Stratification-based Instrumental Variable Analysis Framework for Nonlinear Effect Analysis

2025-07-09

AI Summary

Nonlinear causal effect estimation under continuous exposure is challenged by bias and inefficiency arising from unmeasured confounding and weak instrumental variables (IVs). Method: We propose a hierarchical IV analysis framework that integrates stratification with IV identification principles, enabling robust estimation of nonlinear causal effect functions and accurate detection of change points (thresholds) without strong functional-form assumptions. Contribution/Results: In simulations, the method consistently recovers true effect shapes across diverse model specifications. In empirical analysis using UK Biobank data, it delivers the first Mendelian randomization evidence of a statistically significant causal threshold for alcohol intake on systolic blood pressure—approximately 7 g/day—aligning closely with current clinical guidelines. The framework substantially improves statistical power and inferential reliability under weak-IV conditions, offering a novel paradigm for nonlinear causal discovery in observational epidemiology.

Technology Category

• Reasoning under Uncertainty: Causality
• Machine Learning: Causal Learning
• Intelligent Robots: State Estimation

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Security and Privacy: Large-scale security measurements

Abstract

Nonlinear causal effects are prevalent in many research scenarios involving continuous exposures, and instrumental variables (IVs) can be employed to investigate such effects, particularly in the presence of unmeasured confounders. However, common IV methods for nonlinear effect analysis, such as IV regression or the control-function method, have inherent limitations, leading to either low statistical power or potentially misleading conclusions. In this work, we propose an alternative IV framework for nonlinear effect analysis, which has recently emerged in genetic epidemiology and addresses many of the drawbacks of existing IV methods. This framework enables study of the effect function while avoiding unnecessary model assumptions. In particular, it facilitates the identification of change points or threshold values in causal effects. Through a wide variety of simulations, we demonstrate that our framework outperforms other representative nonlinear IV methods in predicting the effect shape when the instrument is weak and can accurately estimate the effect function as well as identify the change point and predict its value under various structural model and effect shape scenarios. We further apply our framework to assess the nonlinear effect of alcohol consumption on systolic blood pressure using a genetic instrument (i.e. Mendelian randomization) with UK Biobank data. Our analysis detects a threshold beyond which alcohol intake exhibits a clear causal effect on the outcome. Our results are consistent with published medical guidelines.

Problem

• Analyzing nonlinear causal effects with instrumental variables
• Overcoming limitations of existing IV methods for nonlinear analysis
• Identifying change points in causal effects accurately

Innovation

• Stratification-based IV framework for nonlinear effects
• Identifies change points in causal effects
• Outperforms existing methods with weak instruments