Institutional members access full text with Ovid®

Share this article on:

00005650-200702000-0000800005650_2007_45_154_zeliadt_treatment_2article< 79_0_7_4 >Medical Care© 2007 Lippincott Williams & Wilkins, Inc.Volume 45(2)February 2007pp 154-159Trends in Treatment Costs for Localized Prostate Cancer: The Healthy Screenee Effect[Original Article]Zeliadt, Steven B. PhD, MPH*†; Etzioni, Ruth PhD*†; Ramsey, Scott D. MD, PhD*†; Penson, David F. MD, MPH*‡; Potosky, Arnold L. PhD§From the *Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, Washington; †Department of Health Services, School of Public Health and Community Medicine, University of Washington, Seattle; ‡Department of Urology and Preventive Medicine, Keck School of Medicine, University of Southern California/Norris Cancer Center, Los Angeles, California; and §Division of Cancer Control and Population Sciences, National Cancer Institute, Bethesda, Maryland.Supported by grants CA-92408 and CA-88160 from the National Cancer Institute. This study used the linked SEER-Medicare database, a collaborative effort of the Applied Research Program, NCI; the Office of Research, Development and Information, CMS; Information Management Services (IMS), Inc.; and the Surveillance, Epidemiology, and End Results (SEER) Program tumor registries. The interpretation and reporting of these data are the sole responsibility of the study authors.Reprints: Steven B. Zeliadt, Fred Hutchinson Cancer Research Center; P.O. Box 19024; M2-230; Seattle, WA 98109-1024. E-mail: We sought to obtain estimates of trends in initial treatment costs during the prostate-specific antigen (PSA) era that account for the changing patient case-mix associated with screening.Subjects: We used reimbursement claims for Medicare-eligible subjects diagnosed with nonmetastatic prostate cancer between 1991 and 1999. Patients were grouped by initial treatment, with 17,846 receiving radical prostatectomy (RP), 25,933 receiving external beam radiotherapy (XRT), and 4525 receiving brachytherapy (BT).Methods: Cancer-attributable costs were computed by subtracting noncancer costs from total Medicare reimbursements among newly diagnosed cancer patients. Noncancer costs were estimated in 2 ways: (1) average costs among age-matched, cancer-free control subjects (control method) and (2) projections based on claims from subjects before diagnosis (prediagnosis method). Adjusted annual percent change in cancer-attributable costs was calculated using multivariate generalized linear models.Results: Noncancer costs increased at a much lower rate among men prior to diagnosis (3.8% annually) than among the general Medicare population (10.9%). The 2 approaches yielded different results; RP costs declined by 2.4% annually (prediagnosis method) versus 6.2% (control method); XRT costs declined by 1.5% versus 5.8%; and BT costs declined by 4.1% versus 8.3%.Conclusions: Because of self-selection of PSA screening, men diagnosed with prostate cancer today are now healthier overall than men in the general population and are considerably healthier than men diagnosed previously. Estimates of cancer-attributable costs that do not account for this healthy selection effect are likely to be biased. Declines in cancer-attributable treatment costs are evident even after accounting for a healthy screenee effect, suggesting that there has been a real reduction in cancer treatment costs.Prostate cancer is the single most expensive cancer in terms of Medicare reimbursement costs, even considering the cost of lung cancer among both men and women.1 The majority of prostate cancer care costs are related to initial treatment. There has not been a comprehensive examination of trends in treatment costs over time since the introduction of PSA, although several studies have considered costs of individual treatments for periods of up to 4 years.2 Some studies have suggested that initial treatment costs for prostate cancer have been contained by clinical care pathways and a move to performing procedures in the outpatient setting.3–5 However, changing technologies may have mitigated these cost containment measures.2Another explanation for observations of declining costs may be that the composition of the case population has become healthier as a result of increased prostate-specific antigen (PSA) screening, because men participating in screening tend to be relatively healthy. Registry data indicate that prostate cancer cases are now being diagnosed in patients at earlier stages of disease and at younger ages. The impact of this changing patient mix on prostate cancer treatment costs is not known. The goal of this study is to determine to what extent changes in the case-mix of patients diagnosed with prostate cancer as the result of PSA screening can explain trends in initial treatment costs. The selective adoption of screening among healthy subjects has been previously observed and has been termed the healthy screenee effect.6To estimate the treatment-related or cancer-attributable portion of the costs of care, we subtract the expected noncancer costs of care from total costs incurred in the 6 month period following diagnosis. The standard method for estimating expected noncancer costs is to compute the average costs for age-matched controls without cancer in the general population.1,7,8 However, this matched-control approach will produce biased estimates of cancer-attributable costs if men diagnosed with cancer are more or less healthy than aged-matched controls in the general population.Because we are concerned about dissimilarities between screen-detected cases and the general population, we also estimate noncancer costs using data on medical claims among cases prior to their cancer diagnosis. We compare general cost trends among claims of age-matched prostate cancer-free controls with cost trends for prediagnosis claims, which should be similar as both trends reflect the experience of Medicare subjects in the absence of prostate cancer. This comparison allows us to assess the extent to which the prostate cancer case population diverges from the general population over the course of the PSA era.MATERIALS AND METHODSThe data for this study come from the linked Surveillance, Epidemiology and End Results (SEER)-Medicare database, which links SEER records on patients diagnosed with cancer to files containing their Medicare treatment claim history.9 We used data from 11 SEER areas, covering approximately 14% of the U.S. population. We included men aged 65 and older who had a new, primary diagnosis of prostate cancer between 1991 and 1999 (n = 114,308). Patients from the Los Angeles and San Jose registries were included beginning in 1992. We excluded subjects with metastatic disease (7618) and unknown stage of disease (13,279). Subjects were also excluded if they were not continuously eligible for both Medicare Part A and Part B or were enrolled in an HMO during the 12-month period before diagnosis (6065) or the 12 month period after diagnosis (2731).We examined only costs for patients who were actively treated with radical prostatectomy (RP), brachytherapy (BT), or external beam radiation therapy (XRT) within the first 6 months after diagnosis. Initial treatment was determined based on the combination of treatment procedure codes identified in the Medicare claims files and initial treatment recorded by SEER.10 A total of 33,749 men were excluded because they did not receive these 3 treatments. Subjects who received BT with any additional XRT were included in the BT group. Men who received a combination of RP and XRT were excluded (1516) because these men likely have locally advanced disease. Evaluating treatment costs for this population is beyond the scope of this analysis. Subjects who died within the first year of diagnosis (1046) were excluded to ensure that our cost estimates only included treatment costs and not costs associated with death. The final sample included 48,304 subjects.We examined total costs for eligible subjects based on all available Medicare claims for the 6-month period after diagnosis, including the month of diagnosis. Total monthly cost histories were generated for both cases and controls for Part A and Part B costs based on the reimbursement amount paid by Medicare. Part A covers charges for inpatient, skilled nursing facility, home health services following an inpatient stay, and hospice care. Part B includes charges for outpatient services, physician services, diagnostic tests, laboratory services, home health services not following a hospital stay, and other medical services and supplies. Medicare does not pay 100% of expected costs of medical services based on prospectively determined rates that vary by individual institutions. Center for Medicare and Medicaid Services has estimated that expected costs can be determined by increasing Part A payments by approximately 9% and Part B payments by 30%.1 To be consistent with prior studies of Medicare claims, the results we present are actual Medicare reimbursements and do not reflect the expected cost adjustments. The National Cancer Institute has constructed price-adjustors to take into account differences in purchasing power that vary in the Medicare payment system by geographic region and time. We applied these adjustors to ensure that costs for all time periods are adjusted for inflation and are expressed in terms of 2002 dollars.1Noncancer control subjects were identified from a random 5% sample of Medicare patients who never had cancer and whose claim histories are provided as part of the SEER-Medicare database. We identified 5 control subjects for each cancer case. Controls were matched on race, age (±1 year), had to be eligible for both Medicare Part A and Part B and not enrolled in an HMO at the time the case was diagnosed, and had to survive at least as long as the case subject.1,7,8,11 Terminal care costs occurring within 12 months of death were not included for either cases or matched controls. Matched noncancer control costs for each case were calculated as the empirical average monthly costs among the 5 identified controls matched to each case.To estimate noncancer costs based on prediagnosis claims, we computed average prediagnosis monthly costs among cases in the 12-month period before diagnosis. We then used these estimates to project future monthly costs for the 6-month interval corresponding to the period of initial treatment. Prediagnosis claims are used to project probable noncancer cost estimates during the initial treatment period, in effect simulating a scenario in which subjects have their same health but are free from prostate cancer. The objective of this method is to match patients to controls with all dimensions of “health” equal except the prostate cancer diagnosis. The 12-month period before diagnosis was chosen to ensure the health of the subjects was as similar before diagnosis as after diagnosis. Alternative periods were explored (2 to 14 months, 6 to 12 months, and 6 to 18 months before diagnosis). The choice of prediagnosis period did not alter the results.Projections were based on the results of a Generalized Estimating Equations (GEE) model fit to the logarithm of the monthly prediagnosis costs.12 We included the covariates age, race, Charlson comorbidity score, disease grade based on Gleason score (<6, 6–7, >7), and median income of each subject's zip code area from the 2000 census. Charlson comorbidity score was calculated by examining inpatient and physician Part B claims within the 12 month period prior to diagnosis for treatment of any major illness using the Klabunde algorithm.13Given the noncancer costs estimated either by the matched-control method or the prediagnosis claims method, we computed the cancer-attributable, initial treatment cost for each case over the 6 months following diagnosis. We used GEE to estimate the annual percent change in cancer-attributable costs, given clinical and demographic covariates. Time (calendar month) was included in the models as a linear variable; alternative parameterizations did not significantly alter the findings. Separate models were used for each treatment modality to estimate annual percent change, and results from the matched-control and prediagnosis claims methods were compared.RESULTSThe patient characteristics of the study subjects are presented in Table 1. The majority of study subjects (53.7%) were treated with XRT, with the use of BT increasing in later years. Treatment patterns for this population have been described elsewhere.10 The majority of patients did not have a claim for treatment of any major illnesses in the year prior to their diagnosis: 90.2% among patients receiving RP, 90.8% among patients XRT, and 91.1% among patients receiving BT. The proportion of men with no major illnesses did not change considerably over time.TABLE 1. Subject CharacteristicsThe use of hormone therapy increased during the 1990s. Patients receiving XRT and BT were nearly 4 times as likely to be receiving adjuvant hormone therapy as part of their initial treatment in the later period of the study (1997–1999) compared with the initial study period (1991–1993; Table 2).TABLE 2. Change in Study CharacteristicsPatient characteristics also changed over time (Table 2). Overall, the mean age of diagnosis declined from 71.5 years of age to 67.0 years of age. Patients also were increasingly likely to be from a zip code area with a greater median income over the course of the study. There was a small change in the percent of men dying from any cause within 2 years of diagnosis, suggesting that men diagnosed later in the study period may have been healthier than men diagnosed in previous periods.Total costs for the first 6 months of treatment, which include both cancer-attributable costs and noncancer costs, are presented in Figure 1. Average total costs for the 6-month period after diagnosis for men treated with RP were $16,119 in 1991 and $13,569 in 1999. For XRT average total costs declined from $12,894 to $10,996 between 1991 and 1999 and for BT total costs declined from $16,563 to $12,699.FIGURE 1. Total costs and cancer-attributable costs among cases stratified by initial treatment. A, total costs; B, cancer-attributable costs (matched control method); C, cancer-attributable costs (prediagnosis method).Cancer-attributable costs assessed using the matched-control method declined considerably for all treatment modalities from 1991 to 1999 (Fig. 1). Cancer-attributable costs using the prediagnosis claims method also declined, but at a considerably slower rate. The cancer-attributable costs for the initial treatment period using the matched-control method declined from $13,914 in 1991 to $8113 in 1999 for patients treated with RP; from $10,537 to $6116 for patients treated with EBRT, and from $14,278 to $7596 for BT patients. Cancer-attributable costs using the prediagnosis claims method declined from $14,866 in 1991 to $12,135 in 1999 for RP patients, from $11,172 to $9182 for XRT patients and from $15,137 to $11,088 for BT patients.Monthly noncancer cost estimates for the 2 methods are presented in Figure 2. Matched controls show a sharp increase in medical costs over time, whereas the noncancer costs projected from prediagnosis claims among prostate cancer cases are fairly constant. For example, average costs for control subjects without cancer matched by age and race to XRT patients increased, on average, 10.1% per year throughout the 1990s. Prediagnosis costs for XRT patients in the 12 months prior to diagnosis increased only 1.6% per year.FIGURE 2. Two methods for estimating noncancer costs.The average noncancer cost for matched-controls corresponding to the initial treatment phase for RP patients were $2205 in 1991 and $5456 in 1999. For controls matched to XRT patients, noncancer costs were $2357 in 1991 and $4879 in 1999 and costs for controls matched to BT patients were $2284 in 1991 and $5103 in 1999.In contrast, when the prediagnosis claims were used to project noncancer costs during the initial treatment phase, the average noncancer costs for RP patients were $1254 in 1991 and $1434 in 1999; for XRT patients the noncancer were $1722 and $1814, and for BT patients the costs were $1426 and $1611.As a result of these widely differing patterns of noncancer costs, the estimated annual percent change in costs for the matched-control and the prediagnosis claims methods were considerably different. The GEE model using matched controls estimated the annual percent change to be −7.8% (95% confidence interval −8.6 to −7.1) for men treated with RP, −6.2% (−6.8 to −5.6) for men treated with XRT and −7.5% (−9.0 to −5.9) for men treated with BT. When the prediagnosis claims method was used to estimate noncancer costs, the annual percent change was −3.7% (−4.3 to −3.1) among men treated with RP, −1.8% (−2.3 to −1.4) among men treated with XRT, and −3.5% (−4.5 to −2.5) among men treated with BT. The matched-control approach thus overestimates declines in costs among men treated with RP by a factor of 2.6, compared with the prediagnosis claims method. Similarly, declines in costs among men treated with XRT are overestimated by a factor of 3.9, and declines in costs among men treated with BT are overestimated by a factor of 2.0. Declines using the matched-control method are clearly a product of both real declines in total costs, as well as increases in noncancer costs in the general population. However, these increases are not matched by the trends in noncancer costs in the newly diagnosed case population because the case population is getting progressively healthier over the study period.DISCUSSIONIn this study, we observed a dramatic healthy screenee effect in initial treatment cost trends among men diagnosed with prostate cancer during the PSA era. Men who have opted to receive PSA screening appear to be healthier on average than age-matched men in the general population and patients previously diagnosed with cancer. This healthy screenee effect explains the majority of declines in treatment costs noted in previous studies.3–5 Failure to account for this healthy screenee effect results in trends that are biased by as much as 400%. However, even after controlling for this healthy screenee effect, there appears to be true declines in initial treatment costs for localized prostate cancer over the past decade which may be attributed to cost-containment strategies such as integrated care pathways and decreased length of inpatient hospital stays.2Although our results are limited to trends in costs, a profound healthy screenee effect can also be observed in overall survival from data available from the SEER cancer registry.14 Since 1991, men diagnosed with prostate cancer have experienced lower overall mortality rates compared with similar aged men in the U.S. without prostate cancer, even after accounting for their increased prostate-cancer mortality risk.15We attempted to control for the healthy screenee effect through the assessment of comorbidity. We did not observe noticeable differences in costs by comorbidity score, nor did we observe changes in comorbidity scores over time. In a secondary analysis we matched control subjects based on age, race and comorbidity score, as well as stratifying costs by comorbidity level, with the findings remaining consistent to our primary analysis. It is possible that claims-based assessments of comorbidity are not sufficiently sensitive for detecting subtle health status differences in the context of screening where most men have few pre-existing conditions. However, we did notice that the overall proportion of patients dying from any cause within 2 years of diagnosis declined over time, indicating that subjects diagnosed in later years were healthier than in earlier years (Table 2). We note that previous studies have observed that claims-based comorbidity assessments have been shown to have little influence on prostate cancer costs.8Our study represents costs only for Medicare-eligible subjects; cost patterns for men younger than age 65 may be different. Also, we only examined treatment costs for men who received primary therapy. Men selecting watchful waiting or other conservative-management strategies were excluded, which may have contributed to the healthy screenee effect we observed for men receiving primary treatment. We only examined direct medical care costs through reimbursement claims and did not include indirect costs such as a patient's own time spent receiving and recovering from treatment.16,17 Our primary analysis excluded costs for men who died of any cause within 12 months of their diagnosis to ensure that our cancer-attributable cost estimates included only costs directly related to prostate cancer treatment and not costs related to dying. In a secondary analysis we included 1046 men who died within 12 months of being diagnosed and treated. Including these costs increased total cost estimates by approximately $100, with the declining trend in costs remaining identical to the primary analyses.The healthy screenee effect has been previously noted to be important in assessing outcomes in other screening contexts.6,18 Prostate cancer screening was rapidly adopted in the 1990s, with nearly 80% of eligible men having been screened by the end of the decade.19 The dramatic healthy screenee effect we observe may or may not apply to other health care settings. Prostate cancer screening targets a disease that primarily occurs among older men, a population who may be experiencing many different types of health problems. Therefore, the self-selection of healthy subjects to screening may be more pronounced compared with other screening modalities that target younger populations who are more homogenously healthy.Our results are consistent with previous estimates of treatment costs from specific time points during the 1990s.2 A study of Medicare charges found the mean cost of RP between 1993 and 1996 was $18,312 and average charges for XRT were $15,159.20 The higher costs in this study compared with our findings likely reflect the difference between the amount charged to Medicare used by Brandeis et al and the amount of the final Medicare reimbursement which was the cost measure we used. A study of Medicare reimbursements examining total costs for a 10 month period of initial treatment resulted in findings similar to total costs presented in Figure 2.21Even after accounting for the healthy screenee effect, costs of initial treatment have declined. There are several plausible explanations for these declines. First, efforts to reduce length of stay and perform procedures including radiation therapy in the outpatient setting likely contribute to declining costs. Earlier diagnosis due to screening may also avoid the need for node dissection for staging, lowering the extent and cost of surgery. Clinical care pathways for RP have been shown to significantly reduce average length of stay and total costs.3,4 We did observe that costs for XRT patients did not decline as much as costs for RP and BT patients. This may be due to the heavy use of adjuvant hormone therapy in the late 1990s among XRT patients.10In the United States, studies of prostate cancer outcomes (including costs and survival) during the PSA era must account for the influence of a healthy screenee effect or observed changes may be overestimated. Other cancers or diseases in which screening or surveillance is used to identify cases may also be subject to this healthy screenee effect. We were able to compare prediagnosis medical care claims for men with prostate cancer to claims from men in the general population as a method for quantifying the extent to which the case-mix of newly diagnosed patients has changed during the PSA era. Future studies of outcomes in the context of screening should account for a healthy screenee effect.ACKNOWLEDGMENTSWe would like to thank Marie Topor at IMS for creating the analysis dataset and Martin Brown at NCI for comments on a preliminary draft of the study and for preparing the Medicare cost-adjustors used in this study.REFERENCES1. Brown ML, Riley GF, Schussler N, et al. Estimating health care costs related to cancer treatment from SEER–Medicare data. Med Care. 2002;40:104–117. [Medline Link] [Context Link]2. Saigal CS, Litwin MS. The economic costs of early stage prostate cancer. Pharmacoeconomics. 2002;20:869–878. [Context Link]3. Leibman BD, Dillioglugil O, Abbas F, et al. Impact of a clinical pathway for radical retropubic prostatectomy. Urology. 1998;52:94–99. [CrossRef] [Medline Link] [Context Link]4. Litwin MS, Smith RB, Thind A, et al. Cost-efficient radical prostatectomy with a clinical care path. J Urol. 1996;155:989–993. [CrossRef] [Full Text] [Medline Link] [Context Link]5. Koch MO. Cost-efficient radical prostatectomy. Semin Urol Oncol. 1995;13:197–203. [Medline Link] [Context Link]6. Weiss NS, Rossing MA. Healthy screened bias in epidemiologic studies of cancer incidence. Epidemiology. 1996;7:319–322. [Full Text] [Medline Link] [Context Link]7. Baker MS, Kessler LG, Urban N, et al. Estimating the treatment costs of breast and lung cancer. Med Care. 1991;29:40–49. [CrossRef] [Full Text] [Medline Link] [Context Link]8. Taplin SH, Barlow W, Urban N, et al. Stage, age, comorbidity, and direct costs of colon, prostate, and breast cancer care. J Natl Cancer Inst. 1995;87:417–426. [CrossRef] [Medline Link] [Context Link]9. 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. [CrossRef] [Full Text] [Medline Link] [Context Link]10. Zeliadt SB, Potosky AL, Etzioni R, et al. Racial disparity in primary and adjuvant treatment for nonmetastatic prostate cancer: SEER-Medicare trends 1991 to 1999. Urology. 2004;64:1171–1176. [CrossRef] [Medline Link] [Context Link]11. Brown ML, Riley GF, Potosky AL, et al. Obtaining long-term disease specific costs of care: application to Medicare enrollees diagnosed with colorectal cancer. Med Care. 1999;37:1249–1259. [Context Link]12. Zeger SL, Liang KY. An overview of methods for the analysis of longitudinal data. Stat Med. 1992;11:1825–1839. [CrossRef] [Medline Link] [Context Link]13. Klabunde CN, Potosky AL, Legler JM, et al. Development of a comorbidity index using physician claims data. J Clin Epidemiol. 2000;53:1258–1267. [CrossRef] [Medline Link] [Context Link]14. SEER*Stat: Surveillance Research Program, National Cancer Institute SEER*Stat software, version 5. 2.2, 2004. Available at: . Accessed September 1, 2006. [Context Link]15. Zeliadt SB, Etzioni RD, Kristal AR. Estimated impact of the prostate cancer prevention trial on population mortality. Cancer. 2005;104:1556–1557. [Context Link]16. Yabroff KR, Warren JL, Knopf K, et al. Estimating patient time costs associated with colorectal cancer care. Med Care. 2005;43:640–648. [CrossRef] [Full Text] [Medline Link] [Context Link]17. Fryback DG, Craig BM. Measuring economic outcomes of cancer. J Natl Cancer Inst Monogr. 2004;33:134–141. [CrossRef] [Medline Link] [Context Link]18. Morrison AS. Screening in Chronic Disease. New York, NY: Oxford University Press; 1985. [Context Link]19. Jerant AF, Franks P, Jackson JE, et al. Age-related disparities in cancer screening: analysis of 2001 Behavioral Risk Factor Surveillance System data. Ann Fam Med. 2004;2:481–487. [Context Link]20. Brandeis J, Pashos CL, Henning JM, et al. A nationwide charge comparison of the principal treatments for early stage prostate carcinoma. Cancer. 2000;89:1792–1799. [CrossRef] [Medline Link] [Context Link]21. Burkhardt JH, Litwin MS, Rose CM, et al. Comparing the costs of radiation therapy and radical prostatectomy for the initial treatment of early-stage prostate cancer. J Clin Oncol. 2002;20:2869–2875. [CrossRef] [Full Text] [Medline Link] [Context Link] prostate; prostatic neoplasms; costs; cancer-attributable costs; healthy screenee bias;|00005650-200702000-00008#xpointer(id(R1-8))|11065405||ovftdb|SL0000565020024010411065405P52[Medline Link]|00005650-200702000-00008#xpointer(id(R3-8))|11065213||ovftdb|SL000079591998529411065213P54[CrossRef]|00005650-200702000-00008#xpointer(id(R3-8))|11065405||ovftdb|SL000079591998529411065405P54[Medline Link]|00005650-200702000-00008#xpointer(id(R4-8))|11065213||ovftdb|00005392-199603000-00053SL00005392199615598911065213P55[CrossRef]|00005650-200702000-00008#xpointer(id(R4-8))|11065404||ovftdb|00005392-199603000-00053SL00005392199615598911065404P55[Full Text]|00005650-200702000-00008#xpointer(id(R4-8))|11065405||ovftdb|00005392-199603000-00053SL00005392199615598911065405P55[Medline Link]|00005650-200702000-00008#xpointer(id(R5-8))|11065405||ovftdb|SL0004547819951319711065405P56[Medline Link]|00005650-200702000-00008#xpointer(id(R6-8))|11065404||ovftdb|00001648-199605000-00022SL000016481996731911065404P57[Full Text]|00005650-200702000-00008#xpointer(id(R6-8))|11065405||ovftdb|00001648-199605000-00022SL000016481996731911065405P57[Medline Link]|00005650-200702000-00008#xpointer(id(R7-8))|11065213||ovftdb|00005650-199101000-00004SL000056501991294011065213P58[CrossRef]|00005650-200702000-00008#xpointer(id(R7-8))|11065404||ovftdb|00005650-199101000-00004SL000056501991294011065404P58[Full Text]|00005650-200702000-00008#xpointer(id(R7-8))|11065405||ovftdb|00005650-199101000-00004SL000056501991294011065405P58[Medline Link]|00005650-200702000-00008#xpointer(id(R8-8))|11065213||ovftdb|SL0000224219958741711065213P59[CrossRef]|00005650-200702000-00008#xpointer(id(R8-8))|11065405||ovftdb|SL0000224219958741711065405P59[Medline Link]|00005650-200702000-00008#xpointer(id(R9-8))|11065213||ovftdb|00005650-200208001-00001SL00005650200240iv11065213P60[CrossRef]|00005650-200702000-00008#xpointer(id(R9-8))|11065404||ovftdb|00005650-200208001-00001SL00005650200240iv11065404P60[Full Text]|00005650-200702000-00008#xpointer(id(R9-8))|11065405||ovftdb|00005650-200208001-00001SL00005650200240iv11065405P60[Medline Link]|00005650-200702000-00008#xpointer(id(R10-8))|11065213||ovftdb|SL00007959200464117111065213P61[CrossRef]|00005650-200702000-00008#xpointer(id(R10-8))|11065405||ovftdb|SL00007959200464117111065405P61[Medline Link]|00005650-200702000-00008#xpointer(id(R12-8))|11065213||ovftdb|SL00001375199211182511065213P63[CrossRef]|00005650-200702000-00008#xpointer(id(R12-8))|11065405||ovftdb|SL00001375199211182511065405P63[Medline Link]|00005650-200702000-00008#xpointer(id(R13-8))|11065213||ovftdb|SL00005077200053125811065213P64[CrossRef]|00005650-200702000-00008#xpointer(id(R13-8))|11065405||ovftdb|SL00005077200053125811065405P64[Medline Link]|00005650-200702000-00008#xpointer(id(R16-8))|11065213||ovftdb|00005650-200507000-00002SL0000565020054364011065213P67[CrossRef]|00005650-200702000-00008#xpointer(id(R16-8))|11065404||ovftdb|00005650-200507000-00002SL0000565020054364011065404P67[Full Text]|00005650-200702000-00008#xpointer(id(R16-8))|11065405||ovftdb|00005650-200507000-00002SL0000565020054364011065405P67[Medline Link]|00005650-200702000-00008#xpointer(id(R17-8))|11065213||ovftdb|SL0000224220043313411065213P68[CrossRef]|00005650-200702000-00008#xpointer(id(R17-8))|11065405||ovftdb|SL0000224220043313411065405P68[Medline Link]|00005650-200702000-00008#xpointer(id(R20-8))|11065213||ovftdb|SL00002808200089179211065213P71[CrossRef]|00005650-200702000-00008#xpointer(id(R20-8))|11065405||ovftdb|SL00002808200089179211065405P71[Medline Link]|00005650-200702000-00008#xpointer(id(R21-8))|11065213||ovftdb|00005083-200206150-00015SL00005083200220286911065213P72[CrossRef]|00005650-200702000-00008#xpointer(id(R21-8))|11065404||ovftdb|00005083-200206150-00015SL00005083200220286911065404P72[Full Text]|00005650-200702000-00008#xpointer(id(R21-8))|11065405||ovftdb|00005083-200206150-00015SL00005083200220286911065405P72[Medline Link]12065564Trends in Treatment Costs for Localized Prostate Cancer: The Healthy Screenee EffectZeliadt, Steven B. PhD, MPH; Etzioni, Ruth PhD; Ramsey, Scott D. MD, PhD; Penson, David F. MD, MPH; Potosky, Arnold L. PhDOriginal Article245