Rounds, September 1, 2001; 6(3): 33 - 34. [Full Text] [PDF]