This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Ho, V. Articles by Howard, L. Search for Related Content PubMed PubMed Citation Articles by Ho, V. Articles by Howard, L.

Trends in Hospital and Surgeon Volume and Operative Mortality for Cancer Surgery

¹ Department of Economics, Baker Institute, Rice University, MS 22, 6100 Main St. Houston, Texas 77005
² Department of Medicine, Baylor College of Medicine, Houston, Texas
³ Department of Surgery, The University of Alabama at Birmingham, KB 321, 1530 3rd Avenue South, Birmingham, Alabama 35294-0022
⁴ Department of Health Care Organization and Policy, The University of Alabama at Birmingham, RP HB 330, 1530 3rd Avenue South, Birmingham, Alabama 35294-0022

Correspondence: Address correspondence and reprint requests to: Vivian Ho, PhD; E-mail: vho{at}rice.edu.

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

 
Background: We measured 13-year trends in operative mortality for six cancer resections. We then examined whether these trends are driven by changes in hospital and surgeon volume or by changes that occurred among all providers, regardless of volume.

Methods: We analyzed administrative discharge data on patients who received one of six cancer resections in Florida, New Jersey, and New York for three time periods: 1988 to 1991, 1992 to 1996, and 1997 to 2000. Descriptive statistics and nested regression models were used to test for changes in the association between inpatient mortality and annual hospital and annual surgeon volume over time, adjusting for patient and hospital characteristics.

Results: Unadjusted inpatient mortality rates for the six cancer resections declined between .8 and 4.0 percentage points between the time periods 1988 to 1991 and 1997 to 2000. Over this time period, annual hospital and surgeon volumes for the six cancer operations increased an average of 24.3% and 24.2% , respectively. The logistic regressions indicated a relatively stable relationship over time between both increased hospital and surgeon volume and lower inpatient mortality. Simulations suggest that increases in hospital and surgeon procedure volume over time led to a reduction in inpatient mortality ranging from .1 percentage points for rectal cancer to 2.3 percentage points for pneumonectomy.

Conclusions: Persistence of the volume-outcome relation and increasing hospital and surgeon volumes explain much of the decline over time in inpatient mortality for five of the six cancer operations studied. Concentrating cancer resections among high-volume providers should lead to further reduced inpatient mortality.

Key Words: Neoplasms • Surgery • Mortality • Provider volume • Outcomes assessment

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

 
Several federal and nongovernment organizations collaborate to provide information on trends in cancer mortality. These reports indicate that overall cancer death rates declined by 1.1% per year from 1993 to 2001.¹ Declines in operative mortality have been associated with improvements in long-term survival in past studies.^2,3 However, we lack a comparable report on long-term trends in operative mortality for cancer patients. This study measured population-based trends in operative mortality over a 13-year period for six cancer resections.

We also know little about what factors determine trends in operative mortality. Some studies have associated increased experience or increased provider volume over time with reduced mortality.^4–6 However, these studies were focused on only one operation, and some were based on samples of a limited number of surgeons or hospitals. In contrast, this study analyzed data on six cancer operations with an extensive sample of surgeons and hospitals to determine whether increases in provider volume can explain trends in operative mortality.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

 
We first combined data from three large US states on six cancer resections to examine long-term trends in operative mortality and annual hospital and surgeon volume. We then estimated logistic regressions to test whether the association between provider volume and inpatient mortality has remained stable over time. The regression estimates were then used to simulate mortality reductions associated with provider volume increases while patient characteristics were held constant. These simulations allow one to compare mortality reductions solely attributable to volume, versus total changes in mortality that occurred over time.

Patient data were obtained from statewide hospital discharge abstract files for Florida, New Jersey, and New York for the years 1988 to 2000. In accordance with past studies,^6–10 we used International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) procedure codes to identify all patients who underwent resection for colorectal cancer (45.7x, 45.8, 48.4x, 48.5, and 48.6x), pulmonary lobectomy (32.4), pneumonectomy (32.5), esophagectomy (42.40, 42.41, 42.42, or 43.99), and pancreaticoduodenectomy (the Whipple procedure; 52.7). Colorectal cancer patients were further identified and subdivided by ICD-9-CM diagnosis codes: colon cancer (153.0–153.9) and cancer of the rectum (154.0–154.1). For lobectomy, pneumonectomy, and esophagectomy, we also excluded patients who did not have a cancer diagnosis code related to the procedure that was performed.¹¹

Outcomes and Provider Volume
The outcome variable of interest was inpatient mortality. The explanatory variables of interest were hospital procedure volume and surgeon procedure volume. For each hospital, the number of times each specific operation was performed was calculated for each year. Only one occurrence of each procedure was counted for each admission.

We identified the operating surgeon for each procedure based on unique physician state license numbers that were provided in each state database. Records containing missing physician license numbers were excluded from the analysis. To check for the effects of possible miscoding, we matched information from state Department of Health Web sites, the ABMS Medical Specialists Plus directory, and the American Medical Association physician file to identify license numbers that were associated with a surgery specialty (e.g., general surgery or abdominal surgery). The percentage of patients with an operating surgeon whose specialty was unrelated to surgery or unidentifiable ranged from a low of 8.1% for Whipple patients to 12.1% for patients who underwent pneumonectomy. The results we report include these patients. Excluding these patients from the analyses decreased the precision of some of the volume estimates, but the main conclusions of the study remained the same.

Statistical Analysis
The relationship between provider volume and inpatient mortality was analyzed over three time periods: 1988 to 1991, 1992 to 1996, and 1997 to 2000. A logistic regression model was used to examine the relationship between provider volume and mortality, adjusting for potential confounders. The unit of analysis was the patient, with annual volume measured at the level of the hospital and the operating surgeon. In accordance with previous studies of the volume-outcome relation, the logarithms of hospital and surgeon volume were used as the explanatory variables of interest in the regressions.^12–14 Both prior studies and this analysis show that log(volume) seems to be linearly associated with the logit of in-hospital mortality rates.

Indicator variables for the time periods 1992 to 1996 and 1997 to 2000 were included in the regression to compare mortality rates in these time periods relative to 1988 to 1991. We estimated preliminary specifications, including interaction terms between each of the time period indicator variables with both hospital and surgeon volume. These interaction terms capture potential differences in the slope of the volume-outcome relation, which may have changed over time. To avoid potential problems with multicollinearity if the volume-outcome relation did not shift across time periods, interaction terms that had a P value >.20 in the preliminary specifications were dropped from the final regressions.¹⁵

The logistic regressions were estimated by using multilevel generalized linear modeling to adjust for the clustering of patients within surgeons and the clustering of surgeons within hospitals.¹⁶ Surgeons who performed the same procedure in more than one hospital were modeled as contributing a separate random effect in each hospital where they treated patients.¹⁷ All estimation was performed with Stata 8. The GLLAMM command was used to conduct the multilevel modeling.¹⁸

The logistic regressions adjusted for patient and hospital characteristics. Patient characteristics used as independent variables included age (< 60, 60–69, 70–79, and 80 years), sex, race (white, black, or other), urgent or emergency admission, health insurance (Medicare, Medicaid, private, or other/none), state of residence, cancer stage, and the component variables of the Charlson comorbidity index.¹⁹ Cancer stage was defined using ICD-9-CM diagnosis codes (nodal involvement, 196.x; metastasis, 197.x–198.x). Length of stay was also included as an explanatory variable to control for potential censoring of inpatient mortality for patients with shorter lengths of stay.²⁰ In addition, we defined indicator variables to distinguish between specific operations and operation sites that were specific to each procedure. These variables were included to adjust for disease-specific differences in the difficulty of each procedure and patient case mix. For example, the logistic regression for esophagectomy patients includes indicator variables for patients who underwent gastrectomy or total esophagectomy versus a partial esophagectomy, as well as indicator variables for whether the reported location of the tumor was distal, middle, upper, or other. Procedure and tumor locations were defined on the basis of previous studies of these operations that used ICD-9-CM procedure and diagnosis codes.^2,6,8–10 An indicator variable was also included to identify patients treated in a hospital that was a member of the Council of Teaching Hospitals as indicated by the American Hospital Association Annual Survey for each year.

For many cancer procedures, patients undergoing surgery at National Cancer Institute–designated cancer centers have had lower surgical mortality rates than those treated at comparable high-volume hospitals.²¹ Because we had data from only three comprehensive cancer centers in our sample period, we did not include cancer center status as a control variable.

The regression estimates were used to assess the relationship between changes in procedure volume and changes in inpatient mortality that occurred over the sample period while patient characteristics were held constant. We first used the estimates to predict inpatient mortality for all patients treated from 1988 to 1991, assuming that they underwent operation in a hospital or under a surgeon’s care with the mean procedure volume for that period. We then predicted inpatient mortality for these same patients, assuming that the hospital volume or surgeon volume was instead the mean values for these procedures in 1997 to 2000. We simulated the effects of changes in hospital or surgeon volume only in those cases in which the odds ratio for provider volume was statistically significant. In cases where interaction terms indicated a change in the volume-outcome relation over time, these changes were also incorporated into the predictions. These predicted changes in adjusted mortality were then compared with the unadjusted differences in inpatient mortality for each procedure observed in these two time periods.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

 
A total of 344,617 patients in Florida, New Jersey, and New York underwent 1 of the 6 cancer operations between 1988 and 2000. Because Florida did not record surgeon identifiers until 1992, 39,683 Floridians between 1988 and 1991 were excluded from the descriptive statistics and regressions. Table 1 reveals decreases in unadjusted inpatient death rates for each of the six cancer resections over the sample period. Pulmonary lobectomy patients experienced the smallest decline in inpatient mortality (.8 percentage points) between 1988 to 1991 and 1997 to 2000 (from 4.1% to 3.3% ). The largest decline in inpatient mortality (4.0% ) occurred for esophagectomy patients, for whom mortality decreased from 14.5% to 10.5% over the sample period.

The percentage of patients aged 75 years who underwent operation increased over the sample period between 1.1 (for esophagectomy) and 8.8 (for pulmonary lobectomy) percentage points. However, the percentage of patients who received nonelective surgery declined between 1.9 (for colon cancer resections) and 13.4 (for Whipple procedures) percentage points for the six operations. In addition, there was no systematic pattern to changes in the Charlson comorbidity index across the six resections. Therefore, patient case-mix severity did not seem to increase or decrease systematically over the sample period.

Table 2 presents adjusted odds ratios for inpatient mortality. Odds ratios for the time periods 1992 to 1996 and 1997 to 2000 (relative to 1988 to 1991) represent time-specific differences in mortality for all patients, regardless of hospital and surgeon volume. Table 2 also lists adjusted odds ratios for inpatient mortality by hospital and procedure volume, as well as interactions between provider volume and time period, when appropriate.

Although the odds ratios for the association between hospital volume and inpatient mortality were < 1.00 for each of the six resections, the point estimates were nonsignificant for pneumonectomy and esophagectomy. For the remaining procedures, the odds ratios for the association between increased hospital volume and reduced inpatient mortality ranged from .85 for the Whipple procedure (P = .017) to .94 for resection for colon cancer (P = .010). Preliminary specifications indicated that none of the interactions between hospital volume and the time period dummy variables had a P value < .20. Therefore, for these six cancer operations, the relationship between hospital volume and inpatient mortality remained stable over the 13-year time period.

The relationship between increased surgeon volume and reduced inpatient mortality was precisely estimated in at least one time period for each of the six cancer procedures. The statistically significant odds ratios ranged from .67 (P = .004) for patients who underwent pneumonectomy in 1997 to 2000 to .94 (P = .040) for patients who underwent resection for colon cancer in 1997 to 2000. For rectal cancer resection, pulmonary lobectomy, esophagectomy, and the Whipple procedure, the association between increased surgeon volume and reduced inpatient mortality remained stable throughout the 13-year period. The odds ratios for interactions between surgeon volume and time period for the remaining two procedures tended to decline in magnitude in later time periods. This result suggests that each additional procedure performed in later time periods was associated with an even larger reduction in inpatient mortality for colon cancer resection and pneumonectomy.

Table 3 presents predictions that allow one to assess the magnitude of the decline in inpatient mortality over the sample period that is due to increases in hospital and surgeon volume. The predictions are based on mean procedure volumes in 1988 to 1991 vs. 1997 to 2000 while patient characteristics are held constant at their 1988 to 1991 levels. The effects of increases in hospital and/or surgeon volume were included in these predictions only if they were statistically significant in the adjusted mortality regressions listed in Table 2. In most cases, the predicted inpatient mortality rates based on mean procedure volumes in 1988 to 1991 were slightly lower than the actual death rates for this time period. The very high procedure volumes of a few hospitals and surgeons can skew these volume distributions to the right, so that simulations based on mean volumes will underestimate mortality relative to the actual distribution of patients among hospitals of various sizes. Nevertheless, simulations based on changes in mean provider volumes provide a useful gauge of the potential effect of increasing provider volumes on lower in-hospital death rates.

For pulmonary lobectomy, which experienced the largest increases in hospital and surgeon volume over time, the decrease in predicted mortality was equivalent to the observed decrease in mortality that occurred over the sample period. The result suggests that increases in provider volume were primarily responsible for the observed reduction in mortality for pulmonary lobectomy. For the remaining cases, the predicted decrease in inpatient mortality associated with observed increases in provider volume over the sample period was smaller than the actual decline in inpatient mortality. However, as noted in Table 1, the case mix for each of the six resections has also changed over time. Therefore, changes in case mix or other changes in provider characteristics are likely to explain the remaining observed declines in mortality rates.

The particularly small predicted decline in death rates for rectal cancer resection is noticeable, given that the observed death rate decreased 1.5 percentage points over the sample period. Recall from Table 2 that this is the one procedure for which inpatient mortality declined for all patients, regardless of hospital or procedure volume. For this procedure, provider volume played a relatively small role in explaining mortality declines over time.

For pneumonectomy, the absolute value of the decline in predicted mortality is larger than the actual decrease, even though the average surgeon volume did not increase during the sample period. The regression results revealed that the association between surgeon volume and mortality increased in magnitude over time. Therefore, even though surgeon volume did not increase for pneumonectomy during the sample period, the beneficial effect of a higher surgeon volume increased so much that it led to a noticeable decrease in inpatient death rates. The fact that the predicted decrease exceeds the actual decrease in mortality for pneumonectomy is attributable to a highly skewed distribution of procedure volumes among hospitals, which leads to relatively high mean provider volumes for the simulations.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

 
A previous study by Goodney et al.²² concluded that there were no significant changes in operative mortality between 1994 and 1999 for seven of eight cancer procedures that were examined. However, this past study spanned only 6 years, and patient characteristics and hospital volume were available only for the Medicare population. In contrast, we analyzed data on patients of all ages treated over a longer time period (1988 to 2000). Although we lack data from most states, the states of Florida, New Jersey, and New York comprise 15% of the US population.

Whereas Goodney et al.²² found that unadjusted mortality rates increased for colectomy and pulmonary lobectomy between 1994 and 1999, we observed declining operative mortality for all six cancer resections. Moreover, the decline was evident in our data between 1992 to 1996 and 1997 to 2000, a time span comparable to the study by Goodney et al. Furthermore, our analysis suggests that this decline is in part attributable to a persistent volume-outcome relation and an increase in procedure volume for both hospitals and surgeons over the study period. Simulations based on characteristics of patients treated between 1988 and 1991 suggest that increases in hospital and/or surgeon volume over the sample period led to reductions in inpatient mortality ranging between 2.6% and 23.8% for the six cancer operations. Thus, increases in procedure volume are contributing to safer surgery over time.

Contrary to Goodney et al.,²² we also found that risk-adjusted inpatient death rates for resection for rectal cancer were lower in 1992 to 2000 than in 1988 to 1991 for all patients regardless of hospital and surgeon volume. In addition, inpatient mortality for patients who underwent resection for colon cancer was lower in 1992 to 1996 vs. 1988 to 1991, although this mortality reduction did not persist through the end of the sample period. Resections for colon and rectal cancer were by the far the most common procedures of the six operations we studied, with 175,202 and 58,568 admissions in our sample, respectively. The greater frequency of these procedures may have led to "knowledge spillovers" or systemwide learning effects that benefited all providers regardless of their size. Reductions in inpatient mortality over time for hospitals of all sizes have been identified for angioplasty,²⁰ another frequently performed procedure, but have not been identified for the less common cancer resections in this study.

Our analysis faced some significant limitations. Linking of physician license numbers to American Medical Association and American Board of Medical Specialties data revealed that 10.4% of the patients in our sample were treated by operating physicians who did not report surgery as a specialty. We were reluctant to exclude these patients from our sample, because some nonsurgeon specialists may in fact have performed these procedures. For example, we could not rule out the possibility that a gynecologist oncologist might occasionally perform a colectomy or that a physician who performed a partial excision of the large intestine also listed his or her primary specialty as gastroenterology. However, we repeated our analysis excluding those patients with operating physicians that did not list surgery as a primary specialty. The results remained virtually the same, although the precision of the estimates decreased slightly in some cases.

We did not control for the specialty of thoracic surgery, which has been associated with lower operative mortality for lung resection in past research.²³ Preliminary regressions revealed an association between the thoracic surgery specialty and lower operative mortality for lobectomy and pneumonectomy. However, we lacked a sufficient sample size to estimate a precise effect. Nevertheless, even with inclusion of an indicator variable for thoracic surgeons, the coefficients on provider volume, the time period effects, and the interaction terms remained virtually unchanged.

Our analysis focused on inpatient mortality as the outcome of interest. This variable may lead to biased measures of the volume-outcome relationship if, for example, high-volume hospitals discharge patients at a high risk of death sooner than low-volume hospitals. To adjust for this potential confounding effect, we included length of stay as a regressor in our multivariate regressions. In addition, past studies using 30-day mortality to study outcomes for these cancer operations confirmed the presence of a volume-outcome effect.^7,17,22,24

The hospital discharge abstract data examined in this study lack complete information on all clinical characteristics that determine inpatient mortality for these six cancer resections. Nevertheless, past analyses using more detailed clinical data from the Surveillance, Epidemiology, and End Results/Medicare databases also confirmed the presence of a significant hospital volume–outcome effect for several of these procedures.^7,24 In addition, there is little evidence from more detailed clinical studies that the volume-outcome relationship is driven by systematic differences in case mix by provider size.¹⁷

Past studies have shown that at any given point in time, higher-volume hospitals have lower inpatient mortality rates than lower-volume hospitals. However, this is the first study to document increases in procedure volume over time for six cancer resections. Furthermore, we demonstrate that these volume increases play a significant role in explaining observed declines in operative mortality over time. These findings provide strong support for efforts to regionalize provision of these cancer operations. The Leapfrog Group, a coalition of large employers and other health care purchasers, is aiming to direct their employees to high-volume providers. The organization recommends that patients needing cancer surgery should look for hospitals that perform 11 pancreatic resections and 13 esophagectomies per year.²⁵ In our sample for the year 2000, only 8% of hospitals performing the Whipple procedure and 5% of hospitals performing esophagectomies met these criteria. Therefore, efforts to encourage patients to follow the Leapfrog criteria when choosing a hospital should lead to further significant reductions in cancer surgery mortality.

Others suggest that further reductions in inpatient mortality could be achieved by encouraging administrators and leaders in the field of surgery to actively manage the distribution of cancer resections within hospitals to achieve higher surgeon volumes.¹⁷ In addition, several states report procedure volumes for the Whipple procedure and other operations on their Web sites and advise readers that volume is one potential indicator of quality. Insurers are also introducing tiered pricing for hospitals into their insurance plans: a mechanism that may be used to direct patients toward high-volume hospitals in the future.²⁶

However, all of these efforts to regionalize operations leave the final choice of hospital and surgeon to the patient and referring physician, and this still allows the possibility for some patients to be treated at low-volume hospitals. Many states enforce Certificate of Need regulations for open heart surgery and transplantation.²⁷ These rules forbid hospitals from providing these services unless they demonstrate to the state that there is sufficient need in the patient population to meet minimum volume standards. To date, no such rules have been proposed for cancer surgery. Given the magnitude and persistence of the volume-outcome relation for the surgical procedures reviewed in this study, perhaps the application of state Certificate of Need rules or other mandatory regionalization policies should be considered.

	ACKNOWLEDGMENTS

 
Supported by grant RSGHP-03-076-01-PBP from the American Cancer Society.

Received for publication July 22, 2005. Accepted for publication December 6, 2005.

	REFERENCES

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

 

1. Jemal A, Clegg LX, Ward E, et al. Annual report to the nation on the status of cancer, 1975–2001, with a special feature regarding survival. Cancer 2004; 101:3–27.[CrossRef][Medline]
2. Birkmeyer JD, Warshaw AL, Finlayson SRG, Grove MR, Tosteson ANA. Relationship between hospital volume and late survival after pancreaticoduodenectomy. Surgery 1999; 126:178–83.[Medline]
3. Mitry E, Bouvier AM, Esteve J, Faivre J. Benefit of operative mortality reduction on colorectal cancer survival. Br J Surg 2002; 89:1557–62.[CrossRef][Medline]
4. Sutton DN, Wayman J, Griffin SM. Learning curve for oesophageal cancer surgery. Br J Surg 1999; 85:1399–402.
5. Ho V, Heslin MJ. Effect of hospital volume and experience on in-hospital mortality for pancreaticoduodenectomy. Ann Surg 2002; 237:509–14.
6. Birkmeyer JD, Finlayson SRG, Tosteson ANA, Sharp SM, Warshaw AL, Fisher ES. Effect of hospital volume on inhospital mortality with pancreaticoduodenectomy. Surgery 1999; 125:250–6.[Medline]
7. Begg CB, Cramer LD, Hoskins WJ, Brennan MF. Impact of hospital volume on operative mortality for major cancer surgery. JAMA 1998; 280:1747–51.[Abstract/Free Full Text]
8. Harmon JW, Tang DG, Gordon TA, et al. Hospital volume can serve as a surrogate for surgeon volume for achieving excellent outcomes in colorectal resection. Ann Surg 1999; 230:404–13.[CrossRef][Medline]
9. Patti MG, Corvera CU, Glasgow RE, Way LW. A hospital’s annual rate of esophagectomy influences the operative mortality rate. J Gastrointest Surg 1998; 2:186–92.[CrossRef][Medline]
10. Romano PS, Mark DH. Patient and hospital characteristics related to in-hospital mortality after lung cancer resection. Chest 1992; 101:1332–7.[Abstract/Free Full Text]
11. Finlayson EVA, Birkmeyer JD. Operative mortality with elective surgery in older adults. Eff Clin Pract 2001; 4:172–7.[Medline]
12. Ho V. Evolution of the volume-outcome relation for hospitals performing coronary angioplasty. Circulation 2000; 101:1806–11.[Abstract/Free Full Text]
13. Birkmeyer JD, Siewers AE, Finlayson SR, et al. Hospital volume and surgical mortality in the United States. N Engl J Med 2002; 346:1128–37.[Abstract/Free Full Text]
14. 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; 238:161–7.[CrossRef][Medline]
15. Hosmer DW, Lemeshow S. Applied Logistic Regression. 2nd ed. New York: Wiley Inter-Science, 2000.
16. McCulloch CE, Searle RS. Generalized, Linear and Mixed Models. New York: Wiley, 2001.
17. Birkmeyer JD, Stukel TA, Siewers AE, Goodney PP, Wennberg DE, Lucas FL. Surgeon volume and operative mortality in the United States. N Engl J Med 2003; 349:2117–27.[Abstract/Free Full Text]
18. Rabe-Hesketh S, Skrondal A, Pickles A. Reliable estimation of generalized linear mixed models using adaptive quadrature. Stata J 2002; 2:1–21.
19. Romano PS, Roos LL, Jollis JG. Adapting a clinical comorbidity index for use with ICD-9-CM administrative data: differing perspectives. J Clin Epidemiol 1993; 46:1075–9.[CrossRef][Medline]
20. Ho V. Evolution of the volume-outcome relationship for hospitals performing coronary angioplasty. Circulation 2000; 101:1806–11.[Abstract/Free Full Text]
21. Birkmeyer NJO, Goodney PP, Stukel TA, Hillner BE, Birkmeyer JD. Do cancer centers designated by the National Cancer Institute have better surgical outcomes? Cancer 2005; 103:435–41.[CrossRef][Medline]
22. Goodney PP, Siewers AE, Stukel TA, Lucas FL, Wennberg DE, Birkmeyer JD. Is surgery getting safer? National trends in operative mortality. J Am Coll Surg 2002; 195:219–27.[CrossRef][Medline]
23. Goodney PP, Lucas FL, Stukel TA, Birkmeyer JD. Surgeon specialty and operative mortality with lung resection. Ann Surg 2005; 241:179–84.[Medline]
24. Schrag D, Cramer LD, Bach PB, Cohen AM, Warren JL, Begg CB. Influence of hospital procedure volume on outcomes following surgery for colon cancer. JAMA 2000; 284: 3028–35.[Abstract/Free Full Text]
25. The Leapfrog Group for Patient Safety. Evidence-based hospital referral fact sheet. Available at: http://www.leapfroggroup.org/media/file/leapfrog-evidence-based_hospital_referral_fact_sheet.pdf.
26. Robinson JC. Hospital tiers in health insurance: balancing consumer choice with financial incentives. Health Aff 2003; W3:135–46.[CrossRef]
27. American Health Planning Association. National Directory of Health Planning, Policy and Regulatory Agencies. 13th ed. Falls Church, VA: American Health Planning Association, 2002.

This article has been cited by other articles:

				A. D. Auerbach, J. F. Hilton, J. Maselli, P. S. Pekow, M. B. Rothberg, and P. K. Lindenauer Shop for Quality or Volume? Volume, Quality, and Outcomes of Coronary Artery Bypass Surgery Ann Intern Med, May 19, 2009; 150(10): 696 - 704. [Abstract] [Full Text] [PDF]

				R. L. Gruen, V. Pitt, S. Green, A. Parkhill, D. Campbell, and D. Jolley The Effect of Provider Case Volume on Cancer Mortality: Systematic Review and Meta-Analysis CA Cancer J Clin, May 1, 2009; 59(3): 192 - 211. [Abstract] [Full Text] [PDF]

				N. N. Baxter and J. Garcia-Aguilar Organ Preservation for Rectal Cancer J. Clin. Oncol., March 10, 2007; 25(8): 1014 - 1020. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Ho, V. Articles by Howard, L. Search for Related Content PubMed PubMed Citation Articles by Ho, V. Articles by Howard, L.