TY - JOUR
T1 - Individual treatment effect estimation in the presence of unobserved confounding using proxies
T2 - a cohort study in stage III non-small cell lung cancer
AU - van Amsterdam, Wouter A C
AU - Verhoeff, Joost J C
AU - Harlianto, Netanja I
AU - Bartholomeus, Gijs A
AU - Puli, Aahlad Manas
AU - de Jong, Pim A
AU - Leiner, Tim
AU - van Lindert, Anne S R
AU - Eijkemans, Marinus J C
AU - Ranganath, Rajesh
N1 - Funding Information:
WA was supported by the Alexandre Suerman personal PhD stipendium. The Alexandre Suerman stipend had no role in any part of the study design, conduct or reporting.
Publisher Copyright:
© 2022, The Author(s).
PY - 2022/4/7
Y1 - 2022/4/7
N2 - Randomized Controlled Trials (RCT) are the gold standard for estimating treatment effects but some important situations in cancer care require treatment effect estimates from observational data. We developed "Proxy based individual treatment effect modeling in cancer" (PROTECT) to estimate treatment effects from observational data when there are unobserved confounders, but proxy measurements of these confounders exist. We identified an unobserved confounder in observational cancer research: overall fitness. Proxy measurements of overall fitness exist like performance score, but the fitness as observed by the treating physician is unavailable for research. PROTECT reconstructs the distribution of the unobserved confounder based on these proxy measurements to estimate the treatment effect. PROTECT was applied to an observational cohort of 504 stage III non-small cell lung cancer (NSCLC) patients, treated with concurrent chemoradiation or sequential chemoradiation. Whereas conventional confounding adjustment methods seemed to overestimate the treatment effect, PROTECT provided credible treatment effect estimates.
AB - Randomized Controlled Trials (RCT) are the gold standard for estimating treatment effects but some important situations in cancer care require treatment effect estimates from observational data. We developed "Proxy based individual treatment effect modeling in cancer" (PROTECT) to estimate treatment effects from observational data when there are unobserved confounders, but proxy measurements of these confounders exist. We identified an unobserved confounder in observational cancer research: overall fitness. Proxy measurements of overall fitness exist like performance score, but the fitness as observed by the treating physician is unavailable for research. PROTECT reconstructs the distribution of the unobserved confounder based on these proxy measurements to estimate the treatment effect. PROTECT was applied to an observational cohort of 504 stage III non-small cell lung cancer (NSCLC) patients, treated with concurrent chemoradiation or sequential chemoradiation. Whereas conventional confounding adjustment methods seemed to overestimate the treatment effect, PROTECT provided credible treatment effect estimates.
KW - Carcinoma, Non-Small-Cell Lung/drug therapy
KW - Chemoradiotherapy
KW - Cohort Studies
KW - Humans
KW - Lung Neoplasms/therapy
UR - http://www.scopus.com/inward/record.url?scp=85128037907&partnerID=8YFLogxK
U2 - 10.1038/s41598-022-09775-9
DO - 10.1038/s41598-022-09775-9
M3 - Article
C2 - 35393451
SN - 2045-2322
VL - 12
JO - Scientific Reports
JF - Scientific Reports
IS - 1
M1 - 5848
ER -