| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2007;83:1837-1843
© 2007 The Society of Thoracic Surgeons
a Department of Surgery, Division of Thoracic Surgery, Taipei Medical University and Hospital, Taipei, Taiwan
b Department of Surgery, Division of General Surgery, Taipei Medical University and Hospital, Taipei, Taiwan
c School of Health Care Administration, Taipei Medical University, Taipei, Taiwan
Accepted for publication December 6, 2006.
* Address correspondence to Dr Lin, School of Health Care Administration, Taipei Medical University, 250 Wu-Hsing St, Taipei 110, Taiwan (Email: henry11111{at}tmu.edu.tw).
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Methods: We use pooled data for the years 2001 through 2004 obtained from the National Health Insurance Research Database in Taiwan. A total of 4,841 patients, identified as having undergone pulmonary resections for lung or bronchial tumors during the period of this study, were treated by 377 surgeons in 79 hospitals. Multivariate logistic regression analyses were then employed to assess the crude and adjusted odds ratio of in-patient fatalities between surgeon and hospital lung cancer resection volume groups.
Results: Patients treated by low-volume surgeons had significantly higher in-hospital fatality rates than those treated by either medium-volume surgeons (2.3% versus 1.0%; p < 0.001) or high-volume surgeons (2.3% versus 0.6%; p < 0.001). However, hospital case volume alone is not a significant predictor of hospital in-patient fatalities for lung cancer resections. With increasing surgeon volume, there was a decline in the adjusted odds ratio of hospital in-patient deaths. The odds of hospital in-patient deaths for those patients treated by low-volume surgeons were 2.04 times those of medium-volume surgeons, and 2.63 times those of high-volume surgeons.
Conclusions: We conclude that after adjusting for patient, surgeon, and hospital characteristics, an inverse volume-outcome relationship does exist for surgeons, but not for hospitals, in Taiwan.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Some studies have reported no significant trend toward lower operative mortality rates for hospitals with a high volume of lung cancer resections [6–8], while others have reported significantly lower in-hospital mortality rates for high-volume providers undertaking similar resections [9, 10]. Most of these studies were, however, conducted on hospital-level volume alone whereas we have examined the simultaneous contribution to patient outcomes from both hospital and surgeon volume. Furthermore, we have found no examples of similar lung cancer volume-outcome studies having been conducted in Asia.
In this study, we use 4-year population-based data on Taiwan to examine the association between the volume of lung cancer resections (by both surgeons and hospitals) and the subsequent in-hospital mortality rates. We hypothesize that high-volume providers will be associated with superior treatment outcomes for patients undergoing lung cancer resections.
|
Material and Methods |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Study Sample
The study sample was identified from the pooled database by the diagnostic code 162.XX. We identified a total of 157,683 hospitalizations (45,610 patients) for malignant bronchus and lung neoplasms covering the period from January 2001 to December 2004. Of these, a total of 4,841 patients, treated by 377 surgeons in 79 hospitals, were identified as having undergone pulmonary resections for lung or bronchial tumors during the period under examination.
Surgeon and Hospital Lung Cancer Resection Volume Groups
As unique physician and hospital identifiers are available within the NHIRD for each medical claim submitted, that enabled us to identify the same physician, or the same hospital, carrying out one or more lung cancer resections between January 2001 and December 2004.
Surgeons and hospitals were sorted, in ascending order of their total volume of lung cancer resections, with the cutoff points (high, medium, and low) being determined by the volume that most closely sorted the sample patients into three groups that were roughly equal in size. This method is consistent with the methodologies adopted in many of the prior studies.
The sample of 4,841 patients was subsequently divided into three surgeon volume groups: 46 cases or fewer (hereafter referred to as low volume), 47 to 131 cases (medium volume), and 132 cases or more (high volume); whereas the three hospital volume groups were 135 cases or fewer (hereafter referred to as low volume), 136 to 467 cases (medium volume), and 468 cases or more (high volume).
Key Variables of Interest
Adjustments were made for surgeon, hospital, and patient characteristics in the overall assessment of the relationship between in-hospital deaths and volume groups. The primary study outcome was in-hospital mortality, with patient as the unit of analysis, and the key independent variables were the lung cancer resection volume groups for both surgeons and hospitals. We define in-hospital mortality as the death of a patient at any time after operation either at the operating hospital or upon transfer to another hospital.
Surgeon characteristics included the surgeon’s age (as a surrogate for practice experience) and sex; hospital characteristics included hospital ownership, hospital level, and geographical location, with the hospital ownership variable being recorded as one of three types: public, private not-for-profit, and private for-profit hospitals. Within the hospital level variable, each hospital was classified as a medical center (with a minimum of 500 beds), a regional hospital (minimum 250 beds), or a district hospital (minimum 20 beds); hospital level can therefore be used as a proxy for both hospital size and clinical service capabilities. Hospital teaching status was not included in this study since all medical centers and regional hospitals in Taiwan are teaching hospitals.
Patient characteristics consisted of age, sex, severity of illness, and type of operation. Because no illness severity index is available in Taiwan, we used the Charlson Comorbidity Index (CCI) to quantify preexisting comorbidities as a means of adjusting for the higher mortality risks associated with comorbidities (the higher the score, the greater the comorbidity). The type of operation consisted of wedge resection, segmental resection, lobectomy, and pneumonectomy.
Statistical Analysis
The SAS statistical package (SAS System for Windows, version 8.2; SAS Institute, Cary, North Carolina) was used to perform the statistical analysis of the data in this study. Global 
2 analyses were conducted to examine the relationship between surgeon and hospital volume lung cancer resection groups and the unadjusted hospital in-patient fatality rates. Multivariate logistic regression analyses were also employed to assess the crude and adjusted odds ratio of hospital in-patient fatalities between surgeon and hospital lung cancer resection volume groups.
Finally, the generalized estimated equation method was also adopted as a means of accounting for any clustering of the sampled patients among particular surgeons or hospitals. Clustering would indicate the greater likelihood for a given provider’s patient outcomes to be similar to each other, as opposed to being similar to the patient outcomes of a different provider. A two-sided p value of 0.05 or less was considered to be statistically significant.
|
Results |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Details of the distribution of surgeons and patients for lung cancer resections, by surgeon volume, are provided in Table 1, which reveals that lung cancer resections were performed by 377 surgeons between 2001 and 2004, at a mean volume per surgeon of 13 operations. The surgeons in the high-volume group were more likely to be older (p < 0.001) and performed more lobectomies and pneumonectomies (p < 0.001). The 2 analyses also indicate that significant relationships exist between surgeon volume groups and patient sex (p = 0.004), age (p = 0.003), and CCI scores (p < 0.001).
|
|
|
|
|
Comment |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
One likely reason for our departure from the US findings on pulmonary resections is the potential confounding effect of the health insurance system in Taiwan. About 97% of all Taiwanese citizens have been covered under the NHI system since its inception in March 1995, with all patients having free access to any health care provider of their choice in Taiwan. This contrasts sharply with many other health care delivery systems around the world, where there may often be a tendency to limit a patient’s choice to certain providers. Such limitations on patient choice may well result in confounding the relationship between provider volumes and patient outcomes. Furthermore, the study sample in the study by Hannan and associates [10] paper was limited to those patients who had undergone lobectomies, as compared with the four types of resections examined in this study.
The literature suggests two possible explanatory hypotheses for the inverse volume-outcome relationship [11, 12]. The first of these is "selective referral," which suggests that selective referral either by physicians or by the patients themselves will ultimately lead to the referral of more patients to providers renowned for superior treatment outcomes; thus, these providers would inevitably find themselves performing higher volumes of lung cancer resections.
As noted earlier, under the health insurance system in Taiwan, patients have the freedom to choose their preferred provider; thus, based upon word-of-mouth recommendations from relatives or friends, physicians with good reputations for superior outcomes, even those within the departments of the same hospital, will tend to attract greater numbers of patients [13]. Therefore, leaving aside physician referrals, self-referrals by patients may be a major contributor to the inverse relationship between patient outcomes and surgeon volumes, particularly in Taiwan.
The second hypothesis, "practice makes perfect," is based upon the rationale that a larger volume of patients allows providers to develop better levels of skill and expertise in the management of their operations or treatment procedures. Therefore, high-volume providers are more likely to achieve better clinical performance owing to their greater skills and experience.
In our study, the physician-specific volume is a stronger predictor of outcomes than hospital volume. Furthermore, such highly skilled, high-volume surgeons are more likely to be older and to perform more complex procedures, such as lobectomies and pneumonecotomies. Such results add support to the practice makes perfect hypothesis.
Of the total of 45,610 patients in Taiwan diagnosed with lung cancer between 2001 and 2004, only 4,841 patients (10.6%) received surgical lung or bronchial neoplasm resections. Although, all pathologic subtypes of lung cancer was included in our series, the resection rate of lung cancer was far lower than the 16.5% resection rate recently reported in Norway based upon population-based data [14]. While it is difficult to substantiate the reasons for this low resection rate in our study design, the prevalence of pulmonary tuberculosis in Taiwan may be one of the main reasons for the low lung cancer resection rate.
Surgical resection for lung cancer is still regarded as the most effective method for early-stage nonsmall-cell lung cancers, and is generally offered to all patients at stage I and II, and specific groups of patients at stage IIIa (N2) of the disease, but is not usually introduced to those at stage IIIb (N3). Ipsilateral (N2) or contralateral (N3) mediastinal lymphadenopathy is therefore a critical point with regard to the decision on whether the lung cancer can be resected.
According to the literature, patients with pulmonary tuberculosis had about 90% mediastinal lymphadenopathy [15], and even at that stage, it was difficult to differentiate between benign and malignant abnormalities by positron emission tomography [16]. In an area with a high prevalence of pulmonary tuberculosis, such as that which exists in Taiwan, the rationale for tuberculosis-related mediastinal lymphadenopathy complicates the evaluation of the preoperative clinical stage of lung cancer. As we can expect to see more patients being initially diagnosed with clinical N2/N3 lung cancer diseases than the actual numbers, that may explain why the lung cancer resection rate in Taiwan is so low.
There have been a number of recent reports documenting low postoperative mortality after lung cancer surgery [9, 10, 14, 17]. Of these, a Japanese survey showed much improved 30-day mortality rates (at 3.2% for pneumonectomies, 1.2% for lobecotmies, and only 0.8% for lesser resections) [17]. In this study, we also find improved in-hospital mortality rates, at 3.17% for pneumonectomies and just 0.57% for lobectomies (not shown in the Tables), with these figures being much lower than those contained in many of the prior reports.
This study has also examined all of the different operative types of lung cancer resections that resulted in significant hospital in-patient deaths. As compared with lobectomies, wedge resections demonstrated poor short-term outcomes; however, this finding does not gain support from the results reported by any single institute [14]. One reason for this major difference may be that more sublobar resections were performed by low-volume surgeons with high in-hospital mortality rates; that could potentially lead to wedge resections resulting in higher in-hospital mortality rates than those reported for lobectomies.
Another likely explanation for such disparity may be the differences in characteristics between lung cancer lobecotomy and wedge resection patients. In general, a wedge resection is carried out on lung cancer patients who have poor cardiopulmonary function, which would make such patients intolerant to a lobectomy of the lung. Actually, about 39% of the sample patients were in the CCI score category of "6 or more"; 44% of them received sublobar resection, a percentage that is consistent with the documented literature (not shown on Table). Although we have controlled for patient comorbidities, the administrative database adopted for this study was extremely limited in its ability to account for differences in such severity of illness.
This study suffers from two limitations that should be addressed. First of all, although the outcome measure in this study was in-hospital mortality, that did not include postdischarge deaths occurring as a direct result of the surgical procedure. The differences in mortality rates could be partly attributable to this factor. Secondly, the information on the stage of the disease, tumor size, grade, and differentiation are not available in the NHIRD. The stage of the cancer could play a crucial role in lung cancer resection outcomes.
Despite these limitations, we have provided evidence to show that after adjusting for patient, surgeon, and hospital characteristics, an inverse volume-outcome relationship does exist for surgeons, but not for hospitals, in Taiwan. However, we suggest that low volume, as an overall indicator of poor quality, must be used with considerable caution, particularly with regard to policy decision making. We should acknowledge that there are likely to be low-volume surgeons providing excellent pulmonary resection outcomes, who, according to the practice makes perfect and selective referral hypotheses, with advancing age, may ultimately become high-volume surgeons.
Because the causal mechanisms linking volume and outcomes remain unclear, we suggest that further studies should be carried out to identify the differences in the learning processes, perioperative care structures, and surgical techniques between high-volume surgeons with excellent outcomes and low-volume surgeons with poor outcomes. That may well help to reduce the differences between surgeon outcomes and improve the overall quality of care provided to their patients.
|
Acknowledgments |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  H.-C. Lin, S. Xirasagar, C.-H. Chen, and Y.-T. Hwang Physician's Case Volume of Intensive Care Unit Pneumonia Admissions and In-Hospital Mortality Am. J. Respir. Crit. Care Med., May 1, 2008; 177(9): 989 - 994. [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 |
