Abstract
Background
While the clinical implications of advanced melanoma have been extensively documented, little is known about the direct medical costs associated with the disease, particularly for elderly patients who carry the highest disease incidence and morbidity.
Objective
To document resource utilization and costs to the Medicare system for elderly patients with high-risk (stages IIB/C, IIIA/B, IIIC) or metastatic (stage IV) melanoma.
Methods
Data were taken from the Surveillance, Epidemiology, and End Results (SEER)-Medicare linked database combining clinical information on incident cancer cases in the US between 1991 and 2002 with longitudinal (1991–2005) administrative Medicare claims. Subjects aged ≥65 years with at least one stage IIB or higher melanoma diagnosis were selected. An index date was identified corresponding to the first observed stage IIB or higher diagnosis. Subjects were then categorized into mutually exclusive index disease stages, based on the SEER-reported melanoma stage observed at the index date. All subsequent analyses were stratified according to the index disease stage. Subjects without a record of death were required to have at least 6 months of continuous Medicare Part A and Part B benefits coverage before and after their index date. Subjects who died <6 months after their index date were retained for analysis. Resource utilization and costs were evaluated for each patient from index date until death, benefits cessation or end of the database (31 December 2005). Cost data were inflated to 2007 $US and stratified by the care setting in which they were incurred: inpatient hospital, skilled nursing facility, emergency room, physician office, home healthcare, hospice and other ancillary.
Results
6470 subjects met all inclusion criteria. Index stage distribution was: IIB/C (38%), IIIA/B (46%), IIIC (1%) and IV (15%). Median follow-up was 56, 39, 16 and 6 months, respectively. Patients with stage IV disease had 3.1 hospital days per month, compared with 0.5, 0.6 and 1.1 days for stage IIB/C, IIIA/B and IIIC patients, respectively. Adjusted inpatient costs for stage IV subjects were $US5565 per patient per month versus $US1031, $US1440 and $US2275 for stage IIB/C, IIIA/B and IIIC patients, respectively (p < 0.0001). Adjusted total costs were $US11 471 per month for stage IV subjects, compared with $US2338, $US3395 and $US6885 for stages IIB/C, IIIA/B and IIIC, respectively (p < 0.0001).
Conclusion
The per-patient cost of advanced melanoma is high. Hospital services are the largest component of these costs. Monthly costs for subjects with stage IV melanoma were 67% higher than costs for subjects with stage IIIC disease and >3-fold higher than costs for patients with stages IIIA/B and IIB/C. However, when combining estimated monthly costs with median follow-up duration (a proxy for survival time), total costs incurred by Medicare appear to be highest for patients diagnosed at stage IIIA/B.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
Background
Skin cancer has become the most common neoplasm in the US, with more than 1 million cases occurring annually.[1] Most of these cases are basal cell or squamous cell cancers, which have a high cure rate. However, the more serious form of skin cancer is melanoma, a proliferation of transformed melanocytes or pigment-producing cells. These tumours occur primarily on the skin, but also may arise in other tissues where pigmented cells are found. Overall, melanoma is the sixth most common cancer among both men and women in the US and causes 1–2% of all cancer deaths.[1,2] For localized melanoma, the 5-year survival rate is 99%, but this drops sharply to 65% and 15% for regional- and distant-stage diseases, respectively.[1]
While melanoma can be diagnosed in any age group, it is most common among persons of older age. Data collected from the Surveillance, Epidemiology, and End Results (SEER) programme of the National Cancer Institute (NCI) indicate that 44% of all new cases of invasive cutaneous melanoma between 2000 and 2004 occurred in adults 60 years of age or older.[3] Furthermore, the incidence rate of melanoma from 2000 to 2004 was highest among persons aged 70–79 years (53 per 100000 population), nearly six times the incidence rate found among those aged 20–29 years.[3]
With few effective therapeutic options available for advanced melanoma, the disease carries significant morbidity. The clinical implications of advanced melanoma have been extensively documented,[4,5] but little is known about the direct medical costs associated with the disease. One previous study audited medical charts from a regional referral centre to estimate per-patient expenditures for melanoma treatment,[6] while another study constructed an economic model using literature-based data to estimate the total annual costs of diagnosing and treating the disease.[7] However, no previous study has examined patient-level administrative data from real-world practice settings to estimate total all-cause direct medical costs incurred by third-party payers for patients with advanced melanoma, particularly elderly patients who carry the highest disease incidence and morbidity. To address this knowledge gap, we analysed a retrospective claims database with linked clinical information from a large cancer registry to document total per-patient monthly healthcare utilization and costs incurred by the Medicare system for elderly enrollees diagnosed with high-risk (stage IIB/C, IIIA/B or IIIC) or metastatic (stage IV) malignant melanoma.
Methods
Data Source
Data for this study were taken from the SEER-Medicare linked database. The SEER cancer registry was initiated in 1972 by the NCI and currently comprises nine state and three metropolitan, population-based registries that periodically collect information on nearly all (98%) newly diagnosed cancer cases, including malignant melanoma, in persons residing in SEER areas.[8] The current SEER areas have been shown to be nationally representative[9] and to capture approximately 24% of the total US population.[8] The information collected on each incident cancer diagnosis in the SEER registry includes patient-level demographic characteristics, date of diagnosis, clinical data about the tumour (e.g. histology, morphology, topography, stage, grade), selected types of first-line treatments (surgery and radiation, but not chemotherapy) provided within 4 months of diagnosis, and date and cause of death.
The NCI and the Centers for Medicare and Medicaid Services began linking SEER registry data with Medicare claims in 1991. This linkage is updated approximately every 3 years. The result is a database of persons aged 65 years and older with detailed clinical information on incident cancer, linked with comprehensive longitudinal healthcare utilization and cost data from Medicare. At the time of this study, Medicare claims were linked for elderly SEER patients with an incident cancer diagnosis between 1991 and 2002. For these patients, Medicare claims through 2005 were available.
Medicare claims linked to SEER registry data provide information on all services and associated billings covered under Medicare Part A benefits, including inpatient, skilled nursing facility, home healthcare and hospice care. The database also provides data on approximately 95% of services and billings covered under Medicare Part B, including physician visits, outpatient care, durable medical equipment and home healthcare. Data on services not covered by Medicare prior to 2006, including routine physical examinations, long-term residential care and oral drugs dispensed through outpatient pharmacies, are not available from the database. In addition to the patient-level demographic information available from SEER, the linked Medicare database contains information on each patient’s Medicare benefits eligibility, including start and stop dates of coverage, reason for Medicare entitlement and managed-care enrolment. Using the SEER-Medicare linked database, researchers can document patterns and associated costs of care among elderly patients with cancer, including initial and adjuvant chemotherapy regimens, radiation therapy, surgery and other downstream services, as well as comorbidities and healthcare utilization prior to the initial cancer diagnosis.[10–13]
Study Population
Subjects aged 65 years or older with a new diagnosis of malignant melanoma (International Classification of Diseases for Oncology, 2nd Edition [ICD-O-2] code in the range of C44.x) between 1 January 1991 and 31 December 2002 were selected for initial study inclusion. The stage observed for each SEER-reported melanoma diagnosis was determined based on clinical criteria set forth by the Tumor, Node, Metastases staging system for melanoma used by the American Joint Committee on Cancer (AJCC).[14] The AJCC stage for each diagnosis was assigned using an algorithm comprising the raw SEER variables HSTST (historical stage: in situ, localized, regional or distant), E10PN (number of positive lymph nodes), E10SZ (tumour size in mm) and E10EX (extent of disease: with or without ulceration). Cases meeting the AJCC criteria for high-risk or metastatic melanoma were identified as follows:
-
• stage IV: HSTST = distant
-
• stage IIIC: HSTST = regional and E10PN ≥4
-
• stage IIIA/B: HSTST = regional and E10PN <4
-
• stage IIB/C: HSTST = localized and E10SZ >2mm and E10EX = with ulceration.
Because our study focused on subjects diagnosed with high-risk or metastatic melanoma, we excluded subjects with a diagnosis of stage 0,I or IIA and no evidence of stage IIB/C, IIIA/B, IIIC or IV disease.
For each subject, an index date was defined as the date of the first observed melanoma diagnosis of stage IIB or higher. Subjects then were categorized into mutually exclusive index disease stages, based on the melanoma stage observed at the index date. All subsequent analyses were stratified according to the index disease stage. Subjects without a record of death were required to have at least 6 months of continuous Medicare Part A and Part B benefits coverage before and after their index date. Furthermore, subjects who died <6 months after their index date were retained for analysis. All resource utilization and cost outcomes were analyzed for each subject from their index date until the earliest out of death, interruption of benefits coverage or end of the database (31 December 2005).
In addition to the inclusion criteria described above, subjects were also required to have no evidence of health maintenance organization (HMO) enrolment following their index date, because HMOs are not obligated to provide the Medicare system with information about treatments (i.e. administrative procedure codes) and their associated costs. This exclusion is a widely accepted practice in analyses of Medicare claims data.[15,16]
Baseline Subject Characteristics
Subject characteristics that were measured at the index date included age, sex, race, geographic location, urban status of subjects’ residency and follow-up duration (measured as days between the index date and the earliest out of death, Medicare disenrolment or end of the database). The percentage of subjects who died at any point post-index was calculated, as were 1-year post-index survival rates. In addition to measuring these characteristics, the Charlson Comorbidity Index (CCI) was also computed to obtain a measure of subjects’ overall co-morbidity burden. The CCI includes 17 categories of co-morbidities, as defined by the International Classification of Diseases (9th Edition), Clinical Modification diagnosis codes, with associated weights corresponding to the severity of the co-morbid condition of interest.[17,18] A higher CCI score represents a higher overall co-morbidity burden. A CCI score ranges from 0, for subjects with no evidence of any comorbidity, to 33, for very rare cases in which all co-morbidities are present. CCI scores were calculated for each subject based on evidence of the relevant diagnoses during the 6 months prior to their index date. Because the objective of the CCI score was to evaluate underlying co-morbidity burden independent of malignant melanoma, diagnosis codes 172.xx and 173.xx for malignant melanoma were excluded from the CCI calculation.
Outcome Measures
All-cause resource utilization and costs were stratified by the major service sector in which they occurred, based on the place of service, revenue and other relevant codes associated with each Medicare claim. Specific categories that were evaluated included hospitalizations, skilled nursing facility admissions, emergency room (ER) visits, physician office visits, home healthcare and hospice care, all other outpatient ancillary care, and total resource utilization and costs. As noted, data on utilization and costs of prescription drugs dispensed through outpatient retail pharmacies were not available. All utilization and cost outcomes were aggregated across the entire follow-up period available for each subject and were reported at the level of per subject per month. As typically practiced in studies of healthcare costs, all cost data were inflated to 2007 $US at the claim level using year-specific inflators from the medical component of the US Consumer Price Index. For example, the observed reimbursed amount for any specific claim in 1995 was multiplied by the ratio of the average annual medical price index for 2007 to the corresponding index for 1995. Price indices for medical and other service sectors of the US economy are publicly available from the US Bureau of Labor Statistics.
Statistical Analyses
All analyses were carried out using SAS statistical software (version 9, Cary, NC, USA). Baseline subject characteristics and volume of healthcare utilization (e.g. hospital days, percentage of subjects with a hospital admission) were analysed with univariate descriptive statistics. Monthly per-subject healthcare costs reimbursed by the Medicare system were evaluated using multi-variate generalized linear models (GLM) with a log link function to resolve the issue of skewed cost distribution that is common in healthcare claims data.[19] In contrast to the traditional log ordinary least squares regression, GLM provides more robust coefficient estimates.[20] In addition, GLM allows for adjusted, predicted mean costs for subjects in each index disease stage to be directly calculated and statistically compared in the dollar scale, thereby avoiding the issue of potentially biased estimates that may result from the Duan smearing method for back-transformation of logged coefficients.[21] Co-variates included in the GLM models were age, sex, race, geographic location, urban status of subjects’ residency, follow-up duration and the year of subjects’ index date.
Results
We identified a total of 6470 subjects who met all study inclusion criteria (table I). The distribution of stage at index date was 2431 subjects (38%) with stage IIB/C disease, 2971 (46%) with stage IIIA/B, 88 (1%) with stage IIIC and 980 (15%) with stage IV. Subjects were mostly male, almost exclusively White, and evenly distributed across age categories. Sex, race and age distributions did not vary substantially by the index disease stage, with the exception of stage IIIC subjects who were more disproportionately male (69%). As expected, stage IV subjects had the highest post-index mortality rate (95%) and the lowest 1-year post-index survival rate (35%). Median follow-up duration for the overall study population was 40 months, with follow-up decreasing substantially with increasing stage: 56, 39, 16 and 6 months for stages IIB/C, IIIA/B, IIIC and IV, respectively.
Tables II and III, along with figure 1, present the results of our analysis of all-cause healthcare utilization and costs among Medicare enrollees with high-risk or metastatic melanoma. On average, subjects with stage IV disease had 3.1 hospital days per month, compared with 0.5, 0.6 and 1.1 days per month for subjects with stage IIB/C, IIIA/B and IIIC melanoma, respectively. However, the percentage of subjects with at least one hospital admission was highest among subjects with stage IIIC disease at index (figure 1). Adjusted mean inpatient costs among subjects with stage IV disease derived from the GLM regression models were $US5565 per subject per month versus $US1031, $US1440 and $US2275 for subjects with stages IIB/C, IIIA/B and IIIC disease, respectively (all p<0.0001). Subsequent analyses, in which we examined only hospitalizations with a primary or secondary discharge diagnosis of malignant melanoma, indicated that >50% of all hospitalization costs among stage IIIC or stage IV subjects were directly attributable to the disease. The proportion of all hospitalization costs attributable to melanoma in subjects with stage IIB/C and stage IIIA/B disease was only 11% and 25%, respectively.
Physician office visits and other ancillary services also constituted a significant proportion of total healthcare utilization and costs. Unlike hospital services, office visits were utilized most intensely by subjects with stage IIIC melanoma. On average, subjects with stage IIIC disease at index incurred 2.4 office visits per month, compared with 1.3, 1.5 and 2.0 visits per month for subjects with stage IIB/C, IIIA/B and IV melanoma, respectively. Adjusted mean office visit costs incurred by the Medicare system for subjects with stage IIIC disease were $US2415 per subject per month versus $US634, $US921 and $US2132 for stages IIB/C, IIIA/B and IV, respectively (all p<0.05). Adjusted per-subject monthly costs for other ancillary services, a residual category of care that excludes all inpatient, office, ER, home and hospice services, were estimated at $US3697 for subjects with stage IV disease at index, compared with $US661, $US934 and $US1531 for stages IIB/C, IIIA/B and IIIC, respectively (all p<0.0001). Adjusted total costs, inclusive of all noted service categories, were $US11471 per month for stage IV subjects, compared with $US2338, $US3395 and $US6885 for stages IIB/C, IIIA/B and IIIC, respectively (all p<0.0001).
Discussion
The magnitude of cost differences we found between subjects with stage IV melanoma and subjects with less advanced disease is consistent with cost data in previous literature that were derived under alternative methods. Tsao et al.[7] for example, estimated that stage IV disease represented approximately 55% of annual expenditures for treatments in the broader melanoma population. Stage I and II disease, by comparison, collectively represented only 5% of total melanoma expenditures. With stage III representing 34% of total expenditures, Tsao et al.[7] concluded that nearly 90% of annual melanoma costs in 1997 were attributable to the relatively small proportion (approximately 20%) of subjects diagnosed with advanced-stage disease.
Despite the findings of our study and those of Tsao et al.[7] regarding stage-specific cost differences, it remains unclear whether subjects with stage IV melanoma present higher long-term costs when considering the substantially shorter survival times typical of patients with metastatic disease. Extrapolation of our monthly per patient cost estimates to the median follow-up months shown in table I, which crudely proxies survival time, suggests that Medicare incurs the highest total cost for subjects diagnosed at stage IIIA/B ($US3395 per month × 39 months = $US132 405), and the lowest total cost for individuals diagnosed at stage IV disease ($US11417 per month × 6 months = $US68 826). While such calculations are informative, further study is needed to more formally compare the stage-specific costs of equivalent episodes of care. Furthermore, our study identifies and follows patients from the point of their first observed diagnosis of high-risk melanoma (i.e. stage IIB or higher) and does not examine subsequent disease progression or the costs associated with each disease stage within a given patient. Future study to explore the costs associated with disease progression is needed.
Our study is subject to several limitations inherent in analyses of Medicare claims data. First, because of the lack of a Medicare prescription drug (i.e. Part D) benefit prior to 2006, data on the use and costs of orally administered prescription drugs dispensed via outpatient retail pharmacies were not captured in our study data. Without such a pharmacy component, our analysis may understate all-cause medical costs incurred by patients with high-risk or metastatic melanoma. Second, our study relies primarily on administrative claims submitted solely for purposes of Medicare reimbursement and not for purposes of research, with no access to information collected from either the attending physician or the subject. Coding errors, for example, may have caused some utilization and cost measures to be incorrectly identified, and may also have caused some hospitalizations in the noted sub-analysis to be incorrectly attributed to melanoma. The impact of potential misclassification bias stemming from analyses of claims data has been described in previous research.[22,23] Third, our cost analyses represent direct medical costs incurred by the Medicare system. Our study therefore ignores the costs paid by non-Medicare payers (e.g. managed-care organizations, Medicaid) and the broader societal costs of melanoma, including caregiver burden and lost workplace productivity. Fourth, the extent of clinical trial enrolment in our study population is unknown. Data on services received by trial enrollees that were reimbursed by study sponsors, and not by Medicare, were not captured in our data, thereby causing further potential for underestimation of all-cause costs.
Finally, this study includes only those subjects aged 65 years or older. Findings presented here therefore may not be representative of the general population of patients with high-risk or metastatic melanoma. Younger patients diagnosed with high-risk or metastatic melanoma, for example, are likely to engage in more active, aggressive and potentially more costly treatment regimens than those undertaken by elderly patients due to better underlying health and greater tolerance of adverse treatment effects. Other utilization data examined in our study suggest that elderly patients with melanoma do, in fact, follow a generally conservative treatment approach: less than half (42%) of all subjects with stage IV melanoma received radiation therapy following their index date, while slightly more than one-quarter (27%) received chemotherapy. Chemotherapy usage was even lower among patients diagnosed at stages IIB/C (14%) and IIIA/B (21%), although it was substantially higher in the small sample of patients diagnosed at stage IIIC (45%). However, without data on subjects aged <65 years, we cannot empirically compare treatment patterns and associated costs between younger and elderly patients with advanced melanoma.
Despite these limitations, our large study cohort, with linked clinical and healthcare utilization data, provided a unique opportunity to examine the direct economic burden associated with high-risk and metastatic melanoma. Future studies including data on prescription drug utilization, subjects aged <65 years of age and information on indirect societal costs are needed to fully evaluate the economic implications of advanced melanoma.
Conclusions
This study is the first to provide a comprehensive, stage-specific evaluation of the real-world direct cost burden of high-risk and metastatic malignant melanoma using administrative claims data. Our findings suggest that the per-subject cost burden of advanced melanoma to the Medicare system is high and that hospital services are the largest component of these costs. Not surprisingly, subjects with stage IV melanoma imposed the highest burden, with costs nearly 5-fold higher than those for stage IIB/C subjects, more than 3-fold higher than those for stage IIIA/B subjects and approximately 67% higher than those for stage IIIC subjects. Further study is needed to assess whether patients with stage IV disease carry higher long-term or overall costs than individuals with lower-staged high-risk melanoma, in light of the substantially shorter survival time typical in metastatic melanoma.
References
American Cancer Society (ACS). Cancer facts and figures 2007. Atlanta (GA): American Cancer Society, 2007
Goldstein BG, Goldstein AO. Diagnosis and management of malignant melanoma. Am Fam Physician 2001; 63(7): 1359–68
Lachiewicz AM, Berwick M, Wiggins CL, et al. Epidemiologic support for melanoma heterogeneity using the Surveillance, Epidemiology, and End Results Program. J Invest Dermatol 2008; 128(5): 1340–2
Dale PS, Foshag LJ, Wanek LA, et al. Metastasis ofprimary melanoma to two separate lymph node basins: prognostic significance. Ann Surg Oncol 1997; 4(1): 13–8
Markovic SN, Erickson LA, Rao RD, et al. Malignant melanoma in the 21st century, part 2: staging, prognosis, and treatment. Mayo Clin Proc 2007; 82(4): 490–513
Hillner BE, Kirkwood JM, Agarwala SS. Burden of illness associated with metastatic melanoma: an audit of 100 consecutive referral center cases. Cancer 2001; 91(9): 1814–21
Tsao H, Rogers GS, Sober AJ. An estimate of the annual direct cost of treating cutaneous melanoma. J Am Acad Dermatol 1998; 38(5 Pt 1): 669–80
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 2002a; 40(8 Suppl.): IV3–18
Nattinger AB, McAuliffe TL, Schapira MM. General-izability of the surveillance, epidemiology, and end results registry population: factors relevant to epidemiologic and health care research. J Clin Epidemiol 1997; 50(8): 939–45
Warren JL, Harlan LC, Fahey A, et al. Utility of the SEER-Medicare data to identify chemotherapy use. Med Care 2002b; 40(8 Suppl.): IV55–61
Virnig BA, Warren JL, Cooper GS, et al. Studying radiation therapy using SEER-Medicare linked data. Med Care 2002; 40(8 Suppl.): IV49–54
Cooper GS, Virnig B, Klabunde CN, et al. Use of SEER-Medicare data for measuring cancer surgery. Med Care 2002; 40(8 Suppl.): IV43–8
Brown ML, Riley GF, Schussler N, et al. Estimating health care costs related to cancer treatment from SEER-Medicare data. Med Care 2002; 40(8 Suppl.): IV104–17
Greene FL, Page DL, Fleming ID, et al. editors. AJCC cancer staging manual. 6th ed. Wickford: Lippincott Raven Publishers, 2002
Hershman D, Hall MJ, Wang X, et al. Timing of adjuvant chemotherapy initiation after surgery for stage III colon cancer. Cancer 2006; 107(11): 2581–8
Sundararajan V, Hershman D, Grann VR, et al. Variations in the use of chemotherapy for elderly patients with advanced ovarian cancer: a population-based study. J Clin Oncol 2002; 20(1): 173–8
Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis 1987; 40(5): 373–83
Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992; 45(6): 613–9
Wedderburn RWM. Quasi-likelihood functions, generalized linear models, and the Gauss-Newton method. Biometrika 1974; 61: 439–47
Manning WG, Mullahy J. Estimating log models: to transform or not to transform? J Health Econ 2001; 20: 461–94
Manning WG. The logged dependent variable, heteroscedasticity, and the retransformation problem. J Health Econ 1998; 17: 283–95
Losina E, Barrett J, Baron JA, et al. Accuracy of Medicare claims data for rheumatologic diagnoses in total hip replacement recipients. J Clin Epidemiol 2003; 56(6): 515–9
Newcomer R, Clay T, Luxenberg JS, et al. Misclassification and selection bias when identifying Alzheimer’s disease solely from Medicare claims records. J Am Geriatr Soc 1999; 47(2): 215–9
Acknowledgements
This study and the preparation of this article were funded by Bristol Myers Squibb (BMS) Co. Keith Davis and Debanjali Mitra were responsible for the acquisition, management and analysis of all study data. Mr Davis and Ms Mitra also assisted with development of the study design and interpretation of the analysis results, and were primary authors of the article text. Srividya Kotapati assisted in developing the study design, interpretation of analysis results and drafting of the article text. Ramy Ibrahim and Jedd Wolchok assisted with interpretation of the analysis results and drafting of the article text. Mr Davis and Ms Mitra are employees of RTI Health Solutions, the research organization contracted by BMS to conduct this study. Drs Kotapati and Ibrahim are employees of BMS, and Dr Ibrahim also owns stocks in BMS. Dr Wolchok is an employee of Memorial Sloan-Kettering Cancer Center and is a paid advisory board member for BMS and Medarex. Portions of the study data presented in this article were previously presented in poster format at the 2008 North American meeting of the International Society for Pharmacoeconomics and Outcomes Research (ISPOR). The authors would also like to thank Izzy Pike, MD, for his valuable clinical insights provided throughout this study on issues of melanoma staging, treatments and other parameters relevant to the variable definitions used in the study.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Davis, K.L., Mitra, D., Kotapati, S. et al. Direct economic burden of high-risk and metastatic melanoma in the elderly. Appl Health Econ Health Policy 7, 31–41 (2009). https://doi.org/10.1007/BF03256140
Published:
Issue Date:
DOI: https://doi.org/10.1007/BF03256140