The current recommendation is that antidepressants should be continued for 4-6 months after the resolution of an acute episode of depression. This recommendation probably originates with Kuhn’s advice but subsequently several long-term studies appeared to show that people are more likely to relapse after stopping antidepressants compared to if they continue to take them. However these studies all use a discontinuation design and involve patients who have done well on antidepressants in the first place. Thus people who have recovered from depression and remained well for some time are randomised either to continue taking their antidepressant, or to have it withdrawn and replaced by an inert placebo. It is now recognised that withdrawal from antidepressants of all classes produces a discontinuation syndrome that among other symptoms includes adverse effects on mood, anxiety and sleep. These effects have not been distinguished from recurrence of depression in people who are withdrawn to placebo. Thus some of the relapses in the placebo group may simply be symptoms of drug withdrawal. In addition, discontinuation symptoms are also likely to have a substantial psychological impact. The presence of these symptoms is likely to unblind participants, who will be able to guess whether or not they have been allocated to placebo substitution. Given that people included in such trials are people who have done well with antidepressants in the first place, they are likely to have negative expectations of the outcome of stopping treatment. People who suspect they have been put onto placebo may therefore have a worse outcome because they believe they will do badly.
Viguera and colleagues reviewed 27 discontinuation studies and found that relapse rates were as expected substantially higher in those who had their antidepressants discontinued compared with those who did not. The increased risk of relapse was much higher in people who had had recurrent episodes of depression. Diagnosing relapse in depression is even more subjective than in psychotic disorders. In the studies reviewed it consisted of a subjectively defined worsening of depression severe enough to warrant resumed antidepressant drug treatment. As with the neuroleptic trials, the meta-analysis showed that the increased risk of relapse was highest immediately after discontinuation and that the difference in relapse rates between people who had their medication stopped versus those who continued fell progressively over time. This suggests that the act of discontinuation itself influenced relapse. However the average time before relapse occurred was much longer than after neuroleptic discontinuation, at around 14 months, and gradual withdrawal did not reduce the risk of relapse compared with abrupt cessation of treatment. Another observation in this analysis was that the time for which people had taken antidepressants and remained stable prior to discontinuation did not influence relapse rates. People who had been stable for only three weeks relapsed at around the same rate after discontinuation as people who had been stable for four years. In fact if anything there was a slight trend for people who had been stable for longer to have higher relapse rates after discontinuation. As people who are stable for longer would normally be at lower risk of a subsequent relapse, this is unexpected and again suggests that the process of withdrawal of the antidepressant itself acts as a precipitant to relapse.
The Harvard group who published this study proposed that pharmacological stress is responsible for relapse after antidepressant discontinuation, the same as they suggest for neuroleptics. There is little evidence that could support or refute this suggestion. One animal study found that rats withdrawn from long-term imipramine had a depressive reaction 40 days later. On the other hand, the fact that relapses took a considerable time to occur and the fact that no protective effect was observed for gradual discontinuation would argue against this explanation. Psychological mechanisms are likely to be equally or more important in a condition like depression. People who believe that their recovery is attributable to antidepressant drugs are likely to feel anxious and vulnerable if those drugs are withdrawn. The next time they encounter problems they will worry about having a recurrence of their depressive state, in what can soon become a self-fulfilling prophecy. The more often people turn to drugs to help them the greater their insecurity will be. Psychological explanations would fit with the longer time to relapse and might explain how people who have been stable for longer periods are more at risk, since they are likely to be more psychologically dependent on the drugs.