Correspondence: Developing a Model Framework to Assess Cost of Therapy

Web Exclusives - Letters to the Editor, In the News

THE AUTHORS RESPOND: In their letter, Abouzaid and colleagues (December 2015)1 criticized several of the assumptions made by Roy and colleagues (June 2015)2 during the development of the model framework designed to assess the cost of therapy and the cost of therapy per 12 months of progression-free survival (PFS). The assumptions made by Roy and colleagues were necessary to create a generalizable model framework, and they resulted in the introduction of several limitations, which were acknowledged and addressed in the original publication.2

The model framework was intended to calculate the budget impact over a theoretical 1-year time horizon from the perspective of a health plan, during which individual patients may be at different points in therapy (ie, on or off therapy). Because some regimens have a duration of treatment that is much less than 1 year (eg, bortezomib plus dexamethasone) and some much more than 1 year (eg, carfilzomib plus lenalidomide and dexamethasone), adjustments were required so that the treatments with a median duration of 3 months and a short corresponding PFS did not appear extremely unfavorable compared with treatments with a long median duration of treatment and a long PFS. To adjust the patient’s individual duration of treatment and PFS data to a more population-based payer perspective, the duration of treatment to PFS ratio was used.

Furthermore, we acknowledged in the original publication that the data used in the analysis are based on clinical trial data, whereas in a real-world setting, patients may switch to an alternative treatment on progression. However, individual treatment pathways are highly varied and depend on patient history,3 clinical characteristics, preference, tolerance, among other considerations, which would create a very complicated model. Moreover, as previously acknowledged, this model in no way intends to suggest that treatment of an individual patient for a shorter or longer duration of therapy than recommended is appropriate or will result in the PFS gains. This model was not intended to directly compare the efficacy of various treatment regimens, but rather to use a standardized outcome measure (ie, PFS) to compare cost relative to outcome.

We know that patients are administered some treatment at progression, and the model attempts to account for that in a simple and practical way. The Table lists the results using various assumptions regarding treatment length and cost. In our base-case analysis (Table, column A), the duration of treatment to PFS ratio is used to calculate how many treatment months in a year would be necessary from a population-based perspective. Column B in the Table lists the cost per month without progression, assuming that patients receive only 1 course of treatment with no adjustment made to annualize the cost; column C lists the cost per month without disease progression, assuming patients return to using a regimen equal in cost to the average across all regimens ($17,703 per month).


For example, a patient would experience 6.1 months of treatment with the bortezomib plus dexamethasone regimen, accompanied by 8 months of PFS. After 8 months, the remaining 4 months of the year (to total 12 months), the patient would be treated with an average regimen cost of $17,703 per month and would remain progression-free. As shown in the Table, the results remain robust to varying assumptions regarding treatment, with the cost per month without progression for patients treated with panobinostat plus bortezomib and dexamethasone lower than all comparators except bortezomib plus dexamethasone, regardless of what assumptions were made for the cost of treatment after progression.

In their letter, Abouzaid and colleagues also mentioned a lack of adjustment for differences in underlying patient population. Clinical data for PFS and the duration of treatment for each comparator were taken from a population with relapsed or refractory disease (where available). Given that the patient populations of the clinical trials for all comparators were conducted in populations with relapsed or refractory disease, it was assumed that the impact of any underlying differences in the patient populations, particularly those relating to previous lines of therapy, would be negligible. We recognize the heterogeneity and corresponding differences that may be introduced through this simplifying assumption, and we consequently note that the results may favor regimens that have been studied in a less refractory population. Although therefore fully scientific robust comparisons are not being depicted by this model, we state as much and acknowledge these limitations.

Despite these limitations, the overall objective of the model was to provide a framework for consideration and future use. The inputs used in the analysis were secondary to the primary objective of detailing the methodologic framework. If adjusted efficacy estimates become available, such as through an indirect treatment comparison, they can easily be used as inputs into this framework.

The rates of adverse events (AEs) were retrieved from the same clinical trials that were used for the other outcomes in the model. Although the model assumes an equal likelihood of an AE each month of therapy, the calculation used for the standardization to a monthly percentage has the same overall AE rate as was reported in the clinical trial, thus not inflating or deflating the impact of the AE costs. Because no patients discontinue therapy early in the model, the results are not affected by whether all the AEs are likely to happen in the first month or throughout therapy. Furthermore, the standardization of the monthly AE percentage allows for flexibility in the model that would otherwise not be possible.

Finally, this model framework was not intended to calculate the cost offsets associated with any individual treatment regimen, and thus it did not consider the downstream economic outcomes associated with prolonged PFS or delaying disease progression.


1. Abouzaid S, Gibson C, Nagarwala Y. Cost of treatment for relapsed/refractory multiple myeloma. Am Health Drug Benefits. 2015;8(9):670.
2. Roy A, Kish JK, Bloudek L, et al. Estimating the costs of therapy in patients with relapsed and/or refractory multiple myeloma: a model framework. Am Health Drug Benefits. 2015;8(4):204-215.
3. National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines): multiple myeloma. Version 3.2016. September 22, 2015. Accessed January 15, 2016.


Anuja Roy, PhD, MBA
Denise Globe, PhD
Emil T. Kuriakose, MD
Novartis Pharmaceuticals, East Hanover, NJ

Jonathan K. Kish, PhD, MPH
Lisa Bloudek, PharmD, MS
Kristen Migliaccio-Walle, BS
Xcenda, Palm Harbor, FL

David S. Siegel, MD, PhD
Hackensack University Medical Center, NJ

Sundar Jagannath, MD
Tisch Cancer Institute, Mount Sinai Hospital, New York

View the Letter to the Editor

Related Items
Dose Modification of Subcutaneous Tocilizumab in Patients with Rheumatoid Arthritis
May 2020 Vol 13, No 2 published on May 20, 2020 in Letters to the Editor
CMS Expands Access to Telehealth Benefits During COVID-19 Outbreak
Web Exclusives published on March 19, 2020 in Health Policy and Reform, In the News
Tazverik Receives FDA Approval as First Treatment Specifically for Metastatic or Locally Advanced Epithelioid Sarcoma
Web Exclusives published on January 28, 2020 in FDA Approvals, In the News, Select Drug Profiles
Mixed Findings in Annual Cancer Statistics Report
Web Exclusives published on January 21, 2020 in FDA Approvals, In the News, Select Drug Profiles
FDA Grants Approval to Avapritinib for Patients with GIST
Web Exclusives published on January 14, 2020 in FDA Approvals, In the News, Select Drug Profiles
Last modified: August 30, 2021
Copyright © The Lynx Group, LLC. All rights reserved.