Variation in Surgical-Readmission Rates and Quality of Hospital Care

Transcription

1 T h e n e w e ngl a nd j o u r na l o f m e dic i n e special article Variation in Surgical-Readmission Rates and Quality of Hospital Care Thomas C. Tsai, M.D., M.P.H., Karen E. Joynt, M.D., M.P.H., E. John Orav, Ph.D., Atul A. Gawande, M.D., M.P.H., and Ashish K. Jha, M.D., M.P.H. A bs tr ac t From the Departments of Health Policy and Management (T.C.T., K.E.J., A.A.G., A.K.J.) and Biostatistics (E.J.O.), Harvard School of Public Health; the Departments of Surgery (T.C.T., A.A.G.), Medicine, Cardio vascular Division (K.E.J.), and General Internal Medicine (E.J.O., A.K.J.), Brigham and Women s Hospital; and the Veterans Affairs Boston Healthcare System (K.E.J., A.K.J.) all in Boston. Address reprint requests to Dr. Jha at the Harvard School of Public Health, 677 Huntington Ave., Boston, MA 02115, or at ajha@hsph.harvard.edu. N Engl J Med 2013;369: DOI: /NEJMsa Copyright 2013 Massachusetts Medical Society. Background Reducing hospital-readmission rates is a clinical and policy priority, but little is known about variation in rates of readmission after major surgery and whether these rates at a given hospital are related to other markers of the quality of surgical care. Methods Using national Medicare data, we calculated 30-day readmission rates after hospitalization for coronary-artery bypass grafting, pulmonary lobectomy, endovascular repair of abdominal aortic aneurysm, open repair of abdominal aortic aneurysm, colectomy, and hip replacement. We used bivariate and multivariate techniques to assess the relationships between readmission rates and other measures of surgical quality, including adherence to surgical process measures, procedure volume, and mortality. Results For the six index procedures, there were 479,471 discharges from 3004 hospitals. The median risk-adjusted composite readmission rate at 30 days was 13.1% (interquartile range, 9.9 to 17.1). In a multivariate model adjusting for hospital characteristics, we found that hospitals in the highest quartile for surgical volume had a significantly lower composite readmission rate than hospitals in the lowest quartile (12.7% vs. 16.8%, P<0.001), and hospitals with the lowest surgical mortality rates had a significantly lower readmission rate than hospitals with the highest mortality rates (13.3% vs. 14.2%, P<0.001). High adherence to reported surgical process measures was only marginally associated with reduced readmission rates (highest quartile vs. lowest quartile, 13.1% vs. 13.6%; P = 0.02). Patterns were similar when each of the six major surgical procedures was examined individually. Conclusions Nearly one in seven patients hospitalized for a major surgical procedure is readmitted to the hospital within 30 days after discharge. Hospitals with high surgical volume and low surgical mortality have lower rates of surgical readmission than other hospitals. 1134

2 Reducing hospital-readmission rates is a priority for both policymakers and clinical leaders. The focus on readmissions has been driven by a belief that reducing the frequency with which patients return to the hospital can both improve care and lower costs. To date, much of the focus has been on readmissions after hospitalization for medical conditions, in which discharge planning and care coordination are often suboptimal. 1-4 The Centers for Medicare and Medicaid Services (CMS) plans to include surgical procedures as it expands its re admissions penalty program. 5 Using medical-readmission rates as a measure of hospital quality is controversial. Although hospitals vary substantially in their medical-readmission rates, these rates are generally uncorrelated (or even inversely correlated) with measures often used to identify high-quality hospitals, including volume, mortality, and adherence to process measures. 6 This evidence, coupled with studies showing that hospitals with the sickest and poorest patients have the highest readmission rates, 7,8 raises the question of whether rates of readmission for medical conditions actually measure hospital quality or whether they instead reflect social and clinical factors that are unrelated to hospital care. 1,7 The relationship between readmission rates and surgical care may be different. Because most patients undergo nonurgent major surgery when they are clinically stable, surgical readmissions are more likely to result from complications of the care the patient received during the index hospitalization. Therefore, one would expect hospitals that excel in surgical care to generally have fewer readmissions. However, we have relatively little information on the types of hospitals that perform well or poorly with regard to readmission rates; one study showed inconsistent relationships between hospital volume and readmissions after surgery during the 1990s. 9 Patterns of surgical care have changed substantially since then. Whether high-quality hospitals, as defined by commonly used measures of hospital surgical performance, have lower surgical-readmission rates on the basis of current patterns of care is largely unknown. Given that the extension of the Hospital Readmissions Reduction Program to include surgical patients is predicated on the notion that inadequate inpatient care and a poor transition from inpatient to postacute care are responsible for surgical readmissions, an understanding of the degree to which readmissions are related to hospital quality would be helpful. In this study, we sought to answer three questions. First, what are the patterns of surgical readmissions among Medicare patients across a set of major procedures in a national sample of hospitals? Second, are key structural characteristics of hospitals, such as size or teaching status, associated with higher surgical-readmission rates? Finally, to what extent is the performance of a hospital on well-established measures of surgical quality, such as adherence to surgical process measures, procedure volume, and mortality, correlated with its surgical-readmission rate? We hypothesized that hospitals with greater adherence to surgical process measures and higher volume would have fewer readmissions. We also predicted that hospitals with lower 30-day perioperative mortality would have lower readmission rates. Me thods Data We linked three data sources: the 2009 Medicare Inpatient 100% file and 2010 Medicare Provider Analysis and Review (MEDPAR) File, the American Hospital Association (AHA) annual survey on hospital characteristics, and Hospital Compare data, including Hospital Quality Alliance (HQA) Surgical Care scores. We focused on patients undergoing any one of six major surgical procedures: coronary-artery bypass grafting (CABG), pulmonary lobectomy, endovascular repair of abdominal aortic aneurysm, open repair of abdominal aortic aneurysm, colectomy, and hip replacement (see Table S1 in the Supplementary Appendix, available with the full text of this article at NEJM.org). These procedures were selected because they are common and costly for Medicare beneficiaries and because they reflect several surgical subspecialties (cardiac, thoracic, vascular, colorectal, and orthopedic), enhancing the generalizability of results across the spectrum of surgical care and hospitals. In addition, cardiovascular procedures such as CABG and open repair of abdominal aortic aneurysm may be included in the CMS Readmissions Reduction Program penalties by 2015, and an understanding of patterns of readmission after 1135

3 T h e n e w e ngl a nd j o u r na l o f m e dic i n e these procedures will provide specific insight into the implications of federal policy efforts. Patients undergoing concurrent valve repair were excluded from the CABG sample. Lobectomy, endovascular and open repairs of abdominal aortic aneurysm, and colectomy were restricted by use of International Classification of Diseases, Ninth Revision (ICD-9), diagnostic codes to those procedures performed for lung cancer, nonruptured aneurysms, and colorectal cancer, respectively, in order to preserve the homogeneity of the samples (see Table S1 in the Supplementary Appendix for a complete list of ICD-9 codes used to identify each procedure). Patients undergoing procedures performed during December of each year were also excluded, because we lacked data on readmissions occurring in the following calendar year. Of the 4650 hospitals providing acute care services to Medicare patients, 3004 performed at least one of the six selected procedures and thus were included in our analyses. Our final patientlevel cohort was composed of 479,471 discharged patients who underwent one of the six procedures of interest. The study was approved by the Office of Human Research Administration at the Harvard School of Public Health. Informed consent was deemed unnecessary by the institutional review board. Variables Our primary predictors were structural characteristics of hospitals and measures of surgical quality. Structural characteristics were taken from the AHA data set and included hospital size, teaching status, region, ownership (nonprofit, for profit, or public), and proportion of patients living below the federal poverty line. We categorized hospitals with fewer than 100 beds as small, hospitals with 100 to 399 beds as medium size, and hospitals with 400 or more beds as large. The percentage of the population living below the federal poverty line based on ZIP Code for residence was determined from U.S. Census data, and an average was calculated for each hospital on the basis of its patient census. We used three well-established measures of surgical quality: the HQA surgical score, procedure volume, and 30-day surgical mortality. The HQA surgical score is calculated according to the evidence-based Surgical Care Improvement Project process measures, which are designed to reduce perioperative surgical complications (Table S2 in the Supplementary Appendix). These process measures are included in the CMS Hospital Value-Based Purchasing Program. 10 Procedure volume was calculated with the use of primary ICD-9 procedure codes from Medicare claims data; patients who underwent multiple procedures of interest were excluded. We calculated the procedure-specific 30-day surgical mortality rates for each hospital with the use of the Elixhauser risk-adjustment scheme, a well-validated tool developed by the Agency for Healthcare Research and Quality. 11,12 We then applied indirect standardization to create a composite mortality rate across the six procedures. 13 For ease of presentation, we divided hospitals into quartiles of performance for each of the three indicators of surgical quality. Our primary outcome measure was a hospital-level composite of the six procedure-specific risk-adjusted readmission rates at 30 days. We used the Elixhauser risk-adjustment approach to calculate expected procedure-specific odds of readmission within 30 days after discharge for each hospital. To calculate the final risk-adjusted procedure-specific readmission rates for each hospital, we multiplied the observed-to-expected readmission ratio for the hospital by the overall readmission rate for that surgical procedure from our national sample. We used indirect standardization to calculate the composite readmission rate for each hospital by averaging the observedto-expected readmission ratios for the six procedures for the hospital and weighting each ratio on the basis of the number of cases for that procedure. Statistical Analysis We compared characteristics of patients who were readmitted within 30 days after discharge with those of patients who were not readmitted. Next, we compared the characteristics of the hospitals that had composite readmission rates above the median with those of hospitals that had rates below the median; characteristics included hospital size, region, teaching status, ownership, and percentage of the population living below the federal poverty line. The distribution of risk-adjusted composite readmission rates was plotted to illustrate variation in readmission rates across hospitals, and we superimposed a plot of procedure volume on this distribution. We then examined bivariate associations between hospital structural charac- 1136

4 Table 1. Characteristics of Patients Readmitted and Those Not Readmitted within 30 Days after Hospitalization for a Major Surgical Procedure.* Characteristic Patients Readmitted (N = 61,785) Patients Not Readmitted (N = 417,686) P Value Age (yr) 78.0± ±7.6 <0.001 Sex (%) <0.001 Male Female Race (%) <0.001 White Black Other Coexisting condition (%) Congestive heart failure <0.001 Chronic obstructive pulmonary disease <0.001 Hypertension <0.001 Diabetes mellitus <0.001 Renal failure <0.001 Obesity <0.001 Depression <0.001 Length of stay (days) <0.001 Discharge status (%) <0.001 Home Home with services Skilled nursing facility Rehabilitation facility In-hospital death 2.8 Other Patients below poverty line according to residence ZIP Code (%) <0.001 Procedure (%) <0.001 Coronary-artery bypass grafting Lobectomy Endovascular repair of abdominal aortic aneurysm Open repair of abdominal aortic aneurysm Colectomy Hip replacement * Plus minus values are means ±SD. Percentages may not sum to 100 because of rounding. Race was determined from the Medicare Beneficiary Summary File. teristics size, teaching status, region, location, percentage of the population living below the federal poverty line, and ownership and the composite readmission rates at 30 days. We subsequently built a multivariate model that adjusted for all the hospital characteristics. Next, we built three separate multivariate models (one for each of the three quality measures), adjusting each for all the hospital characteristics described above. Finally, we built a model adjusted for all three quality measures (HQA surgical score, volume, and mortality) to identify the independent 1137

5 T h e n e w e ngl a nd j o u r na l o f m e dic i n e Table 2. Characteristics of Hospitals According to Performance and Readmission Rates.* Characteristic Hospitals below Median Readmission Rate Hospitals above Median Readmission Rate P Value Ownership (%) 0.04 For profit Private nonprofit Public Size (%) 0.02 <100 beds beds beds Major teaching hospital (%) <0.001 Rural (%) Region (%) <0.001 Northeast Midwest South West ICU present (%) No. of full-time nurses per 1000 patient-days <0.001 Median Interquartile range Patients covered by Medicaid (%) Median Interquartile range HQA surgical score 0.41 Median Interquartile range Composite surgical volume 0.02 Median Interquartile range Composite surgical mortality (%) <0.001 Median Interquartile range * Percentages may not sum to 100 because of rounding. ICU denotes intensive care unit. The Hospital Quality Alliance (HQA) surgical score ranges from 0 to 100% and reflects the percentage of patients receiving evidence-based process-of-care measures. The composite surgical volume is the number of procedures of interest performed during the study period. relationships between each of these measures and readmission rates. All regression models were weighted by hospital-level procedure volume. Although many of the analyses we performed could lead to false positive results through multiple testing, our primary purpose was to investigate the relationship between the composite readmission rates at 30 days and the three quality measures (HQA surgical score, volume, and mortality). These three relationships are most accurately captured in our final, fully adjusted model. In that model, a Bonferroni-adjusted P value of was considered to indicate statistical significance. In sensitivity analyses, modeling HQA surgical score, volume, and mortality as continuous or categorical variables did not have a significant 1138

6 effect on the overall findings; therefore, we present results showing our quality measures in quartiles. We also examined the relationship between quality measures and readmission rates for each of the surgical procedures individually. For multilevel categorical predictors and covariates, P values for trend are presented. All analyses were performed with the use of Stata software, version 12.0 (StataCorp). R esult s Characteristics of the Patients and Hospitals In 2009 and 2010, there were 3004 U.S. hospitals that performed at least one of the six surgical procedures in our study (Table S3 in the Supplementary Appendix). These hospitals accounted for 90.7% of all discharges from acute care hospitals for Medicare fee-for-service beneficiaries in the United States during the study period. There were 479,471 discharges after the index procedures, and 30-day readmission rates ranged from 10.5 to 17.4% across the six procedures (Table S4 in the Supplementary Appendix). The median risk-adjusted composite readmission rate at 30 days was 13.1% (interquartile range, 9.9 to 17.1). As expected, patients who were readmitted were older than patients who were not readmitted (mean, 78.0 vs years of age; P<0.001) and had more coexisting conditions (Table 1). Hospitals with readmission rates below the median were more likely than those with rates above the median to be nonprofit (66.3% vs. 63.6%, P = 0.04), nonteaching (93.2% vs. 89.1%, P<0.001), and located in the West (24.1% vs. 14.8%, P<0.001). These hospitals had a higher number of full-time nurses per 1000 patient-days (median, 7.2 vs. 6.9; P<0.001) and a lower proportion of Medicaid patients (median, 16.2% vs. 17.0%; P = 0.005) than did hospitals with readmission rates above the median (Table 2). No. of Hospitals >36.0 Readmission Rate (%) Structural Characteristics of Hospitals and Readmission Rates There was marked variation in the risk-adjusted composite (six-procedure) readmission rates at 30 days across hospitals (Fig. 1). We also found substantial variation in readmission rates across hospitals in each of the six procedure-specific analyses. In a multivariate model that accounted for each of the structural characteristics (Table 3), we found that nonprofit hospitals had a lower composite readmission rate than for-profit hospitals (13.1% vs. 13.7%, P<0.001). Similarly, nonteaching hospitals had a lower composite readmission rate than major teaching hospitals (12.9% vs. 13.9%, P<0.001). There was no significant difference in composite readmission rates between urban and rural hospitals (13.2% and 13.4%, respectively; P = 0.99). Hospitals in the highest quartile for proportion of patients residing in highpoverty ZIP Code areas had a higher composite readmission rate than hospitals in the lowest quartile (14.0% vs. 13.1%, P<0.001). Trends were similar in each of the six procedure-specific analyses. Quality Measures and Readmission Rates When we used multivariate models to examine the relationship between surgical quality measures and readmission rates, accounting for hospital characteristics, we found that hospitals in the highest quartile for HQA surgical score had a somewhat lower risk-adjusted composite readmission rate than those in the lowest quartile (13.1% vs. 13.6%, P = 0.02). Similarly, hospitals with the highest procedure volumes had a lower composite readmission rate than did hospitals with the lowest procedure volumes (12.7% vs. All hospitals High-volume hospitals Figure 1. Distribution of Risk-Adjusted 30-Day Readmission Rates for Six Surgical Procedures across U.S. Hospitals. Rates of readmission for any reason are shown for all hospitals and for hospitals with a high volume of procedures. 1139

7 T h e n e w e ngl a nd j o u r na l o f m e dic i n e Table 3. Thirty-Day Composite Readmission Rates According to Selected Hospital Characteristics. Characteristic Multivariate-Adjusted Readmission Rate* P Value Size 0.04 <100 beds beds beds 13.0 Setting 0.99 Rural 13.4 Urban 13.2 Major teaching hospital <0.001 Yes 13.9 No 12.9 Ownership <0.001 For profit 13.7 Private nonprofit 13.1 Public 13.5 ICU present 0.12 Yes 13.1 No 13.5 Region <0.001 Northeast 14.4 Midwest 12.7 South 13.5 West 11.9 Average proportion of patients below poverty line according to residence ZIP Code Lowest 13.1 Highest %, P<0.001), and hospitals in the lowest quartile for mortality had a lower composite readmission rate than hospitals in the highest quartile for mortality (13.3% vs. 14.2%, P<0.001) (Table 4). In our final multivariate model, which simultaneously accounted for all three quality measures in addition to hospital structural characteristics, volume and mortality remained independently associated with risk-adjusted composite readmission rates. The highest-volume hospitals had a % <0.001 * Rates are the six-procedure composite rates and were calculated from a single multivariate analysis that simultaneously adjusted for demographic characteristics of the patients, coexisting conditions, and each of the hospital characteristics listed above (r 2 = 0.057). lower composite readmission rate than the lowestvolume hospitals (12.8% vs. 16.5%, P<0.001), and the lowest-mortality hospitals had a lower composite readmission rate than the highestmortality hospitals (13.0% vs. 13.8%, P = 0.001) (Table 4). However, the best-performing hospitals, as assessed on the basis of the HQA surgical score, had only a marginally lower composite readmission rate than the poorest-performing hospitals, a difference that was not significant (13.1% and 13.5%, respectively; P = 0.08). When we examined the readmission rates for the six surgical procedures individually, the overall relationships with hospital quality were consistent with those observed for the composite readmission rates. All three quality measures HQA surgical score, procedure volume, and surgical mortality were generally associated with the procedure-specific readmission rates, although the differences were not always significant. For open repair of abdominal aortic aneurysm, we found no relationship between mortality and readmission rate (Table S5 in the Supplementary Appendix). Discussion We found that approximately one in seven patients discharged after undergoing a major surgical procedure is readmitted within 30 days, although there is substantial variation across U.S. hospitals. With performance measured according to procedure volume and 30-day mortality rates, high-performing hospitals had lower surgicalreadmission rates than low-performing hospitals, and the patterns were consistent for each of the six procedures we examined. HQA surgical score, a reflection of adherence to surgical process measures, was weakly associated with readmission rates, a relationship that was markedly diminished when other quality measures were accounted for. Taken together, these findings offer evidence that surgical-readmission rates are indeed associated with measures of surgical quality. These findings, although quite different from what has been found for medical readmissions, are not surprising. The reasons that bring surgical patients back to the hospital soon after discharge are probably different from those that bring medical patients back. Whereas medical patients may come back because of poor social support at home, inability to access primary care, or general poor health, surgical patients are more likely to return as a consequence of complications arising 1140

8 Table 4. Multivariate Analysis of Quality Measures and 30-Day Composite Readmission Rates. Quality Measure Multivariate-Adjusted Readmission Rate, Model 1* P Value Multivariate-Adjusted Readmission Rate, Model 2 P Value % % HQA surgical score Lowest quartile Second quartile Third quartile Highest quartile Surgical volume <0.001 <0.001 Lowest quartile Second quartile Third quartile Highest quartile Surgical mortality < Lowest quartile Second quartile Third quartile Highest quartile * Model 1 rates are risk-adjusted composite (six-procedure) readmission rates calculated from multivariate analyses that adjusted for hospital size, teaching status, ownership, region, and location (r 2 = 0.059, 0.085, and for HQA surgical score, volume, and mortality, respectively). Model 2 rates are risk-adjusted composite (six-procedure) readmission rates calculated from multivariate analyses that adjusted for hospital size, teaching status, ownership, region, and location, as well as for the other quality measures (r 2 = 0.091). from the surgery. 14 Hospitals with higher volumes of procedures and those with lower mortality rates are better versed in protecting patients from postoperative complications. 15,16 These high-quality hospitals may also use systematic approaches to determine when patients are ready for discharge and engage in readmission-prevention strategies that may have a greater effect on surgical patients than on medical ones. Although we could not specifically determine why higher-volume or lowermortality hospitals have lower readmission rates, we suspect that these factors may be at play. We found that adherence to best-practice guidelines, as reflected by the HQA surgical score, was weakly associated with marginally lower readmission rates. The absence of an independent relationship between the HQA surgical score and readmission rates may be a result of the low variation in performance on this measure (median score, 98.5% in the highest quartile vs. 91.4% in the lowest quartile). Prior studies have also shown an inconsistent relationship between reported process measures and important outcomes, such as mortality. 17,18 Our finding that surgical-readmission rates have a modest but consistent relationship with measures of surgical quality, especially procedure volume and mortality, should offer some reassurance to policymakers who wish to use surgical-readmission rates to grade and pay hospitals, especially given that our findings were generally consistent across all the surgical procedures we considered. However, whether improved surgical outcomes (e.g., reduced mortality) or a greater concentration of surgical procedures in a few institutions (higher volume) would translate into fewer readmissions is unclear. Although we found significant associations between readmission rates and hospital characteristics (e.g., ownership), their clinical significance is unclear. Several studies have identified patient-level factors, such as patient age and severity of underlying illness 14,19,20 or length of stay, that are predictive of readmission. However, the data on the relationship between hospital characteristics or surgical-quality measures and readmissions are less robust. Studies have shown that rates of readmission after certain high-risk procedures are either inversely related 24 or unrelated 9 to procedure volume. Both these studies focused on pat- 1141

9 terns of care in the 1990s, before the widespread use of minimally invasive approaches and postsurgical care algorithms. We suspect that as surgical care has changed, factors such as procedure volume have become more important for recognizing the risk of complications, and this may underlie the relationship between volume and readmission rates that we observed. There are limitations to our study beyond those described above. Because we used administrative data, we were unable to adequately account for factors not captured by billing codes. However, we expect that inadequate risk adjustment would probably diminish our ability to find an association. Another concern is that some have questioned the value of treating volume as a categorical variable Our results did not qualitatively change in various sensitivity analyses, including those in which volume was treated as a continuous variable. Finally, we focused on the Medicare population and therefore cannot say whether our findings would extend to nonelderly persons in the United States. In conclusion, we found variation in surgicalreadmission rates across U.S. hospitals. Readmission rates were associated with procedure volume and mortality rate, two well-established measures of hospital surgical quality. Surgicalreadmission rates may be a relevant and valid way for policymakers, clinical leaders, and patients to measure the surgical performance of U.S. hospitals. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. References 1. Joynt KE, Orav EJ, Jha AK. Thirty-day readmission rates for Medicare beneficiaries by race and site of care. JAMA 2011; 305: Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med 2009;360: Jha AK, Orav EJ, Epstein AM. Public reporting of discharge planning and rates of readmissions. N Engl J Med 2009;361: Voss R, Gardner R, Baier R, Butterfield K, Lehrman S, Gravenstein S. The care transitions intervention: translating from efficacy to effectiveness. Arch Intern Med 2011;171: Readmissions Reduction Program ( -for-service-payment/acuteinpatientpps/ Readmissions-Reduction-Program.html). 6. Chen LM, Jha AK, Guterman S, Ridgway AB, Orav EJ, Epstein AM. Hospital cost of care, quality of care, and readmission rates: penny wise and pound foolish? Arch Intern Med 2010;170: Joynt KE, Jha AK. Who has higher readmission rates for heart failure, and why? Implications for efforts to improve care using financial incentives. Circ Cardiovasc Qual Outcomes 2011;4: Joynt KE, Jha AK. Characteristics of hospitals receiving penalties under the Hospital Readmissions Reduction Program. JAMA 2013;309: Goodney PP, Stukel TA, Lucas FL, Finlayson EV, Birkmeyer JD. Hospital volume, length of stay, and readmission rates in high-risk surgery. Ann Surg 2003; 238: CMS issues final rule for first year of hospital value-based purchasing program ( press/factsheet.asp?counter=3947). 11. Fry DE, Pine M, Jordan HS, et al. Combining administrative and clinical data to stratify surgical risk. Ann Surg 2007;246: Pine M, Jordan HS, Elixhauser A, et al. Enhancement of claims data to improve risk adjustment of hospital mortality. JAMA 2007;297: Pouw ME, Peelen LM, Lingsma HF, et al. Hospital standardized mortality ratio: consequences of adjusting hospital mortality with indirect standardization. PLoS One 2013;8(4):e Kassin MT, Owen RM, Perez SD, et al. Risk factors for 30-day hospital readmission among general surgery patients. J Am Coll Surg 2012;215: Birkmeyer JD, Skinner JS, Wennberg DE. Will volume-based referral strategies reduce costs or just save lives? Health Aff (Millwood) 2002;21(5): Ghaferi AA, Birkmeyer JD, Dimick JB. Hospital volume and failure to rescue with high-risk surgery. Med Care 2011;49: Jha AK, Li Z, Orav EJ, Epstein AM. Care in U.S. hospitals the Hospital Quality Alliance program. N Engl J Med 2005;353: Garcia N, Fogel S, Baker C, Remine S, Jones J. Should compliance with the Surgical Care Improvement Project (SCIP) process measures determine Medicare and Medicaid reimbursement rates? Am Surg 2012;78: Schneider EB, Hyder O, Brooke BS, et al. Patient readmission and mortality after colorectal surgery for colon cancer: impact of length of stay relative to other clinical factors. J Am Coll Surg 2012;214: Greenblatt DY, Greenberg CC, Kind AJ, et al. Causes and implications of readmission after abdominal aortic aneurysm repair. Ann Surg 2012;256: Vorhies JS, Wang Y, Herndon J, Maloney WJ, Huddleston JI. Readmission and length of stay after total hip arthroplasty in a national Medicare sample. J Arthroplasty 2011;26:Suppl: Cram P, Lu X, Kates SL, Singh JA, Li Y, Wolf BR. Total knee arthroplasty volume, utilization, and outcomes among Medicare beneficiaries, JAMA 2012; 08: Hendren S, Morris AM, Zhang W, Dimick J. Early discharge and hospital readmission after colectomy for cancer. Dis Colon Rectum 2011;54: Hannan EL, Racz MJ, Walford G, et al. Predictors of readmission for complications of coronary artery bypass graft surgery. JAMA 2003;290: LaPar DJ, Kron IL, Jones DR, Stukenborg GJ, Kozower BD. Hospital procedure volume should not be used as a measure of surgical quality. Ann Surg 2012;256: Livingston E, Cao J, Dimick JB. Tread carefully with stepwise regression. Arch Surg 2010;145: Livingston EH, Cao J. Procedure volume as a predictor of surgical outcomes. JAMA 2010;304: Livingston EH, Elliot A, Hynan L, Cao J. Effect size estimation: a necessary component of statistical analysis. Arch Surg 2009;144: Copyright 2013 Massachusetts Medical Society. 1142