Financial Incentives for Quality in Breast Cancer Care
Diana M. Tisnado, PhD, Danielle E. Rose-Ash, PhD, [...], and Katherine L. Kahn, MD
Abstract
Objectives
To examine the use of financial incentives related to performance on quality measures reported by oncologists and surgeons associated with a population-based cohort of patients with breast cancer in Los Angeles County, California, and to explore the physician and practice characteristics associated with the use of these incentives among breast cancer care providers.
Study Design
Cross-sectional observational study.
Methods
Physician self-reported financial arrangements from a survey of 348 medical oncologists, radiation oncologists, and surgeons caring for patients with breast cancer in Los Angeles County (response rate, 76%). Physicians were asked whether they were subject to financial incentives for quality (ie, patient satisfaction surveys and adherence to practice guidelines). We examined the prevalence and correlates of incentives and performed multivariate logistic regression analyses to assess predictors of incentives, controlling for other covariates.
Results
Twenty percent of respondents reported incentives based on patient satisfaction, and 15% reported incentives based on guideline adherence. The use of incentives for quality in this cohort of oncologists and surgeons was modest and was primarily associated with staff- or group-model health maintenance organization (HMO) settings. In other settings, important predictors were partial physician ownership interest, large practice size, and capitation.
Conclusions
Most cancer care providers in Los Angeles County outside of staff- or group-model HMOs are not subject to explicit financial incentives based on quality-of-care measures. Those who are, seem more likely to be associated with large practice settings. New approaches are needed to direct financial incentives for quality toward specialists outside of staff- or group-model HMOs if pay-for-performance programs are to succeed in influencing care.
Healthcare quality is of concern nationally, and healthcare structural arrangements have been evolving rapidly to respond to increasing financial pressures and demands to enhance quality. These changes have been shown to affect primary care delivery,1–4 yet little is known about how these new organizational and financial arrangements affect the delivery of specialty care, particularly in the context of cancer.
It has long been recognized that financial incentives can have a powerful influence on physician and patient behavior. Associations have been found between implicit financial incentives (ie, provider compensation type) and productivity,5 resource use,6 propensity to use particular treatment options such as breast-conserving surgery7 and growth factors,8 continuity, compliance, patient satisfaction,9 and the use of evidence-based medicine practices.10 Explicit incentives for quality (ie, direct financial incentives tied to performance on quality measures or “pay for performance”) have been viewed as a mechanism to stimulate improvements in healthcare quality and patient outcomes by better aligning financial incentives with the delivery of higher-quality care. Although the effect of explicit financial incentives on primary care physicians has been the subject of several studies,4,11–14 understanding of these financial arrangements among specialists remains limited. Even less is known about the financial incentives used among specialists who provide cancer care in today’s managed care–penetrated environment.
Breast cancer serves as an excellent condition to use as a model for the study of structural characteristics and how they influence quality of care in general and cancer care more specifically. Breast cancer is a complex expensive disease with multiple treatments delivered by multiple providers over time, making patients vulnerable to problems with quality such as poor continuity and coordination of care.
Practice guidelines for breast cancer care issued by the National Comprehensive Cancer Network have been in place since the 1990s. However, a 2002 review15 of the patterns-of-care literature found notable variations in axillary lymph node dissection, hormone receptor status documentation, radiation therapy after breast-conserving surgery, and the use of tamoxifen citrate. The National Initiative for Cancer Care Quality,16 an in-depth study of quality of care for patients with breast or colorectal cancer in 5 states across the United States from 1998 to 2002, revealed substantial opportunities for improvement. Adherence was less than 85% for half of the quality measures, and substantial variation was found across metropolitan areas. Significant proportions of eligible patients did not complete an indicated course of tamoxifen17 and had no documented discussion about postmastectomy reconstruction. 18 In a population-based study19 conducted in 2000 in Los Angeles County, California, 60% of black women, 55% of Spanish-speaking Hispanic women, and 34% of white women did not receive desired help for symptoms. Concerns about financial incentives motivating overuse of certain cancer treatments such as growth factors have also been expressed.8
Based on theoretical grounds20 and evidence from previous interventions,13,14 it is reasonable to expect that financial incentives for quality have the potential to influence physician behavior, aligning financial incentives with desired outcomes. At the time that we conducted the present study, we were unaware of any specific pay-for-performance systems in oncology, but with the substantial energy and enthusiasm mounting behind pay for performance as a way to motivate and reinforce quality in general medicine, as well as the rising concerns regarding quality in oncology, we set out to assess the state of uptake among oncologists and surgeons treating patients with breast cancer.
We developed and conducted a physician survey to describe the structure of breast cancer care in Los Angeles County21 and to evaluate the effect of structure on the quality of care that patients with breast cancer receive.19 In this article, we present findings regarding the prevalence and types of financial incentives tied to performance on quality measures reported by oncologists and surgeons associated with a population-based cohort of patients with breast cancer in Los Angeles County. To assess the determinants of financial incentives tied to performance on quality measures, we explore the physician and practice characteristics associated with the use of these incentives among breast cancer care providers. Based on the literature and cognitive interviews conducted to inform the development of the survey, we hypothesized that explicit incentives for quality, although available and potentially useful for motivating and reinforcing cancer quality assessment and improvement efforts, would be rarely used except for large group-model health maintenance organizations (HMOs) or large multi-specialty medical group practices.
METHODS
We conducted a cross-sectional study of the structure and organization of care associated with the physicians identified among a population-based sample of women with incident breast cancer participating in the Los Angeles Women’s Health Study, which included a population-based sample of women 50 years and older identified as having a new diagnosis of breast cancer in Los Angeles County in 2000 by the Los Angeles County Cancer Surveillance Program Rapid Case Ascertainment system. A computer-assisted telephone interview (CATI) was administered to 1224 women who agreed to participate. The detailed CATI queried women about healthcare providers who fulfilled various roles in their care, including delivering, recommending, or discussing the possible use of breast cancer treatments. We targeted all Los Angeles County medical oncologists, radiation oncologists, and surgeons reported by their patients in the CATI.
We developed items to query physicians about financial reimbursement arrangement and incentives. These were based on the literature5,9,10,22–27 and on cognitive interviews with cancer care specialists.
Patient Case Mix
Physicians were asked to report the proportions of patients in their main practice setting who were primarily covered by each of the following: Medicare fee for service (FFS), Medicare managed care, MediCal FFS, MediCal managed care, private traditional FFS health insurance, private managed care health insurance, other health coverage, and no coverage (uninsured).
Physician Compensation Method
We asked physicians to indicate the proportions of clinical compensation that are paid to them in the following ways: salary that is fixed for the year or contract period; salary that may vary during the year or contract period according to factors such as productivity, performance review, and net income of the practice, FFS or discounted FFS, capitated reimbursements, and other.
Financial Incentives for Quality
Physicians were asked whether in the previous 12 months their pay was affected in any way as a result of scoring well on the following: (1) patient satisfaction surveys or (2) other specific measures of quality of care such as adherence to practice guidelines or protocols.
Practice Setting
We accounted for the importance in Southern California of network-model managed care involvement. To do so, we derived new practice setting categories by dividing single-specialty medical groups, multispecialty medical groups, and solo practices into high and low managed care, using as a cut point the median reported percentage (36%) of network-model managed care patients.
Study Design
In 2004, we mailed the finalized self-administered survey instrument, approved by the institutional review board at the University of California, Los Angeles, to 477 physicians named by their patients with contact information, including medical oncologists (n = 175), radiation oncologists (n = 75), and surgeons (n = 227). We received 348 surveys from physicians associated with 298 unique office addresses and determined 19 of the original 477 physicians to be ineligible, for a final physician response rate of 76% (67% for medical oncologists, 89% for radiation oncologists, and 80% for surgeons).
Response weights were calculated as the inverse of the probability of response based on a logistic regression model that included physician sex, specialty type, study patient volume, and sharing an office with another surveyed physician. Significant predictors of response were specialty type (odds ratio [OR], 0.24; P <.001 for medical oncologists; OR, 0.44; P = .04 for surgeons compared with radiation oncologists) and sharing an office with another surveyed physician compared with not sharing (OR, 1.70; P = .02).
We examined the prevalence, correlates, and predictors of use of financial incentives for quality using the following dependent variables for the physician financial incentives: (1) financial incentives based on the results of patient satisfaction surveys, (2) financial incentives linked to other measures of quality such as adherence to practice guidelines, and (3) financial incentives linked to patient satisfaction and guideline adherence. Covariates examined included practice characteristics (managed care involvement and practice setting type and size) and physician characteristics (specialty, sex, and ownership interest). Descriptive and bivariate analyses were performed that included comparisons of categorical variables using χ2 test.
We performed multivariate logistic regression analyses to assess the predictors of physician incentives for quality of care. We developed 3 models using the 3 dependent variables for the physician financial incentives already described. Multivariate models controlled for physician ownership interest in their practice, practice setting type, practice size, whether the physician was paid predominantly FFS, and whether the physician received any capitated payments. All analyses were weighted for nonresponse and were adjusted for clustering of physicians within shared offices.28
RESULTS
Table 1 gives results describing characteristics of the physician sample by specialty type and overall. Overall, 20% of respondents reported being subject to incentives for quality based on the results of patient satisfaction surveys and 15% based on the results of other quality measures such as adherence to practice guidelines. Twenty-two percent reported being subject to either, and 13% reported being subject to both types of incentives. These rates did not differ significantly by specialty type.
Table 2 gives bivariate associations between physician and practice characteristics and incentives based on patient satisfaction, other quality measures, and both types of incentives. Bivariate results indicate that, for all measures, physicians’ reported use of financial incentives for quality in their main practices varied significantly by physician ownership interest, practice type, practice size, being paid predominantly FFS, and (only in the case of reporting both types of incentives) being paid any capitation. The use of incentives for quality was most likely among physicians working in staff- or group-model HMO practice settings and in large practice (>50 physicians) settings. Other physicians who were more likely to report the use of incentives for quality were those with partial ownership interest in their practice, those who were not paid predominantly according to FFS, and (in the case of incentives based on both types of quality measures) those not receiving any reimbursement in the form of capitation. Physician age and sex were not significantly associated with the use of incentives for quality and were not included in further analyses.
Table 3 gives the results of multivariate analyses. In the multivariate analyses, 2 sets of logistic regression analyses were performed for each of the 3 dependent variables, namely, incentives based on the results of (1) patient satisfaction surveys, (2) other quality measures such as adherence to practice guidelines, and (3) both incentive types. We predicted the use of financial incentives for quality as a function of ownership interest, practice setting type, practice size, FFS reimbursement type, and acceptance of capitation among all physicians surveyed.
In logistic regression analyses predicting incentives based on patient satisfaction (Table 3), group-model HMO practice type was far more likely to be associated with reported use of incentives for quality compared with solo practice type (OR, 33.20; P <.001), controlling for ownership interest, practice size, and reimbursement type. Other significant predictors were partial ownership interest compared with full ownership (P = .04) and large practice size compared with small practice size (P = .005). Physician sex and specialty type were not significantly associated with the use of incentives for quality and were excluded from final models.
Model 1 in Table 3, including all physicians, illustrates that the use of incentives based on patient satisfaction was overwhelmingly associated with staff- or group-model HMO practice setting type. Model 2 (n = 284) restricted the analysis to physicians not associated with group-model HMOs using the same covariates. In model 2, significant predictors of incentives based on patient satisfaction were partial ownership interest versus full ownership (OR, 4.94; P = .002) and any reimbursement in the form of capitation versus none (OR, 3.26; P = .01).
Results of the 2 models predicting incentives based on other quality measures such as adherence to quality measures demonstrate that significant predictors of incentives based on quality measures other than patient satisfaction were staff- or group-model HMO practice setting type (OR, 8.58; P = .04) and large practice size (OR, 4.41; P = .006) (Table 3). Reimbursement predominantly in the form of FFS was negatively associated with the use of such incentives (OR, 0.23; P = .03). Ownership interest was not significant. In the model excluding staff- or group-model HMO physicians, only large practice size was a significant predictor (OR, 4.78; P = .04).
Table 3 also gives results of analyses that specify as dependent variables the use of financial incentives based on both types of quality measures. The results of model 1 were consistent with the results in which the dependent variable was incentives based on quality measures other than patient satisfaction, except that ownership interest was not significant. Excluding staff- or group-model HMO physicians, no independent variables were significant in model 2.
DISCUSSION
To our knowledge, this is the first study to examine the use of financial incentives for performance on quality measures specifically among cancer care providers. We found that the use of incentives for quality among surgeons and medical and radiation oncologists caring for patients with breast cancer in Los Angeles County was modest and was primarily associated with staff- or group-model HMO practice settings.
We found that study physicians in staff- or group-model HMO settings were much more likely to report being subject to pay for performance. However, we note that few staff- or group-model HMOs exist outside of California, the Pacific Northwest, and Minnesota. In light of the small number of such settings in our sample and elsewhere, we performed analyses that excluded physicians in staff- or group-model HMO settings to identify predictors of pay-for-performance use in other settings. Among physicians in other settings, important predictors were partial physician ownership interest, large practice size, and physician receipt of capitated reimbursements.
These findings could be accounted for by group-model HMOs responding to concerns about quality with strong quality initiatives. Staff- or group-model HMOs are often well equipped to administer pay-for-performance programs with their strong link between financing and care delivery, information technology capabilities for data capture and attribution of care to a specific physician, and potential to influence organizational culture. In contrast, the more prevalent network model faces challenges in delivering incentives, with weaker links among physicians, particularly specialists.
Our findings might also reflect that oncology specialists are less likely to contract directly with health plans as opposed to with physician organizations, which in turn may be less likely than health plans to implement pay-for-performance schemes directed at specialists with whom they contract. In our study, 59% of all physicians not in staff- or group-model HMOs and 70% of 133 physicians in single-specialty medical groups reported at least 1 contract with a medical group or independent practice association; it is unknown how many might also contract directly with health plans.
Differences were also found across levels of physician ownership interest in their practice. Respondents with partial ownership interest in their practice were more likely to report receipt of incentives for quality than full owners, despite controlling for practice setting type and size. We believe that this reflects differences across practice setting types in the source of the incentive program. In some settings, incentives for quality are administered by the medical group, whereas in other settings incentives come directly from the health plan to the physicians. Unfortunately, we did not have data regarding this level of complexity and were unable to tease out any such differences. We were also somewhat surprised to find no significant differences by specialty type, as many domains of structure of care under study in this physician cohort have varied across specialty type.21
Clearly, there has been great interest in paying for performance as a way to motivate improvements in the quality of care in the United States, and perhaps nowhere more so than in California. Although no 2 medical markets are alike, experiences in Southern California have been viewed as a bellwether and may offer insights for plans and providers in other markets. Managed care dominates the Southern California medical market, although enrollment is down slightly from the highs of the 1990s. The Southern California managed care market is dominated by a few large HMOs. Although the single-group–model HMO remains the largest managed care plan in the state, with a large effect on the market, most are network-model HMOs. Following substantial consolidation throughout the 1990s, medical groups and independent practice associations are now the most prominent form of physician organization in Los Angeles. Increasingly, physician organizations face demands from health plans aiming to link compensation to performance. Although only a small proportion of payments from health plans was tied to clinical performance in 2004, that proportion was expected to grow, particularly given the strong value-based purchasing environment and efforts of the Integrated Healthcare Association in California, as well as other initiatives around the country such as Bridges to Excellence and Rewarding Results.29
The Centers for Medicare & Medicaid Services are moving closer to implementation of incentives, but little is known about the prevalence of their use among specialists. Rosenthal30 documents that the percentage of enrollees in plans with various types of pay-for-performance systems targeting specialists increased substantially between 2003 and 2006, although systems targeting primary care providers have remained dominant. In 2007, only 11 of the Centers for Medicare & Medicaid Services Physician Quality Reporting Initiative quality measures targeted cancer care, with just 2 focusing on breast cancer.31
A good deal of medical care, particularly for chronic conditions among older persons, is and will continue to be delivered by specialists. Like primary care physicians, they should be accountable for the quality of care that they provide; they should also have opportunities to reap rewards when they are delivering good care. However, the modest use of incentive systems among specialists caring for patients with cancer outside of staff- or group-model HMO settings suggests that among the major challenges to be faced will be how to apply such incentive systems to specialists in smaller independent practices. An additional challenge will be the evaluation of systems that are implemented in different ways such as those in which incentives are delivered at the medical group level compared with the individual physician level. Furthermore, in conditions such as cancer in which a team of physicians may be treating the patient and certain patient needs are fulfilled by more than 1 specialty type, methods must be developed for the appropriate attribution of “credit” and responsibility when guideline-concordant care is and is not delivered. In addition, most providers have contracts with multiple plans or physician organizations and face differing arrangements with each, negatively affecting the likelihood that incentives would significantly influence physician behavior.
This study is subject to several limitations. The study is limited in geographic scope. Results come from Los Angeles County physicians named as care providers to a population-based cohort of patients with breast cancer and reflect practices of a small number of plans and physician organizations in a managed care–penetrated market facing strong purchaser interest in quality improvement initiatives. Experiences in Southern California have been viewed as a bellwether and may offer insights for plans and providers in other markets, particularly in markets with high managed care penetration such as in the Pacific Northwest, Northeast, mid-Atlantic, and Midwest (eg, Washington, Oregon, Massachusetts, New York, Pennsylvania, Maryland, Minnesota, Michigan). In addition, narrow geographic focus provides a control for unmeasured market characteristics.
We did not have data allowing us to separate out the specific system level associated with the administration of incentives for quality. Therefore, we could not discern which incentives for quality were administered by health plans versus by the physician’s medical group. However, we measured incentives directly affecting the pay of respondent physicians based on the results of their performance compared with incentives that may be received by the medical group and never dispersed to individual physicians. Nonresponse could have biased results if the reason for nonresponse was differentially associated with the use of financial incentives for quality. However, at 76%, our response rate was high relative to most physician surveys, and analyses were weighted for nonresponse based on specialty type, physician sex, practice volume, and sharing an office with another study participant. In addition, this analysis did not assess whether financial incentives for quality are associated with quality of care and patient outcomes in the context of cancer care.
CONCLUSIONS
Most cancer care providers in Los Angeles County outside of staff- or group-model HMOs are not subject to financial incentives based on quality-of-care measures, and those who are seem more likely to be associated with large practice settings. New approaches will be needed to direct financial incentives for quality toward specialists outside of staff- or group-model HMOs if pay-for-performance programs are to succeed in improving quality. However, further research is needed to determine whether the use of financial incentives can indeed influence cancer care and outcomes and what, if any, associations exist between implicit financial incentives (eg, reimbursement based on FFS, capitation, or salary), explicit financial incentives linked to performance on quality measures, and actual outcomes of patients with cancer.
Acknowledgments
Funding Source: This work was supported by a grant from the California Breast Cancer Research Program (#7PB-0126). Dr Tisnado received additional grant support from a Department of Defense Breast Cancer Research Post-doctoral Program (#DAMS17-03-1-0328) and the UCLA Center for Health Improvement of Minority Elderly/Resource Centers for Minority Aging Research (NIH/NIA #P30AG021684). Dr Rose-Ash received grant support from a UCLA Cancer Education and Career Development National Cancer Institute (#R25 CA 087949).
Footnotes
Author Disclosure: The authors (DMT, DER-A, JLM, JAL, PAG, KLK) report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.
This study was presented in part at the annual sessions of the 24th AcademyHealth Research Meeting; June 3–5, 2007; Orlando, FL; and at the California Breast Cancer Research Program; September 7–9, 2007; Los Angeles, CA.
Authorship Information: Concept and design (DMT, DER-A, JLM, PAG); acquisition of data (DMT, DER-A); analysis and interpretation of data (DMT, DER-A, JLM, PAG); drafting of the manuscript (DMT); critical revision of the manuscript for important intellectual content (DMT, DER-A, JLM, PAG); statistical analysis (DMT, DER-A); provision of study materials or patients (DMT); obtaining funding (DMT); and administrative, technical, or logistic support (DMT).