Department of Medical Sciences, Uppsala University, Uppsala, Sweden.
Scand J Public Health. 2011 Jun;39(4):439-41. doi: 10.1177/1403494810396556. Epub 2011 Jan 26.
New risk factors for various diseases are suggested at an increasing pace, with promise of clinical usefulness for risk prediction, but almost unvaryingly without formal testing of that property. We propose that a risk factor clinically relevant for risk prediction can be defined as one that correctly alters predicted risk to a clinically relevant extent in persons with a relevant absolute risk, such that it affects clinical decision making. We recommend that investigators who suggest a new risk factor for clinical use investigate if the new risk factor adds capacity to discriminate between persons who will subsequently experience the outcome from those who will not. For that purpose, we provide tools for calculating the net reclassification improvement, NRI, and the integrated discrimination improvement, IDI, using major statistical packages.
新的疾病风险因素不断被提出，有望用于风险预测的临床应用，但几乎无一例外地没有对该特性进行正式测试。我们提出，对于风险预测具有临床相关性的风险因素，可以定义为在具有相关绝对风险的人群中，正确改变预测风险至具有临床相关程度的因素，从而影响临床决策。我们建议，那些为临床应用提出新风险因素的研究人员，应该调查新的风险因素是否增加了区分随后会经历结局的人与不会经历结局的人的能力。为此，我们提供了使用主要统计软件包计算净重新分类改善（NRI）和综合判别改善（IDI）的工具。
