Jan Willem Kuenen
Addressing Unwarranted Differences in Treatment Decisions
A patient walks into a doctor’s office. He has severe back pain and is diagnosed with a herniated disc. Treatment options range from rest and physical therapy to invasive back surgery.
What the patient does not realize is just how much his treatment depends on the doctor he happens to see. A patient who visits one doctor can be up to five times more likely to receive surgery than a patient who visits another doctor. The option that will yield the better result is not always clear to the doctor, and this lack of scientific data on outcomes produces huge “practice variations” in treatment decisions–—with serious consequences in terms of health care costs and quality. This is a big problem not only for patients but also for payers, which end up footing the bill for what could be suboptimal care.
Back pain is just one example. Practice variation is commonly seen in the treatment of many other conditions, such as prostate cancer, and in decisions to perform such procedures as cataract surgery, cesarean section, and hip and knee replacement. Depending on where a patient lives and the doctor he sees, there can be more than a fivefold variation in decisions regarding hospitalization and surgery or other invasive treatments, even after correcting for differences in patient populations. The variation exists among countries, among regions within a country, among hospitals within a region, and even among doctors at the same hospital.
Payers—both public payers and private health insurers—are extremely well positioned to help reduce unwarranted practice variation. Payers have the data, the incentives, and the role in the value chain to make a difference. In our work with payers and our ongoing research in this area, we have looked at the underlying drivers of unwarranted practice variation and mapped out a strategy to help payers mitigate its effects and thereby to improve health care quality and reduce health care costs. By facilitating better decision making, payers can lower costs by €100 million to €200 million for every million lives covered, while at the same time improving patient outcomes. (This assumes an average health-care cost of €2,000 per individual covered for payers in the developed world.)
Variation in medical treatments, or practice variation, is very common.1 Some studies reveal more than a fivefold variation in treatment decisions. Most studies compare regions with the highest and the lowest rates for particular modes of treatment; however, they do not necessarily correct for population differences such as age, gender, and socioeconomic status; failing to do so can lead to overestimates. We used data corrected for these population differences and evaluated differences between regions—in multiple countries—in the tenth and ninetieth percentiles for the use of particular treatment modes, in order to exclude extreme outliers.
Using this more conservative approach, we still found that treatment intensity for a given condition routinely varies by a factor of two or more among regions. For instance, a Medicare patient suffering from back pain who lives in Salt Lake City (a region in the ninetieth percentile for treatment intensity for back pain) will be 2.1 times more likely to receive back surgery than a comparable Medicare patient living in Napa, California (in the tenth percentile). And these differences are just as common and significant in other developed markets across the world.
These practice variations can be seen in the case of most diseases. The exception is where treatment is unavoidable, where the treatment choice is clear, and where differences in provider judgment are negligible, such as in hospital admission rates for hip fractures. However, these scenarios account for only about 15 percent of provided care.2
The 2014 Organisation for Economic Co-operation and Development (OECD) report Geographic Variations in Health Care confirms the existence of significant variations in treatment between and within 13 OECD member countries.3 (See Exhibit 1.)
Optimal treatment requires informed decision making, for which three factors must converge: clear scientific guidance, patient involvement, and physician judgment and experience. (See Exhibit 3.) When these three factors come together, physicians make more-informed treatment decisions for their patients, diminishing unwarranted practice variation. It also helps to have a system with financial incentives in place to reward high-value care rather than high-volume care, though this alone is not enough.
Health care payers, public and private, can shed light on unwarranted practice variation by analyzing and leveraging their huge and comprehensive data sets. This data can help create transparency of outcomes, inform providers on best practices, and guide contracting decisions. More generally, payers can use their unique position within the health care value chain to facilitate optimal treatment decisions and minimize perverse incentives that erode quality and value. We have identified four levers that can help reduce practice variation: inform and align providers, empower and involve patients, improve scientific guidance, and align incentives. (See Exhibit 5.) For each lever, we recommend several concrete actions that health payers can take.
Although cost saving is just one of the benefits that will come from reducing practice variation (maximizing value to members is the ultimate goal), it provides a useful shorthand to illustrate the potential rewards for payers. We have seen that full implementation of the four levers can lead to savings of €100 million to €200 million per million covered lives over three to five years, or 5 to 10 percent of total health-care costs. Even for payers that implement just one or two of the levers, the impact can be very significant. Here is a breakdown of potential cost savings:
Not all of these recommendations can be achieved in the short term. Levers one and two could lead to short- to medium-term results (one to three years), whereas levers three and four likely won’t deliver results for three or more years. (See Exhibit 9.) The precise savings over time will also vary depending on several factors, including market specifics (such as system dynamics and the level of trust between the payer and providers), the forcefulness with which the payer implements the levers, and the order of implementation. Nonetheless, it is important to make these investments now—and leading payers are indeed doing so already.
The authors would like to thank their BCG colleagues Jennifer Clawson, Joachim Engelhard, Josephine Linthorst, Gabriel Osterdahl, and Neil Soderlund for their contributions to the research.
Payers have historically focused on a very narrow definition of their core business: reimbursement and administration of health care costs. However, some are now broadening their role to ensure that their members receive optimal outcomes for all spending. Leading payers are changing from passive health-care reimbursers to active health-care facilitators—and rightfully so. (See the exhibit below.) By focusing only on reimbursement and administration, payers can optimize just 3 to 10 percent of their business, but by addressing the value of care delivered, they cover the full scope of what they pay for on behalf of their members. Leading payers are becoming more active in three areas:
In developed nations, prostate cancer is the most common cancer in men, and 3 percent of men overall in those nations die from this disease. However, an astonishing 50 percent of men who die from another cause have been living with prostate cancer meaning many more die with it than from it. That is, men over a certain age who have been diagnosed with localized disease are more likely to die from something else.
Physicians treating prostate cancer have a complex set of options. The most intensive treatment—prostatectomy—leads to better survival in many cases, especially for high-risk patients, but also runs risks such as incontinence and sexual dysfunction. Other treatment options (such as radiation therapy, chemotherapy, and hormone therapy) present a different set of risks and likely outcomes. Watchful waiting has much lower complication risks but a potentially higher mortality rate for specific groups.
Because individual preferences differ, it is essential that patients have the opportunity to consider how treatments will affect their quality of life. One patient may opt for surgery based on a strong desire to see his grandchildren grow up, while another may choose watchful waiting owing to a strong desire to avoid incontinence or sexual dysfunction. These preferences should be part of the treatment discussion.
Unfortunately, this is often not the case, as we can see from the high regional practice variation across and within countries. Dartmouth Atlas of Health Care data shows that prostatectomy rates in the U.S. vary by as much as 2.5 times (age-standardized and corrected for outliers) from one region to the next, and analyzing treatment mix reveals all kinds of variations, including emphasis on delayed treatment, emphasis on one treatment option, and an equal mix of all treatment options. (See the exhibit below.)
Tools to counteract these variations exist.
First, the International Consortium for Health Outcomes Measurement has developed a standard set—that is, minimum sets of outcomes and risk factors that it recommends all providers track—for localized prostate cancer. This tool provides an internationally agreed upon method for measuring outcomes that matter to patients, so providers and payers can compare performance globally and the impact of treatment choices on patient outcomes can be better understood.
Second, decision aids can be used to help patients weigh pros and cons. The National Health Service in the UK already provides Web-based decision-making tools for localized prostate cancer and these aids should provide up-to-date outcome data per treatment option.
Third, when providers actively solicit patient preferences, decisions are more in line with the patient’s life goals. Typically, this reduces costs as well, as patients tend to choose more conservative treatments.
The prostate cancer example shows how practice variation can be managed within a single patient group. In the future, international, data-based benchmarks, communicated through decision aids, may be used to engage patients in a sensitive treatment-decision process. This can consistently produce greater total health-care value at a lower overall system cost with higher satisfaction for individual patients.
Health care leaders around the world are increasingly committed to providing value-based health care. However, standardized metrics on patient outcomes for many common medical conditions still don’t exist. To fill this gap, BCG, Sweden’s Karolinska Institute, and professor Michael E. Porter of Harvard Business School jointly established the International Consortium for Health Outcomes Measurement (ICHOM; www.ichom.org). ICHOM is a not-for-profit organization that brings together many of the most experienced clinical experts in the world to establish standard sets of outcomes for the most important diseases.
At the time of this writing, ICHOM has developed 12 standard sets—for coronary artery disease, localized prostate cancer, low-back pain, cataracts, Parkinson’s disease, and more. By measuring outcomes based on the ICHOM standard sets, providers around the world will be able to generate data to help improve physician performance, drive quality-focused funding by health care payers, and—most important—improve patient outcomes. ICHOM supports implementation of these standard sets and will develop additional sets in the coming months and years. By 2017, ICHOM aims to have implemented standard sets for more than 50 of the diseases with the highest disease burden worldwide. In this way, ICHOM is working to improve patients’ quality of life worldwide.