how can the Medicare system (and the health care system more generally) be reformed to improve both the quality of care and the efficiency of the health care system? Do Differences In Illness Levels Explain Higher Medicare Spending? Health services use is, of course, strongly related to health status. Data from the Medicare Current Beneficiary Survey (MCBS) show that those who reported excellent health spent an average of 1.5 days per year in the hospital, while those in poor health spent an average of 4.2 days in the hospital.® There also are differences in health status across regions. We created an “illness index” that uses regional rates of heart attack, stroke, hip fracture, cancer, gastrointestinal hemorHEALT H AFFAIR S – 1 3 Februar y 200 2MEDICAR E “Greater Medicare spending does not purchase the infrastructure needed to ensure compliance with evidence-phased medicine^ rhage, and death of Medicare beneficiaries to quantify the underlying disease burden in a region. These measures were chosen because the hospitahzation records for the illnesses are accurate reflections of their true incidence in the population; nearly every elderly person with a hip fracture ends up in the hospital. (Not surprisingly, the Social Security Administration is assiduous about measuring mort’ality accurately.) Using regression analysis, we found that the health of enrollees in Grand Junction, Colorado, one of the healthiest regions in the United States, imphes that their per capita Medicare spending should be about 20 percent below the national average. By contrast, the regression suggests that those living in Birmingham, Alabama, one of the least healthy regions, should receive about 24 percent above the national average.” These estimated differences in underlying health are substantial and could be used, for example, in “risk-adjusted” regional capitation payments for ^^^^^ ^ Medicare enrollees. Still, they explain just 27 percent of the REFORM (weighted) variation in Medicare spending across regions. Conse – quently, illness-adjusted Medicare spending differs greatly across regions.^ Other studies with homogeneous patient populations (such as those with hip fracture or heart attack) confirm that substantial differences in Medicare use and spending across U.S. regions are largely independent of beneficiaries’ need for services.’ How Do Practice Patterns Differ in Higii-Spending Regions? We considered these questions by examining variations in three categories of services: effective care, preferences-sensitive care, and supply-sensitive care. The categories of care are distinguished by the relative roles of medical theory and opinion, medical evidence, the per capita supply of medical resources, and the importance and appropriateness of patients’ preferences in choosing a treatment option (Exhibit 1). • Effective care. Effective care comprises services whose use is supported by well-articulated medical theory and strong evidence for efficacy, as determined by clinical trials or valid cohort studies. The category is further restricted to interventions that virtually all patients should want as part of the contract they make with their health care systems. Effective-care indicators, based on Health Plan Employer Data and Information Set (HEDIS) measures and exHEALT H AFFAIR S – We b Exclusiv eMED/C/^R E REEOR M EXHIBIT 1 Categories Of Medical Effective care Preference-sensitive care Suppiy-sensitive care Services Factors that Influence utilization Medical theory Strong Strong Weak Medical evidence Strong Variabie Weak Per capita supply of resources Weak Variable Strong Importance of patients’ preferences Weak Strong Variable SOURCE: Authors’ analysis. NOTES: Effective care refers to services of proven effectiveness that involve no significant trade-offs—all patients with specific medical needs should receive them. Confiict between patients and providers over the value of care is minimai. Preferencesensitive care invoives trade-offs; decisions should therefore be based on patients’ preferences and values. Aithough opinions are strongly held by clinical advocates, supporting scientific evidence may be weak or strong. The effect of supply on rates of discretionary care is variable. Patients’ and providers’ vaiues are often in confiict. Suppiy-sensitive care is generaliy provided in the absence of specific clinicai theories of benefit governing the relative frequency of use. Medical texts provide littie or no guidance on when to scheduie a revisit, perform a diagnostic test, hospitalize, or admit to intensive care. However, utilization rates are strongiy infiuenced by the suppiy of resources, in some cases, patients’ preferences and values should play a central role, particuiariy for end-of-iife care. panded for the Dartmouth Atlas of Health Care, include vaccination for pneumococcal pneumonia; mammography screening for breast cancer and screening for colon cancer; eye examinations for diabetics; HgAlc and blood lipid monitoring for diabetes; and, for heart attack ••••••••••• victims, the prescription of aspirin therapy, beta-blockers, angioten- MEDICARE W99 sin converting enzyme (ACE) inhibitors and early reperfusion with thrombolytic agents, or percutaneous transluminal coronary angioplasty (PTCA), For each of these services, use rates vary extensively among hospital referral regions. For example, among patients with heart attacks who were considered “ideal candidates” for betablockers, those who actually got the needed drug ranged from 5 percent to 92 percent of patients among the 306 Dartmouth Atlas Hospital Referral Regions (HRRs), Unfortunately, most regions exhibited substantial underuse: Compliance with evidence-based practice guidehnes exceeds 80 percent of patients in only eight regions; in ten regions, compliance was less than 20 percent. The percentage of female Medicare beneficiaries (ages 65-69) who received a mammogram at least once over a two-year period (as recommended by the U.S. Preventive Services Task Force) ranged from 21 percent to 77 percent, wath all regions falling below the “bestpractice” benchmark provided by Kaiser Permanente South. The most important explanation for such variation in effective care appears to be the lack of infrastructure to ensure compliance with well-accepted (evidence-based) standards of practice. The important question for our purpose is. Does higher Medicare spending buy better quahty? Fxhibit 2 suggests that it does not. On average, there is as much underuse in high-cost as in low-cost regions, which suggests that greater spending does not purchase the HEALT H AFFAIR S – 1 3 Februar y 200 2MEDICAR E EXHIBIT 2 Use Of Effective Care, Preference-Sensitive Care, And Suppiy-Sensitive Care Among Hospital Referral Regions, Grouped By Per Enroiiee Spending Level Ratio to lowest-spending decile 2.9 2.7 2.5 2.3 2.1 1.9 _ 1.7 / ^ / y A _ l Medical specialist ” visits^ Hospitai days^ ^ ^ ^^ ^ ^^ ^ __ Percent admitted 1.5 ~ ” ‘ 1.3 Preference-sensitive 1.1 S5 ^ ™ care index Effective care index 3,000 3,500 4,000 4,500 5,000 5,500 6,000 6,500 7,000 Total Medicare spending (1996 dollars) SOURCE: Dartmouth Atlas of Health Care, 1998 and 1999. NOTES: Hospital referral regions were ranked according to per enrollee spending adjusted for age, sex, and race and put into ten groups. The exhibit gives the average per enroiiee spending in each group. Use rates for each category of utilization were calculated and expressed as a ratio to rates in thirty-one hospital service areas with lowest spending. Medical specialist visits, hospital days, and percent admitted to ICU are all measures of supply-sensitive care. The index for effective care use is the sum for rates for the eleven indicators cited in the text; the index for preference-sensitive care is the sum for rates for the ten surgical procedures profiled in the 1999 Dartmouth Atlas of Health Care. ICU is intensive care unit. ^ Care provided per decedent in the last six months of life. infrastructure needed to ensure compliance with the standards of practice dictated by evidence-based medicine. • Preference-sensitive care. Preference-sensitive care is chnical services where for many patients at least two valid alternative treatment strategies are available. Since the risks and benefits of the options differ, the choice of treatment involves trade-offs. In theory, these treatment choices should depend on informed patients’ making decisions based on the best cHnical evidence. In practice, however, treatment choices appear to be determined largely by local medical opinion concerning the value of surgery or its alternatives. For example, cardiac bypass surgery rates exhil3it about a fourfold range of variation, from three per thousand (adjusted for age, sex, and race) in Albuquerque, New Mexico, to more than eleven per HEALT H AFFAIR S – We b Exclusiv eMEDICAR E REFOR M thousand in Redding, California. The rates are strongly correlated with the numbers of per capita cardiac catheterization labs in the regions but not with illness rates as measured by the incidence of heart attacks in the region. Surgery for back pain varies even more, but the rates are not strongly correlated with supply of beds or surgeons. While there is a large body of research oh bypass surgery, there is much less for other surgical procedures. For example, the surgical decision regarding treatment of low back pain must be made in the absence of evidence from cHnical trials. It seems hkely that individual physicians’ opinions, rather than patients’ preferences, explain the more than sixfold variation in surgery rates among the 306 hospital referral regions. Indeed, regions do not show consistently high or low rates across surgical procedures, and for most procedures the patterns are not explained by the supply of surgeons. Rather, the patterns are idiosyncratic, with high rates for some discretionary procedures and low rates for others—a phenomenon we refer to as the “surgical signature.” The use of discretionary surgery is, on average, not higher in regions with greater spending (Exhibit 2). • Supply-sensitive services. In contrast to effective care and aBMHa^ preference-sensitive care, the medical theory governing decisions MEDICARE wioi about the use of hospitals as a site of care or the frequency of physician visits and diagnostic tests is much less well developed. Medical texts and journals, for example, are silent on the incremental value of three-month versus six-month intervals between physician visits for patients with such conditions as diabetes or hypertension. These sources are similarly uninformative with regard to the indications for hospitahzation, use of intensive care, and use of imaging and other diagnostic tests for patients with a host of chronic illnesses. Regions differ greatly in these measures of intensity. These variations are particularly pronounced during the last six months of hfe, a period of time when many Medicare enroUees are quite sick and which accounts for more than 20 percent of total Medicare expenditures.’” During 1995-96 the average numbers of visits to medical specialists ranged from two per decedent in Mason City, Iowa, to more than twenty-five in Miami, Florida.” The average number of days per decedent spent in hospital ranged from 4.6 in Ogden, Utah, to 21.4 in Newark, New Jersey. A similar pattern holds for admissions to intensive care units (ICUs) in the last six months of life, with nearly half of all decedents experiencing an ICU admission in Miami, Florida, compared with only 14 percent in Sun City, Arizona. These variations cannot reasonably be attributed to differences in illness: During the last six months of hfe most people are iU, regardless of where they hve. HEALT H AFFAIR S – 1 3 Februar y 200 2MEDICAR E Moreover, similarly situated communities often have strikingly dif’ ferent rates. For example, while in Sun City, Arizona, only 14 percent of decedents experience an ICU admission in the last six months of life, 49 percent and 45 percent of decedents in Sun City, California, and Sun City, Florida, respectively, do so. The local supply of medical speciahsts and acute care hospital capacity explains 41 percent of the variation in end-of-life care intensity across HRRs.’^ We therefore adopt the term “supply-sensitive” to capture these indicators of health care intensity for chronically iU patients.” The incremental Medicare dollar spent in regions with higherthan-average spending tends to be for medical specialist visits, diagnostic tests, and use of intensive care and hospitalizations for medical conditions.’”• Fxhibit 2 shows the close cori:elation between per capita Medicare spending for the entire Medicare population and the average number of specialist visits for those in their last six months of Me. Thus we view the incremental Medicare doUar as flowing not simply toward more speciahst visits in the general elderly population but, more specifically, toward specialist visits concentrated among the population with chronic and ultimately Me- ••••^^^H threatening diseases. Many of these patients do not survive and are W102 MEDICARE thus weU represented in our sample of people in their last six REFORM month s of hfe.’5 The strong associations between higher spending and greater use of supply-sensitive care, and the lack of association between more spending and more preference-sensitive or effective care, can be seen in the medical care of residents of four regions that represent either very high or very low levels of overall spending: Miami, Florida; Orange County, California; Portland, Oregon; and Minneapolis, Mirmesota (Fxhibit 3). Age-, sex-, and race-adjusted spending in Miami, for example, is 2.45 times greater than in Minneapohs. During the last six months of life the “extra” spending purchases 6.55 times more visits to medical specialists, 2.13 times more hospital days, and 2.16 times more admissions to an ICU. By contrast, rates for effective care and preference-sensitive care are slightly lower in Miami than in Minneapolis. Is More Better? We considered this question for each of the three categories of service. It seems clear that for our eleven indicators of effective care, more is better. One study suggested that regions with better quality are associated with better survival rates in the Medicare population.’* On these measures of quahty, all regions in the United States are practicing subpar medicine—use rates are too low.