Sequential Designs for Monitoring Two Endpoints in a Clinical Trial
- PDF / 765,212 Bytes
- 10 Pages / 504 x 720 pts Page_size
- 83 Downloads / 216 Views
Drug Informotion Journal, Vol. 33. pp. 417426. 1999 Printed in the USA. All rights reserved.
Copyright Q I999 Drug Information Association Inc.
SEQUENTIAL DESIGNS FOR MONITORING TWO ENDPOINTS IN A CLINICAL TRIAL SUSANTODD,PHD Research Fellow, The Medical and Pharmaceutical Statistics Research Unit, The University of Reading, Reading, United Kingdom
When more than one response is of primary interest in a clinical trial, use of univariate stopping rules for allowing interim analyses is inappropriate. This paper considers the case where two endpoints are of interest andpresents general methodology for the design, monitoring, and analysis of such trials. Two examples are presented of situations in which the technique could be implemented; the first is a trial in AlzheimerS patients, the second concerns women with osteoporosis. ”he appropriate application of the procedure can lead to the same ethical and economic benefits of univariate sequential designs, and so carries the potential to lead to more efJicient clinical trials. Key Words: Alzheimer’s disease; Correlated bivariate endpoints; G r o u p sequential designs; Interim analyses; Osteoporosis
might reveal that the new therapy has already proved itself superior, or indeed inferior, to STUDIES INVOLVING HUMANS are the standard. Is it then ethical to continue the unique. Ethical considerations continually trial? Unfortunately, without suitable adjustprompt research into new ways of accurately ment for the process of repeated inspection, evaluating potential treatments, while ade- such interim analyses, although advantaquately safeguarding patients. Most clinical geous, may lead to unreliable statistical findtrials recruit patients over a period of time, ings. and data from the study accumulate steadily The ethical (and economical) desirability throughout its duration. A data and safety of stopping a trial which shows an early admonitoring board may exist to assess this vantage for one or other of the treatments safety and efficacy information and recom- has motivated the development of trial procemend modification or early stopping of the dures incorporating techniques of sequentid trial as appropriate. If the data are examined analysis. Data are periodically inspected as before the trial is due to terminate, then they the trial progresses, perhaps after every patient or alternatively, at just one or two interim looks, with treatment efficacy being Presented at the DIA 34th Annual Meeting ‘Thinking assessed. The characteristic feature of seGlobally: Product Development, Registration, and Mar- quential clinical trials is that the number of keting in the New Millennium,” June 7-11, 1998. Bos- observations required is not predetermined. ton, Massachusetts. The decision to terminate depends, at any Reprint address: Susan Todd, The Medical and stage, on the results of the observations prePharmaceutical Statistics Research Unit, Department of viously made and the trial is stopped as soon Applied Statistics, The University of Reading, PO Box as sufficient evidence to re
Data Loading...
