Modifying Clinical Trial Designs
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, 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. 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. 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.
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, 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. 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. 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.
SHARE
