Scholarlink

<p>Guideline-driven timing of treatment initiation plays a central role in clinical decision-making and health policy, yet conventional analyses of randomized clinical trials typically estimate only an average guideline effect and may fail to recover the timing-specific impacts that are most relevant for practice. In this work, we address two key challenges in evaluating guideline effects: (1) treatment initiation times vary substantially across individuals under the reference guideline, and (2) counterfactual outcomes for patients who would initiate treatment at a given time under the guideline are unobserved under alternative initiation strategies. We introduce a causal estimand that captures the average guideline effect for individuals who would initiate treatment at a given time under the reference guideline, and we establish its identification conditions. We then propose a weighting estimator that combines nonparametric outcome regression with estimated treatment-initiation densities to recover these timing-specific effects. We investigate the estimator’s theoretical properties and evaluate its finite-sample performance through simulations. Finally, we apply the proposed method to a randomized HIV treatment trial to estimate the heterogeneous effects of early versus guideline-based antiretroviral therapy initiation on tuberculosis risk.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Heterogeneous effects of guideline-based treatment initiation: a causal framework for timing-specific evaluation

  • 期刊论文

  • 发表日期：

    2026年5月14日

查看全文
  • Sanghee Kim,
  • Seonjin Kim,
  • Myung Hee Lee,
  • Hyunkeun Cho

摘要

Guideline-driven timing of treatment initiation plays a central role in clinical decision-making and health policy, yet conventional analyses of randomized clinical trials typically estimate only an average guideline effect and may fail to recover the timing-specific impacts that are most relevant for practice. In this work, we address two key challenges in evaluating guideline effects: (1) treatment initiation times vary substantially across individuals under the reference guideline, and (2) counterfactual outcomes for patients who would initiate treatment at a given time under the guideline are unobserved under alternative initiation strategies. We introduce a causal estimand that captures the average guideline effect for individuals who would initiate treatment at a given time under the reference guideline, and we establish its identification conditions. We then propose a weighting estimator that combines nonparametric outcome regression with estimated treatment-initiation densities to recover these timing-specific effects. We investigate the estimator’s theoretical properties and evaluate its finite-sample performance through simulations. Finally, we apply the proposed method to a randomized HIV treatment trial to estimate the heterogeneous effects of early versus guideline-based antiretroviral therapy initiation on tuberculosis risk.