Temporal Trends in the Management of Potentially Resectable Lung Cancer

Farhood Farjah; Douglas E. Wood; David Yanez; Rebecca G. Symons; Bahirathan Krishnadasan; David R. Flum

doi:10.1016/j.athoracsur.2007.12.081

Temporal Trends in the Management of Potentially Resectable Lung Cancer

Farhood Farjah, Douglas E. Wood, David Yanez, Rebecca G. Symons, Bahirathan Krishnadasan, David R. Flum

Research output: Contribution to journal › Article › peer-review

27 Scopus citations

Abstract

Background: Standardized, evidence-based guidelines recommend lung resection for patients with stage I or II nonsmall-cell lung cancer (NSCLC), and select patients with stage IIIA disease. We hypothesized that the proportion of patients operated on would increase over time coincident with increasing adherence to practice guidelines and improved patient/provider education over time. Methods: This investigation was a cohort study of tumor-registry data linked to Medicare claims. Results: Between 1992 and 2002, 24,030 patients-mean age 75 ± 6 years, 55% men-were diagnosed with NSCLC. In each stage, the proportion of patients undergoing resection was lower in 2002 compared with 1992: stage I (68% versus 80%, p < 0.001), II (59% versus 74%, p < 0.001), and IIIA (23% versus 35%, p < 0.001). The mean age and comorbidity index of the cohort was higher in 2002 compared with 1992 (76 versus 74 years, p < 0.001; and 0.47 and 0.82, p < 0.001, respectively). The unadjusted odds of resection decreased by 6% per year (odds ratio 0.94, 99% confidence interval: 0.93 to 0.95), and adjustment for age, comorbidity index, race, and stage resulted in a slightly smaller (4% per year) but significantly decreasing trend in operative management over time (odds ratio 0.96, 99% confidence interval: 0.95 to 0.97). Conclusions: Unexpectedly, the use of resection for lung cancer has decreased dramatically over time, and this decline is not fully accounted for by an older cohort with more comorbid conditions. Future investigations should determine whether increasing unmeasured contraindications to resection, barriers to accessing specialty care, an inadequate supply of thoracic surgeons, or bias against operative therapy are responsible.

Original language	English (US)
Pages (from-to)	1850-1856
Number of pages	7
Journal	Annals of Thoracic Surgery
Volume	85
Issue number	6
DOIs	https://doi.org/10.1016/j.athoracsur.2007.12.081
State	Published - Jun 2008
Externally published	Yes

ASJC Scopus subject areas

Surgery
Pulmonary and Respiratory Medicine
Cardiology and Cardiovascular Medicine

Access to Document

10.1016/j.athoracsur.2007.12.081

Cite this

@article{061c534f0ac54341b59e657f41531ddf,

title = "Temporal Trends in the Management of Potentially Resectable Lung Cancer",

abstract = "Background: Standardized, evidence-based guidelines recommend lung resection for patients with stage I or II nonsmall-cell lung cancer (NSCLC), and select patients with stage IIIA disease. We hypothesized that the proportion of patients operated on would increase over time coincident with increasing adherence to practice guidelines and improved patient/provider education over time. Methods: This investigation was a cohort study of tumor-registry data linked to Medicare claims. Results: Between 1992 and 2002, 24,030 patients-mean age 75 ± 6 years, 55% men-were diagnosed with NSCLC. In each stage, the proportion of patients undergoing resection was lower in 2002 compared with 1992: stage I (68% versus 80%, p < 0.001), II (59% versus 74%, p < 0.001), and IIIA (23% versus 35%, p < 0.001). The mean age and comorbidity index of the cohort was higher in 2002 compared with 1992 (76 versus 74 years, p < 0.001; and 0.47 and 0.82, p < 0.001, respectively). The unadjusted odds of resection decreased by 6% per year (odds ratio 0.94, 99% confidence interval: 0.93 to 0.95), and adjustment for age, comorbidity index, race, and stage resulted in a slightly smaller (4% per year) but significantly decreasing trend in operative management over time (odds ratio 0.96, 99% confidence interval: 0.95 to 0.97). Conclusions: Unexpectedly, the use of resection for lung cancer has decreased dramatically over time, and this decline is not fully accounted for by an older cohort with more comorbid conditions. Future investigations should determine whether increasing unmeasured contraindications to resection, barriers to accessing specialty care, an inadequate supply of thoracic surgeons, or bias against operative therapy are responsible.",

author = "Farhood Farjah and Wood, {Douglas E.} and David Yanez and Symons, {Rebecca G.} and Bahirathan Krishnadasan and Flum, {David R.}",

note = "Funding Information: Lung resection provides optimal therapy for patients with early-stage NSCLC and select patients with stage IIIA disease. We hypothesized that the dissemination of evidence-based practice guidelines would have resulted in an increase in the use of resection. Contrary to expectation, the odds of resection decreased significantly over time (6% per year). The changing characteristics of the cohort over time (older and sicker patients) accounted for only a fraction of this trend, with an adjusted estimate suggesting a 4% per year decrease in the odds of resection. Several factors may explain the observed decline in operative management. Lung function is a strong determinant of therapy [ 15 ]. Although it is unknown whether the severity of underlying emphysema among lung cancer patients has changed over time, an increase in the prevalence of prohibitively low lung function might explain a decline in the use of resection. Increasing barriers to accessing appropriate care may provide another explanation. Certain providers appear to deliver higher quality surgical care, as evidenced by the associations between provider type and short- and long-term outcomes [ 16–20 ]. Although all patients in our sample had health insurance, their ability to access thoracic surgeons and comprehensive cancer centers was not known and cannot be determined using this dataset. Lung cancer treatment guidelines recommend that appropriate surgical consultation consists of review by “thoracic surgical oncologists who perform lung cancer surgery as a prominent part of their practice” [ 5 ]. It is not known whether the supply of specialists in general thoracic surgery has changed over time to meet the demands of lung cancer patients and nonsurgical providers. Appropriate referrals are another determinant of accessing thoracic surgeons. Survey data suggest that provider type (surgical versus nonsurgical), years since completing training, and case load influence beliefs regarding appropriate treatment and expected survival for lung cancer patients [ 21 ]. Lack of specialty training in thoracic surgery, low volume experience in thoracic surgery, or lack of experience in managing higher risk patients or complex resections may result in patients being declined appropriate surgical resection where they may have been a candidate for surgery in another center or with another surgeon. Differing views of optimal management may impact surgical referrals. Nonsurgical specialists may be reluctant to refer a patient for appropriate surgical consultation for a variety of reasons, including negative perceptions about surgical morbidity or cancer survival, misunderstanding of practice guideline recommendations, remunerative benefits of nonsurgical therapy, or lack of an adequate partnership with a legitimate thoracic surgical oncologist. Finally, our study spanned a period before and after publication of the earliest practice guidelines [ 6, 7 ]. As we found no evidence that temporal trends in operative management were impacted by these and subsequent guidelines, it raises a question about the effectiveness of practice guidelines on physician behavior. It appeared that advancing age and an increasing burden of comorbid conditions accounted for a fraction of the declining trend in operative management over time. Several investigations have shown an acceptable level of perioperative risk after resection for lung cancer among select elderly patients [ 22–28 ], supporting the notion that age alone is not a contraindication to therapy. The number, type, and severity of comorbidities appear to be associated with higher morbidity and lower survival among resected patients [ 29–33 ], and yet it is debatable whether that risk sufficiently outweighs the benefits of resection. Disentangling the effects of age, comorbidity, and other preoperative predictors of outcome is challenging and requires a multidisciplinary team that includes an experienced thoracic surgeon. If providers limit referrals and surgical options on the basis of age and comorbidity alone, then there may be an opportunity to improve the quality of care through further education, advocacy, and encouraging the expansion of multidisciplinary teams that include a experienced thoracic surgical oncologist. These findings must be considered in the context of several important limitations. Our analysis was limited to elderly, Medicare beneficiaries with full fee-for-service coverage and no concurrent health maintenance organization enrollment. If patterns of care were different for younger patients and those under different insurance plans, then our findings may not be generalizable. Theoretically, our exclusion criteria could have biased our results. We concluded that any such bias was unlikely because a sensitivity analysis revealed decreasing rates of resection over time in an cohort without exclusions and among the subgroup who were excluded. Our exclusions were necessary because they allowed for adjustment for comorbid conditions and a description of all modes of therapy, including chemotherapy. Another limitation is that the modified Charlson comorbidity index has not been validated for measuring temporal trends in comorbid conditions. It is possible that better coding over time has resulted in an apparent increase in comorbidities over time. Furthermore, this index does not measure the severity of underlying disease, and therefore we may not have sufficiently accounted for the relationship between comorbidity and the receipt of optimal therapy. It would have been preferable to measure well-established determinants of resection, such as pulmonary function, but these data were not available in SEER-Medicare. Because the stage of disease was based on the highest level of information within 4 months of diagnosis, it is possible that differential accuracy of staging by treatment (operative versus nonoperative) might have biased our results. Some patients with clinical stage I, II, or IIIA were likely upstaged to stage IIIB/IV as a result of intraoperative findings. These patients would have been excluded from our analysis. Such exclusions would bias our results if the rate of intraoperative upstaging to stage IIIB/IV increased over time. Yet, it is more likely that intraoperative upstaging has decreased over time as a result of better pretreatment staging (namely, with positron emission tomography) and patient selection. Finally, 7.3% of our cohort had missing covariate data, possibly biasing our findings; however, the proportion of missing data did not appear to change over time (5.9% in 2002 versus 5.7% in 1992, p = 0.79) nor were differences in the proportion of missing variables between operated and nonoperated patients clinically important (7.0% versus 7.8%, p = 0.02). To the extent to which our results our generalizable, it is important to consider the implications of our findings in absolute rather than relative terms. Current estimates forecast 213,380 new cases of lung cancer in the United States in 2007 [ 1 ]. According to SEER estimates, 80% will have NSCLC and of those, 33% will have stage I, II, or IIIA disease for a total of 56,332 patients with potentially resectable NSCLC. A conservative estimate of the proportion of resected cases in 2007 would be the proportion of resected cases in 2002 (57%). Accordingly, 32,109 patients would be expected to undergo resection in 2007. Had the proportion of resected cases remained constant at levels observed in 1992 (70%), then 39,433 patients would be expected to undergo resection in 2007. The difference between these two estimates suggests that up to 7,323 patients might not receive appropriate therapy in 2007. Assuming that all 7,323 patients would otherwise be eligible for resection, and an operative mortality rate of 5.2% [ 2 ], 6,870 lives that could be saved in 2007 would unnecessarily be lost. Ultimately, change in therapy over time is important if it results in change in outcomes over time. Overall mortality rates for lung cancer have been declining since 1992 [ 1 ], suggesting perhaps that the declining trend in operative therapy is inconsequential. However, these improvements in outcome have been attributed to smoking intervention programs that prevent lung cancer [ 34 ]. Other potential reasons for improved overall mortality rates include improvements in patient and treatment safety, earlier incidental detection of lung cancer associated with the increasing use of computed tomography for other indications, better staging with subsequent stage migration effect, and new institution of adjuvant therapy for resected lung cancer. Supporting this view is the observation that the overall mortality rates for all cancers combined have improved over the same period as lung cancer [ 35 ]. Since lung resection is widely considered primary therapy with curative intent, it will be important to reverse the declining trend in resection to further improve lung cancer outcomes. In conclusion, the optimal care of potentially resectable lung cancer patients appears to be diminishing rather than improving over time. Future studies should aim to identify factors responsible for this trend, and provide information on how to improve access to appropriate thoracic surgical care and adherence to lung cancer practice guidelines. This study used the linked Surveillance, Epidemiology, and End Results (SEER)–Medicare database. The interpretation and reporting of these data are the sole responsibility of the authors. The authors acknowledge the efforts of the Applied Research Program, National Cancer Institute; the Office of Research, Development and Information, CMS; Information Management Services (IMS), Inc.; and the SEER program tumor registries in the creation of the SEER-Medicare database. Farhood Farjah was supported by a Cancer Epidemiology and Biostatistics Training Grant (T32 CA09168-30) and Ruth L. Kirschstein by a National Research Service Award (F32 CA130434-01) from the National Cancer Institute. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute or the National Institutes of Health. The authors are also grateful for additional resources made available through the University of Washington's Department of Surgery and the Surgical Outcomes Research Center, and the generosity of the Schilling Family. ",

year = "2008",

month = jun,

doi = "10.1016/j.athoracsur.2007.12.081",

language = "English (US)",

volume = "85",

pages = "1850--1856",

journal = "Annals of Thoracic Surgery",

issn = "0003-4975",

publisher = "Elsevier USA",

number = "6",

}

TY - JOUR

T1 - Temporal Trends in the Management of Potentially Resectable Lung Cancer

AU - Farjah, Farhood

AU - Wood, Douglas E.

AU - Yanez, David

AU - Symons, Rebecca G.

AU - Krishnadasan, Bahirathan

AU - Flum, David R.

N1 - Funding Information: Lung resection provides optimal therapy for patients with early-stage NSCLC and select patients with stage IIIA disease. We hypothesized that the dissemination of evidence-based practice guidelines would have resulted in an increase in the use of resection. Contrary to expectation, the odds of resection decreased significantly over time (6% per year). The changing characteristics of the cohort over time (older and sicker patients) accounted for only a fraction of this trend, with an adjusted estimate suggesting a 4% per year decrease in the odds of resection. Several factors may explain the observed decline in operative management. Lung function is a strong determinant of therapy [ 15 ]. Although it is unknown whether the severity of underlying emphysema among lung cancer patients has changed over time, an increase in the prevalence of prohibitively low lung function might explain a decline in the use of resection. Increasing barriers to accessing appropriate care may provide another explanation. Certain providers appear to deliver higher quality surgical care, as evidenced by the associations between provider type and short- and long-term outcomes [ 16–20 ]. Although all patients in our sample had health insurance, their ability to access thoracic surgeons and comprehensive cancer centers was not known and cannot be determined using this dataset. Lung cancer treatment guidelines recommend that appropriate surgical consultation consists of review by “thoracic surgical oncologists who perform lung cancer surgery as a prominent part of their practice” [ 5 ]. It is not known whether the supply of specialists in general thoracic surgery has changed over time to meet the demands of lung cancer patients and nonsurgical providers. Appropriate referrals are another determinant of accessing thoracic surgeons. Survey data suggest that provider type (surgical versus nonsurgical), years since completing training, and case load influence beliefs regarding appropriate treatment and expected survival for lung cancer patients [ 21 ]. Lack of specialty training in thoracic surgery, low volume experience in thoracic surgery, or lack of experience in managing higher risk patients or complex resections may result in patients being declined appropriate surgical resection where they may have been a candidate for surgery in another center or with another surgeon. Differing views of optimal management may impact surgical referrals. Nonsurgical specialists may be reluctant to refer a patient for appropriate surgical consultation for a variety of reasons, including negative perceptions about surgical morbidity or cancer survival, misunderstanding of practice guideline recommendations, remunerative benefits of nonsurgical therapy, or lack of an adequate partnership with a legitimate thoracic surgical oncologist. Finally, our study spanned a period before and after publication of the earliest practice guidelines [ 6, 7 ]. As we found no evidence that temporal trends in operative management were impacted by these and subsequent guidelines, it raises a question about the effectiveness of practice guidelines on physician behavior. It appeared that advancing age and an increasing burden of comorbid conditions accounted for a fraction of the declining trend in operative management over time. Several investigations have shown an acceptable level of perioperative risk after resection for lung cancer among select elderly patients [ 22–28 ], supporting the notion that age alone is not a contraindication to therapy. The number, type, and severity of comorbidities appear to be associated with higher morbidity and lower survival among resected patients [ 29–33 ], and yet it is debatable whether that risk sufficiently outweighs the benefits of resection. Disentangling the effects of age, comorbidity, and other preoperative predictors of outcome is challenging and requires a multidisciplinary team that includes an experienced thoracic surgeon. If providers limit referrals and surgical options on the basis of age and comorbidity alone, then there may be an opportunity to improve the quality of care through further education, advocacy, and encouraging the expansion of multidisciplinary teams that include a experienced thoracic surgical oncologist. These findings must be considered in the context of several important limitations. Our analysis was limited to elderly, Medicare beneficiaries with full fee-for-service coverage and no concurrent health maintenance organization enrollment. If patterns of care were different for younger patients and those under different insurance plans, then our findings may not be generalizable. Theoretically, our exclusion criteria could have biased our results. We concluded that any such bias was unlikely because a sensitivity analysis revealed decreasing rates of resection over time in an cohort without exclusions and among the subgroup who were excluded. Our exclusions were necessary because they allowed for adjustment for comorbid conditions and a description of all modes of therapy, including chemotherapy. Another limitation is that the modified Charlson comorbidity index has not been validated for measuring temporal trends in comorbid conditions. It is possible that better coding over time has resulted in an apparent increase in comorbidities over time. Furthermore, this index does not measure the severity of underlying disease, and therefore we may not have sufficiently accounted for the relationship between comorbidity and the receipt of optimal therapy. It would have been preferable to measure well-established determinants of resection, such as pulmonary function, but these data were not available in SEER-Medicare. Because the stage of disease was based on the highest level of information within 4 months of diagnosis, it is possible that differential accuracy of staging by treatment (operative versus nonoperative) might have biased our results. Some patients with clinical stage I, II, or IIIA were likely upstaged to stage IIIB/IV as a result of intraoperative findings. These patients would have been excluded from our analysis. Such exclusions would bias our results if the rate of intraoperative upstaging to stage IIIB/IV increased over time. Yet, it is more likely that intraoperative upstaging has decreased over time as a result of better pretreatment staging (namely, with positron emission tomography) and patient selection. Finally, 7.3% of our cohort had missing covariate data, possibly biasing our findings; however, the proportion of missing data did not appear to change over time (5.9% in 2002 versus 5.7% in 1992, p = 0.79) nor were differences in the proportion of missing variables between operated and nonoperated patients clinically important (7.0% versus 7.8%, p = 0.02). To the extent to which our results our generalizable, it is important to consider the implications of our findings in absolute rather than relative terms. Current estimates forecast 213,380 new cases of lung cancer in the United States in 2007 [ 1 ]. According to SEER estimates, 80% will have NSCLC and of those, 33% will have stage I, II, or IIIA disease for a total of 56,332 patients with potentially resectable NSCLC. A conservative estimate of the proportion of resected cases in 2007 would be the proportion of resected cases in 2002 (57%). Accordingly, 32,109 patients would be expected to undergo resection in 2007. Had the proportion of resected cases remained constant at levels observed in 1992 (70%), then 39,433 patients would be expected to undergo resection in 2007. The difference between these two estimates suggests that up to 7,323 patients might not receive appropriate therapy in 2007. Assuming that all 7,323 patients would otherwise be eligible for resection, and an operative mortality rate of 5.2% [ 2 ], 6,870 lives that could be saved in 2007 would unnecessarily be lost. Ultimately, change in therapy over time is important if it results in change in outcomes over time. Overall mortality rates for lung cancer have been declining since 1992 [ 1 ], suggesting perhaps that the declining trend in operative therapy is inconsequential. However, these improvements in outcome have been attributed to smoking intervention programs that prevent lung cancer [ 34 ]. Other potential reasons for improved overall mortality rates include improvements in patient and treatment safety, earlier incidental detection of lung cancer associated with the increasing use of computed tomography for other indications, better staging with subsequent stage migration effect, and new institution of adjuvant therapy for resected lung cancer. Supporting this view is the observation that the overall mortality rates for all cancers combined have improved over the same period as lung cancer [ 35 ]. Since lung resection is widely considered primary therapy with curative intent, it will be important to reverse the declining trend in resection to further improve lung cancer outcomes. In conclusion, the optimal care of potentially resectable lung cancer patients appears to be diminishing rather than improving over time. Future studies should aim to identify factors responsible for this trend, and provide information on how to improve access to appropriate thoracic surgical care and adherence to lung cancer practice guidelines. This study used the linked Surveillance, Epidemiology, and End Results (SEER)–Medicare database. The interpretation and reporting of these data are the sole responsibility of the authors. The authors acknowledge the efforts of the Applied Research Program, National Cancer Institute; the Office of Research, Development and Information, CMS; Information Management Services (IMS), Inc.; and the SEER program tumor registries in the creation of the SEER-Medicare database. Farhood Farjah was supported by a Cancer Epidemiology and Biostatistics Training Grant (T32 CA09168-30) and Ruth L. Kirschstein by a National Research Service Award (F32 CA130434-01) from the National Cancer Institute. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute or the National Institutes of Health. The authors are also grateful for additional resources made available through the University of Washington's Department of Surgery and the Surgical Outcomes Research Center, and the generosity of the Schilling Family.

PY - 2008/6

Y1 - 2008/6

N2 - Background: Standardized, evidence-based guidelines recommend lung resection for patients with stage I or II nonsmall-cell lung cancer (NSCLC), and select patients with stage IIIA disease. We hypothesized that the proportion of patients operated on would increase over time coincident with increasing adherence to practice guidelines and improved patient/provider education over time. Methods: This investigation was a cohort study of tumor-registry data linked to Medicare claims. Results: Between 1992 and 2002, 24,030 patients-mean age 75 ± 6 years, 55% men-were diagnosed with NSCLC. In each stage, the proportion of patients undergoing resection was lower in 2002 compared with 1992: stage I (68% versus 80%, p < 0.001), II (59% versus 74%, p < 0.001), and IIIA (23% versus 35%, p < 0.001). The mean age and comorbidity index of the cohort was higher in 2002 compared with 1992 (76 versus 74 years, p < 0.001; and 0.47 and 0.82, p < 0.001, respectively). The unadjusted odds of resection decreased by 6% per year (odds ratio 0.94, 99% confidence interval: 0.93 to 0.95), and adjustment for age, comorbidity index, race, and stage resulted in a slightly smaller (4% per year) but significantly decreasing trend in operative management over time (odds ratio 0.96, 99% confidence interval: 0.95 to 0.97). Conclusions: Unexpectedly, the use of resection for lung cancer has decreased dramatically over time, and this decline is not fully accounted for by an older cohort with more comorbid conditions. Future investigations should determine whether increasing unmeasured contraindications to resection, barriers to accessing specialty care, an inadequate supply of thoracic surgeons, or bias against operative therapy are responsible.

AB - Background: Standardized, evidence-based guidelines recommend lung resection for patients with stage I or II nonsmall-cell lung cancer (NSCLC), and select patients with stage IIIA disease. We hypothesized that the proportion of patients operated on would increase over time coincident with increasing adherence to practice guidelines and improved patient/provider education over time. Methods: This investigation was a cohort study of tumor-registry data linked to Medicare claims. Results: Between 1992 and 2002, 24,030 patients-mean age 75 ± 6 years, 55% men-were diagnosed with NSCLC. In each stage, the proportion of patients undergoing resection was lower in 2002 compared with 1992: stage I (68% versus 80%, p < 0.001), II (59% versus 74%, p < 0.001), and IIIA (23% versus 35%, p < 0.001). The mean age and comorbidity index of the cohort was higher in 2002 compared with 1992 (76 versus 74 years, p < 0.001; and 0.47 and 0.82, p < 0.001, respectively). The unadjusted odds of resection decreased by 6% per year (odds ratio 0.94, 99% confidence interval: 0.93 to 0.95), and adjustment for age, comorbidity index, race, and stage resulted in a slightly smaller (4% per year) but significantly decreasing trend in operative management over time (odds ratio 0.96, 99% confidence interval: 0.95 to 0.97). Conclusions: Unexpectedly, the use of resection for lung cancer has decreased dramatically over time, and this decline is not fully accounted for by an older cohort with more comorbid conditions. Future investigations should determine whether increasing unmeasured contraindications to resection, barriers to accessing specialty care, an inadequate supply of thoracic surgeons, or bias against operative therapy are responsible.

UR - http://www.scopus.com/inward/record.url?scp=43649084819&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=43649084819&partnerID=8YFLogxK

U2 - 10.1016/j.athoracsur.2007.12.081

DO - 10.1016/j.athoracsur.2007.12.081

M3 - Article

C2 - 18498783

AN - SCOPUS:43649084819

SN - 0003-4975

VL - 85

SP - 1850

EP - 1856

JO - Annals of Thoracic Surgery

JF - Annals of Thoracic Surgery

IS - 6

ER -

Temporal Trends in the Management of Potentially Resectable Lung Cancer

Abstract

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this