文献速递术前评估与围术期医学神经外科与老年麻醉

SNAP-3 研究:用有向无环图减少偏倚并优化因果推断

Directed acyclic graphs to minimise bias and optimise causal inference in SNAP-3: an observational cohort study of frailty, multimorbidity, and delirium in older surgical patients.
PubMed 文献 · Br J Anaesth · 2025-05-27

SNAP-3 研究旨在描述老年手术患者衰弱、多病共存及谵妄及其管理对术后结局的影响。本文介绍了该队列研究中有向无环图(DAG)的构建方法,DAG 依据证据合成法构建并经独立临床专家修订,用于明确变量间因果关系并指导统计调整。DAG 提供了透明框架,有助于减少偏倚并优化因果推断。

要点

  • DAG 可直观呈现变量间假设关系并指导统计模型选择。
  • SNAP-3 为衰弱、多病共存、谵妄与术后结局关系构建了 DAG。
  • DAG 构建采用证据合成法并经临床专家修订,提高透明度。

核心结论

  • DAG 为观察性研究中的统计决策提供了透明框架,有助于减少偏倚并优化因果推断。

研究设计与样本

  • SNAP-3 为观察性队列研究,本文描述其 DAG 的构建方法。
  • DAG 依据「构建有向无环图的证据合成」方法创建,并经独立临床专家输入修订。

主要发现

  • DAG 直观呈现了衰弱、多病共存、谵妄作为暴露与术后结局(住院时间、谵妄、并发症、死亡)关系的假设。
  • 所有 DAG 的关键节点包括年龄、痴呆、遗传易感性、听力与视力损害、住院时间、恶性肿瘤、手术严重程度、多重用药、术后围术期医学服务、术前门诊评估、性别、社会剥夺、 urgency、谵妄、衰弱、多病共存、交互作用、并发症等。

临床怎么用

  • 临床研究者可参考 SNAP-3 的 DAG 构建方法,明确因果假设并指导统计调整。

局限性

  • DAG 基于假设构建,需经临床专家评审,并可在未来研究中进一步修正。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

BACKGROUND: The 3rd Sprint National Anaesthesia Project (SNAP-3) aims to describe the impact of frailty, multimorbidity, and delirium, and their management, on outcomes after surgery within the older surgical population. Causal diagrams, such as directed acyclic graphs (DAGs), are a useful tool for visually representing relationships between variables and for clarifying the causal assumptions underlying a chosen statistical model. METHODS: A description of how the SNAP-3 cohort study's DAGs were developed is provided. DAGs have been created for the exposure-outcome relationships between frailty, multimorbidity, and delirium (as an exposure) with postoperative outcomes (length of hospital stay, delirium, morbidity, mortality). DAGs were created following the approach of 'Evidence synthesis for constructing directed acyclic graphs', and revised after independent clinical expert input. RESULTS: DAGs provide visual representations of assumptions made, and provide an objective approach to appropriate statistical adjustments. Key nodes within all the DAGs included age, dementia, genetic predisposition, hearing and visual impairment, length of stay, malignancy, operative severity, polypharmacy, postoperative perioperative medicine service, preoperative clinic review, sex, social deprivation, urgency, with delirium, frailty, multimorbidity, interaction, morbidity acting as exposures, or outcomes in certain DAGs. CONCLUSIONS: DAGs provide a transparent framework for statistical decision-making in observational research. We provide an overview of DAGs using the SNAP-3 DAGs as examples to explain fundamental concepts for developing and using causal diagrams. This overview acknowledges the complexities of exploring clinical relationships and the assumptions that are necessary, providing an opportunity for critique of the relationships described and refinements for future studies.

来源:Br J Anaesth,查看原文。版权归原作者所有。

引用

  1. Swarbrick CJ, Blake HA, Martin P, et al. Directed acyclic graphs to minimise bias and optimise causal inference in SNAP-3: an observational cohort study of frailty, multimorbidity, and delirium in older surgical patients. Br J Anaesth. 2025. PMID: 40425394. doi:10.1016/j.bja.2025.04.027

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。