PHARMACEUTICAL STATISTICS

Pharmaceut. Statist. 2009; 8: 253–254

Published online 22 July 2009 in Wiley InterScience

(www.interscience.wiley.com) DOI: 10.1002/pst.399 LITERATURE REVIEWS

Literature Review April– June 2009

S. Krishna Padmanabhan1 and Andrew Stone2,�,y1Wyeth Research and Development, 500 Arcola Rd, Collegeville, PA 19426, USA2AstraZeneca Pharmaceuticals, Clinical Development, Alderley Park, Macclesfield,SK10 4TG, UK

INTRODUCTION

This review covers the following journals receivedduring the period from April to June 2009:

� Applied Statistics, volume 58, part 2.� Biometrical Journal, volume 51, part 2.� Biometrics, volume 65, part 2.� Biometrika, volume 96, part 2.� Biostatistics, volume 10, part 2.� Clinical Trials, volume 6, parts 2–3.� Contemporary Clinical Trials, volume 30, part 3.� Drug Information Journal, volume 43, parts 3–4.� Journal of Biopharmaceutical Statistics,

volume 19, parts 3–4.� Journal of the Royal Statistical Society,

Series A, volume 172, parts 2–3.� Journal of the Royal Statistical Society,

Series B, volume 71, parts 2–3.� Statistics in Medicine, volume 28, parts 9–16.� Statistical Methods in Medical Research,

volume 18, parts 2–3.

SELECTED HIGHLIGHTS FROMTHE LITERATURE

Interim analyses

In this paper the authors provide a new take onbias estimation when a group sequential design isstopped at an interim. They do this by comparing

the effect inflation with similarly extreme resultsfrom a fixed sample design. They conclude that‘Except for very early interim analyses (p25% ofthe information), concerns about the inflation ofthe treatment effect should be minor’.

� Freidlin B and Korn EL. Stopping clinicaltrials early for benefit: impact on estimation.Clinical Trials 2009; 6:119–125.

Endpoint adjudication

As clinical trials become increasingly complex, thecost effectiveness of some procedures is beingscrutinized. In this paper the authors perform asystematic review of event adjudication from largecardiovascular outcome trials comparing thetreatment effects observed from adjudicated andsite reported outcomes.

� Pogue J et al. Evaluating the benefit of eventadjudication of cardiovascular outcomes in largesimple RCTs. Clinical Trials 2009; 6:239–251.

Multiregional trials

Multiregional trials, and especially those that includeJapanese patients, are becoming increasingly com-mon. The following paper explores different app-roaches to sample-size allocation between regions.

� Uesaka H. Sample size allocation to regions ina multiregional trial. Journal of Biopharmaceu-tical Statistics 2009; 19:580–594.

A multifunctional group of experts within theScientific and Regulatory technical groups ofyE-mail: Andrew.stone@astrazeneca.com

*Correspondence to: Andrew Stone AstraZeneca Pharma-ceuticals, Clinical Development, Alderley Park, Macclesfield,SK10 4TG, UK.

Copyright r 2009 John Wiley & Sons, Ltd.

PhRMA member companies was convened as theSimultaneous Global Development Committee tomake recommendations on simultaneous globaldrug development. This article describes the initialobservations of the committee.

� Saillot JL et al. Industry efforts on simulta-neous global development. Drug InformationJournal 2009; 43:339–347.

Adaptive trials

Adaptive clinical trials are becoming very popularbecause of their flexibility in allowing mid-streamchanges of sample size, endpoints, populations,etc. At the same time, they have been regardedwith mistrust because they can produce bizarreresults in very extreme settings. Understandingthe advantages and disadvantages of theserapidly developing methods becomes a necessity.This paper reviews flexible methods for samplesize re-estimation when the outcome iscontinuous.

� Proschan MA. Sample size re-estimation inclinical trials. Biometrical Journal 2009; 51:348–357.

Statistical practice and education

The authors present the results of their study thatinvestigated doctors’ use of, and attitudes toprobability and statistics with a view to informingeducation of medical school students in probabil-ity and statistics. The results of this study can helpguide how these techniques should be taught tomedical undergraduates and should encourage

today’s medical students of the subjects’ relevanceto their future careers.

� Swift L et al. Do doctors need statistics?Doctors’ use of and attitudes to probabilityand statistics. Statistics in Medicine 2009;28:1969–1981.

Statistical methodology

The Wilcoxon–Mann–Whitney test is a popularnon-parametric technique used to compare themean or median of two distributions. Based on awide-ranging simulation study, the authors showthat the problem of lack of robustness (todifferences in the shapes of the distributions) ofthis test is more serious than is thought to be thecase. They offer theoretical reasoning for thisbreakdown and practical advice on examining thesamples prior to applying the test.

� Morten W et al. The Wilcoxon-Mann-Whitneytest under scrutiny. Statistics in Medicine 2009;28:1487–1497.

Sample size formulae are developed to test meandifferences between a treatment and control groupwhen noncompliance and selection bias areexpected. Selection bias in the form of systematicdifferences in mean and variance among latentadherence subgroups are incorporated and theseare illustrated via examples.

� Shardell MD et al. Calculating sample sizefor studies with expected all-or-none non-adherence and selection bias. Biometrics 2009;65:635–639.

Copyright r 2009 John Wiley & Sons, Ltd. Pharmaceut. Statist. 2009; 8: 253–254DOI: 10.1002/pst

254 Literature Review