College of Nursing, MSC09 5350 1 University of New Mexico, Albuquerque, NM 87131, USA.
Heart Lung. 2013 May-Jun;42(3):221-6. doi: 10.1016/j.hrtlng.2013.01.006. Epub 2013 Mar 13.
Statistics estimated from a 2 × 2 contingency table, such as sensitivity, specificity, and likelihood ratios, are usually presented in contexts of diagnostic decision making or evaluation of screening tests. However, their use is not restricted to those contexts. The purpose of this article is to summarize for clinicians and researchers in nursing and other health care disciplines the conceptual basis, calculation, application, and limitations of statistics derivable from a 2 × 2 table and to provide a worked example based on previous research. It will be shown that these statistics can potentially be used for analyzing any binary outcome (e.g., occurrence vs. nonoccurrence of some event) with a binary predictor (e.g., present vs. absent or above vs. below some cut point). Most can be estimated with a hand-held or online calculator, and, apart from chi-square statistics, without first converting observed data to a test-statistic. Clearer understanding of the derivation and clinical meaning of these statistics will encourage their wider use in health research and clinical practice.
从 2×2 列联表中估计的统计数据，如敏感性、特异性和似然比，通常用于诊断决策或评估筛选试验的背景下。然而，它们的使用并不限于这些情况。本文的目的是为护理和其他医疗保健学科的临床医生和研究人员总结从 2×2 表中得出的统计学的概念基础、计算、应用和局限性，并提供一个基于先前研究的实例。结果表明，这些统计数据可潜在地用于分析任何具有二项预测变量（如某个事件的出现与不出现）的二项结果（如存在与不存在或超过与低于某个切点）。其中大多数可以使用手持或在线计算器来估计，并且除了卡方统计量外，无需先将观察数据转换为检验统计量。更清楚地了解这些统计数据的推导和临床意义将鼓励它们在健康研究和临床实践中的更广泛应用。
