Total Medicare Costs Associated With Diagnosis and Treatment of Prostate Cancer in Elderly Men

This large cohort study uses the SEER-Medicare database to evaluate the Medicare costs associated with diagnosis and treatment of nonmetastatic prostate cancer in men 70 years or older.

Key Points

Question

What are the costs to the US Medicare program from the diagnosis and treatment of localized prostate cancer among elderly men?

Findings

In this SEER-Medicare study of 49 692 men aged 70 years or older, the median per-patient cost related to diagnosis and workup, treatment, follow-up, and morbidity management was $14 453 within 3 years after diagnosis; for those with a Gleason score of 6 or lower who received no active treatment within 1 year of diagnosis, the median cost was $1914. The estimated total 3-year cost to Medicare associated with the annual detection of prostate cancer in men 70 years or older is $1.2 billion.

Meaning

Reducing detection of localized prostate cancer in elderly patients represents a potential source of significant cost savings for the US Medicare program.

Abstract

Importance

Localized prostate cancer diagnosis and treatment among elderly men who are not likely to benefit represents a potential source of low-value health care services.

Objective

To quantify the costs to the Medicare program associated with detection and treatment of prostate cancer among elderly men in the United States.

Design, Setting, and Participants

This nationwide, population-based, retrospective cohort study uses the Surveillance, Epidemiology, and End Results (SEER)-Medicare linked database to identify men 70 years or older diagnosed with localized prostate cancer between 2004 and 2007 and to ascertain Medicare costs associated with diagnosis and workup, treatment, follow-up, and morbidity management of the disease. National Medicare costs were estimated using per-person costs, stage-adjusted prostate cancer incidence rates by age from SEER 2007 through 2011, and 2010 Census population estimates by age.

Main Outcomes and Measures

Estimated costs to the Medicare program overall, and in each (mutually exclusive) category related to diagnosis and workup, treatment, follow-up, and morbidity management.

Results

This nationwide, population-based, retrospective cohort study included 49 692 men with nonmetastatic prostate cancer from the SEER-Medicare database (all participants were 70 years or older; 25 981 [52.3%] were 76 years or older). The median per-patient cost within 3 years after prostate cancer diagnosis was $14 453 (interquartile range [IQR], $4887-$27 899). The majority of this cost was attributable to treatment costs (median, $10 558; IQR, $1990-$23 718). Patients with a Gleason score of 6 or lower who pursued initial conservative management (no treatment within 12 months of diagnosis) had a 3-year median total cost of $1914 per patient. The estimated total 3-year cost to the Medicare program associated with the annual detection of prostate cancer in men 70 years or older is approximately $1.2 billion. Increasing active surveillance use in those with Gleason score of 6 or lower could reduce this cost by $320 million.

Conclusions and Relevance

There is substantial cost to the Medicare program associated with the diagnosis and treatment of localized prostate cancer among elderly men in the United States, despite the fact that these men are unlikely to die of prostate cancer. The majority of costs are related to treatment. Reducing provision of low-value health care services among this patient population could result in significant health care savings.

Introduction

Prostate cancer is the most commonly diagnosed solid tumor in men,¹ and the majority of cases are detected through screening. Given the long natural history of this disease, many patients die with, rather than from prostate cancer.² Whether routine prostate cancer screening in the general population is beneficial is a subject of debate; however, there is wide consensus that screening should not occur among elderly men and those with limited life expectancy^3,4,5,6 because screening does not provide a significant survival benefit within 10 years.^7,8 The lack of potential benefit in diagnosing prostate cancer in elderly men also relates to the unclear benefit from aggressive treatment when a patient is diagnosed. In the Prostate Cancer Intervention vs Observation Trial (PIVOT), patients with prostate cancer were randomized to watchful waiting vs radical prostatectomy.⁹ After a median follow-up of 10 years, there was no survival benefit from prostatectomy (53% survival prostatectomy vs 50% watchful waiting, P = .22) in patients with a mean age of 67 years. On the other hand, treatment is associated with sexual dysfunction, bowel problems, urinary adverse effects, and decreased health-related quality of life.^10,11

Reduction in early prostate cancer diagnosis and overtreatment among elderly men represents an opportunity to reduce provision of low-value health care services that can result in net harm to patients and also waste resources in the US health care system. The purpose of the present study is to estimate the costs to the Medicare program associated with detection of prostate cancer among elderly men in the United States. We chose to study men 70 years or older, following published guidelines by the US Preventive Services Task Force, American College of Physicians, and the American Urological Association recommending cessation of screening at this age.^3,12,13 All associated costs within 3 years after prostate cancer detection were ascertained, including costs of diagnosis and workup, treatment, follow-up, and morbidity management. We further analyzed costs associated with detection and treatment of patients with likely nonlethal cancer (Gleason score ≤6), for whom many published guidelines recommend active surveillance instead of aggressive treatment.¹⁴

Methods

Data Source and Patient Cohort

The National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) program is an epidemiologic surveillance system collecting demographic information (age at diagnosis, race, marital status), clinical and tumor characteristics (prostate-specific antigen [PSA], clinical stage, Gleason score), and vital status for all individuals diagnosed with cancer and residing within 1 of the 18 SEER regions, which currently covers 28% of the population in United States.^15,16 The SEER-Medicare data are an electronic linkage of persons in SEER with their Medicare enrollment and claims data.¹⁶ Medicare Part A and B claims data provide information on medical services delivered in the inpatient and outpatient settings, including service dates, diagnoses, procedures, and treatments delivered during medical encounters. Furthermore, using Medicare data in the 12 months prior to prostate cancer diagnosis, we calculated Klabunde index—a validated modified Charlson comorbidity measure.¹⁷ The Klabunde index is a refined comorbidity measurement algorithm appropriate for analyses of commonly occurring cancers when using administrative claims databases.¹⁷ The study was approved by the University of North Carolina at Chapel Hill institutional review board, waiving written informed consent for deidentified retrospective data.

The analytic cohort consisted of men residing within a SEER region who were diagnosed from 2004 through 2007 at 70 years or older with a first primary, nonmetastatic prostate cancer. A portion of these patients may have developed disease progression within the subsequent 3 years, but this information is not captured in the SEER-Medicare data. Men diagnosed at autopsy or death were excluded from analysis. To ensure as complete capture of information as possible from claims data, analysis was restricted to all men with continuous fee-for-service plan enrollment (Medicare Parts A and B) and no managed care plan enrollment for the 12 months prior to prostate cancer diagnosis (set to the first day of the month and year of diagnosis from SEER) and 3 years after diagnosis. Men with disenrollment due to death within the 3 years after diagnosis were included in the analytic cohort. The final analytic cohort included 49 692 men.

Outcome Measures and Statistical Analysis

The primary goal of this study was to assess costs associated with and within 3 years after prostate cancer diagnosis. We used a microcosting approach to attribute Medicare payments from Parts A and B (inpatient, outpatient, and carrier files) to prostate cancer. The microcosting approach isolates payments specifically for prostate cancer care using diagnosis and procedure codes (see the eTable in the Supplement). Prostate cancer care was divided into 4 categories: (1) diagnosis and workup, (2) treatment (including initial and all subsequent treatments), (3) management of morbidity commonly attributed to prostate cancer treatments,¹⁸ and (4) posttreatment follow-up. As defined for this study, costs in these categories are mutually exclusive, so no cost is counted more than once.

Diagnosis and workup included claims from 30 days prior to the date of prostate cancer diagnosis through the earlier of initial treatment or 6 months after diagnosis. During this period, we included Medicare payments for claims with diagnosis and/or procedure codes for the PSA test, digital rectal examination, biopsy, pathologic analysis, imaging (including computed tomography, magnetic resonance imaging, and bone scans, which are commonly used for prostate cancer staging), and associated physician visits (eTable in the Supplement).

Treatment costs included claims for radical prostatectomy (including cost for anesthesia), radiation therapy including external beam radiation and brachytherapy (including cost for radiation planning), cryotherapy, androgen deprivation therapy, and chemotherapy from the date of prostate cancer diagnosis through the end of the follow-up period (ie, death or 3 years after diagnosis). Thus, this category captures costs from initial treatment as well as treatment for recurrent disease within 3 years.

Follow-up costs included claims from the end of the diagnosis-workup period through the end of the follow-up period. This category included claims with a diagnosis of prostate cancer and codes for PSA test, digital rectal examination, imaging, biopsy, pathologic analysis, and associated physician visits.

Morbidity costs included claims from the end of the diagnosis-workup period through the end of the follow-up period (identical time frame as follow-up costs). This category included claims for procedures related to morbidity commonly associated with prostate cancer treatments: urinary, gastrointestinal, sexual, and hip problems—all based on published studies.¹⁸

Many elderly patients with prostate cancer have urinary, bowel, and/or sexual dysfunction even before prostate cancer diagnosis. To ensure that we did not overcount prostate cancer treatment-related morbidity costs, we calculated costs in these categories in the year prior to prostate cancer diagnosis. We then calculated expected costs related to baseline dysfunction in these categories as the product of the prediagnosis period costs and the mean follow-up time after diagnosis in our cohort (2.75 years). Note that follow-up time for cost calculations was limited to 3 years after diagnosis, and is shorter for men who died earlier. Finally, the excess cost (observed posttreatment morbidity costs minus expected costs from baseline dysfunction) was attributed to treatment-related morbidity and reported; we set treatment-related morbidity cost to $0 as a minimum if the subtraction resulted in a negative amount.

All payments were adjusted to 2013 US dollars using personal health care and component price indices as reported by the Agency for Healthcare Research and Quality.¹⁹ Specifically, we used the hospital care component for the Medpar (inpatient) claims, the physician and clinical services component for the National Claims History (clinician) claims, and the overall personal health care index for the Outpatient claims.

We report costs by category of care (diagnosis and workup, treatment, morbidity, and follow-up) over the entire 3-year follow-up period. We then scaled actual costs from the SEER-Medicare population to estimate the national Medicare cost using per-person cost estimates from our analysis, prostate cancer incidence rates by age from the SEER 2007-2011 data²⁰ (adjusted to remove metastatic cancers), and 2010 Census population estimates by age. Our projections to national Medicare costs do not include beneficiary cost sharing but do assume that men with prostate cancer covered under Medicare Advantage have similar diagnosis and treatment costs to those under traditional, fee-for-service Medicare.

Results

Patient characteristics are detailed in Table 1. About half of the cohort was age 70 to 75 years, and the other half older than 75 years. There were 20 982 men (42%) with a Gleason score of 6 or lower, 16 927 (34%) with a Gleason score of 7, and 9018 (18%) with Gleason score from 8 to 10.

Table 1. Patient Characteristics.

Characteristic	Patients, No. (%)
Age, y
70-75	23 711 (47.7)
≥76	25 981 (52.3)
Comorbidity, Klabunde index
0	30 478 (61.3)
>0	17 864 (35.9)
Unknown	1350 (2.7)
Race/ethnicity
White	41 285 (83.1)
Black	4967 (10.0)
Other/unknown	3440 (6.9)
Married	32 181 (64.8)
Year of diagnosis
2004	13 050 (26.3)
2005	12 043 (24.2)
2006	12 339 (24.8)
2007	12 260 (24.7)
US regiona
Northeast	10 584 (21.3)
South	11 979 (24.1)
Central	6022 (12.1)
West	21 107 (42.5)
Regional educationb
Quartile 1 (highest)	12 451 (25.1)
Quartile 2	12 403 (25.0)
Quartile 3	12 425 (25.0)
Quartile 4	12 413 (25.0)
Medicaid dual eligible	1379 (2.8)
Residence
Nonmetropolitan	8424 (17.0)
Metropolitan	41 253 (83.0)
Gleason score
≤6	20 982 (42.2)
7	16 927 (34.10)
8-10	9018 (18.10)
Unknown	2765 (5.6)

^aGeographic regions are as follows: Northeast (Connecticut, New Jersey), Central (Detroit, Iowa, Utah, New Mexico), South (Atlanta, Rural Georgia, Kentucky, Louisiana), and West (California, Seattle, Hawaii).

^bQuartiles of non–high school graduates in census tract.

Costs associated with prostate cancer diagnosis and treatment are detailed in Table 2. For the overall cohort, the median per-patient cost within 3 years after prostate cancer diagnosis was $14 453, with treatment costs making up the majority ($10 558). We then examined costs by subgroups of patients based on age, comorbidity, and Gleason score. Men older than 75 years had lower median per-patient total costs than those aged 70 to 75 years, with cost differences between these groups mostly driven by differences in the treatment category. For men with Gleason scores of 6 or lower, median per patient cost was $12 616. However, patients with these lower Gleason scores who pursued initial conservative management (received no treatment within 12 months of diagnosis) had a 3-year median total cost of only $1914 per patient.

Table 2. Per-Patient Costs Associated With Detection and Treatment of Prostate Cancer in Men 70 Years or Older.

Characteristic	Median (IQR) Cost of Care, 2013 $US
	Diagnosis and Workup	Treatment	Morbiditya	Follow-up	Net Total
Age, y
70-75	1787 (1102-2667)	13 692 (5225-25 650)	147 (0-671)	257 (104-570)	17 906 (9272-29 893)
>75	1554 (908-2380)	6796 (1008-21 621)	12 (0-559)	227 (67-536)	10 641 (3503-25 728)
>75, Klabunde index >0	1583 (932-2423)	5758 (930-20 613)	0 (0-581)	211 (57-524)	9584 (3450-25 129)
Gleason score ≤6	1637 (972-2532)	8993 (819-20 323)	4 (0-511)	219 (79-486)	12 616 (3438-24 512)
Age 70-75 y	1732 (1050-2667)	11 399 (2513-22 213)	99 (0-596)	234 (96-507)	15 031 (6701-26 324)
Age >75 y	1512 (896-2389)	4903 (222-16 866)	0 (0-390)	198 (60-461)	8342 (2483-21 501)
Gleason score 7-10	1730 (1076-2552)	13 724 (4057-26 267)	156 (0-733)	272 (106-625)	18 133 (7276-30 682)
Gleason score ≤6 and no active treatment	1276 (411-2023)	0 (0-544)	0 (0-0)	125 (0-443)	1914 (695-3866)
Age 70-75 y	1140 (12-1959)	0 (0-547)	0 (0-0)	100 (0-481)	1800 (150-4015)
Age >75 y	1332 (748-2070)	0 (0-543)	0 (0-0)	133 (0-421)	2002 (998-3729)
Overall	1664 (995-2519)	10 558 (1990-23 718)	81 (0-620)	242 (90-552)	14 453 (4887-27 899)

Abbreviation: IQR, interquartile range.

^aMorbidity from 1 year before diagnosis, scaled to the average length of follow-up in the 3 years after diagnosis, subtracted from morbidity costs.

Projecting from the analytic sample of patients living in SEER regions, the total 3-year cost to the Medicare program associated with the annual detection of prostate cancer in men 70 years or older is approximately $1.2 billion (Table 3), including $451 million spent on men with Gleason scores of 6 or lower. However, if all men with these lower Gleason scores pursued initial conservative management (no treatment within the first 12 months after treatment), then it would lead to a reduction of total costs by $320 million. The Medicare cost for detection of prostate cancer in men older than 75 years is $601 million; if all patients with a Gleason score of 6 or lower in this age group pursued initial conservative management, then this cost would be reduced by $132 million. Men with a Klabunde score greater than 0 have worse than population-median comorbidities. The total Medicare cost for detection of prostate cancer in men older than 75 years and with a Klabunde score greater than 0 (the patient group expected to have <10-year life expectancy) is $230 million.

Table 3. Aggregate 3-Year Medicare Costs of Prostate Cancer Diagnosis in Elderly Men.

Patient Group	Medicare Costs, Millions of 2013 $US
	SEERa	National (Estimated)
Age >70 y	225	1228
Age 70-75 y	122	627
Age >75 y	104	601
Age >75 y, Klabunde index >0	40	230
Age ≥70 y, Gleason score ≤6	83	451
Age ≥70 y, Gleason score 7-10	134	732

^aCalculated from average annual number of incident cases, 2004-2007.

A breakdown of costs by whether patients received aggressive treatment initially and by age group is illustrated in Figure 1. A breakdown of costs by whether patients received aggressive treatment initially and by Gleason score is illustrated in Figure 2. Sizable costs are evident, especially among men receiving treatment, even in groups with limited life expectancy.

Figure 1. Mean 3-Year Per-Patient Costs for Prostate Cancer by Patient Age and Whether Patient Received Aggressive Initial Treatment.

Note that the 3-year average per-patient treatment costs reflect the total treatment costs incurred by the proportion of patients who received treatment in years 2 and 3 divided by all patients who pursued initial active surveillance; expected annual nonprostate morbidity costs subtracted from reported morbidity costs. Comorbid indicates comorbidities; Tx, aggressive initial treatment.

Figure 2. Mean 3-Year Per-Patient Costs for Prostate Cancer by Patient Age, Gleason Score, and Whether Patient Received Aggressive Initial Treatment.

Note that the 3-year average per-patient treatment costs reflect the total treatment costs incurred by the proportion of patients who received treatment in years 2 and 3 divided by all patients who pursued initial active surveillance; expected annual nonprostate morbidity costs subtracted from reported morbidity costs. Comorbid indicates comorbidities; Tx, aggressive initial treatment.

Discussion

Each diagnosed case of prostate cancer is associated with downstream costs, and the main goal of the present study was to calculate associated costs for 3 years after diagnosis. This study found that diagnosing men 70 years or older with prostate cancer each year in the US is associated with 3-year costs to the Medicare program in excess of $1.2 billion, including $451 million for men diagnosed with Gleason scores of 6 or lower. Treatment costs were the primary driver of costs associated with prostate cancer diagnosis. Prostate cancer is often a slow-growing disease, with a 95% 15-year relative survival among all diagnosed.²¹ Thus, elderly patients, especially those with comorbidities, are unlikely to die of prostate cancer or benefit from screening. This is reflected in published guidelines, including the 2018 recommendation from the USPSTF, which continues to recommend against PSA screening in men 70 years or older.^3,5,13,14 Diagnosing prostate cancer in this patient population may result in a net harm to patients, and further represent a misallocation of limited health care resources.

To our knowledge, this is the first study to comprehensively examine costs associated with and after prostate cancer diagnosis using direct measurement and microcosting methodology. An alternative approach estimates costs associated with diagnosis by comparing the health care costs incurred by cancer patients with costs incurred by individuals without cancer,^22,23 but this approach assumes that the 2 populations have comparable health care costs outside of the cancer in question. In contrast, the microcosting approach of the present study did not require a comparison sample of noncancer patients and additionally enabled detailed calculations of costs attributable to different categories of care. A prior study examined Medicare cost implications of prostate cancer screening, which included costs of PSA testing and prostate cancer diagnosis.²⁴ That study found annual expenditures for patients aged 66 to 99 years related to screening to be $447 million in 2009 US dollars, but this did not include costs of treatment and subsequent follow-up, which we found to represent the majority of spending related to prostate cancer detection. Overall, our findings were consistent with this prior study. Furthermore, our study and the microcosting methodology revealed additional insights that warrant further discussion.

First, with the continued controversy regarding prostate cancer screening, the present study provides insight on an important but understudied piece of this debate—cost. We used age 70 years as the primary analysis based on American Urological Association guidelines,³ but we also provide additional subgroup analyses to inform this debate. For example, detecting prostate cancers in men older than 75 years—per the US Social Security Index, this is the age at which men have a median life expectancy of 10 years—costs Medicare $601 million over 3 years for each annual cohort. Age 75 years is also cited in a Choosing Wisely recommendation for cessation of PSA testing.²⁵ Furthermore, the annual Medicare cost was $230 million for detecting prostate cancers in men older than 75 years who have a worse than median comorbidity score. These men have less than a 10-year life expectancy and are the least likely to benefit from prostate cancer screening; they are more likely to die from a comorbid illness than from prostate cancer.¹⁷ Randomized trials that have assessed the potential benefits of prostate cancer screening^26,27 and treatment (radical prostatectomy vs observation)^9,28 have consistently shown minimal to no overall survival benefit within 10 years of screening and treatment.

Second, with increasing recognition of the issue of overdiagnosis of prostate cancer, population-level incidence and treatment have decreased in recent years. Comparing SEER age-specific rates between the 2007-2011 and 2009-2013 periods, all age groups experienced a decline in incidence of prostate cancer.^29,30 The age groups with the largest declines were ages 70 to 74 years (a decrease of 112 men per 100 000) and 75 to 79 years (a decrease of 120 men per 100 000). If we were to scale up our per-person cost to the national level using this more recent incidence data, the total 3-year costs to Medicare for diagnosing prostate cancer in men 70 years or older would be $1 billion, a decrease of approximately $200 million relative to the earlier incidence rates.

Third, active surveillance has become a standard option for men with Gleason scores of 6 or lower.³¹ Borza et al³² reported that prostate cancer treatment rates have declined in recent years by 42%. Based on data from the present study, treatment costs were about 80% of the overall average cost per man 70 years or older. Combining these facts, we estimate that the decrease in overtreatment of prostate cancer would be associated with a 34% decrease (42% of 80%) in Medicare spending, totaling $413 million over 3 years for each annual cohort. This may be a conservative estimate of savings, as recent guidelines have supported use of active surveillance in select patients with a Gleason score of 7 as well.¹⁴ While we do not expect the adoption of active surveillance to be 100%, our findings provide insight on potential cost savings to Medicare relative to the proportion of patients who choose active surveillance.

With rising health care costs, it is imperative for research to identify areas of lower value health care services. Cessation of aggressive treatment of low-risk prostate cancer among elderly patients represents an opportunity to benefit the patient—by preventing harm from treatments that are unlikely to lead to improved survival—while also providing large annual cost savings to Medicare. While the debate regarding prostate cancer screening has focused largely on the overdetection and overtreatment of low-risk cancers,³³ it is also well established that high-risk cancers are life-threatening, and treatment can improve survival.^34,35 Thus, screening men with at least a 10-year estimated life expectancy and selectively treating those detected with high-risk cancers may be one approach which eliminates potential health care waste while allowing men who develop aggressive cancers to be detected and receive treatment. Therefore, we agree with published screening and treatment guidelines consistent with this approach.^3,12,13 Admittedly, accurately estimating life expectancy is difficult, but this is an issue that applies to many screening recommendations for men and women.

Strengths and Limitations

Strengths of this study include analyzing a population-based patient cohort, as well as using claims data for cost calculations that are broken down into distinct phases of care. Our results using Medicare data are timely, given the ongoing discussions regarding screening and treatment for prostate cancer, and have direct policy implications. There are also several potential limitations of this study. Our study did not include men with managed care plans, and it is possible that there is a lower incidence of prostate cancer diagnosis and treatment among men in managed care plans, where Healthcare Effectiveness Data and Information Set (HEDIS) measures may be used to minimize PSA testing in older men. Also, we do not know if the cancer cases analyzed in this study were detected through screening or after development of symptoms. However, the majority of US prostate cancer cases are detected through screening, and we excluded from analysis patients who were diagnosed with metastatic (M1) disease, which is the group most likely to present with symptoms. Furthermore, we analyzed costs within 3 years after prostate cancer detection; with longer follow-up, it is likely that more patient treatment and/or morbidity management will be observed, which would further increase costs in this patient population. As an aging population contributes to US health care spending growth, reducing provision of low-value services among elderly patients will remain an important aspect of containing health care spending.³⁶ We specifically designed this study to provide a conservative estimate of costs associated with and after prostate cancer detection. Additional studies are needed to examine the extent of financial burdens experienced by patients as a result of low-value screening and treatment practices.

Conclusions

Diagnosing men 70 years or older with prostate cancer each year is associated with $1.2 billion in Medicare costs, but there are substantial cost savings from active surveillance (compared to aggressive treatment) for men with Gleason scores of 6 or lower. Continued efforts to minimize overdiagnosis and overtreatment of prostate cancer will have a significant impact on reducing waste of limited health care resources.

Supplement.

eTable. Diagnoses and procedure codes