Department of Clinical Epidemiology, Leiden University Medical Center (LUMC), Leiden, The Netherlands; Department of Internal Medicine, Leiden University Medical Center (LUMC), Leiden, The Netherlands.
Department of Clinical Epidemiology, Leiden University Medical Center (LUMC), Leiden, The Netherlands.
J Clin Epidemiol. 2020 Jul;123:69-79. doi: 10.1016/j.jclinepi.2020.03.015. Epub 2020 Mar 30.
The objective of this study was to systematically review and externally assess the predictive performance of models for ischemic stroke in incident dialysis patients.
Two reviewers systematically searched and selected ischemic stroke models. Risk of bias was assessed with the PROBAST. Predictive performance was evaluated within The Netherlands Cooperative Study on the Adequacy of Dialysis (NECOSAD), a large prospective multicenter cohort of incident dialysis patients. For discrimination, c-statistics were calculated; calibration was assessed by plotting predicted and observed probabilities for stroke, and calibration-in-the-large.
Seventy-seven prediction models for stroke were identified, of which 15 were validated. Risk of bias was high, with all of these models scoring high risk in one or more domains. In NECOSAD, of the 1,955 patients, 127 (6.5%) suffered an ischemic stroke during the follow-up of 2.5 years. Compared with the original studies, most models performed worse with all models showing poor calibration and discriminative abilities (c-statistics ranging from 0.49 to 0.66). The Framingham showed reasonable calibration; however, with a c-statistic of 0.57 (95% CI 0.50-0.63), the discrimination was poor.
This external validation demonstrates the weak predictive performance of ischemic stroke models in incident dialysis patients. Instead of using these models in this fragile population, either existing models should be updated, or novel models should be developed and validated.
本研究旨在系统地回顾和外部评估用于预测透析患者中缺血性卒中的模型的预测性能。
两名审查员系统地搜索并选择了缺血性卒中模型。采用 PROBAST 评估偏倚风险。在荷兰透析充足性合作研究（NECOSAD）中评估预测性能，该研究是一项针对新发生透析患者的大型前瞻性多中心队列研究。为了评估区分度，计算了 c 统计量；通过绘制卒中的预测和观察概率以及大校准来评估校准。
共确定了 77 个用于卒中预测的模型，其中 15 个进行了验证。偏倚风险很高，所有这些模型在一个或多个领域都被评为高风险。在 NECOSAD 中，在 1955 名患者中，有 127 名（6.5%）在 2.5 年的随访期间发生了缺血性卒中。与原始研究相比，大多数模型的表现都较差，所有模型的校准和区分能力都较差（c 统计量范围为 0.49 至 0.66）。Framingham 模型具有合理的校准；然而，其区分能力较差，c 统计量为 0.57（95%置信区间 0.50-0.63）。
这项外部验证表明，用于透析患者中缺血性卒中的模型的预测性能较弱。在这个脆弱的人群中，不应该使用这些模型，要么更新现有的模型，要么开发和验证新的模型。
