1. Pay-for-Performance in Safety Net Settings: New Evidence from the Agency for Healthcare Research and Quality (AHRQ) Gary Young, J.D., Ph.D., Bert White M.B.A., Karen Sautter, M.P.H., Jason Silver, Barbara Bokhour, Ph.D., Mark Meterko, Ph.D. Boston University School of Public Health and Department of Veterans Affairs, Health Services Research and Development Service National Healthcare Incentives Institute October 21, 2008 Financial support from the Agency for Healthcare Research and Quality

2. Definition of Safety-Net Provider Various criteria have been proposed: Service to high levels of Medicaid and uninsured patients. Service to high levels of Medicaid and uninsured patients. Public ownership Public ownership Rural setting Rural setting

3. Pay-for-Performance (P4P) in Safety-Net Settings* As of July 2006, 28 state Medicaid agencies were operating P4P programs. As of July 2006, 28 state Medicaid agencies were operating P4P programs. 15 Medicaid agencies plan to start P4P programs. 15 Medicaid agencies plan to start P4P programs. By 2011, there will be an estimated 82 P4P programs in 43 states. By 2011, there will be an estimated 82 P4P programs in 43 states. * Source: Center for Health Care Strategies

4. P4P in Safety-Net Settings: Theoretical Considerations Performance = Motivation + Skill Financial incentive as a motivator: external rewards vs. intrinsic motivation Financial incentive as a motivator: external rewards vs. intrinsic motivation Skills for improving quality: Learning vs. Performing; Resources for improving quality (added pressures from complexity of case mix, high need for care coordination) Skills for improving quality: Learning vs. Performing; Resources for improving quality (added pressures from complexity of case mix, high need for care coordination)

5. P4P: Will it Work? Recent evidence points to modest gains from P4P in terms of provider adherence. Selected Findings : Rosenthal et al. (2006) Relative increase of 3.6 percentage points for cervical cancer screening.Rosenthal et al. (2006) Relative increase of 3.6 percentage points for cervical cancer screening. Levin-Scherz et al. (2006) Relative increase of 2-19 percentage points for diabetes measures.Levin-Scherz et al. (2006) Relative increase of 2-19 percentage points for diabetes measures. Lindenauer et a. (2007) CMS Premier demonstration: Relative increase of 2.6 percentage points for AMI measures; 3.4 points for pneumonia measures; 4.1 points for heart failure measures.Lindenauer et a. (2007) CMS Premier demonstration: Relative increase of 2.6 percentage points for AMI measures; 3.4 points for pneumonia measures; 4.1 points for heart failure measures. Young et al. (2007)Absolute increase of 7 percentage points for diabetes measure (e.g., eye exam).Young et al. (2007)Absolute increase of 7 percentage points for diabetes measure (e.g., eye exam). Pearson et al. (2008) Relative increase of.04 to.07 percentage points for certain diabetes measures and well child visits. But relatively less improvement for other measures (e.g., Chlamydia screening).Pearson et al. (2008) Relative increase of.04 to.07 percentage points for certain diabetes measures and well child visits. But relatively less improvement for other measures (e.g., Chlamydia screening).

6. P4P and Safety-Net Providers: Existing Research Felt-Lisk et al. (2007) Absolute increase of 4 to 22 percentage point increase among 4 CA Medicaid plans (i.e., Local Initiative Rewarding Results) for well child visits (documentation-driven improvement). Felt-Lisk et al. (2007) Absolute increase of 4 to 22 percentage point increase among 4 CA Medicaid plans (i.e., Local Initiative Rewarding Results) for well child visits (documentation-driven improvement). Werner et al. (2008) Hospitals with high Medicaid caseloads (> or = 40%) exhibit relatively less improvement on Medicare Compare measures (e.g., aspirin at discharge) than hospitals with low Medicaid caseloads( or = 40%) exhibit relatively less improvement on Medicare Compare measures (e.g., aspirin at discharge) than hospitals with low Medicaid caseloads(< or =5% ). Goldman et al (2007) Survey of 37 executives at safety-net hospitals about public reporting and P4P. Major concerns: case mix, lack of resources, and socio-economic problems of patients (e.g., inability to speak English). Goldman et al (2007) Survey of 37 executives at safety-net hospitals about public reporting and P4P. Major concerns: case mix, lack of resources, and socio-economic problems of patients (e.g., inability to speak English).

7. AHRQ Research Three Key Questions What is the potential for pay-for-performance to improve quality in safety net settings? What is the potential for pay-for-performance to improve quality in safety net settings? Are there unique challenges to designing and implementing pay-for-performance in safety net settings? Are there unique challenges to designing and implementing pay-for-performance in safety net settings? Does applying pay-for-performance to safety net settings carry substantial risks for unintended consequences? Does applying pay-for-performance to safety net settings carry substantial risks for unintended consequences?

8. Study Setting Safety Net Setting A (SNSa) Safety Net Setting A (SNSa) Insurer-sponsored program (new)Insurer-sponsored program (new) Community health center as unit of accountabilityCommunity health center as unit of accountability Medicaid populationMedicaid population Safety Net Setting B (SNSb) Safety Net Setting B (SNSb) Provider-sponsored program (mature; full- risk contracts)Provider-sponsored program (mature; full- risk contracts) Individual physician as unit of accountabilityIndividual physician as unit of accountability Medicaid and uninsured populationMedicaid and uninsured population

9. Sources of Data Survey of physicians Survey of physicians -- SNSa (61/108; 56% response rate; 44% aware of incentive). --SNSb (89/141; 63% response rate; --SNSb (89/141; 63% response rate; 50% aware of incentive). Interviews with senior leaders Interviews with senior leaders Clinical performance data Clinical performance data -- Administrative data (SNSa) -- Chart reviews (SNSb)

10. Key Findings No definitive evidence of quality improvement in short term. No definitive evidence of quality improvement in short term. Higher adherence to clinical process correlated with better patient outcomes. Higher adherence to clinical process correlated with better patient outcomes. No evidence of unintended consequences based on survey, interviews, and clinical performance data. No evidence of unintended consequences based on survey, interviews, and clinical performance data. Physicians accepting of concept, but Physicians accepting of concept, but financial incentive not a direct motivator for quality. financial incentive not a direct motivator for quality. Achievement of pay-for-performance program goals complicated by socio-economic status of patients. Achievement of pay-for-performance program goals complicated by socio-economic status of patients. Financial incentives for quality can be undercut by larger incentives for productivity. Financial incentives for quality can be undercut by larger incentives for productivity.

11. Key Findings No definitive evidence of quality improvement in short term. No definitive evidence of quality improvement in short term.

12. Safety Net Setting A - Performance Measures adherence scores, 13 community health centers *Diabetic Eye Exam no incentive in 2007 **** From admin data vs. National Medicaid from hybrid data (except Asthma) **HbA1c incentive in 2007 ***No Data Available Safety Net Setting A (SNSa) Health Measure**** 200520062007 IncentivizedSNSa National Medicaid SNSa SNSaMedicaid*** Asthma87%86%92%87%88% Well Child 29%70%35%73%42% Diabetic Eye Exam * 48%49%56%51%49% HbA1c Testing ** 84%76%87%71%87% Non-Incentivized Adolescent Well Child 46%42%50%51%49% LDL Screening 82%81%84%81%76% Nephropathy54%49%58%75%81%

13. Key Findings Higher adherence to clinical process correlated with better patient outcomes (based on sample of 51 physicians at Safety Net Setting B from 2002 to 2006). Higher adherence to clinical process correlated with better patient outcomes (based on sample of 51 physicians at Safety Net Setting B from 2002 to 2006). HbA1c:.33 --.69HbA1c:.33 --.69 LDL:.22 --.47LDL:.22 --.47

14. Key Findings No evidence of unintended consequences based on survey, interviews and clinical performance data. No evidence of unintended consequences based on survey, interviews and clinical performance data.

15. Physician Survey Data Scale Score (Min=1/Max=5) Physician Perceptions of P4Q In Three Healthcare Settings 2.91 2.37 4.29 4.11 3.33 2.67 1.98 2.24 2.46 2.89 2.79 3.11 2.61 3.76 3.86 2.57 2.94 2.72 3.62 3.51 2.84 1 1.5 2 2.5 3 3.5 4 4.5 5 AwarenessFinancial SalienceClinical RelevanceNo Unintended Consequences ControlCooperationImpact SNSb (n=45)SNSa (n=27) Comm. Setting (n=234) Scale Score (Min=1/Max=5)

16. Key Findings Physicians accepting of concept, but Physicians accepting of concept, but financial incentive not a direct motivator for quality. financial incentive not a direct motivator for quality. “…just pay us appropriately to begin with. Why should you have to incentivize a doctor for quality if you pay them enough.”

17. Key Findings Achievement of pay-for-performance program goals complicated by socio- economic status of patients. Achievement of pay-for-performance program goals complicated by socio- economic status of patients. “Many of these people we care for, part of their economic, social and psychological experience is that they lack value…So that the whole process of communicating to a person that they are a human being of value, part of that occurs in the communication between a physician and that of a patient.”

18. Key Findings Financial incentives for quality can be undercut by larger incentives for productivity. Financial incentives for quality can be undercut by larger incentives for productivity. “You feel like sometimes you’re running an assembly line…there is an inherent conflict between time and quality.”

19. Conclusions and Directions for Future Research While our research suggests that P4P is not necessarily antithetical to the values of safety-net providers, the effectiveness of the concept for improving quality in such settings is not very apparent. While our research suggests that P4P is not necessarily antithetical to the values of safety-net providers, the effectiveness of the concept for improving quality in such settings is not very apparent. In designing P4P programs to improve quality, careful consideration must be given to other incentive and compensation arrangements that may conflict or undermine quality-related incentives. In designing P4P programs to improve quality, careful consideration must be given to other incentive and compensation arrangements that may conflict or undermine quality-related incentives. As our investigation consisted of two case studies, research is needed to test the validity of the findings in a large sample of safety-net providers. As our investigation consisted of two case studies, research is needed to test the validity of the findings in a large sample of safety-net providers.