In the past few years, BCG has conducted extensive international research on national outcomes registries—databases that systematically capture information on outcomes for all or nearly all of the patients in a country, region, or network of clinical sites with regard to a particular disease, condition, or medical procedure. The critical finding of our research is that these data make it possible to identify variations in health outcomes across clinical sites, analyze the root causes of those variations, and codify best practices (that is, those that produce the highest-quality outcomes). Making the data and analysis transparent—in the first instance, to clinicians, and eventually also to the public at large—then becomes the catalyst for the definition of new standards, the dissemination of best practices across the system, reductions in outcomes variation, and improvements in median health outcomes over time. (See Exhibit 2.) In addition, systematic quality improvement of this type often has the positive side effect of lowering total health-care costs for some medical conditions and procedures, because unnecessary procedures are eliminated, expensive complications occur less frequently, and repeat treatments are avoided by “getting it right the first time.”
Although the U.S. has a number of excellent outcomes registries, the complexity and fragmentation of the U.S. health-care system has limited their spread. At the moment, there is no coordinated national system in U.S. health care for tracking health outcomes by clinical site.
What would be the potential economic impact of a comprehensive health-outcomes reporting system in the U.S.? The very lack of systematic outcomes data at the disease level makes such a calculation extremely difficult. Still, it is possible to put some broad parameters around the question.
We know that there is tremendous variability in health care outcomes and health care costs across the U.S. health system. Both Medicare and the federal government’s Agency for Healthcare Research and Quality (AHRQ) have reported considerable variation in outcomes across regions and clinical sites.4 And the Dartmouth Atlas of Health Care has documented enormous variation in average Medicare spending per patient across U.S. health-referral regions.
One much-studied condition for which there are reasonably good data is acute myocardial infarction (AMI), or heart attack. Not only does the 30-day mortality rate for Medicare AMI patients vary widely by hospital, but there is little correlation between outcomes and costs. (See Exhibit 3.)
AMI is also one condition for which the existence of an outcomes registry has been shown to contribute to improvements in health outcomes. For example, the existence of a comprehensive AMI registry since 1998 in Sweden correlates with improvements in best practice and a continuous decline in that country’s 30-day AMI mortality rate. Although some inevitable variation across Swedish clinical sites still exists, the standard deviation has also narrowed considerably over time—from a high of 3.49 percentage points in 2000 to a low of 1.31 percentage points in 2011. (See Exhibit 4.) For comparison purposes, Sweden reduced its mean 30-day mortality rate by 62 percent from 2000 through 2008, compared with a reduction of only 26 percent in the U.S. during the same time period.
What would be the impact of similar improvements in the U.S.? We estimate that if the poorer-performing half of U.S. hospitals treating Medicare patients in 2011 had been able to achieve the 2009 U.S. median 30-day AMI mortality rate, the result would have been roughly 6,300 additional quality-adjusted statistical life-years, representing an economic value of $1.1 billion.
Additional value could also be expected in the form of costs avoided. For example, total Medicare spending on hospital-based AMI-related procedures was approximately $1.4 billion in 2008. In that year, if the more expensive half of U.S. hospitals had achieved the median cost per AMI admission, Medicare would have saved $146 million—about 11 percent of Medicare spending on AMI procedures. And if the poorer-performing half of U.S. hospitals had achieved the median AMI 30-day readmission rate, Medicare would have saved an additional $34 million in avoided readmission costs, another 2.5 percent of spending.
At first glance, these absolute dollar amounts may seem small. However, the diagnosis-related group (DRG) categories for AMI include patients receiving only diagnostic procedures (and therefore exclude angioplasty and bypass) and apply only to the hospital-related reimbursement for the condition—not to the total cost, including transportation, emergency-room charges, physician payments, additional diagnostics and procedures, and follow-up care. Nor do these DRG categories encompass the broader costs of treating the underlying disease. Thus, our analysis addresses only the tip of the iceberg—but even here we see considerable potential value in reducing variation and improving outcomes.
What might be the economic impact of extending such savings throughout the entire U.S. health-care system? For an order-of-magnitude estimate, consider that if Medicare spending in the country’s 306 health-referral regions that are currently above the median were reduced to the 2009 median, the result would be an 8 percent savings in total Medicare spending. Applying that 8 percent figure to total U.S. health-care spending would result in cost savings in the neighborhood of $200 billion. This amount is a conservative estimate in that systematic measurement of clinical outcomes and costs per patient group would also significantly contribute to additional savings by eliminating unnecessary costs.
Based on calculating the value of lowering the cost of treating Medicare patients in the most expensive health-referral regions to the median cost levels. Data on cost variation were 2009 data, adjusted for price, age, race, and sex, from the Dartmouth Atlas of Health Care