Strategies for dealing with fraud in clinical trials

Research output: Contribution to journalReview articlepeer-review

Abstract

Research misconduct and fraud in clinical research is an increasing problem facing the scientific community. This problem is expected to increase due to discoveries in central statistical monitoring and with the increase in first-time clinical trial investigators in the increasingly global reach of oncology clinical trials. This paper explores the most common forms of fraud in clinical trials in order to develop offensive and defensive strategies to deal with fraud. The offensive strategies are used when fraud is detected during a trial and the defensive strategies are those design strategies that seek to minimize or eliminate the effect of fraud. This leads to a proposed fraud recovery plan (FRP) that would be specified before the start of a clinical trial and would indicate actions to be taken upon detecting fraud of different types. Statistical/regulatory issues related to fraud include: dropping all patients from a site that committed fraud, or just the fraudulent data (perhaps replacing the latter through imputation); the role of intent-to-treat analysis; effect on a planned interim analysis; effect on stratified analyses and model adjustment when fraud is detected in covariates; effect on trial-wide randomization, etc. The details of a typical defensive strategy are also presented. It is concluded that it is best to follow a defensive strategy and to have an FRP in place to follow if fraud is detected during the trial.

Original languageEnglish (US)
Pages (from-to)22-27
Number of pages6
JournalInternational Journal of Clinical Oncology
Volume21
Issue number1
DOIs
StatePublished - Feb 1 2016

Keywords

  • Clinical trial methodology
  • Fraud recovery plan
  • Regulatory strategies
  • Research misconduct

ASJC Scopus subject areas

  • Surgery
  • Hematology
  • Oncology

Fingerprint Dive into the research topics of 'Strategies for dealing with fraud in clinical trials'. Together they form a unique fingerprint.

Cite this