In light of strong relationships between procedure volume and outcomes

Transcription

1 Regional Availability Of High- Volume For Major Surgery Many patients continue to undergo high-risk surgery at hospitals with inadequate experience in performing their procedure. by Justin B. Dimick, Samuel R.G. Finlayson, and John D. Birkmeyer ABSTRACT: Despite evidence of increased risks, a large number of patients still have surgery in low-volume hospitals. To better understand why, we used Medicare data to study the regional availability of high-volume hospitals. More than half of patients undergoing three procedures in low-volume hospitals lived in regions lacking a high-volume hospital. Some regions simply lacked enough cases to support a high-volume hospital. Other regions had enough cases but too many hospitals performing them. Although consolidation of surgical services may be feasible in some settings, volume-based referral strategies are impractical for many U.S. regions. In light of strong relationships between procedure volume and outcomes for high-risk surgery, there have been numerous efforts aimed at directing patients to high-volume hospitals or surgeons. 1 The Leapfrog Group, a large coalition of public and private employers, is using a variety of financial incentives to steer patients to high-volume hospitals for five selected procedures. More broadly, a diverse collection of groups including Internet provider-rating services (such as patient advocacy groups, and the lay media are advising patients to consider volume as they decide where to undergo other procedures. 2 Despite growing attention to this issue, however, many patients continue to undergo high-risk procedures at low-volume hospitals. 3 For example, between 1994 and 1999 nearly half of Medicare patients who had pancreatic resection underwent surgery in hospitals that performed three or fewer of these operations each year. Patients in these hospitals experienced mortality rates severalfold higher than their counterparts in higher-volume hospitals. 4 Although other factors are no doubt at work, many surgical patients, particu- Justin Dimick (Justin.B.Dimick@Dartmouth.edu) is a postdoctoral research fellow and Samuel Finlayson, an assistant professor of surgery, in the Veterans Affairs (VA) Outcomes Group, VA Medical Center, in White River Junction, Vermont. John Birkmeyer is the George D. Zuidema Professor and Chair, Surgical Outcomes Research, University of Michigan Medical Center, in Ann Arbor. HEALTH AFFAIRS ~ Web Exclusive VAR-45 DOI /hlthaff.var Project HOPE The People-to-People Health Foundation, Inc.

2 Regional Variations larly those residing in rural areas, may simply not have ready access to hospitals that have adequate experience with high-risk surgery. To test this hypothesis empirically, we used methods from the Dartmouth Atlas of Health Care to explore the availability of high-volume hospitals across U.S. tertiary health care markets. 5 Study Data And Methods Study population. We studied the delivery of three high-risk operations (coronary artery bypass graft, or CABG, elective abdominal aortic aneurysm repair, and pancreatic resection) in the national Medicare population from 1999 to These operations have well-known volume-outcome associations and represent three of the five procedures targeted by the Leapfrog Group s evidence-based hospital referral initiative. 6 Leapfrog, a coalition of more than 150 private and public purchasers of health care, rates hospitals according to three practices that will improve the safety of health care provided to employees. One of these practices, evidence-based hospital referral, recommends selective referral to high-volume hospitals for five procedures: CABG, abdominal aortic aneurysm repair, pancreatic resection, esophagectomy, and percutaneous coronary intervention.. We defined high-volume hospitals as those meeting minimum volume thresholds set forth by the Leapfrog Group in the 2003 update to its criteria for evidence-based hospital referral. 7 We first determined the average annual Medicare volume for each hospital performing at least one of the three procedures. To extrapolate Medicare volumes to total volumes, we used data from the Nationwide Inpatient Sample to determine the proportion of the total number of cases (all payers) performed among Medicare patients. 8 Finally, the total volume was estimated by multiplying each hospital s observed Medicare volume by the overall total/medicare ratio for each operation. that had an average total volume above the Leapfrog criteria from 1999 to 2001 were considered high-volume. We studied patients access to high-volume hospitals in each of the 306 U.S. hospital referral regions (HRRs). As defined for the Dartmouth Atlas of Health Care, HRRs reflect distinct, naturally occurring markets for tertiary health care services. 9 The methods for defining HRRs are described in detail elsewhere, but we briefly review the two main steps. In the first step, a series of hospital service areas (HSAs) representing local health care markets were defined. ZIP codes were assigned to HSAs based on where the plurality of Medicare beneficiaries received inpatient care for common medical conditions. In the second step, these HSAs were grouped into HRRs based on where the plurality of beneficiaries went for major cardiac operations. Further checks verified that all of these HRRs contained at least one hospital that performed major neurosurgical procedures. Because hospitals that perform major cardiac and neurosurgical procedures also tend to perform other high-risk surgical procedures, HRRs provide a useful approximation of the geographic boundaries of the health care markets for the procedures in our analysis. Prior analyses by the Dartmouth Atlas group have demonstrated a median local- VAR-46 7 October 2004

3 ization index (proportion of all hospitalizations occurring within a region) of 88 percent across HRRs, with a range of percent. 10 Procedures. The three included procedures vary widely according to how often they are performed, ranging from very common (CABG) to relatively rare (pancreatic resection). Patients undergoing each operation were identified using the appropriate combination of codes from the International Classification of Diseases, Ninth Revision (ICD-9). For each of the three procedures, we categorized HRRs according to the presence or absence of at least one high-volume hospital for that procedure. If the region had none and the combined total number of procedures performed by all hospitals located within the HRR fell below the Leapfrog volume threshold, we considered the HRR to have inadequate caseloads to support a high-volume hospital. Otherwise, we considered the HRR to have sufficient cases but too many hospitals. Results The total number of cases in the United States, the number of hospitals performing each procedure, and the Leapfrog volume thresholds are shown in Exhibit 1. Overall, more than half of surgical cases were performed in low-volume hospitals during for all three procedures. The availability of high-volume hospitals varied widely across geographic regions for each procedure. More HRRs had at least one high-volume hospital for CABG (42 percent) and abdominal aortic aneurysm repair (44 percent) than for pancreatic resection (16 percent) (Exhibit 2). Although many HRRs along the Pacific Coast and parts of the Intermountain West lacked high-volume hospitals for all three procedures, regional availability of high-volume hospitals tended to vary by procedure. 11 EXHIBIT 1 Description Of Three High-Risk Surgical Procedures Performed In The National Medicare Population, Procedure Total U.S. cases Number of hospitals performing procedure Leapfrog definition of high-volume hospital Percent of cases in low-volume hospitals CABG Abdominal aortic aneurysm repair Pancreatic resection 818, ,048 14,288 1,102 2,571 1, per year 50 per year 11 per year SOURCE: Centers for Medicare and Medicaid Services, Medicare Provider Analysis and Review (MEDPAR) files, NOTES: The MEDPAR files contain hospital discharge abstracts for the acute care hospitalizations of all Medicare recipients covered by the hospital care program (Part A). Because only patients covered by fee-for-service arrangements are included, our sample excludes the approximately 10 percent of Medicare patients who were enrolled in risk-bearing health maintenance organizations (HMOs) during this period. These numbers were further adjusted to represent the total number of U.S. cases using the Nationwide Inpatient Sample as described in the Methods section. CABG is coronary artery bypass graft. HEALTH AFFAIRS ~ Web Exclusive VAR-47

4 Regional Variations EXHIBIT 2 Geographic Availability Of High-Volume For Major Surgery In The United States, By Percentage Of Regions And Proportion Of Surgical Cases In Each Type Of Region, In many regions that lacked a high-volume hospital, there were enough overall cases to meet the volume threshold, but there were too many hospitals performing them. With CABG, most HRRs without a high-volume hospital had sufficient overall caseloads, but too many hospitals were performing the procedure (Exhibit 2). In contrast, with pancreatic resection (and to a lesser extent abdominal aortic aneurysm repair), most HRRs lacked sufficient overall caseloads to support even a single high-volume hospital. Most of the three procedures were performed in regions with existing highvolume centers (Exhibit 2). Overall, 71 percent of patients resided in HRRs with high-volume hospitals for CABG, with a similar proportion for abdominal aneurysm repair and a lower proportion for pancreatic resection. When only patients treated at low-volume centers were considered, however, a disproportionate number lived in HRRs that lacked high-volume hospitals for CABG (56 percent), pancreatic resection (64 percent), and abdominal aortic aneurysm repair (45 percent) (Exhibit 3). Discussion Despite growing evidence of strong, inverse relationships between volume and mortality rates with certain procedures, many patients continue to undergo highrisk surgery at hospitals with inadequate experience. As a result, there are a large number of potentially avoidable surgical deaths in the United States. For example, the Leapfrog Group estimates that more than 2,000 deaths could be averted each VAR-48 7 October 2004

5 EXHIBIT 3 Number Of Patients Having Surgery In Low-Volume Within Each Type Of Region, Availability of high-volume hospitals within the region Total cases in lowvolume hospitals ( ) At least one highvolume hospital Enough cases but too many hospitals Inadequate cases for high-volume hospital Procedure Number Percent Number Percent Number Percent Number Percent CABG Abdominal aortic aneurysm repair Pancreatic resection 428,762 77,533 8, ,152 43,068 3, ,152 26,221 3, ,458 8,244 2, SOURCE: Centers for Medicare and Medicaid Services, Medicare Provider Analysis and Review (MEDPAR) files, See Exhibit 1 Notes for details on this data set. NOTE: CABG is coronary artery bypass graft. year in the United States with five procedures alone if patients had surgery at hospitals that met their volume standards. 12 While volume-based referral may save lives, these policies have been criticized on several fronts. 13 Some argue that volume standards create obvious incentives for hospitals to perform more cases, which could lead to unnecessary surgery for some discretionary procedures. Others argue that hospitals (and their surgeons) may become less proficient with related procedures or emergency operations that must be handled locally. For example, the surgeon no longer performing elective abdominal aortic aneurysm repair may be less prepared to manage patients with ruptured aneurysms. Our study addresses another, more practical limitation of volume-based referral: Patients may not have access to high-volume hospitals. For three high-risk procedures, we found that approximately half of patients in low-volume hospitals lived in regions without a high-volume center. Many of these regions had enough overall cases but too many hospitals sharing them. In other regions, however, there were simply not enough cases to support a single high-volume center. These findings suggest the need to consider access and geography as volume-based referral strategies are implemented. Study limitations. We should consider several limitations of our analysis. First, the geographic units on which our analysis were based HRRs from the Dartmouth Atlas project were created based on referral patterns for cardiac surgical and neurosurgical care. Although they may not perfectly reflect health care markets for other types of tertiary care, it is likely that surgical treatment of peripheral vascular and major cancer surgery share similar referral patterns. Further, previous analyses from the Dartmouth Atlas project suggest that a small proportion of Medicare patients are hospitalized outside their HRRs. 14 Second, the availability of a high-volume hospital within an HRR may be considered a blunt indicator of access. HEALTH AFFAIRS ~ Web Exclusive VAR-49

6 Regional Variations Despite considerable media attention to this issue, many patients are likely unaware of the importance of procedure volume. HRRs vary extensively in their geographic size, from small urban centers to more rural regions that are hundreds of miles wide. It is likely that many patients residing in HRRs without high-volume centers live within short distances to such hospitals of adjoining HRRs. Conversely, high-volume hospitals in large, rural HRRs may be too far away for some patients living in those regions. These issues could be addressed more directly by examining patient travel times. 15 Policy implications. Room to move on volume-based referral. Almost half of all patients undergoing surgery in low-volume settings live in regions already served by high-volume hospitals. It is worth considering why this is the case. First, some patients may be making informed decisions to undergo surgery at their local lowvolume hospital. One study, based on hypothetical scenarios, showed that some patients would accept increased operative mortality risks locally instead of traveling to a distant referral center. 16 However, it is not clear whether patients would have voiced the same preferences if faced with real decisions or if the designated referral center was nearby. Second, patients may be making uninformed decisions. Despite considerable media attention to this issue, many patients are likely unaware of the importance of procedure volume with some procedures or how to obtain volume information for hospitals or surgeons in their area. 17 This problem could be addressed by better educational efforts by private health plans, public-sector health agencies, and consumer advocacy groups. Third, and perhaps most likely, patients at low-volume hospitals may be deferring decisions about where to have surgery to their doctors. The advice of regular or referring physicians is the most important determinant of where patients receive their hospital-based care. 18 Referrals made by many physicians are no doubt influenced more by habit and their desires to support their local hospitals and surgeons and less by information about comparative hospital performance. 19 If physicians are primarily responsible for steering patients to low-volume centers for high-risk surgery, educational efforts aimed at patients and their families will have limited effect. Creating real change in referral patterns may require selective contracting and other financial disincentives set by payers and purchasers. However, our findings suggest that volume-based referral initiatives with teeth should be targeted at regions already served by high-volume hospitals. To minimize access problems, the Leapfrog Group exempts hospitals in rural areas from its evidence-based hospital referral initiative. Although correct in concept, this rural-versus-nonrural dichotomy may be inadequate. Our analyses (including data not shown) suggest that many primarily rural HRRs have existing highvolume centers, while numerous nonrural HRRs do not. VAR-50 7 October 2004

7 Our study suggests that many more regions would have high-volume hospitals if high-risk cases were consolidated in fewer hospitals. However, it is not clear how to make this happen. Educational efforts and public reporting may influence some patients, but these measures are unlikely to bring about wholesale redistributions in surgical caseloads. 20 States could impose more stringent certificateof-need criteria for high-risk procedures, but this tool is better designed for limiting the proliferation of new centers than for curtailing the practices of existing centers. Many smaller and less populated regions simply lack enough overall caseloads to support even a single high-volume center. Patients in these regions could conceivably travel out of region for selected high-risk procedures. Many patients, particularly those who are relatively affluent and educated, already travel considerable distances to undergo specific procedures at national centers of excellence. Payers and policymakers could consider models for extending this opportunity to other patients in areas lacking centers with appropriate experience. Need for alterative approaches to improving surgical quality. Efforts to improve surgical quality will need to look beyond volume-based referral alone. As outlined in this paper, such strategies are impractical in many parts of the United States lacking high-volume centers. Moreover, although volume is clearly linked to lower mortality rates with many procedures on average, it remains a poor predictor of individual hospital performance. Many low-volume centers have very good performance, while some high-volume hospitals are clearly poor performers. Even if this result could be achieved, getting all patients to high-volume centers would only go so far in improving surgical outcomes. For these reasons, it is worth considering alternatives to volume-based referral. Obviously, other criteria could be used to identify high-quality hospitals in regions lacking high-volume centers. For example, hospitals with high volumes for procedures other than the targeted operation could be chosen. New evidence has shown that the volume-outcome effect may not be specific to single procedures. 21 Alternatively, hospitals with certain technological capabilities, high nurse-to-bed ratios, and intensive care units (ICUs) run by board-certified intensivists are also markers for better surgical outcomes. 22 staffed by surgeons with high procedure volumes, or subspecialty fellowship training, also have excellent outcomes. 23 Unfortunately, these attributes all tend to be most associated with large, high-volume hospitals and thus may not be useful for identifying high quality in smaller hospitals. In addition to structural measures such as volume, process-of-care or direct outcome measures could be used to signify high-quality hospitals, as reflected in the latest revision of surgical standards adopted by Leapfrog. Unfortunately, validated process measures of quality are not available for most surgical procedures. And when available, they often relate to secondary outcomes (for example, prophylactic antibiotics for wound infection) instead of being directly related to mor- HEALTH AFFAIRS ~ Web Exclusive VAR-51

8 Regional Variations tality rates. Moreover, process measures often require access to clinical data, which are expensive and not widely available. Direct outcome measures also require clinical data (for adequate risk adjustment) and thus share the same limitation. Moreover, outside cardiac surgery, most operations are not performed often enough to measure outcomes precisely even if the data were available. 24 Given the numerous obstacles facing selective referral strategies, we must not overlook efforts aimed at improving surgical care where it is already happening. To the extent that superior outcomes with some procedures can be linked to discrete, exportable processes of care, outcomes could be improved by educational efforts aimed at low-volume surgeons. For example, a nationwide effort to disseminate a specific surgical technique (total mesorectal excision) to surgeons in Norway greatly reduced recurrence rates after surgery for rectal cancer. 25 The likelihood of achieving similar success with other procedures is not known. However, the existence of wide variations in surgical quality across the United States suggests that we should at least make the attempt. Justin Dimick was supported by a Veterans Affairs Special Fellowship Program in Outcomes Research. The views expressed herein do not necessarily represent the views of the Department of Veterans Affairs. NOTES 1. R. Galvin and A. Milstein, Large Employers New Strategies in Health Care, New England Journal of Medicine 347, no. 12 (2002): ; and A Milstein et al., Improving the Safety of Health Care: The Leapfrog Initiative, Effective Clinical Practice 3, no. 6 (2000): HealthGrades Inc., How to Choose a Hospital: HealthGrades Guidelines for Choosing a High-Quality Hospital, HowToChoose&re_modtype=HRC (30 July 2004); Center for Medical Consumers, 2002 Reported Volume for Selected Procedures Performed in New York State Licensed and Ambulatory Surgery Centers, (30 July 2004); and R. Mishra, Study Cites Risks of Low-Volume Surgeries, Boston Globe, 1 March J.D. Birkmeyer et al., Hospital Volume and Surgical Mortality in the United States, New England Journal of Medicine 346, no. 15 (2002): Ibid. 5. J.E. Wennberg and M.M. Cooper, eds., The Quality of Medical Care in the United States: A Report on the Medicare Program, The Dartmouth Atlas of Health Care 1999 (Chicago: AHA Press, 1999). Complete descriptions of the methods and previous research can be found on the Dartmouth Atlas Web site, 6. Milstein et al., Improving the Safety of Health Care. 7. Leapfrog Group, Evidence-Based Hospital Referral Fact Sheet, EHR_FactSheet.PDF (30 July 2004). 8. For a more detailed description of the method used to convert Medicare volume to total, all-payer volume, see Birkmeyer et al., Hospital Volume and Surgical Mortality. 9. Wennberg and Cooper, eds., The Quality of Medical Care in the United States. 10. J.E. Wennberg and M.M. Cooper, eds., The Dartmouth Atlas of Health Care 1996 (Chicago: AHA Press, 1996). 11. This geographic variation can be seen in Supplemental Online Exhibits 1 3, available at content.healthaffairs.org/cgi/content/full/hlthaff.var.45/dc J.D. Birkmeyer and J.B. Dimick, Potential Benefits of the 2003 Leapfrog Standards: Effect of Process and Outcomes Measures, Surgery 135, no. 6 (2004): J.D. Birkmeyer, Should We Regionalize Major Surgery? Potential Benefits and Policy Considerations, Journal of the American College of Surgeons 190, no. 3 (2000): VAR-52 7 October 2004

9 14. Wennberg and Cooper, eds., The Dartmouth Atlas of Health Care J.D. Birkmeyer et al., Regionalization of High-Risk Surgery and Implications for Patient Travel Times, Journal of the American Medical Association 290, no. 20 (2003): S.R. Finlayson et al., Patient Preferences for Location of Care: Implications for Regionalization, Medical Care 37, no. 2 (1999): Mishra, Study Cites Risks ; A. Comarow, Higher Volume, Fewer Deaths, U.S. News and World Report (17 July 2000): 68 70; and L. Marsa, Not Every Hospital Is Created Equal, Los Angeles Times, 25 September M.N. Marshall et al., The Public Release of Performance Data: What Do We Expect to Gain? A Review of the Evidence, Journal of the American Medical Association 283, no. 14 (2000): E.C. Schneider and A.M. Epstein, Influence of Cardiac-Surgery Performance Reports on Referral Practices and Access to Care: A Survey of Cardiovascular Specialists, New England Journal of Medicine 335, no. 4 (1996): P.S. Romano and H. Zhou, Do Well-Publicized Risk-Adjusted Outcomes Reports Affect Hospital Volume? Medical Care 42, no. 4 (2004): ; and E.C. Schneider and A.M. Epstein, Use of Public Performance Reports: A Survey of Patients Undergoing Cardiac Surgery, Journal of the American Medical Association 279, no. 20 (1998): D.R. Urbach and N.N. Baxter, Does It Matter What a Hospital Is High Volume For? Specificity of Hospital Volume-Outcome Associations for Surgical Procedures: Analysis of Administrative Data, British Medical Journal 328, no (2004): D.A. Alter et al., Long-Term Myocardial Infarction Outcomes at With or Without On-Site Revascularization, Journal of the American Medical Association 285, no. 16 (2001): ; L.H. Aiken et al., Hospital Nurse Staffing and Patient Mortality, Nurse Burnout, and Job Dissatisfaction, Journal of the American Medical Association 288, no. 16 (2002): ; and P.J. Pronovost et al., Physician Staffing Patterns and Clinical Outcomes in Critically Ill Patients: A Systematic Review, Journal of the American Medical Association 288, no. 17 (2002): J.D. Birkmeyer et al., Surgeon Volume and Operative Mortality in the United States, New England Journal of Medicine 349, no. 22 (2003): ; and B.E. Hillner et al., Hospital and Physician Volume or Specialization and Outcomes in Cancer Treatment: Importance in Quality of Cancer Care, Journal of Clinical Oncology 18, no. 11 (2000): J.B. Dimick et al., Surgical Mortality as an Indicator of Hospital Quality: The Problem with Small Sample Size, Journal of the American Medical Association (forthcoming) 25. A. Wibe and the Norwegian Rectal Cancer Group, Total Mesorectal Excision for Rectal Cancer What Can Be Achieved by a National Audit? Colorectal Disease 5, no. 5 (2003): HEALTH AFFAIRS ~ Web Exclusive VAR-53