|
|
||||||||
Section of Thoracic Surgery, Yale University School of Medicine, New Haven, Connecticut
Accepted for publication June 14, 2011.
* Address correspondence to Dr Kim, Yale University School of Medicine, Section of Thoracic Surgery, 330 Cedar St, BB 205, New Haven, CT 06520 (Email: anthony.kim{at}yale.edu).
| Abstract |
|---|
|
|
|---|
Methods: The 2008 Healthcare Utilization Project-Nationwide Inpatient Sample database was culled using the International Classification of Diseases (9th Clinical Modification) procedure codes specifically distinguishing VATS vs open lobectomies (32.41 and 32.49, respectively) available only after October 2007. High hospital VATS volume was defined as 95th percentile or higher (> 20 VATS/year). Univariable and multivariable analyses were used to identify independent predictors of the following outcome measures: 30-day in-hospital morbidity and mortality, hospital length of stay (LOS), and hospital costs.
Results: We identified 6,292 primary lung cancer patients undergoing pulmonary lobectomy, including 1,523 undergoing VATS (24%). Compared with open, VATS patients had fewer complications (38% vs 44%, p < 0.001) and median LOS (5 vs 7 days; p < 0.001). In multivariable analysis, VATS was an independent predictor of fewer total complications (odds ratio, 0.83; p = 0.004) and shorter LOS (2.3 ± 0.3-day difference, p < 0.001). Patients undergoing VATS at high-volume VATS hospitals had shorter median LOS (4 vs 6 days, p = 0.001) compared with low-volume VATS hospitals. Multivariable analysis showed high hospital VATS volume independently predicted shorter LOS (0.9 ± 0.4-day difference, p = 0.001).
Conclusions: In a national database, VATS lobectomy was associated with fewer complications and shorter LOS than open lobectomy in primary lung cancer patients. Among patients undergoing VATS, high hospital volume was also associated with shorter LOS.
| Introduction |
|---|
|
|
|---|
| GENERAL THORACIC SURGERY:
The Annals of Thoracic Surgery CME Program is located online at http://cme.ctsnetjournals.org. To take the CME activity related to this article, you must have either an STS member or an individual non-member subscription to the journal.
|
Interest in minimally invasive surgical interventions, including video-assisted thoracoscopic surgery (VATS), has grown steadily since the first report of VATS lobectomy for lung cancer in 1994 [1]. This procedure was initially controversial, however, due to concerns about inferior oncologic results. Multiple studies have since supported the long-term oncologic equivalence and perioperative benefit in complications, hospital length of stay (LOS), and costs of VATS compared with conventional open thoracotomy, particularly after it was demonstrated that VATS could be performed without rib spreading [2–7].
The primary criticism of these studies is that these comparisons were made at single-institution, high-volume centers of excellence in VATS. Given the lack of a large prospective, randomized trial comparing VATS vs open lobectomy and the low likelihood of such a trial being performed in the future, a population-based comparison of perioperative clinical and economic outcomes is needed. Two previous analyses were published by two independent groups in 2010 using a national inpatient database [8, 9], but both studies used incorrect procedure codes to identify the VATS vs open lobectomy cohorts, thus limiting the ability to draw any conclusion from these studies.
Another important question for a population-based comparison of VATS lobectomy is the role of hospital volume on perioperative and oncologic outcomes. In 2001 Bach and colleagues [10] published a seminal study demonstrating that on a national scale, patients with stage I to IIIA non-small cell lung cancer resected at high-volume hospitals had fewer postoperative complications, a lower 30-day mortality rate, and increased 5-year survival than similar patients treated at low-volume hospitals. Several other reports since then have shown a positive association between increased hospital volume and improved patient outcomes after several other types of high-risk cardiothoracic operations, including coronary artery bypass grafting, aortic valve replacement, and esophagectomy [11–13].
To date, the effect of hospital volume on outcomes for lung cancer patients specifically after VATS lobectomy is unknown. The aims of this population-based analysis were (1) to compare perioperative clinical and economic outcomes after VATS vs open lobectomy using accurate procedure codes, and (2) among patients undergoing VATS, to compare perioperative clinical and economic outcomes after VATS at high-volume vs low-volume hospitals.
| Material and Methods |
|---|
|
|
|---|
Patient Selection Criteria
Procedure codes are based on the International Classification of Diseases, 9th edition, Clinical Modification (ICD-9-CM), Volume 3, which was created by the National Center for Health Statistics (NCHS) as an extension of the ICD-9 classification initially published in 1977 by the World Health Organization (WHO). All changes and modifications to the ICD-9-CM are overseen by the NCHS and the Center for Medicare and Medicaid Services. Before October 2007, lung lobectomy was coded nonspecifically with ICM-9-CM procedure code 32.4, but beginning in October 2007, this code was replaced by two new codes specifying thoracoscopic (32.41) vs all other surgical approaches (32.49) to lung lobectomy. We chose 2008 was chosen because it was the only full calendar year in which the new codes were used and that was available in the HCUP-NIS at the time of the study. Patients undergoing redo thoracotomy, indicated by ICD-9-CM procedure code 34.03, were excluded.
Independent Variables
The primary independent variables, surgical technique (VATS vs open) and hospital VATS volume, were modeled as categoric variables. Hospitals were designated as high-volume VATS centers if they were in the 95th percentile if more than 20 VATS were performed there per year and as high-proportion VATS centers if their VATS/total lobectomy ratio exceeded 50%.
Independent patient demographic variables included age, sex, race, median household income, and primary payer, defined as private/Health Maintenance Organization, Medicaid, self-pay, Medicare, no charge, and other. Categories for median household income were $1 to $38,999 (low), $39,000 to $47,999 (medium-low), $48,000 to $63,000 (medium), and more than $63,000 (high).
Comorbidity scores were calculated using an adaptation of the Charlson Comorbidity Index [14]. Because all patients in our cohort had a minimum Charlson score of 2 due to their diagnoses of primary lung cancer, Charlson scores of 2 to 3 were categorized as "low," 4 to 5 as "medium," and 6 or more as "high."
Other hospital-provider variables included hospital region (Northeast, Midwest, South, and West), location (urban and rural), and teaching status (teaching and nonteaching). All independent variables were treated as categoric variables.
Outcome Variables
The outcomes of interest were (1) in-hospital total and pulmonary complications, (2) mean LOS, (3) total inpatient hospital charges, and (4) in-hospital death. Perioperative complications were categorized as wound-related, infectious, cardiovascular, intraoperative, systemic, gastrointestinal, urinary, and pulmonary (adapted from the Appendix in Gopaldas and colleagues [8]). Information regarding complication severity was not available, so complications were treated as a dichotomous variable (0 vs
1).
Statistical Analysis
Univariable analyses of the independent variables with our outcomes of interest were performed by
2 statistical analysis to compare proportions of total complications and pulmonary complications, and by Mann-Whitney nonparametric analysis to compare medians of LOS and costs. Multivariable logistic regression models were used to adjust for independent variables for in-hospital total and pulmonary complications, and multivariable linear regression models were used to adjust for significant independent variables for LOS and total in-patient hospital costs. These analyses were adjusted for demographic and clinical patient and provider distributions. A backward elimination procedure was applied for independent variable selection. Reference comparisons were selected according to standard procedure for the HCUP database. Data analysis and management were performed using SPSS 16.0 software (SPSS Inc, Chicago, IL) and Excel 2007 (Microsoft Corp, Redmond, WA). All tests were two-sided, with statistical significance set at a value of p < 0.05.
| Results |
|---|
|
|
|---|
|
|
|
|
|
|
| Comment |
|---|
|
|
|---|
Perioperatively, open lobectomy has been reported to have a complication rate of 32% to 37%, according to the prospective American College of Surgeons Oncology Group Z0030 clinical trial of 766 patients [17] and a retrospective analysis of The Society of Thoracic Surgeons database of 6,042 patients [18]. For VATS lobectomy, the prospective Cancer and Leukemia Group B 39802 clinical trial of 127 patients reported major perioperative (grade
3) morbidity in only 7% of patients, although total perioperative morbidity was not reported [8]. When VATS and open lobectomy were compared in a small prospective study of 55 patients, total complications were significantly less frequent when VATS was used as the initial approach [5]. These results have been supported by other recent studies [6, 7], including a propensity-matched analysis from The Society of Thoracic Surgeons database [19].
Two other studies attempting to compare VATS vs open lobectomy using the HCUP-NIS database identified patients in the two cohorts with incorrect procedure codes [8, 9]. Because the procedure codes distinguishing VATS lobectomy from open lobectomy only began to be used in October 2007, the study periods of 2004 and 2006 by Gopaldas and colleagues [8] that compared diagnostic transpleural thoracoscopy vs nonspecific lobectomy and 2007 by Hannan and colleagues [9] that compared VATS lobectomy vs nonspecific lobectomy (the latter of which also includes VATS patients from January to September 2007) were too early to perform this analysis using the HCUP-NIS database. Therefore, our study period of 2008 was the first full year available to analyze this comparison during these data. We believe that our findings have more face validity than those of Gopaldas and colleagues [8] and Hannan and colleagues [9] because our reported rates of VATS lobectomy are more consistent with those previously reported in the literature (24% vs 5% and 7%, respectively).
Of note, men with Medicare or Medicaid insurance had consistently worse perioperative outcomes overall, even after multivariable adjustment of demographic and clinical factors such as age, income, and comorbidity. Although unavailable covariates, such as tumor stage, may play a strong role, this represents a curious finding that may be multifactorial in etiology. In addition, the variability in inpatient LOS and costs across different geographic regions of the United States may be influenced by a variety of factors not measured in the database, including but not limited to differences in billing practices, expenses of perioperative staff and resources, and attitudes of patients and providers regarding discharge variables.
In the second part of our study, among the 1,523 VATS patients, patients who underwent VATS at hospitals with high annual VATS volume (> 20 VATS per year) had a significantly shorter hospital LOS than those at hospitals with low annual VATS volume, but had equivalent total hospital costs and short-term rates of morbidity and mortality. Interestingly, patients at hospitals that used VATS for most of their annual lung cancer lobectomies had both fewer complications and shorter hospital LOS than those at hospitals with a low ratio of VATS to total lobectomy.
Using VATS proportion rather than VATS volume as a proxy for experience may reflect a decrease in case selection bias, because surgeons at hospitals with mature VATS programs may be less inclined to avoid using VATS for high-risk patients. For lung cancer operations overall, hospital volume has been shown to be associated with both short-term morbidity and long-term mortality rates on a national scale [10]. There may also be other provider-related factors involved in lung cancer operations. Recent evidence from HCUP-NIS suggests that in-hospital lung cancer deaths may be reduced in teaching hospitals compared with nonteaching hospitals at all but the highest volume institutions [20] and that general thoracic surgeons may achieve better mortality rates and hospital LOS outcomes after decortications, segmentectomies, lobectomies, and pneumonectomies than surgeons not specializing in cardiothoracic surgery [21]. The provider-related covariates that we examined, including hospital geography, urban vs rural location, and teaching status, appeared to have minimal effect on in-hospital morbidity and mortality rates among VATS patients specifically.
The limitations of this study include those inherent to any retrospective analysis of a large administrative database, although HCUP-NIS is widely used and has been well validated. Some degree of treatment selection bias is possible in the VATS vs open cohorts and the high-volume vs low-volume cohorts. However, the available demographic and clinical characteristics that could have potentially confounded the results of this study were adjusted for in multivariable analysis, which can isolate the independent predictors of outcomes. Long-term outcomes cannot be assessed from the database, and readmissions are not captured; thus, the observed complication rates may be underestimated.
The hospital LOS was recorded as the date of admission to the date of discharge because the date of the operation was not recorded; thus, specific postoperative LOS could not be ascertained. However, most lobectomies are performed as same-day admissions, so it is likely that only a small number of patients were actually hospitalized before the operation. Other potential modifying factors not captured by the database include number of years in practice, surgeon specialty or board certification, cumulative surgeon or hospital experience with thoracoscopic or open lobectomy, and pathologic characteristics and staging of the lung cancers.
In conclusion, our study has demonstrated, on a national scale, that VATS lobectomy is independently associated with fewer total complications and shorter LOS for patients with primary lung cancer compared with open lobectomy. For patients undergoing VATS lobectomy, high hospital VATS volume is also associated with shorter LOS, whereas a high hospital VATS/total lobectomy ratio is associated with fewer total complications and shorter LOS.
Although this type of retrospective analysis has the potential for selection bias, the findings of this population-based study are largely consistent with other published studies and support the notion that VATS is a reasonable and perhaps preferred approach for lung cancer lobectomy given the benefit in short-term morbidity and hospital LOS stay and equivalence in total hospital costs and short-term mortality. In addition, experienced VATS centers may also be recommended given a similar benefit in short-term morbidity and hospital LOS. As the national experience with VATS and the subsequent collection of VATS-specific data increases, further research will be necessary to validate this study.
| Acknowledgments |
|---|
|
|
|---|
| References |
|---|
|
|
|---|
Related Article
This article has been cited by other articles:
![]() |
H. Launer, D. V. Nguyen, and D. T. Cooke National perioperative outcomes of pulmonary lobectomy for cancer in the obese patient: A propensity score matched analysis J. Thorac. Cardiovasc. Surg., May 1, 2013; 145(5): 1312 - 1318. [Abstract] [Full Text] [PDF] |
||||
![]() |
L. Konge, R. H. Petersen, H. J. Hansen, and C. Ringsted No extensive experience in open procedures is needed to learn lobectomy by video-assisted thoracic surgery Interact CardioVasc Thorac Surg, December 1, 2012; 15(6): 961 - 965. [Abstract] [Full Text] [PDF] |
||||
![]() |
D. T. Cooke and D. H. Wisner Who Performs Complex Noncardiac Thoracic Surgery in United States Academic Medical Centers? Ann. Thorac. Surg., October 1, 2012; 94(4): 1060 - 1064. [Abstract] [Full Text] [PDF] |
||||
![]() |
A. W. Kim, F. C. Detterbeck, D. J. Boffa, R. H. Decker, P. R. Soulos, L. D. Cramer, and C. P. Gross Characteristics Associated With the Use of Nonanatomic Resections Among Medicare Patients Undergoing Resections of Early-Stage Lung Cancer Ann. Thorac. Surg., September 1, 2012; 94(3): 895 - 901. [Abstract] [Full Text] [PDF] |
||||
![]() |
K. Kernstine A Lobectomy by Any Other Name J. Clin. Oncol., August 10, 2012; 30(23): 2803 - 2804. [Full Text] [PDF] |
||||
![]() |
T. L. Demmy Invited Commentary Ann. Thorac. Surg., February 1, 2012; 93(2): 379 - 380. [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 |