Pay‐for‐performance for chronic medical conditions, such as hypertension, is widely cited as an important step in potential health care reform. By re‐aligning financial incentives, pay‐for‐performance has the potential to improve quality and reward practitioners who provide that quality, but it also has the potential to encourage unintended consequences that could negatively impact overall quality of care. One concern is that practitioners who see more complicated patients with multiple comorbid conditions will have greater difficulty in achieving adequate blood pressure (BP) control. If true, this could lead to the unintended consequence of having practitioners, who provide care to the most complicated and difficult patients, be financially penalized. In the extreme, one could even set up a perverse incentive for “patient shopping” where practitioners would be incentivized to seek out patients with uncomplicated hypertension and resist attending to those who have more difficult‐to‐control disease. While pay‐for‐performance in not currently a part of the Veterans Administration (VA) health care delivery system, the goal of this analysis from 8 VA centers across the United States was to determine the impact of different coexisting chronic medical conditions on measured quality of care for hypertension.
Using the VA National Patients Care database, VA investigators retrospectively identified veterans with documented hypertension who received primary care at 1 of 8 VA facilities in 2005. Using diagnosis and procedure codes as well as laboratory and pharmacy data, the investigators identified specific coexisting chronic medical conditions for each patent with hypertension. These diagnoses were termed either concordant or discordant with the diagnosis of hypertension. The concordant conditions chosen (somewhat arbitrarily) were diabetes, dyslipidemia, and ischemic heart disease. The discordant conditions (also chosen somewhat arbitrarily) were arthritis, chronic obstructive pulmonary disease, and depression. For the purposes of this analysis, patients with hypertension were subsequently divided into 4 groups: (1) no other comorbid conditions, (2) only hypertension‐concordant conditions, (3) only hypertension‐discordant conditions, and (4) both hypertension‐concordant and hypertension‐discordant conditions. The primary objective of the study was to determine and compare the rates of BP control and quality care for hypertension in each of these 4 groups. Quality care for hypertension was defined as BP at the index visit of <140/90 mm Hg or any one of the following occurring within 6 months of the index visit: initiation of an antihypertensive medication, increasing the dose of current antihypertensive medications, adding an additional antihypertensive medication, patient receiving at least 4 antihypertensive medications, or follow‐up BP <140/90 mm Hg. The same BP goal of <140/90 mm Hg was used for patients with diabetes or chronic kidney disease, in keeping with the VA performance assessment program in place since 2005. In addition to the above measures of quality, patient satisfaction was also determined and compared for each group using a standardized questionnaire administered by mail.
Overall, there were 141,609 hypertensive patients in the identified cohort. Of these, 16.0% had no other comorbid conditions, 49.5% had only concordant comorbid conditions, 8.7% had only discordant comorbid conditions, and 25.9% had both types of conditions. On average these patients had 3.7 primary care and 0.8 diagnosis‐related specialty care visits during the year prior to the index visit. Not surprisingly, the mean number of primary care visits was highest in the group with both types of comorbid conditions and lowest in the group with none. BP control rates (<140/90 mm Hg at index visit) were highest in the group with both types of comorbid conditions (69.2%) and lowest in the group with none (57.3%). BP control rates were intermediate in patients with only concordant (64.7%) and only discordant (63.0%) comorbid conditions. When looking at overall quality care for hypertension, rates were once again highest in those with both types of comorbid conditions (90.4%) and lowest in those with none (80.2%); while those having only concordant or discordant comorbid conditions had intermediate quality of care. Overall, only 27.3% of the cohort did not receive quality care for hypertension.
In subsequent analyses adjusted for age alone or age and illness burden, BP control was positively associated with having identified comorbid medical conditions, and the odds of receiving quality care for hypertension increased as the number of comorbid medical conditions increased. Adjustment for the number of yearly primary care or specialty care visits did not affect these results. The majority of patients (74.6%) who completed the patient satisfaction survey reported that their hypertension‐related care was either “excellent” or “very good.” Reported satisfaction did not appear to be related to the presence or absence of BP control, the quality of hypertension care, or the number of coexisting medical conditions.
This retrospective observational study of hypertensive individuals receiving care in the VA system demonstrates that complex patients with multiple chronic medical conditions are more likely to achieve adequate BP control and receive quality care for their hypertension. These findings should reassure those who care for the most medically complex patients with hypertension but remain concerned that they will be penalized if they enroll in programs that incentivize pay‐for‐performance care.—Petersen LA, Woodard LD, Henderson LM, et al. Will hypertension performance measures used for pay‐for‐performance programs penalize those who care for medically complex patients? Circulation. 2009;119:2978–2985.
Comment
Pay‐for‐performance initiatives have the potential to align financial incentives with quality patient care and could potentially help increase BP control rates in the United States. An often‐cited example of the promise of pay‐for‐performance is the Quality Outcomes Framework (QOF) initiated in the United Kingdom in 2004. Under QOF, general practitioners receive additional compensation for achieving various clinical targets, including BP control (defined as a BP of <150/90 mm Hg). In fact, a recent study suggests that BP awareness, treatment, and control rates improved dramatically in the United Kingdom between 2003 and 2006. In 2004, only 48% of men and 44% of women had BP under control, but by 2006, 52% of men and 53% of women had achieved BP control as defined above. While other factors may have played a substantial role in achieving these improved rates of awareness, treatment, and control, it seems likely that pay‐for‐performance was an important contributor.
Although pay‐for‐performance can take many forms, the one most commonly discussed is a bonus payment for practitioners who hit certain clinical benchmarks, such as BP control. As seen in the United Kingdom, while such a system may improve quality, at least as measured by specific benchmarks, it may also have certain drawbacks. By using specific, but sometimes somewhat arbitrary benchmarks, pay‐for‐performance paradoxically may actually increase health care expenditures while not improving the overall health of the population. This may be especially true if practitioners concentrate on certain aspects of care for which they are financially incentivized (eg, BP control), but ignore other aspects of care that are not captured in the system and financially incentivized (eg, teaching to the test).
Pay‐for‐performance can also predispose to other perverse incentives, so the question posed in the title of the study is very relevant: Will hypertension performance measures used for pay‐for‐performance programs penalize those who care for medically complex patients? While the results of this analysis, which demonstrate that in the VA system more complex patients actually have higher rates of BP control and quality measures of hypertension treatment, are encouraging, this type of study cannot adequately answer the fundamental question due to unavoidable methodologic limitations. It is important to keep in mind that this was not a study of a pay‐for‐performance system, but rather an analysis of a group of patients and clinicians within the largest federal health care system that may not be representative of the US population at large. Since these were all VA patients, the cohort was older and almost exclusively male with extremely affordable access to health care providers and prescription medications. Additionally, the clinicians in the VA system are supported by a free, integrated electronic medical record, various types of electronic decision support, and clear clinical care paths and BP goals. While it did not appear that the results were confounded by the number of primary care visits each patient had, it is possible that other important potential confounders were not measured, including duration of care in the VA system, length of time since diagnosis of hypertension, number of lifetime visits for hypertension, race and ethnicity, and other important comorbid conditions that may lead practitioners to treat BP more aggressively (such as chronic kidney disease or cerebrovascular disease).
These limitations aside, the findings of this study are reassuring that a system of pay‐for‐performance, if properly designed, may be able to incentivize quality care without penalizing practitioners who care for more complicated patients. The question then becomes: what type of pay‐for‐performance system should clinicians who treat hypertension advocate?
In our opinion, there are a few important attributes that need to be addressed in any future health care delivery system for optimal hypertension treatment. First, clinicians need universal access to the types of tools that are available at the VA, including affordable electronic medical records systems, automated decision support (including clinical reminders that automatically prompt the physician when BP is above goal), clear and definitive guidelines, and affordable medication access for their patients. Second, the performance measures need to be carefully defined. Since it is easiest to measure, there will be great temptation to simply use office‐measured BP control rates to define quality. However, this runs the risk of penalizing clinicians who take care of those with the most difficult‐to‐control BP and other barriers to BP control (including patient adherence to medications). Clinicians should be compensated not just for achieving adequate BP control, but also for avoiding therapeutic inertia in patients whose BP is not controlled. A good faith effort to uptitrate antihypertensive therapy, either by increasing doses or adding another antihypertensive medication in a timely manner should also be incentivized. Third, in order to avoid overtreatment of patients who may not benefit from more aggressive care, determination of BP control needs to be based not just on office readings, but instead needs to incorporate appropriate out‐of‐office readings as well. Finally, the amount of additional compensation given for quality care should be sufficiently large to allow clinicians (particularly primary care practitioners) to shift their business model from a fee‐for‐service paradigm, where they are only compensated for performing a face‐to‐face visit in their office, to a quality‐based medical home environment where coordination of care and use of allied health personnel to help deliver more cost‐effective care, including telephone, fax, and Internet become financially viable.
By aligning financial incentives with quality care, pay‐for‐performance has the potential to improve outcomes in patients with hypertension and to fundamentally change the way clinicians deliver care for their patients with hypertension. Since hypertension is the most common chronic medical condition for which patients take daily prescription medications, it seems apparent that some type of pay‐for‐performance for BP management is going to be a core component of health care reform in the near future. Clinicians treating hypertension, therefore, need to become actively involved in helping to determine the details of a pay‐for‐performance system for the treatment of hypertension. From the results of this study, it would appear that physicians who care for the most medically complex patients with hypertension will not be penalized if they participate in programs that incentivize pay‐for‐performance care.