The Response of Physician Groups to P4P Incentives

May 1, 2007

Objectives:

Objectives: Despite substantial enthusiasmamong insurers and federal policy makers forpay-for-performance incentives, little is knownabout the current scope of these incentives ortheir influence on the delivery of care. To assessthe scope and magnitude of pay-for-performance(P4P) incentives among physician groups and toexamine whether such incentives are associatedwith quality improvement initiatives.

Study Design: Structured telephone survey ofleaders of physician groups delivering primarycare in Massachusetts.

Assessed Methods: Prevalence of P4P incentivesamong physician groups tied to specific measuresof quality or utilization and prevalence ofphysician group quality improvement initiatives.

P

Results: Most group leaders (89%) reported P4Pincentives in at least 1 commercial health plancontract. Incentives were tied to performanceon Health Employer Data and Information Set(HEDIS) quality measures (89% of all groups),utilization measures (66%), use of informationtechnology (52%), and patient satisfaction (37%).Among the groups with P4P and knowledge ofall revenue streams, the incentives accountedfor 2.2% (range, 0.3%-8.8%) of revenue. P4Pincentives tied to HEDIS quality measures werepositively associated with groups' quality improvementinitiatives (odds ratio, 1.6; = .02). Thirty-sixpercent of group leaders with P4P incentivesreported that they were very important or moderatelyimportant to the group's financial success.

Conclusions: P4P incentives are now commonamong physician groups in Massachusetts, andthese incentives most commonly reward higherclinical quality or lower utilization of care.Although the scope and magnitude of incentivesare still modest for many groups, we found anassociation between P4P incentives and the useof quality improvement initiatives.

(Am J Manag Care. 2007;13:249-255)

Despite widespread implementation of pay-for-performance (P4P) incentives,little is known about physician groups' response to P4P incentives andwhether they foster quality improvement.

  • A majority of physician group leaders viewed P4P incentives favorably.
  • Only one third of leaders reportedly thought that their current P4Pincentives were financially important. To increase the emphasis on qualityimprovement, 91% of respondents believed that incentives of 5% or morewould be necessary.
  • P4P incentives were associated with a higher likelihood that a grouphad quality improvement initiatives targeting measures with incentives.

Commercial health plans are introducing pay-for-performance(P4P) incentives into physician contracts at an acceleratingpace.1-3 These incentives tie a portion of a physician's reimbursementto measures of clinical quality or other measures such as theuse of diagnostic imaging or the use of preferred medications. Proponentsexpect that these incentives will motivate providers to address documentedproblems with the quality of care and unexplained variations in the deliveryof care.4-6 The US Congress is considering inclusion of P4P incentives inMedicare's traditional fee-for-service payment system, and several Medicaredemonstration projects already include such incentives.7-9

Despite this enthusiasm, most previous evaluations of P4P incentiveprograms have shown little, if any, improvement in performance on qualitymeasures.10-18 An exception is a recent study19 that found that hospitalsengaged in a P4P program achieved modestly greater improvementin quality. Potential explanations for these generally negative prior resultsinclude the small magnitude of incentives from a single insuranceplan,12,20,21 resistance or indifference among physicians,22 or a lag timebetween the start of quality improvement initiatives and actual improvementsin quality.10,11,21 As this payment approach evolves, it is importantto understand how providers view P4P incentives and whether theybelieve the incentives motivate quality improvement.

Massachusetts is one of the few states where health plans have widelyimplemented performance incentives,23-25 as well as one of the only statesto have publicly available reports on quality for all physician groups. Assuch, the experience of the state's physician groups in this setting mayoffer valuable insights on how physician groups respond to P4P incentivesin a setting with public reporting on quality. We conducted a surveyof physician group leaders to characterize the scope and financial magnitudeof current P4P incentives and of leaders' views of these incentives asa mechanism to improve quality, as well as to test the hypothesis thatthere is an association between P4P incentives and the use of qualityimprovement initiatives.

METHODS

Sample of Physician Groups

Massachusetts Health Quality Partners, a nonprofit collaborative ofphysicians, health plans, consumers,and government agencies, maintainsa roster of physician groups (based onaggregated data from 5 major Massachusettshealth plans) to enable publicreporting on the clinical quality ofphysician groups. The roster includes all groups with 3 or morephysicians who provided outpatient primary care during theprior year. The 2005 roster listed 134 physician groups thatincluded 86% of 5804 primary care physicians in the state.

We made an initial contact with each of the groups on theMassachusetts Health Quality Partners roster to identify aleader eligible for the survey and to assess the independence ofthe group with respect to management. Some groups consideredindependent according to the Massachusetts HealthQuality Partners roster were in fact closely related and weremerged. For example, 17 entities on the Massachusetts HealthQuality Partners roster, although geographically separated,had neither a local medical director nor a local financialmanager and instead comprised a single large group. Afteraccounting for these arrangements, our sample consisted of104 autonomous physician groups. Four groups no longer inoperation during 2005 were excluded, producing the finalstudy sample of 100 groups. In our analyses, we account for thefact that some of these 100 groups affiliate with one anothersolely for contracting with commercial health plans, formingphysician networks.

Interview Guide

We developed a structured interview guide addressing thefollowing 5 areas: (1) the prevalence, scope, and magnitude ofP4P incentives in the group's health plan contracts, (2) thegroup's use of quality improvement initiatives, (3) the leader'sassessment of the financial importance and effect of P4P incentiveson the group, (4) the leader's general views on P4P,and (5) characteristics of the group.

The interview guide included a question about whether thephysician group had P4P incentives in its 2004 commercialhealth plan contracts. The question explicitly named the following4 categories of measures tied to P4P incentives: (1)Health Employer Data and Information Set (HEDIS) measures(eg, mammography), (2) patient survey measures (eg, satisfaction),(3) utilization measures (eg, the use of formulary medications),and (4) the use of information technology (eg, the use ofelectronic medical records and the use of electronic prescribing).The guide included questions about the percentage ofcommercial health plan revenue tied to these measures.Respondents were instructed to consider all types of P4P incentives,including bonuses and withholds, in their estimates.

Chlamydia

We selected 8 commonly used HEDIS measures of primarycare quality that might be the subject of a P4P incentive.These included measures of preventive care (percentage of eligiblepatients who received mammography, percentage thatreceived screening, and percentage that receivedappropriate well-child visits) and measures of chronic diseasemanagement (percentage of persons with diabetes mellitusreceiving appropriate glycosylated hemoglobin testing, percentageof persons with moderate-to-severe asthma prescribeda controller medication, percentage of patients screened forhyperlipidemia after an acute cardiac event, percentage ofpatients who had their low-density lipoprotein cholesterollevel appropriately controlled after an acute cardiac event,and percentage of patients with hypertension with blood pressureless than 140/90 mm Hg).

We defined 10 categories of quality improvement initiatives(Appendix available online at www.ajmc.com). Foreach HEDIS measure, all respondents were asked an openendedquestion to elicit whether initiatives were in placeto improve the group's performance on that measure. Foreach reported initiative, the interviewer probed for additionalinformation to enable classification into 1 of 10 categoriesof initiatives. Initiatives were counted only if theyinvolved concrete actions. For example, "the group's physiciansdiscuss improving performance at regular meetings"was not counted, but "an ongoing program to collect andanalyze performance data and give the results to physicians"was counted. Initiatives in the planning stage wereexcluded.

The interview guide also included questions about othergroup characteristics that might be associated with qualityimprovement initiatives such as the fraction of the group'sphysicians that were employees of the group (or in a partnershipagreement).26,27 The survey included a question aboutoperating margin (profit vs loss), but many group leaders wereunable to answer this question. In many groups, revenue issimply passed on to the member physicians, and there is nooverall group operating margin. We refined the interviewguide based on feedback from local physician leaders whowere not part of our sample.

Interview Guide Administration

Initial telephone calls yielded the contact information ofthe leader most likely to have detailed knowledge of the topicsof the interview. This person was a medical director (75%of groups), a nonphysician executive (19%), or an office manager(6%). One of us (AM) used the interview guide to conducta 30-to 60-minute telephone interview with each leader.A trained project assistant listened to each interview andrecorded detailed notes. All interviews were conductedbetween May and September 2005.

Statistical Analysis

For each of 8 HEDIS quality measures, we calculated separatelythe percentage of groups reporting P4P incentivestied to a measure and the percentagereporting quality improvement initiativestargeting a measure. To testthe association between P4P incentivesand quality improvement initiatives,we specified a single logisticregression that encompassed the 8HEDIS measures in our study. Theunit of analysis was a quality measurewithin a group. Each physician groupcould contribute up to 8 observations.The dependent variable waswhether or not the respondentreported a quality improvement initiativetargeting the given measure.The independent variable waswhether or not the respondentreported a P4P incentive tied to thegiven measure.

To address the within-group correlationamong the quality improvementinitiatives, we used a generalizedlinear mixed model assuming anunstructured covariance matrix.28 Toaddress the varying frequency of qualityimprovement initiatives across thedifferent measures (eg, mammographyinitiatives were more common thanhypertension initiatives), the modelincluded a unique intercept for eachmeasure. The model included thefollowing other factors that might influence the group'slikelihood of undertaking quality improvement initiatives:the number of primary care physicians (above vs belowthe median of 39 physicians), the proportion of physiciansemployed by the group (majority vs less than majority),affiliation with a physician network, whether themajority of the group's physicians were specialists,whether the majority of the group's physicians used anelectronic medical record, and whether greater than 25%of the group's commercial health plan revenue was in theform of capitation.

In analyses of the HEDIS measures, we excluded a group ifthe measure was irrelevant (eg, pediatric groups were excludedfrom analyses involving the mammography measure).Generalized linear mixed-model results were estimated usingProc Glimmix in SAS version 9.2 software (SAS Institute,Inc, Cary, NC). Items lacking a response were excluded fromthe analyses.

RESULTS

We completed interviews with leaders of 79 of 100 physiciangroups (Table 1). Among participating physician groups,77% provided both adult and pediatric primary care, 47%were composed of all or mostly primary care physicians, and53% were affiliated with a physician network. In 15% ofgroups, the majority of physicians used an electronic medicalrecord.

Prevalence and Focus of P4P Incentives

During 2004, 89% of the groups had a P4P incentive in atleast 1 commercial health plan contract (Table 2). HEDISmeasures were the most frequently reported target of the P4Pincentives, with the same 89% of groups reporting 1 or moreincentives tied to HEDIS measures. In the other categories,66% of groups reported incentives tied to utilization measures,52% reported incentives tied to the use of information technology, and 37% reported incentives tied to the group's performanceon patient satisfaction surveys. Only 1 group reportedan incentive tied to a measure outside of these 4 categories.Among the groups with P4P incentives, 54% of leaders wereable to track revenue from noncommercial health plansources such as Medicare or Medicaid. For this subset ofgroups, the mean percentage of total revenue tied to P4Pincentives was 2.2% (range, 0.3%-8.8%). Among all thegroups with P4P incentives, 81% had received at least half oftheir incentives in the previous year, and 42% had received atleast three quarters of their incentives.

Association Between P4P Incentives andQuality Improvement Initiatives

P

P

P

P

Among all groups, 72% reported at least 1 ongoing qualityimprovement initiative focused on a HEDIS measure (Table 2).The percentage of groups reporting initiatives targeting a measureranged from 12% (initiatives to improve hypertension control)to 61% (initiatives to increase glycosylated hemoglobinmeasurement). The most common type of quality improvementinitiative was the development of an internal registry and feedbacksystem for physicians about their performance on a givenmeasure. Adjusting for other group characteristics, we foundan independent positive associationbetween P4P incentivestied to a HEDIS measureand quality improvement initiativestargeting that measure(odds ratio [OR], 1.6; =.02) (Table 3). Other groupcharacteristics independentlyassociated with quality improvementinitiatives includedhaving a majority of physiciansemployed by the group (OR,3.9; = .002), affiliation with aphysician network of physiciangroups (OR, 2.4; =.03), and having more thanthe median number of physiciansin the group (OR, 2.6; = .03).

Reported Impact of P4PIncentives on the Group

P

P

Among the groups reportingP4P incentives, 36% ofleaders reported that the incentiveswere moderately orvery important to the group'soverall financial performance, and 56% reported that theincentives had a moderate or significant impact on the group.Compared with those with less than 1% of overall revenue tiedto incentives, group leaders with more than 3% were morelikely to report that the incentives were financially important(56% vs 11%, = .01 for test of trend) and that incentives hadan impact on the group (89% vs 37%, = .02 for test of trend).

Leaders' Views of P4P Incentives

Overall, 77% of physician group leaders expressed supportfor the notion of paying physician groups based on their performanceon HEDIS measures, and 79% reported that P4Pincentives would lead to quality improvement during the next3 years. Ninety-one percent reported that the ideal percentageof revenue tied to P4P incentives should be 5% or more. Only12% reported that their group's physicians perceive poor qualityof care as a major issue in healthcare.

DISCUSSION

P4P incentives are increasingly viewed as a cornerstoneof efforts to improve quality, despite limited evidence oftheir effectiveness.29 Few prior studies22 have assessed theviews of physician group leadersabout P4P incentive programs,and, to our knowledge,ours is the first study to examinethe relationship betweenP4P incentives and quality improvementinitiatives. In 2005in Massachusetts, where 89% ofphysician groups faced at least1 P4P incentive in their commercialhealth plan contracts,incentive programs accountedfor approximately 2% of totalrevenue, and we found that P4Pincentives were associated witha higher likelihood that a grouphad quality improvement initiativestargeting measures tied toincentives.

Our results shed light on theopen question of whether currentP4P incentives are large enough to motivate qualityimprovement.30 Only one third of the group leaders viewedincentives of the magnitude observed in our study as importantto their group's overall financial status, suggesting thatlarger incentives may be necessary to engage more groups.Most physician group leaders reported that incentives of 5%or more would be necessary to increase the emphasis on qualityimprovement. Nevertheless, the magnitude of incentives isonly one of many factors influencing the decision to undertakequality improvement. Other factors include the perceivedclinical importance of the quality measure, the costsand effectiveness of an available quality improvement initiative,the group's operating margin, the fraction of revenuefrom the payer with the incentive, and the structure of thegroup. Structure may be especially important. A large physiciangroup that employs its physicians and makes central useof the incentive dollars may make different decisions than asmaller group that divides and distributes incentive dollars toindependent individual practitioners. These factors mightexplain why the Medicare hospital P4P demonstration projectis the only P4P program that has demonstrated improvement.19 Hospitals may be better able than outpatient practicesto garner the resources for quality improvement. Also, hospitalsface one dominant payer, Medicare, while outpatientpractices face many different payers with potentially differentincentives.

P4P incentives are frequently described as a mechanism toimprove clinical quality,21,29 but they can also be applied toreign in the escalating cost of care.31 We found that P4Pincentives are applied for both purposes. A majority of thephysician groups reported incentives tied to both clinicalquality measures and utilization measures targeting reductionof expensive care such as emergency department visits.Anecdotally, some leaders expressed concern that healthplans' inclusion of utilization measures revealed the primarygoal of P4P programs to be monetary savings rather than qualityimprovement. The 2 goals could work at cross purposes.Given limited capacity for change, groups might choose tofocus on utilization reduction initiatives rather than on qualityimprovement. Of note, the multivariate model suggests anegative (but not statistically significant) association betweenP4P incentives on utilization and the likelihood of reporting aquality improvement initiative (OR, 0.6; 95% confidenceinterval, 0.2-1.3).

Our findings highlight the importance of physician grouporganizational characteristics in understanding how incentiveprograms and quality improvement can be optimized. Likeothers, we found that larger groups were more likely to undertakequality improvement initiatives,32 as were groups thatemploy a majority of their physicians and those affiliated witha network of physician groups. These findings support thecontention that the integration of primary care physiciansinto larger groups may enable them to undertake more qualityimprovement activities.33

The views of physician group leaders and practicing physiciansmay not be aligned. The majority of leaders in our studyreported that the group's physicians do not view quality as amajor problem in healthcare. This result is consistent withprior research suggesting that practicing clinicians remainskeptical of the national quality of care agenda outlined by theInstitute of Medicine.22,34,35 In the face of resistance by practicingclinicians, physician group leaders may look to P4Pincentives as a way to engage practicing physicians in effortsto improve quality.

Some commentators worry that P4P incentives may haveadverse consequences such as motivating physicians to avoidclinically complex or noncompliant patients or divertingphysicians' attention away from important but unmeasuredmedical services.20 In the United Kingdom, there is evidencethat some family physicians responded to a national P4P programby inappropriately excluding patients from the measures.36 In our interviews, which included a probe aboutadverse and unintended consequences, physician group leadersdid not note these concerns. However, if P4P incentivesbecome more financially important, evidence of adverse consequencesmay surface.

Our study has some limitations. First, the study design doesnot allow us to assert that P4P incentives "caused" physiciangroups to undertake quality improvement initiatives. That amajority of leaders reported that P4P incentives have had aneffect on their group is compelling, but it is plausible thatphysician groups with existing quality improvement initiativesnegotiated to obtain P4P incentives relevant to thoseinitiatives. Second, we assessed the group's use of qualityimprovement initiatives, not the performance of groups, andquality improvement activities may not always improve thequality of care.37 Third, although Massachusetts physiciangroups are similar in size to groups across the United States,38our results may not generalize to states with fewer incentiveprograms or where other configurations of health plans andphysician groups predominate.

P4P incentives are an increasingly popular innovation.Private insurers commonly include these incentives in contractswith physicians, and the federal government ispoised to implement P4P incentives in the Medicareprogram. There is little doubt that P4P will play anincreasing role in physician payment. Massachusettsphysician groups have been at the forefront of thismovement. Our results suggest that physician groupleaders are open to an expansion of these incentiveprograms and that incentives may stimulate qualityimprovement initiatives. Whether these initiativeswill yield improvements is unclear. However, if theydo, then P4P incentive programs could be an importantstep toward a higher-quality healthcare system.

Acknowledgments

We thank Susan Burwick, BA, for her invaluable assistance and the physiciangroup leaders for their gracious contribution of time.

Author Affiliations:

From RAND Health, Santa Monica, Calif, andUniversity of Pittsburgh Medical School, Pittsburgh, Pa (AM); andDepartment of Ambulatory Care and Prevention, Harvard Medical School(SDP, KPK), Harvard Pilgrim Health Care (KLC), Massachusetts HealthQuality Partners (JAS, BR), and Department of Health Policy and Management,Harvard School of Public Health (ECS), Boston, Mass.

Funding Source:

This study was supported by the Robert Wood JohnsonFoundation Rewarding Results Initiative and by National Research ServiceAward 5 T32 HP11001-15. The funders had no role in the design, analyses, ordecision to submit the manuscript for publication.This study was presented at the Annual Meeting of the Society of GeneralInternal Medicine, April 27, 2006, Los Angeles, Calif, and at the AnnualResearch Meeting of Academy for Health Services Research and HealthPolicy, June 28, 2006, Seattle, Wash.

Correspondence Author:

Eric C. Schneider, MD, MSc, Department ofHealth Policy and Management, Harvard School of Public Health, KresgeBldg, Rm 406, 677 Huntington Ave, Boston, MA 02115. E-mail:eschneid@hsph.harvard.edu.

Author Disclosure:

The authors (AM, SDP, KLC, KPK, JAS, BR, ECS)report no relationship or financial interest with any entity that would pose aconflict of interest with the subject matter discussed in this manuscript.

Authorship Information:

Concept and design (AM, SDP, KLC, JAS, BR,ECS); acquisition of data (AM, JAS, BR, ECS); analysis and interpretation ofdata (AM, SDP, KLC, KPK, ECS); drafting of the manuscript (AM, JAS, BR,ECS); critical revision of the manuscript for important intellectual content(AM, SDP, KLC, KPK, JAS, BR, ECS); statistical analysis (AM, KLC, ECS);obtaining funding (KLC, JAS, BR, ECS); administrative, technical, or logicalsupport (JAS, BR); supervision (ECS).

Health Aff

(Millwood).

1. Rosenthal MB, Fernandopulle R, Song HR, Landon B. Paying forquality: providers' incentives for quality improvement. 2004;23:127-141.

2. Leapfrog Group. Incentive & reward compendium guide & glossary.Available at: http://ir.leapfroggroup.org/compendium/. AccessedOctober 27, 2005.

N Engl J Med.

3. Rosenthal MB, Landon BE, Normand SL, Frank RG, Epstein AM.Pay for performance in commercial HMOs. 2006;355:1895-1902.

4. McClellan M. Statement of The Honorable Mark McClellan, M.D.,Ph.D., Administrator, Centers for Medicare and Medicaid Services, U.S.Department of Health and Human Services. Testimony before theSubcommittee on Health of the House Committee on Ways andMeans. September 29, 2005. Available at: http://waysandmeans.house.gov/hearings.asp?formmode=view&id=3817. Accessed April 8, 2007.

N Engl J Med.

5. McGlynn EA, Asch SM, Adams J, et al. The quality of health caredelivered to adults in the United States. 2003;348:2635-2645.

Ann Intern Med.

6. Fisher ES,Wennberg DE, Stukel TA, Gottlieb DJ, Lucas FL, Pinder EL.The implications of regional variations in Medicare spending, part 2:health outcomes and satisfaction with care. 2003;138:288-298.

7. Miller ME. Pay for performance in Medicare, July 27, 2005. Statementof Mark E. Miller, Ph.D., Executive Director, Medicare PaymentAdvisory Commission, before the Committee on Finance, U.S. Senate.Available at: http://www.medpac.gov/publications/congressional_testimony/Testimony_P4P.pdf. Accessed April 8, 2007.

8. Centers for Medicare and Medicaid Services. Medicare begins performance-based payments for physician groups. January 31, 2005.Available at: http://www.cms.hhs.gov/apps/media/press/release.asp?Counter=1341. Accessed April 8, 2007.

N Engl J Med.

9. Iglehart JK. Linking compensation to quality—Medicare paymentsto physicians. 2005;353:870-872.

JAMA.

10. Rosenthal MB, Frank RG, Li Z, Epstein AM. Early experience withpay-for-performance: from concept to practice. 2005;294:1788-1793.

Pediatrics.

11. Hillman AL, Ripley K, Goldfarb N,Weiner J, Nuamah I, Lusk E.Theuse of physician financial incentives and feedback to improve pediatricpreventive care in Medicaid managed care. 1999;104(pt 1):931-935.

Am J Public Health.

12. Hillman AL, Ripley K, Goldfarb N, Nuamah I,Weiner J, Lusk E.Physician financial incentives and feedback: failure to increase cancerscreening in Medicaid managed care. 1998;88:1699-1701.

Am J Prev Med.

13. Kouides RW, Bennett NM, Lewis B, Cappuccio JD, Barker WH,LaForce FM; Primary-Care Physicians of Monroe County. Performancebasedphysician reimbursement and influenza immunization rates inthe elderly. 1998;14:89-95.

Am J Public Health.

14. Fairbrother G, Hanson KL, Friedman S, Butts GC.The impact ofphysician bonuses, enhanced fees, and feedback on childhood immunizationcoverage rates. 1999;89:171-175.

Ambul Pediatr.

15. Fairbrother G, Siegel MJ, Friedman S, Kory PD, Butts GC. Impactof financial incentives on documented immunization rates in the innercity: results of a randomized controlled trial. 2001;1:206-212.

Prev Med.

16. Roski J, Jeddeloh R, An L, et al. The impact of financial incentivesand a patient registry on preventive care quality: increasing provideradherence to evidence-based smoking cessation practice guidelines.2003;36:291-299.

Pediatrics.

17. Davidson SM, Manheim LM,Werner SM, Hohlen MM,YudkowskyBK, Fleming GV. Prepayment with office-based physicians in publiclyfunded programs: results from the Children's Medicaid Program.1992;89(pt 2):761-767.

J Am Pharm Assoc (Wash

18. Christensen DB, Holmes G, Fassett WE, et al. Influence of a financialincentive on cognitive services: CARE project design/implementation.). 1999;39:629-639.

N Engl J Med.

19. Lindenauer PK, Remus D, Roman S, et al. Public reporting and payfor performance in hospital quality improvement. 2007;356:486-496.

Med Care Res Rev.

20. Rosenthal MB, Frank RG. What is the empirical basis for paying forquality in health care? 2006;63:135-157.

JAMA.

21. Dudley RA. Pay-for-performance research: how to learn what cliniciansand policy makers need to know. 2005;294:1821-1823.

Am J Manag Care.

22. Rolnick SJ, Margolis KL, Fortman KK, Maciosek MV, Grimm RH Jr.How acceptable are financial incentives and written feedback forimproving hypertension control? Perspectives from physicians, clinicadministrators, and patients. 2002;8:441-447.

Health & Medicine Week.

23. HMO's physician incentive programs earn national recognition.February 16, 2004:731.

Health Care Market Stabilizes,

but Rising Costs and State Budget Woes Loom in Boston.

24. Hoadley JF, Draper DA, Kuo S, et al. Washington,DC: Center for Studying Health System Change; 2003. Communityreport 12.

Can Money

Buy Quality? Physician Response to Pay for Performance.

25. Bodenheimer T, May JH, Berenson RA, Coughlan J. Washington,DC: Center for Studying Health System Change; 2005.

Health Aff (Millwood).

26. Robinson JC, Casalino LP. Vertical integration and organizationalnetworks in health care. 1996;15:7-22.

Health Aff (Millwood).

27. Rittenhouse DR, Grumbach K, O'Neil EH, Dower C, Bindman A.Physician organization and care management in California: from cottageto Kaiser. 2004;23:51-62.

J Am Stat Assoc.

28. Breslow N, Clayton D. Approximate inference in generalized linearmixed models. 1993;88:9-25.

N Engl J Med.

29. Epstein AM, Lee TH, Hamel MB. Paying physicians for high-qualitycare. 2004;350:406-410.

Assessment of Pay-for-Performance Options for Medicare Physician Services: Final Report.

30. Sorbero ME, Damberg CL, Shaw R, et al. Washington, DC: US Dept of Health and Human Services; May 2006.RAND working paper series, prepared for the Assistant Secretary forPlanning and Evaluation.

31. Commonwealth Fund. The Commonwealth Fund Health CareOpinion Leaders Survey: assessing health care experts' views on healthcare costs. April 2005. Available at: http://www.cmwf.org/surveys/surveys_show.htm?doc_id=275633. Accessed April 8, 2007.

JAMA.

32. Casalino L, Gillies RR, Shortell SM, et al. External incentives,information technology, and organized processes to improve healthcare quality for patients with chronic diseases. 2003;289:434-441.

Health Aff (Millwood).

33. Enthoven AC,Tollen LA. Competition in health care: it takes systemsto pursue quality and efficiency. 2005;suppl Web exclusives:W5-420-33.

N Engl J Med.

34. Blendon RJ, DesRoches CM, Brodie M, et al. Views of practicingphysicians and the public on medical errors. 2002;347:1933-1940.

Health Aff (Millwood).

35. Audet AM, Doty MM, Shamasdin J, Schoenbaum SC. Measure,learn, and improve: physicians' involvement in quality improvement.2005;24:843-853.

N Engl J Med.

36. Doran T, Fullwood C, Gravelle H, et al. Pay-for-performance programsin family practices in the United Kingdom. 2006;355:375-384.

Health Aff (Millwood).

37. Shojania KG, Grimshaw JM. Evidence-based quality improvement:the state of the science. 2005;24:138-150.38. Smart DR. Medical Group Practices in the US: Chicago, Ill:American Medical Association; 2005.