Health Care Costs of Anal Cancer in a Commercially Insured Population in the United States

BACKGROUND: The incidence and death rate of anal cancer in the United States has been increasing on average 2%-3% per year over the past 10 years. Human papillomavirus (HPV) vaccination is a potentially viable prevention strategy, since about 80% of anal cancers are attributable to HPV. To understand the effect of HPV vaccination, it is important to estimate the treatment costs for the HPV-related disease. OBJECTIVE: To estimate the 2-year per patient mean direct health care costs associated with anal cancer in a commercially insured population in the United States. METHODS: All newly diagnosed anal cancer patients were identified from a 2011-2014 Truven MarketScan database. Matched population controls were selected from the database with a 2-step matching method using demographic, comorbidity, and health care cost variables. Costs for the first 2 years were measured for cancer patients and controls. The difference in costs between the groups was calculated to estimate the costs associated with anal cancer treatment. A generalized linear model with gamma distribution and log link function was applied to estimate the costs for censored months for the patients who did not have at least 2 years of follow-up. RESULTS: 1,976 patients with anal cancer and 1,976 controls were included in the study. The first 2-year per patient adjusted mean cost associated with anal cancer treatment was $127,531 (SD = $189,064). Male sex, cancer diagnosis, higher Charlson Comorbidity Index score, and higher prediagnosis costs were significantly associated with higher monthly costs. Higher psychiatric diagnostic group scores were associated with lower monthly costs. Anal cancer treatment costs were highest in the first 6 months after diagnosis (per patient per month [PPPM] mean = $12,846), leveling off at a much lower monthly cost during the subsequent 18 months of the 2-year period (PPPM mean = $3,717). CONCLUSIONS: The first 2-year costs associated with anal cancer treatment were substantial. Given that approximately 80% of anal cancers are attributable to HPV infection, this study provides important parameters for estimating the potential economic benefit of HPV vaccination.

A nal cancer is rare, accounting for 0.5% of all new cancer cases in the United States. In 2017, there were 8,200 new cases of anal cancer diagnosed in the United States, and it was estimated that 1,100 people would die from this cancer. 1 The incidence of anal cancer has more than doubled since 1975 and has been increasing on average 2.2% each year over the past 10 years. 2 The death rate of anal cancer has been increasing on average 3.2% each year over the past 10 years, with no improvement in survival over the past 4 decades. 2 Anal cancer affects more women than men, showing a gender ratio of 1.4 to 1 in the U.S. population. Most cases are diagnosed in individuals aged 50 years or older. 2 Almost all anal cancers are attributable to infection with oncogenic types of human papillomavirus (HPV). In a study of anal squamous cell carcinomas diagnosed between 1993 and 2005, HPV deoxyribonucleic acid was detected in 91% of anal cancers, and in 79% of those cases, the HPV type was 16 or 18. 3 Human immunodeficiency virus (HIV) infection is another risk factor associated with anal cancer occurrence, but it is unclear whether HIV infection directly causes anal carcinogenesis or if its effect is mediated through HPV. 4 There is a markedly high incidence of anal cancer in HIV-positive men who have sex with men, and a meta-analysis of 21 studies showed that the prevalence of anal HPV infection in HIV-positive men who had sex with men was 93%. 5 Cost-effectiveness studies provide information about the potential efficiency of using pharmacists and/or other alternative settings for increasing the HPV vaccination rate. 6 However, cost-effectiveness studies are limited by a lack of current real-world estimates of treatment costs for HPV-related cancers. For example, leading U.S. economic analyses are based on a study by Hu and Goldie (2008) that approximated the lifetime costs of anal cancer from a study of rectal cancer in a Canadian population, [7][8][9] and the leading U.S. HPV economic

Health Care Costs of Anal Cancer in a Commercially
Insured Population in the United States estimates will serve as an important input to decision models that inform policymakers about the potential cost offsets associated with expansion of U.S. HPV immunization rates, which are low compared with other developed countries and the Healthy People 2020 goal of 80% coverage. 14 13 However, the findings of this study are not directly applicable to the United States given the differences between the 2 countries in populations, cost structure, and treatment protocols. The purpose of this study was to estimate the per patient treatment costs over the first 2 years after diagnosis associated with incident cases of anal cancer from a large U.S. database of commercial insurance claims. Current real-world cost  health care experiences. The MarketScan CCAE database was particularly useful for studying the costs of treatment for anal cancer because this disease affects many individuals who are aged younger than 65 years. CCAE data were collected from more than 100 large employers and health insurance plans for employees, early retirees, and Consolidated Omnibus Budget Reconciliation Act beneficiaries and their dependents who were covered by private insurance. Costs were measured for inpatient admissions, outpatient services, and drug prescriptions. The population in the CCAE represents more than 15% of the employer-sponsored, privately insured, and noncapitated U.S. population younger than aged 65 years. In 2011, there were more than 50 million subjects in the CCAE database (49% male and 51% female). Forty-one percent of subjects were aged 18-44 years; 33% were aged 45-64 years; and about 26% were aged younger than 18 years. Since the MarketScan data were de-identified, the study was exempt from institutional human subjects review.

Case Identification
The Patients were included if they had at least 1 primary or secondary inpatient diagnosis claim for anal cancer or at least 2 primary or secondary outpatient diagnosis claims for anal cancer that were more than 30 days apart. The index date was defined as the first date an anal cancer diagnosis appeared in the data. Patients were excluded from the study if (a) they did not have continuous enrollment for 6 months before and after the index date, (b) they were diagnosed with cancer in any site (ICD-9-CM codes 140 through 208) during the 6 months before the index date, or (c) they were aged younger than 18 years.
We found 2 patients whose 2-year health care costs exceeded $1,000,000 ($2,087,635 in 1 patient and $1,129,032 in the other). Because these costs were extremely high compared with other patients' costs, we excluded these 2 outlier patients to avoid overestimating the mean treatment cost.

Control Selection
Controls were selected from among patients in the MarketScan database without a diagnosis of anal cancer or other HPVrelated cancer (i.e., cervical cancer, oropharyngeal cancer, penile cancer, vulvar cancer, and vaginal cancer). Controls were randomly assigned index dates that matched case index  After application of the exclusion criteria, each case was matched with controls on the basis of (a) age ± 5 years, (b) sex, (c) 9 geographic areas, and (d) index date. Propensity scores were computed for each case and candidate control for the following variables: (a) Charlson Comorbidity Index score during the 6 months before the index date, (b) psychiatric diagnostic group score during the 6 months before the index date, 17 (c) prediagnosis costs, and (d) type of health insurance plan. The prediagnosis costs were defined as the health care costs during the first 3 months of the 6-month period before the index date. For example, if an individual's index date was July 1, 2012, the prediagnosis costs were the health care costs from January 1, 2012, through March 31, 2012. A propensity score and Mahalanobis distance were used to produce 1 to 1 case and control matching. 18,19 Once a control was selected for a given case, the control was removed from consideration for additional matches.

Study Outcomes
Since the longest follow-up period in the study population was 3.5 years and the mean follow-up duration was approximately 20 months, we defined the primary study outcome as the per patient estimated 2-year health care costs for treatment of anal cancer. These costs were measured by calculating the difference in mean health care costs between the cases and controls in the first 2 years after the index date. A generalized linear model with log link function and gamma distribution was used for the cost estimates. Costs were adjusted for covariates including age, sex, having at least 1 year of follow-up or not, type of health insurance plan, geographic area, case/control status, Charlson Comorbidity Index score, psychiatric diagnostic group score, prediagnosis costs, 2-years' month index from 1 to 24, squared month index, the interaction terms between case/control status and month index, and the interaction terms between case/control status and squared month index.
The dependent variable for the model was the monthly health care costs during the 2-year follow-up period, which resulted in 24 partitions of estimated cost for each patient. To estimate the 2-year per patient health care costs, the censored data for the patients who did not have at least 2 years of follow-up were replaced with the model estimates. The monthly differences in health care costs between cases and controls allowed partitioning of costs between the initial treatment phase and the continuing phase of cancer treatment. 20 Cost was defined as the total cost of copayments, coinsurance, deductibles, and coordination of benefits and other adjustments for inpatient admissions, outpatient services, and prescription medications. We included all the health care costs incurred, including insurer payments and patient outof-pocket payments. All costs were adjusted to year 2015 U.S. dollar values using the Consumer Price Index from the U.S. Bureau of Labor Statistics. 21

■■ Results
A total of 1,976 patients diagnosed with anal cancer during 2011-2014 were identified (Figure 1). The mean age of cases was 54 years, and there were more women than men (63.82% vs. 36.18%; Table 1). The majority of cases and controls were covered by a preferred provider organization. The South Atlantic region (Delaware, Maryland, Virginia, West Virginia,

Observed and Model Adjusted Health Care Costs in Cases and Controls and Differences in Costs in the First 2 Years After the Index Date a
North Carolina, South Carolina, Georgia, Florida, and the District of Columbia) had the highest percentage of the patient population (20.14%). Except for employment status and enrollment duration, there were no significant differences between cases and controls in other population characteristics.
The cost difference between cases and controls was $106,579 (standard deviation [SD] = $87,491) for the first 2 years after the index date without adjustment for covariates and censoring ( Table 2). The greatest difference in costs between cases and controls in the first 2 years was the difference in costs of outpatient services (70%), followed by inpatient admissions (27%) and prescription drugs (3%; Table 2). The top 3 highest cost items among the outpatient services were radiation treatment After model adjustments for censored cases, the estimated total per patient health care cost for the first 2 years for patients with anal cancer was $127,531 (Table 2). Sex, case/control status, Charlson Comorbidity Index score, psychiatric diagnostic groups score, prediagnosis costs, and interaction terms between case/control status and month index and between case/control status and squared month index were significantly associated to 2015 U.S. dollars. Because that study was based on the experience from a single-payer health care system using older medical technology, the total cost may significantly underestimate today's cost of treating anal cancer in the United States. 9 Chesson et al. (2012) synthesized the results from Hu and Goldie and a previous HPV vaccine economic evaluation and then applied scenario analysis to estimate per patient anal cancer treatment cost at $39,800, adjusted to 2015 U.S. dollars. 9,23,24 In contrast, our study used real-world health care claims data with adjustments for covariates and non-normal cost distribution. Deshmukh et al. (2015) estimated the average lifetime cost per patient with anal cancer at $51,200 (2015 U.S. dollars). 12 However, since Deshmukh et al. focused on the Medicare-insured population, their results are not directly comparable to those of the present study.
Male sex, cancer diagnosis, higher Charlson Comorbidity Index score, and higher prediagnosis costs were significantly associated with higher per patient monthly costs in our study. An unexpected result was that higher psychiatric diagnostic group scores were associated with lower per patient monthly costs. While psychiatric conditions may present an access to care barrier, this could not be determined from the claims data.
Most of the cost in our population was for outpatient services (70%), followed by inpatient admissions (27%) and prescription drugs (3%). The outpatient service with the highest cost was radiation treatment, with a mean cost of $12,000 per patient. Patients incurred the highest costs in the first 6 months after the index date. After the sixth month, the cost pattern remained stable until the end of follow-up. This cost pattern matches what we expected in terms of anal cancer with the 2-year health care costs ( Table 3). The adjusted 2-year per patient health care costs for anal cancer patients were 8.7 times greater than the per patient costs for the matched noncancer patients. The case costs were highest in the first 6 months after diagnosis (per patient mean cost = $12,846), leveling off at much lower monthly costs during the subsequent 18 months of the 2-year period (per patient mean cost = $3,713; Figure 2). An example calculation of the adjusted cost using the model is provided in the Appendix (available in online article).

■■ Discussion
By searching the 2011-2014 MarketScan CCAE database, we identified nearly 2,000 patients with anal cancer, with a mean age of 54 years. This mean age is in line with the 2010-2014 statistics of the SEER Program, 22 which showed that the 10-year age group with the highest percentage of anal cancer cases diagnosed (29.4%) comprised individuals aged 55-64 years. In our study population, the ratio of women to men was 1.5 to 1. This gender distribution was similar to what has previously been reported in the literature, which indicated that anal cancer affects more women than men, showing a respective gender ratio of 1.4 to 1 in the U.S. population. 2 We found that after adjustment for the non-normal cost distribution and censoring due to disenrollment, the estimated per patient treatment cost for anal cancer was $127,531 in the first 2 years, or $63,766 per year. The literature on anal cancer treatment cost for the privately insured population in the United States is limited. Hu and Goldie applied a Canadian costing study of rectal cancer to estimate the discounted lifetime per patient cost of anal cancer at $36,300, adjusted  treatment. After diagnosis, the primary treatments for patients were combined chemotherapy and radiation. Those treatments accounted for the bulk of the cost. Treatment duration varied by the individuals who experience differences in disease progression. However, high costs in the first 6 months were also observed for other HPV-related cancer. [25][26][27] Following the primary treatments, regular examination (e.g., clinic visits or anoscopy) was provided, which resulted in stable costs during the continuing care phase of treatment. Deshmukh et al. defined the first 6 months after diagnosis as the initial phase and the last 12 months before death as the terminal phase and found that patients with anal cancer had the highest costs in the initial and terminal phases. 12 Our future studies will focus on estimating the lifetime cost to provide a more complete estimate of the treatment cost for anal cancer. Much of the cost and pain and suffering associated with anal cancer and other HPV-related cancers can be prevented by expanding the low HPV vaccination rates in the United States. Out-of-pocket cost is not a barrier because under the Affordable Care Act, HPV vaccination was assessed to be highly effective and cost-effective and, therefore, must be fully covered by qualified health insurance programs. 28 The vaccine is also covered fully by the Vaccines for Children Program. 29 Community-based pharmacies along with primary schools (through twelfth grade) have been identified as the most promising alternative sites for expanding the vaccination rates due to their convenience, experience with vaccinations, and generally high trust among the population. 6 These alternative sites are especially good for adolescent boys who make relatively few visits to primary care physicians. Ideally, pharmacists will collaborate with primary care physicians to educate and encourage parents to vaccinate their children in the most convenient and cost-effective settings.

Limitations
This study has a number of limitations. First, MarketScan CCAE data covered the population aged younger than 65 years that was insured by private insurance plans. Therefore, the results do not generalize to populations aged 65 years or older or people covered by Medicare or Medicaid. Second, since the death information in the MarketScan database was not comprehensive, an estimate of the per patient cost for the terminal phase of illness was not feasible. Third, cancer stage was not available in the database, which precluded analysis of cost by stage at diagnosis. Fourth, data from anal adenocarcinoma may skew the results. However, information on pathology confirmation to differentiate squamous cell carcinoma and anal adenocarcinoma was not available in the database.
Despite these limitations, this study used real-world claims payment data, which are broadly representative of the U.S. privately insured population.