Ovarian cancer is the eighth most common cancer among women, but the disease ranks fifth in cancer-related causes of death among women.1 This translates to 22,440 new ovarian cancer cases diagnosed annually in the United States.2 Of these cases, 60% are detected in late stages of disease, after the cancer has metastasized. The 5-year survival rate for late-stage ovarian cancer is only 28.9%, but when detected in an early stage, the survival rate increases to 92.2%.3 The current standard of care for assessing the risk for malignancy when evaluating a complex ovarian mass relies heavily on the single-protein biomarker CA125, which is exceedingly poor at identifying early stages of ovarian cancer.4,5
An additional concern with the existing standard of care is the unnecessary referral of low-risk patients to the limited supply of gynecologic oncologists. Although referral to a gynecologic oncologist is common practice, the majority of pelvic masses are benign and can safely and cost-effectively be removed by a general gynecologic surgeon. Moreover, it is estimated that approximately 20% of women will develop a pelvic mass or a benign ovarian cyst during their lifetime6; therefore, referring all such women to the limited number of gynecologic oncologists is not a feasible practice.
The American College of Obstetricians and Gynecologists (ACOG) released updated guidelines in November 2016 of its recommended criteria for referring patients with a pelvic mass to a gynecologic oncologist.7 Previous ACOG guidelines from 2007 recommended referral to a gynecologic oncologist based on the results of ultrasound imaging studies, the presence of ascites, the existence of a nodular or fixed pelvic mass, a family history of breast or ovarian cancer, evidence of abdominal or distant metastasis, and elevated CA125 levels.8 An elevated CA125 level indicates an increased risk for malignancy; however, a woman’s CA125 level may be elevated for reasons other than ovarian cancer.9
The 2007 ACOG guidelines provided a recommended CA125 level for premenopausal women informed by expert opinion, not by clinical evidence.8 However, this recommendation was removed in the 2016 guidelines and was replaced with the statement, “no evidence-based threshold is currently available.”7 The updated guidelines now include the use of a multivariate index assay, such as OVA1, Risk of Malignancy Index, Risk of Ovarian Malignancy Algorithm, or International Ovarian Tumor Analysis, as alternatives to CA125.7 Specifically, the guidelines state that CA125 or a multivariate index assay should only be performed when level A recommendations (ultrasound and physical examination) are inconclusive.
When used as a stand-alone diagnostic to determine the risk for malignancy, the overall sensitivity of CA125 is between 68.4% and 77.0%, and declines to 61.0% to 65.7% in patients with early-stage ovarian cancer.4,5,10-12 When CA125 is used to determine the risk for malignancy in conjunction with results from ultrasound imaging and physical examination, the sensitivity rises only to 77.0% to 79.3% across all women11,13 and to 76.7% in women with early-stage cancer (47.1% for premenopausal women vs 88.2% for postmenopausal women).4,13
Early detection and appropriate intervention improve member clinical outcomes, and evidence shows that treatment costs for late-stage ovarian cancer are considerably more expensive.14 The cost of treating advanced ovarian cancer, especially in older individuals, is substantial, averaging $65,908 for the initial treatment of a Medicare beneficiary.15 Accordingly, the current dependence on a single-protein biomarker for assessing the risk for malignancy is arguably inadequate.
OVA1 combines the results of 5 single-protein biomarkers (ie, CA125, apolipoprotein A1, beta 2 microglobulin, prealbumin, and transferrin) to determine the risk for malignancy, and has an overall sensitivity of 92.2% as a stand-alone test and rises to 98.1% when accompanied by ultrasound imaging and physical examination.4,13 This represents a 23.8% increase in sensitivity versus CA125. Moreover, the sensitivity and negative predictive value of OVA1 are considerably higher for patients with early-stage ovarian cancer.13 The negative predictive value (ie, the proportion of true negative test results) for OVA1 ranges from 92.0% to 96.9%,4,5,10-13 allowing physicians to maintain clinical responsibility more confidently for patients who are at low risk for ovarian cancer rather than referring to a gynecologic oncologist for evaluation of pelvic masses that are predominately benign.
The sensitivity of OVA1 in premenopausal women is 88.2% when combined with ultrasound imaging and physical evaluation, whereas CA125 only reaches 47.1%.13 Therefore, premenopausal women are at an increased risk for failure to be diagnosed at early-stage disease, resulting in an increased mortality risk associated with advanced disease.3
Two studies have examined the economic impact of using OVA1 to determine which health plan members should be referred to a gynecologic oncologist for removal of a pelvic mass. One study found a “refer all” strategy to be least expensive, but did not account for quality-adjusted life-years (QALYs) or for patient survival rates.16 The second study, which did account for member-perceived quality of life as well as survival rates, showed that OVA1 had an incremental cost-effectiveness ratio (ICER) of $35,094 per QALY gained compared with the 2007 modified ACOG guidelines and an ICER of $12,189 per QALY gained compared with CA125 alone.17 In the United States, an ICER of <$62,000 is thought to illustrate an efficient intervention, supporting the cost-effectiveness of OVA1.18 Moreover, in the second study the “refer all” strategy was determined to be less expensive, but was associated with reduced QALYs.17
Although there is a robust body of literature to support the improved test sensitivity of OVA1 compared with the standard of care, few studies have examined the economic impact of OVA1 from the perspective of payers. The cost of cancer care is projected to increase by 27% from 2010 to 2020,19 partially because of the introduction of new treatments, and payers need to better understand the economic impact on a per-member per-month (PMPM) basis. The present study is intended to help inform health plan decision makers about the true cost of incorporating the use of OVA1 to aid the diagnosis of female members who currently rely on CA125 to determine their risk for pelvic mass malignancy.
The budget impact model in this study was constructed from the perspective of health plan policymakers and comprises commercially insured and Medicare members. The model starts with a current-state scenario, in which CA125 is used along with the current clinical guidelines to determine the risk for pelvic mass malignancy.7 In this scenario, the model applies a sequence of filters to identify the relevant subset of plan members for inclusion. The Figure provides a flow diagram of the filtering process for this study. The filtering process distributes plan members who receive a CA125 into 2 pathways: the first pathway analyzes the course of treatment for health plan members whose CA125 results were nonelevated, and the second pathway analyzes the course of treatment for those with elevated CA125 results whose pelvic mass was determined to be in the early versus late stages of malignancy.
The model then devises a future-state scenario, in which OVA1 is substituted for CA125 in a proportion of members to determine their risk for pelvic mass malignancy. Relevant plan members are split into the 2 pathways using the same filtering logic that was applied in the current-state scenario; however, several assumptions are incorporated, including (1) the OVA1 market penetration rate, (2) the improved accuracy of OVA1 compared with CA125, and (3) assumed variation in key parameters pertaining to clinical resource utilization.
The first parameter variation captures a shift in procedures that were performed from the inpatient setting to the outpatient setting, which was enabled by an increase in diagnostic accuracy, allowing physicians to manage the patient in a less intense setting because of a low risk for malignancy based on OVA1. Second, increased physician confidence in OVA1 is assumed to also drive a shift in the number of procedures performed by physician type—moving away from gynecologic oncologists to general gynecologists. Finally, the model incorporates the possibility that a proportion of patients with a very low risk for malignancy may avoid surgery, based on OVA1 results combined with clinical assessment and ultrasound imaging instead of surgery.
The economic outcomes of the budget impact model (ie, estimated total costs and PMPM computations) for each pathway of the model are calculated separately for commercially insured and Medicare members. First, we estimate the total procedure costs accumulated in the current-state and future-state scenarios for members filtered into the nonelevated CA125/OVA1 pathway. The PMPM economic impact is then calculated for each scenario. Future-state PMPMs are subtracted from respective current-state PMPMs to determine the overall budget impact of OVA1 at the health plan population level.
Outcomes were calculated using the same method for the elevated CA125/OVA1 pathway. However, instead of total procedure costs, this analysis used the 24-month total episode-of-care costs to examine the impact of OVA1 on the total cost of early- and late-stage cancer, because the total cost of managing these groups of patients is very different.
The model culminates in a sensitivity analysis in which we permit uncertainty around the assumptions made in the future-state scenario to vary across a range of values to examine the resulting impact on PMPM expenditures.
The metrics used to conduct the filtering processes in operationalizing the current- and future-state scenarios originated from real-world claims data and literature sources. Claims data were used to filter the total payer population to include only females tested with CA125. Because claims compendia do not contain the entire set of data elements required (particularly the results of diagnostic tests) to complete the filtering process, literature sources were utilized to obtain values for the missing elements, including the percent of CA125 results that are elevated versus nonelevated, the percent of cases with nonelevated CA125 results who did and did not proceed with surgery, the percent of cases with elevated CA125 results whose pelvic mass is determined to be benign versus malignant, and the percent of malignant cases detected in an early versus late stage of cancer. In addition, CA125 and OVA1 sensitivities were taken from the literature. A complete list of values and their respective data sources are available in the Figure.
Data Sources for Procedure Mix and Costs
The breakdown of the procedure mix and the associated costs utilized in the first pathway of the analyses (ie, nonelevated CA125/OVA1 analysis) were derived from insurance claims. Commercial claims data were extracted from the OptumInsight, Inc (Eden Prairie, MN) database comprising approximately 25 million members. Claims used to derive the procedure costs were required to meet certain criteria, including (1) female members who had CA125 testing, as identified by Current Procedural Terminology (CPT) code 86304 from April 1, 2013, through March 31, 2015, and (2) the absence of a cancer diagnosis at the time of the procedure or within 6 months after the procedure (to ensure that members receiving CA125 testing to monitor ovarian cancer treatment response were not included in the analysis). Appendix A provides a list of the diagnosis codes used to make this distinction.
Surgical cases were classified as a hysterectomy, salpingectomy and/or oophorectomy, cystectomy, or no surgery based on the procedure codes identified during the 6 months after the initial CA125 test. Because it is possible for members to receive multiple procedures, a hierarchy was established (in the same order as the procedures listed above) to restrict consideration to the most invasive procedure for each member (eg, a claim with hysterectomy and oophorectomy procedure codes would be classified as a member who received a hysterectomy). A complete list of CPT procedure codes used in the classification scheme is reported in Appendix B.
Claims for these procedures were further categorized by place of service (ie, inpatient and outpatient), physician type (ie, gynecologist and gynecologic oncologist), and menopausal status. Because claims data do not specify menopausal status, member age was used as a proxy. Women aged 18 to 51 years were classified as premenopausal, and women aged 52 to 64 years were classified as postmenopausal. Women aged >64 years were excluded from the commercial data to differentiate from Medicare. The paid claims amounts were used to represent the actual reimbursement levels by payers.
The commercial cost data utilized for the elevated CA125/OVA1 pathway were gathered through a 24-month retrospective claims analysis of female members with ovarian cancer (International Classification of Diseases, Ninth Revision [ICD-9] code 138.0) who received CA125 testing between July 1, 2013, and June 30, 2014. ICD-9 diagnosis codes for ovarian cancer do not distinguish between early- and late-stage cancers; therefore, clinician experts were consulted on the appropriate method for classifying members.
Late-stage ovarian cancer was defined as the metastasis of cancer outside of the pelvic region or to lymph nodes. Claims meeting these criteria between July 1, 2013, and June 30, 2014, were classified as late-stage ovarian cancer. Conversely, early-stage cancer was defined as the absence of these diagnosis codes suggesting spread of the disease. A list of CPT codes used in the classification scheme is shown in Appendix B. The episode-of-care costs comprised reimbursement for the healthcare expenditures, including inpatient care, outpatient care, physician services, and pharmacy.
The claims used to derive the costs for Medicare calculations were identified from the Medicare Inpatient Standard Analytic File and the Medicare Outpatient Standard Analytic File between January 1, 2014, and December 31, 2015, for the nonelevated CA125/OVA1 analysis and between January 1, 2013, and December 31, 2013, for the elevated CA125/OVA1 analysis.
The Medicare files contain data for approximately 37 million fee-for-service members. The claims data for women aged ≥65 years were extracted from Medicare data sets using the same criteria for the commercially insured members, and these women were deemed to be postmenopausal. However, because the cost data for physician, pharmacy, and outpatient services were not available in the Medicare data sets, the cost for these categories was estimated to be 80% of the commercial payment rate.
Finally, the total cost estimates in the nonelevated and elevated CA125/OVA1 analyses are inclusive of the CA125 and OVA1 costs. Reimbursement for CA125 testing was derived from claims data ($36, commercial; $31, Medicare). OVA1 pricing was provided by the manufacturer—$552 for commercial, and $225 for Medicare.
A total of 92,843 plan members were included in the analysis, which comprised 48,113 commercially insured members and 44,730 Medicare beneficiaries. In the nonelevated CA125/OVA1 pathway of the model, 44,620 commercially insured members and 38,214 Medicare beneficiaries were retained for analysis. The elevated CA125/OVA1 pathway comprised 3493 commercial insurance members (ie, 2646 with early-stage cancer and 847 with late-stage cancer) and 6516 Medicare beneficiaries (ie, 2747 with early-stage cancer and 3769 with late-stage cancer).
Nonelevated CA125/OVA1 Pathway
The economic impact of OVA1 use on the subset of members currently tested with CA125 was calculated subject to the following conservative assumptions:
- OVA1 market penetration of 20%
- 5% shift of procedures from inpatient to outpatient setting
- 5% of procedures currently performed by gynecologic oncologists shift to general gynecologists
- 5% of patients whose procedures are currently performed by a gynecologist will avoid surgery.
Our model assumptions were developed in consultation with the lead author (who is a clinician who actively manages patients with ovarian cancer and a former medical director of a large health plan) who urged conservatism in our estimate; the model assumptions were further informed by expert opinions and market intelligence offered by the manufacturer.
As a result of these assumptions, the future-state scenario estimates that commercially insured patients will experience a 0.11% decrease in the total procedures performed by gynecologic oncologists, a 0.20% decrease in the total procedures performed by gynecologists, and a 0.30% increase in surgery avoidance. The Medicare population experienced a 0.22% decrease in the total procedures performed by gynecologic oncologists, a 0.04% increase in the total procedures performed by gynecologists, and a 0.18% increase in surgery avoidance.
Table 1 provides full details on the procedure incidence rates by setting and by physician type for the commercially insured and Medicare populations. Average payer reimbursements per procedure type and by care setting are shown in Table 2.
As expected, the costs for inpatient procedures were substantially higher than for outpatient procedures. In the commercially insured group, the total cost of a hysterectomy performed by a gynecologist as an inpatient procedure averaged $14,932 compared with $7421 when performed as an outpatient procedure. The cost difference remained consistent among the Medicare population, in whom the total payments for an inpatient hysterectomy performed by a gynecologist averaged $8743 compared with $3342 for the equivalent outpatient procedure.
Furthermore, procedures performed by gynecologic oncologists were more expensive than when performed by a gynecologist. An inpatient hysterectomy procedure averaged $14,932 when completed by a gynecologist and $16,689 when performed by a gynecologic oncologist. This pattern persisted in the Medicare population, where an inpatient hysterectomy cost averaged $8743 for a gynecologist and $9947 for a gynecologic oncologist.
Elevated CA125/OVA1 Pathway
Early- and late-stage cancer incidence rates for the current- and future-state scenarios by payer population are presented in Table 3. In the commercially insured population, the future adoption of OVA1 at the 20% level produced a 3.27% shift in cancer detection from late stage to early stage among premenopausal women and a shift of 0.38% among postmenopausal women. In the Medicare population, future utilization of OVA1 at the 20% adoption rate produced a 0.38% shift of cancer detection from late stage to early stage, where all women are presumed to be postmenopausal.
Table 4 presents the 24-month episode-of-care costs for early-stage and late-stage cancers, by payer. The costs were substantially higher for late-stage cancer compared with early-stage cancer in the commercially insured and Medicare populations. Commercially insured premenopausal women with early-stage cancer had 24-month episode costs that averaged $35,754, whereas premenopausal women with late-stage cancer had an average 24-month episode cost of $224,922.
PMPM Calculations and Sensitivity Analysis
PMPM calculations were completed for the commercially insured and Medicare populations, combining the economic impact of OVA1 on both pathways that were analyzed in the model. In the future-state scenario (using our base-case assumptions), the commercial population experiences a $0.05 PMPM savings and the Medicare population experiences a $0.01 PMPM savings.
Table 5 presents a sensitivity analysis of PMPM savings that was conducted to determine the range of the economic impact that OVA1 may have on a payer, given variations in the key model inputs. The model assumptions that were allowed to vary in the sensitivity analysis included (1) an OVA1 market penetration rate ranging from 10% to 50%, (2) a shift of procedure location ranging from 2.5% to 20%, and (3) a shift of physician type (which includes surgery avoidance) from 2.5% to 20%.
The results of the sensitivity analyses revealed that increased adoption of OVA1 to 50% market penetration could yield payer savings up to $0.17 PMPM in the commercial population and $0.05 PMPM in the Medicare population. It is best to calculate the total dollar savings at the plan-specific level.
More than 22,000 new cases of ovarian cancer are diagnosed annually in the United States, with the majority of them being late-stage disease at the time of diagnosis. The results of our budget impact model estimate that average treatment costs for late-stage cancer could be as high as $224,922 over a 24-month episode. This finding, coupled with extremely high mortality rates for this disease, poses a formidable challenge for physicians and payers.
Our study findings suggest that, because of its increased sensitivity and negative predictive value, the use of OVA1 may result in savings for payers of $0.05 PMPM by enabling physicians to better manage patients in the appropriate setting of care. These savings can be attributed to a reduction in the total procedures performed by gynecologic oncologists (up to 0.22%), decreases in total procedures performed by gynecologists (up to 0.20%), a 0.30% increase in surgery avoidance for patients with nonelevated test results, as well as a shift in cancer detection from late-stage to early-stage cancer reaching 3.27%.
The results of the budget impact model support the use of OVA1 by indicating that modest cost-savings can be achieved by health plans while reaping the clinical benefits of improved diagnostic accuracy, early disease detection, and reductions in multiple, and possibly unnecessary, referrals to gynecologic oncologists. The observed favorable differential of savings in the commercial population is likely attributable to the increased sensitivity of OVA1 in premenopausal women compared with the current standard of care. Overall, the addition of a multivariate index assay is cost-neutral, and possibly cost-sparing, across the commercially insured and Medicare populations.
This study has its limitations. First, reliance on administrative claims data makes it difficult to categorize precisely the clinical conditions and the chronologic occurrence of events.
Second, using 80% of commercial reimbursements as a proxy for Medicare physician, pharmacy, and outpatient payments might have underestimated or overestimated Medicare costs, depending on the region of the country.
Third, our assumption that a portion of low-risk cases may be managed by gynecologists in the outpatient setting may be imprecise, especially in the Medicare population, because the model is not able to accurately account for clinical modifiers such as comorbidities and surgical risks, which could drive these patients to inpatient settings if present.
Fourth, our assumptions regarding adoption rates, changes in referral patterns, and changes in setting of care are subject to challenge; however, such uncertainties are addressed by the sensitivity analysis. Our analyses of direct medical costs did not include coinsurance or deductible amounts paid by Medicare patients, but patient liability is included in commercial data.
Finally, the analysis did not consider the indirect costs associated with time off from work, travel time to visit the offices of thinly distributed gynecologic oncologists, and the psychological stress associated with seeing a cancer specialist.
The budget impact model presented in this study is offered as a first step toward assisting health plan policymakers in weighing the financial consequences—that is, differences between inpatient and outpatient procedure reimbursements and early- versus late-stage cancer reimbursements in premenopausal versus postmenopausal women—of expanded utilization of multivariate assay testing in relation to its clinical advantages in treating this silent killer, which often presents at a late stage. Future investigations of this type should further refine our model, by extending the analyses longitudinally and by more closely coupling economic data with clinical measures. We also encourage testing of the model using regional and plan-specific data to assess its robustness.
We would like to thank Brittany Blau for her assistance with literature research and drafting of the manuscript, and David Gregory for conceptualization of the study.
Source of Funding
This study was funded by Vermillion Inc, Austin, TX, which developed the OVA1 test.
Author Disclosure Statement
Dr Brodsky is on the Speaker’s Bureau for Myriad Genetics. Dr Owens is a consultant to Eli Lilly, Genentech, Inspire Medical Systems, Johnson & Johnson, Novartis, Roche Diagnostics, and Sun Pharmaceuticals. Dr Scotti, Mr Needham, and Ms Cool are consultants to Vermillion.
Dr Brodsky is Assistant Professor of Obstetrics and Gynecology, The University of Toledo Medical Center, Toledo, OH; Dr Owens is President, Gary Owens Associates, Ocean View, DE; Dr Scotti is Alfred E. Driscoll Professor of Healthcare & Life Sciences Management, Fairleigh Dickinson University, Teaneck, NJ; Mr Needham is Senior Consultant, Baker Tilly LLP, New York, NY; Ms Cool is Manager, Baker Tilly LLP.
1. Ovarian Cancer Research Fund Alliance. Statistics. www.ocrfa.org/members/about-ovarian-cancer/statistics/. Accessed November 22, 2016.
2. American Cancer Society. What are the key statistics about ovarian cancer? Revised January 6, 2017. www.cancer.org/cancer/ovariancancer/detailedguide/ovarian-cancer-key-statistics. Accessed February 16, 2017.
3. National Cancer Institute. SEER cancer stat facts: ovarian cancer. www.seer.cancer.gov/statfacts/html/ovary.html. Accessed September 7, 2017.
4. Longoria TC, Ueland FR, Zhang Z, et al. Clinical performance of a multivariate index assay for detecting early-stage ovarian cancer. Am J Obstet Gynecol. 2014;210:78.e1-78.e9.
5. Ueland FR, Desimone CP, Seamon LG, et al. Effectiveness of a multivariate index assay in the preoperative assessment of ovarian tumors. Obstet Gynecol. 2011;117:1289-1297.
6. Moore RG, Bast RC Jr. How do you distinguish a malignant pelvic mass from a benign pelvic mass? Imaging, biomarkers, or none of the above. J Clin Oncol. 2007;25:4159-4161.
7. American College of Obstetricians and Gynecologists’ Committee on Practice Bulletins—Gynecology. Practice Bulletin No. 174: evaluation and management of adnexal masses. Obstet Gynecol. 2016;128:e210-e226.
8. American College of Obstetricians and Gynecologists. ACOG Practice Bulletin. Management of adnexal masses. Obstet Gynecol. 2007;110:201-214.
9. Mayo Clinic. Tests and procedures: CA 125 test: definition. April 14, 2014. www.mayoclinic.org/tests-procedures/ca-125-test/basics/definition/prc-20009524. Accessed February 7, 2017.
10. Bristow RE, Smith A, Zhang Z, et al. Ovarian malignancy risk stratification of the adnexal mass using a multivariate index assay. Gynecol Oncol. 2013;128: 252-259.
11. Bristow RE, Hodeib M, Smith A, et al. Impact of a multivariate index assay on referral patterns for surgical management of an adnexal mass. Am J Obstet Gynecol. 2013;209:581.e1-581.e8.
12. Goodrich ST, Bristow RE, Santoso JT, et al. The effect of ovarian imaging on the clinical interpretation of a multivariate index assay. Am J Obstet Gynecol. 2014;211:65.e1-65.e11.
13. Ware Miller R, Smith A, DeSimone CP, et al. Performance of the American College of Obstetricians and Gynecologists’ ovarian tumor referral guidelines with a multivariate index assay. Obstet Gynecol. 2011;117:1298-1306.
14. Yabroff KR, Lamont EB, Mariotto A, et al. Cost of care for elderly cancer patients in the United States. J Natl Cancer Inst. 2008;100:630-641.
15. Urban RR, He H, Alfonso-Cristancho R, et al. The cost of initial care for Medicare patients with advanced ovarian cancer. J Natl Compr Canc Netw. 2016;14:429-437.
16. Kim KH, Zsebik GN, Straughn JM Jr, Landen CN Jr. Management of complex pelvic masses using a multivariate index assay: a decision analysis. Gynecol Oncol. 2012;126:364-368.
17. Forde GK, Hornberger J, Michalopoulos S, Bristow RE. Cost-effectiveness analysis of a multivariate index assay compared to modified American College of Obstetricians and Gynecologists criteria and CA-125 in the triage of women with adnexal masses. Curr Med Res Opin. 2016;32:321-329.
18. Shiroiwa T, Sung YK, Fukuda T, et al. International survey on willingness-to-pay (WTP) for one additional QALY gained: what is the threshold of cost effectiveness? Health Econ. 2010;19:422-437.
19. Mariotto AB, Yabroff KR, Shao Y, et al. Projections of the cost of cancer care in the United States: 2010-2020. J Natl Cancer Inst. 2011;103:117-128. Erratum in: J Natl Cancer Inst. 2011;103:699.
20. OptumInsight. Single payer database. 2014. Data on file at Baker Tilly, LLP.