What can health care organizations do to support physicians in practicing value-based care? A combination of financial incentives and data-driven tools and capabilities may help doctors to align their activities with value-based care principles.
I am in favor of progress; it’s change I don’t like.–Mark Twain1
Transitioning from volume-based to value-based payment and care delivery models in health care has been one of the most important industry-wide efforts over the past few years, but the pace of change has been slow. For instance, in a 2016 survey of executives at provider organizations, 94 percent indicated that they are on the path to value-based care, yet only 27 percent have completed pilots or are at some stage of rollout. These numbers show little change from 2015.2
What can health care organizations do to stimulate wider adoption of value-based care? They can try to gain a better understanding of physician perspectives: Physicians are obviously affected by industry changes, and they can have great influence on the cost and quality of care.
The Deloitte 2016 Survey of US Physicians, a nationally representative sample of 600 US primary care and specialty physicians, confirms the slow pace of adoption of value-based payment models among physicians: Currently, there is little focus on value in physician compensation, and physicians are generally reluctant to bear financial risk for care delivery. At the same time, however, many physicians conceptually endorse some of the principles behind value-based care, such as quality and resource utilization measurement. The survey results suggest that:
The survey findings suggest that, to stimulate the adoption of value-based care and support physicians in delivering on the “Triple Aim,”3—lower cost, better health, and improved patient experiences—a combination of financial incentives and data-driven tools and capabilities may help. Specifically, organizations could seek to:
“The most expensive piece of medical equipment, as the saying goes, is a doctor’s pen,” writes Atul Gawande, a prominent physician, writer, and health services researcher.4 In fact, the majority of US physicians are currently reimbursed based upon volume under a system known as fee-for-service (FFS).5 FFS encourages the use of more tests, procedures, and treatments, not all of which might be supported by evidence on quality and value.6 It is not surprising, therefore, that many efforts to improve the performance of the US health care system are focused on physicians. Health systems, physician organizations, health plans and government payers, and life sciences companies want to better understand how to influence physician behavior to realize success under value-based payment models.
Transitioning from volume to value has been slow. Many health systems and medical groups still make the majority of their revenue under FFS.7 Only 3 percent of health systems provide more than one-half of all care under value-based contracts.8 And while health systems may have value-based contracts that put the organization as a whole under some risk for meeting quality and cost goals, many still compensate their physicians based primarily upon volume.
The Medicare Access & CHIP Reauthorization Act of 2015 (MACRA) intends to encourage the adoption of value-based care in the United States. MACRA will base Medicare payments to clinicians on their performance against certain cost and quality measures, starting in 2019. It encourages participation in alternative payment models that require financial risk-sharing while also improving quality.9
Although clinicians aspire to improve health outcomes, the FFS model does not reward physicians for achieving these improvements. FFS, the most common payment system in the United States today, rewards physicians and hospitals for furnishing a high volume and intensity of services. Many experts agree that FFS works against population health goals of using “health care resources effectively and efficiently to improve the lifetime health and well-being of a specific population.”10
Facing rising and unsustainable costs and subpar quality, employers, private health insurers, and government purchasers of health care are pushing for value-based payment models. Under these models, providers are paid based upon performance, measured in terms of cost, quality, and outcomes, not volume. (For an overview of specific value-based payment arrangements, please refer to Appendix 2.)
MACRA aims to accelerate the adoption of value-based payment models by setting two reimbursement tracks for physicians in Medicare.11 The default track is the Merit-Based Incentive Payment System (MIPS), which varies payments to physicians based upon their individual performance on cost and quality indicators as well as their use of health information technology and clinical improvement activity. Additional payments go to physicians (and other clinicians) who participate in advanced Alternative Payment Models (APMs), or certain value-based payment models that carry both upside and downside financial risk.
Given the increasing efforts to transition to value-based payment models (see sidebar, “About MACRA”),12 how can health care organizations stimulate wider adoption rates among physicians? To explore this question, we draw on results from the Deloitte 2016 Survey of US Physicians, which provides unique insights on value-based care from a nationally representative sample of 600 primary care and specialty physicians. The survey sheds light on physicians’ current sources of compensation, preferences regarding compensation arrangements, perceptions regarding care transformation efforts, needed tools and capabilities, and readiness for change.
Topics covered in the Deloitte 2016 Survey of US Physicians
The survey asked physicians about a range of topics related to MACRA, consolidation, and health information technology. Since 2014, we have asked questions on value-based care.
The survey explored the following topics:
Even though many stakeholders in the health care industry are committed to value-based care, our findings indicate that most individual physicians are not yet compensated under value-based models. Understanding physicians’ views on various compensation methods can help guide health systems and other organizations working with physicians in structuring physician compensation to better align with value-based care principles.
Value-based payments make up a small proportion of physician compensation
Similar to our 2014 findings, a majority of physicians (more than 8 in 10) still report being compensated under FFS or salary (figure 1). While physician participation in value-based payment models is increasing (30 percent in 2016 vs. 25 percent in 2014), few physicians participate in models that have the greatest downside risk (10 percent in capitation and 4 percent in shared-risk arrangements).
Even for organizations participating in Centers for Medicare and Medicaid (CMS) pilots, such as accountable care organizations (ACOs), a study revealed that the structure of physician compensation was similar to that in organizations that were not part of an ACO. The study found that physicians in ACOs and those not in ACOs earned 49 percent of their compensation from salary, 46 percent from productivity (volume), and only about 5 percent from quality and other factors.13
Not only are value-based sources of payments an uncommon source of physician compensation, but the proportion of compensation tied to performance, such as better quality or lower cost, is also small (figure 2). One-half of physicians in the survey reported performance bonuses less than or equal to 10 percent of their compensation, and one-third reported that they were ineligible for performance bonuses. These numbers are well below the threshold (20 percent of total compensation) that the literature suggests would be effective in incentivizing physicians and producing behavior change.14
Interestingly, physicians reported that they would be willing to accept sizeable proportions of compensation at risk, if required to. The median reported proportion is 15 percent, meaning one-half of surveyed physicians would put more than 15 percent of compensation at risk and the other half would accept less than 15 percent.
Most physicians reported that they prefer FFS and/or salary (figure 3). As in 2014, few physicians preferred value-based payment models that carry significant financial risk (such as capitation and shared risk). However, compared to 2014, more physicians preferred models that include some upside risk component, such as shared-savings models.
Studies show that incentives are most effective when they go directly to the clinician (rather than to the medical group or treatment team), and outcomes are most likely to improve when financial incentives are sufficiently large—at least 20 percent.15 Geisinger Health System, known for its employed-physician model and strong cost and quality performance, uses an 80/20 compensation model (20 percent of physician compensation is based upon cost and quality performance). Since instituting this compensation structure in 2006, Geisinger has seen improved health outcomes and lower costs for 18 common treatment interventions for conditions such as congestive heart failure.16
What should you consider?
Regardless of financial incentives to reduce costs and improve care quality, physicians would have a difficult time meeting these goals if they lack data-driven tools. These tools can give them insight on cost and quality metrics, and can help them make care decisions that are consistent with effective clinical practice. Our survey explored physicians’ needs for and usage of such tools: clinical protocols to inform decisions, care pattern data for performance measurement and improvement, care pattern data for outside referrals, and electronic health records (EHR) technology. Most physicians reported having some access to these tools, but the access varied by type of physician and tool.
Variations in care lead to variations in cost and quality. Based on an analysis of this variation, the Dartmouth Atlas of Care estimates that as much as 20 percent to 30 percent of all US health care spending may be unnecessary.17 For organizations participating in value-based payment models, reducing unwarranted variation in care is an important clinical priority since doing so can help improve quality and cost performance. Some approaches to reducing unwarranted clinical variation include:
Physicians reported that they have access to clinical protocols and acknowledge their value. Three out of four surveyed physicians reported having clinical protocols, with 36 percent of all respondents having access to comprehensive protocols for many medical conditions. Anecdotally, care standardization has a longer history in inpatient settings (that is, hospitals), and our survey results confirm greater protocol availability among inpatient-based (90 percent) compared to outpatient-based (69 percent) physicians. We also see large differences between employed and independent physicians: 92 percent of employed physicians vs. 68 percent of physicians in independently owned practices have protocols (see figure 4).
Since implementing clinical protocols at the point of care can require EHRs and clinical decision support, the cost of these technologies may explain the lower adoption rate among independently owned physician practices.19
Overall, physicians held positive views about clinical protocols and the idea of reducing clinical variation. Three in five (60 percent) physicians reported that on balance, the positive aspects of having protocols outweigh the negatives (figure 5) and nearly one-half (48 percent) think that reducing clinical variation could help improve the performance of the US health care system (figure 6).
Physicians with access to clinical protocols tended to have more favorable views about controlling costs and quality, measuring performance, and reducing clinical variation. And those with access to protocols were more likely to favor value-based payment models and public reporting of individual physicians’ performance, even though these strategies are generally unpopular.
Some physicians expressed concern that clinical protocols limit physicians’ ability to make clinical decisions (43 percent agree and 35 percent disagree with this statement). According to other research on this topic, concern about losing clinical autonomy has been a major barrier to clinical protocol adoption.20 Our results show that physicians with access to protocols generally viewed them more favorably. We also find that even physicians with access to clinical protocols had concerns over loss of clinical autonomy: 37 percent agreed and 39 percent disagreed that protocols limit physicians’ ability to make clinical decisions.
Physicians’ favorable attitudes about using clinical protocols suggest that the industry is making strides in bringing the evidence-based approach to the point of care. Nevertheless, the following hurdles remain: adopting protocols among physicians who do not currently have them; broadening the protocol scope to include more conditions; and reviewing and updating existing clinical guidelines to ensure that they accurately reflect the evidence base. Reasons for low adoption of clinical protocols, especially for small independent practices, may include absent or inadequate HIT systems, skepticism about the utility of clinical guidelines, and distrust of the guideline development process.21
What should you consider?
Care pattern reports are available to most physicians, but challenges remain
Care pattern reports provide physicians with feedback on their clinical practices. They may contain information on the patient experience, the quality of care, resource use, or cost, and can be used for continuous quality improvement or for performance-based compensation.
Sixty-five percent of surveyed physicians reported receiving care pattern information. However, the survey also reveals some gaps between the reported availability and the perceived usefulness of these tools. Physicians noted that care pattern reports should contain information on clinical outcomes, patient experience measures, and cost. In practice, though, physicians reported mostly receiving information on “process” measures, such as quality-of-care information, rather than clinical outcomes. Patient experience is a frequently available quality metric, and many physicians find it useful.
Compiling and reporting accurate care pattern data can be challenging, due to time lags in data collection and attribution when patients see multiple doctors. While many measures exist for primary care, there are few for specialties like oncology and nephrology.22 Quality reporting is expensive, with an estimated cost of around $40,000 per physician per year.23 Our survey results show that:
Physicians with access to some types of advanced capabilities (for example, clinical protocols and/or care pattern information) were less likely to say they feel underprepared for quality reporting requirements such as those considered under MACRA. But even in this group, most say that quality reporting is burdensome. For instance, 72 percent of physicians with access to clinical protocols vs. 82 percent of those without described quality reporting as burdensome, and 84 percent vs. 87 percent said they would need additional resources to comply with reporting requirements.
When asked about improvements to care pattern reports, physicians cited that they would like the data to be adjusted for patient complexity or severity (60 percent), to be trustworthy and consistent with their experience (51 percent), and to have a stronger focus on outcomes instead of processes (36 percent). Some of the desired features had more to do with the delivery and usability of care pattern reports than with their actual content (see figure 8).
Under MACRA, performance on resource utilization and quality measures will be factors affecting the level of physician reimbursement in Medicare. The fact that only one in five physicians reported receiving resource utilization data (figure 7) points to the need to develop these reporting capabilities further.24 Not only do physicians need to receive this type of information, but the data need to be presented in a way that is useful, easy to understand, and actionable.25
Similarly, quality data used in setting performance benchmarks should be reliable, reproducible, and focused on outcomes within physicians’ control. Methodological details and rationale about outcome measurements (such as severity adjustments, or lack thereof) and patient attribution should be clearly explained, as well as the implications for physicians’ work and what they would need to do to improve.
What should you consider?
Physicians report low use of cost or quality data in informing patient referrals
For organizations building value-based care capabilities, understanding physician referral behaviors and patterns of referrals can be a way to find savings or improve outcomes.26 One study found that, in cases where treatment guidelines were unclear, physicians in high-spending regions were much more likely to choose intensive clinical approaches than physicians in low-spending areas, and a number of those approaches involved referrals (for example, referrals to specialists for one-time consultations or for ongoing management, to tests and diagnostic procedures, to hospitals or intensive care units).27 There is also high variation in health care prices that is unrelated to quality; this variation exists even within the same markets, where the prices for the same procedure can vary by a factor of three or four.28
Our surveyed physicians cited trust or working relationship (75 percent) and specialized expertise (69 percent) as the top two criteria in patient referrals (see figure 9). Other studies also show that physicians value clinical expertise.29
Consistent with the literature, patient access considerations (51 percent) are also prominent in referral decisions. This is especially true among primary care (58 percent) and nonsurgical specialists (60 percent).
Our survey results suggest data-driven and evidence-driven referral patterns are uncommon: Only 15 percent of physicians said they take into account outcomes or quality ratings when they make referrals. Cost considerations were also infrequent, as 15 percent of physicians considered patient co-pays and insurance in-network status and only 1 percent took into account physicians’ fees.
Not surprisingly, physicians were interested in different types of information for different types of referrals. The rate of complications would be useful for 64 percent of physicians in referrals to procedural specialists. For referrals to specialists who mostly provide consultations, patient experience (55 percent) and rate of diagnostic errors (42 percent) were considered most useful. For referrals to treatment facilities, patient experience (44 percent) topped the list. And for referrals to outpatient diagnostic facilities, cost to the patient (52 percent) was the number one choice, followed by the rate of diagnostic errors (45 percent).
Many physicians are interested in using data on quality in their referrals; in its absence, they rely on habitual referrals. Given this interest and MIPS incentives tied to resource use measures, physicians may see value in referring patients to providers who routinely use low-intensity (or conservative) approaches. Additionally, if the current trend of increased patient cost sharing due to high deductibles continues, physician interest in cost-related information may grow, since many physicians are attuned to patient access considerations.
What should you consider?
To better understand the factors most likely to contribute to physicians’ participation in value-based payment models, we built a regression model that used a combination of demographics, practice-setting characteristics, and measures of tools and resource availability (see Appendix for methodology).
We found that, with regard to their willingness to adopt value-based payment models, physicians can be classified in three broad segments:
Our analysis shows important differences in demographics and practice-setting characteristics of the physicians in the three segments (see figure 10, and Appendix for full details):
Our analysis also revealed large differences among segments in attitudes, experience with performance-based compensation, and risk tolerance. For instance, 36 percent of physicians in the willing segment already receive some compensation from a value-based source of payment vs. 24 percent of physicians who are on the fence and 21 percent of resistant respondents.
The demographic and practice setting differences between the segments have particular implications for value-based care efforts. For instance, solo practitioners and those in small practices might be more difficult to engage effectively as they tend to be more resource-constrained, both in terms of staffing and technology availability.
Interestingly, however, our analysis shows that the availability of tools and resources helps mitigate the effects of non-modifiable demographic characteristics. For instance, when physicians have care pattern information, clinical protocols, and Stage 3 Meaningful Use EHRs, their willingness to participate in value-based care increases. Making these tools available could help move physicians from the on the fence to the willing category.
Physicians’ views on needed resources and capabilities provide further insights into what would increase the likelihood that they would accept risk-based compensation. For instance, while many physicians consider the ability to track costs as a prerequisite for accepting risk-based compensation, two-thirds (67 percent) of willing physicians thought so. On-the-fence physicians gave much greater weight to patient engagement tools than the other segments, while resistant physicians were more skeptical of risk-based compensation; 28 percent said none of the presented options would make physicians more likely to accept more risk (table 1).
While financial incentives and supporting tools are important, this analysis indicates that there are additional factors associated with a physician’s propensity to bear financial risk.
What should you consider?
Financial incentives, when coupled with supporting tools and capabilities, can potentially increase the pace of physicians’ transition to value-based care. Organizations in each of the health care sectors should consider how they can best help physicians make this transition.
Health systems: Opportunity to grow value-based contracting, increase incentives, and improve tools for physicians
Health plans: Opportunity to serve as a data and analytics resource for cost and referral information to support value-based care
Biopharma and medtech companies: Opportunity to build economic evidence and partner for value
Having physicians engaged and involved is critical for value-based care since their decisions impact treatment, costs, and quality. The various stakeholders should consider how they will each play a part in helping physicians transform care delivery. Only when all of the stakeholders are working together toward the same goal can the Triple Aim of lower cost, better health, and improved patient experience truly be within reach.
Since 2011, the Deloitte Center for Health Solutions has surveyed a nationally representative sample of US physicians on their attitudes and perceptions about the current market trends impacting medicine and predictions about the future state of the practice of medicine. The general aim of the survey is to understand physician adoption and perception of key market trends of interest to the health plan, health care provider, life sciences, and government sectors. The 2016 survey included 600 US primary care and specialty physicians and had new questions on MACRA. The national sample is representative of the American Medical Association (AMA) Masterfile with respect to years in practice, gender, geography, practice type, and specialty, so as to reflect the national distribution of US physicians.
The AMA is the major association for US physicians and its Masterfile is a census of all US physicians (not just AMA members). The database contains records of more than 1.4 million US physicians and is based upon graduating medical school and specialty certification records. It is used for both state and federal credentialing, as well as for licensure purposes.34 This database is widely regarded as the gold standard for health policy work among primary care physicians and specialists, and is the source used by the federal government and academic researchers for survey studies among physicians. We selected a random sample of physician records with complete mailing information from the AMA Masterfile, and stratified it by physician specialty, to invite participation in an online 20-minute survey.
Value-based payment models and CMS pilot definitions
To better understand the factors most likely to contribute to physicians’ participation in value-based payment models, we built a regression model that uses a combination of demographics, practice setting characteristics, and measures of tools and resource availability to predict willingness.
We first classified physicians into three segments based on their willingness to participate in value-based payment models under two hypothetical scenarios. In the first scenario, participation is incentivized: If physicians accept any of the value-based options, they are guaranteed a 5 percent increase in reimbursement. By contrast, the second scenario includes the possibility of reduced reimbursement if physicians remain in FFS.
We classify physicians as willing (56 percent in our sample) if they chose to participate in value-based care models under both scenarios. The resistant segment (24 percent) is defined as those who opted for FFS under both scenarios. Physicians who are on the fence (20 percent) chose value-based arrangements in one scenario but not in the other.
The classification of physicians into segments is a robust way to define willingness to participate. For instance:
We then use an ordinal logistic regression model to predict willingness to participate in value-based payment models on the basis of demographic, practice setting, and resource availability characteristics. The regression-based approach to predict segment membership helps identify the individual contribution of each variable when several factors are related to each other. Our final model included eight independent variables, of which seven were statistically significant (individually and jointly) and the overall model is statistically significant (based on the likelihood ratio test and score test). The regression model’s percent concordance is 65 percent and c-statistic value is 0.65, suggesting a moderate level of prediction. The model suggests a significant association of demographics and practice setting with the willingness to adopt new value-based payment models. The model also highlights the ability of certain levers, such as access to care pattern reports and clinical protocols, to assist physicians in adoption of value-based care models. Table 2 shows the results of the regression model to predict segment membership.
An analysis of the demographics of the three segments shows similarities and important differences (table 3). The largest differences in demographics for the three segments are with practice setting and practice size. The resistant segment has a higher proportion of physicians in solo practice and small practice than the willing and on the fence segments.
Physicians in the willing segment are more likely to have access to data-driven tools than physicians in the other two segments.