1513
Centralization of Care for Patients With Advanced-Stage Ovarian Cancer A Cost-effectiveness Analysis
Robert E. Bristow, MD1,2 Antonio Santillan, MD1 Teresa P. Diaz-Montes, MD, MPH1 Ginger J. Gardner, MD1 Robert L. Giuntoli II, MD1 Benjamin C. Meisner, MHS, MBA1 Kevin D. Frick, PhD3 Deborah K. Armstrong, MD1,2
BACKGROUND. The objective of this study was to evaluate the cost-effectiveness of centralized referral of patients with advanced-stage epithelial ovarian cancer who underwent primary cytoreductive surgery and adjuvant chemotherapy. METHODS. A decision-analysis model was used to compare 2 referral strategies for patients with advanced-stage ovarian cancer: 1) referral to an expert center, with a rate of optimal primary cytoreduction of 75% and utilization of combined intraperitoneal and intravenous adjuvant chemotherapy, and 2) referral to a less experienced center, with a rate of optimal primary cytoreduction of 25% and adjuvant treatment that consisted predominantly of intravenous chemotherapy
1
The Kelly Gynecologic Oncology Service, Department of Obstetrics and Gynecology, The Sidney Kimmel Comprehensive Cancer Center, The Johns Hopkins Medical Institutions, Baltimore, Maryland. 2
Department of Oncology, The Sidney Kimmel Comprehensive Cancer Center, The Johns Hopkins Medical Institutions, Baltimore, Maryland. 3
Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, The Johns Hopkins Medical Institutions, Baltimore, Maryland.
alone. The cost-effectiveness of each strategy was evaluated from the perspective of society.
RESULTS. A cost-effectiveness analysis revealed that the strategy of expert center referral had an overall cost per patient of $50,652 and had an effectiveness of 5.12 quality-adjusted life years (QALYs). The strategy of referral to a less experienced center carried an overall cost of $39,957 and had an effectiveness of 2.33 QALYs. The expert center strategy was associated with an additional 2.78 QALYs at an incremental cost of $10,695 but was more cost-effective, with a cost-effective ratio of $9893 per QALY compared with $17,149 per QALY for the less experienced center referral strategy. Sensitivity analyses and a Monte Carlo simulation confirmed the robustness of the model.
CONCLUSIONS. According to results from the decision-analysis model, centralized referral of patients with ovarian cancer to an expert center was a cost-effective healthcare strategy and represents a paradigm for quality cancer care, delivering superior patient outcomes at an economically affordable cost. Increased efforts to align current patterns of care with a universal strategy of centralized expert referral are warranted. Cancer 2007;109:1513–22. 2007 American Cancer Society.
KEYWORDS: centralized care, cost-effectiveness, ovarian cancer, quality-adjusted life years.
Supported by the Pam McDonald Ovarian Cancer Research Program. Address for reprints: Robert E. Bristow, MD, The Kelly Gynecologic Oncology Service, Department of Obstetrics and Gynecology, 600 North Wolfe Street, Phipps 281, Baltimore, MD 21287; Fax: (410) 614-8718; E-mail:
[email protected] Received November 1, 2006; revision received December 9, 2006; accepted January 2, 2007.
ª 2007 American Cancer Society
W
orldwide, 204,449 women are newly diagnosed with ovarian cancer each year, and there are an estimated 124,860 diseaserelated deaths.1 Approximately 65% of patients will be diagnosed with International Federation of Gynecology and Obstetrics (FIGO) stage III (T3N0/N1M0) or stage IV (any T, any N, M1) disease.2 For this group, the most important clinician-driven prognostic factors are the extent of residual disease after primary cytoreductive surgery and the administration of adjuvant platinum-based chemotherapy.3,4 The association between initial surgical care by a gynecologic oncologist and a higher likelihood of optimal primary cytoreductive surgery, close adherence to recommended chemotherapy treatment
DOI 10.1002/cncr.22561 Published online 12 March 2007 in Wiley InterScience (www.interscience.wiley.com).
1514
CANCER
April 15, 2007 / Volume 109 / Number 8
FIGURE
1.
Decision-analysis
model for the referral of patients with advanced-stage ovarian cancer to either an expert center or a less experienced center. IP indicates intraperitoneal; IV, intravenous.
guidelines, and resultant superior clinical outcomes has been well documented.5–10 In addition to subspecialty surgical care, recent attention has focused on centralized referral of patients with ovarian cancer to high-volume expert centers that offer a multidisciplinary expertise to the global treatment program.5,9,11,12 However, to date, the clinical benefits and associated costs of centralized care have not been quantified precisely. With current concerns over the escalating economic burden of healthcare and a concomitant desire to optimize clinical outcomes, such information would be of value to public policy and healthcare administrators as well as third-party payers. Therefore, the objective of the current study was to simulate the projected clinical benefits and economic costs associated with referral to an expert center with a high rate of optimal primary surgical cytoreduction and utilization of intraperitoneal (IP) chemotherapy measured against referral to a less experienced center with a comparably low rate of optimal surgical resection and treatment with intravenous (IV) chemotherapy alone in the majority of patients.
MATERIALS AND METHODS Overall Model A decision-analysis model was created to evaluate 2 management strategies for patients with advancedstage epithelial ovarian cancer from primary cytoreductive surgery through completion of front-line chemotherapy (Fig. 1). The 2 strategies were: 1) referral to an expert center, with achievement of optimal residual disease (1 cm) in 75% of patients and administration of initial chemotherapy using a combination of IP cisplatin and IP and IV paclitaxel (IP/
IV) in the majority (80%) of optimally cytoreduced patients, and 2) referral to a less experienced center, with a 25% rate of optimal primary cytoreduction and with adjuvant chemotherapy consisting predominantly (90%) of IV carboplatin and IV paclitaxel (IV/IV). Within each strategy, it was assumed that patients had undergone an attempt at maximal cytoreductive surgery followed by a planned 6 cycles of chemotherapy. For patients who underwent optimal cytoreductive surgery, chemotherapy could consist of either the IP/IV regimen or the IV/IV regimen. For patients who were left with suboptimal residual disease, chemotherapy consisted of the IV/IV regimen. The model and analyses conformed to the 10 basic principles that should be incorporated into a cost-effectiveness analysis identified by the Panel on Cost-Effectiveness in Health and Medicine convened by the U.S. Public Health Service.13 The model was generated from the perspective of society and incorporated all relevant cost associated with primary management of advanced-stage epithelial ovarian cancer.
Model Estimates: Clinical Assumptions Clinical assumptions were obtained from a review of the published English-language literature. Whenever possible, data from Phase III trials was utilized; however, when such data were unavailable, clinical assumptions were generated from Phase II trials, case-control studies, retrospective case series, and expert opinion (Table 1). For the expert center strategy, the rate of optimal primary cytoreduction (residual disease measuring 1 cm) was estimated at 75%.6,14–23 Among the patients who underwent optimal resection, the rate of
Centralization of Ovarian Cancer Care/Bristow et al.
1515
TABLE 1 Model Estimates: Clinical Assumptions Clinical parameter Expert center Rate of optimal primary cytoreduction Rate of complete cytoreduction* Proportion receiving IP/IV chemotherapy* Operative mortality of primary surgery Less experienced center Rate of optimal primary cytoreduction Rate of complete cytoreduction* Proportion receiving IP/IV chemotherapy* Operative mortality of primary surgery IP/IV chemotherapyy Median survival: No gross residual, mo Undiscounted Discounted{ Median survival: Residual 0.1–1.0 cm, mo Undiscounted Discounted{ Quality of well-being index: Mo 1–5 Quality of well-being index: Mo 6–12 Incidence of hospitalization for fever Incidence of hospitalization for infection Incidence of hospitalization for gastrointestinal toxicity Incidence of hospitalization for metabolic events Incidence of hospitalization for thrombocytopenia Incidence of hospitalization for renal or genitourinary event IV/IV chemotherapy§ Median survival: No gross residual, mo Undiscounted Discounted{ Median survival: Residual 0.1–1.0 cm, mo Undiscounted Discounted{ Median survival: Residual >1.0 cm, mo Undiscounted Discounted{ Quality of well-being index: Mo 1–5 Quality of well-being index: Mo 6–12 Incidence of hospitalization for fever Incidence of hospitalization for infection Incidence of hospitalization for gastrointestinal toxicity Incidence of hospitalization for metabolic events Incidence of hospitalization for thrombocytopenia Incidence of hospitalization for renal or genitourinary event
Model estimate, %
Reference(s)
75 50 80 3
614–23 2224–28
25 10 10 3
5729–34 5729–34
113.3 84.3
35
52.6 46.7 0.66 0.71 9 8 12 7 3 2
35
78.2 65.6
35
39.1 35.8
35
26.6 25.1 0.73 0.76 3 3 6 2.0 1.0 0.5
36
35 35 35 35 35 35 35 35
35 35 3536 3536 35 35 35 35
IP indicates intraperitoneal; IV, intravenous. * Among patients with residual disease that measured 1 cm. y IP cisplatin plus IP and IV paclitaxel. { Survival time was discounted at 3% per year. § IV carboplatin plus IV paclitaxel.
complete cytoreduction (no gross residual disease) was estimated at 50%.22,24–28 For patients who received treatment at a less experienced center, the rate of optimal primary cytoreduction was estimated at 25%, and 10% of patients who underwent optimal resection were left with no gross residual disease.5,7,29–34 An operative mortality rate of 3% was
derived from a review of 200 consecutive patients with ovarian cancer from the Maryland Health Services Cost Review Commission (HSCRC) database who underwent hysterectomy with salpingo-oophorectomy (All Patient Diagnostic-Related Groups [APDRG] 684, 686, 687, and 688) and cytoreduction (APDRG 544) between fiscal years (FY) 2004 and 2006.
1516
CANCER
April 15, 2007 / Volume 109 / Number 8
Model estimates for the proportion of optimally resected patients who received a contemporary IP/IV chemotherapy regimen were 80% for patients who attended an expert center based on the current utilization rate at The Johns Hopkins Medical Institutions (JHMI), and 10% for patients who attended a less experienced center, based on HSCRC data for the administration of IP chemotherapy for ovarian cancer at non-JHMI hospitals during FY 2006. The treatment completion rate for IP/IV chemotherapy was modeled after the experimental arm of Gynecologic Oncology Group (GOG) protocol 172: 6 cycles (42%), 5 cycles (5%), 4 cycles (5%), 3 cycles (7%), 2 cycles (15%), 1 cycle (19%), and 0 cycles (8%).35 The balance of incomplete IP treatment cycles was substituted with cost estimates for the IV treatment regimen up to a total of 82.9% of patients completing 6 cycles of adjuvant chemotherapy.35 For optimally resected patients who received the IV/IV chemotherapy regimen, the treatment completion rate was modeled after the control arm of GOG protocol 172: 6 cycles (90%), 5 cycles (2%), 4 cycles (2%), 3 cycles (2%), 2 cycles (2%), and 1 cycle (2%).35 The treatment completion rate for patients with suboptimal residual disease who received IV/IV chemotherapy was modeled after GOG protocol 132: 6 cycles (81%), 5 cycles (2%), 4 cycles (5%), 3 cycles (4%), 2 cycles (3%), and 1 cycle (5%).36 Treatment-related complications that were most likely to result in hospitalization were assumed to be fever, infection, gastrointestinal toxicity, metabolic events, renal or genitourinary events, and thrombocytopenia. The frequencies of hospitalization for these events, according to residual disease and treatment regimen, were calculated at 50% of the incidence of grade 3 or 4 fever and infectious morbidity and 25% of the incidence of grade 3 or 4 gastrointestinal toxicity, metabolic events, renal or genitourinary events, and thrombocytopenia in GOG protocols 172 and 132 (Table 1).35,36 All survival estimates were generated from the actual or projected median survival from Phase III trials of patients with advanced-stage ovarian cancer (GOG protocols 172 and 132) and were incorporated into cost-effectiveness analysis first as undiscounted outcomes and then discounted at a rate of 3% per year according to the estimated median survival for each treatment program (Table 1).27,28,35,36 Quality-of-life adjustments for the toxicity of chemotherapy during the first year of survival were estimated from longitudinal quality-of-life assessments using the Functional Assessment of Cancer Treatment-Oncology (FACT-O) survey in GOG protocol 172.35 Quality-of-life index values were calculated based on a maximum FACT-O score of 156, with
TABLE 2 Model Estimates: Costs Clinical parameter
Model estimate, $
Primary surgery* Placement of IP catheter and first cycle of IP chemotherapy* Inpatient IP/IV chemotherapy cycle* Outpatient IP/IV chemotherapy cycle* Removal of IP catheter* Outpatient IV/IV chemotherapy cycle* Total cost surgery and IP/IV therapy (treatment received)y Total cost optimal surgery and IV/IV therapy (treatment received)y Total cost suboptimal surgery and IV/IV therapy (treatment received)y Cost of each hospitalization for fever or infection* Cost of each hospitalization for gastrointestinal toxicity* Cost of each hospitalization for metabolic event* Cost of each hospitalization for thrombocytopenia* Cost of each hospitalization for renal or genitourinary event* Lost wages{ Caregiver support IP/IV chemotherapy{ Caregiver support IV/IV chemotherapy{
21,133 6080 2860 1655 1957 1494 45,172 29,633 29,184 6830 7184 6654 13,756 14,730 8440§ 1381§ 1036§
IP indicates intraperitoneal; IV, intravenous. * Costs based on 40% collection rate for professional fee charges and 60% collection rate for hospital charges. y Not all patients completed 6 cycles of specified regimen; incorporates costs of surgery and chemotherapy. { Based on an average hourly wage of $16.68 adjusted for a 5-day work week and a 57.5% employment rate. § According to the Bureau of Labor Statistics.37,38
quality-adjusted life years (QALYs) adjusted for differences between the IP/IV regimen and the IV/IV regimen during the initial 5-month period and the subsequent 7 months after diagnosis.
Model Estimates: Costs Model estimates for the costs of care were derived from the Maryland HSCRC database for hospital charges and from the common-use JHMI administrative databases for professional fee charges for patients who underwent initial surgical management and received primary chemotherapy for FIGO stage IIIC and IV ovarian cancer between FY 2004 and FY 2006. All charges were adjusted to 2006 U.S. dollars using the consumer price index, and estimated costs were calculated as 60% of total hospital charges and 40% of professional fees (based on the average ratio of total hospital direct and indirect costs to charges and the average departmental professional fee-collection rate) (Table 2). Primary cytoreductive surgery Direct and indirect hospital charges were obtained for patient services with an International Classification of Disease, 9th Revision, Clinical Modification (ICD-9-CM) code of 183.0 and Common Procedural Terminology (CPT) codes of 58951, 58952, 58953, or
Centralization of Ovarian Cancer Care/Bristow et al.
58954. Total hospital charges (mean, $32,889; median, $26,541; range $18,686-85,430) were calculated for the index admission of 40 consecutive patients undergoing primary cytoreductive surgery. The mean total hospital charge was used to generate model cost estimates. Professional fee surgeon charges were estimated as 40% of $3500 ($1400) to approximate the contractual Medicare reimbursement of $1379 for a CPT code of 58951 during FY 2006.
IP chemotherapy regimen The IP/IV chemotherapy regimen was based on the treatment program reported in GOG protocol 172 and currently in use at the JHMI.35 Each cycle consisted of a 24-hour infusion of paclitaxel on Day 1; an IP infusion of cisplatin in Day 2, which required a 2day inpatient hospitalization; and IP paclitaxel administered in the ambulatory setting on Day 8. Charges were incorporated for placement of an IP infusion port as a separate procedure (CPT code 44942) and coinciding with the first cycle of chemotherapy (CPT code 99221). Mean hospital and professional fee charges for these procedures were calculated at $9284 and $1275, respectively. For each subsequent treatment cycle, mean inpatient hospital and professional charges were $4584 and $275, respectively, for treatment on Days 1 and 2, whereas the mean ambulatory hospital and professional charges were $2665 and $140, respectively, for treatment on Day 8. Removal of the intraperitoneal catheter on an ambulatory basis was associated with a mean hospital charge of $2261 and a professional charge (CPT code 44942) of $1000. The total cost of surgery and IP/IV chemotherapy based on treatment received, excluding hospitalization for treatment-related toxicity, was $45,172. IV chemotherapy regimen The IV/IV chemotherapy treatment regimen consisted of IV carboplatin and IV paclitaxel (as a 3-hour infusion) administered on an ambulatory basis on Day 1 of a 21-day treatment cycle. For this calculation, mean hospital charges were $2396 per treatment cycle. The professional charge associated with each treatment was $140. The total costs of surgery and IV/IV chemotherapy, based on treatment received, excluding hospitalization for treatmentrelated toxicity, were $29,633 after optimal cytoreduction and $29,184 after suboptimal cytoreduction. Hospitalization for treatment-related toxicity All cost estimates for treatment-related toxicity were obtained from a review of the Maryland HSCRC database for patients who were admitted to JHMI during FY 2005 to FY 2006 for neutropenic fever
1517
(ICD-9-CM code 288.0) resulting from chemotherapy, gastrointestinal toxicity (ICD-9-CM code 560.81), chemotherapy-related metabolic events (ICD-9-CM code 276.9), and chemotherapy-induced thrombocytopenia (ICD-9-CM code 287.5).
Lost wages and caregiver support Lost patient wages were calculated based on an average hourly earning of $16.68 and were adjusted for a 5-day work week and an employment-to-population ratio of 57.5%, as reported by the Bureau of Labor Statistics as of April 2006 for female residents of the United States aged 20 years.37,38 Lost wages were estimated for a 6-week period surrounding surgery plus an additional 16 weeks to complete the full treatment program. Caregiver support costs were calculated similarly by using estimates of 2 hours per day during the postoperative recovery period, 14 hours per IP/IV treatment cycle, and 8 hours per IV/ IV treatment cycle. Cost-Effectiveness and Sensitivity Analyses Cost effectiveness was calculated as the overall cost per patient, in terms of economic impact and effect on quality-adjusted survival time, of implementing each of the 2 strategies. The cost-effectiveness ratio was calculated as the overall cost per QALY associated with each strategy, and the incremental cost-effectiveness ratio was calculated as the additional cost per QALY for the more costly strategy. A series of 1-way sensitivity analyses of each relevant variable across a reasonable range was performed, and a tornado diagram was developed for the entire model. Both cost-effectiveness analyses and 1-way sensitivity analyses were conducted from the perspective of society. Probabilistic sensitivity analysis was performed with Monte Carlo simulation with 10,000 repetitions, and triangular distributions were assumed for all variables. For cost estimates and survival time assumptions, 95% confidence intervals (95% CIs) were used to define variable distributions, whereas baseline values 25% were used to define the distributions for rates of optimal and complete cytoreductive surgery. All modeling and calculations were performed using a commercially available decision-analysis software program (TreeAge Pro Healthcare module; Treeage Software Inc., Williamston, Mass), and all costs are reported as 2006 U.S. dollars.
RESULTS Cost-Effectiveness Analysis Cost-effectiveness analysis revealed that the strategy of referral to an expert center had an overall cost per patient of $50,652 and effectiveness of 5.12 QALYs
1518
CANCER
April 15, 2007 / Volume 109 / Number 8
TABLE 3 Cost-effectiveness Comparison of Expert Center Referral Strategy Versus Less Experienced Center Referral Strategy Strategy Undiscounted survival time Less experienced center Expert center Discounted survival time* Less experienced center Expert center
Cost, $
Incremental cost, $
Effectiveness, QALYs
Incremental Effectiveness, QALYs
C/E, $
Incremental C/E, $
39,957 50,652
10,695
2.33 5.12
2.78
17,149 9893
3809
40,116 50,592
10,476
2.14 4.24
2.11
18,746 11,932
5029
QALYs indicates quality-adjusted life years; C/E, cost-effective ratio. Survival time discounted at 3% per year.
FIGURE 2. Sensitivity analyses: tornado diagram for the overall model. IV
indicates
intravenous;
chemo,
chemotherapy; IP, intraperitoneal; exper., experienced; cytoreduct., cytoreduction; wtp, willingness to pay.
(Table 3). The strategy of referral to a less experienced center carried an overall cost of $39,957 and effectiveness of 2.33 QALYs. The expert center referral strategy was associated with an additional 2.78 QALYs at an incremental cost of $10,695 but was more cost effective, with a cost-effectiveness ratio of $9893 per QALY compared with $17,149 for the less experienced center referral strategy. The incremental cost-effectiveness ratio (ICER) of the more effective option is the ratio of the mean incremental cost and mean incremental effectiveness (in terms of $ per QALY). Expert center referral, which was the more effective strategy, had an ICER of $3809. In other words, each additional QALY gained by referral to an expert center cost $3809. Repeating the cost-effectiveness analysis with survival time discounted at a rate of 3% per year yielded similar findings, with the strategy of expert center referral associated with an incremental gain of 2.11 QALYs at an additional cost of $10,476 per patient (Table 3).
Sensitivity Analyses The tornado diagram depicted in Figure 2 shows that the variables with the most influence on the cost of the different strategies in the overall model were the cost of IV/IV chemotherapy for optimally and suboptimally resected patients, the cost of IP/IV chemotherapy, and the probability of IP/IV chemotherapy at a less experienced center. The only threshold value was observed for the cost of treatment with IP/IV chemotherapy; at a cost of $28,000, IP/IV chemotherapy and IV/IV chemotherapy would have an equivalent value total cost of $39,000. For a cost of initial treatment
