Lecture 12 - 10/14/99

Prevention, Quality of Care, and Enrollee Satisfaction

• See next Miller-Luft article

• Concepts: Why access may decrease? (cross-subsidies)
• Special populations: chronically ill; poor; minorities; rural residents
• Some empirical evidence that HMOs attract relatively healthy individuals initially

Does Managed Care Lead to Better or Worse Quality of Care?

Robert Miller and Harold Luft

Satisfaction (Exhibit 2)

• Most better satisfaction in HMO than FFS including Sisk (1996)

Quality of Care (Exhibit 3)

• 20 studies; 24 observations
• Quality in HMO better and results statistically significant: 2/20 studies; 3/24 observations
• Quality in HMO worse and results statistically significant: 3/20 studies; 3/24 observations

Quality of Care Results Favorable to HMO Plans, Compared with FFS (Exhibit 4)

• Corresponds to the top half of observations column in Exhibit 3
• Includes chronically ill nonelderly and chronicaly ill nonpoor from Ware (1996), but results not significant

Interpreting and Generalizing the Results

• Evidence is quite limited
• Few studies, different methods used, some quite dated
• HMOs are heterogeneous, including in how they pay physicians, how clinical care is organized
• Focus on health plan performance rather than on market performance ("spillovers")

Are Expectations about HMO's Quality Too High?

• Perverse payment incentives: plans/providers face disincentive to enroll most vulnerable; if develop reputation for treating vulnerable well, get selected against
• Inadequate information

Slow change in development of clinical processes

Differences in 4-Year Health Outcomes: HMO v. FFS

John E. Ware et al.

Overview

• Issue: how do outcomes for chronically ill persons compare?
• Used data from Medical Outcomes Study
• 3 cities: Boston, Los Angeles, Chicago
• 5 conditions: hypertension, non-insulin-dependent diabetes, AMI, CHF, depression
• Focus on Medicare beneficiaries and persons near or below poverty line

Medical Outcomes Study

• Sampling of HMOs, participation rates
• Sample of 3,569 completed baseline assessment (1986). Randomly selected 2,235 for follow-up. 4-year follow-up obtained for 1,574 patients.
• Samples of persons interviewed: diagnosis; gender; age
• Each patient sent a baseline questionnaire by mail
• Used 36-item Short-Form Health Survey (SF-36)
• Changes in health status estimated two ways:

1.  baseline scores subtracted from 4-year follow-up scores with death assigned 0.
2.  Classified patients into

• same
• better
• worse (better than assign 0 for death)

• Deaths more common for CHF and over 65

Statistical Analysis

• Covariates used in estimation of regression adjusted health change scores: system of care; age at baseline; gender; race; poverty status; condition; comorbid conditions; initial physical or mental health; MOS design variables (e.g., study site); interactions (e.g., HMO and age 65+)
• Used multinomial logit for analysis of health change categories

Results: Table 3

• On average, physical health patients worsened; mental health patients improved for both continuous/discrete measures.
• Over 4 years, no difference in change in physical or mental status by HMO v. FFS
• Average adjusted physical decline > elderly than for nonelderly. For mental, nonelderly improved over time. No change for elderly.
• Both poverty and nonpoverty groups declined in physical health, but mental health only improved for those nonpoor.

Results: Table 4

• On physical outcomes, nonelderly fared better in HMOs than in FFS
• For elderly, physical outcomes better in FFS than in HMO
• On mental health, elderly did better in HMOs than in FFS
• For nonelderly, no statistically significant difference between HMO and FFS on mental

Results: Tables 5 & 6

• For a subgroup of poor patients who were ill at baseline, those in HMOs experienced average decline of –2.0 in physical health while those in FFS experienced 5.4 point increase.

Policy Implications

• Although outcomes in HMO about the same on average, patients who were elderly or poor were twice as likely to decline in health in HMO than in FFS.
• Poverty status rather than Medicaid better marker of risk of poor outcome in an HMO
• Caution about expanding HMOs for vulnerable groups and need for replication

Limitations

• Results may be out of date
• Not a randomized trial
• Only studied outcomes in 3 cities and only for 5 conditions
• Loss to followup (studied by not problem)
• Plan switching—20% switched but switching not disproportionate for elderly v. nonelderly