# 40. Discard unfavourable data ## 40.5. Methodology/Refinements/Sub-species ### 40.5.1. Dumping data When data contradict a central theme, it is often more convenient for a manipulator to simply dump that data and pretend that they never existed. This is hard to detect since the data are not published and hardly anyone knows that the data even existed. It's a hard case to prove for a victim. Reboxetine trials: Very often pharmaceutical companies dump unfavourable data, as in the trials of the antidepressant drug Reboxetine. Trials comparing Reboxetine against other similar drugs in three small studies (507 patients in total), showed that Reboxetine was just as good as any other drug, but 1,657 patients' worth of data was left unpublished, and these data showed that patients using Reboxetine did worse than those on other similar drugs. Worse still was the handling of the side-effects data. The drug looked fine in the trials that appeared in headline academic literature, but the unpublished studies revealed that patients were more likely to have side-effects and withdraw from the trial if they were taking Reboxetine rather than one of its competitor products. Obviously, this kind of unflattering information was not for general release and was effectively "buried" by the pharmaceutical company. In the published data, Reboxetine was a safe and effective drug, but in reality it was no better than a sugar pill and, worse, it did more harm than good. According to a systematic review and meta-analysis by IQWiG, the German Institute for Quality and Efficiency in Health Care, "published data on Reboxetine overestimated the benefit of Reboxetine versus placebo by up to 115% and Reboxetine versus (other similar drugs) SSRIs by up to 23%". It also underestimated harm, concluding that Reboxetine was an ineffective and potentially harmful antidepressant. The study also showed that nearly three quarters of the data on patients who took part in trials of Reboxetine was not published by Pfizer until much later. By early 2007, Reboxetine was licensed worldwide in over 50 countries, including Italy, Germany and the United Kingdom. But in May 2007, however, the US FDA shockingly declined Pfizer - Pharmacia's license application for the United States market. **Data dumping by religious investigators:** It's not just big corporations that get tempted to dump contradictory data. In the fundamental Christian religious hothouses of the United States, the use of selective data is quite common. A certain infamous study on "distance healing using prayer" was "widely acknowledged as the most scientifically rigorous attempt ever to discover if prayer can heal". This was a rather incredible claim to outside observers, given that the study only had 40 participants. The study has since been discredited for improprieties in mining data. The study was originally designed to test whether prayer could have an effect on the death rate from AIDS. However, the objective was changed after the study was completed, to see if there were any significant correlations between prayer and about two dozen illnesses associated with AIDS such as diarrhoea and oral thrush. The statistician was finally able to find about half a dozen significant correlations. These correlations were then selected as if they were the main purpose of the study, after it was apparent that prayer had no effect on the survival of the AIDS victims. This is the same manipulative concept as "switching dependent variables" discussed below. Normally a dependant variable is defined and fixed in the original hypothesis, like "the advance of infection of AIDS" in this case. When this didn't show any correlation with prayer, the researchers switched to another bundle of diseases, until they finally stumbled on some spurious coincidental correlations. They then used these apparently positive results to prove their original point, as if this was always the intention. They quietly dumped the "uncooperative" test results. Finally they had "evidence" that distance healing could be helped by prayer. Thank God.