Journal of Thoracic Oncology:
Redistribution of Health Care Costs after the Adoption of Positron Emission Tomography among Medicare Beneficiaries with Non–Small-Cell Lung Cancer, 1998–2005
Dinan, Michaela A. PhD*†; Curtis, Lesley H. PhD*‡; Carpenter, William R. PhD†; Biddle, Andrea K. PhD†; Abernethy, Amy P. MD*‡§; Patz, Edward F. Jr MD§‖; Schulman, Kevin A. MD*‡; Weinberger, Morris PhD†
*Duke Clinical Research Institute, Duke University School of Medicine, Durham, North Carolina; †Department of Health Policy and Administration, Gillings School of Global Public Health, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina; ‡Department of Medicine, Duke University School of Medicine, Durham, North Carolina; §Duke Cancer Institute, Duke University School of Medicine, Durham, North Carolina; and ‖Department of Radiology, Duke University School of Medicine, Durham, North Carolina.
Disclosure: Dr. Weinberger was supported by a Veterans Affairs career development award. All other authors declare no conflict of interest.
Address for correspondence: Michaela A. Dinan, PhD, Duke Clinical Research Institute, PO Box 17969, Durham, NC 27715. E-mail: firstname.lastname@example.org
Introduction: Treatment patterns and cost implications of increased positron emission tomography imaging use since Medicare approval in 1998 are not well understood. We examined rates of surgery, radiotherapy, and chemotherapy and inpatient and total health care costs between 1998 and 2005 among Medicare beneficiaries with non–small-cell lung cancer.
Methods: Patients in this retrospective cohort study were 51,374 Medicare beneficiaries diagnosed with non–small-cell lung cancer between 1996 and 2005. The main outcome measures were receipt of surgical resection, radiotherapy, and chemotherapy and inpatient and total health care costs within 1 year of diagnosis.
Results: Between 1996–1997 and 2004–2005, the proportion of patients undergoing surgical resection decreased from 29% to 25%, the proportion receiving radiation therapy decreased from 49% to 43%, and inpatient costs decreased from $28,900 to $26,900. The proportion of patients receiving chemotherapy increased from 25% to 40% and total costs increased from $47,300 to $52,200 (p < 0.001 for all comparisons). Changes in use and costs remained after adjustment for shifting demographic characteristics during the study period.
Conclusions: Adoption of positron emission tomography between 1998 and 2005 was accompanied by decreases in rates of surgery and radiotherapy and in short-term inpatient costs among Medicare beneficiaries with non–small-cell lung cancer, although there was an increase in chemotherapy and overall costs.
Positron emission tomography (PET) is an advanced imaging modality used in the clinical diagnosis, staging, and follow-up of non–small-cell lung cancer (NSCLC). Randomized controlled trials of PET use in initial staging of NSCLC have suggested that PET can detect occult metastatic disease and avoid futile resection.1–3 Epidemiologic studies suggest that PET is associated with upstaging of NSCLC in clinical practice.4,5 The presumption that appropriate detection of occult metastatic disease should reduce futile rates of localized tumor control and associated health care costs has been examined in a single randomized controlled trial6 and remains an open question at the population level. Whether increased PET use among Medicare beneficiaries with NSCLC has affected rates of localized control or its associated costs remains an important policy question.
In this study, we examined rates of local surgical resection, radiotherapy, and chemotherapy and inpatient and total health care costs among Medicare beneficiaries with NSCLC. Specifically, we tested the hypotheses that, between 1998 and 2005, the Medicare population with NSCLC experienced reductions in the use of surgical resection and radiotherapy and in short-term inpatient surgery costs and overall health care costs.
PATIENTS AND METHODS
Data are from the Surveillance Epidemiology and End Results (SEER)-Medicare linked data. SEER-Medicare is a collaborative effort between the National Cancer Institute and the Centers for Medicare & Medicaid Services that links routinely collected population-based data from cancer registries across the United States to Medicare administrative data and health care claims.7 The SEER data include demographic and incident cancer characteristics, including grade and stage, for approximately 25% of the U.S. population with cancer. Medicare provides health insurance for 97% of people 65 years and older in the United States, and the data reflect health care services used and comorbid conditions. SEER-Medicare data have been used previously to examine factors that affect the quality of care for patients with cancer, including sociodemographic characteristics, physician and hospital characteristics, surgery, chemotherapy, radiation, comorbid conditions, complications, screening, relapse, and costs.8–18 The Office of Human Research Ethics at the University of North Carolina at Chapel Hill approved this study.
We analyzed SEER-Medicare data from the 12 SEER registries that were active from 1996 onward. We included all patients from these registries who had a diagnosis of cancer of the lung and bronchus with microscopically confirmed NSCLC histology between 1996 and 2005, were 66 years or older at the time of the diagnosis, and had Medicare Part A and Part B coverage without participating in a health maintenance organization or in Medicare Part C during the year before and after diagnosis or until death. We excluded patients for whom a diagnosis was determined at autopsy or death or who had another diagnosis of malignancy in the year before the NSCLC diagnosis. To help ensure full acquisition of cancer-related claims, we also required that patients have a primary diagnosis of lung cancer (International Classification of Diseases, Ninth Revision, Clinical Modification 162.2–162.9 or 231.2) on an inpatient, outpatient, or carrier-based Medicare claim within 2 months of the SEER diagnosis.
The primary study outcomes were surgical resection, receipt of radiotherapy, and inpatient health care costs within 1 year of the SEER diagnosis. Additional treatment and cost variables included receipt of chemotherapy, use of PET, and overall health care costs. We obtained treatment and health care costs from claims data from 2 months before through 12 months after the month of the NSCLC diagnosis in the SEER data. We identified receipt of each treatment modality using previously defined sets of Healthcare Common Procedure Coding System codes and International Classification of Diseases, Ninth Revision, Clinical Modification codes (Supplemental Table, Supplementary Digital Content 1, http://links.lww.com/JTO/AXX).19,20 We identified surgical procedures using claims for lung resection from inpatient, outpatient, or carrier claims files.19 We identified receipt of chemotherapy using claims from outpatient, carrier, and durable medical equipment claims files.20 We used outpatient and carrier claims data to identify receipt of radiation therapy.20 We also used outpatient and carrier claims from 2 months before through 4 months after the SEER diagnosis to identify use of PET to coincide with the 4-month follow-up period used by SEER to provide cancer stage.21 Medicare payment information was available in each claim file; we summed line-item payments (from home health, hospice, and outpatient files), total claim payment amounts (from carrier and durable medical equipment files), or total reimbursement plus total daily per diem charges (from inpatient files) and adjusted all payments to 2008 dollars using the health care component of the Consumer Price Index.
We obtained all remaining variables from the SEER Patient Entitlement and Diagnosis Summary File. We extrapolated SEER cancer stage to the American Joint Committee on Cancer third edition staging system to provide a common staging system throughout the study period.5 Survival at 2 years was obtained from the SEER-based death date. Demographic variables included age, sex, race, ethnicity, marital status, and local census tract characteristics (i.e., metropolitan urban or rural status, percentage not finishing high school, percentage below the poverty line, and percentage of black race).
We defined discrete cohorts to represent the pre-PET period (1996–1997), the initial PET period (2000–2001), and the late PET period (2004–2005). We compared demographic and treatment characteristics using chi-square tests for categorical variables and Kruskal-Wallis nonparametric tests for continuous variables. We used multivariable logistic regression models (logit) to examine the use of surgical resection, radiation therapy, and chemotherapy over time using 1996–1997 as the reference period and modeling subsequent years as categorical variables.
We analyzed health care costs using multivariable ordinary least squares regression models of total Medicare payments in the inpatient, non-inpatient, and total claims files between 1996 and 2005. We modeled Medicare payments as logged costs to adjust for left-skewed cost data and to avoid nonnormally distributed error terms. We calculated relative percentage differences in costs as described by Kennedy.22
All regression models controlled for disease stage and demographic characteristics. In order for all observations to use sociodemographic variables from the 2000 census, we used 1996 as the earliest year in the regression analyses. Patient SEER registry was included as a control variable in all regressions to mitigate the effects of regional differences in treatment practices and costs.7
We performed sensitivity analysis of cost regressions by analyzing short-term costs (within 4 mo of diagnosis) to provide a more direct surrogate for initial surgical costs. All results were considered significant at p less than 0.05. We used SAS version 9.2 (SAS Institute, Cary, NC) for all analyses.
Of the 159,201 Medicare beneficiaries diagnosed with cancer of the lung and bronchus by SEER between 1996 and 2005, 58,575 met the eligibility criteria for the analysis (Supplemental Figure, Supplementary Digital Content 1, http://links.lww.com/JTO/AXX). Later NSCLC cohorts had a larger percentage of patients who were older than 80 years, were women, had more than 1 comorbid condition, and were not married (Table 1; all p < 0.001).
As shown in Table 2, between 1996–1997 and 2004–2005, decreasing proportions of patients underwent surgical resection (29.1% versus 24.7%; p < 0.001) and radiation (49.4% versus 42.9%; p < 0.001). By contrast, the proportion of patients receiving chemotherapy increased significantly from 25.1% to 40.4% (p < 0.001). During the same period, total costs per patient increased from $47,300 to $52,200 (p < 0.001). Increasing costs were entirely driven by non-inpatient costs ($18,400 versus $25,300; p < 0.001). Inpatient costs in the year after diagnosis decreased by $2200 per patient by 2000–2001 (p < 0.001). The overall proportion of patients undergoing surgical resection or radiation began to decrease by 2000, whereas chemotherapy use increased throughout the study period (Figure 1). Inpatient surgery costs decreased between 1998 and 2003, whereas nonsurgical inpatient costs remained stable over the study period and total costs increased steadily between 2000 and 2004 (Figure 2). The average cost of chemotherapy and radiation therapy combined constituted a relatively small and stable cost over the course of the study.
After adjustment for shifting patient demographic characteristics, and using 1996–1997 as the reference, the odds of undergoing surgical resection or radiation therapy decreased and the odds of receiving chemotherapy increased between 2001 and 2005 (Table 3). Surgical resection was more likely among patients who had localized disease or were married and was less likely among older patients, men, patients from census tracts with lower levels of education, and patients with advanced disease or multiple comorbid conditions. Both radiation therapy and chemotherapy were more likely among patients with advanced disease and were less likely among patients older than 80 years (all p < 0.001).
Total non-inpatient and overall mean costs per patient increased between 1996–1997 and 2005. After adjustment for shifting demographic characteristics and using 1996–1997 as the pre-PET baseline, 12-month inpatient costs in the year after diagnosis did not change significantly, but the coefficients for years 2000, 2001, and 2004 suggested a trend toward decreasing inpatient costs (Table 4). Sensitivity analyses of inpatient costs, in which we limited the cost analysis to the 4 months after diagnosis, showed a significant 12% decrease in inpatient costs. Total health care costs increased by 10% by 2005, driven entirely by increases in non-inpatient costs. Localized disease, black race, and comorbidity increased both inpatient and overall health care costs.
In sensitivity analyses, the presence of localized disease was more strongly associated with inpatient costs within 4 months than at 12 months. Overall costs and non-inpatient cost estimates were unaffected by analyses using a 4-month versus 12-month cost window.
After controlling for shifting demographic characteristics, we found that Medicare beneficiaries with NSCLC were less likely after the adoption PET to undergo surgery or radiation therapy and had lower mean inpatient costs between 1998 and 2005. However, during the same period, the use of chemotherapy and non-inpatient expenditures increased rapidly, more than offsetting potential savings in inpatient expenditures.
Estimates from 2003 through 2005 suggest a relatively stable reduction of 12% in inpatient expenditures, which in 2004–2005 would have amounted to an approximately $3200 reduction per patient. During the same period, increased non-inpatient costs counteracted any net cost savings. Interpretation of cost redistribution may be complicated by other changes in oncology practice that occurred during the study period, including a global shift of cancer care from inpatient to outpatient settings23 and evolving guidelines supporting increased chemotherapy use in patients with NSCLC.24 We observed a consistent increase in chemotherapy use before, during, and after the study period that did not appear temporally correlated with PET adoption. Alternatively, the observed decrease in surgical and associated inpatient costs could be due, in part, to a growing proportion of patients with multiple comorbid conditions in later cohort years who might be denied a surgical procedure for medical reasons unrelated to cancer. However, we did not observe a concomitant substitution or increase in radiation therapy, typically used in medically inoperable patients, which would argue against an increase in comorbidity as the sole explanation for our findings.
It has been largely accepted that PET may reduce futile thoracotomies that might otherwise occur when a patient who presents with occult metastatic disease undergoes local, definitive treatment for incurable disease.1 It should be noted, however, that PET scans can also result in false-positive results and could potentially cause cancellation of surgical resection for some patients. Because of appropriate, selective PET administration among patients with early-stage, less aggressive disease, demonstrating a direct association of PET with reduced thoracotomy and radiotherapy rates is difficult using a nonexperimental approach. Nonetheless, our study provides supporting evidence that PET-induced stage migration may have reduced rates of futile thoracotomy and radiation, accompanied by a decrease in subsequent inpatient health care costs.
Alternative explanations for decreasing surgery rates include changes in general oncology practice patterns unrelated to PET use,23 changing guidelines,24 and evidence suggesting the benefit of chemotherapy, and improvements in management of side effects. Alternatively, the use of higher-resolution computed tomographic scans or mediastinoscopy, increased appreciation of surgical morbidity in patients with advanced disease, or an aging and sicker population may have shifted the risks and benefits, actual or perceived, of local versus systemic treatment during the study period. In theory, a reduction in futile thoracotomy could lead to an inadvertent increase in the use of chemotherapy. Making this scenario less likely is the observation that chemotherapy rates increased before PET adoption. Nonetheless, the potential for PET to shift care from thoracotomy to chemotherapy has not previously been appreciated within small randomized trials and represents an important consideration when evaluating the overall effect of PET use on health care expenditures.
Our study has several limitations. First, we were only able to identify PET scans reimbursed by Medicare. To minimize missed claims, we limited all analyses to Medicare beneficiaries who were likely to have complete claims data. Second, the SEER registry overrepresents patients who are nonwhite, live in areas with less poverty, and live in urban areas, which may limit the generalizability of the findings.7 In addition, changes in treatment patterns and costs between SEER registries may vary. To help mitigate these differences, we included registry as a control variable in all regression analyses. Third, Medicare claims include intravenous chemotherapy and oral equivalents, but not chemotherapy or supportive medications filled as outpatient prescription. Oral chemotherapy can pose a substantial cost to patients and outside insurers, and our analysis likely underestimates the overall costs of oral chemotherapy and nonchemotherapy prescriptions for patients with cancer.
In conclusion, surgical resection, radiotherapy, and corresponding inpatient costs decreased among Medicare with NSCLS after the widespread adoption of PET. Consistent with previous research, this study suggests that PET has played a role in the upstaging of early-stage NSCLC and a corresponding reduction in futile attempts at local control of occult metastatic disease. The ability of PET to affect patient management, health care resource use, and costs remain important areas of ongoing research that may change as new treatments become available.
This study used the linked 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, Centers for Medicare & Medicaid Ser vices; Information Management Services (IMS); and the Surveillance, Epidemiology, and End Results (SEER) Program tumor registries in the creation of the SEER-Medicare database. Damon M. Seils, MA, Duke University, provided editorial assistance and prepared the manuscript. Mr. Seils did not receive compensation for his assistance apart from his employment at the institution where the study was conducted.
1. Fischer B, Lassen U, Mortensen J, et al. Preoperative staging of lung cancer with combined PET-CT. N Engl J Med. 2009;361:32–39
2. Maziak DE, Darling GE, Inculet RI, et al. Positron emission tomography in staging early lung cancer: a randomized trial. Ann Intern Med. 2009;151:221–228, W248
3. van Tinteren H, Hoekstra OS, Smit EF, et al. Effectiveness of positron emission tomography in the preoperative assessment of patients with suspected non-small-cell lung cancer: the PLUS multicentre randomised trial. Lancet. 2002;359:1388–1393
4. Chee KG, Nguyen DV, Brown M, Gandara DR, Wun T, Lara PN Jr. Positron emission tomography and improved survival in patients with lung cancer: the Will Rogers phenomenon revisited. Arch Intern Med. 2008;168:1541–1549
5. Dinan MA, Curtis LH, Carpenter WR, et al. Stage migration, selection bias, and survival associated with the adoption of positron emission tomography among medicare beneficiaries with non-small-cell lung cancer, 1998-2003. J Clin Oncol. 2012;30:2725–2730
6. Verboom P, van Tinteren H, Hoekstra OS, et al.PLUS study group. Cost-effectiveness of FDG-PET in staging non-small cell lung cancer: the PLUS study. Eur J Nucl Med Mol Imaging. 2003;30:1444–1449
7. Warren JL, Klabunde CN, Schrag D, et al. Overview of the SEER-Medicare data: content, research applications, and generalizability to the United States elderly population. Med Care. 2002;40:IV-3–IV-18
8. Schrag D, Bach PB, Dahlman C, Warren JL. Identifying and measuring hospital characteristics using the SEER-Medicare data and other claims-based sources. Med Care. 2002;40(8 Suppl):IV-96–IV-103
9. Cooper GS, Virnig B, Klabunde CN, Schussler N, Freeman J, Warren JL. Use of SEER-Medicare data for measuring cancer surgery. Med Care. 2002;40(8 Suppl):IV-43–IV-48
10. Warren JL, Harlan LC, Fahey A, et al. Utility of the SEER-Medicare data to identify chemotherapy use. Med Care. 2002;40(8 Suppl):IV-55–IV-61
11. Virnig BA, Warren JL, Cooper GS, Klabunde CN, Schussler N, Freeman J. Studying radiation therapy using SEER-Medicare-linked data. Med Care. 2002;40(8 Suppl):IV-49–IV-54
12. Klabunde CN, Warren JL, Legler JM. Assessing comorbidity using claims data: an overview. Med Care. 2002;40(8 Suppl):IV-26–IV-35
13. Potosky AL, Warren JL, Riedel ER, Klabunde CN, Earle CC, Begg CB. Measuring complications of cancer treatment using the SEER-Medicare data. Med Care. 2002;40(8 Suppl):IV-62–IV-68
14. Freeman JL, Klabunde CN, Schussler N, Warren JL, Virnig BA, Cooper GS. Measuring breast, colorectal, and prostate cancer screening with medicare claims data. Med Care. 2002;40(8 Suppl):IV-36–IV-42
15. Earle CC, Nattinger AB, Potosky AL, et al. Identifying cancer relapse using SEER-Medicare data. Med Care. 2002;40(8 Suppl):IV-75–IV-81
16. Baldwin LM, Adamache W, Klabunde CN, Kenward K, Dahlman C, Warren JL. Linking physician characteristics and medicare claims data: issues in data availability, quality, and measurement. Med Care. 2002;40(8 Suppl):IV-82–IV-95
17. Bach PB, Guadagnoli E, Schrag D, Schussler N, Warren JL. Patient demographic and socioeconomic characteristics in the SEER-Medicare database applications and limitations. Med Care. 2002;40(8 Suppl):IV-19–IV-25
18. Brown ML, Riley GF, Schussler N, Etzioni R. Estimating health care costs related to cancer treatment from SEER-Medicare data. Med Care. 2002;40(8 Suppl):IV-104–IV-117
19. Cooper GS, Yuan Z, Stange KC, Dennis LK, Amini SB, Rimm AA. Agreement of Medicare claims and tumor registry data for assessment of cancer-related treatment. Med Care. 2000;38:411–421
20. Farjah F, Flum DR, Ramsey SD, Heagerty PJ, Symons RG, Wood DE. Multi-modality mediastinal staging for lung cancer among Medicare beneficiaries. J Thorac Oncol. 2009;4:355–363
21. Shahinian VB, Kuo YF, Gilbert SM. Reimbursement policy and androgen-deprivation therapy for prostate cancer. N Engl J Med. 2010;363:1822–1832
22. Kennedy PE. Estimation with correctly interpreted dummy variables in semilogarithmic equations. Am Econ Rev. 1981;71:801
23. Tangka FK, Trogdon JG, Richardson LC, Howard D, Sabatino SA, Finkelstein EA. Cancer treatment cost in the United States: has the burden shifted over time? Cancer. 2010;116:3477–3484
24. Pfister DG, Johnson DH, Azzoli CG, et al.American Society of Clinical Oncology. American Society of Clinical Oncology treatment of unresectable non-small-cell lung cancer guideline: update 2003. J Clin Oncol. 2004;22:330–353
Carcinoma; Non–small-cell lung cancer; Health care costs; Positron emission tomography
Copyright © 2014 by the International Association for the Study of Lung Cancer
Highlight selected keywords in the article text.