In January 1994 the head of a British university department of psychology wrote to a pharmaceutical company in France suggesting that a research project be conducted in his department on the effects of an investigational drug in a realistic animal model of stress. Following consideration of this letter, the company drafted an agreement between itself and the professor, taking into account that the professor had developed an animal model inducing chronic mild stress in the rat and that the company was interested in the testing of a new medicine for the treatment of stress. For performing this study the professor's department was to be paid an agreed sum of money.
The member of staff who was designated by the professor to conduct this study was a postgraduate student in psychology, who was interested in the animal model that was to be used for this study. Her own experimental findings, however, led her to conclude that, whereas the drinking of sucrose solution was said to be reduced when rats were subjected to a battery of mild stresses, the drinking of sucrose solution was significantly increased when the rats were further stressed. After subsequent discussion with the professor she repeated her experiment using, first, outbred rats - with the same result - and then with inbred rats.The experiment with these last rats demonstrated a transient reduction in sucrose drinking after mild stress, but this returned to normal after three weeks, and after two further weeks the stressed rats increased their sucrose drinking just as the rats in her two previous experiments had done.
When she conducted the experiment with the rats yet again, she found no difference in sucrose drinking at all between the two groups. She sought to find a reason why her experimental findings failed to match the model and came up with a possible explanation. Regardless of this, however, the professor ordered the student to repeat the study with the control rats becoming the experimental rats and vice versa. On this occasion there was a reduction in sucrose drinking in the rats which had previously been the controls - but as there were, in effect, now no controls, the observation was considered meaningless. Nevertheless she was further instructed to begin treatment with the investigational drug but this made no observable difference to the drinking patterns of either groups of rats. Her conclusions were that:
• the differences observed between the results she obtained in her experiments and those described by the professor in his original papers were profound, suggesting that the chronic mild stress model was flawed;
• the switch between controls and experimental rats in the repeat experiment rendered that experiment invalid;
• there was no evidence that the investigational drug effected any changes in the experimental rats.
The case was submitted to MedicoLegal Investigations and a detailed comparison was made between the reports submitted to the company by the professor in December 1994 and April 1996 with the data yielded by the postgraduate student when she conducted the study herself, and a number of differences were noted:
• Inappropriate sucrose data appear to have been used and analysed.
• Critical sucrose data have been omitted.
• The fact that an effect, which had disappeared by week 2, was hidden.
• Experimental groups were not matched.
• So-called drug-induced differences were likely to be the result of differences between groups that existed before the drug was administered.
• Inappropriate mean sucrose intake values (and standard errors of the mean) were used.
• Values were inserted for missing values from the original data set without justification.
c. Clear guidelines as to the path to be followed once fraud is suspected.
d. Clear guidelines as to the right of appeal if a complainant feels that his/her concern is being inappropriately addressed.
e. Access to appropriate statistical screening methods which can be used efficiently on a routine basis.
f. A policy that all cases involving clinical trials on medicinal products, which give rise to serious concern regarding a specific investigator, should be notified at an early stage to the ABPI Director of Medical Affairs.
Clearly it is important that anyone involved in the detection of suspected research misconduct needs advice and guidance, given that, in spite of the thrust of this book, a suspicion of research misconduct will remain a rare event. Advice on where such guidance is available can come from the ABPI, which is absolutely committed to giving all possible assistance in advising on action against fraud. Appropriate publicity should be given to the outcome of such action, to act as a deterrent, in a determination to minimise fraud and misconduct in clinical research. Only the utmost vigour in applying this policy will be successful; it is in the ultimate interests of patient safety to do so. Would that official bodies recognised these priorities as well.
Was this article helpful?