Newspaper Article

Special report: Antidepressants: Analysis: For some, drugs are only first aid

Cantopher, Tim

2013

Overview

It's part of our culture to take up polarised positions. Our political and legal systems are based on this premise and our media rely on it. The middle ground isn't interesting and is rarely aired. It is this environment in which the pros and cons dominate the literature on antidepressants. Meanwhile, sufferers from the depressive illness don't know which way to turn. The cons are equally selective. They point to meta-analyses showing that antidepressants don't out-perform placebo sufficiently to reach this arbitrary level of significance. They conclude that antidepressants don't work. This is the oldest misuse of statistics in the book. An insignificant difference doesn't mean that you've proven no difference; it just means that you haven't proved that there is one. Another issue is failing to exclude the outlier. As I've already pointed out, one expects occasional misleading results from research. A meta-analysis should deal with this problem by excluding from the analysis any study with results wildly different from all the others. The pros say this hasn't always happened in the cons' analyses, potentially producing misleadingly negative results.

• Publisher: Guardian News & Media Limited