This article has Open Peer Review reports available.
Design, rationale, and baseline characteristics of a cluster randomized controlled trial of pay for performance for hypertension treatment: study protocol
© Petersen et al; licensee BioMed Central Ltd. 2011
Received: 29 July 2011
Accepted: 3 October 2011
Published: 3 October 2011
Despite compelling evidence of the benefits of treatment and well-accepted guidelines for treatment, hypertension is controlled in less than one-half of United States citizens.
This randomized controlled trial tests whether explicit financial incentives promote the translation of guideline-recommended care for hypertension into clinical practice and improve blood pressure (BP) control in the primary care setting. Using constrained randomization, we assigned 12 Veterans Affairs hospital outpatient clinics to four study arms: physician-level incentive; group-level incentive; combination of physician and group incentives; and no incentives (control). All participants at the hospital (cluster) were assigned to the same study arm. We enrolled 83 full-time primary care physicians and 42 non-physician personnel. The intervention consisted of an educational session about guideline-recommended care for hypertension, five audit and feedback reports, and five disbursements of incentive payments. Incentive payments rewarded participants for chart-documented use of guideline-recommended antihypertensive medications, BP control, and appropriate responses to uncontrolled BP during a prior four-month performance period over the 20-month intervention. To identify potential unintended consequences of the incentives, the study team interviewed study participants, as well as non-participant primary care personnel and leadership at study sites. Chart reviews included data collection on quality measures not related to hypertension. To evaluate the persistence of the effect of the incentives, the study design includes a washout period.
We briefly describe the rationale for the interventions being studied, as well as the major design choices. Rigorous research designs such as the one described here are necessary to determine whether performance-based payment arrangements such as financial incentives result in meaningful quality improvements.
Despite compelling evidence of the benefits of treatment, hypertension is controlled in less than one-half of United States (US) citizens with the disease . Inadequate blood pressure (BP) control results in excess morbidity and mortality from cardiac, renal, and peripheral vascular disease . While some cases of poor BP control relate to patients' compliance with treatment, there is significant under-treatment of hypertension on the part of physicians. In one study, hypertension patients received less than 65% of care that was indicated for this condition .
Under-treatment of hypertension is puzzling, because good evidence exists about the efficacy of antihypertensive medications. Despite widespread dissemination of guidelines for treatment of hypertension (including the Seventh Report of the Joint National Committee (JNC 7) on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure), and widespread comparative effectiveness trials such as the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) , translation into clinical practice has been incomplete .
Awareness of barriers to such clinical research translation has raised enthusiasm about using novel methods, such as financial incentives, to overcome them . A number of pay-for-performance programs have been implemented in the United Kingdom (UK) and the US [7, 8]. However, research evidence of effectiveness of pay-for-performance programs, particularly randomized trials, is limited [9, 10]. Using a cluster randomized controlled trial (RCT), we are testing the effect of explicit financial incentives to promote the translation of guideline-recommend care for hypertension into clinical practice and thereby improve BP control in the primary care setting. This trial addresses the needs of policy makers, payers, physicians, administrators, and others for information about a clinically relevant intervention in routine practice. We are not aware of other ongoing randomized trials of pay for performance directed at physicians and provider groups .
Design of trial
Because our goal was to evaluate the impact of financial incentives on individual physicians as well as primary care provider groups, we implemented a cluster randomized controlled design and clustered at the facility level .
Design of incentives
Should individuals, groups, or some combination receive financial incentives for their performance? One could anticipate that with group- or practice-team-level incentives, individual physicians would not capture the full returns on their individual effort to improve the quality of their care. Traditional economic theory suggests that the potential for some physicians to 'free-ride' on the efforts of others may lead many to reduce their individual efforts [10, 13]. Conversely, the problem with rewarding individuals, but not the individual's team or group, is that cooperation among the group is not necessarily encouraged. Moreover, group incentives may be important in supporting and rewarding infrastructure improvements to the healthcare delivery system. Studies evaluating the chronic care model suggest that multi-disciplinary teams produce better patient outcomes [14, 15]. Yet, traditional fee-for-service payment models create disincentives for making organizational changes in care delivery [15, 16]. Thus, theory suggests the potential for group-level incentives to support organizational and team-based efforts to improve quality, but it is not yet known how these effects compare to individual-level incentives.
The design of incentives for quality also raises the question, what should we reward? Would incentives targeting processes (i.e., what clinicians do) or outcomes (i.e., what ultimately happens to patients) be expected to produce the highest quality? Of course, the best process-of-care measures are those for which there is evidence that better performance leads to better outcomes. But it is important to note that process-of-care measures may be more sensitive to quality differences than are measures of outcomes, because a poor outcome does not necessarily occur every time there is a quality problem. Furthermore, outcomes are often dependent upon events that are outside of the control of the clinician. For example, one can prescribe guideline-recommended medications, but if the patient is not able to adhere to these medications, the BP may not be controlled. One theoretical problem with using incentives solely based upon process-of-care measures is that physicians may attempt 'gaming' and focus solely on the measure (i.e., use of a guideline-recommended medication), while ignoring the intended outcome, which is the attainment of a target BP.
To avoid the problems associated with the exclusive use of process-of-care or outcome measures described above and to dampen the gaming incentive, both a process-of-care measure (use of guideline-recommended medications in patients who have no other compelling indications) and an intermediate outcome measure (achievement of the JNC 7 BP goal) will be used. This approach may avoid the pitfalls of process-of-care measures alone that can encourage gaming, while avoiding the disadvantage of basing incentives solely on outcomes that may be relatively rare or difficult to achieve and somewhat beyond the control of the provider. Thus, a combined approach potentially capitalizes upon the advantages and complementary nature of both types of quality-of-care measures . The design of our trial reflects the following logic: other things being equal, process-based incentives will create stronger incentives for improvement in process (which the physician can directly control); similarly, process-based incentives may also produce better outcomes (assuming that the processes being incentivized improve outcomes). We believe that combining outcome-based with process-based incentives has the potential to produce even greater quality improvement than process measures alone .
A third design question is whether the incentive payments should be distributed according to relative performance (i.e., the participant's percentile ranking compared to peers) or absolute performance (i.e., strictly according to performance relative to the same standard)? The former approach may be problematic because participants cannot directly control the behavior or outcomes of others on whom a relative comparison is based. The latter approach rewards behavior that the individual or group can directly control [10, 13]. Therefore, we designed our incentives to reward participants each time they provided guideline-recommended medications and/or their patient met the guideline-recommended BP threshold or, in the absence of controlled BP, the participant appropriately responded to uncontrolled BP .
In this article, we describe the methods used to assess the effects of physician- and group-level financial incentives on processes and outcomes of care for outpatients with hypertension.
The goals of this study were to: determine the effect of physician-level financial incentives on processes and outcomes of care for outpatients with hypertension; assess the impact of group-level financial incentives; ascertain whether there are additive effects of physician- plus group-level financial incentives; evaluate the persistence of the effect of financial incentives; and identify any unintended consequences of these explicit financial incentives.
This study is a four-arm cluster RCT of individual physician incentives, group-level financial incentives, and performance audit and feedback to improve the translation of guideline-recommended care for hypertension in the primary care setting. Participants at study sites randomized to one of the three intervention arms earned either physician-level, group-level, or physician- and group-level incentives (combined incentive arm). Both intervention and control arm (no incentives) participants received audit and feedback reports.
Study sites and constrained randomization
We partnered with five Veterans Affairs (VA) regional networks to implement the study. Within these networks, hospital-based VA outpatient clinics that could conduct human subjects' research and had eight or more eligible primary care physicians were potential study sites. Of the 22 potential study sites, only 12 met the requirements for study implementation: hospital director approval; Institutional Review Board (IRB) and VA Research and Development (R&D) approval; having a credentialed individual willing to serve as the site's PI; and having at least five consented physician participants at time of randomization. We stratified these 12 study sites on the following characteristics expected a priori to be associated with responses to financial incentives and the study outcomes of interest: teaching status, geographic location, participation in the ALLHAT study  (a very large trial of various medications used to treat hypertension in both VA and non-VA settings that included intensive education regarding hypertension control and use of evidence-based hypertension treatment at participating sites), and the degree of clinic geographic proximity within the primary care setting at each study site. We identified a hospital as a teaching facility if it was listed in the Association of American Medical College's (AAMC) Council of Teaching Hospitals (COTH) directory or if the American Medical Association's (AMA) Fellowship and Residency Electronic Interactive Database (FREIDA) listed the VA facility as having a 'major' affiliation with a medical school. To determine geographic region, we used US Census Bureau information to identify the corresponding Census Division for each site. Study site investigators provided clinic layout information, and we designated sites as being integrated if the layout was amenable to group cohesion (e.g., the primary care clinic offices were located on the same floor at the study site).
Study site characteristics
US Census Division
ALLHAT study site
Primary care geographic integration†
VA Boston HCS
VA Connecticut HCS
Charlie Norwood VAMC
Ralph H. Johnson VAMC
E. South Central
Aleda E. Lutz VAMC
E. North Central
John D. Dingell VAMC
E. North Central
G.V. (Sonny) Montgomery VAMC
E. South Central
Michael E. DeBakey VAMC
W. South Central
Oklahoma City VAMC
Oklahoma City, OK
W. South Central
W. North Central
Power and sample size
The anticipated effect sizes were based upon a systematic review of financial incentives in healthcare . In that review, the range of increases in performance using physician-level incentives was 5.9% to 25.3%, most of which were for preventive services such as immunizations. Because this study assesses treatment of a chronic disease requiring multiple types of interventions (i.e., medications, counseling, monitoring, lifestyle modification), rather than a single immunization, the increase in performance is anticipated to be somewhat less than those trials at the high end of this range.
The sample size calculation must take into account several factors. First, this is a cluster design, randomized at the facility level. Second, there is a measurement error associated with estimating each physician's proportion, because we can only sample a finite number of patients per physician. Finally, there must be a correction for multiple comparisons because we are testing two effects and an interaction. The sample size computation uses the non-central t-distribution and is based on a SAS® program from Donner and Klar . An iterative solution is required because the number of degrees of freedom depends on the sample size. The variance in measurements is the sum of the variance between hospitals, the remaining variation between physicians and the error variance of the measurement. Estimates of the variance between hospitals and the variance between physicians were obtained from pilot data. The error variance of measurement of the physician's proportion was calculated from the binomial distribution. Using these data, we estimated values of the intraclass correlation of 0.39 for appropriate medication and 0.14 for BP control.
We provide power calculations for the process and outcome measures. We calculated effect sizes for various values of the difference in percentage use of appropriate medication and BP control we could detect between the study arms with 80% power using a two-sided t-test with 95% significance. Greater increases in power result from increasing the number of clusters than by increasing the number of cases within clusters . However, it is much more difficult and costly to recruit more hospitals. We chose to use three hospitals per study arm with five physicians per hospital and 40 patient charts per physician. We determined that we could detect a difference of 17 percentage points between the mean proportions of appropriate medications in the arms, for an effect size of 1.59. Similarly, for BP control, we determined we could detect a difference of 15 percentage points between the mean proportions in the study arms for an effect size of 1.30.
We adjusted the sample size to account for anticipated physician participant attrition. Using VA physician workforce planning data, we estimated 5.8% would leave VA employment during the study. We adjusted our initial sample size, five physicians per site, to seven per site to account for attrition using the formula proposed by Lachin , defined as 1/(1-R)2 where R is the drop-out rate.
Baseline characteristics of study participants
Primary care physicians who worked at least 0.6 full-time equivalents (approximately three days per week related to clinical activities) or had a panel size of at least 500 patients were eligible to participate. Research assistants on the Houston coordinating center study team obtained informed consent from at least five eligible participants (see Power and sample size section) at each study site prior to randomization. We sought to avoid coercion by informing the individuals that participation was strictly voluntary, that their decision to participate would not impact their employment status, and that their supervisors and other hospital officials would not have access to their performance data. Following randomization of study sites to study arms, we continued to consent eligible physician participants as necessary to meet our recruitment goal of seven physicians per site (see Power and sample size section). We enrolled a total of 83 physicians from 12 VA hospitals. At two sites, more than seven eligible physicians consented for the study, so we randomly chose seven among those who consented.
Characteristics of physicians and non-physician primary care personnel enrolled at the start of the intervention
Primary care physicians (n = 83)
(n = 42)
Male, n (%)
Age at start of study, mean (SD), y
Race/ethnicity, n (%)
Board certified, n (%)
Primary specialty internal medicine, n (%)
Trained in a subspecialty/secondary specialty, n (%)
Years practicing since residency completion, mean (SD)
Proportion of professional time spent delivering patient care, mean (SD)
Role in patient care delivery, n (%)
Licensed practical nurse
Medical support assistant
Physician assistant or nurse practitioner
Registered nurse case manager or care coordinator
Between February and April 2008, we provided all participants with the current JNC 7 hypertension guidelines during live educational webinars presented by a cardiologist and an internist. At the sessions, participants learned their study arm assignment. Participants in the intervention arms received information about the magnitude and criteria for the incentives.
Audit and feedback
The financial incentive intervention phase of the study started in April 2008 and consisted of disbursements of incentive payments earned over each of five performance periods. For each performance period, trained chart abstractors extracted relevant data from the VA electronic medical records of 40 randomly selected patients in each physician's panel who met eligibility criteria. Only those patients with hypertension who had a face-to-face healthcare encounter where hypertension could be addressed during the performance interval were eligible for inclusion in the sampling frame. For the first performance period, we examined a four-month period prior to the start of the incentive intervention phase to assess participants' baseline performance. We anchored performance periods two to five from the start of the incentive intervention phase. Incentive payments arrived in participants' VA paychecks approximately every four months and typically followed the posting of the feedback report for that performance period. After each data collection period, we notified participants via e-mail the date of the paycheck in which the payment was to appear.
Incentive payment structure
The five participating VA Networks' contributions to the incentive fund totalled $250,000. Based on this amount, we then simulated study results using pilot data and estimated rates of improvement to determine the greatest per-outcome incentive that we could provide per physician. The results indicated a per-patient maximum amount of $18.20, one-half from use of guideline-recommended medications and one-half from either BP control or appropriate response to uncontrolled BP.
Intervention components and incentive payouts
Total estimated* incentive over entire study period
Per outcome per patient incentive award
Physician and group
No incentives (control)
We developed a data abstraction tool and electronically collected data on the physicians' patients from the VA Computerized Patient Records System (CPRS). Abstractors at the Houston coordinating center collected data on patient demographics, vital signs, diabetes mellitus, cardiovascular conditions, renal conditions, terminal illness, and other relevant co-morbidities, laboratory values and medications, including dosages, allergies, contraindications, and patient refusals for all study sites. Additionally, we collected lifestyle modifications recommended by the provider or other medical staff members and information on study outcomes addressed at follow-up visits within a given timeframe. We collected hemoglobin A1c (HbA1c) levels in patients with diabetes, low-density lipoprotein cholesterol (LDL-C) levels in patients with hyperlipidemia, and colorectal cancer screening procedures among eligible patients to assess unintended consequences, i.e., neglect of these conditions or missed screening opportunities, that may have resulted from the intervention's focus on hypertension care.
After a training session, chart abstractors were required to complete 20 practice charts with 95% agreement on key data elements with all other abstractors and the trainer. During each period of data collection, the abstraction manager spot-checked charts for accuracy in abstracting key variables, addressing problem areas with abstracters as necessary. At the end of each period of data collection, we randomly selected at least three charts from each site in order to evaluate abstracter agreement. Also, we blinded chart abstractors to the study's objectives and to study arm assignments to ensure impartiality in the data collection process.
Interviews - team planning, pay-for-performance perspectives, and unintended consequences of incentives
After participants received their second incentive payments and audit and feedback reports, the study team, including an industrial/organizational psychologist, interviewed 17 physician participants and 11 non-physician participants across the study sites to collect data about the strategies that they used to improve hypertension care, including clinic planning and team interaction shared mental models. Twelve physician participants and four non-physicians also completed a similar interview after the fourth disbursement of incentive payments and audit and feedback reports. These distinct interviews were intended to provide insights on how care strategies and planning may have evolved over time, and the drivers behind any observed changes.
Following the posting of the participants' final audit and feedback reports in April 2010, we interviewed 28 physicians and 10 non-physician participants across the study sites to ascertain their perspectives on financial incentives, the intervention's impact on team dynamics and care delivery for hypertension and non-related conditions, and organizational changes that occurred at the study site during the intervention period. Focusing on similar concept domains, we also interviewed primary care personnel, 24 physicians and 25 non-physicians across the sites, who did not participate in the intervention and primary care leadership at the 12 sites.
Washout period and post-washout period data collection
The unit of analysis will be the physician. We will perform a repeated measures longitudinal analysis using mixed models to evaluate the effect of the intervention. We will evaluate three different predictors for each outcome: each incentive arm versus the control arm; individual-level incentive arms versus arms with no individual incentives; and group-level incentive arms versus arms with no group incentives. Models will be developed independently for each outcome. First, using scientifically relevant covariates selected a priori, we will construct a maximal model for each outcome as described by Cheng et al. . The maximal model provides the flexibility to evaluate both the covariance structure and the list of covariates for inclusion in the final model. The relationship between each continuous covariate and the outcome will be explored to assess the need for covariate transformations. We also will determine whether any site-to-site variation exists and include the facility (cluster) as a random effect as necessary. We then will perform backward elimination to delete variables of no value, arriving at our final model.
In this paper, we describe our rationale, methods, and baseline participant characteristics for a cluster-randomized trial to assess the effectiveness of pay for performance in improving hypertension control and use of guideline-recommended medications in the primary care setting. The potential impact of this study is great. Seventy-two million Americans have hypertension , but hypertension is controlled in less than one-half of those who carry this diagnosis . A variety of methods, such as audit and feedback, academic detailing, reminders, guidelines, and combinations of interventions have not succeeded in eliminating the under-treatment of hypertension . Given the track record of other interventions, the effectiveness and cost-effectiveness of financial incentives in overcoming such barriers to translation of research into practice must be rigorously evaluated .
Other studies of the use of financial incentives for improving healthcare quality have suffered from a dilution of the incentive due to multiple payers or competing incentives that make it difficult to assess effectiveness [9, 26]. Relative to these situations, studying incentives in the VA healthcare system provides several advantages. First, the VA is both the insurer and the provider, so organizational responses to incentives are easier to anticipate and assess. Second, the VA uses a single payment approach, rather than a diverse mix as seen elsewhere. The VA employs salaried physicians to care for its enrollees. The VA healthcare system consists of 21 networks that operate on a global budget appropriated by Congress. Funding is distributed to the networks via a form of capitation in which payments are made per veteran meeting eligibility criteria . The VA has a common national electronic medical information system, making the collection of common practice data across widely disparate geographic sites and types of facilities feasible. Thus, carrying out the study in the VA allows an evaluation of the effectiveness of financial incentives that is free from some of the limitations encountered in other settings.
Paying more for healthcare services shown to improve quality could have a tremendous impact on care delivery. However, despite great potential, numerous questions are unanswered. How effective (and cost-effective) are financial incentives for quality? Can we expect the effect of financial incentives to persist after they are stopped? Will important patient care activities that are not rewarded financially be neglected? Thus, despite enthusiasm about the potential for aligning financial incentives with high quality healthcare, there are a number of fundamental unanswered questions about their optimal design, effectiveness, and implementation that we will address in this trial .
While our study design has numerous strengths, we must acknowledge some limitations. First, as part of the VA Healthcare Personnel Enhancement Act of 2004 (implemented in 2006) , the VA healthcare system instituted a new payment system that includes performance pay based on the accomplishment of specific clinical quality goals and objectives which may be established at the local, network, or national level. However, although this program provides financial incentives for improvements in quality of care, the specific measures (i.e., hypertension control, diabetes management, colorectal cancer screening) for which payments are given have not been implemented uniformly across VA facilities. In addition, VA performance pay encompasses a wide array of measures, while our incentive focuses solely on management of hypertension, thus maximizing the effect of our intervention. Further, by using a RCT design, we limit the likelihood of confounding by other concurrent quality improvement programs. Second, the VA has an extensive system of clinical reminders (i.e., use of thiazide diuretics in patients with hypertension or use of aspirin in patients with ischemic heart disease) designed to promote the provision of guideline-recommended care. However, while the VA has implemented a variety of quality improvement strategies, our pilot data indicate that there remains room for further improvement. Third, although the VA healthcare system cares for few women patients, the subjects of this study are physicians. There is little reason to believe that the effect of a financial incentive to a physician would result in different treatment of a woman or a man with hypertension. Thus, findings from this study will be relatively generalizable to budgeted systems and staff-model health maintenance organizations (HMOs) that serve many millions of patients, to Centers for Medicare and Medicaid Services (CMS) with 40.5 million beneficiaries, and indirectly applicable to other healthcare delivery models. Finally, it is possible that there may be a 'volunteer effect' among participants. For example, physicians with a bias toward use of financial incentives may have been more likely to participate in our study. However, we expect that the randomized control design of our study will limit the effect of this potential source of bias.
In this paper, we have given a brief description of the rationale for the interventions being studied in this trial of pay for performance, as well as some of the design choices. Rigorous research designs such as this one are necessary to determine whether performance-based payment arrangements result in meaningful quality improvements. In this large cluster-RCT of pay for performance, we are seeking to provide such evidence for one of the most common chronic conditions affecting US citizens.
We gratefully acknowledge the VA Network Directors who provided the financial incentives for this study as well as the VA staff who participated. This work is supported in part by Veterans Affairs Health Services Research & Development (HSR&D) Investigator-Initiated Research (IIR) 04-349 (PI Laura A. Petersen, MD, MPH), NIH RO1 HL079173-01 (PI Laura A. Petersen, MD, MPH), the American Recovery and Reinvestment Act of 2010 (NHLBI 1R01HL079173-S2), and the Houston VA HSR&D Center of Excellence HFP90-020 (PI Laura A. Petersen, MD, MPH). Dr. Petersen was a recipient of the American Heart Association Established Investigator Award (Grant number 0540043N) and was a Robert Wood Johnson Foundation Generalist Physician Faculty Scholar (Grant number 045444) at the time that this study was planned and funded. Dr. Hysong was a recipient of an NHLBI Investigator Research Supplement to Promote Diversity in Health-Related Research (1R01HL079173-S1) during the early stages of the study and is currently a VA HSR&D Career Development Awardee (CDA 07-0181). Dr. Profit's contribution is supported in part by the Eunice Kennedy Shriver National Institute of Child Health and Human Development #1 K23 HD056298-01 (PI: Jochen Profit, MD, MPH). Dr. Conrad is an awardee of the Robert Wood Johnson Foundation Health Care Financing and Organization Program (Grant number 63214). Dr. Dudley is a Robert Wood Johnson Investigator Awardee in Health Policy. The views expressed are solely of the authors, and do not necessarily represent those of the VA.
- Wang TJ, Vasan RS: Epidemiology of uncontrolled hypertension in the United States. Circulation. 2005, 112: 1651-1662. 10.1161/CIRCULATIONAHA.104.490599.View ArticlePubMedGoogle Scholar
- Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo JL, Jones DW, Materson BJ, Oparil S, Wright JT, Roccella EJ, National Heart and Blood Institute Joint National Committee on Prevention, and Treatment of High Blood Pressure; National High Blood Pressure Education Program Coordinating Committee: The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure: the JNC 7 report. JAMA. 2003, 289: 2560-2572. 10.1001/jama.289.19.2560.View ArticlePubMedGoogle Scholar
- McGlynn EA, Asch SM, Adams J, Keesey J, Hicks J, DeCristofaro A, Kerr EA: The quality of health care delivered to adults in the United States. N Engl J Med. 2003, 348: 2635-2645. 10.1056/NEJMsa022615.View ArticlePubMedGoogle Scholar
- The ALLHAT Officers and Coordinators for the ALLHAT Collaborative Research Group: Major cardiovascular events in hypertensive patients randomized to doxazosin vs chlorthalidone: the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT). JAMA. 2000, 283: 1967-1975. 10.1001/jama.283.15.1967.View ArticleGoogle Scholar
- Cabana MD, Rand CS, Powe NR, Wu AW, Wilson MH, Abboud PA, Rubin HR: Why don't physicians follow clinical practice guidelines? A framework for improvement. JAMA. 1999, 282: 1458-1465. 10.1001/jama.282.15.1458.View ArticlePubMedGoogle Scholar
- Berwick DM, DeParle NA, Eddy DM, Ellwood PM, Enthoven AC, Halvorson GC, Kizer KW, McGlynn EA, Reinhardt UE, Reischauer RD, Roper WL, Rowe JW, Schaeffer LD, Wennberg JE, Wilensky GR: Paying for performance: Medicare should lead. Health Aff (Millwood). 2003, 22: 8-10.View ArticleGoogle Scholar
- Doran T, Fullwood C, Gravelle H, Reeves D, Kontopantelis E, Hiroeh U, Roland M: Pay-for-performance programs in family practices in the United Kingdom. N Engl J Med. 2006, 355: 375-384. 10.1056/NEJMsa055505.View ArticlePubMedGoogle Scholar
- Lindenauer PK, Remus D, Roman S, Rothberg MB, Benjamin EM, Ma A, Bratzler DW: Public reporting and pay for performance in hospital quality improvement. N Engl J Med. 2007, 356: 486-496. 10.1056/NEJMsa064964.View ArticlePubMedGoogle Scholar
- Petersen LA, Woodard LD, Urech T, Daw C, Sookanan S: Does pay-for-performance improve the quality of health care?. Ann Intern Med. 2006, 145: 265-272.View ArticlePubMedGoogle Scholar
- Conrad DA, Perry L: Quality-based financial incentives in health care: Can we improve quality by paying for it?. Annu Rev Public Health. 2009, 30: 357-371. 10.1146/annurev.publhealth.031308.100243.View ArticlePubMedGoogle Scholar
- Dudley RA: Pay-for-performance research: how to learn what clinicians and policy makers need to know. JAMA. 2005, 294: 1821-1823. 10.1001/jama.294.14.1821.View ArticlePubMedGoogle Scholar
- Donner A, Klar N: Design and Analysis of Cluster Randomization Trials in Health Research. 2000, London: ArnoldGoogle Scholar
- Kuhn M: Quality in primary care: economic approaches to analyzing quality-related physician behavior. 2003, London: Office of Health Economics Research, 92-Google Scholar
- Bodenheimer T, Wagner EH, Grumbach K: Improving primary care for patients with chronic illness. JAMA. 2002, 288: 1775-1779. 10.1001/jama.288.14.1775.View ArticlePubMedGoogle Scholar
- Bodenheimer T, Wagner EH, Grumbach K: Improving primary care for patients with chronic illness: the chronic care model, Part 2. JAMA. 2002, 288: 1909-1914. 10.1001/jama.288.15.1909.View ArticlePubMedGoogle Scholar
- Wagner EH, Austin BT, Davis C, Hindmarsh M, Schaefer J, Bonomi A: Improving chronic illness care: translating evidence into action. Health Aff (Millwood). 2001, 20: 64-78. 10.1377/hlthaff.20.6.64.View ArticleGoogle Scholar
- Petersen LA, Woodard LD, Henderson LM, Urech TH, Pietz K: Will hypertension performance measures used for pay-for-performance programs penalize those who care for medically complex patients?. Circulation. 2009, 119: 2978-85. 10.1161/CIRCULATIONAHA.108.836544.View ArticlePubMedPubMed CentralGoogle Scholar
- Donner A, Klar N: Statistical considerations in the design and analysis of community intervention trials. J Clin Epidemiol. 1996, 49: 435-439. 10.1016/0895-4356(95)00511-0.View ArticlePubMedGoogle Scholar
- Lachin JM: Introduction to sample size determination and power analysis for clinical trials. Control Clin Trials. 1981, 2: 93-113. 10.1016/0197-2456(81)90001-5.View ArticlePubMedGoogle Scholar
- Kluger AN, DeNisi A: The effects of feedback interventions on performance: a historical review, a meta-analysis, and a preliminary feedback intervention theory. Psychological Bulletin. 1996, 119: 254-284.View ArticleGoogle Scholar
- Hysong SJ: Meta-analysis: audit and feedback features impact effectiveness on care quality. Med Care. 2009, 47: 356-363. 10.1097/MLR.0b013e3181893f6b.View ArticlePubMedPubMed CentralGoogle Scholar
- Cheng J, Edwards LJ, Maldonado-Molina MM, Komro KA, Muller KE: Real longitudinal data analysis for real people: building a good enough mixed model. Stat Med. 2010, 29: 504-520.View ArticlePubMedPubMed CentralGoogle Scholar
- Rosamond W, Flegal K, Friday G, Furie K, Go A, Greenlund K, Haase N, Ho M, Howard V, Kissela B, Kittner S, Lloyd-Jones D, McDermott M, Meigs J, Moy C, Nichol G, O'Donnell CJ, Roger V, Rumsfeld J, Sorlie P, Steinberger J, Thom T, Wasserthiel-Smoller S, Hong Y, American Heart Association Statistics Committee and Stroke Statistics Subcommittee: Heart disease and stroke statistics--2007 update: a report from the American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Circulation. 2007, 115: e69-171. 10.1161/CIRCULATIONAHA.106.179918.View ArticlePubMedGoogle Scholar
- Grimshaw JM, Thomas RE, MacLennan G, Fraser C, Ramsay CR, Vale L, Whitty P, Eccles MP, Matowe L, Shirran L, Wensing M, Dijkstra R, Donaldson C: Effectiveness and efficiency of guideline dissemination and implementation strategies. Health Technol Assess. 2004, 8: iii-iv. 1-72View ArticlePubMedGoogle Scholar
- Epstein AM, Lee TH, Hamel MB: Paying physicians for high-quality care. N Engl J Med. 2004, 350: 406-410. 10.1056/NEJMsb035374.View ArticlePubMedGoogle Scholar
- Frolich A, Talavera JA, Broadhead P, Dudley RA: A behavioral model of clinician responses to incentives to improve quality. Health Policy. 2007, 80: 179-193. 10.1016/j.healthpol.2006.03.001.View ArticlePubMedGoogle Scholar
- Petersen LA, Urech TH, Byrne MM, Pietz K: Do financial incentives in a globally budgeted healthcare payment system produce changes in the way patients are categorized? A five-year study. Am J Manag Care. 2007, 13: 513-522.PubMedGoogle Scholar
- Department of Veterans Affairs Health Care Personnel Enhancement Act of 2004. Public Law 108-445. [http://frwebgate.access.gpo.gov/cgi-bin/getdoc.cgi?dbname=108_cong_public_laws&docid=f:publ445.108]
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.