Patterns of Surgical Care of Lung Cancer Patients

Transcription

1 Patterns of Surgical Care of Lung Cancer Patients Alex G. Little, MD, Valerie W. Rusch, MD, James A. Bonner, MD, Laurie E. Gaspar, MD, Mark R. Green, MD, W. Richard Webb, MD, and Andrew K. Stewart, MA Department of Surgery, Wright State University School of Medicine, Dayton, Ohio; Memorial Sloan-Kettering Center, New York, New York; Radiation Oncology, University of Alabama, Birmingham, Alabama; Anschutz Cancer Pavilion, University of Colorado Health Science Center, Aurora, Colorado; Hollings Cancer Center, Medical University of South Carolina, Charleston, South Carolina; Department of Radiology, University of California, San Francisco, California; and American College of Surgeons and the National Cancer Data Base, Chicago, Illinois GENERAL THORACIC Background. This survey was performed to determine the patterns of surgical care provided patients with non-small cell lung carcinoma (NSCLC). Methods. In 2001, the American College of Surgeons carried out a patient care survey of 729 hospitals to retrieve information of NSCLC patients history, evaluation, pathology, and surgical treatment. Results. Inclusion criteria were met by 40,090 patients: of whom 11,668 (29.1%) were treated surgically; 74.2% alone and 25.8% as part of multimodality therapy. Of these patients, 59.5% were in stage I, 17.5% in stage II, 17.0% in stage III, and 6.0% in stage IV. Surgery patient demographics were the following: 55% male and 45% female; 46.8% 70 years or older; and 76.3% had significant comorbidities. Tumor characteristics: squamous 28%, adenocarcinoma 37.6%, other 34.4%. Staging: in addition to radiologic examinations, preoperative mediastinoscopy was performed in 27.1% of operated patients with node biopsy in only 46.6% of these procedures. Operations: wedge resection 15.6%, lobectomy 70.8%, pneumonectomy 13.6%. Surgical margins were positive in 7.8%, but only 65.2% had frozen section analysis. Perioperative mortality was 5.2%, but was 4.0% in nontransfused patients and 12.7% in transfused patients and was 3.2% in high-volume (more than 90 operations per year) versus 4.8% in low-volume hospitals (p < 0.001). Conclusions. (1) Patients being operated for NSCLC are elderly with significant comorbid conditions. (2) More patients than previously are female and have adenocarcinoma. (3) Mediastinoscopy is infrequently performed and lymph nodes are biopsied in less than 50% of them. (4) Lobectomy is the most common operation, and positive surgical margins are too frequent. (5) Operative mortality is reasonable but transfusion is a marker for increased risk and outcomes are superior in high-volume hospitals. (6) Hospitals with higher volume had fewer perioperative deaths. (Ann Thorac Surg 2005;80:2051 6) 2005 by The Society of Thoracic Surgeons The Commission on Cancer of the American College of Surgeons (CCACS) for many years has conducted national surveys of hospitals to evaluate the patterns of care of patients with specific cancer diagnoses through patient care evaluations (PCE). The most recent previous PCE for non-small cell lung cancer (NSCLC) patients was in 1990 [1]. This paper reports a CCACS PCE for patients diagnosed with NSCLC carried out in the year The goal was to determine the demographic characteristics of patients receiving surgical care for NSCLC, to retrieve information about the type and extent of the cancer in these patients, to determine the type and frequency of utilization of various staging modalities, and to determine patterns and perioperative outcomes of surgical therapy. Lung cancer accounts for 13% of all new cancers, Accepted for publication June 27, Presented at the Forty-first Annual Meeting of The Society of Thoracic Surgeons, Tampa, FL, Jan 24 26, Address correspondence to Dr E. Greer Gay, The American College of Surgeons, Commission on Cancer, 633 N. Saint Clair Street, Chicago, IL 60611; ggay@facs.org. approximately 87% of which are NSCLC. It is the leading cause of cancer death among men and women in the United States. This disease is usually not diagnosed early, hence the five-year survival rate for all stages combined is only 15% [2]. Treatment depends on the extent of disease, the location of the primary tumor, and the presence or absence of medical comorbidities. Appraisal of lymphatic and distant metastases is critical for staging and patient evaluation [3]. Material and Methods A total of 729 hospitals participated in the PCE study of NSCLC patients, diagnosed between January 1, 2001 and December 31, All geographic regions were included. As fully defined elsewhere [2], there were four cancer program categories: community cancer centers, comprehensive community cancer centers, teachingresearch facilities, and other. A total of 40,909 cases met the inclusion criteria of microscopically confirmed primary NSCLC, classified topographically as C34.0, C34.1, C34.2, C34.3, C34.4, C34.8, and C34.9 with histologies coded in the range of 8012 to 8576 and a behavior code of 2005 by The Society of Thoracic Surgeons /05/$30.00 Published by Elsevier Inc doi: /j.athoracsur

2 GENERAL THORACIC 2052 LITTLE ET AL Ann Thorac Surg PATTERNS OF SURGICAL CARE IN NSCLC 2005;80: , and staged according to the 5th edition of the American Joint Committee on Cancer (AJCC) [3]. In addition to the data routinely submitted to the National Cancer Data Base (NCDB), described in the Commission on Cancer s Registry Operations and Data Standards [4, 5], this PCE collected more detailed information on patient histories, tumor identification, tumor evaluation, pathology, staging, and the first course of treatment. The NCDB is a hospital cancer registry-based, national clinical surveillance and outcomes database supported by the American College of Surgeons and the American Cancer Society. The database captures 70% to 75% of all newly diagnosed cancer cases in the United States annually. The data are submitted for the purpose of quality improvement efforts of participating hospitals. To ensure standardized data, coding schema followed those published in the fifth edition of the Manual for Staging Cancer, the Registry Operations and Data Standards (ROADS), and the third edition of the International Classification of Diseases for Oncology [3, 5, 6]. In addition to the use of a standard clinical and pathologic stage, a combined AJCC stage was used to report pathologic stage when available, supplemented by clinical stage when pathologic stage was not available or was unknown. Data quality checks were conducted at the local as well as the depository level, using GenEdits, a public domain program developed by the Centers for Disease Control, modified by the North American Association of Central Cancer Registries (NAACCR) in conjunction with the Commission on Cancer, to facilitate data quality checks on large blocks of data [7]. Reports generated by computer edit checks performed by the NCDB were returned to participating hospitals for review, correction, and resubmission when data errors or data conflicts were found. After the case abstractor s review of all the special study data items requested for the study, but not ordinarily submitted in the NCDB, the chairman or a member of the institutional cancer committee reviewed the quality of the data of all cases entered into the study. No patient identifiers were collected as part of the study. Patient demographic information was obtained from the hospital admission form. Case identification information (facility identification number and accession number) were collected for administrative purposes only and in accordance with the North American Association of Central Cancer Registries Data Submission Standards [8]. Neither patients nor physicians were identified. Treatment modality reflected the initial course of cancer-directed therapy and included combinations of surgery, chemotherapy, and radiotherapy. The various surgical procedures were classified according to the standardized coding schema defined in ROADS. A volume measure was developed to identify the impact of hospital volume, if any, on patient care. All thoracic surgery cases identified in the 1999, 2000, and 2001 NCDB Calls for Data were used to establish an average annual operative caseload for each facility. All data were analyzed using a standardized statistical software package, SPSS software version 11.5 (SPSS, Table 1. NSCLC Patients, Surgical Cohort Patient Age (Years) Number of Patients % of Total Chicago, IL). Missing data and unknown values were excluded from the analyses to assure appropriate denominators and this is so noted in the tables. The Pearson s 2 test, Fisher s exact one-sided test, and confidence intervals were calculated to test significance. Results , , , , NSCLC non-small cell lung cancer. The survey identified a total of 40,090 patients who met inclusion criteria. The analysis of the entire PCE population is currently under review [9]. This manuscript addresses the 11,668 patients (29.1% of the total) whose initial management included surgical therapy. There were 279 hospitals classified as community cancer centers with 2,609 (22.4%) patients; 266 were comprehensive community cancer centers with 5,359 (46.0%) patients and 141 were teaching/research hospitals with 3,649 (31.3%) patients. Six patients did not have a category of cancer program listed; 45 received surgery at a mixture of 10 free-standing, integrated cancer, or hospital associate programs. Patient Characteristics Men outnumbered women as 55% of patients were male and 45% were female. Only 10.4% of patients had never smoked. The average age at diagnosis was a little older than 67 years, and 46.8% of patients were 70 or older. The age distribution is shown in Table 1. In terms of race/ ethnicity, though 84.1% of the patients in the larger study were white, 10.1% black, and 4.7% non-white, within this surgical cohort 87.8% were classified as white, 7.6% as black, and 4.7% non-white. Many patients had symptoms at the time of their initial presentation. Cough was reported by 53.8% of patients, shortness of breath by 45.1%, and weight loss by 23.9%. Relevant comorbid conditions were present in 76.3% of patients. Of those with comorbidities, 25.0% had only one while 75.0% had two or more. The most common comorbid conditions were chronic obstructive pulmonary disease, high blood pressure, atrial fibrillation, and coronary artery disease. In addition, 3,016 or 25.8% of patients had a prior history of another invasive malignancy. Of these, the most common were head and neck cancers in 18.7%, male genital cancers in 15.3%, breast cancer in 12.3%, and lung cancer in 12.1%.

3 Ann Thorac Surg LITTLE ET AL 2005;80: PATTERNS OF SURGICAL CARE IN NSCLC Table 2. Noninvasive Staging Examinations Chest CT Scan Performed a 92.3% Mediastinal adenopathy b 26.9% Metastases b 14.6% PET scan Performed a 26.4% Mediastinal positivity b 27.7% Metastases b 15.5% Bone Scan Performed a 33.3% Metastases b 6.7% Brain CT Scan Performed a 22.5% Metastases b 7.1% Brain MRI Performed a 10.4% Metastases b 13.9% a Percent of all surgical patients. b Percent of exam patients with this radiologic abnormality. CT computed tomographic; MRI magnetic resonance imaging; PET positron emission tomography. Tumor Characteristics Squamous cell carcinoma histology accounted for 28% of this cohort (33.5% male, 21.3% female), adenocarcinoma 37.6% (33.8% male, 42.1% female), and 34.4% other subhistologies not otherwise specified. The primary tumor was located in the upper lobe in 59.3% of patients, in the lower lobe in 29.3%, in the middle lobe in 4.6%, and was described as overlapping in 2.5%. Only 1.3% of patients had tumor in a main stem bronchus. Preoperative Staging Examinations Table 2 displays the frequency with which various radiologic staging examinations were performed. The most commonly utilized modality was computerized tomographic (CT) scan of the chest (92.3%). Positron emission tomography (PET) scans were obtained in only 26.4% of patients. The mediastinum was evaluated preoperatively with mediastinoscopy in only 27.1% of these surgical patients. Of those undergoing this procedure, only 46.6% of patients had documented evidence of lymph node biopsy material submitted to pathology. That practice significantly differed depending on the type of cancer program in which the mediastinoscopy was performed. Lymph node biopsy was performed more frequently in teaching/ research facilities, 51.9% (confidence interval [CI] 49.0% to 54.8%), than in either community cancer centers, 40.2% (CI 36.4% to 44.0%), or comprehensive community cancer centers, 45.1% (CI 42.3% to 47.8%) (p 0.05). Final Staging Table 3 provides the final stage; ie, the stage at time of hospital discharge of these patients. In stage I, 59.5% patients comprised the majority of the surgical cohort, followed by 17.5% for stage II and 17.0% for stage III. Six percent of the patients in this surgical cohort were in stage IV. Therapy Table 4 displays the initial treatment utilized in these patients both overall and by stage. The majority, 74.2%, received only surgical treatment. The remaining 25.8% received some form of multimodality therapy with surgery plus either radiation therapy (9.8%), chemotherapy (5.7%), or both (10.4%). Patients may have received subsequent treatment after their initial treatment but that information is unavailable. Table 5 displays the frequency with which specific operations were performed both according to the stage of disease and overall. Lobectomy was the most common operation, being performed in 70.8% of patients, with the second most common operation being a wedge resection in 15.6% of patients. Only 13.6% of patients were treated with pneumonectomy. During operation, only 57.8% of patients had lymph nodes either sampled or removed from the mediastinum. The remaining 42.2% of patients only had lymph nodes submitted that were attached to the surgical specimen with no separately submitted lymph nodes from any mediastinal station. On analysis across categories of cancer programs, mediastinal lymph nodes were sampled more frequently in teaching-research facilities (67.9%, CI 65.8% to 69.9%) than in either comprehensive community cancer centers (55.6%, CI 53.9% to 57.4%) or community cancer centers (48.1%, CI 45.5% to 50.6%) (p 0.01). At least one surgical margin, not including the visceral pleural margin, of the surgical specimen was positive for carcinoma in 7.8% of patients. However, only 65.2% of patients had frozen section analysis of surgical margins at time of operation. Frozen section analysis was significantly more likely to be performed in teaching-research facilities (66.8%, CI 65.1% to 68.5%) and comprehensive community cancer centers (65.7%, CI 64.3% to 67.2%) than community cancer centers (61.8%, CI 59.7% to 63.9%) (p 0.05). Outcomes Perioperative mortality, defined as death within 30 days of surgery, was 5.2% for all patients for whom death was Table 3. Stage at Time of Hospital Discharge 2053 Stage Number of Patients % Total I % IA % IB % II % IIA % IIB % III % IIIA % IIIB % IV % Note: For 142 patients (1.2%) stage is unknown; 10 patients (0.1) were staged occult. GENERAL THORACIC

4 GENERAL THORACIC 2054 LITTLE ET AL Ann Thorac Surg PATTERNS OF SURGICAL CARE IN NSCLC 2005;80: Table 4. Initial Treatment Treatment Stage I II III IV Total Surgery alone 6,280 (91.8%) 1,228 (61.1%) 743 (38.1%) 257 (38.1%) 8,508 (74.2%) Surgery RT 273 (4.0%) 346 (17.2%) 353 (18.1%) 149 (22.1%) 1,121 (9.8%) Surgery chemo 157 (2.3%) 153 (7.6%) 214 (11.0%) 133 (19.7%) 657 (5.7%) Surgery chemo RT 132 (1.9%) 282 (14.0%) 639 (32.6%) 135 (20.0%) 1,188 (10.4%) % % of patients in the stage. Note: 142 patients (1.2%) stage, where stage is unknown; 10 patients, staged occult and 42 patients received surgery plus immuno not included in analysis. Pearson s Chemo chemotherapy; RT radiation therapy. ascertained. By surgical procedure, perioperative mortality for lobectomy was 4.5%, for pneumonectomy 8.5%, and for wedge resection 4.9%. Category of hospital did not have a significant effect as mortality rates were similar for the three hospital types: community cancer centers 5.3%, comprehensive community cancer centers 5.3%, and teaching-research institutions 5.1% (p 0.895). Hospital volume was associated with outcomes. There was a significant difference in mortality rates when hospitals with 90 or fewer operations per year were compared with those with more than 90 per year: 4.8% vs 3.2% (p 0.037). Mortality was associated with the utilization of perioperative blood transfusion. For those patients for whom transfusion information was documented (n 8,669) perioperative mortality was 4.0% in nontransfused patients, but rose to 12.7% in the 1,185 (13.7%) patients who received a perioperative transfusion. Comment The CCACS PCE process that sponsored this study was not hypothesis driven. The PCE is designed to determine simply what is being done in hospitals. There is no access to physician information and these data represent only a single year in time and thus do not permit long-term follow-up analysis. However, as with previous PCEs, there is utility to identifying patterns of patients, their disease, and hospital care. First is simply tracking this information to identify and allow reaction to changes in these variables over time. Second, some comparisons and observations of our study demonstrate several clinical issues clearly, and others, which need further, hypothesis driven, studies based on surveys of both hospitals and physicians. Not surprisingly, most patients were smokers; the prevalence of 89.6% of our patients that were smokers is consistent with others reporting the finding of 85% to 90% of all lung cancer patients having a smoking history [10]. This report shows that patients being operated on for NSCLC are complicated. First, they are elderly, as also demonstrated in the overall sample of 40,090 identified with NSCLC in which 50.7% were age 70 or older [9]. In the patients operated on, 46.8% were age 70 or older. In addition, most patients had significant comorbidities and many had been treated for previous cancers. It appears the thoracic surgery community is increasingly going to be challenged to deal with and treat elderly patients with significant disease. Also true is the observation that increasingly more patients are female and have adenocarcinoma. As recently reported, lung cancer in females behaves intrinsically differently biologically and its natural history also is different. What impact this may have on treatment is as yet undetermined [10]. Why the surgical cohort proportionally included fewer blacks, 7.6%, than found in the larger study from which this cohort was drawn, where blacks made up 10.1% of the population, cannot be explained. The seeming inequity hopefully will prompt others to probe this matter further to determine if it reflects access to surgeons or is an issue of patient or physician selection of therapy. Table 5. Frequency of Operations Operation Performed Stage I II III IV Operation: Total (%) Lobectomy 5,104 (75.3%) 1,416 (71.2%) 1,100 (61.6%) 249 (44.6%) 7,869 (70.8%) Wedge resection 1,156 (17.1%) 111 (5.6%) 258 (14.4%) 212 (38.0%) 1,737 (15.6%) Pneumonectomy 520 (7.7%) 461 (23.2%) 429 (24.0%) 97 (17.4%) 1,507 (13.6%) Note: % % of patients in the stage. 142 patients (1.2%) stage, where stage is unknown and 10 patients staged occult not included in analysis. Also not included were 403 patients whose surgery was less specifically coded (Pearson s % of stage IV patients had at least two metastatic sites.

5 Ann Thorac Surg LITTLE ET AL 2005;80: PATTERNS OF SURGICAL CARE IN NSCLC This survey shows that the chest computerized axial tomography scan is the staple examination for staging. The PET scans were only performed in 26.4% of patients but the year 2001 was fairly early in their utilization and this may not reflect current utilization. The relative infrequency with which mediastinal lymph nodes are sampled, either before, by mediastinoscopy, or during the surgical procedure is concerning. Although an agreed upon gold standard for the need for lymph node sampling and/or removal is not established, the benefits to the patient of accurate staging to define prognosis and determine the appropriate therapy is unquestioned [11]. Therefore, it is distressing that only 27.1% of patients had preoperative mediastinoscopy and of them only 46.6% actually had lymph node pathology findings. Additionally troublesome is that only 42.2% had nodes sampled at any mediastinal level during surgery. Lymph node biopsy was performed more frequently preoperatively as well as intraoperatively in teaching/research facilities, perhaps because mediastinoscopy is more likely to be performed by thoracic surgeons in these hospitals, although without physician information from our hospital survey this is speculative. These findings suggest that there is a need for continuing education of all surgeons who perform lung resection for NSCLC and perhaps an opportunity for the Society of Thoracic Surgeons (STS) to develop formal guidelines. Also troubling is the finding that 7.8% of patients left the operating room with what were eventually shown to be surgical margins positive for carcinoma. This compares unfavorably to reported rates of positivity ranging from 1.6% to 3.4% [12 15]. This observation is coupled with the finding that only 65.2% of patients had frozen section analysis of the surgical margin before the patient left the operating room. With the capability of performing bronchial sleeve resections and/or bronchoplastic operations, this seems a higher rate of positive surgical margins than ideal and suggests the benefit of more frequent use of frozen section analysis of margins to decrease this incidence. This is another area in which there is room for continuing education, by the STS, of surgeons performing these operations. The perioperative mortality rates reported from this operative experience seem reasonable. They exceed, but are comparable with, previously reported mortality rates of 2.9% for lobectomy and 6.2%, for pneumonectomy [16]. It is interesting that the mortality of 4.9% for wedge resection actually exceeds the perioperative mortality rate of 4.5% for lobectomy. Although we cannot identify from this survey the rationale for the surgeon s selection of operative procedure, this finding suggests that less physiologically fit patients are undergoing this operation, rather than that wedge resection is a more dangerous technical undertaking than lobectomy. Perioperative mortality was higher in transfused patients than those who did not require transfusion. While there are issues related to transfusion and immunosuppression in terms of long-term cancer-free survival [17], here the need for transfusion probably is a marker for either a difficult operation or one in which there was excessive blood loss. Lastly, as others have reported, outcomes seem related to hospital volume [18]. In hospitals that performed 90 operations or more per year, or approximately two lung resections a week, perioperative mortality was significantly less likely to occur. Finally, the findings of this PCE underscore the lack of consensus as to the optimum treatment of patients with advanced local disease; ie, those in stage II, stage III, or stage IV. There is inconsistency as to whether patients are entered into a multimodality therapeutic protocol and, if so, whether it consists of surgery plus radiation therapy or chemotherapy or both. Potentially alarming is that 16.9% of the stage IV cohort, for whom either a wedge, lobectomy, or pneumonectomy were performed, had at least two metastatic sites, thus essentially would not have been treated according to the NCCN guidelines that define when surgery of the primary site for this patient may be recommended [19]. Hopefully, the outcomes of ongoing clinical trials will bring consistency into the treatment of these patients. The authors express their gratitude to E. Greer Gay, PhD, of the American College of Surgeons and the National Cancer Data Base who carried out the data analyses. References Humphrey EW, Smart CR, Winchester DP, et al. National survey of the pattern of care for carcinoma of the lung. J Thoracic Cardiovasc Surg 1990;100: American Cancer Society. Cancer Reference Information (2004) X What are the key statistics for lung cancer. 3. American Joint Committee on Cancer. Cancer Staging Manual, 5th edition. Philadelphia; Lippincott-Raven: Commission on Cancer. Cancer Program Standards. Chicago, IL: American College of Surgeons; 2004: Registry operations and data standards (ROADS), 1996, 1998 supplement to ROADS, vol. II; ROADS vol. II revised 1/1/98; ROADS vol. III edits, CoC, Chicago: American College of Surgeons Commission on Cancer. 6. International classification of diseases for oncology. 3rd edition. AJCC Cancer Staging Manual, 5th edition. New York, NY: Springer-Verlag; Williamson TJ, McKelvey LW. Computer principles. In: Hutchison CL, Menck HR, Burch M, Gottschalk R, eds. Cancer registrar management principles and practice. 2nd ed. Dubuque: Kendall/Hunt; 2004: Data standards and data dictionary. 8th ed. North American Association of Central Cancer Registries (NAACCR): Little AG, Bonner JA, Gaspar LE, et al. National survey of the pattern of patient care for non-small cell carcinoma of the lung. Lung Cancer (submitted for publication). 10. Patel JD. Lung cancer in women. J Clin Oncol 2004;23: Graham ANJ, Chan KJM, Pastorino V, Goldstraw P. Systematic nodal dissection in the intrathoracic staging of patients with non-small cell lung cancer. J Thorac Cardiovasc Surg 1999;117: Kaiser LR, Fleshner P, Keller S, Martini N. Significance of extramucosal residual tumor at the bronchial resection margin. Ann Thorac Surg 1989;47: Liewald F, Hatz RA, Dienemann H, Sunder-Plassmann. Importance of microscopic residual disease at the bronchial margin after resection for non-small cell carcinoma of the lung. J Thorac Cardiovasc Surg 1992;104: GENERAL THORACIC

6 GENERAL THORACIC 2056 LITTLE ET AL Ann Thorac Surg PATTERNS OF SURGICAL CARE IN NSCLC 2005;80: Ghiribelli C, Voltolini L, Paladini, et al. Treatment and survival after lung resection for non-small cell lung cancer in patients with microscopic residual disease at the bronchial stump. Eur J Cardiothorac Surg 1999;16: Snijder RJ, de la Riviere AB, Elbers HJJ, van den Bosch. Survival in resected stage I lung cancer with residual tumor at the bronchial resection margin. Ann Thorac Surg 1998;65: Ginsberg RJ, Hill LD, Eagan RT, et al. Modern thirty-day operative mortality for surgical resections in lung cancer. J Thorac Cardiovasc Surg 1983;86: Little AG, Wu HS, Ferguson MK, et al. Perioperative blood transfusion adversely affects prognosis of patients with stage I non-small cell lung cancer. Am J Surg 1990;160: Goodney PP, Lucas FL, Stukel TA, Birkmeyer JD. Surgeon specialty and operative mortality with lung resection. Ann Surg 2005;241: National Comprehensive Cancer Network (NCCN), NSCLC, NCCN Network, Inc, NSCLC-12. Available at: org/docroot/cri/content/cri_2_4_1x_what_are_the_key_ statistics_for_lung_cancer_26.asp (accessed May 20, 2005). DISCUSSION DR CAROLYN E. REED (Charleston, SC): Dr Pairolero, Dr Murray, members and guests. Dr Little has nicely presented the results of a survey performed by the American College of Surgeons Commission on Cancer to determine pattern of care for patients undergoing surgical resection of nonsmall cell lung cancer. Although there are the usual concerns about an abstracted database and the survey is a snapshot in time, some of the astounding results cannot be ignored. The results are an indictment of the present care of patients with nonsmall cell lung cancer, and it is very unfortunate that the deficiencies in the database do not allow us to be more precise in directing our corrections. Surprisingly, operative mortality is fairly appropriate, and I agree that transfusion is probably a marker of increased complexity. The 1990 paper did not report comorbidities to allow for comparison. Serious deficiencies of the data include the inability to identify community versus university hospitals, cardiothoracic versus general surgeons. 34.4% of the tumors are classified as other. We do not know what clinical stages underwent mediastinoscopy or why. Because of the lack of mediastinal staging, it is possible that the number of patients undergoing operation who were stage III or IV was much higher than 22%. It is unclear why patients with metastases identified on bone scan or MRI/CT of the brain underwent operation. There is no indication that induction therapy was used in stage III disease, and this is somewhat surprising for However, even with the flaws, this survey renders some astounding findings. In two-thirds of the 27.6% of patients undergoing mediastinoscopy no lymph node tissue was submitted; 42% of the patients had no mediastinal staging at the time of operation; about half of the patients had no frozen section of the bronchial margin, which undoubtedly accounts for the high rate of positive margins; a conservative estimate of one quarter of the patients had locally advanced disease at the time of surgery. This survey leads me to conclude that we are not giving excellent care to our lung cancer patients. What are the possible causes? Unfortunately, we do not know if patterns of care differ in community versus university hospitals, in the hands of general surgeons versus general thoracic surgeons versus cardiac surgeons, in high volume versus low volume hospitals, in rural versus urban settings, et cetera, et cetera. We do not know how staging was directed in the individual clinical stages. Without this information it is difficult to precisely mount educational efforts to improve care, but it is clear that we need to do better. Dr Little, I have a few questions for you. Why were patients with bone and brain metasases going to surgery? Did supporting confirmatory tests support false positivity? I calculate that as many as one quarter of patients could have had a nontherapeutic thoracotomy. Why do you think only 7.6% of the patients were African-American and 1.9% Hispanic when 2000 census data indicates 12% of the population are black and about 12% are of Hispanic origin? Did a surgeon actually check the data? Would you comment on the robustness of the database and its weaknesses? Is there any possibility that we could know more about the surveyed hospitals and providers? And finally, what are your thoughts about why the reported pattern of care in 2001 is so poor? Dr Little, you are to be commended for outlining the dismal care of lung cancer patients. It will be up to the STS community to address the findings of this survey and institute corrective action. Thank you for the floor. DR LITTLE: Thank you, Dr Reed. Dr Reed clearly knows she asked some questions I can t answer but I think asked them particularly to make sure we all continue to think about the implications of the questions ourselves. Why did stage IV patients or patients with metastatic disease have operation? We will never be able to answer that, but we can observe none of them should, and whether it is through the Society of Thoracic Surgeons or other bodies, these messages of what constitutes appropriate care need to be transmitted to all surgeons doing these operations. Why clear racial unevenness, if you will, I hesitate to say inequality? I don t know. We are carrying out further analyses of some of the data, looking at things such as payor status, for example, and maybe we will be able to answer some of that, and certainly if we can, I think that will be useful. The database is credible because of the number, variety and wide geographic distribution of the surveyed hospitals. There are perhaps two weaknesses. One is the inability to get physician data. Our data comes from the hospital. Physicians aren t required to identify their specialty when they do an operation, so we can t determine what the specialty was of surgeons. The other is that the way this study was carried out, it didn t explore a hypothesis. It simply wanted to determine what was being carried out. So any question that starts with why is going to be hard to answer. And that leads me to your last question, why were, overall, the data so disappointing in terms of either staging or patient care? I can only speculate here, but I am happy to do so. I think it has to do with surgeons who are not sitting in this room thinking to themselves, debating the issue raised by Dr Allen of whether they should dissect or biopsy mediastinal lymph nodes. There are surgeons who go and take out a lobe and leave the operating room thinking they have done the appropriate job but haven t. I think it is our responsibility as individuals and as members of this Society to deliver the message to all surgeons doing these operations.