: The costs of diagnostic, treatment, and end-of-life phases for advanced squamous cell carcinoma of the head and neck (SCCHN) are unknown, particularly after the approval of the first biologic agent for treatment, cetuximab. We aimed to characterize costs for each of the 3 disease phases and total cost of care for patients with advanced SCCHN before and after the approval of cetuximab.

: Retrospective, observational, cohort study.

: This was a retrospective analysis of the PharMetrics Choice insurance claims database. Patients 20 years or older with International Classification of Diseases, Ninth Revision, Clinical Modification diagnostic codes suggestive of advanced SCCHN between March 1, 2003, and March 1, 2008, were included. Patients were divided into 2 cohorts by date of diagnosis: a “prebiologic” period and a “postbiologic” period. Descriptive statistics were used to summarize baseline patient characteristics, monthly and total medical costs, and cost drivers. The Mann-Whitney U test was used to compare medical costs between segments and cohorts.

: A total of 365 patients with SCCHN met the study criteria. Median monthly medical costs were: diagnosis ($2199), treatment ($4161), end-of-life ($6614), and total ($4167). Total medical costs were driven primarily by outpatient costs (23%), inpatient costs (18%), and the cost of radiation therapy (16%). In the treatment segment, median monthly costs were significantly lower in the prebiologic era compared with the postbiologic era ($3301 vs $4381; P = .0024).

: Median total costs exceeded $4000 per month for patients with SCCHN, driven primarily by the end-of-life segment. The median monthly costs of the treatment segment increased by approximately $1000 following cetuximab’s approval.

Head and neck cancers account for approximately 3% of all cancer diagnoses and 2% of cancer-related deaths in the United States.

 In 2015, an estimated 45,780 Americans were diagnosed with head and neck cancers, of which 90% were of squamous cell histology.

Currently, the estimated 5-year survival rate for all patients diagnosed with squamous cell carcinoma of the head and neck (SCCHN) is 60%. Of those diagnosed with SCCHN, 60% are diagnosed with locally or regionally advanced disease, generally denoting a poorer prognosis.

Fortunately, recent data from the American Cancer Society demonstrate a steady annual decrease in SCCHN mortality of approximately 2% for women. Mortality in men has since stabilized.

The decrease in mortality is likely attributable to advances in the management of head and neck cancers.

 The National Comprehensive Cancer Network clinical practice guidelines recommend combined modality therapy for patients diagnosed with locally or regionally advanced disease.

Currently, first-line recommendations include combining radiation therapy with chemotherapy (eg, cisplatin) or with biologic therapy.

 Cetuximab is a biologic therapy, an epidermal growth factor receptor inhibitor approved in 2006 for use in patients with locally or regionally advanced SCCHN in combination with radiation therapy.

Second-line options include surgery, radiation (if not previously given), or systemic therapy with various single-agent chemotherapeutic options: cisplatin, carboplatin, docetaxel, cetuximab, or fluorouracil.

With the rising cost of cancer care in the United States, new therapies must be critically evaluated in regard to their economic value. Multiple factors, including extent of care, advancements in diagnostic capabilities, and increases in cancer prevalence rates, have contributed to this rise in recent years.

 The growing use of newer drug therapies, such as biologic agents, is often cited as one of the primary drivers of increasing cancer costs.

 However, available cost analyses of head and neck cancers have been limited to time periods preceding approval of these newer treatment modalities.

Differing patterns of care across disease segments (ie, diagnosis, treatment, end-of-life) make it difficult to quantify the costs of managing specific cancers

; however, a longitudinal evaluation of incidence-based direct medical costs for specific cancers can be used to estimate long-term costs for managing patients with cancer.

 The objective of this study was to describe the segmented (diagnosis, treatment, end-of-life) and total direct medical costs associated with the management of SCCHN.

This was a retrospective, observational, cohort study utilizing data from a private payer claims database (PharMetrics Choice) spanning the period from September 1, 2002, through March 1, 2010. The study was approved by the institutional review board at the University of Texas Health Science Center at San Antonio.

The PharMetrics database is an integrated patient-level database of pharmacy and medical claims from more than 90 commercial health plans in the United States. The database is representative of commercially insured patients across the United States and captures various claims rendered, such as laboratory and diagnostic tests, outpatient physician visits, hospitalizations, outpatient clinic visits, and pharmacy-related costs.

The PharMetrics Choice database (derived from the PharMetrics database) excludes claims for carve-outs, patients who lack integrated pharmacy and medical benefits, plans that have failed to submit data for the previous 2 years, and plans that lack membership tables. In 2010, more than 55 million patients were included in the PharMetrics Choice database.

All patients 20 years or older and enrolled in an eligible health plan for more than 1 day between March 1, 2003, and March 1, 2008, (the enrollment period) were considered for initial sampling. Individuals with advanced or locally advanced SCCHN were identified using an algorithm developed for use with healthcare claims data (Figure 1).

Diagnosis of new SCCHN cases was determined by the presence of at least 1 inpatient or 2 outpatient claims suggestive of head and neck malignancies at sites with a high proportion of squamous cell histology. These claims were identified using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes 140.X to 148.X, 161.X, 195.0, and 196.0. The date of the first claim for SCCHN was assigned as the index date.

Eligible patients were required to have 6 months of continuous enrollment preindex date and 24 months of continuous enrollment postindex date to account for patient progression through all segments of disease. Additionally, patients were required to have claims and procedure codes indicative of receipt of therapy for squamous cell histology and recurrent and/or advanced disease (ie, stage II, III, IV). Patients with another primary malignancy, as evidenced by at least 1 inpatient or 2 outpatient claims, were excluded.

Variables evaluated at the time of the index date included patient age, sex, geographic region, and National Cancer Institute (NCI) Comorbidity Index score. This score was derived from the Charlson Comorbidity Index and numerically represents the risk of impending mortality according to the presence of various chronic diseases at the time of cancer diagnosis.

 Additional data, such as race, ethnicity, tobacco and alcohol use, and human papilloma virus status, were not reported for the study population due to a lack of reporting of such information in the claims database.

Patients were followed from the index date until patient death, end of continuous enrollment, or end of the study period (March 1, 2010). Patient follow-up was evaluated in total and as segments (ie, diagnosis, treatment, and end-of-life). These 3 disease segments were derived from prior published studies evaluating the economic burden of cancer across phases of care.

The diagnosis segment was defined as the period 90 days immediately preceding the index date plus the time until the receipt of active treatment, identified as claims for surgery, radiation therapy, biologic therapy, or chemotherapy.

The treatment segment consisted of all claims identified following the initiation of active treatment until the study's end or the start of the end-of-life segment. The end-of-life segment was defined as the period 90 days preceding death. As reporting of mortality in healthcare claims is limited, we used a previously published algorithm for identifying deaths within insurance claims databases.

Patients with claims ending prior to the study end (March 1, 2010) and coding of a “likely fatal event” (eg, hospitalization, hospice, emergency department [ED], resuscitation, or cardiac arrest) within 30 days of the last date of active enrollment were considered to have died. Those patients who did not experience a likely fatal event during the study period would have remained in the treatment phase and not experienced the end-of-life phase.

The primary outcome was direct medical costs of SCCHN from the payer perspective, according to the specific disease segment: diagnosis, treatment, and end-of-life. Cost estimates were based on direct medical costs paid and included total reimbursed amounts and all paid medical and pharmacy claims.

Claims for hospitalization, ED visits, outpatient physician visits, outpatient treatments, skilled nursing care, hospice direct costs, radiation therapy, and drug costs made during the follow-up period were included. Costs were reported as monthly costs for each segment. Secondary outcomes included total monthly cost, cost per disease segment and total cost, percentage of costs for individual claims, and cost comparison among the prebiologic and postbiologic cohorts.

The approval of new treatments allowed stratification across 2 time periods. Patients were categorized into cohorts surrounding the approval of the biologic agent, cetuximab. Patients with an index date between March 1, 2003, and February 28, 2006, were assigned to the prebiologic cohort whereas patients with an index date on or following March 1, 2006, were assigned to the postbiologic cohort. Costs were compared among individuals according to prespecified cohort.

Additionally, those patients who experienced all segments of care (ie, diagnosis, treatment, and end-of-life) were assessed separately. This analysis was performed to evaluate costs incurred in patients with a likely fatal event, as these costs may differ from those for patients experiencing only the first 2 of the 3 segments of care associated with advanced SCCHN. These patients were referred to as the “full-care” group.

Categorical data were reported as numbers and percentages. Data were compared using the 

2 method. Continuous data were reported using descriptive methods and compared across cohorts using the Mann-Whitney U test. All costs were adjusted to 2009 US dollars based on the medical services component of the Consumer Price Index and were presented as medians and interquartile ranges (IQRs).

All statistical analyses were conducted using JMP version 9.0 (SAS Corporation; Cary, North Carolina).

Initial sampling identified 38,601 patients with head and neck malignancies. A total of 365 patients met the study inclusion criteria for SCCHN: 100 in the prebiologic period and 265 in the postbiologic period (Figure 1). The eAppendix (available at www.ajpb.com) describes patient demographics for the study population in composite and among the prebiologic and postbiologic cohorts. Patients were predominately male (78%), with a median age of 57 years. The mean NCI Comorbidity Index score was 0.53 (SD = 1.04).

The median duration of follow-up after the index date for all patients with SCCHN was 30.1 months (eAppendix), predominated by the treatment segment (median = 27.3 months). Overall, the total median cost incurred for patients with SCCHN was $110,793 (IQR, $73,769-$156,836) (Table 1). The median cost for care of patients with SCCHN incurred in this study was $4167 per month over the follow-up period.

The end-of-life segment was associated with the highest monthly cost (median = $6614; IQR, $3172-$17,158), followed by the treatment segment (median = $4161; IQR, $2540-$7915); however, total costs (median = $99,247; IQR, $63,117-$142,653) were highest in the treatment segment due to the longer duration of follow-up. The diagnostic segment incurred the lowest total (median = $6341; IQR, $2749-$11,121) and monthly (median = $2199; IQR, $1077-$4677) costs.

Outpatient clinic visits, radiation, and inpatient therapy costs were the 3 main cost drivers, representing 23%, 18%, and 16% of total medical costs, respectively. During the treatment segment, outpatient visits (23%) were the primary cost driver (

Radiation therapy, when combined with intensity-modulated radiation therapy (IMRT), accounted for 39.5% of the total costs incurred. Composite oncology drug (chemotherapy plus cetuximab) claims accounted for 7% of treatment costs, with chemotherapy accounting for 2.6% and biologics 4.4%. Surgery accounted for only 0.6%.

Patient characteristics were similar for prebiologic and postbiologic cohorts, with the exception of geographic region (P <.0036) (eAppendix). The overall duration of diagnostic and end-of-life segments (if present) were similar, with median durations of 2.8 and 2.9 months, respectively (Table 2). Patients in the prebiologic period had a significantly longer median duration of treatment, resulting in a longer observation period compared with the postbiologic period (29.1 vs 26.8 months; P = .0023).

Median monthly costs in the treatment segment were increased in the postbiologic cohort compared with the prebiologic cohort ($3301 vs $4381; P = .0024), whereas costs in the diagnostic and end-of-life segments were not statistically different (Table 2). Total monthly costs were higher in the postbiologic cohort compared with the prebiologic cohort ($3236 vs $4421; P = .0006). Medical claims between cohorts remained unchanged, with the exception of IMRT (8% vs 14%; P <.0001), chemotherapy (4% vs 2%; P = .001), and cetuximab (0% vs 4%; P <.0001) use.

Twelve percent of patients (n = 44) in this study were included in the full-care group: 14 prebiologic patients and 30 postbiologic patients. Baseline characteristics of patients in the full-care group were similar to patients in the non—full-care group: median age (58.5 vs 57 years; P = .0943), percentage male (81.8% vs 77.8%; P =.5521), and mean NCI comorbidity index score (0.73 vs 0.50; P = .1476). Full-care patients in the prebiologic and postbiologic cohorts had similar baseline characteristics.

Median monthly costs in the full-care group according to segment were: diagnostic segment ($1733; IQR, $660-$3614), treatment segment ($8265; IQR, $3946-$17,851), and end-of-life segment ($6614; IQR, $3172-$17,157). Total median costs for the follow-up period were $122,041 (IQR, $73,582-$168,461), with the highest proportion of costs incurred in the treatment segment ($89,281; IQR, $46,598-$121,067).

In comparing prebiologic and postbiologic cohorts, median monthly costs and median total costs were similar across all disease segments. The total median monthly cost for the follow-up period ($7817; IQR, $4053-$12,243) was higher compared with the non—full care group, ($3889; IQR, $2549-$6757; P = .0002).

To our knowledge, this is the first study to characterize the segmented direct medical costs of SCCHN across disease segments during the era of biologic therapy. The results of this study illustrate the economic impact of SCCHN across different segments of care and capture the impact of cetuximab by evaluating the time surrounding the approval of this agent.

Earlier studies evaluated the cost of head and neck cancers utilizing data from the 1990s.13,14 In a study by Lang et al, the economic cost of SCCHN in patients 65 years or older used data collected from the Surveillance, Epidemiology, and End Results database.

The study results demonstrated a significant increase in the median cost of SCCHN compared with controls ($1428 vs $446 per month; P < .001), with most costs incurred from hospitalization. The investigators found that patients with regional and distant disease had higher average costs ($2059-$2547); however, chemotherapy and radiation accounted for only 8% and 10% of regional and distant disease costs, respectively.

In comparison, the median cost of care of patients with SCCHN incurred in this study was $4167 per month over the follow-up period. When converted to 2009 dollars, the median cost per month incurred for advanced-stage disease in the study by Lang et al

 would be approximately $3300, which is comparable with the costs incurred in our study.

Although our study is the first to our knowledge to evaluate the segmented costs of SCCHN, other studies have conducted similar analyses in other malignancies.

 Song et al evaluated the costs incurred for metastatic colorectal cancer during the era of biologic therapy.

Articles

Original Research

The cost of squamous cell cancer of the head and neck is driven primarily by the end-of-life phase. Treatment costs increased following the approval of cetuximab.

Cost of Squamous Cell Cancer of the Head and Neck

We are in the midst of a massive shift when the government, private sector, and all stakeholders within healthcare are trying to move us from a system based on reimbursing for the volume of services to one that is pinned to the value of those services.

One component of this shift in the pharmaceutical arena is the practice of developing value-based contracts, under which biopharmaceutical manufacturers and payers agree to link coverage and reimbursement levels to a drug’s effectiveness and/or how frequently it is used.

To date, in the United States, value-based contracting activity has been limited; however, the desire to ensure patient access to important treatments is driving interest in agreements among both payers and biopharmaceutical manufacturers, and our changing healthcare environment may generate more activity in this arena in the future. Nevertheless, activity could remain limited unless we address several operational barriers, as well as regulatory and communications concerns.

Compelling Reasons to Engage in Value-Based Contracting

There are convincing arguments for manufacturers and payers to both consider value-based contracts: 1) manufacturers can use value-based agreements to differentiate and demonstrate the effectiveness of their product versus their competitors, which can assist payers in making formulary decisions; 2) payers can use these agreements to gain experience with a product, reducing uncertainty regarding clinical value, performance and financial impact.

There are a variety of agreements that have been struck between payers and manufacturers. Although most are based on providing discounts to payers if the drug does not meet a particular clinical endpoint, the contract details are all different. In one agreement, for example, Novartis will provide Harvard Pilgrim with a discount if its heart failure drug Entresto does not provide a given level of reduction in hospitalizations for congestive heart failure.

 In another example, Cigna reached independent agreements with Sanofi and Amgen for their PCSK9 inhibitor drugs, but they share the same overall objective.

According to the agreement, “[i]f Cigna’s customers aren’t able to reduce their [low-density lipoprotein cholesterol] LDL-C levels at least as well as what was experienced in clinical trials, the 2 biopharmaceutical companies will further discount the cost of the drugs. If the drugs meet or exceed expected LDL-C reduction, the original negotiated price remains in place.”

 At the same time, Eli Lilly and Company and Anthem have forged a partnership to begin a public dialogue about ways to move toward an “environment conducive to allowing health plans and manufacturers to enter into a variety of value-based contracting arrangements, aligned with the shift toward value-based payment and the goal of promoting access to high-value care.”

Challenges With Structuring and Implementing Value-Based Contracts

There are many difficulties inherent in implementing value-based contracts, particularly those based on particular outcomes or indications of a drug. These fall into 3 main areas: operational, communications and regulatory.

. Significant payer and provider infrastructure is necessary to monitor an outcomes-based agreement. For example, payers may not have the capability to monitor and track pharmaceutical usage and outcomes relevant to the contract due to a lack of a robust information technology, data infrastructure, or staff expertise, among other issues.

Without proper tracking mechanisms in place, it can be difficult to ensure medication adherence. Manufacturers are reluctant to accept financial consequences when they cannot control how a drug is prescribed or used. In addition, agreeing on outcomes that are meaningful and measurable within a reasonable timeframe is challenging.

. Payers may be interested in lowering expenditures for a particular patient population, such as reducing hospital visits and costs. Yet, this type of clinical or economic outcome might not be included on a drug’s label, which must be approved by the FDA.

As a result, payers and manufacturers cannot agree to a contract based on an outcome that is not listed on the drug’s label, even if strong data exists for that outcome, and this limits their ability to effectively share risk. FDA is currently reviewing how it regulates the exchange of information between payers and manufacturers, so this concern could be addressed in the near term.

. Medicaid Best Price was enacted as a way to keep federal spending in check and guarantee that Medicaid is getting the lowest price that the manufacturer can afford to sell it, which is roughly one-fourth of a drug’s average manufacturer price.

Manufacturers are reluctant to develop value-based contracts that might result in an effective greater discount than Medicaid Best Price, because the lower “discount” or price would apply not only to the population in the agreement, but to the entire Medicaid population as well.

Fear of inadvertently running afoul of the federal anti-kickback statute, which prohibits the exchange or offer of anything of value in an effort to reward the referral of business, is another reason some manufacturers and payers shy away from value-based contracting.

Value-added services could include activities such as nurse coaches, case management support or adverse event monitoring, all of which have benefits in the administration of a value-based contract. The federal government would need to provide clarity to address this situation.

Given the need to provide patients with access to effective medications, it is likely that we will see more value-based contracts in the future. This expansion could be fueled by: 1) improved data systems, which can help create the mechanisms to monitor patient usage of biopharmaceuticals and their effect on health outcomes, and 2) the changing of incentives within the healthcare system toward value-based orientations.

The increased use of accountable care organizations, for example, may lead to enhanced health system alignment with value-based agreements. Yet for value-based contracts to truly take off, we will need to address some key operational, communications and regulatory barriers sooner, rather than later.

: National Pharmaceutical Council, Washington, DC.

: Mr Leonard is an employee of the National Pharmaceutical Council, an industry-funded health policy research group that is not involved in lobbying or advocacy, and an officer of the Board of National Pharmaceutical Council.

: Concept and design (DL); drafting of the manuscript (DL); critical revision of the manuscript for important intellectual content (DL); and supervision (DL).

: Dan Leonard, MA, President, National Pharmaceutical Council, 1717 Pennsylvania Ave, NW, Washington, DC 20006. E-mail: dleonard@npcnow.org.

1. Garrison LP, Carlson JJ, Bajaj PS, et al. Private sector risk-sharing agreements in the United States: trends, barriers, and prospects. Am J Manag Care. 2015;21(9):632-640.

2. Pay-for-performance deal with Novartis [press release]. Wellesley, MA: Harvard Pilgrim HealthCare; June 27, 2016. 

https://www.harvardpilgrim.org/public/news-detail?nt=HPH_News_C&nid=1471911695877

3. Cigna’s two new value-based contracts with pharma for PCSK9 inhibitor cholesterol drugs tie financial terms to improved customer health [news release]. Bloomfield, CT: Cigna; May 11, 2016. 

https://www.cigna.com/newsroom/news-releases/2016/cignas-two-new-value-based-contracts-with-pharma-for-pcsk9-inhibitor-cholesterol-drugs-tie-financial-terms-to-improved-customer-health

4. Eli Lilly and Company; Anthem. Value-based contracting arrangements. Eli Lilly and Company website. 

https://lillypad.lilly.com/WP/wp-content/uploads/LillyAnthemWP2.pdf

. Published January 29, 2016. Accessed May 4, 2017. 

5. Drug and device manufacturer communications with payors, formulary committees, and similar entities—questions and answers; draft guidance for industry and review staff; availability. Federal Register website. 

https://www.federalregister.gov/documents/2017/01/19/2017-01011/drug-and-device-manufacturer-communications-with-payors-formulary-committees-and-similar

. Published January 19, 2017. Accessed May 4, 2017.

6. Sachs R, Bagley N, Lakdawalla D. Value-based pricing for pharmaceuticals in the Trump administration. Health Affairs blog website. 

http://healthaffairs.org/blog/2017/04/27/value-based-pricing-for-pharmaceuticals-in-the-trump-administration/

. Published April 27, 2017. Accessed May 4, 2017.

Commentary

Given the need to provide patients with access to effective medications, it is likely that we will see more value-based contracts in the future; however, some key operational, communications, and regulatory barriers will need to be addressed first.

Opportunities and Challenges With Value-Based Contracting

: The heterogeneity of treatment effect (HTE) refers to the nonrandom variability in response to treatment. This study describes how US payers use HTE evidence when formulating coverage policies for oncology drugs.

: We employed a qualitative approach using semistructured, in-depth interviews with 15 payers.

: An interview guide was developed based on theory and pilot interviews. Themes that emerged from this content analysis were summarized along with verbatim quotes from payers.

: All payers agreed that HTE is important to incorporate into coverage policies of oncology treatments. The FDA label is the most overwhelming determinant of whether HTE evidence gets incorporated into a coverage policy. If not in the FDA label, payers find it difficult to use HTE evidence due to the inability to precisely differentiate responders from nonresponders and the logistical difficulty to operationalize HTE. All payers reported that randomized controlled trials are the most trusted source to establish HTE evidence. In addition to the FDA label, payers also consider treatment guidelines, the quality/magnitude of HTE evidence, the availability of effective alternative substitutes, treatment line, cancer aggressiveness, and politics. When a biomarker and a companion diagnostic are involved, the degree to which HTE evidence is incorporated into coverage policies will also depend on the clinical and analytic validity of the test and the ability to accurately and pragmatically distinguish responders from nonresponders.

: Payers’ oncology coverage decisions are affected by a myriad of factors. Payers require more definitive HTE evidence to make more efficient coverage decisions.

The FDA approval of costly anticancer medications that yield a modest incremental survival benefit has become more prevalent in the past few years, with the price range reaching $10,000 per month in 2013 compared with $4500 per month in 2003.

 Between October 2012 and October 2013, the FDA approved 9 anticancer drugs.

Although this significant number of new drug approvals in a 12-month period indicates increased drug development activity and improvements in regulatory processes (eg, accelerated approvals), most approved drugs were associated with high price tags and a modest survival benefit.

Healthcare payers are facing great challenges when making coverage decisions, with uncertainty surrounding the value of oncology treatments specific to their plan members.

 Therefore, it is important that payers appropriately assess patient heterogeneity in response to treatment and use that information to inform coverage policies for oncology drugs.

Heterogeneity of treatment effect (HTE) refers to the nonrandom variability in the response to a treatment among different individuals given their characteristics. Kravitz et al define HTE as the “magnitude of the variation of individual treatment effects in the population”

; they further explain that, when HTE is present, the modest average effects realized in clinical trials often reflect a mixture of patients, where a group experiences the average effect, others experience less than or more than the average effect, and some may be harmed by treatment.

Failing to account for HTE could mean that healthcare payers may initially be paying for treatments that are ineffective, and for additional downstream costs attributed to initially ineffective or harmful therapies.

 If HTE evidence is used efficiently, payers can spare patients unnecessary adverse effects if they are not likely to clinically benefit and encourage the use of appropriate therapies in patients with markers suggestive of good response, such as patients with acute lymphoblastic leukemia (ALL) who have a positive Philadelphia chromosome and would respond better to imatinib.

Strategies have been incorporated by healthcare payers to restrict access to anticancer medications depending on an individual’s response.

 For example, if a biomarker predicts whether a drug is expected to work, payers may decide to pay for that test to determine whether the drug should be given.

Subsequently, payers would use the test result to determine whether to reimburse for the drug where a negative finding typically would not be reimbursed. Similarly, if a specific biomarker predicts duration of therapy, payers would test that biomarker and reimburse for the duration of therapy as indicated by the test result. Step-therapy programs have been used if a marker predicts which drugs should be used first from several alternatives.

Although restriction strategies have been commonly used in coverage policies, the process by which HTE evidence informs the implementation of these strategies in coverage policies is understudied. The purpose of this study was to further understand the payers’ decision-making process, particularly to elicit their views on how they use HTE evidence when formulating coverage policies for oncology drugs.

A qualitative approach using semi-structured in-depth interviews was employed to answer our research question.

Payers were selected using a 2-stage sampling approach. In the first stage, we used an expert sampling method (a specific case of purposive sampling) and contacted 20 payers through our professional network.

 We approached payers based on the following criteria: 1) they had a pharmacy or medical background, 2) they had been involved in pharmacy and therapeutics (P&T) committee meetings in their organization, and 3) they had reviewed HTE evidence in their organization to any extent. We received responses from 18 payers (90% response rate), of whom 16 satisfied the criteria and were interested in participating; of these 16 payers, 3 initially agreed but then dropped out.

The second stage—snowball sampling&mdash;was used with payers recruited from the first stage to further recruit other payers to increase our sample size.

 Fifteen payers participated in this study.

Our sample represented payers with pharmacy and medical backgrounds and a variety of payer groups. Although it is true that many P&T committee members are not oncologists and do not have detailed knowledge of oncology, the health plans make sure that there are oncologists on their committees as either standing or ad hoc members to assist the committee in their exploration of new oncology agents.

We developed an interview guide containing 3 sections: orientation, interview questions, and closing remarks (

). The orientation included asking payers whether they agreed to be voice recorded, provided a rationale and context of why this study was important, and provided an orientation to technical terms. Payers were also asked if they had any questions before the interview proceeded. The interview guide had 4 types of questions: background, high-level, detailed, and closing. The third section involved closing remarks to thank interviewees for their time.

The interview questions were developed to understand the factors that affect the value judgments involved when making a coverage decision based on the analysis of a decision framework developed by Eddy.

 Pilot interviews were conducted to narrow the focus of the questions and to ensure their relevance to oncology. Although the interview guide was semi-structured, it was not uncommon that interviewees would switch the order of questions according to what they felt was most important to address during the interview.

 Interview questions were sent to the interviewees beforehand to ensure that they understood the questions and were prepared during the interview.

Each interview was scheduled to last approximately 30 to 60 minutes. During the interview, payers were provided with a decision-making framework and asked in question 5 (Table 1) to modify it to reflect real-world decision making. After the first interview, and in cases where major modifications took place, subsequent interviewees reviewed the updated conceptual framework as informed by the preceding interview.

The content analysis involved reviewing the written notes from the interviews and complementing them with the audio recordings. Major thematic areas that emerged from the answers were coded and categorized under 1 of the 4 domains: 1) the importance of incorporating HTE evidence in coverage policies of anticancer drugs, 2) the factors that would impact the decision to restrict access to anticancer drugs, 3) sources of HTE evidence, and 4) the future of incorporating HTE evidence into coverage policies. Themes that emerged from the analysis were summarized along with verbatim quotes from payers to support each theme.

Our target was to recruit 12 to 15 individuals representing the various payer groups. Of the 20 individuals who were contacted, 15 were interviewed. The final sample included 15 payers and related organizations representing 11 payer groups located in the United States providing health insurance to 172.5 million individuals (Table 2). The following section describes the themes that emerged during the interviews and are supplemented with verbatim quotes from payers (Table 3).

Payers’ Views on the Role of Incorporating HTE

Theme 1: HTE is important to incorporate into coverage policies, yet most payers do not delve into HTE details unless it is specifically indicated in the FDA label.

All payers were familiar with the concept of HTE (although not the term) and its application in coverage policies. They unanimously agreed that HTE has a critical role in defining coverage policies in oncology; however, there is variation in the extent to which they incorporate HTE evidence in coverage policies.

Ten payers indicated that they do not delve into the details of HTE unless it is in the FDA label. Two of the 10 backed this approach explicitly, responding, “If HTE was not indicated in the FDA label, it was hard to justify a coverage policy that incorporates HTE.”

Small-sized payers indicated that they frequently rely on their pharmacy benefit managers’ recommendations. One payer indicated that they are heavily regulated and legally required to cover anticancer drugs. Only 1 payer stated that they are very sophisticated in their approach and have in-house technology assessment groups that thoroughly investigate HTE.

 Theme 2: The inability to precisely differentiate responders from nonresponders and the logistical difficulty to operationalize HTE are reasons why most payers do not focus on HTE. The payers interviewed for our study did not believe that science has advanced far enough that it can be used to clearly differentiate responders from nonresponders. For example, 1 payer gave the example that a negative finding from testing a biomarker does not exclusively mean lack of benefit; instead, it indicates that a patient will not benefit as much compared with a patient with a positive result.

When asked about clinical and sociodemographic factors, all payers agreed that although it does not make sense biologically to differentiate responders from nonresponders, they are used in prior authorizations. For example, Eastern Cooperative Oncology Group (ECOG) performance status is used by many payers in coverage policies for patients taking sipuleucel-T for metastatic prostate cancer.

Patients must be in the ECOG 0 or 1 group to qualify for coverage. Additionally, for large payers, it is logistically difficult to come up with different workloads and clinical pathways for all patients. Therefore, P&T committees trust that physicians will steer patients to their most effective treatment option without searching for HTE evidence and incorporating it into coverage policy.

Payers’ Views on Sources of HTE Evidence

Theme 3: Randomized controlled trials (RCTs) are the most trusted source to establish evidence on treatment effectiveness, and this extends to evaluating HTE evidence. All payers agree that well-conducted RCTs in peer-reviewed journals remain the most reliable source for establishing HTE evidence. However, 5 payers cautioned that RCTs cannot answer all types of HTE questions. For example, due to the narrowly defined population included in RCTs, some HTE questions cannot be answered, such as, “How does the treatment effectiveness vary by comorbidity status?”

Interestingly, none of the payers were able to weigh in on the statistical methods as proof of HTE within RCTs (eg, interaction terms, growth mixture models); however, post hoc subgroup analyses seemed to be the most familiar method. Questions that require longer follow-up would be expensive to answer using RCTs, in which case observational studies may be of importance.

In addition to RCTs, all 15 payers used some combination of Medicare-approved compendia, observational studies, systematic reviews, meta-analyses, and pragmatic clinical trials. Three payers indicated that observational studies are not enough to support coverage decisions, but are more compelling as evidence to support appeals.

Theme 4: Although most payers agreed that biomarkers are most useful in predicting response to treatment, all payers stressed that the quality and magnitude of evidence determine the usability of any predictor.

There is a continuum of the predictive value of the predictor categories (ie, biomarkers are generally more useful than clinical factors) that are more useful than sociodemographic factors in predicting response to treatment. However, the quality and magnitude of evidence is always the key to the usability of these factors. Payers unanimously indicated that they prefer the “cut and dry” predictors and tend to avoid the controversial gray area.

One reason why payers think biomarkers are more appealing is that they are more related to the pathophysiology of the disease and the mechanism of action of the drug compared with other factors. Five payers added that there is a lot of HTE evidence that is “no good” and does not give a clear-cut answer.

 The consensus was that if there is not good enough HTE evidence, payers would be less likely to use the evidence until more definitive HTE evidence arises. Others indicated that in oncology, payers must use the best available evidence. For example, prior to the breast cancer Oncotype DX panel, 2 payers indicated that they used the patient’s age and tumor size at point of surgery for prior authorization criteria.

Payers’ Views on Factors That Could Affect Coverage Restrictions of Oncology Drugs

Theme 5: Factors that influence the general health technology assessment process also would come into play when assessing HTE evidence. The FDA label was the most overwhelming determinant for restricting coverage. All payers stated that it is almost impossible for an oncology drug to be covered for certain subgroups unless that specific indication was on the FDA labeling.

In the situation where the label has a companion diagnostic test built into it, payers found it easy to formulate a coverage policy based on the companion diagnostic. However, when the HTE was in relation to a sociodemographic feature, it was somewhat harder to translate that HTE information into policy. The example that was given by 2 payers was belimumab, a drug used to treat systemic lupus erythematosus.

Two RCTs demonstrated the effectiveness and safety of belimumab in patients with lupus, excluding previously treated patients and those with the more severe lupus.

 African American patients participating in the 2 RCTs did not appear to respond to belimumab, leading the FDA to caution against its use in this patient population in its labeling. Both payers indicated that they would still cover it for African Americans but would use a “watch and wait” approach.

Other highly weighted factors that affect the incorporation of HTE evidence into coverage policy are the oncology treatment guidelines (mostly National Comprehensive Cancer Network [NCCN]), the quality and magnitude of evidence (RCTs weighted more than other sources), the availability of effective alternative therapies, line of treatment, and the aggressiveness of cancer. Thirteen of the respondents were involved in the clinical evaluation team and not in the economic evaluation; therefore, most were reluctant to speak about the costs of drugs or companion diagnostic tests.

Theme 6: The degree to which HTE evidence is incorporated into coverage policies depends on the predictive value of the characteristic in question and the ability to accurately and pragmatically distinguish responders from nonresponders.

One of the most cited factors that would determine the degree to which HTE evidence can be incorporated into coverage policy is the predictive value of the characteristic in question. Three payers stated that testing for Philadelphia chromosome positivity in patients with ALL when taking imatinib is a good example of a characteristic that is highly predictive of response to treatment; it is also considered one of the “easier” decisions for payers to make.

The clinical and analytic validity of the test used to detect the presence of a specific genotype is one of the major factors that payers talked about. If a test can, to a reasonable extent of sensitivity and specificity, identify the specific genetic arrangement that is associated with improved or worsened response to treatment, then this test would be used to guide therapy in clinical pathways and would be covered.

One payer also stated that operationalization of the test at the point of prescribing can impact whether the test gets covered. The payer identified 3 questions to determine whether the test is easy or difficult to operationalize: is the test FDA approved?

How complicated is it to perform the test? How long to get the results? Another payer identified the background prevalence of the mutation that predicts response to treatment: when the background prevalence of the desired mutation is low, the payer indicated that it makes more sense to use the companion diagnostic test to identify those small numbers of patients (eg, crizotinib for non—small cell lung cancer).

 Another payer pointed out the concept of the physician practice gap. A physician practice gap is a clinically important difference between the distribution of treatment practices occurring in a given provider population compared with the pattern that would be expected in that setting if providers followed evidence-based best practice guidance. If physicians follow guidelines (eg, NCCN), then there is no need for a policy to restrict access; however, if there is a significant practice gap, a coverage policy becomes necessary.

When making coverage decisions in oncology, payers find difficulties in translating evidence on treatment effect heterogeneity into coverage policies.    

Payers' Views on the Heterogeneity of Treatment Effect in Oncology

The world is confused—no, this column is not about politics&mdash;rather, I asked 20 people I know what they believe a “pharmacy benefit” covered, and all but 1 of them stated that they believed “benefit” covered all the medications that they are taking. They did not say that it covered only those medications paid for by the pharmacy benefit or only those that are pills; they perceived this as a benefit that singularly addressed all of their medicines.

It is time that we listen to the consumer. I would suggest that those who make benefit decisions should start to think of the benefit the same way their employees and families think about it. Today it is often 3 different organizations that are overseeing, managing, and potentially distributing a person’s medications: one for traditional small molecule medications, a second that does similar activities for self-administered biologics, and a third that manages infused medications.

This is commonly known as the “best in class” model. A second model uses a 2-organization approach, with 1 organization to oversee, manage, and distribute the traditional pharmacy benefit medications and self-injectables (these are paid for under the pharmacy benefit), but they do not have any direct interaction with those medications that are infused and often delivered through the physician’s office or hospital (these are most often covered under the medical benefit).

Companies where you see this most often are organizations such as CVS Health or Express Scripts/Accredo. Recently, some of these organizations have also begun to interject themselves in the management of the infusible “medical benefit medications.” I would suggest that this is a model that is beginning to align itself with the public’s view of what a pharmacy benefit should be. I would also suggest that this management model still has a long way to go.

Although one company may be responsible for all the management of medications regardless of payment source, none of them have a total patient-centric view of the medications. In fact, they often have little or no overlap of clinical or administrative teams and often do not have a single view of the patient in their systems or data. It is time that the pharmacy world begins to address care coordination and management the same way we are asking the medical community to do so.

Imagine a patient that had multiple sclerosis with common comorbidities of diabetes and depression. In order for this patient to have optimal outcomes and feel their best, it is important that a single person or team be responsible for the entirety of the medication management.

The following are just a few things that should be considered with a patient such as this:

Are any of these medications negatively interacting?

Do any of these medications cause side effects that could impact 1 or more of the comorbidities?

Is the patient adherent to all their medications?

If the patient cannot afford to take all the medications, which are the most important?

The pharmacy benefit has begun to attract greater attention by the public over the last few years for a number of reasons, including the increase in medications that are available and the cost of many of these medications.

It is time we push ourselves and the benefit managers to evolve the model into a single integrated management (and when possible distribution) model. I believe this would better address the needs of the patient, as well as the needs of society. What do you think?

From the Editor

The pharmacy benefit has begun to attract greater attention due to the increase in medications that are available and their cost. It is time we refocus on the needs of the patient and push ourselves and the benefit managers to evolve into a single integrated management and distribution model.

It Is Time We Create a Pharmacy Benefit That Is Patient Centric

: We studied the independent association of adher- ence to oral hypoglycemic medications with poor glycemic con- trol among a population of adults with type 2 diabetes (T2D), adjusting for demographics, health behaviors, and clinical and treatment characteristics.

: This was a retrospective cohort design.

: We studied a population of Kaiser Permanente Northwest (KPNW) members with T2D who either had: 1) good glycemic control (glycated hemoglobin [A1C] <8.0%; n = 15,891) or 2) poor glycemic control (A1C >9.0%; n = 3709). The primary independent variable was medication adherence to 1 or more oral hypoglycemic medications. High medication adherence was defined as at least 80% of days covered in the 12 months prior to the A1C test date (yes vs no). Multiple logistic regression was used to analyze the independent association of medication adherence with poor glycemic control, adjusting for demographics, health behaviors, medical comorbidities, healthcare utilization, receipt of diabetes care management services, and intensity of diabetes treatments. All measures were constructed via KPNW’s electronic heath record.

: Increased adherence to oral hypoglycemic medications was associated with a lower likelihood (OR, 0.54; 95% CI, 0.50-0.59; P <.0001) of having poor glycemic control, after adjusting for demographics, health behaviors, comorbidities, healthcare utilization, receipt of diabetes care management services, and intensity of diabetes treatments.

: Higher adherence to oral hypoglycemic medica- tions is independently associated with a lower likelihood of having poor glycemic control among an adult population with T2D. Studies of the effects of measures to improve medication adherence on population-level glycemic control are needed.

Type 2 diabetes (T2D) is a chronic, progressive condition characterized by the failure of insulin secretion to compensate for resistance to insulin’s actions, resulting in hyperglycemia. The long-term complications of diabetes contribute to its status as a leading cause of premature illness and mortality.1 In 2012, 29.1 million individuals in the United States had diabetes (approximately 90% of the patients had T2D), and the prevalence appears to have increased in the past 2 decades.2 In 2012, diabetes resulted in $245 billion in costs, including $176 billion in direct medical costs and $69 billion in reduced productivity.3

The long-term control of hyperglycemia is essential in order to improve health outcomes for populations with diabetes. Poor glycemic control has been associated with microvascular complications4-6 and mortality,7,8 whereas good control reduces the development of microvascular and long-term cardiovascular complications.9 Current clinical guidelines to manage glycemic control utilize glycated hemoglobin (A1C) target levels, with adjustment for age and length of illness10; however, A1C levels greater than 75 mmol/mol (9.0%) are universally considered to reflect poor control.11 Of particular concern, according to data from a recent National Health and Nutrition Examination Survey, the proportion of adults with diabetes with poor glycemic control (A1C >9.0%) increased from 17.9% in the 2005-to-2008 period to 21.0% in the 2009-to-2012 period.12

Despite clinical evidence of the efficacy of several glucose- lowering medications, adherence levels often remain sub- optimal.13-17 Moreover, several studies have found that lower medication adherence can be associated with poorer glycemic control outcomes.14,18-23 However, these studies have several limitations, including small sample sizes and analyses not conducted on a population level.24,25

With this background, the primary objective of this study was to analyze the independent association of adherence to oral hypoglycemic medications with poor glycemic control among adults with T2D on a population level. Such research has significant clinical implications—given that medication adherence is a potentially modifiable factor&mdash;that may be influenced by quality improvement initiatives. A secondary objective was to identify demographics, health behaviors, comorbidities, healthcare utilization, diabetes care management services, and treatment intensity mea- sures associated with poor glycemic control, with an emphasis on factors that were modifiable.

The institutional review board of Kaiser Permanente Northwest (KPNW) in Portland, Oregon, approved the study protocol and waived the need for patient consent for data use, as is the protocol for data-only studies.

ThestudywasconductedatKPNW,anonprofit,group-model healthcare system in northwest Oregon and southwest Washington. The system consists of 15 medical clinics and 2 hospitals that serve more than 500,000 members. KPNW membership demographics (age, sex, race/ethnicity) are similar to the larger population in the geographic area. We used the electronic health record (EHR) (Epic Systems Corporation; Madison, Wisconsin) and administrative data systems to ascertain information on patient membership; demographics; clinical data, including height and weight; laboratory results; and medical healthcare utilization.

We conducted a retrospective cohort of KPNW members who met all the following inclusion

fasting glucose >125 mg/dL, nonfasting glucose >200 mg/ dL), or 3) 1 or more elevated A1C test results (>6.5%). In addition, those with T2D had an International Classification of Diseases, Ninth Revision (ICD-9)-related encounter (in any medical setting) as far back as membership existed. T2D ICD-9 codes include any of the following: 250.00, 250.02, 250.12, 250.20, 250.22, 250.30, 250.32, 250.40, 250.42, 250.52,250.60,250.62,250.70,250.72,250.80,250.82, 250.90, and 250.02.

Uncontrolled and Good Glycemic Control Groups

The process flow describing the study population is shown in the FIGURE. We identified 37,514 members who met the initial inclusion criteria. Of this population, 4759 members had an A1C value greater than 9.0% and formed the poor glycemic control group, while 27,514 had an A1C value less than 8.0% and made up the good glycemic control group. Another 5250 members with A1C values between 8.0% and 9.0% were not included in the analysis. The final analytic population (n = 19,600) included poor (3709) and good (15,891) glycemic control populations that received 1 or

criteria: 1) had an A1C test between July 1, 2014, and June 30, 2015; 2) were 18 years or older on the A1C test date; 3) were on the KPNW diabetes registry with a T2D indication at the time of the A1C test; and 4) had continuous health plan enrollment for 12 or more months prior to the A1C test date, which was defined as the index date. If the results of multiple A1C tests were available between July 1, 2014, and June 30, 2015, the latest value was used in the analysis.

KPNW members enter the diabetes registry through any of 3 ways: 1) receipt of treatment with oral hypo- glycemic medications, 2) 2 or more abnormal fasting or nonfasting glucose test results in the previous 3 years (eg, more oral hypoglycemic medications in the 12 months prior to the A1C test. Generally, adherence to oral hypoglycemic medications and other study covariates was measured in the year prior to the A1C test date, for a complete study period between July 1, 2013, and June 30, 2015.

. The primary dependent variable was poor glycemic control (A1C >9.0%) versus good glycemic control (A1C <8.0%).

. Adherence to oral hypoglycemic medications was the primary independent variable in the analysis. We used the proportion of days covered (PDC) to construct medication adherence—a measure well described in the literature26 and considered to be more consistent than another adherence measure commonly used with electronic pharmacy information, such as the medication possession ratio.27 The denominator included the total number of days between the first hypoglycemic medication fill and the end of the 365-day observation period, prior to the A1C test date. The numerator included total days covered by prescription fills during the denominator period. High medication adher- ence was defined as 80% or more days covered during the denominator period. If multiple medications were prescribed, a measure of total average days covered was constructed. No insulin use was included in the adherence measure.

. We identified 6 groups of covariate measures: demographics, health behaviors, comorbidities, healthcare utilization, use of diabetes care management services, and diabetes treatment-intensity measures. The rationale for selecting these measures was 2-fold. First, they are important factors previously associated with poor glycemic control in the literature. Second, adjusting for these measures on a population level increased the ability to better assess the independent association of medication adherence with poor glycemic control.

. Using KPNW’s EHR, we measured age (continuous: assessed upon A1C test date), sex (female vs male), race/ethnicity, and primary language status (non—English speaker: yes vs no).

Among a population of adult patients with type 2 diabetes, adherence to oral hypoglycemic medications is independently associated with poor glycemic control, even after adjusting for patient-level covariates.

May/June 2017

Cost of Squamous Cell Cancer of the Head and Neck

Opportunities and Challenges With Value-Based Contracting

Payers' Views on the Heterogeneity of Treatment Effect in Oncology

It Is Time We Create a Pharmacy Benefit That Is Patient Centric

Assessing the Association Between Medication Adherence and Glycemic Control