Trends in Regionalization of Inpatient Care for Urological Malignancies

1. Trends in Regionalization of InpatientCare for Urological Malignancies Matthew R. Cooperberg Sanjukta Modak Badrinath R. Konety Department of Urology University of California, San Franicsco AHRQ Annual Conference Bethesda, MD September 10, 2008 A Heath Care Utilization Project Nationwide Inpatient Sample Study

2. Introduction Surgical Volume and Outcomes

3. Associating volume and outcomes • Hospital surgical volume associated with better outcomes, first noted 1979 [Luft et al. NEJM 1979; 301:1364] • Major cancer surgery [Begg et al. JAMA 1998; 280:1747] • Urologic oncologic surgery [Joudi et al. J Urol 2005; 174:432] • HCUP NIS cystectomy studies: postop mortality 2.9% vs 6.4% in highest vs lowest quintile volume hospitals [Birkmeyer et al. NEJM 2002; 346:1128] • Higher volume assoc with shorter LOS, lower charges, lower complication rates [Konety et al. J Urol 2005; 175:1695. Konety et al. Urology 2006; 68:58] • HVH status for other urologic or non-urologic surgery not assoc with outcomes [Konety et al. J Clin Oncol 2006; 24:2006]

4. “Getting under the hood” • Is hospital volume or surgeon volume more important? • Medicare analysis: surgeon volume accounts for 39% of the effect of hospital volume [Birkmeyer et al. NEJM 2004; 349:2117] • What else drives the association? • Hospital size (beds / capacity) • Urban location • Teaching mission • Staffing ratios • Patient age, LOS, other procedures Hollenbeck et al. J Urol 2007; 177:2095 Konety et al. J Urol 2004; 172:1056 Konety et al. J Urol 2005; 173:1695

5. Volume-outcomes continued… • IOM systemic review 2002: 135 studies across 27 diagnoses / procedures [Halm et al. Ann Intern Med 2002; 137:511] • In general higher volumes associate with better outcomes, but magnitude of association varies widely, as does methodological quality of studies • Provider variables may be outweighed by patient variables, perhaps insufficiently reflected in claims-based data (NSQIP investigators) [Khuri et al. World J Surg 2005; 29:1222, Best et al. J Am Coll Surg 2002; 194:257] • Secular / temporal trends • e.g., NIS analysis CABG: during period of declining volume, 50% decline in proportion of HVH, mortality declined consistently with greatest decline among LVHs [Ricciardi et al. Arch Surg 2008; 143:338] • Regionalization already supported by policy (public and private, mostly reimbursement-driven) in some cases

6. Methods

7. 15-year Trends in Regionalization • HCUP NIS data 1988-2002: accessible source of population-based data on health services trends • Bladder cancer • Renal cancer • Prostate cancer • Hospitals ranked to tertiles in each year by numbers of discharges (excluding those with no discharges) • Separate analyses of surgical and non-surgical admissions • Subset analyses by geographic region and primary payer

8. Results

9. Surgical volume thresholds

10. Trends in Admissions

11. Trends in Admissions

12. Trends in Admissions

13. Trends: Medicare / Medicaid

14. Prostate Cancer HVH Admissions

15. Prostate Cancer HVH Admissions

16. Prostate Cancer HVH Admissions

17. Prostate Cancer HVH Admissions

18. Conclusions and Implications

19. Summary of findings • About 2/3 of urologic oncology admissions at HVHs • Relative increase in regionalization • 4.5% for bladder cancer • 8.9% for renal cancer • No increase for prostate cancer but higher baseline • Substantial regional variation

20. Is regionalization a good trend? • HCUP study: hospitals meeting Leapfrog Group volume standards had similar in-hospital mortality to others. Volume standards would adversely impact low volume hospitals and increase patient travel time. [Ward et al. J Rural Health 2004; 20:344] • Many rural areas lack the referral base to support even one HVH for some procedures. [Dimick et al. Health Aff 2004; web VAR45] • For invasive bladder cancer a delay of >3 months from diagnosis to cystectomy is associated with increased mortality. [Chang et al. J Urol 2003; 170: 1085. Sanchez-Ortiz et al. J Urol 2003; 169:110]

21. Is regionalization a good trend? • Proportion of US hospitals performing cystectomy varied from 45 to 50% between 1988 and 1996, then fell to 39% by 2000.[Taub et al. J Urol 2006; 176:2612] • Nonwhite patients, those with Medicaid / no insurance less likely to receive complex surgical care at HVH (in general and cystectomy)[Liu et al. JAMA 2006; 296:1973. Konety et al. Cancer 2007; 109:542] • Bladder cancer patients tend to be older and low SES; radical cystectomy generally perceived to be under-compensated. Regionalization increases burden of uncompensated care on HVHs[Soloway. Cancer 2005; 104:1559]

22. Urologist Distribution Data from HRSA Area Resource File, 2006 Odisho et al. J Urol, in press

23. Urologists <45 Urologist Distribution Data from HRSA Area Resource File, 2006 Odisho et al. J Urol, in press

24. Conclusions • Regionalization of bladder and renal cancer care has occurred over the past 15 years • Trend is likely to continue given provider demographic trends • Policy decisions must balance (possible) benefit due to regionalization with (likely) harm if access is reduced • Alternative: identify and promulgate HVH processes of care • HCUP/NIS invaluable for descriptive health services research; clinically rich data needed to better define volume-outcomes associations