Modifying clinical trial designs to test treatments for clinical significance in individual patients
Modifying clinical trial designs to test treatments for clinical significance in individual patients
复制标题
DOI：
10.2165/00044011-200121100-00007
复制
发表时间：
2001-01-01
影响因子：
3.2
通讯作者：
Becker, RE
中科院分区：
文献类型：
--
作者：
Becker, RE
本文的目的是设计临床试验，测试治疗对个体患者的临床意义。在重复测量设计中，标量汇总统计描述了每个个体的临床过程。为了为每个患者的临床过程提供令人满意的可靠性作为分析单位，试验前的测量可靠性研究构成了使用结果测量的评估计划的基础。如果治疗组之间患者病程出现统计学显着差异，则拒绝原假设。使用临床和统计显着性标准，研究者评估每个患者的结果：首先是对治疗的反应；其次分配在治疗和安慰剂条件下患者结果发生的概率。这种分析方法为从业者提供了基于临床试验的模型证据，以评估患者护理中的治疗效果。分析方法开发了一种概率模型，可将个体识别为对治疗条件的反应者或非反应者。修改后的临床试验设计对当前临床试验的确认进行了细化和更加具体化，即通过一个模型确定治疗有效，该模型可识别患者对试验中的治疗有反应和无反应，并提供每个患者结果发生的概率。这些方法为临床试验设计提供了实用的修改。经过修改，临床试验证明了个体患者对试验中治疗的反应。用于判断个体患者反应的临床试验模型为医生提供了判断个体患者对治疗的反应的证据基础，与当前实践相比，其可靠性得到了提高。临床试验设计变化带来的临床护理优势在阿尔茨海默氏症治疗中显而易见，测量中的巨大误差差异妨碍了对大多数患者治疗反应的准确评估。建议对临床试验设计和分析进行的修改适用于其他疾病类别，因为临床评估和评估个体患者对治疗反应的临床试验证据基础的可靠性得到了提高。Becker 和 Markwell。([1]) 在研究了用于评估阿尔茨海默病 (AD) 患者药物反应的方法的准确性后，得出的结论是，当前的临床试验报告不足以让医生做好充分准备，就临床使用的治疗做出明智的决定。 AD 患者评估方法中的误差方差使得大多数患者的治疗反应评估不可靠。([1]) 在 AD 和其他疾病类别中，临床试验报告通常将分析限制为证明具有 p 值的组之间具有统计显着性差异。大多数报告没有提供评估临床试验临床意义所需的分析：效应大小；患者的假阳性和假阴性识别率；结果分布的同质性、双峰性或偏斜研究；结果测量的可靠性研究。即使包含这些更详细分析的临床试验报告也无法充分满足从业者的需求——为每位患者确定最有效的治疗方法。临床试验证明各组之间的疗效存在差异；执业医师必须就每个患者的治疗效果做出决定。执业医师使用非系统化的临床判断来应用来自临床试验的科学证据。（[2]）我寻求一种设计和分析临床试验的方法，以更好地解决执业医师的问题。临床试验的证据目标进行了修改，以确定对个体的治疗效果。不在分组比较中。临床试验方法的这种扩展首先确定了患者评估方法的可靠性，其次测试了个体患者治疗的临床意义，第三提供了患者反应是治疗结果的可能性。从业者使用个体患者评估的扩展临床试验模型来减少基于证据的患者护理中非系统临床判断的影响。
The aim of this article was to design clinical trials that test treatments for clinical significance in individual patients, In a repeated measures design a scalar summary statistic describes the clinical course of each individual. To provide satisfactory reliability for each patient's clinical course to be the unit of analysis, a study of measurement reliability prior to the trial forms the basis of the plan of assessment using outcome measures. The null hypothesis is rejected if statistically significant differences in patient courses arise among the treatment arms. Using criteria of clinical and statistical significance, the investigator evaluates each individual patient's outcome: first for response to the treatment; second to assign a probability for occurrence of the patient's outcome under treatment and placebo conditions. This method of analysis provides the practitioner with a model evidence based in a clinical trial to evaluate treatment effects in patient care. The method of analysis develops a probabilistic model that identifies individuals as responders or non-responders to a treatment condition. The modified clinical trial design refines and makes more specific the current clinical trial confirmation that a treatment is effective with a model that identifies patients as responding and non-responding to the treatments in the trial and provides a probability of occurrence for each patient outcome.These methods offer a practical modification of clinical trial design. With modification, clinical trials evidence individual patient responses to treatments in the trial. The clinical trial model for judging individual patient responses provides the physician with an evidence base for judging individual patient responses to treatment with improved reliability compared with current practices. The clinical care advantages from the design changes in clinical trials are readily apparent in Alzheimer's therapy where large error variances in measurements preclude accurate assessments of treatment response for most patients. The suggested modifications to clinical trial design and analysis apply in other disease categories because of improved reliability in clinical assessments and the clinical trial evidence base for evaluating individual patient responses to treatment.Becker and Markwell.([1]) after studying the accuracy of methods used to assess drug response in patients with Alzheimer's disease (AD), concluded that current clinical trial reports do not adequately prepare the physician to reach informed decisions about treatments used in the clinic. Error variance in methods of assessment of patients with AD makes evaluations of treatment response unreliable for the majority of patients.([1]) In AD and other disease categories, clinical trial reports often limit analyses to a demonstration of statistically significant differences between groups with p-values. Most reports do not provide the analyses needed to estimate the clinical significance of the clinical trial: effect size; rates of false-positive and false-negative identifications of patients; studies of homogeneity, bimodality or skew in the distributions of outcomes; studies of the reliability of outcome measures. Even clinical trial reports with these more detailed analyses do not adequately address the practitioner's needs - to identify the most effective treatment for each patient. Clinical trials evidence efficacy in differences among groups; the practising physician must reach decisions about the efficacy of a treatment in each individual patient.The practitioner applies the scientific evidence from the clinical trial using unsystematised clinical judgements.([2]) I sought a method of design and analysis of clinical trials that better addresses the practitioner's problems. Evidentiary aims for clinical trials are revised to identify treatment effects in individuals. not in group comparisons. This extension of clinical trial methods first establishes reliability for methods of patient assessment, second tests treatments for clinical significance in individual patients, and third offers a probability that a patient's response is a consequence of treatment. A practitioner uses the extended clinical trial model of individual patient assessment to reduce the influence from unsystematic clinical judgements in evidence-based patient care.