Detecting possible misconduct and whistleblowing

Each trial management group will need to decide how assiduously, if at all, it tests data for possible misconduct in a trial. In multi-centre trials, there are circumstances, such as the following, that should arouse suspicion; and statistical methods are available to detect the plausibility of data received from centres. These rely on the difficulty of fabricating plausible data: human beings are poor generators of random numbers. For a full description, see the report prepared by the International Society for Clinical Biostatistics Subcommittee on Fraud [11].

♦ Data are unconvincingly tidy. Data from one centre can be compared with data from all centres with respect to the distribution of variables taken either in isolation (univariate) or jointly (multivariate). False data may lie 'too close' to the mean and lack variance and outliers. Multivariate observations may lack consistency and fail to correlate with other data in ways that would be expected.

♦ False data may exhibit digit preference.

♦ Data may be normally distributed when in fact normal distribution is implausible.

♦ Compliance with the protocol may seem too good to be true.

♦ Changes over time and dates of attendances and records may be unlikely.

If a trial management group has possible evidence of suspect data, then it should first consider raising the matter with the person who provided the data. It needs to be recognized that assuming the role of whistle-blower is difficult. Most of us would be loath to make unjustified allegations, but willfully concealing misconduct is itself misconduct. Junior researchers, in particular, may fear reprisals, and there might even be the threat of libel. Both whistle-blowers and those they accuse must be treated justly; allegations can ruin the careers of both accused and accusers. Nevertheless, it is very important to report observed, suspected or apparent misconduct to the director of the group concerned. This should initially be done informally and confidentially. If it seems that misconduct may have occurred, the standard procedures for the group concerned should be set in motion.

Whistle-blowers are responsible for making allegations in good faith, maintaining confidentiality, and cooperating in any assessment or investigation that maybe necessary.

0 0

Post a comment