van der Schouw Y T, Verbeek A L, Ruijs S H
Department of Medical Informatics and Epidemiology, University of Nijmegen, The Netherlands.
Invest Radiol. 1995 Jun;30(6):334-40. doi: 10.1097/00004424-199506000-00002.
Because new diagnostic tests become available rapidly, the authors determined a need for proper assessment of tests before their implementation in clinical practice. Three factors are of pivotal importance: the selection of the proper study population, the determination of the diagnostic power including its related statistical analysis, and the relation of the new test to current diagnostic tools. Patients suspected of having a disease are those who would benefit from the application of a new test. Therefore, only those patients need be involved in the assessment study.
Summary measures of diagnostic power other than sensitivity and specificity are advocated because these conventional measures depend on cutoff points and are susceptible to selection bias. The relation between the new test and existing diagnostic tools must be established to determine if the new test contributes to the diagnostic process.
To avoid waste of effort and money, the authors suggest a prudent assessment approach in phases. Whereas the initial challenge consists of selection of an adequate patient population, subsequently all determinants of diseases (signs, symptoms, comorbidity, and other diagnostic factors) and factors influencing the decision to use a test (patient burden and cost) are considered.
由于新的诊断测试迅速出现，作者确定在临床实践中实施这些测试之前需要对其进行适当评估。有三个因素至关重要：选择合适的研究人群、确定诊断能力及其相关的统计分析，以及新测试与当前诊断工具的关系。疑似患有疾病的患者是那些将从新测试的应用中受益的人。因此，只有这些患者需要参与评估研究。
提倡使用除敏感性和特异性之外的诊断能力汇总指标，因为这些传统指标依赖于截断点，并且容易受到选择偏倚的影响。必须建立新测试与现有诊断工具之间的关系，以确定新测试是否有助于诊断过程。
为避免精力和金钱的浪费，作者建议分阶段采取谨慎的评估方法。最初的挑战在于选择足够的患者群体，随后要考虑所有疾病决定因素（体征、症状、合并症和其他诊断因素）以及影响使用测试决策的因素（患者负担和成本）。
