0
We're unable to sign you in at this time. Please try again in a few minutes.
Retry
We were able to sign you in, but your subscription(s) could not be found. Please try again in a few minutes.
Retry
There may be a problem with your account. Please contact the AMA Service Center to resolve this issue.
Contact the AMA Service Center:
Telephone: 1 (800) 262-2350 or 1 (312) 670-7827  *   Email: subscriptions@jamanetwork.com
Error Message ......
Original Investigation |

Primary Care Quality in the Medicare Program:  Comparing the Performance of Medicare Health Maintenance Organizations and Traditional Fee-for-Service Medicare FREE

Dana Gelb Safran, ScD; Ira B. Wilson, MD, MSc; William H. Rogers, PhD; Jana E. Montgomery, ScM; Hong Chang, PhD
[+] Author Affiliations

From The Health Institute, Division of Clinical Care Research, New England Medical Center (Drs Safran, Wilson, Rogers, and Chang and Ms Montgomery), and the Department of Medicine, Tufts University (Drs Safran, Wilson, Rogers, and Chang), Boston, Mass.


Arch Intern Med. 2002;162(7):757-765. doi:10.1001/archinte.162.7.757.
Text Size: A A A
Published online

Background  Since 1972, Medicare beneficiaries have had the option of enrolling in a Medicare-qualified health maintenance organization (HMO). Little information exists to inform beneficiaries' choices between the traditional fee-for-service (FFS) Medicare program and an HMO.

Objectives  To compare the primary care received by seniors in Medicare HMOs with that of seniors in the traditional FFS Medicare program, and among HMOs, and to examine performance differences associated with HMO model-type and profit status.

Methods  Data were derived from a cross-sectional observational survey of Medicare beneficiaries 65 years or older in the 13 states with mature, substantial Medicare HMO markets. Only beneficiaries continuously enrolled for 12 months or more in traditional FFS Medicare or a qualified Medicare HMO were eligible. Data were obtained using a 5-stage protocol involving mail and telephone (64% response rate). Analyses included respondents who identified a primary physician and had all required data elements (N = 8828). We compared FFS and HMO performance on 11 summary scales measuring 7 defining characteristics of primary care: (1) access, (2) continuity, (3) integration, (4) comprehensiveness, (5) "whole-person" orientation, (6) clinical interaction, and (7) sustained clinician-patient partnership.

Results  For 9 of 11 indicators, performance favored traditional FFS Medicare over HMOs (P<.001). Financial access favored HMOs (P<.001). Preventive counseling did not differ by system. Network-model HMOs performed more favorably than staff/group–model HMOs on 9 of 11 indicators (P<.001). Few differences were associated with HMO profit status.

Conclusions  The findings are consistent with previous comparisons of indemnity insurance and network-model and staff/group–model HMOs in elderly and nonelderly populations. The stability of results across time, geography, and populations suggests that the relative strengths and weaknesses of each system are enduring attributes of their care. Medicare enrollees seem to face the perennial cost-quality trade-off: that is, deciding whether the advantages of primary care under traditional FFS Medicare are worth the higher out-of-pocket costs.

SINCE 1972, Medicare beneficiaries in many parts of the country have had the option of enrolling in a Medicare-qualified health maintenance organization (HMO) rather than receiving their care through the traditional fee-for-service (FFS) Medicare program. The decision to allow seniors to enroll in HMOs was heralded by economists for the potential cost savings, and by geriatricians for the potential improvements in patient care in a more highly integrated system.14 Integration of care assumes heightened importance in the care of patients with chronic illness, as is predominantly the case with Medicare beneficiaries (87%).5 A hallmark of the HMO model is its pairing of patients with a single, primary physician who assumes responsibility for coordinating and approving all aspects of the patient's care.

From 1972 until 1990, enrollment in Medicare HMOs was modest and grew slowly. However, over the next decade, the number of Medicare-authorized HMOs tripled (from 95 to 262), and enrollment in them more than tripled (from 2 million to 6.2 million), with most of the growth occurring after 1992.6 With enrollment now expanded well beyond a few isolated geographic areas with a few dozen plans and limited participation, there is renewed interest in evaluating the performance of Medicare HMOs relative to the traditional FFS Medicare program. For example, has the promise of better-integrated health care been realized in Medicare HMOs? Is there evidence that primary care performance overall (not just integration of care) is superior in Medicare HMOs, which emphasize the primary care physician's role and require beneficiaries to select a primary physician to coordinate their care?

Primary care has been defined by the Institute of Medicine and others to require the following 7 core characteristics: (1) accessibility, (2) continuity, (3) comprehensiveness, (4) integration, (5) clinical management, (6) a "whole-person" orientation, (7) and a sustained clinician-patient partnership.712 Numerous studies have compared performance of indemnity and managed care systems on 1 or more of these attributes, and some have compared performance across different models of managed care. However, little information exists with which to compare the traditional FFS and managed Medicare systems on these attributes, and that information available derives largely from 2 evaluations conducted in 1985-1986 (National Medicare Competition Evaluation [NMCE]) and 1990 (Tax Equity and Fiscal Responsibility Act [TEFRA] Evaluation), respectively.1316 The dramatic proliferation of Medicare HMOs since 1990, including their emergence in states with little or no previous involvement, compels analysis of more current data. In addition, while studies of non-Medicare populations have identified substantial primary care performance differences across different models of managed care,1720 studies of managed Medicare have focused almost exclusively on its comparison with traditional FFS Medicare, and have not examined performance differences among different types of Medicare HMOs. There have been no studies comparing the primary care performance of for-profit and not-for-profit Medicare HMOs.

The present study compares performance on the 7 defining characteristics of primary care in traditional FFS Medicare and Medicare HMOs, and, among Medicare HMOs, compares the performance by HMO model-type (staff/group vs independent practice association [IPA]/network) and by HMO tax status (for-profit vs not-for-profit).

Data for these analyses are the first to be published from the Study of Choice and Quality in Senior Health Care, a longitudinal observational study of elderly Medicare beneficiaries in 13 states (Arizona, California, Colorado, Florida, Illinios, Massachusetts, Minnesota, New Mexico, New York, Oregon, Pennsylvania, Texas, and Washington). The study compares the primary care performance and outcomes of the traditional Medicare and Medicare HMO systems. These analyses use baseline data (1998).

STUDY DESIGN AND SAMPLING

Sampling for the study occurred in 3 consecutive stages: first, the states were selected; second, we selected HMOs within the chosen states; and finally, beneficiaries were selected from qualifying HMOs and from the traditional FFS Medicare program. Because the study goal was to evaluate Medicare HMOs in their mature—not nascent—stages, we excluded states and plans within states where managed Medicare was newly emerging.

The inclusion criteria for states were that there be at least 5% enrollment in managed Medicare as of 1994 or earlier, and that there be at least 1 market containing 2 or more mature Medicare HMOs. All states meeting these criteria were included (n = 13). Nationwide, 80% of Medicare HMO enrollees and 28% of FFS Medicare enrollees reside in these 13 states.

The inclusion criteria for Medicare HMOs within qualifying states were as follows: (1) a risk contract with the Health Care Financing Administration (HCFA) (now called the Centers for Medicare and Medicaid Services); (2) substantial enrollment (≥2500 enrollees) since 1994 or before; and (3) at least 3000 enrollees at the time of sampling (September 1998). All Medicare HMOs meeting these criteria (n = 121) were included. These HMOs cover 99.2% of Medicare HMO enrollees in the 13 study states. Rural areas were almost entirely excluded from the study owing to the absence of any such areas with 2 or more well-established Medicare HMOs.

Eligibility criteria for beneficiaries were the following: (1) age 65 years or older; (2) payment of Medicare Part B premium (by beneficiary or other payee); (3) residence in one of the 13 qualifying states; and (4) continuous enrollment for 1 year or more in the traditional FFS Medicare program or in one of the qualifying Medicare HMOs. The latter criterion was imposed to assure that study participants had experience on which to evaluate their care and to avoid the high rates of disenrollment that occur during the initial months of participation in a Medicare HMO,13,21 which would impede longitudinal analyses. In addition, because data collection was principally by self-administered questionnaires, we excluded patients meeting any of the following 3 criteria indicated on the HCFA data file: (1) principal language other than English; (2) residence in a skilled nursing facility; or (3) representative payee listed, indicating that the patient did not manage his or her own affairs.

The HCFA provided data files for sampling. The files included listings for all elderly beneficiaries from the 13 study states who had been continuously enrolled for 1 year or more in the traditional FFS Medicare program (n = 6 236 793) or in their current HMO (n = 2 903 002). From these files, a starting sample of 10 666 Medicare HMO enrollees and 5332 traditional FFS Medicare enrollees was drawn (n = 15 998). The HMO enrollees were randomly selected from the 121 qualifying plans, with sample sizes defined in accordance with the study's allocation model results. An allocation model is a sampling method that assigns the exact numbers of individuals to be sampled from each entity (eg, HMO) by simulating the study's principal regression analyses and minimizing the estimation variance for these analyses.22 Using this method, the largest HMOs contributed approximately 0.2% of their members to the sample, and the smallest HMOs contributed approximately 0.4% of their members. Under straight proportional sampling, each plan would have contributed roughly 0.26% of its members to the sample.

Once the HMO sample was drawn, FFS beneficiaries were matched to HMO beneficiaries based on age, sex, and residential ZIP code, and then randomly sampled in a 1:2 ratio (1 FFS enrollee to 2 HMO enrollees). Matching at baseline enabled the study to more appropriately compare enrollees in the FFS and HMO systems in certain types of analyses, including outcomes over the ensuing years.

DATA COLLECTION

Baseline data were obtained using a standard 5-stage survey protocol involving mail and telephone.23 The protocol included (1) an advance letter signed by the HCFA administrator; (2) an initial survey mailing; (3) a reminder post-card; (4) a second survey mailing; and (5) telephone follow-up using up to 10 attempts. An abbreviated survey was administered in telephone interviews. The abbreviated survey was mailed to nonrespondents with no telephone number available. Long-form surveys were completed between October 1998 and December 1998, and short-form surveys were completed between January 1999 and March 1999. This protocol yielded 9625 responses—8118 long-forms and 1507 short-forms (404 by mail and 1103 by telephone)—for a baseline response rate of 64% after deceased beneficiaries and bad addresses were removed (9625/15 140). Administrative data from the HCFA revealed that nonrespondents were older and included more nonwhites and women than respondents (data available at: http://www.nemc.org/general; accessed January 29, 2002). Similarly, delayed respondents (short-form mail and telephone)—whose characteristics are presumed to approximate those of nonrespondents—were more likely than early respondents (long-form mail) to be nonwhite, have less education and income, and be in poor health (short-form mail only) (data available at: http://www.nemc.org/general; accessed January 29, 2002).

The long-form survey contained 190 items covering 7 topic areas: (1) primary care quality, (2) health care coverage, (3) health care spending, (4) medication regimen, (5) health status, (6) health care utilization, and (7) sociodemographic characteristics. The short-form survey contained 103 items covering the same 7-topic areas in a more limited way. All modules derived from instruments that had been extensively tested and validated.2430 The primary care module included the Primary Care Assessment Survey (PCAS), a validated questionnaire that uses 11 summary scales to measure 7 defining characteristics of primary care named by the Institute of Medicine and others712: (1) accessibility, (2) continuity, (3) integration, (4) comprehensiveness, (5) a whole-person orientation, (6) clinical management, and (7) a sustained clinician-patient partnership. These analyses group the 11 PCAS scales into 2 categories: structural features of care and quality of physician-patient interactions. Table 1 summarizes the item content for the PCAS scales. The PCAS does not assess technical aspects of clinical quality because of limitations of patient-provided information about these issues.3134 Further detail about the development and validation of the PCAS is available elsewhere.26,35

Table Graphic Jump LocationTable 1. Summary of Item Content for Primary Care Assessment Survey Scales
STATISTICAL ANALYSES

Analyses included all baseline respondents who reported having a primary physician and who had all required data elements (n = 8828). Analyses excluded those without a primary physician (n = 245 from FFS; n = 408 from HMOs) and those for whom essential data were missing (n = 144). Beneficiaries' enrollment status (traditional FFS Medicare vs Medicare HMO) was known from administrative data provided by the HCFA; and for HMO enrollees, HCFA administrative data indicated the HMO model-type (IPA/network vs staff/group) and tax status (for-profit vs not-for-profit). All other data for analyses were obtained from beneficiaries through their completed questionnaires.

We compared the sociodemographic and health profiles of beneficiaries in traditional FFS Medicare with those in Medicare HMOs overall and by HMO model-type and tax status. Next, we used multivariable linear regression to model performance on each of the 11 PCAS scales as a function of Medicare delivery system (traditional FFS Medicare vs Medicare HMO), controlling for beneficiaries' sociodemographic profile (age, sex, race, and years of education), state, and 11 chronic conditions that are sensitive to good primary care management and that are highly prevalent among US adults 65 years and older (hypertension, angina, congestive heart failure, recent myocardial infarction, stroke, asthma, emphysema, chronic obstructive pulmonary disease, diabetes, arthritis, and cancer).36

The effect of HMO model-type was tested in separate models with binary indicators of HMO model-type (IPA/network, staff/group) and traditional FFS Medicare as the reference group, with control variables as noted above. We tested for interactions between state and system (FFS, IPA/network, staff/group). For each HMO model-type (IPA/network, staff/group), we tested for individual plan effects using analysis of variance methods, modeling each PCAS scale as a function of patient characteristics and plan identifiers. The effect of HMO tax status was tested in models with binary indicators of for-profit and not-for-profit HMOs controlling for HMO model-type and the set of variables noted above.

We tested the sensitivity of the analytic results to patients' years of experience with their current primary care physician (relationship duration) and years of enrollment in their current Medicare delivery system (FFS or current HMO) by adding indicators of these 2 characteristics to all regression models.

To test for effects of the long-form vs short-form PCAS scales, all analyses were performed in 3 ways, and the results compared: (1) long-form scales with long-form respondents only; (2) short-form scales with the full analytic sample; and (3) long-form scales with the full analytic sample, where long-form values for short-form respondents were estimated using seemingly unrelated regression methods.37 The results were the same in all 3 methods. We therefore report the results from the third method to make use of the full analytic sample and to permit comparison of the study data to other published studies that have used the long-form scales.18,26,3840

Finally, for each model, we tested the sensitivity of the system effects to patient health status through the use of interaction terms. We tested 2 sets of models including interactions between system (traditional FFS Medicare, IPA/network–model HMO, or staff/group–model HMO) and health status. The first set evaluated the interaction between the system and the count of chronic medical conditions (a binary indicator differentiating patients with 2 or more conditions from those with 1 or no conditions). The second set evaluated the interaction between the system and a summary indicator of health status from the Short-Form 36-Item Survey24 (a binary indicator differentiating patients with physical health index scores below the national mean for adults 65 years and older [41 points]41 from those at or above the national mean for this age cohort).

Probability sampling weights, derived as the inverse of sampling probabilities, were applied to all analyses to correct for the effects of differences across the strata. The statistical software used in our analyses (STATA 6.0) takes these weights into account when computing the SE. P values in regression analyses were corrected for multiple comparison by means of the Bonferroni method for individual t tests to account for analysis of 11 dependent variables.

Table 2 summarizes the sociodemographic and health characteristics of the analytic sample, comparing those in the traditional FFS Medicare program (n = 2545) with those in Medicare HMOs (n = 6283), and among HMO enrollees, comparing those in IPA/network–model HMOs (n = 3913) with those in staff/group–model HMOs (n = 2370). On average, HMO enrollees were younger, more likely to be poor (yearly household income <$20 000), more likely to be Hispanic, and healthier than their counterparts in traditional FFS Medicare. Traditional FFS Medicare enrollees had more chronic health conditions (P<.001) and lower functional status as measured by basic activities of daily living (P<.001), instrumental activities of daily living (P<.001), and the physical component summary index of the SF-36 (P<.05). There were few demographic or health differences between the 2 HMO model-types, except with respect to race.

Table Graphic Jump LocationTable 2. Characteristics of the Analytic Sample*

Primary care performance comparisons among the Medicare delivery systems are summarized in Table 3. Overall, performance favored traditional FFS Medicare over Medicare HMOs on 9 of 11 PCAS scales in analyses that did not differentiate between HMO model-type. Financial access to care was significantly better among Medicare HMO enrollees (P<.001), and preventive counseling did not differ by delivery system. The traditional FFS Medicare system performed significantly better on organizational access to care, duration of primary care relationships, visit-based continuity, and all scales pertaining to the quality of the physician-patient interaction (P<.001). The magnitude of the differences between systems—measured by the standardized effect size (ES) —ranged from 0.17 to 0.38 for structural features of care and from 0.16 to 0.25 for scales measuring the quality of the physician-patient interaction (ES = [mean FFS score − mean HMO score]/scale SD). Overall, the largest differences were those associated with financial access (ES = 0.38 favoring HMO), relationship duration (ES = 0.30 favoring FFS), and visit-based continuity (ES = 0.28 favoring FFS). The smallest significant differences were those associated with the thoroughness of physical examinations (ES = 0.16 favoring FFS), organizational access to care (ES = 0.17 favoring FFS), and trust (ES = 0.17 favoring FFS). The results were unchanged when probability sampling weights were excluded from analyses and were insensitive to controls for primary care relationship duration and tenure in current Medicare delivery system. Interactions between delivery system and patient health status showed no significant effects.

Table Graphic Jump LocationTable 3. Mean Adjusted Primary Care Assessment Scores by Medicare Delivery System*

Differences between the 2 HMO model-types and traditional FFS Medicare showed the identical pattern in terms of direction and statistical significance of effects, although the magnitude of the differences between IPA/network–model HMOs and FFS were consistently smaller than the differences between staff/group–model HMOs and FFS. The most substantial difference between IPA/network–model HMOs and FFS were in the areas of financial access to care (ES = 0.33 favoring HMO) and duration of primary care relationships (ES = 0.27 favoring FFS). The largest differences between staff/group–model HMOs and FFS were in financial access (ES = 0.47 favoring HMO), visit-based continuity (ES = 0.45 favoring FFS), and physician's knowledge of the patient (ES = 0.37 favoring FFS). Comparing IPA/network–model HMOs and staff/group–model HMOs with each other, the IPA/network models performed significantly better on 9 of 11 PCAS scales (P<.001). Financial access to care was more favorable in staff/group–model HMOs (P<.001), and the model-types did not differ with respect to preventive counseling.

Analysis of variance results revealed no significant plan effects within HMO model-type. Models including state-system interaction terms revealed no significant state-system interactions in IPA/network vs FFS comparisons and limited state-system interactions in staff/group vs FFS analyses. The latter interactions were accounted for almost exclusively by the performance of 1 staff/group–model HMO in Arizona, which scored significantly higher than FFS (and IPA/networks) in that geographic area on 3 dimensions of care (organizational access, integration, and thoroughness of physical examinations).

Health maintenance organization tax status was associated with significant performance differences for 4 of 5 structural features of primary care and 1 element of the physician-patient interaction (Table 4). The largest differences associated with tax status were in visit-based continuity (ES = 0.26 favoring for-profit HMOs) and relationship duration (ES = 0.17 favoring not-for-profit HMOs). For organizational access, financial access, and preventive counseling, the effect sizes associated with tax status were small (ES = 0.09, 0.08, and 0.08, respectively).

Table Graphic Jump LocationTable 4. Adjusted Comparison of Not-for-Profit and For-Profit Medicare HMOs*

In the year 2000, nearly 20% of elderly Medicare beneficiaries were enrolled in a Medicare HMO compared with 6% in 1990.6 While new initiatives provide comparative data about individual Medicare HMOs to inform beneficiaries' choices,42 there is little information comparing the performance of Medicare HMOs with the traditional FFS Medicare program. Two large-scale evaluations of Medicare HMOs vs traditional FFS Medicare (NMCE and TEFRA)1316 were conducted more than a decade ago, when there were only a few dozen Medicare HMOs nationwide and enrollment in them was limited to select populations in select geographic areas. In this context, the present study affords the first comparison of the traditional and managed Medicare systems since Medicare HMOs have become a substantial player in the US health care delivery system.

Overall, the findings reveal that primary care performance favors the traditional FFS Medicare system over Medicare HMOs for 9 of 11 features of care that were examined. A salient advantage of Medicare HMOs was their superior financial access to care (ie, fewer cost-related barriers to care). Among HMOs, IPA/networks emerged with more favorable performance than staff/group– model HMOs on 9 features of care, but performed less favorably than traditional FFS Medicare on those same features. The results did not differ for patients in better or worse health, and the relative performance of the systems was the same in all 13 states, with the exception of 1 staff/group–model HMO that outperformed FFS and IPA/networks in its area (Arizona) on 3 aspects of care. The findings are markedly consistent—in both direction and magnitude of effect—with previous comparisons of indemnity and managed care systems' performance on the defining elements of primary care. We summarize this literature below.

STRUCTURAL FEATURES OF CARE

Like the present study, previous studies of access and continuity have consistently found HMO enrollees to report more favorable financial access than patients with indemnity insurance, while organizational access and continuity of care have favored indemnity insurance over HMOs.1520,31,4348 Studies that have differentiated HMOs by model-type have found, as we did, that financial access favored staff/group–model HMOs, while organizational access and primary care relationship duration were least favorable in staff/group–model HMOs.17,18,47,49 Although few studies have examined visit-based continuity, the present study replicates findings from a study of commercially insured adults, which found visit-based continuity to be highest among those with indemnity insurance, intermediate in IPA/network–model HMOs, and the lowest in staff/group–model HMOs.18

Previous studies' findings regarding integration of care also accord with our observations. The TEFRA evaluation found higher levels of integration in traditional FFS Medicare than in Medicare HMOs among patients seen for joint pain (n = 4252) or chest pain (n = 1080).16 The TEFRA evaluation did not differentiate among HMO model-types. Recent studies of commercially insured adults found better integration of care reported by patients with indemnity insurance than by HMO enrollees,18,50 and when HMO model-type was distinguished, found more integrated care reported by IPA/network–model HMO members than by those in staff/group–model HMOs.18

QUALITY OF PHYSICIAN-PATIENT INTERACTIONS

Our findings with respect to the quality of physician-patient interactions are also consistent with previous research in Medicare and non-Medicare study populations. Both the NMCE and TEFRA evaluations found that communication quality favored traditional FFS Medicare over Medicare HMOs—with Medicare HMO enrollees more likely to report that clinicians did not explain medical problems sufficiently,13 and that explanations were poor.15 The TEFRA study found less favorable interpersonal treatment reported by Medicare HMO enrollees than by those in traditional FFS Medicare.15 Among non-Medicare study populations, patients' assessments of both communication quality and interpersonal treatment have consistently favored patients in indemnity insurance over those in HMOs, and among HMO members, have found those in staff/group–model HMOs to report significantly less favorable communication and interpersonal treatment than their counterparts in IPA/network–model HMOs.17,18,20,31,44,45,48 Only 1 previous study has compared system performance with regard to a whole-person orientation and patient trust,18 and that study's findings in a commercially insured adult population mirrored those observed here among Medicare enrollees.

Finally, evidence regarding preventive care remains mixed. In 1985-1986, the NMCE found higher rates of preventive screening in staff/group–model HMOs than in IPA/systems or in traditional FFS Medicare,14 while in 1990, the TEFRA evaluation found more preventive counseling reported among traditional FFS Medicare enrollees than among those in Medicare HMOs.15 The present study found no significant differences in the rates of preventive counseling (smoking, alcohol use, diet, exercise) among Medicare beneficiaries by delivery system, although rates of counseling seemed low in all systems.

LIMITATIONS

There are several relevant limitations to this study. First, we did not evaluate technical elements of clinical quality. Studies comparing system performance on technical elements of quality are needed to further our understanding of the relative strengths and weaknesses of the systems.

Second, because of nonresponse, the study underrepresents seniors from racial and ethnic minority groups, those who are in frail health, and those who are socioeconomically disadvantaged because of low levels of education and/or income. Detailed assessment of the primary care experiences and views of Medicare beneficiaries in these vulnerable population subgroups would provide a fuller understanding of the relative performance of the traditional and managed Medicare systems.

Third, there are only limited benchmarks available with which to frame the observed system differences in terms of their clinical and/or market significance. While the PCAS scales have been shown to predict outcomes, including patients' adherence to medical advice and voluntary disenrollment from a primary care physician's practice,40,51,52 we do not yet know whether the magnitude of the observed differences between Medicare systems will be associated with clinically important outcome differences. In absolute terms, however, the magnitude of the observed system differences (ie, standardized effect sizes averaging 0.17 for traditional FFS Medicare vs IPA/network–model HMO comparisons and 0.32 for traditional FFS Medicare vs staff/group–model HMOs) are comparable with the effect sizes observed for common pharmacologic therapies on patients' functional health status. For example, the effect sizes associated with omeprazole (vs placebo) in the treatment of reflux symptoms, fluticasone (vs placebo) in the treatment of asthma, and salmeterol (vs placebo) in the treatment of chronic obstructive pulmonary disease are 0.30, 0.27, and 0.17, respectively, when net changes in physical health status are assessed.5355

Finally, the study is restricted to states and areas within states with at least 2 large, mature Medicare HMOs. While 99% of Medicare HMO enrollees nationally reside in the states and plans studied, only 28% of traditional FFS Medicare beneficiaries reside here. It is unclear whether the observed performance of the traditional FFS Medicare system generalizes to areas in which Medicare HMO presence in the market is more limited or is altogether absent.

The results of this study, together with 25 years of research evaluating performance differences among delivery systems, suggest that the observed differences between indemnity insurance and HMOs (and within HMOs, between IPA/network– and staff/group–model plans) are enduring features of primary care performance that persist across time, geography, and population. The consistent findings across multiple studies suggest that there is something structurally different about the way primary care is provided in HMOs and, in particular, in staff/group–model HMOs, that leads to these differences. Members of our group have previously hypothesized that the observed performance differences arise from multiple factors including a different orientation toward the role of the "team" in open vs closed-model practice settings, and differences in the professional expectations and preferences of physicians attracted to practice in these different types of organizations.18,56Closed-model practice settings refer to those in which physicians work exclusively for 1 health plan (ie, staff/group–model HMOs), while open-model settings refer to those in which physicians accept patients from multiple health plans (eg, indemnity insurance, IPA/network–model HMOs, and preferred provider organizations). For whatever combination of reasons, open- and closed-model practices seem to consistently produce very different rates of visit-based continuity, and perhaps related to that, patients in closed-model settings consistently report less rich and less enduring clinician-patient relationships (as measured by shorter primary care relationships, lower scores on interpersonal treatment, physician's knowledge of the patient, quality of communication, and patient trust).17

The perennial question remains: are the differences in the quality of relationships, quality of the clinical encounter, organizational access, and integration worth the higher out-of-pocket costs paid by beneficiaries in traditional FFS Medicare vs Medicare HMOs? What individuals are willing to spend on quality will continue to depend on their individual values and preferences. At a minimum, society can offer information to inform beneficiaries about the trade-offs. At a maximum, policymakers can intervene if quality does not meet an agreed-on standard. These data suggest that we can reliably tell patients what key differences in primary care will be in indemnity insurance arrangements, IPA/network–model HMOs, and staff/group–model HMOs. The advantage of the traditional FFS Medicare system—as discerned from the experiences of enrollees—is more personalized, patient-oriented, integrated primary care. The advantages of Medicare HMOs are lower out-of-pocket costs, minimal paperwork, and in some cases expanded benefit package (eg, drug coverage, eyeglasses, and dental care). The latter advantage has been eroding in the wake of the 1997 Balanced Budget Amendment because Medicare HMOs face declining margins with which to offer discretionary benefits. As the benefit packages of traditional FFS Medicare and Medicare HMOs converge, our results suggest that beneficiaries are increasingly left with a trade-off involving differences in cost and quality.

Accepted for publication August 2, 2001.

This research was supported by grant R01 HS09622 from the Agency for Healthcare Research and Quality, Rockville, Md, and the National Institute on Aging, Bethesda, Md.

We gratefully acknowledge Renee Mentnech, Kim Elmo, and Russ Patterson at HCFA, whose commitment to this research and facilitation of our data requests have made the study possible. We are also indebted to Carlos Zarabozo (HCFA) and Joseph P. Newhouse, PhD (Harvard Medical School), who have been invaluable resources and offered keen insights throughout the project.

We also gratefully acknowledge Brian Clarridge, PhD (Center for Survey Research, University of Massachusetts), and Tim Nanneman, PhD (Roper Starch Worldwide), and their colleagues for their technical expertise and commitment to excellence in obtaining the data from Medicare enrollees, and Barbara Seltzer for her dedicated assistance in preparing the manuscript.

Corresponding author and reprints: Dana Gelb Safran, ScD, The Health Institute, 750 Washington St, Box 345, Boston, MA 02111 (e-mail: dsafran@lifespan.org).

Gillick  MR The impact of health maintenance organizations on geriatric care. Ann Intern Med. 1987;106139- 143
Link to Article
Iglehart  JK Health policy report: second thoughts about HMOs for Medicare patients. N Engl J Med. 1987;3161487- 1492
Link to Article
Bonanno  JBWetle  T HMO enrollment of Medicare recipients: an analysis of incentives and barriers. J Health Polit Policy Law. 1984;941- 62
Link to Article
Rubenstein  LZKane  RL Geriatric assessment of programs: their time has come. J Am Geriatr Soc. 1985;33646- 647
Hoffman  CRice  DSung  H-Y Persons with chronic conditions: their prevalence and costs. JAMA. 1996;2761473- 1479
Link to Article
Health Care Financing Administration (Office of Managed Care), Group Health Plan Master File, 1989-2000.  Washington, DC Health Care Financing Administration2000;
Institute of Medicine, Primary Care: America's Health in a New Era.  Washington, DC National Academy Press1996;
Institute of Medicine, Report of a Study: A Manpower Policy for Primary Health Care.  Washington, DC National Academy of Sciences1978;
Starfield  B Primary Care: Concept, Evaulation and Policy.  New York, NY Oxford University Press1992;
Millis  JS The Millis Commission Report.  Chicago, Ill American Medical Association1966;
Alpert  JCharney  E The Education of Physicians for Primary Care.  Washington, DC US Dept of Health, Education, and Welfare1973;
Safran  DG Defining Primary Care: A Background Paper Prepared for the Insititute of Medicine Committee on the Future of Primary Care.  Boston, Mass New England Medical Center1994;
Rossiter  LFLangwell  KWan  TTRivnyak  M Patient satisfaction among elderly enrollees and disenrollees in Medicare health maintenance organizations: results from the National Medicare Competition Evaluation. JAMA. 1989;26257- 63
Link to Article
Retchin  SMBrown  B The quality of ambulatory care in Medicare health maintenance organizations. Am J Public Health. 1990;80411- 415
Link to Article
Clement  DGRetchin  SMBrown  RS Satisfaction with access and quality of care in Medicare risk contract HMOs. Luft  HSHMOs and the Elderly Ann Arbor, Mich Health Administration Press1994;
Clement  DGRetchin  SMBrown  RSStegall  MH Access and outcomes of elderly patients enrolled in managed care. JAMA. 1994;2711487- 1492
Link to Article
Safran  DGTarlov  ARRogers  WH Primary care performances in fee-for-service and prepaid health care systems: results from the Medical Outcomes Study. JAMA. 1994;2711579- 1586
Link to Article
Safran  DGRogers  WHTarlov  AR  et al.  Organizational and financial characteristics of health plans: are they related to primary care performance? Arch Intern Med. 2000;16069- 76
Link to Article
Luft  HS Assessing the evidence on HMO performance. Milbank Mem Fund Q Health Soc. 1980;58501- 536
Link to Article
Miller  RHLuft  HS Managed care plan performance since 1980: a literature analysis. JAMA. 1994;2711512- 1519
Link to Article
Langwell  KStearns  SNelson  J  et al.  Disenrollment Experience in the TEFRA/CMP Program.  Washington, DC Mathematica Policy Research Inc1989;
Morris  C A finite selection model for experimental design of The Health Insurance Study. J Econometr. 1979;1143- 61
Link to Article
Dillman  DA Mail and Telephone Surveys: The Total Design Method.  New York, NY John Wiley1978;
McHorney  CAWare  JEJRaczek  AE The MOS 36-Item Short-Form Health Survey (SF-36), II: psychometric and clinical tests of validity in measuring physical and mental health constructs. Med Care. 1993;31247- 263
Link to Article
McHorney  CAWare  JE  JrLu  JFSherbourne  CD The MOS 36-Item Short-Form Health Survey (SF-36), III: tests of data quality, scaling assumptions, and reliability across diverse patient groups. Med Care. 1994;3240- 66
Link to Article
Safran  DGKosinski  MTarlov  AR  et al.  The Primary Care Assessment Survey: tests of data quality and measurement performance. Med Care. 1998;36728- 739
Link to Article
Shea  SStein  ADLantigua  RBasch  CE Reliability of the behavioral risk factor survey in a triethnic population. Am J Epidemiol. 1991;133489- 500
Katz  JNChang  LCSangha  OFossel  AHBates  DW Can comorbidity be measured by questionnaire rather than medical record review? Med Care. 1996;3473- 84
Link to Article
Charlson  MEPompei  PAles  KLMacKenzie  CR A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40373- 383
Link to Article
Adams  PFMorano  MA Current Estimates From the National Health Interview Survey, 1994: Supplement on Aging. 193rd ed. Hyattsville, Md National Center for Health Statistics1995;Vital and Health Statistics No. 10.
Davies  ARWare  JEBrook  RHPeterson  JRNewhouse  JP Consumer acceptance of prepaid and fee-for-service medical care: results from a randomized controlled trial. Health Serv Res. 1986;21429- 452
Ware  JESnyder  MKWright  WR Results regarding scales constructed from the patient satisfaction questionnaire and measure of other health perceptions. Development and Validation of Scales to Measure Patient Satisfaction With Health Care Services Springfield, Va National Technical Information Service1976;NTIS publication PB:288-330.
US Congress Office of Technology Assessment. Washington, DC: office of Technology Assessment, The Quality of Medical Care Information for Consumers: WARE. 1988;231- 243
Sox  HCMargulies  ISox  CH Psychologically mediated effects of diagnostic tests. Ann Intern Med. 1981;95680- 685
Link to Article
Murray  ASafran  DG The Primary Care Assessment Survey: a tool for measuring, monitoring, and improving primary care. Maruish  MEHandbook of Psychological Assessment in Primary Care Settings Mahwah, NJ Lawrence Erlbaum Associates Inc2000;623- 651
Department of Health and Human Services, National Center for Health Statistics, Centers for Disease Control and Prevention, Health United States, 1992 and Health People, 2000.  Hyattsville, Md US Dept of Health and Human Services1993;
Green  WH Econometric Analysis.  New York, NY MacMillan Publishing Co2001;
Taira  DASafran  DGSeto  TBRogers  WHTarlov  AR The relationship between patient income and physician discussion of health risk behaviors. JAMA. 1997;2781412- 1417
Link to Article
Murphy  JChang  HMontgomery  JRogers  WHSafran  DG The quality of physician-patient relationships: patients' experiences 1996-1999. J Fam Pract. 2001;50123- 129
Safran  DGMontgomery  JEChang  HMurphy  JRogers  WH Switching doctors: predictors of voluntary disenrollment from a primary physician's practice. J Fam Pract. 2001;50130- 136
Ware  JEKosinksi  MKeller  SK SF-36 Physical and Mental Health Summary Scales: A User's Manual.  Boston, Mass Health Institute1994;
Cooper  JKHaffer  SCMichael  JAKohlmann  TStevic  M Measuring Medicare health outcomes: methodology, baseline data and issues. Int J Qual Health Care. 2001;139- 16
Link to Article
Francis  AMPolissar  LLorenz  AB Care of patients with colorectal cancer: a comparison of a health maintenance organization and fee-for-service practices. Med Care. 1984;22418- 429
Link to Article
Murray  JP A follow-up comparison of patient satisfaction among prepaid and fee-for-service patients. J Fam Pract. 1988;26576- 581
Rubin  HRGandek  BRogers  WHKosinski  MMcHorney  CAWare  JEJ Patients' ratings of outpatient visits in different practice settings: results from the Medical Outcomes Study. JAMA. 1993;270835- 840
Link to Article
Mechanic  D The organization of medical practice and practice orientations among physicians in prepaid and nonprepaid primary care settings. Med Care. 1975;13189- 204
Link to Article
Wolinsky  FDMarder  WD Waiting to see the doctor: the impact of organizational structure on medical practice. Med Care. 1983;21531- 542
Link to Article
Mechanic  DWeiss  NCleary  PD The growth of HMOs: issues of enrollment and disenrollment. Med Care. 1983;21338- 347
Link to Article
Wolinsky  FDMarder  WD Spending time with patients: the impact of organizational structure on medical practice. Med Care. 1982;201051- 1059
Link to Article
Roulidis  ZCSchulman  KA Physician communication in managed care organizations: opinions of primary care physicians. J Fam Pract. 1994;39446- 451
Safran  DGMurray  AChang  HMontgomery  JEMurphy  JRogers  WH Linking doctor-patient relationship quality to outcomes. J Gen Intern Med. 2000;15 (suppl) 116
Link to Article
Safran  DGTaira  DARogers  WHKosinski  MWare  JETarlov  AR Linking primary care performance to outcomes of care. J Fam Pract. 1998;47213- 220
Watson  RGTham  TCJohnston  BTMcDougall  NI Double blind cross-over placebo controlled study of omeprazole in the treatment of patients with reflux symptoms and physiological levels of acid reflux: the "sensitive oesophagus." Gut. 1997;40587- 590
Mahajan  POkamoto  LJSchaberg  AKellerman  DSchoenwetter  WF Impact of fluticasone propionate powder on health-related quality of life in patients with moderate asthma. J Asthma. 1997;34227- 234
Link to Article
Jones  PWBosh  TK Quality of life changes in COPD patients treated with salmeterol. Am J Respir Crit Care Med. 1997;1551283- 1289
Link to Article
Murray  AMontgomery  JChang  HRogers  WHInui  TSafran  DG Doctor discontent: a comparison of physician satisfaction in different delivery system settings, 1986-1997. J Gen Intern Med. 2001;16452- 459
Link to Article

Figures

Tables

Table Graphic Jump LocationTable 1. Summary of Item Content for Primary Care Assessment Survey Scales
Table Graphic Jump LocationTable 2. Characteristics of the Analytic Sample*
Table Graphic Jump LocationTable 3. Mean Adjusted Primary Care Assessment Scores by Medicare Delivery System*
Table Graphic Jump LocationTable 4. Adjusted Comparison of Not-for-Profit and For-Profit Medicare HMOs*

References

Gillick  MR The impact of health maintenance organizations on geriatric care. Ann Intern Med. 1987;106139- 143
Link to Article
Iglehart  JK Health policy report: second thoughts about HMOs for Medicare patients. N Engl J Med. 1987;3161487- 1492
Link to Article
Bonanno  JBWetle  T HMO enrollment of Medicare recipients: an analysis of incentives and barriers. J Health Polit Policy Law. 1984;941- 62
Link to Article
Rubenstein  LZKane  RL Geriatric assessment of programs: their time has come. J Am Geriatr Soc. 1985;33646- 647
Hoffman  CRice  DSung  H-Y Persons with chronic conditions: their prevalence and costs. JAMA. 1996;2761473- 1479
Link to Article
Health Care Financing Administration (Office of Managed Care), Group Health Plan Master File, 1989-2000.  Washington, DC Health Care Financing Administration2000;
Institute of Medicine, Primary Care: America's Health in a New Era.  Washington, DC National Academy Press1996;
Institute of Medicine, Report of a Study: A Manpower Policy for Primary Health Care.  Washington, DC National Academy of Sciences1978;
Starfield  B Primary Care: Concept, Evaulation and Policy.  New York, NY Oxford University Press1992;
Millis  JS The Millis Commission Report.  Chicago, Ill American Medical Association1966;
Alpert  JCharney  E The Education of Physicians for Primary Care.  Washington, DC US Dept of Health, Education, and Welfare1973;
Safran  DG Defining Primary Care: A Background Paper Prepared for the Insititute of Medicine Committee on the Future of Primary Care.  Boston, Mass New England Medical Center1994;
Rossiter  LFLangwell  KWan  TTRivnyak  M Patient satisfaction among elderly enrollees and disenrollees in Medicare health maintenance organizations: results from the National Medicare Competition Evaluation. JAMA. 1989;26257- 63
Link to Article
Retchin  SMBrown  B The quality of ambulatory care in Medicare health maintenance organizations. Am J Public Health. 1990;80411- 415
Link to Article
Clement  DGRetchin  SMBrown  RS Satisfaction with access and quality of care in Medicare risk contract HMOs. Luft  HSHMOs and the Elderly Ann Arbor, Mich Health Administration Press1994;
Clement  DGRetchin  SMBrown  RSStegall  MH Access and outcomes of elderly patients enrolled in managed care. JAMA. 1994;2711487- 1492
Link to Article
Safran  DGTarlov  ARRogers  WH Primary care performances in fee-for-service and prepaid health care systems: results from the Medical Outcomes Study. JAMA. 1994;2711579- 1586
Link to Article
Safran  DGRogers  WHTarlov  AR  et al.  Organizational and financial characteristics of health plans: are they related to primary care performance? Arch Intern Med. 2000;16069- 76
Link to Article
Luft  HS Assessing the evidence on HMO performance. Milbank Mem Fund Q Health Soc. 1980;58501- 536
Link to Article
Miller  RHLuft  HS Managed care plan performance since 1980: a literature analysis. JAMA. 1994;2711512- 1519
Link to Article
Langwell  KStearns  SNelson  J  et al.  Disenrollment Experience in the TEFRA/CMP Program.  Washington, DC Mathematica Policy Research Inc1989;
Morris  C A finite selection model for experimental design of The Health Insurance Study. J Econometr. 1979;1143- 61
Link to Article
Dillman  DA Mail and Telephone Surveys: The Total Design Method.  New York, NY John Wiley1978;
McHorney  CAWare  JEJRaczek  AE The MOS 36-Item Short-Form Health Survey (SF-36), II: psychometric and clinical tests of validity in measuring physical and mental health constructs. Med Care. 1993;31247- 263
Link to Article
McHorney  CAWare  JE  JrLu  JFSherbourne  CD The MOS 36-Item Short-Form Health Survey (SF-36), III: tests of data quality, scaling assumptions, and reliability across diverse patient groups. Med Care. 1994;3240- 66
Link to Article
Safran  DGKosinski  MTarlov  AR  et al.  The Primary Care Assessment Survey: tests of data quality and measurement performance. Med Care. 1998;36728- 739
Link to Article
Shea  SStein  ADLantigua  RBasch  CE Reliability of the behavioral risk factor survey in a triethnic population. Am J Epidemiol. 1991;133489- 500
Katz  JNChang  LCSangha  OFossel  AHBates  DW Can comorbidity be measured by questionnaire rather than medical record review? Med Care. 1996;3473- 84
Link to Article
Charlson  MEPompei  PAles  KLMacKenzie  CR A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40373- 383
Link to Article
Adams  PFMorano  MA Current Estimates From the National Health Interview Survey, 1994: Supplement on Aging. 193rd ed. Hyattsville, Md National Center for Health Statistics1995;Vital and Health Statistics No. 10.
Davies  ARWare  JEBrook  RHPeterson  JRNewhouse  JP Consumer acceptance of prepaid and fee-for-service medical care: results from a randomized controlled trial. Health Serv Res. 1986;21429- 452
Ware  JESnyder  MKWright  WR Results regarding scales constructed from the patient satisfaction questionnaire and measure of other health perceptions. Development and Validation of Scales to Measure Patient Satisfaction With Health Care Services Springfield, Va National Technical Information Service1976;NTIS publication PB:288-330.
US Congress Office of Technology Assessment. Washington, DC: office of Technology Assessment, The Quality of Medical Care Information for Consumers: WARE. 1988;231- 243
Sox  HCMargulies  ISox  CH Psychologically mediated effects of diagnostic tests. Ann Intern Med. 1981;95680- 685
Link to Article
Murray  ASafran  DG The Primary Care Assessment Survey: a tool for measuring, monitoring, and improving primary care. Maruish  MEHandbook of Psychological Assessment in Primary Care Settings Mahwah, NJ Lawrence Erlbaum Associates Inc2000;623- 651
Department of Health and Human Services, National Center for Health Statistics, Centers for Disease Control and Prevention, Health United States, 1992 and Health People, 2000.  Hyattsville, Md US Dept of Health and Human Services1993;
Green  WH Econometric Analysis.  New York, NY MacMillan Publishing Co2001;
Taira  DASafran  DGSeto  TBRogers  WHTarlov  AR The relationship between patient income and physician discussion of health risk behaviors. JAMA. 1997;2781412- 1417
Link to Article
Murphy  JChang  HMontgomery  JRogers  WHSafran  DG The quality of physician-patient relationships: patients' experiences 1996-1999. J Fam Pract. 2001;50123- 129
Safran  DGMontgomery  JEChang  HMurphy  JRogers  WH Switching doctors: predictors of voluntary disenrollment from a primary physician's practice. J Fam Pract. 2001;50130- 136
Ware  JEKosinksi  MKeller  SK SF-36 Physical and Mental Health Summary Scales: A User's Manual.  Boston, Mass Health Institute1994;
Cooper  JKHaffer  SCMichael  JAKohlmann  TStevic  M Measuring Medicare health outcomes: methodology, baseline data and issues. Int J Qual Health Care. 2001;139- 16
Link to Article
Francis  AMPolissar  LLorenz  AB Care of patients with colorectal cancer: a comparison of a health maintenance organization and fee-for-service practices. Med Care. 1984;22418- 429
Link to Article
Murray  JP A follow-up comparison of patient satisfaction among prepaid and fee-for-service patients. J Fam Pract. 1988;26576- 581
Rubin  HRGandek  BRogers  WHKosinski  MMcHorney  CAWare  JEJ Patients' ratings of outpatient visits in different practice settings: results from the Medical Outcomes Study. JAMA. 1993;270835- 840
Link to Article
Mechanic  D The organization of medical practice and practice orientations among physicians in prepaid and nonprepaid primary care settings. Med Care. 1975;13189- 204
Link to Article
Wolinsky  FDMarder  WD Waiting to see the doctor: the impact of organizational structure on medical practice. Med Care. 1983;21531- 542
Link to Article
Mechanic  DWeiss  NCleary  PD The growth of HMOs: issues of enrollment and disenrollment. Med Care. 1983;21338- 347
Link to Article
Wolinsky  FDMarder  WD Spending time with patients: the impact of organizational structure on medical practice. Med Care. 1982;201051- 1059
Link to Article
Roulidis  ZCSchulman  KA Physician communication in managed care organizations: opinions of primary care physicians. J Fam Pract. 1994;39446- 451
Safran  DGMurray  AChang  HMontgomery  JEMurphy  JRogers  WH Linking doctor-patient relationship quality to outcomes. J Gen Intern Med. 2000;15 (suppl) 116
Link to Article
Safran  DGTaira  DARogers  WHKosinski  MWare  JETarlov  AR Linking primary care performance to outcomes of care. J Fam Pract. 1998;47213- 220
Watson  RGTham  TCJohnston  BTMcDougall  NI Double blind cross-over placebo controlled study of omeprazole in the treatment of patients with reflux symptoms and physiological levels of acid reflux: the "sensitive oesophagus." Gut. 1997;40587- 590
Mahajan  POkamoto  LJSchaberg  AKellerman  DSchoenwetter  WF Impact of fluticasone propionate powder on health-related quality of life in patients with moderate asthma. J Asthma. 1997;34227- 234
Link to Article
Jones  PWBosh  TK Quality of life changes in COPD patients treated with salmeterol. Am J Respir Crit Care Med. 1997;1551283- 1289
Link to Article
Murray  AMontgomery  JChang  HRogers  WHInui  TSafran  DG Doctor discontent: a comparison of physician satisfaction in different delivery system settings, 1986-1997. J Gen Intern Med. 2001;16452- 459
Link to Article

Correspondence

CME
Meets CME requirements for:
Browse CME for all U.S. States
Accreditation Information
The American Medical Association is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The AMA designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 CreditTM per course. Physicians should claim only the credit commensurate with the extent of their participation in the activity. Physicians who complete the CME course and score at least 80% correct on the quiz are eligible for AMA PRA Category 1 CreditTM.
Note: You must get at least of the answers correct to pass this quiz.
You have not filled in all the answers to complete this quiz
The following questions were not answered:
Sorry, you have unsuccessfully completed this CME quiz with a score of
The following questions were not answered correctly:
Commitment to Change (optional):
Indicate what change(s) you will implement in your practice, if any, based on this CME course.
Your quiz results:
The filled radio buttons indicate your responses. The preferred responses are highlighted
For CME Course: A Proposed Model for Initial Assessment and Management of Acute Heart Failure Syndromes
Indicate what changes(s) you will implement in your practice, if any, based on this CME course.
Submit a Comment

Multimedia

Some tools below are only available to our subscribers or users with an online account.

Web of Science® Times Cited: 30

Related Content

Customize your page view by dragging & repositioning the boxes below.

Articles Related By Topic
Related Collections
PubMed Articles