Long-term medical-care costs related to prostate cancer ... - Nature

Prostate Cancer and Prostatic Diseases (2010) 13, 278–284 & 2010 Macmillan Publishers Limited All rights reserved 1365-7852/10

www.nature.com/pcan

ORIGINAL ARTICLE

Long-term medical-care costs related to prostate cancer: estimates from linked SEER-Medicare data ´ Benedict3, CG Roehrborn4 and P Albertsen5 ME Stokes1, L Black2, A 1

United BioSource Corporation, Dorval, Quebec, Canada; 2Global Health Outcomes, GSK, Research Triangle Park, NC, USA; United BioSource Corporation, Budapest, Hungary; 4Department of Surgery (Urology), UT Southwestern Medical Center, Dallas, TX, USA and 5Department of Surgery (Urology), UConn Health Center, Farmington, CT, USA

3

Limited data are available concerning long-term prostate cancer (PCa)-related medical costs for use in assessing PCa prevention strategies. The aim of this study was to examine treatment, long-term survival, and long-term PCa-related costs by cancer stage. Costs in phases of care based on the natural history of PCa were also examined. Our data illustrate that initial care is characteristic of a rapid rate of cost accrual. This rate then decreases during continuing care. For terminal care, only stage IV resumes a rate of cost accrual similar to initial care. With average PCa-related costs of $18 168 observed over an average follow-up of 4 years, prevention strategies may result in a reduction in medical costs. Prostate Cancer and Prostatic Diseases (2010) 13, 278–284; doi:10.1038/pcan.2010.5; published online 9 March 2010

Keywords: prostate cancer; utilization; health-care costs; survival

Introduction Prostate cancer (PCa) is one of the most commonly diagnosed cancers in the United States.1 According to the American Cancer Society, there will be 192 280 newly diagnosed cases in 2009 and 27 360 men will die of the disease, making PCa the second leading cause of cancer death in American men.1 Local and regional PCa account for more than 90% of all PCa cases and the 5-year relative survival rate for these stages approaches 100%.1Survival rates have increased dramatically over the past 25 years.1 These increases are attributed to better treatment and advances in detection leading to earlier diagnosis.1 Pharmacologic therapies are undergoing evaluation for the prevention of PCa. A phase III study examining finasteride has been completed and results have shown a 25% reduction in PCa over 7 years.2 Another 5-alpha reductase inhibitor, dutasteride, is also currently being evaluated in a phase III trial.3 As additional data become available regarding the benefits and risks of these agents, an examination of their economic impact will also be important. This requires an assessment of the additional treatment costs with future savings these strategies offer in terms of preventing or delaying PCa. Although recent studies have examined PCa-related costs from a Medicare perspective using either the

Correspondence: ME Stokes, Health Economics, United BioSource Corporation, 185 Dorval Avenue, Suite 500, Dorval, QC, Canada H9S 5J9. E-mail: [email protected] Received 1 November 2009; revised 24 January 2010; accepted 29 January 2010; published online 9 March 2010

Surveillance, Epidemiology, and End Results (SEER)Medicare data,4–6 the Medicare 5% public use files,7 or the CaPSURE database,8 there are few recent studies examining costs in phases of care beyond initial treatment and stratified by cancer stage. Furthermore, many of the studies identified only examine costs for patients diagnosed with early stage PCa.6–8 A few older studies9,10 have examined costs in phases of care beyond initial treatment and have stratified estimates by cancer stage. However, these studies consider costs over earlier periods (1990–19919 and 1973–199110), which may be less relevant today given the changes in approach to PCa diagnosis and treatment that have occurred. The purpose of this study was to examine treatment, long-term survival, and long-term costs by cancer stage using data from the SEER-Medicare linked database in patients diagnosed with PCa during 1991–2002.

Methods Data source and patient population We identified patients with PCa from a database linking cancer registry information with Medicare claims (SEERMedicare). The database and its contents have been described in detail elsewhere.11 At the time analyses were conducted, data were available on incident cancer cases representing B26% of the US population with Medicare claims available from 1991 through 2004.12 Male patients aged 65 years and older with a first primary diagnosis of PCa between 1 January 1991 and 31 December 2002 were selected from SEER. A sample of male patients from the random 5% sample of Medicare

Costs of prostate cancer ME Stokes et al

beneficiaries without cancer were selected and matched to cases on 5-year age groups to estimate background medical costs unrelated to PCa. Details related to the sample selection procedures for the PCa and non-cancer cohorts have been described in a prior study examining first-year PCa costs.4

each time period. Long-term total population costs were estimated similarly to PCa-related costs. Modelbased estimates of total long-term costs were validated by comparing results with total average costs based on the actual Medicare payments appearing on the claims.

Estimation of monthly treatment phase costs We used a phased-based model similar to a previous study to estimate medical-care costs.13 Using survival data from SEER, the period of time from diagnosis until death (or date of censoring) was divided into distinct phases of care (initial, continuing, and terminal care). The number of months each patient contributed to a particular phase was determined using an algorithm designed to model the natural history of PCa as described in Brown et al.13 For patients surviving X18 months and dying before 2004, the initial phase was defined as the first 6 months after PCa diagnosis, the terminal phase as the last 12 months before death, and the continuing care phase as the time between initial and terminal phases. For patients surviving o18 months, the final 12 months of follow-up were allocated first to the terminal phase and the remaining months were defined as initial phase. The total average monthly cost of health-care utilization was calculated for each patient by summing the payment data on each Medicare claim for each treatment phase and then dividing by the total number of follow-up months patients contributed to each phase. The Medicare claims from 1991 to 2004 were used to estimate phase-specific costs. For patients alive as of December 2004, only cost data through 2003 were analyzed because we could not determine if costs were related to terminal or non-terminal care. During this period, the main initial therapeutic options including prostatectomy, radiation, and hormone therapy were available. All cost data were inflated to 2004 constant dollars using the medical-care services component of the Consumer Price Index. The analysis of monthly PCa treatment phase costs was conducted on patients diagnosed with PCa over a 12-year period (1991–2002). A sensitivity analysis was carried out to ascertain whether monthly costs for patients diagnosed during 1998–2002 were comparable to the primary analysis (years 1991–2002).

Statistical methods Non-parametric bootstrap methods were used to calculate 95% confidence intervals (CIs) around cost estimates.14 We re-sampled with replacement from the study dataset to create 1000 random samples. Total and PCa-related costs were calculated for each of the 1000 replicates. CIs were derived from the sample distribution of costs of the 1000 replicates at the 2.5 and 97.5% quantiles. Life expectancy was estimated by cancer stage using the Kaplan–Meier method. The time to death was calculated as the number of months from PCa diagnosis until death. Patients alive as of their last day of data in the Medicare claims were censored. Statistical significance was evaluated at a ¼ 0.05 using the log-rank test.

Estimation of average long-term costs The incremental differences in average monthly treatment phase costs between PCa cases and the non-cancer cohort were assumed to be PCa-related costs. Phasespecific monthly cost estimates were combined with survival data from SEER to calculate long-term costs for patients from diagnosis until either death or end of follow-up in the Medicare claims. First, the SEER data were used to estimate the proportions of patients, stratified by stage, who survived for each of the different time periods after diagnosis. Next, we estimated PCa-related costs for each survival time by using the corresponding phase-specific monthly cost estimates and then summing these costs across the number of months patients contributed to each treatment phase. As patients survived different lengths of time, we weighted our cost estimates based on the percentage of patients surviving

279

Results Patient characteristics Table 1 compares the baseline characteristics of PCa patients to the non-cancer cohort. Non-cancer patients

Table 1 Characteristics of the prostate cancer and non-cancer cohorts

N Age at study index Mean s.d

Prostate cancer

Non-cancer cohort

81 659

73 481

74.1 6.1

74.7 6.2

Age (5-year categories), N (%) 65–69 21 263 (26.0%) 70–74 26 081 (31.9%) 75–79 19 463 (23.8%) 80–84 9563 (11.7%) 85+ 5289 (6.5%)

17 556 (23.9%) 22 039 (30.0%) 18 510 (25.2%) 9887 (13.5%) 5489 (7.5%)

Age at start of each phase of care, mean (s.d.) Initial care 74.1 (6.1) Continuing care 74.1 (5.7) Terminal care 79.7 (6.6)

74.7 (6.2) 74.6 (5.8) 80.0 (6.7)

Survival (years) Median (95% CI) Mean (s.e.)

10.2 (10.0–10.3) 8.7 (0.019)

9.8 (9.8–10.0) 8.8 (0.022)

Geographic region, N (%): Midwest Northeast South West Unknown

20 089 (24.6%) 14 179 (17.4%) 8598 (10.5%) 38 793 (47.5%) 0 (0.0%)

12 980 (17.7%) 15 107 (20.6%) 11 687 (15.9%) 33 643 (45.8%) 64 (0.1%)

Median household income Mean s.d

36 932 15 487

35 517 14 741

Abbreviations: CI, confidence of interval; N, number of patients; s.e., standard error; s.d., standard deviation.

Prostate Cancer and Prostatic Diseases


280

were comparable to PCa cases with respect to age and survival. A larger proportion of PCa patients resided in the Midwest and West compared to non-cancer patients. Mean age at entry into each phase of care was similar between PCa cases and age-matched controls.

incremental difference averaging $1857 (Figure 1; stage I $1615, stage II $1890, stage III $2331, and stage IV $2212). For the All PCa Sample, monthly costs during continuing care were B$200 higher (Figure 2; stage I $247, stage II $170, stage III $96, and stage IV $344) versus non-cancer patients. Monthly terminal costs for stages I–III were comparable to non-cancer patients with the exception of stage IV. Stage IV terminal costs were over $1000 higher versus stages I–III.

Initial PCa patterns of treatment The use of specific PCa-related medical care observed during the first 6 months of follow-up is shown in Table 2. Almost one-third of all PCa patients did not receive an active therapy during initial treatment. Usage of prostatectomy was highest for stage III (column % 56.4, row % 47.3) and lowest for stage I (column % 6.4, row% 17.4). Treatment with external beam radiation and brachytherapy was more common among lower versus higher staged patients. There was very little use of cryotherapy (All stages column % 0.1) or bilateral orchidectomy (All stages column % 0.2).

Survival estimates Kaplan–Meier survival curves after initial cancer diagnosis are shown in Figure 2. The median survival time was B36 months among patients diagnosed at stage IV. Survival times were considerably longer for men diagnosed in stages I–III. Approximately 61, 56, and 55% of stages III, II, and I patients, respectively, were alive 120 months after the initial PCa diagnosis versus 18% of stage IV patients (Po0.001, all comparisons).

Monthly treatment phase cost estimates Monthly PCa-related and other costs, by treatment phase and cancer stage, are displayed in Figure 1. PCa-related costs in each phase were estimated by subtracting the total monthly costs of patients without cancer from the total monthly costs of PCa patients. The non-cancer cohort was used to estimate background medical costs unrelated to PCa (other costs). These results are based on a subsample of the data excluding monthly costs in the 95–100 percentiles. Investigation of the data led to the conclusion that cost outliers were driving up mean monthly costs for each treatment phase and inflating long-term estimates; when patients with costs in the 95–100 percentiles were excluded from the estimation of mean monthly costs, model-based estimates of long-term costs yielded results that were more consistent with actual mean Medicare payments (Actual $47 153 versus Model $46 751). The greatest difference in monthly costs between PCa and non-cancer patients was in the initial phase with an

Average long-term costs Long-term cost estimates for PCa patients are displayed in Figure 3. At the time of analysis, the Medicare data consisted of claims spanning calendar years 1991 through 2004. Total costs (discounted at 3%) were highest for stage IV ($57 691 95% CI $57 086–58 281) followed by stage III ($54 676 95% CI $54 286–55 030), stage II ($47 023 95% CI $46 585–47 432), and stage I ($40 969 95% CI $40 742–41 218). PCa-related costs (discounted at 3%) ranged from $16 582 to 24 996 (stage I $16 582 95% CI $16 505–16 666; stage II $17 016 95% CI $16 905–17 130; stage III $18 032 95% CI $17 942–18 117; and stage IV $24 996 95% CI $24 729–25 264). PCa costs represented a significant portion of total costs observed in this study and made up 40.5, 36.2, 33.0, and 43.3% of total costs for stages I, II, III, and IV, respectively. Figure 4 plots cumulative PCa-related costs (undiscounted) after cancer diagnosis using mean monthly

Table 2 Numbers of patients receiving initial treatment for prostate cancer and comparison with non-cancer cohort Service

Prostate cancer All stages n

%

Row

Stage I

%

Col

n

%

Row

Non-cancer cohort

Stage II %

Col

n

%

Row

Stage III

%

Col

n

%

Row

Stage IV

%

Col

n

%

Row

N Surgical treatment Prostatectomy Bilateral orchidectomy Cryotherapy of prostate

15 438 100 182 100 117 100

18.9 0.2 0.1

2681 17.4 44 24.2 103 88.0

6.4 3809 24.7 0.1 13 7.1 0.2 14 12.0

26.3 7297 47.3 0.1 14 7.7 0.1 o5 0.0

56.4 1651 10.7 0.1 111 61.0 0.0 o5 0.0

Radiation therapy External beam radiation Brachytherapy

13 073 100 7224 100

16.0 8.8

7614 58.2 5770 79.9

18.2 2675 20.5 13.8 1126 15.6

18.4 1577 12.1 7.8 248 3.4

12.2 1207 1.9 80

16.9 1981 17.3 17.2 2336 16.0 0.5 55 15.0 34.6 3642 14.1

13.7 831 7.2 16.1 1817 12.5 0.4 54 14.7 25.1 2371 9.2

6.4 1598 13.9 14.0 3216 22.0 0.4 64 17.4 18.3 5349 20.7

81 659

Hormone therapy Goserelin 11 483 100 Leuprolide 14 590 100 Triptorelin 367 100 No initial treatment (WW) 25 843 100

41 869

14.1 7073 17.9 7221 0.4 194 31.6 14481

61.6 49.5 52.9 56.0

14 505

12 935

Col

12 350

Abbreviations: % Col, column percentage (percent within cancer stage); N, number of patients; NA, not applicable; % treatment category); WW, watchful waiting (absence of treatment during first 6 months after PCa diagnosis).


%

9.2 1.1

Row

n

%

Row

%

Col

73 481 13.4 28 100.0 0.9 o5 0.0 0.0 o5 0.0

0.0 0.0 0.0

9.8 100 100.0 0.6 20 100.0

0.1 0.0

12.9 26.0 0.5 43.3

104 100.0 185 100.0 o5 0.0 NA NA

0.1 0.3 0.0 NA

, row percentage (percent within

Costs of prostate cancer ME Stokes et al $5,000

281

PCa-related Other†

$4,500 $4,000

$1,185

$37

$3,500

$386

$3,000 $2,500 $2,000 $3,130

$1,500

$2,331

$3,044

$1,890

$1,000

$2,212

$3,130

$3,130

$1,857

$96 $247

$170

$443 $446

$443 $446

$500 $0

$3,056

$1,615

Init Cont Term

$344

Init Cont Term

Stage I

Stage II

$443 $446

$443 $446

Init Cont Term

Init Cont Term

Init Cont Term

Stage III Cont=Continuing Care

Init=Initial Care

$216

$443 $446

Stage IV

All PCa Sample

Term=Terminal Care

Figure 1 Mean monthly costs among prostate cancer (PCa) patients, by stage and treatment phase. wRefers to the background medical costs unrelated to PCa. Other costs estimated using sample of Medicare beneficiaries without cancer.

(Figure 2). Men diagnosed at stage III were also younger versus stages I and II.

100% 90%

Percentage alive

80%

Stage III

Stage II

70% 60% Stage I

50% 40%

Stage IV*

30% 20% 10% 0% 0

10

20

30

40

50

60

70

80

90

100 110

120

Months following PCa diagnosis

Figure 2 Kaplan–Meier survival over 10-years after prostate cancer (PCa) diagnosis, by cancer stage. *Po0.0001, stage IV versus stages I, II, III using the log-rank test, all comparisons.

PCa-related phase specific estimates and the average number of months patients contributed to each treatment phase. Costs after PCa accumulate rapidly for all stages in the first 6 months reflecting a period of relatively high medical resource use characteristic of treating cancer. After initial treatment, the rate of cost accrual slows and levels off for stages I–III throughout continuing and terminal care, respectfully. Stage IV patients had higher terminal costs versus non-cancer patients and also contributed more months of follow-up to the terminal phase compared with stages I–III. This is reflected in the plot as stage IV PCa costs resume a rate of cost accrual similar to initial care. Patients diagnosed at stage III had higher PCa-related costs compared with stages I and II. Stage III patients used PCa-related medical care to a greater extent during the initial treatment phase (Table 2), and although monthly continuing care phase costs were lowest for stage III, this cohort contributed more continuing care months versus stages I and II because of longer survival times

Sensitivity analysis We observed similar relationships between initial, continuing, and terminal care monthly costs for the cohort of PCa patients diagnosed during 1998–2002. As in the primary analysis, PCa-related costs were highest during the initial phase (All $1542, stage I $1470, stage II $1541, stage III $1743, stage IV $1863). Monthly costs during continuing care ranged from $129 to $316 (All $244, stage I $270, stage II $172, stage III $129, stage IV $316). Continuing care costs were lowest for stage III as in the primary analysis. Monthly terminal PCa costs were comparable to the main analysis with the exception of stage IV where costs were over $500 higher (1998–2002 $1764 versus 1991–2002 $1185). Overall, initial care costs were $315 lower for the 1998–2002 cohort versus patients diagnosed during 1991–2002.

Discussion Our results show that the most common initial therapies for treating PCa were prostatectomy (18.9%) and external beam radiation (16.0%). The usage of prostatectomy was highest for stage III. Radiotherapy was more common among lower versus higher staged patients. Nearly one-third of all PCa patients did not receive an active therapy during initial care. Stage IV patients had the poorest prognosis after diagnosis as fewer than 20% survived 10 years compared with 55, 56, and 62% of those in stages I, II, and III, respectively (Po0.001, all comparisons). Total PCa-related costs averaged $18 168 (discounted at 3%) over an average follow-up of 4 years for the all patient sample. PCarelated costs varied by cancer stage and were highest for patients diagnosed at stage IV. The greatest differential in monthly costs between PCa and non-cancer patients Prostate Cancer and Prostatic Diseases


282

$70,000 Other† Costs PCa-related Costs $60,000

$50,000

$32,695

$40,000 $36,644

$28,583

$30,007 $30,000

$24,387

$20,000

$24,996 $10,000

$16,582

$17,016

$18,032

Stage I

Stage II

Stage III

$18,168

$0 Stage IV

All Stages

Figure 3 Average long-term costs (discounted at 3%) among prostate cancer (PCa) patients, by cost type and cancer stage. wRefers to background medical costs unrelated to Pca.

$30,000 37 * $25,000

PCa-related cumulative costs

Stage IV $20,000

72 * 55 * 43 *

$15,000

Stage III

$10,000

Stage II

Stage I

$5,000

$0 0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

Months following PCa diagnosis

Figure 4 Mean prostate cancer (PCa)-related cumulative costs after diagnosis, by cancer stage. The figure represents the average pattern of cost accrual observed during follow-up. Each line is the average number of months patients contributed to the analysis. Mean follow-up varied by cancer stage. *Mean number of months after prostate cancer diagnosis.

occurred during initial treatment with incremental costs ranging from $1615 to $2331 for stages I and III, respectively. There are few studies in the literature that have estimated PCa-related costs, by cancer stage among patients diagnosed with local, regional, and distant PCa in which the staging information has come directly from a clinical source.9,10 Both Taplin and Fireman have shown that remote cancer stages have lower initial costs relative to regionally staged patients. Fireman also concluded that stage effects on 15-year long-term cost estimates were modest and insignificant for PCa.10 Our Prostate Cancer and Prostatic Diseases

results agree with these studies in that our initial care costs for stage IV were lower versus stage III. However, results of our long-term analysis contrast with Fireman in that PCa-related costs for stage IV were nearly $7000 higher versus stage III. This difference may be the result of shifting patterns of care related to the sampling of patients in different time frames. Fireman sampled patients diagnosed with PCa during an earlier period (1973–1991), whereas our study included patients diagnosed from 1991 to 2002. A few studies in the literature were identified which estimate health-care costs during initial treatment in


patients with early stage PCa using Medicare databases.6,7 Our initial care PCa-related cost estimates among stages I (Figure 4, $9006 at month 6) and II (Figure 4, stage II $10 480 at month 6) are at the lower end of the ranges of estimates provided by Zeliadt6 ($9182–15 137) and Brandeis7($11 955–30 338). We attribute this difference primarily to the fact that both Zeliadt and Brandeis included only patients who were actively treated whereas the current analysis includes those receiving expectant management. Furthermore, the study by Brandeis used charges as a proxy for estimating costs rather than actual Medicare reimbursements. This study has several limitations. The study population was limited to Medicare patients aged X 65 years. As older men are more likely to receive expectant management as initial therapy, our cost estimates are probably lower in comparison to younger cohorts. The Medicare claims are primarily used for administrative purposes in obtaining reimbursement for services provided to Medicare-eligible patients; research is not the intended purpose of these data.15 The use of the Medicare claims to estimate medical costs is limited to the services used by patients, which Medicare covers. During the study period, Medicare did not cover oral hormone therapy or prescription drugs commonly used to treat pain or infection. We have defined phase-specific cancer-related monthly costs as the incremental difference between monthly costs among PCa patients and a cohort of patients without cancer matched according to age. This is an indirect approach to estimating costs related to the treatment of a disease. However, in an administrative database it is difficult to identify all costs related to the treatment of a specific condition using procedure and diagnostic codes. This is particularly true if patients are using general, non-cancer medical care to a greater extent versus non-cancer patients. This type of medical care cannot be captured using administrative codes. We did not have complete Medicare cost data from diagnosis until death for many of the patients in this study because of censoring. Patients were censored if they were alive as of the last day of Medicare follow-up. To quantify the impact of censored data on results, we estimated costs using only the survival information for patients diagnosed during 1991 as this patient subgroup had the greatest number of uncensored patients. Results from this analysis showed that stage IV still had the highest PCa-related costs ($35 405). However, in contrast with the full sample results, costs for stages I–III descended with increasing stage at diagnosis (PCa-related costs, stage I $29 841, stage II $24 353, stage III $21 582). This result was expected because stage III patients had lower PCa-related continuing care costs relative to stages I and II. With a longer average period of follow-up among PCa patients diagnosed in 1991, the impact of continuing care costs is greater versus subgroups with shorter follow-up periods. In estimating total and PCa-related costs, we used monthly cost inputs that were based on a sample of cases excluding patients with costs in the 95–100 percentiles. Thus, our estimates of the long-term total and PCa-related burden would have been higher had we included these patients in our estimation of monthly costs. Rather than estimating terminal PCa-related costs

with an incremental approach using non-cancer controls, we could have used the clinical data in SEER to determine whether patients died from cancer and then considered their total terminal phase costs as related to PCa. However, we believe our approach is more appropriate because had these patients not developed PCa, they likely would have died from conditions similar to the non-cancer cohort and accrued substantial terminal costs. Thus, we assumed all PCa costs were those that could potentially be eliminated through PCa prevention strategies. From a Medicare payer perspective these details are important because they illustrate that, with our overall conservative approach, there is still a large cost burden associated with PCa. We can combine our per-patient estimates of PCa costs with incidence data from the American Cancer Society16 to gain an initial impression of the aggregate PCa cost burden imposed on the US health-care system. Our analyses suggest that each PCa case has an average initial PCa-related treatment cost of $11 142 ($1857 average monthly cost for each of the first 6 months after diagnosis). Adjusting this cost estimate to 2009 US dollars and combining it with data from the American Cancer Society estimating that 192 280 new PCa cases were diagnosed during 2009 yield a high aggregate PCa health-care cost burden of upwards of 2.5 billion dollars annually for initial care alone. This estimate does not include treatment costs for patients diagnosed before 2009. Owing to improvements in screening modalities, patients are being diagnosed with PCa at younger ages and/ or earlier stages. Pharmacologic therapies that prevent or delay the onset of PCa could become important strategies for containing costs and improving patient outcomes. Our analyses quantify the medical-care costs related to PCa during initial, continuing, and terminal phases of PCa care. Furthermore, our analyses include patients diagnosed at distant stages as well as those receiving expectant management. To appropriately assess the economic benefits of prevention strategies, it is important to consider costs across all stages and treatment strategies.

283

Conflict of interest Funding for this study was provided by GlaxoSmithKline (GSK) to United BioSource Corporation. Dr Peter Albertsen is a paid consultant to GSK and Blue Cross Blue Shield. Dr Albertsen also participates in a research study sponsored by sanofi-aventis. Libby K Black is an employee of GSK. Dr Claus G Roehrborn is currently a paid consultant to GSK and has been an investigator for ´ gnes Benedict are full-time GSK. Michael E Stokes and A employees of United BioSource Corporation.

Acknowledgements This study used the linked SEER-Medicare database. We owe the sole responsibility of interpreting and reporting of these data. We acknowledge the efforts of the Applied Research Program, National Cancer Institute; the Office of Research, Development, and Information, Centers for Medicare and Medicaid Services; Information Management Services and the Surveillance, Epidemiology, and Prostate Cancer and Prostatic Diseases


284

End Results (SEER) Program tumor registries in the creation of the SEER-Medicare database. Portions of this research were presented at American Urological Association Annual Meeting, 17–22 May 2008, Orlando, FL. This study was funded by GlaxoSmithKline.

References 1 Cancer facts and figures. 2009; Available from http://www.cancer. org/docroot/STT/STT_0.asp. [accessed 15 October 2009]. 2 Thompson IM, Goodman PJ, Tangen CM, Lucia MS, Miller GJ, Ford LG et al. The influence of finasteride on the development of prostate cancer. N Engl J Med 2003; 349: 215–224. 3 Andriole G, Bostwick D, Brawley O, Gomella L, Marberger M, Tindall D et al. Chemoprevention of prostate cancer in men at high risk: rationale and design of the reduction by dutasteride of prostate cancer events (REDUCE) trial. J Urol 2004; 172: 1314–1317. 4 Roehrborn CG, Albertsen P, Stokes ME, Black L, Benedict A. First-year costs of treating prostate cancer: estimates from SEERMedicare data. Prostate Cancer Prostatic Dis 2009; 12: 355–360. 5 Warren JL, Yabroff KR, Meekins A, Topor M, Lamont EB, Brown ML. Evaluation of trends in the cost of initial cancer treatment. J Natl Cancer Inst 2008; 100: 888–897. 6 Zeliadt SB, Etzioni R, Ramsey SD, Penson DF, Potosky AL. Trends in treatment costs for localized prostate cancer: the healthy screenee effect. Med Care 2007; 45: 154–159. 7 Brandeis J, Pashos CL, Henning JM, Litwin MS. A nationwide charge comparison of the principal treatments for early stage prostate cancer. Cancer 2000; 89: 1792–1799.


8 Penson DF, Schonfeld WH, Flanders SC, Henke CJ, Warolin KL, Carroll PR et al. Relationship of first-year costs of treating localized prostate cancer to initial choice of therapy and stage at diagnosis: results from the CAPSURE database. Urology 2001; 57: 499–503. 9 Taplin SH, Barlow W, Urban N, Mandelson MT, Timlin DJ, Ichikawa L et al. Stage, age, comorbidity, and direct costs of colon, prostate, and breast cancer care. J Natl Cancer Inst 1995; 87: 417–426. 10 Fireman BH, Quesenberry CP, Somkin CP, Jacobson AS, Baer D, West D et al. Cost of care for cancer in a health maintenance organization. Health Care Financ Rev 1997; 18: 51–76. 11 Warren JL, Klabunde CN, Schrag D, Bach PB, Riley GF. Overview of the SEER-Medicare data: content, research applications, and generalizability to the United States elderly population. Med Care 2002; 40: IV-I3. 12 National Cancer Institute.. Surveillance Epidemiology and End Results. About SEER. 2005. Available from http://seer.cancer. gov/about. [accessed September 2007]. 13 Brown ML, Riley GF, Potosky AL, Etzioni RD. Obtaining longterm disease specific costs of care: application to Medicare enrollees diagnosed with colorectal cancer. Med Care 1999; 37: 1249–1259. 14 Efron B, Tibshirani R. An Introduction to the Bootstrap. Chapman & Hall: New York, 1993. 15 Iezzoni LI. Assessing quality using administrative data. Ann Intern Med 1997; 127: 666. 16 American Cancer Society. Overview: prostate cancer. Available from http://www.cancer.org/docroot/CRI/content/CRI_2_2_1X_ How_many_men_get_prostate_cancer_36.asp?sitearea ¼ . [accessed January 2010].