Therapeutic Inertia and the Medicare Crisis

According to the National Bipartisan Commission on the Future of Medicare, Medicare Part A (which pays for hospitalization) will go broke in 2008. ¹ A recent article by Okonofua et al. ² stated that 87% of clinic visits during which blood pressure (BP) was >140/90 mm Hg resulted in no change of medications being recommended by the treating physician, a phenomenon referred to as therapeutic inertia (TI). What, if anything, do these two statements have in common? Let us look at each of them in turn.

THE MEDICARE CRISIS

By far the biggest expenditure for Medicare is on people with chronic conditions, who account for 88% of all prescriptions filled, and 72% of physician visits. ³ There are more than 100 million Americans with at least one chronic condition, and with the aging of the baby boomers, that number goes up every year. ⁴ So what are these chronic conditions? By far, the most common is hypertension (26% of the population), followed by arthritis (20%), respiratory disorders (19%), and others such as cholesterol disorders (13%), heart disease (11%), and diabetes (10%); the last 3 of which are, of course, closely related to hypertension. ⁴ The most common cause of hospitalization is heart failure, ⁵ of which the 2 leading causes are coronary heart disease and hypertension (particularly in women). According to the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 7), treatment of hypertension can reduce heart failure by 50%. ⁶

The control rates of hypertension are well known. In the latest analysis ⁷ of the National Health and Nutrition Examination Survey (NHANES) data, 28.7% of the US population was estimated to have hypertension, an increase of 3.7% from the previous estimate 10 years earlier, and hypertension was controlled (BP <140/90 mm Hg) in 31% of people. ⁷

One proposed solution to this problem is the Medicare Modernization Act, which among other things has the goal of incorporating disease management programs into Medicare for the more cost‐effective treatment of patients with chronic conditions. Part of the Medicare reform process is a series of “pay for performance” initiatives that have been developed in collaboration with organizations such as the American Medical Association (AMA) and the Agency for Healthcare Research and Quality (AHRQ) under the umbrella of the Physician Consortium for Performance Improvement (www.physicianconsortium.org). The Consortium's stated vision is to “fulfill the responsibility of physicians to patient care, public health, and safety by becoming the leading source organization for evidence‐based clinical performance measures and outcomes reporting tools for physicians.” The recommendation for hypertension is that 2 or more readings should be taken at each visit, and physician performance is evaluated by both the number of measurements made at each visit and the proportion of patients whose pressure is below the standard goal levels. The Consortium also provides prospective data collection forms in which the BP data can be entered, and 3 options for the plan of care: recheck BP at a later date; initiate or alter medical therapy; and initiate or alter nonpharmacologic therapy.

THERAPEUTIC INERTIA

This brings us back to TI. The Okonofua study ² surveyed clinical records of 7253 hypertensive patients from family practices in South Carolina in 2003, and defined TI as a visit during which the BP was >140 mm Hg systolic or >90 mm Hg diastolic, with no increase of medication. To be included in the study, there had to be at least 4 visits during the year, at one or more of which the BP was elevated. For each patient the TI score was derived as the difference between the observed and expected medication changes—thus, a high score indicated that the patient's physician had recommended few changes. The overall control rate of BP was about 40%, a little below the estimated national average of 53% for treated hypertensive patients. ⁷ The main finding was a linear relationship between the TI scores and the degree of BP control. For the highest quintile of TI (ie, the least likely to increase antihypertensive medications), the control rate was 25% at the first visit and 6% at the last. In contrast, for the lowest TI quintile, the control rate increased from 53% to 75%.

The concept of TI was first put on the map in a much quoted paper by Berlowitz et al, ⁸ who surveyed Veterans Affairs (VA) clinics, and found that 40% of hypertensive patients had BPs >160/90 mm Hg, despite an average of 6 annual clinic visits. Increases in medications were prescribed in only 7% of visits. This state of affairs is not likely to be due to doctors' ignorance about the goals of treatment: in a survey conducted at about the same time as the VA study, 97% of physicians knew very well that the goal BP was 140/90. ⁹ Nonetheless, knowledge does not necessarily dictate behavior. In another study ¹⁰ of 21 primary care physicians in the United States (who were also quite familiar with the guidelines) the systolic pressure of their hypertensive patients was >140 mm Hg at 93% of visits. Various reasons were given by the doctors to explain their inertia, such as saying that they were satisfied with the BP response, it was only the systolic pressure that was high, or that the hypertension was only borderline. Another factor that may account for TI is physicians' overestimation of how well they adhere to guidelines, which has been demonstrated in several studies and for several disease conditions, including hypertension, lipid management, and diabetes. ¹¹ It is well recognized that to change people's behavior, education is necessary but not sufficient. The crucial ingredient here is motivation.

WHAT WILL IT TAKE TO CHANGE PHYSICIANS' BEHAVIOR?

Adopting any new behavior pattern requires 2 components: information and motivation. We must know how to accomplish the change, and we must want to do it. The traditional approach for changing physicians' behavior has been education. The amount of technical information that we need to do our jobs well is huge, and growing every day, and the adoption of new behavior patterns will not occur if we do not know how to achieve them. But it has been shown in numerous studies that education on its own does not necessarily lead to behavior change. In the case of hypertension, numerous surveys have shown that physicians are aware of the guidelines, ¹⁰ which suggests that educating physicians will not necessarily lead to better BP control.

Another basic principle for learning a new behavior pattern or skill is providing feedback. If we do not know how well (or badly) we are performing, it is going to be difficult to make any improvement. This principle is obvious for learning athletic skills, but less so for health‐related behavior patterns. Nevertheless, it is still relevant. A good example is weight loss, where self‐monitoring (regular weighing) is a critical component of success. ¹² The equivalent measure for hypertension treatment is knowing what proportion of patients are currently controlled, a statistic that most of us do not currently have at our fingertips. One of the promises of the introduction of electronic medical records is that such data should be readily available. A study done many years ago in an urban health center found that giving physicians regular reports on the percentage of their hypertensive patients whose BP was controlled resulted in an improvement in the overall rates of control. ¹³ Other studies of providing performance feedback have not been so successful. Two randomized trials, one looking at the rates of cancer screening, ¹⁴ and the other at pediatric preventive care, ¹⁵ found that feedback by itself produced no change in physician behavior. These findings are consistent with the general behavioral principle that information is necessary, but not sufficient. The missing ingredient is motivation.

In general, people's motivation can be changed in two ways—by using reward or punishment—the classic carrot‐and‐stick approach. The most obvious way of increasing the motivation of physicians to prescribe more antihypertensive medications is by rewarding them with money. Our current health care system rewards doctors for performing procedures on their patients rather than talking to them about their lifestyle or need to take medication, which may explain why there are not more hypertension specialists or general physicians in the United States. One of the Centers for Medicare & Medicaid Services (CMS)‐sponsored initiatives is a physician group practice demonstration in which groups of physicians are able to obtain performance‐based payments. The evidence that rewarding doctors financially may improve their behavior is mixed. At least 2 studies have showed a modest increase of immunization rates when doctors were given an additional payment for each shot, ¹⁶ but the financial amounts have been small, and it is reasonable to suppose that more generous payments would have bigger effects. The 2 studies quoted above that failed to find that performance feedback improved physician performance also found that combining feedback with financial rewards had no significant effect. ¹⁴ , ¹⁵ And a survey of 104 physicians from a large health maintenance organization found that only 38% supported the idea of using financial incentives to groups of physicians, and only 24% to individual physicians. ¹⁷ In contrast, 80% of the same physicians supported the idea of providing additional funding to the clinics to create systems to improve hypertension care.

Do these results mean that physicians are insensitive to financial rewards? Not likely, if the recent attention paid to physicians' conflicts of interest and undue financial influence from pharmaceutical companies is to be believed. The amounts used in the studies quoted above were relatively modest, and increased the physicians' reimbursement by about 5%–10%. To have much effect, it seems, the carrot will have to be juicier than that. But what about the stick approach? Not surprisingly, this does not appear to have been tested experimentally. Getting physicians to volunteer for a study that might reduce their reimbursement is a nonstarter. This does not mean that an organization such as CMS will not adopt such an approach, but if it does, it seems unlikely that it will be systematically evaluated before it is introduced.

CONCLUSIONS: WHAT CAN BE DONE TO IMPROVE THE SITUATION?

The implications of TI are huge: if doctors were more aggressive in their prescription of antihypertensive drugs, it should be possible to control hypertension in 65%–70% of patients, rates that actually have been achieved in some large clinical trials such as the Antihypertensive and Lipid‐Lowering Treatment to Prevent Heart Attack Trial (ALLHAT). ¹⁸ Such an improvement would reduce the risk of major cardiovascular events by 15%–20%, ¹⁹ and could significantly impact the Medicare crisis. It has also been estimated that if BP were controlled to 140/90 mm Hg in the United Kingdom (where BP control rates are even worse), the numbers of strokes and heart attacks could be reduced by one third. ²⁰

In a recent review of BP control rates in the United States achieved both in various health care systems and in clinical trials, Krakoff ¹⁸ suggested four types of intervention that might improve national rates of control. These were: (1) providing education and feedback to providers, (2) making greater use of nonphysician providers, (3) wider use of electronic medical records, and self‐monitoring of BP. His argument was that these were all components that led to the superior BP control rates seen in the clinical trials. These are all commendable, but they do not address one critical ingredient that distinguishes a clinical trial from routine practice, and that is TI. By definition, treating patients in a clinical trial requires a rigid adherence to the trial protocol, so that TI is effectively eliminated. Another difference is that patients who enroll in clinical trials are likely to be relatively compliant, so they are not necessarily representative of the general hypertensive population. The big question is, therefore, whether pay‐ for‐performance will actually lead to any improvement in BP control rates by reducing TI. The jury is still out on this, but it is an issue that urgently needs experimental testing.