Healthcare Utilization and Costs Associated with Skeletal-Related Events in Prostate Cancer Patients with Bone Metastases

ORIGINAL ARTICLE Healthcare utilization and costs associated with skeletal-related events in prostate cancer patients with bone metastases

M Hagiwara1, TE Delea1, MW Saville2 and K Chung2

BACKGROUND: Patients with bone metastases secondary to prostate cancer are predisposed to skeletal-related events (SREs), including spinal cord compression, pathological fracture, surgery to bone and radiotherapy to bone. The objective of this study was to document current patterns of healthcare utilization and costs of SREs in patients with prostate cancer and bone metastases. METHODS: This was a retrospective, observational study using the Thomson MedStat MarketScan Commercial Claims and Encounters database from September 2002 to June 2011. Study subjects included all persons with claims for prostate cancer and for bone metastases, and one or more claims for an SRE. Unique SRE episodes were identiﬁed based on a gap of at least 90 days without an SRE claim, and classiﬁed by treatment setting (inpatient or outpatient) and SRE type (spinal cord compression, pathological fracture, surgery to bone or radiotherapy). RESULTS: Of 3919 patients with prostate cancer and bone metastases, 2090 (53%) had one or more SRE episodes. Among 1237 patients who met all other criteria, there were 1623 SRE episodes over a mean (s.d.) follow-up of 16.1 (12.9) months. The percent of episodes that required inpatient treatment ranged from 14% (radiotherapy) to 82% (surgery to bone). On average, inpatient episodes with surgery to bone (n ¼ 36 episodes) were most costly (mean (s.e.) $88 838 ($11 830)/episode), whereas outpatient episodes with surgery to bone (n ¼ 8 episodes) were least costly (mean (s.e.) $4749 ($1690)/episode). Of the total SRE costs (mean (s.e.) $20 984 ($951)/episode), 41% were attributable to outpatient radiotherapy (n ¼ 1169 episodes), 23% to inpatient radiotherapy (n ¼ 184 episodes), and 19% to inpatient treatment of pathological fractures (n ¼ 101 episodes). CONCLUSIONS: In patients with prostate cancer and bone metastases, SREs are associated with high costs and hospitalizations.

Prostate Cancer and Prostatic Disease (2013) 16, 23–27; doi:10.1038/pcan.2012.42; published online 13 November 2012 Keywords: skeletal-related events; prostatic neoplasms; metastases; healthcare costs

INTRODUCTION reported a mean cost of B$12 500 for treatment of SREs in the Prostate cancer is one of the most common cancers in the United year following SRE diagnosis. Barlev et al.10 recently examined the States, with an estimated 240 890 new cases and 33 720 deaths in costs of hospitalizations for SRE in patients with bone metastasis 2011, and it is the second leading cause of cancer death in men secondary to prostate or breast cancer or with multiple myeloma. behind lung cancer.1 The skeleton is the most common site of These authors reported that among patients with prostate cancer, metastasis in men with advanced prostate cancer.2,3 Median the mean health plan payment per hospital admission was survival from diagnosis of bone metastases in men with prostate $59 788 for spinal cord compression, $22 390 for pathological cancer is B3 years.2,4,5 In men with bone-metastases of castration- fracture and $42 094 for surgery to the bone. Although these resistant prostate cancer who have failed at least one hormonal studies provide useful information on the costs of SREs in patients therapy, median survival is B20 months.6 Patients with bone with bone metastases of solid tumors, they have some limitations. metastases of prostate cancer are at risk of a variety of skeletal- Although Barlev et al.10 examined the costs of SREs treated in the related events (SREs), such as pathological fractures, bone inpatient setting, little is known regarding the relative magnitude instability requiring surgery to the bone, radiation to the bone of the costs of SREs treated on both an inpatient and an to control local tumor burden and manage pain, and spinal cord outpatient basis. Also, none of these studies have examined the compression. These events can be associated with considerable costs of episodes of care for SREs. As deﬁned in the literature, an morbidity, impaired health-related quality of life, reduced survival episode of care (EOC) represents the period initiated by patient and increased healthcare costs.7–9 presentation with a diagnosis of a clinical condition and Numerous studies have examined the impact of SREs on costs concluded when the condition is resolved.19 EOC-based analyses in patients with bone metastases secondary to solid tumors.10–18 have been used extensively to assess the costs of treatment of In an analysis of patients with prostate cancer, McKiernan et al.17 diseases such as diabetic foot ulcers,20 abdominal pain21 and low estimated the expected costs of care directly attributable to SREs back pain.22 As many patients experience more than one SRE, to be $9783 per patient. After matching on baseline EOC-based cost analysis can be useful in understanding the characteristics, expected total medical-care costs were $20 484 economic impact of different types of SREs, and in informing greater in patients with SREs vs those without. Lage et al.14 estimates of the potential cost savings that might accrue as a evaluated the costs of SREs in patients with prostate cancer and result of interventions that may prevent them.

1Policy Analysis Inc. (PAI), Brookline, MA, USA and 2Amgen Global Health Economics, Thousand Oaks, CA, USA. Correspondence: Dr M Hagiwara, Policy Analysis Inc. (PAI), Four Davis Court, Brookline, MA 02445 USA. E-mail: [email protected] Received 1 June 2012; revised 17 September 2012; accepted 17 September 2012; published online 13 November 2012 SRE costs in prostate cancer with bone mets M Hagiwara et al 24 MATERIALS AND METHODS nursing facility care, physician office visits, hospital outpatient visits, Study design and data source laboratory tests and other care. In counting these services, multiple claims occurring on the same day were counted once, except for outpatient A retrospective observational longitudinal cohort design was used. Data prescriptions. Also tallied were the costs of inpatient hospitalizations, for this study were from the Thomson MedStat MarketScan Commercial Claims and Encounters database. This database contains information on emergency department visits, home healthcare visits, hospice care, skilled the US health insurance claims of employees of large, self-insured corpora- nursing facility care, physician office visits, hospital outpatient visits, tions and their dependents, along with a few commercial health plans. The laboratory tests and other healthcare services. Costs were approximated database is fully de-identified and compliant with the Health Insurance based on paid amounts (total gross payment after applying pricing guidelines such as fee schedules and discounts, but before applying Portability and Accountability Act of 1996 (HIPAA). Accordingly, institu- deductibles, coinsurance, copayments and coordination of benefit). tional review board approval was not necessary. Data used in this study span the period from 1 September 2002 to 30 June 2011 (‘study period’). Identification of SREs Patient selection For each patient, the follow-up period was scanned to identify all medical Patients were included in the analysis if they met all of the following claims indicative of an SRE (spinal cord compression, pathological fracture, criteria: confirmed diagnosis of bone metastases (International Classifica- surgery to bone or radiotherapy). Diagnosis and procedure codes used to tion of Disease, 9th edition, Clinical Modification (ICD-9-CM) diagnosis identify SREs are listed in Supplementary Appendix B. The codes used to codes 170.XX or 198.5X); confirmed diagnosis of prostate cancer (185.XX) identify SREs were based on coding algorithms used in prior studies,11–13,17 before bone metastases diagnosis; and one or more SRE episodes (detailed supplemented with discussion with clinical and coding experts. When algorithm is found in section below, titled as ‘Identification of SREs’). For there was uncertainty with respect to the inclusion of codes for the the purpose of this study, a patient was considered to have a ‘confirmed’ identification of SREs, the coding algorithm was designed to be as specific diagnosis if the patient had one or more inpatient facility claims with the as possible to avoid the inclusion of events that were not actually SREs. diagnosis code or codes of interest listed as the primary diagnosis on the Regarding the identification of pathological fractures, there are two claim, or two or more medical claims on different days and no more than different sets of ICD-9-CM codes for fractures. Pathological or spontaneous 6 months apart, with the same diagnosis code or codes of interest listed in fractures are classified in the musculoskeletal and connective tissue any diagnosis field on the claims. This approach of identifying confirmed chapter of ICD-9-CM (ICD-9-CM 733.1X), whereas traumatic fractures are diagnoses was used to improve the specificity of diagnoses based in ICD- classified in the injury and poisonings chapter of ICD-9-CM (ICD-9-CM 9-CM diagnosis codes on health insurance claims. These codes may lack 800–824). Because pathological or spontaneous fractures may be specificity if listed on claims for services provided to rule out certain associated with trivial trauma such as a minor fall, many such fractures diseases or as working diagnoses while clinical diagnoses are confirmed by may be incorrectly coded as traumatic fractures. For the purpose of this pathological or other means. study, pathological fracture was therefore identified based on confirmed The ‘index date’ was defined as the date with the first confirmed diagnoses of pathological fracture (ICD-9-CM 733.1X) or a confirmed diagnosis of bone metastases. The 6-month period before the index date diagnosis of traumatic fracture (ICD-9-CM 800–824) (see ‘Patient selection’ was designated the ‘pre-index period’; the period from the index date to section for the definition of confirmed diagnosis). Claims with codes the end of continuous enrollment was designated the ‘follow-up period’. for traumatic fracture (ICD-9-CM 800–824) were only included, however, Patients o18 years of age on the index date, with gender specified as if there were no concurrent claims with codes suggesting accident female, o6 months of continuous enrollment before the index date, a (excluding accident codes for ‘falls on the same level’) or trauma. diagnosis of primary cancer other than prostate cancer before the index ‘Concurrent claims’ were defined as claims occurring during the 2-week date, or with invalid cost data were excluded. Because diagnoses codes for period ending with the claim for the fracture. When identifying a primary cancer may sometimes be used for metastatic disease, patients confirmed diagnosis of fracture, all diagnosis codes for fractures to the were assumed to have other primary cancer only if they did not also have a same site were considered as confirmatory. When identifying surgery to corresponding claim for metastasis to the same site. For example, patients bone, claims with concurrent claims suggesting accident (excluding falls with claims with a diagnosis of primary liver neoplasm (ICD-9-CM 155.XX) on the same level) or significant trauma also were excluded. Because there as well as claims with diagnosis of metastasis to the liver (197.0X, 197.1X, are no body site modifiers for procedure codes for radiotherapy, claims for 197.2X, and 197.3X) were not considered to have ‘other primary cancer’. radiotherapy procedures were considered only if there was a concurrent Conversely, those with claims with diagnoses of primary liver neoplasm but claim with a diagnosis code for bone metastasis, spinal cord compression, without a claim for metastasis to the liver were assumed to have ‘other bone pain (ICD-9-CM 733.90) or pathological fracture. primary cancer’. Codes for identification of primary and secondary For each patient with at least one SRE claim during the follow-up period, malignancies by site are listed in Supplementary Appendix A. SRE episodes were identified using a gap of 90 days or more without an outpatient claim or an inpatient admission for an SRE. Gaps were defined based on the number of days without SRE claims starting with the last SRE Patient characteristics encounter (Figure 1). The choice of the 90-day gap was based on the For each patient, baseline demographic characteristics and comorbidities observation that the mean daily cost of SRE-related care by time since the were assessed as of the index date, including calendar year, age, region, first SRE claim declined rapidly between 1 and 3 months after the first SRE plan type and Deyo’s version of the Charlson Comorbidity Index.23 Pre- claim and leveled off shortly thereafter. index healthcare utilization and costs also were measured including: the Each SRE episode was then classified into mutually exclusive categories. number of outpatient pharmacy claims or procedures for chemotherapy, SREs include both diagnoses (spinal cord compression and pathological systemic corticosteroids and intravenous bisphosphonates; the number of fracture) as well as procedures (surgery to bone and radiotherapy). These outpatient pharmacy claims (that is, prescriptions) for opioids and oral procedures may be used in the treatment of the diagnosed condition(s) bisphosphonates; and the number of claims for inpatient hospitalizations, (for example, surgery to bone may be used for a pathological fracture). emergency department visits, home healthcare visits, hospice care, skilled Spinal cord compression can present as an oncologic emergency,

Episode #1 Episode #2 Episode #3

70 Day Gap 90 Day Gap 90 Day Gap SCC RT RT SCC PF RT SB RT

Time, Days

SCC=Spinal cord compression, PF=Pathological fracture, SB=Surgery to bone, RT=Radiotherapy Figure 1. Example for identifying SRE episodes with 90-day gap. Gaps were deﬁned based on the number of days without SRE claims starting withthelastSREclaim.PF,pathologicalfracture;RT,radiotherapy;SB,surgerytobone;SCC,spinalcordcompression;SRE,skeletal-relatedevent.

Prostate Cancer and Prostatic Disease (2013), 23 – 27 & 2013 Macmillan Publishers Limited SRE costs in prostate cancer with bone mets M Hagiwara et al 25 potentially leading to severe and permanent neurological damage including paresis or paralysis, loss of control of bladder and/or bowel Table 2. Patient characteristics at baseline (N ¼ 1237) function, loss of independence, reduced quality of life, and reduced survival.24–27 Accordingly, in defining mutually exclusive categories, we Characteristic N (%)/Mean (s.d.) used a hierarchy in which the diagnoses of spinal cord compression and Index year, N (%) pathological fracture were at the top, and surgery to bone and 2003 67 (5.4) radiotherapy were at the bottom. This yielded the following hierarchy 2004 98 (7.9) for identifying mutually exclusive categories of SRE episodes based on 2005 113 (9.1) evidence of confirmed diagnoses or procedures: spinal cord compression; 2006 133 (10.8) pathological fracture without spinal cord compression; surgery to bone 2007 196 (15.8) without spinal cord compression and without pathological fracture; and 2008 217 (17.5) radiotherapy without any other types of SRE claims. Episodes were further 2009 214 (17.3) classified as ‘inpatient’ or ‘outpatient’ based on the presence during the 2010 159 (12.9) episode of one or more inpatient hospital claims with any SRE. 2011 40 (3.2) Age, years, mean (s.d.) 58.8 (4.6) Outcome measures and analyses For each SRE episode, all the claims during the episode were scanned and Region, US, N (%) classified as SRE-related or non-SRE-related according to ICD-9-CM South 513 (41.5) diagnosis codes and procedure codes listed in Supplementary Appendix North Central 352 (28.5) B. For each SRE episode, all the SRE-related costs (payments) were tallied West 195 (15.8) and adjusted to calendar year 2010 US dollars.28 The number and Northeast 152 (12.3) percentages of SRE episodes by type of SRE, treatment setting, and SRE Unknown 25 (2.0) claims identified during the episode, and the mean (s.d.) SRE-related cost Plan type, N (%) per episode by type of SRE and treatment setting were calculated. All PPO 758 (61.3) analyses were conducted using SAS Proprietary Software, Release 9.2.3 Comprehensive 114 (9.2) (SAS Institute, Cary, NC, USA). HMO 156 (12.6) Other 209 (16.9) Charlson index, mean (s.d.)a 3.5 (2.6) RESULTS Healthcare utilization, N (%) Of 3919 patients with bone metastases and prostate cancer Chemotherapy 37 (3.0) identified in the dataset, 2090 patients (53%) had at least one SRE. Systemic corticosteroid 304 (24.6) Of these, 1237 patients met all other criteria (Table 1). IV bisphosphonate 2 (0.2) Mean (s.d.) age was 58.8 (4.6) years, 42% of patients resided in Opioid 603 (48.7) the South, followed by North Central (29%), West (16%) and Oral bisphosphonate 28 (2.3) Northeast (12%), and the majority had preferred provider Hospitalization 156 (12.6) organization (PPO) (61%), followed by health maintenance Emergency department visit 281 (22.7) organization (HMO) (13%) plan types (Table 2). The mean (s.d.) Home healthcare 124 (10.0) Hospice 2 (0.2) Charlson index during the pre-index period was 3.5 (2.6), Skilled nursing facility 4 (0.3) consistent with advanced prostate cancer. The mean (s.d.) Physician office visit 1191 (96.3) duration of follow-up was 16.1 (12.9) months. Hospital outpatient visit 1075 (86.9) There were 1623 SRE episodes identified among 1237 patients. Lab visits 462 (37.3) Among spinal cord compression episodes (n ¼ 40), B75% of the Other visit 188 (15.2) episodes included claims for radiotherapy, while 25% included Healthcare claims, number, mean (s.d.)a,b claims for pathological fracture (Table 3). Among pathological Chemotherapy 0.08 (0.57) fracture episodes (n ¼ 186), B51% of the e4pisodes included Systemic corticosteroid 0.59 (1.41) claims for radiotherapy, while 26% included claims for surgery to IV bisphosphonate 0.00 (0.09) bone, and 38% included pathological fracture claims alone. Opioid 1.86 (3.14) Among surgery to bone episodes (n ¼ 44), 52% included claims Oral bisphosphonate 0.06 (0.48) for radiotherapy. Almost all episodes with claims for radiotherapy Hospitalization 0.16 (0.47) (90%) were classified as radiotherapy episodes. Emergency department visit 0.40 (1.18) Home healthcare 0.31 (1.38) Skilled nursing facility 0.02 (0.42) Physician office visit 10.7 (9.8) Hospital outpatient visit 5.4 (6.8) Table 1. Selection of study subjects Other visit 0.43 (2.48) Criteria N (%) Healthcare costs, $, mean (s.d.)a Hospitalization 2668 (12 013) Confirmed diagnosis of bone metastases, and 2090 (100.0) Emergency department visit 334 (1601) confirmed diagnosis of prostate cancer before bone Home healthcare 119 (790) metastases diagnosis, and one or more SRE claims Hospice 7 (196) during the follow-up period Skilled nursing facility 63 (1332) Physician office visit 5826 (11 515) Exclusions Hospital outpatient visit 7471 (14 837) Less than 6 months of continuous enrollment 625 (29.9) Lab visits 141 (461) before index date Other visit 431 (5962) More than one type of primary cancer before 265 (12.7) Duration of follow-up, months, mean (s.d.) 16.1 (12.9) index date Other 10 (0.5) NOTE: Costs are in 2010 US dollars aCalculated over 6 months before index date. Total Excluded 853 (40.8) b Remaining 1237 (59.2) Values are numbers of claims. Except for medications, claims on the same day are counted once. Abbreviation: SRE, skeletal-related event. HMO, health maintenance organization; PPO, preferred provider organization.

& 2013 Macmillan Publishers Limited Prostate Cancer and Prostatic Disease (2013), 23 – 27 SRE costs in prostate cancer with bone mets M Hagiwara et al 26 Twenty one percent of all SREs required inpatient treatment. DISCUSSION Only 14% of radiotherapy episodes required inpatient treatment, This was a retrospective, observational study using health whereas 82% of surgery to bone episodes required inpatient insurance claims data to document patterns of healthcare treatment. utilization and costs associated with SRE episodes in patients The mean (s.e.) cost per episode was $54 751 ($3526) for with prostate cancer and bone metastases. Results of this analysis inpatient episodes and $11 768 ($481) for outpatient episodes suggest that in patients with prostate cancer and bone (Figure 2). On average, inpatient surgery to bone episodes were metastases, SREs episodes are associated with high costs and most costly with mean (s.e.) cost per episode of $88 838 ($11 830), hospitalizations. followed by inpatient pathological fracture episodes with mean This study differs from prior studies of the costs of SREs in that (s.e.) cost per episode of $64 125 ($8089). Outpatient episodes an EOC approach was used for estimating the costs of SREs. EOC with surgery to bone were least costly with mean (s.e.) cost per analysis is appropriate for analyzing the economic burden of episode of $4749 ($1690). clinical events when their treatments involve a series of separate but related healthcare services. By using an EOC approach, a more Table 3. Number of SRE episodes by types of SRE claims identified complete picture of the costs of events can be obtained. Future within each episode among patients with prostate cancer and bone health insurance claims-based studies on the costs of SREs, as well metastases as potential other events such as adverse events of treatment of disease progression in cancer patients, should explore the use of SRE classification Inpatient Outpatient All Episodes EOC analysis. Use of and EOC approach, however, requires a rule for defining N (%) N (%) N (%) each episode. Our decision to use a 90-day gap to define SRE episodes was based on the observation that the mean daily costs SCC only 2 (7.4) 5 (38.5) 7 (17.5) of SRE-related care declined rapidly from 30 to 90 days post SRE, SCC and PF 0 (0.0) 0 (0.0) 0 (0.0) and leveled off shortly thereafter. Some prior clinical trials of bone SCC and SB 2 (7.4) 0 (0.0) 2 (5.0) modifying drugs have used a 21-day window to calculate SCC and RT 12 (44.4) 8 (61.5) 20 (50.0) cumulative incidence of SREs6 or skeletal morbidity rate.29 SCC and PF and SB 1 (3.7) 0 (0.0) 1 (2.5) However, the use of gaps rather than windows to identify SCC and PF and RT 6 (22.2) 0 (0.0) 6 (15.0) unique SRE episodes is consistent with traditional EOC analyses. SCC and SB and RT 1 (3.7) 0 (0.0) 1 (2.5) It is instructive to compare results of this study with those SCC and PF and SB 3 (11.1) 0 (0.0) 3 (7.5) 10 and RT reported by Barlev et al., which also used the MedStat All SCC 27 (100.0) 13 (100.0) 40 (100.0) MarketScan databases to evaluate the costs of inpatient SRE episodes in patients with bone metastasis secondary to prostate PF only 24 (23.8) 46 (54.1) 70 (37.6) or breast cancer. Although the mean cost per hospitalization for PF and SB 18 (17.8) 3 (3.5) 21 (11.3) spinal cord compression in prostate cancer patients reported by PF and RT 36 (35.6) 31 (36.5) 67 (36.0) Barlev et al.10 ($59 788) is similar to the mean cost per inpatient PF and SB and RT 23 (22.8) 5 (5.9) 28 (15.1) episode for spinal cord compression from our study ($54 458), the All PF 101 (100.0) 85 (100.0) 186 (100.0) mean costs per hospitalization for pathological fracture and surgery to bone reported by Barlev et al. ($22 390 and $42 094, SB only 14 (38.9) 7 (87.5) 21 (47.7) SB and RT 22 (61.1) 1 (12.5) 23 (52.3) respectively) are considerably lower than the mean costs per All SB 36 (100.0) 8 (100.0) 44 (100.0) inpatient episode for pathological fracture and surgery to bone reported here ($64 125 and $88 838, respectively). This difference RT only 184 (100.0) 1169 (100.0) 1353 (100.0) may be a consequence of a number of factors. First, inpatient episodes in this study include follow-up outpatient care, whereas Total 348 (100.0) 1275 (100.0) 1623 (100.0) Barlev et al.10 considered only the cost of hospitalization. Second, this study used wider sets of codes to identify pathological Abbreviations: PF, pathological fracture; RT, radiotherapy; SB, surgery to 10 bone; SCC, spinal cord compression; SRE, skeletal-related event. fracture and surgery to bone. Third, Barlev et al. used data from both commercial plans and Medicare. This study focused on

120 30 112.9 28.4

100 25 88.8 80 80.2 20 71.9 64.1 64.8 60 15 14.9 14.7 54.5 50.9 12.8 48.1 11.8 43.0 11.4 40 10.8 37.0 10 35.1 8.7 Episode($, thousands) Episode($, thousands) 8.0 Mean (95%CI) Cost per SRE Mean (95%CI) Cost per SRE 20 5 4.7

1.4 0.8 0 0 Spinal cord Pathological Surgery Radiotherapy Spinal cord Pathological Surgery Radiotherapy compression fracture to bone compression fracture to bone (N= 27) (N= 101) (N= 36) (N= 184) (N= 13) (N= 85) (N= 8) (N= 1169)

Type of SRE Type of SRE Figure 2. Mean (95% CI) cost of inpatient and outpatient SRE episodes among patients with prostate cancer and bone metastases by type of SRE (number of patients ¼ 1237): (a) Inpatient SRE episodes; and (b) outpatient SRE episodes. SRE, skeletal-related event.

Prostate Cancer and Prostatic Disease (2013), 23 – 27 & 2013 Macmillan Publishers Limited SRE costs in prostate cancer with bone mets M Hagiwara et al 27 commercial plans only because of the possibility that some care 7 Coleman RE. Metastatic bone disease: clinical features, pathophysiology and received by Medicare patients may not be captured in the treatment strategies. Cancer Treat Rev 2001; 27: 165–176. MedStat database. 8 Oefelein MG, Ricchiuti V, Conrad W, Resnick MI. Skeletal fractures negatively Limitations of this study should be noted. This study used correlate with overall survival in men with prostate cancer. J Urol 2002; 168: administrative claims data that may be subject to data coding 1005–1007. limitations and data entry error. Also, diagnostic and procedure 9 Saad F, Gleason DM, Murray R, Tchekmedyian S, Venner P, Lacombe L et al. Long- term efficacy of zoledronic acid for the prevention of skeletal complications in codes on claims may have limited sensitivity and specificity for patients with metastatic hormone-refractory prostate cancer. J Natl Cancer Inst clinical diagnoses. This may have affected the distribution of SREs 2004; 96: 879–882. if some diagnosis codes are more sensitive or specific than others. 10 Barlev A, Song X, Ivanov B, Setty V, Chung K. Payer costs for inpatient treatment of As noted above, this study used data from the MedStat pathologic fracture, surgery to bone, and spinal cord compression among commercial database, and the estimates reported herein may patients with multiple myeloma or bone metastasis secondary to prostate or not be generalizable to Medicare or other elderly patients. Further, breast cancer. J Manag Care Pharm 2010; 16: 693–702. the MarketScan commercial database is geographically diverse 11 Delea T, Langer C, McKiernan J, Liss M, Edelsberg J, Brandman J et al. The cost of but not entirely representative of the US population, and patients treatment of skeletal-related events in patients with bone metastases from lung from the database may not be representative of all commercially cancer. Oncology 2004; 67: 390–396. 12 Delea T, McKiernan J, Brandman J, Edelsberg J, Sung J, Raut M et al. Retrospective insured patients in the United States or other settings. Approxi- study of the effect of skeletal complications on total medical care costs in patients mately 13% of patients in the study sample were in HMO plans. As with bone metastases of breast cancer seen in typical clinical practice. J Support a consequence of capitation arrangements, the payment field for Oncol 2006; 4: 341–347. office visits for these patients may result in potential under- 13 Delea TE, McKiernan J, Brandman J, Edelsberg J, Sung J, Raut M et al. Impact of estimation of outpatient SRE costs. The extent of this bias is likely skeletal complications on total medical care costs among patients with bone to be small, however, as cancer patients are likely to receive metastases of lung cancer. J Thorac Oncol 2006; 1: 571–576. treatment from oncologists rather than primary care physicians, 14 Lage MJ, Barber BL, Harrison DJ, Jun S. The cost of treating skeletal-related events and the former are less likely to be in capitated payment in patients with prostate cancer. Am J Manag Care 2008; 14: 317–322. arrangements than the latter. Lastly, for some SRE episode 15 Krupski TL, Foley KA, Baser O, Long S, Macarios D, Litwin MS. Health care cost associated with prostate cancer, androgen deprivation therapy and bone categories (for example, inpatient and outpatient surgery to complications. J Urol 2007; 178(4 Pt 1): 1423–1428. bone), the number of events was small and our estimates may lack 16 Groot MT, Boeken Kruger CG, Pelger RC, Uyl-de Groot CA. Costs of prostate precision. cancer, metastatic to the bone, in the Netherlands. Eur Urol 2003; 43: 226–232. Despite these limitations, this study has important implications. 17 McKiernan JM, Delea TE, Brandman J, Sung J, Raut M, Oster G. Effects of skeletal First, results of this study suggest that SREs are associated with complications on total medical care costs in patients with bone metastases of high costs and hospitalizations among patients with prostate solid tumors. American Society of Clinical Oncology 41st Annual Meeting, May 15, cancer and bone metastases, and that these costs may be greater 2005 Orlando, FL (Poster). J Clin Oncol 2004 ASCO Annual Meeting Proceedings than estimated previously for some events. This study also 2004; 22: 6057. demonstrates the utility of EOC-based analyses for evaluating 18 Hagiwara M, Oglesby A, Chung K, Zilber S, Delea TE. The impact of bone metastases and skeletal-related events on healthcare costs in prostate cancer the costs of clinical events in retrospective studies using health patients receiving hormonal therapy. Community Oncol 2011; 8: 508–515. insurance claims data. 19 Hornbrook MC, Hurtado AV, Johnson RE. Health care episodes: definition, mea- surement and use. Med Care Rev 1985; 42: 163–218. 20 Mehta SS, Suzuki S, Glick HA, Schulman KA. Determining an episode of care using CONFLICT OF INTEREST claims data. Diabetic foot ulcer. Diabetes Care 1999; 22: 1110–1115. May Hagiwara, PhD and Thomas E Delea, MSIA are employees of Policy Analysis Inc. 21 Klinkman MS. Episodes of care for abdominal pain in a primary care practice. (PAI) and have received research funding and consulting fees from Amgen and Arch Fam Med 1996; 5: 279–285. Novartis. Karen Chung, PharmD, MS is an employee of and owns stock options in 22 Shekelle PG, Markovich M, Louie R. Comparing the costs between provider types Amgen. M Wayne Saville, MD was an employee of and owned stock options in Amgen. of episodes of back pain care. Spine (Phila Pa 1976) 1995; 20: 221–226; discussion 227. 23 Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ACKNOWLEDGEMENTS ICD-9-CM administrative databases. J Clin Epidemiol 1992; 45: 613–619. The paper was funded by Amgen. 24 Abrahm JL, Banffy MB, Harris MB. Spinal cord compression in patients with advanced metastatic cancer: ‘all I care about is walking and living my life’. JAMA 2008; 299: 937–946. REFERENCES 25 Conway R, Graham J, Kidd J, Levack P. What happens to people after malignant 1 American Cancer Society. Detailed guide: prostate cancer what are the key sta- cord compression? Survival, function, quality of life, emotional well-being tistics about prostate cancer? Available at http://www.cancer.org/Cancer/Prostate and place of care 1 month after diagnosis. Clin Oncol (R Coll Radiol) 2007; 19: Cancer/DetailedGuide/prostate-cancer-key-statistics. Accessed 31 January 2012. 56–62. 2 Coleman RE. Skeletal complications of malignancy. Cancer 1997; 80(8 Suppl): 26 Loblaw DA, Perry J, Chambers A, Laperriere NJ. Systematic review of the diagnosis 1588–1594. and management of malignant extradural spinal cord compression: the Cancer 3 Mundy GR. Metastasis to bone: causes, consequences and therapeutic opportu- Care Ontario Practice Guidelines Initiative’s Neuro-Oncology Disease Site Group. nities. Nat Rev Cancer 2002; 2: 584–593. J Clin Oncol 2005; 23: 2028–2037. 4 Ohmori K, Matsui H, Yasuda T, Kanamori M, Yudoh K, Seto H et al. Evaluation of 27 Loblaw DA, Laperriere NJ. Emergency treatment of malignant extradural the prognosis of cancer patients with metastatic bone tumors based on serial spinal cord compression: an evidence-based guideline. J Clin Oncol 1998; 16: bone scintigrams. Jpn J Clin Oncol 1997; 27: 263–267. 1613–1624. 5 Soloway MS, Hardeman SW, Hickey D, Raymond J, Todd B, Soloway S et al. 28 Bureau of Labor Statistics. Consumer Price Index. Medical Care, US City Average. Stratification of patients with metastatic prostate cancer based on extent of Series Id: CUUR0000SAM, CUUS0000SAM. US Department of Labor: Washington, DC, disease on initial bone scan. Cancer 1988; 61: 195–202. 2008, Available fromURL http://www.bls.gov/cpi/home.htm. 6 Fizazi K, Carducci M, Smith M, Damiao R, Brown J, Karsh L et al. Denosumab 29 Saad F, Gleason DM, Murray R, Tchekmedyian S, Venner P, Lacombe L et al. versus zoledronic acid for treatment of bone metastases in men with castration- A randomized, placebo-controlled trial of zoledronic acid in patients with resistant prostate cancer: a randomised, double-blind study. Lancet 2011; 377: hormone-refractory metastatic prostate carcinoma. J Natl Cancer Inst 2002; 94: 813–822. 1458–1468.

Supplementary Information accompanies the paper on the Prostate Cancer and Prostatic Diseases website (http://www.nature.com/pcan)

& 2013 Macmillan Publishers Limited Prostate Cancer and Prostatic Disease (2013), 23 – 27