Review of Index Line Selection Methods in Hemato-Oncology Externally Controlled Trials: A Secondary Analysis of the Hermans et al Systematic Review
Author(s)
Michael Wallington, MA1, Iman Fakih, MSc2, Yong Chen, PhD3, Ann-Sophie Demers, MSc4, Yong Chen, PhD5.
1Pfizer UK, London, United Kingdom, 2STATLOG, Montreal, QC, Canada, 3University of Pennsylvania, Philadelphia, PA, USA, 4STATLOG, Québec, QC, Canada, 5Pfizer, Collegeville, PA, USA.
1Pfizer UK, London, United Kingdom, 2STATLOG, Montreal, QC, Canada, 3University of Pennsylvania, Philadelphia, PA, USA, 4STATLOG, Québec, QC, Canada, 5Pfizer, Collegeville, PA, USA.
OBJECTIVES: In external control arm (ECA) trials, control patients could be eligible for selection at various time points. In hemato-oncology, this often occurs when the trial criteria require failure of prior treatments. As later treatment lines are typically associated with greater morbidity and mortality, the choice of index line in an ECA can greatly impact the resultant efficacy estimates. The objective of this study was to review hemato-oncology ECAs and evaluate the indexing methods used.
METHODS: In a 2024 systematic review, Hermans et al.1 identified 32 real-world data (RWD)-derived hemato-oncology ECAs published 1/1/2000 - 23/10/2023. A secondary analysis of these ECAs was conducted. ECAs with (1) retrospective selection into the ECA, (2) failure of first/greater-line therapies inclusion criteria, and (3) access to individual-level data were included. ECA characteristics and indexing method details were extracted and summarized.
RESULTS: Twenty-two ECAs met the secondary analysis inclusion criteria. Prior retrospective chart review studies (n=7; 31.8%) were the most frequent RWD sources, followed by electronic medical records databases (n=4; 18.2%). Only 12 studies (54.5%) explicitly acknowledged that their study design leads to multiple eligible index lines per patient. Most studies (n=15; 68.2%) selected one line per patient in their main analysis, while 6 (27.3%) selected multiple. The most frequently used indexing methods were ‘first eligible line’ (n=6; 27.3%) and ‘all eligible lines’ (n=6; 27.3%). Four studies used methods that introduce bias in overall survival, according to the peer-reviewed literature on the subject, specifically ‘last eligible line’, ‘last available line’, and ‘random line from eligible lines’. Six studies conducted indexing method sensitivities. Most (n=15; 68.2%) did not cite references on indexing methodology.
CONCLUSIONS: The majority of ECAs identified faced the challenge of multiple eligible index lines, but only half acknowledged it. More careful consideration is needed in this matter especially to minimize resultant biases.[1] Hermans. JAMA Oncol. 2024.
METHODS: In a 2024 systematic review, Hermans et al.1 identified 32 real-world data (RWD)-derived hemato-oncology ECAs published 1/1/2000 - 23/10/2023. A secondary analysis of these ECAs was conducted. ECAs with (1) retrospective selection into the ECA, (2) failure of first/greater-line therapies inclusion criteria, and (3) access to individual-level data were included. ECA characteristics and indexing method details were extracted and summarized.
RESULTS: Twenty-two ECAs met the secondary analysis inclusion criteria. Prior retrospective chart review studies (n=7; 31.8%) were the most frequent RWD sources, followed by electronic medical records databases (n=4; 18.2%). Only 12 studies (54.5%) explicitly acknowledged that their study design leads to multiple eligible index lines per patient. Most studies (n=15; 68.2%) selected one line per patient in their main analysis, while 6 (27.3%) selected multiple. The most frequently used indexing methods were ‘first eligible line’ (n=6; 27.3%) and ‘all eligible lines’ (n=6; 27.3%). Four studies used methods that introduce bias in overall survival, according to the peer-reviewed literature on the subject, specifically ‘last eligible line’, ‘last available line’, and ‘random line from eligible lines’. Six studies conducted indexing method sensitivities. Most (n=15; 68.2%) did not cite references on indexing methodology.
CONCLUSIONS: The majority of ECAs identified faced the challenge of multiple eligible index lines, but only half acknowledged it. More careful consideration is needed in this matter especially to minimize resultant biases.[1] Hermans. JAMA Oncol. 2024.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
CO213
Topic
Clinical Outcomes, Methodological & Statistical Research, Study Approaches
Topic Subcategory
Comparative Effectiveness or Efficacy
Disease
Oncology