| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2001;72:334-340
© 2001 The Society of Thoracic Surgeons
a Department of Surgery, The Johns Hopkins University School of Medicine, Baltimore, Maryland, USA
b Department of Anesthesiology/Critical Care Medicine, The Johns Hopkins University School of Medicine, Baltimore, Maryland, USA
c Department of Health Policy and Management, The Johns Hopkins University Schools of Medicine and Hygiene and Public Health, Baltimore, Maryland, USA
Address reprint requests to Dr Heitmiller, The Johns Hopkins Hospital, Department of Surgery, 600 N Wolfe St, Osler 624, Baltimore, MD 21287-4605
e-mail: rheitmil{at}jhmi.edu
Presented at the Thirty-seventh Annual Meeting of The Society of Thoracic Surgeons, New Orleans, LA, Jan 29–31, 2001.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and methodsResultsCommentDiscussionReferences |
|---|
Methods. Statewide database was analyzed for patients who underwent esophageal resection in Maryland (n = 1,136 patients) from 1984 to 1999. Multivariate regression was used to determine the association of hospital volume with in-hospital mortality, length of stay, and charges after adjusting for case mix and time period.
Results. Unadjusted in-hospital mortality rates were lower in high volume hospitals (2.7%) than medium (12.7%) and low (16%) volume hospitals (p < 0.001). High hospital volume was associated with (1) fivefold reduction in the risk of death (odds ratio, 0.21; 95% confidence interval, 0.10 to 0.42; p < 0.001); (2) a 6-day (95% confidence interval, 5 to 7 days; p < 0.001) reduction in length of stay; and (3) $11,673 (95% confidence interval, $9,504 to $12,841; p < 0.001) decrease in hospital charges.
Conclusions. Hospitals that perform high volumes of esophageal resection have superior clinical and economic outcomes. By referring these patients to high volume centers, we may improve quality and reduce costs.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentDiscussionReferences |
|---|
We conducted this study to report the experience with esophagectomy in Maryland over a 15-year period and to examine the relationship between hospital volume and in-hospital mortality, length of stay, and hospital charges. Because Maryland is a relatively small state in which evidence-based referrals is feasible, that state provides a unique opportunity to explore the relationship between volume and outcome and to evaluate the potential impact of evidence-based referrals for esophageal operations.
|
Material and methods |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentDiscussionReferences |
|---|
Hospital volume
Each hospital in Maryland is assigned a unique identification number in the Health Services Cost Review Commission database. We calculated the number of procedures performed during each calendar year for all hospitals that performed esophageal resection in Maryland from 1984 to 1999. The hospitals were then divided into three groups based on annual procedural volume. On the basis of natural breakpoints for volume, the volume groupings were less than or equal to 3 procedures per year (low volume), between 4 and 15 per year (medium volume), and more than 15 per year (high volume).
Outcomes
In-hospital mortality, length of stay, and hospital charges were the primary outcome variables. Hospital charges are tightly regulated by the Health Services Cost Review Commission in Maryland and therefore, are a good approximation of cost. Hospital charges were adjusted for inflation and converted to 1999 dollars using the consumer price index for health care in the United States from 1984 to 1999 [8]. To identify a potential high-risk population to target for regional referral to a high volume esophageal surgery center, we analyzed the effect of hospital volume on in-hospital mortality for patients greater than and less than 65 years old (the cutoff for eligibility of Medicare benefits).
Case mix adjustment
To account for patient difference between hospitals we adjusted for demographics (age, sex, and race), comorbid disease, and severity of illness. Race was encoded as a dichotomous variable (white versus nonwhite). Patient comorbid disease states were retrieved from the secondary diagnosis codes contained in the Health Services Cost Review Commission database. Comorbid disease was adjusted for using Romano’s modification of the Charlson comorbidity index for use with ICD-9-CM discharge codes from an index hospitalization [9–11]. Diseases were encoded and entered into the model as a series of 10 dichotomous variables. For the severity of illness adjustment, we used both the nature of admission field assigned at the time of admission (elective, urgent, or emergent) and the specific operative procedure performed. Time was encoded as a categoric variable with three time periods: 1984 to 1989, 1990 to 1994, and 1995 to 1999.
Statistical analysis
Univariate analyses of outcomes with patient characteristics, hospital volume, and time periods were performed using the 
2 test, simple logistic regression, t test, Wilcoxin rank-sum test, analysis of variance, and simple linear regression when appropriate. Bivariate predictor variables with a p less than 0.1 were included in the multivariate analysis. We used the Shapiro-Wilk test to determine whether variables were normally distributed. Multiple logistic regression was used to test the association of hospital volume with in-hospital mortality while adjusting for important independent variables (univariate predictors). Variables were included in a sequential fashion while testing for goodness of fit using the Homer-Lemeshow method and receiver operator characteristic curves were calculated. The results are reported as adjusted odds ratio for death with 95% confidence intervals (CI). Multiple linear regression of log-transformed length of stay (LOS) and hospital charges was used to estimate the association of hospital volume after adjusting for other independent predictor variables. The log-transformed regression coefficient was exponentiated to obtain the percent change in the given outcome variable. Median regression was used to estimate the number of days and amount of charges associated with independent variables. Significance was set at p less than 0.05. All analyses were performed with STATA 5.0 (Houston, TX).
|
Results |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentDiscussionReferences |
|---|
Study population
From 1984 to 1999 there were 1,136 patients with a primary discharge diagnosis for esophageal resection in the Maryland Health Services Cost Review Commission database. The demographic characteristics of all patients and for each volume group are shown in Table 1. The population was predominantly male (74%) and had a mean age of 61 years. High volume hospitals treated a higher percentage of white patients (84%) than medium (70%) and low (7 2%) volume hospitals. Severity of illness as measured by admission field assignment was not significantly different by hospital volume. Comorbid disease diagnoses were similar across the volume strata except that fewer patients underwent operation for malignancy at the high volume hospitals (77% versus 87% at low and medium volume hospitals; p < 0.001). The concentration of esophageal operations in high volume centers increased dramatically during the study period from only 14% of patients during time period 1 (1984 to 1989) to 46% in time period 3 (1995 to 1999) (p < 0.001). Patients at high volume centers were more likely to be from out-of-state (37%) versus medium (7%) and low (3%) volume providers (p < 0.001).
|
|
|
In the multivariate analysis adjusting for case mix and time period, high hospital volume was associated with a fivefold reduction in the risk of in-hospital death (odds ratio, 0.21; 95% CI, 0.10 to 0.42; p < 0.001) (Table 3). The final logistic regression model accounted for differences among hospitals reasonably well as evidenced by its ability to discriminate between survivors and nonsurvivors (Homer-Lemeshow goodness of fit test, statistic = 0.34) and the area under the receiver operating characteristic curve was 0.730.
|
|
Hospital charges
The median hospital charges expressed in 1999 dollars for all patients were $33,483 (interquartile range, $22,722 to $55,186). High volume hospitals had significantly lower charges compared to low and medium volume hospitals during the 15-year study period (p < 0.001) (Table 2). After adjusting for inflation, there was a dramatic reduction in hospital charges regardless of volume during each time period (p < 0.001 for each time period).
In the univariate analysis, other independent variables associated with increased hospital charges were the following: male sex (p = 0.007), increasing age (p = 0.02), nonwhite race (p < 0.001), urgent admission (p = 0.001), emergent admission (p < 0.001), extent of procedure (p = 0.005), malignancy (p = 0.0004), metastatic disease (p = 0.009), and a history of myocardial infarction (p = 0.01).
In a multivariate analysis adjusting for case mix and time period, high hospital volume was independently associated with a 35% (95% CI, 28% to 41%; p < 0.001) decrease in hospital charges (Table 3). Using median regression, the decrease in charges at high volume hospitals was estimated to be $11,673 (95% CI, $9,504 to $12,841; p < 0.001). In addition, there was a 17% (95% CI, 7% to 26%: p = 0.001) reduction in hospital charges during time period 3 that was independent of hospital volume. Other independent variables that remained significant in the multivariate analysis for hospital charges are shown in Table 3.
|
Comment |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentDiscussionReferences |
|---|
This high volume effect likely represents a complex interaction of organizational and structural differences, processes of care, physician and nurse expertise, access to technology, and easily mobilized resources [7, 12]. Optimizing preoperative and perioperative care for patients with comorbid illness is crucial. Surgical proficiency and experience likely minimize adverse complications. During the past two decades, significant improvements have been made in the surgical technique and perioperative management of patients undergoing esophageal resection [5]. At least in part, this improvement is due to the dissemination and refinement of the transhiatal approach for esophagectomy [9, 10, 13, 14]. Anesthesia and critical care unit staffing are also vital to perioperative care, and therefore, outcome. For example, patients undergoing abdominal aortic operation have a threefold increase in perioperative mortality in hospitals that do not have daily rounds by an intensive care unit physician. In the same study, inadequate nurse to patient ratio was associated with an increase in complications and intensive care unit LOS [15]. Furthermore, certain processes of care, such as critical pathways and protocols, have been shown to decrease cost and reduce the incidence of pulmonary complications after esophageal operation [16, 17]. High volume centers may be likely to develop and incorporate these care processes into their practice, and this may partially explain the relationship between volume and outcome. Additional studies exploring the factors contributing to the relationship of hospital volume to outcomes are needed.
In addition to the effect of hospital volume, there has been overall improvement in both clinical and economic outcomes during the past 15 years. In-hospital mortality, LOS, and inflation-adjusted charges are all significantly reduced over this time period regardless of esophageal surgical volume. Although the improved outcomes can be partially explained by a greater concentration of esophageal operation in high volume hospitals, even after adjusting for volume in the multivariate analysis, a statistically significant improvement in all three outcomes persists over time. These results are consistent with a large single institution report by Orringer and colleagues [8] of 1,085 patients who underwent transhiatal esophagectomy. They reported a reduction in perioperative mortality to 4% and an average LOS of 7 days during the most recent time period.
The effect of a single surgeon at any hospital could not be determined from this study. Three excellent surgeons at the high volume hospitals could have influenced the improved outcome attributed to high volumes. However, the impact of one individual at a low volume hospital is not likely to cause an undue effect outside of the effect of low surgeon volume. These two factors (surgeon and hospital volume) should be colinear.
The strengths of our study are the large number of patients, the diverse character of hospitals and surgeons (academic and community practitioners), the longitudinal analysis of outcomes over three time periods, and the adjustment of patient case mix. Furthermore, because Maryland is a relatively small state, evidence-based hospital referral is feasible and this study estimates the potential impact such a policy could achieve. Analyzing both the effects of time and hospital volume allowed us to correct for changes in outcomes over time, which is an important consideration for esophageal resection. However, we acknowledge several limitations to this study. Although we adjusted for case mix using patient demographics, 10 comorbid disease diagnoses, nature of admission, and the extent of procedure, if patients differ in other important prognostic variables in a systematic fashion, outcome differences may not be attributable solely to volume differences. In addition, a prospectively acquired physiologic score, such as the APACHE III, rather than relying on limited diagnostic coding, would allow for more robust risk adjustment. A second limitation is inferring the direction of causality between outcomes and volume. Patients and referring physicians might seek out high-quality providers and thereby increase hospital volume—the "selective referral" hypothesis—or conversely, providers with higher volume might refine techniques and abilities thereby improving outcome—the "practice makes perfect" hypothesis [18]. Furthermore, we were limited to studying in-hospital mortality instead of 30-day or long-term mortality, the latter of which has been related to hospital volume for pancreaticoduodenectomy in the Medicare population [19]. Also, we did not have any information on adjuvant and neoadjuvant therapy for esophageal malignancy, which can have effects on both in-hospital and long-term mortality [20, 21]. The final limitation is the external validity of our findings. Maryland, like few other states, has a single regional center that performs the highest volume of esophageal operations in the state. Whether our results can be applied to other states with different hospital profiles is unknown.
In summary, in-hospital mortality, LOS, and hospital charges were significantly lower at high volume centers compared to lower volume centers during each time period studied. In addition, we found an overall improvement in these same three outcomes during the entire 15-year study period including an almost 70% reduction in in-hospital mortality. Because outcomes at low and medium volume centers did not improve over time, this statewide reduction in mortality rate in recent years likely results from not only improved surgical technique and perioperative management, but also the increased proportion of esophageal procedures now performed at high volume centers. This reduction in mortality was seen in all ages, but particularly in elderly patients. Consequently, referring patients to a high volume center will reduce mortality and resource utilization for patients undergoing esophageal resection.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentDiscussionReferences |
|---|
The "take-home" messages are clear and profound. Large volume hospitals were associated with a case mix-adjusted, fivefold reduction in mortality after esophagectomy as compared to medium and low volume centers. In addition, length of stay was analyzed as a surrogate for morbidity, and although all hospitals enjoyed length of stay reductions, during this 15-year study, high volume centers had a highly statistically significant length of stay advantage, equivalent to 6 days over their smaller volume colleagues.
When attention was turned to an economic impact analysis, all institutions were successful in reducing charges during the 15-year period. In case mix and inflation-adjusted dollars, this amounted to a 17% reduction in charges independent of hospital volume. But when the high volume centers were analyzed alone, they outperformed these other hospitals 2:1, with a 35% decrease in charges, calculated to be equivalent to more than $11,000 in savings per patient.
Such data provides compelling momentum for the application of evidence-based practice in the management of esophageal disease. Acknowledging that the derived benefits of large volume centers are multifactorial and reflect not only refinement in surgical technique but also the ready availability of experienced multidisciplinary consultant services, these investigators suggest, and I would concur, that we now have evidence to make referral of patients requiring esophageal resection to established high volume centers the standard of care.
It is time to check our surgical egos at the door and realize that patient outcome is not predicted only by how well we perform in the operating room. As they note in their discussion, for every 8 patients referred from a low volume center to a high volume center, one death could be prevented.
I have several questions.
The case mix adjustments were based on whether comorbid diseases in 10 categories were reported. How can you control for variability in coder quality between small, medium, and large reporting centers?
In the manuscript, "extent" of resection was a univariable predictor of mortality and of hospital charges. Could you please elaborate on this stratification? Does this reflect more transhiatal resections or more esophagectomies for benign disease, or both?
Third, esophageal operation, as with much of general thoracic surgery, especially outside of academic centers and large metropolitan areas, continues to be performed by general surgeons or surgeons without formal thoracic training. Do you have any data on the board qualifications of the surgeons performing esophagectomy in Maryland during this time period?
And finally, preoperative nutritional status has been demonstrated to be an important independent predictor of outcome after both pulmonary and esophageal resection. Could this be a confounding factor in your analysis? Was there any information in the database to substantiate or refute the relationship of nutritional status to postsurgical outcome?
I congratulate you on a well-conducted and analyzed study, and I thank the Society for the privilege of the floor.
DR CATTANEO: Thank you, Dr DeCamp, for all of your gracious comments on our paper. To address the initial question about the adjustment for comorbid disease, we did use secondary diagnostic and procedural coding to help establish comorbid diseases for the separate patients. We recognize, however, that this is a limitation of our study. Also, if our patients had differed systematically by any other variable that did not show up in the coding procedure, then our data could be affected.
It was initially significant in the univariate analysis as to whether the extent of resection involved partial or a complete resecton of the esophagus. When we adjusted for extent of resection in the multivariate analysis, however, this fell out as not being an important predictor of our three primary outcome variables.
With regard to benign disease and whether procedures were performed by a transhiatal approach or by a transthoracic approach, we did find that initially high volume hospitals had a significantly decreased percentage of procedures performed for benign disease as compared to low and medium volume hospitals. Once again, when we performed our multivariate analysis, this fell out as not being an important predictor of our outcome variables. We were able to determine in the database whether procedures were performed either transhiatally or transthoracically; however we did not include this as part of our analysis. We looked specifically at whether esophagectomy was performed but not by one particular method.
We would all like to have data as to whether subspecialty training affects in-hospital mortality, length of stay, and hospital charges. This, however, is a limitation again of the Health Services Cost Review Commission database in that we do not know what subspecialty training, if any, the different providers had. I can say from our own institution that more than 80% to 90% of esophagectomies are performed by general thoracic surgeons.
Finally, nutritional status was not a variable that we specifically analyzed. It was, unfortunately, not something we could adequately abstract from our database, but certainly it is important and could potentially be studied in the future.
DR JACK M. MATLOFF (Los Angeles, CA): I would like to just make a couple of observations, one being that this is the issue that I spoke about yesterday in terms of our need to look not only at our individual patient obligations and our experiences but also to populations. The second issue is clearly that we have to have better databases, and Dr Wright’s comments about a thoracic surgical database rather than using administrative data I think is critical at this point. A third point is that the maturity of practices have to make us very wary of endorsing the simple concept of volume and outcome. For instance, in coronary surgery where the argument seems to be proceeding most vociferously at this point in time, the maturity of the practice seems to blunt the effect of the volume. So it is not an all or none situation.
And, finally, more to come!
|
References |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentDiscussionReferences |
|---|
Related Article
This article has been cited by other articles:
|
|
 |
|
  E. S. Weiss, R. A. Meguid, N. D. Patel, S. D. Russell, A. S. Shah, W. A. Baumgartner, and J. V. Conte Increased mortality at low-volume orthotopic heart transplantation centers: should current standards change? Ann. Thorac. Surg., October 1, 2008; 86(4): 1250 - 1260. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. W. Wouters, B. P. Wijnhoven, H. E. Karim-Kos, H. G. Blaauwgeers, L. P. Stassen, W.-H. Steup, H. W.Tilanus, and R. A. Tollenaar High-Volume versus Low-Volume for Esophageal Resections for Cancer: The Essential Role of Case-Mix Adjustments based on Clinical Data Ann. Surg. Oncol., January 1, 2008; 15(1): 80 - 87. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Sundaresan, B. Langer, T. Oliver, F. Schwartz, M. Brouwers, H. Stern, and Expert Panel on Thoracic Surgical Oncology Standards for Thoracic Surgical Oncology in a Single-Payer Healthcare System Ann. Thorac. Surg., August 1, 2007; 84(2): 693 - 701. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. W. Steyerberg, B. A. Neville, L. B. Koppert, V. E.P.P. Lemmens, H. W. Tilanus, J.-W. W. Coebergh, J. C. Weeks, and C. C. Earle Surgical Mortality in Patients With Esophageal Cancer: Development and Validation of a Simple Risk Score J. Clin. Oncol., September 10, 2006; 24(26): 4277 - 4284. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. R. DeMeester Adenocarcinoma of the Esophagus and Cardia: A Review of the Disease and Its Treatment Ann. Surg. Oncol., January 1, 2006; 13(1): 12 - 30. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. B. Dimick, R. M. Wainess, G. R. Upchurch Jr, M. D. Iannettoni, and M. B. Orringer National Trends in Outcomes for Esophageal Resection Ann. Thorac. Surg., January 1, 2005; 79(1): 212 - 216. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Jain, R. Pietrobon, S. Hocker, U. Guller, A. Shankar, and L. D. Higgins The Relationship Between Surgeon and Hospital Volume and Outcomes for Shoulder Arthroplasty J. Bone Joint Surg. Am., March 1, 2004; 86(3): 496 - 505. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. B. Dimick, J. A. Cowan Jr, L. M. Colletti, and G. R. Upchurch Jr Hospital Teaching Status and Outcomes of Complex Surgical Procedures in the United States Arch Surg, February 1, 2004; 139(2): 137 - 141. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. B. Dimick, J. A. Cowan Jr, G. Ailawadi, R. M. Wainess, and G. R. Upchurch Jr National Variation in Operative Mortality Rates for Esophageal Resection and the Need for Quality Improvement Arch Surg, December 1, 2003; 138(12): 1305 - 1309. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. S. Panageas, D. Schrag, E. Riedel, P. B. Bach, and C. B. Begg The Effect of Clustering of Outcomes on the Association of Procedure Volume and Surgical Outcomes Ann Intern Med, October 21, 2003; 139(8): 658 - 665. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. M. Shahian and S.-L. T. Normand The volume-outcome relationship: from Luft to Leapfrog Ann. Thorac. Surg., March 1, 2003; 75(3): 1048 - 1058. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. B. Dimick, P. J. Pronovost, J. A. Cowan, and P. A. Lipsett Surgical volume and quality of care for esophageal resection: do high-volume hospitals have fewer complications? Ann. Thorac. Surg., February 1, 2003; 75(2): 337 - 341. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. B. Dimick, J. A. Cowan Jr, J. A. Knol, and G. R. Upchurch Jr Hepatic Resection in the United States: Indications, Outcomes, and Hospital Procedural Volumes From a Nationally Representative Database Arch Surg, February 1, 2003; 138(2): 185 - 191. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. B. Dimick, P. J. Pronovost, J. A. Cowan Jr, and P. A. Lipsett Postoperative Complication Rates After Hepatic Resection in Maryland Hospitals Arch Surg, January 1, 2003; 138(1): 41 - 46. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. R. Doty, J. D. Salazar, A. A. Forastiere, E. I. Heath, L. Kleinberg, and R. F. Heitmiller Postesophagectomy morbidity, mortality, and length of hospital stay after preoperative chemoradiation therapy Ann. Thorac. Surg., July 1, 2002; 74(1): 227 - 231. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
