PS 4-41 RELIABILITY AND CONSISTENCY OF THREE VALUE FRAMEWORKS FOR ONCOLOGY THERAPEUTICS

Wednesday, October 26, 2016
Bayshore Ballroom ABC, Lobby Level (Westin Bayshore Vancouver)
Poster Board # PS 4-41

Tanya G.K. Bentley, PhD1, Joshua T. Cohen, PhD2, Elena B. Elkin, PhD3, Julie Huynh, MD4, Arnab Mukherjea, DrPH, MPH5, Thanh H. Neville, MD, MSHS6, Ioana Popescu, MD, MPH7, Jenelle M. Zambrano, DNP, CNS, RN1, Eunice Chang, PhD1 and Michael S. Broder, MD, MSHS1, (1)Partnership for Health Analytic Research, LLC, Beverly Hills, CA, (2)Tufts Medical Center, Boston, MA, (3)Memorial Sloan-Kettering Cancer Center, New York, NY, (4)Hematology Oncology Medical Group of San Fernando Valley, Encino, CA, (5)California State University, East Bay, Hayward, CA, (6)UCLA Department of Medicine, Los Angeles, CA, (7)David Geffen School of Medicine at UCLA, Department of Medicine, Los Angeles, CA
Purpose: We evaluated the reliability and consistency of published oncology value assessment frameworks to help physicians and payers better understand how to use these frameworks to achieve greater value in medical decision-making.

Method: Six raters (3 MDs, 1 DNP, 2 PhDs) all rated 2 oncology products for each of 3 cancers (6 product-cancer combinations in all) using 3 frameworks:

  • ASCO Value Framework
  • ESMO Magnitude of Clinical Benefit Scale
  • Institute for Clinical and Economic Review (ICER) Value Assessment Framework.

More prevalent and costly cancers and related products were selected to represent a range of indications (curative and palliative), malignancies (solid and hematologic), and mechanisms (cytotoxic, biologic, immunologic).

Raters received the published clinical data required to complete the evaluations and detailed instructions for each framework, and were provided no formal training. Intraclass correlation coefficients (ICC) were estimated to measure tool reliability. Drugs indicated for advanced disease (5 of the 6) were rank ordered by mean and individual scores, and Kendall’s coefficient was calculated to measure agreement among tools. In sensitivity analyses (SA) for ICC, raters were excluded one at a time. 

Result: There were 6 ratings of 6 products for each of the 3 frameworks (108 ratings total). ICC results (SA range) were: ASCO 0.66 (0.61-0.70); ESMO 0.73 (0.67-0.78); ICER 0.72 (0.65-0.95). Rankings for the 5 advanced disease regimens (A-E) varied by framework:

ASCO

ESMO

ICER

A

B

A

B

D

C

C

A

D

D

C

B

E

E

E

Kendall’s W across all 3 frameworks was 0.69 (range 0.59-0.85 among individual raters). Pairwise, Kendall’s was 0.75 for ASCO-ESMO, 0.85 ASCO-ICER, and 0.70 ESMO-ICER. 

Conclusion: Knowledgeable but untrained raters, provided with key data, produced moderately reliable results using 3 recently published frameworks for assessing the value of cancer treatments. The frameworks had fair to good consistency, indicating convergent validity, although they led to significantly different conclusions about the relative value of treatments for advanced disease. The conclusions suggest it may be premature to use these frameworks in treatment decision-making without further proof of their reliability or a better understanding of how to interpret differential results between frameworks.