Skip navigation

Drug Treatment of Obsessive-Compulsive Disorder (OCD)

Obsessive-compulsive disorder is a common psychiatric condition with a prevalence of up to 1% of adults, in which the patient engages in repetitive actions (compulsions, e.g. hand washing) or trains of thought (obsessions, e.g. counting rituals) which he knows are irrational but which, if he stops, lead to incapacitating anxiety.

Standard first line psychiatric treatment is either clomipramine (an older tricylic antidepressant with strong effects on the neurotransmitter serotonin) or one of the newer specific serotonin re-uptake inhibitors (SSRIs). A recent meta-analysis [1] attempts a systematic review of the evidence for this approach.

Systematic review of RCTs

Since OCD is a chronic, relapsing and remitting condition, where treatment effects are generally modest, the review is rightly restricted to randomised controlled trials. The authors describe a strategy of computer and manual searching to identify as many as possible, and found as many as 53.

Thirty-six were admitted into the analysis, and the reason for each exclusion is described clearly (e.g. "diagnostic criteria not strictly enforced" ( ! )).

The authors tried to obtain any necessary original data, for calculation of effect size. The statistical procedure used was Hedges `g' (an effect size measure), unfamiliar to most colleagues consulted.


The result of the meta-analysis is expressed in an unusual though logical format, as "increase in improvement rate over placebo", given as 61% for clomipramine and 22-28% for SSRIs.

Although the techniques used are difficult for the non-statistician, the results seem broadly supported by scanning the helpful tables summarising each study.

The review addresses publication bias (negative studies selectively unpublished: the "file-drawer" problem), by statistically suggesting the number of negative studies which would be needed to "cancel out" the conclusions of the review. Another possible source of bias, not considered, relates to methodological quality, particularly randomisation: rigorous randomisation is associated with lower treatment effects. In 60% of studies, results analysed excluded dropouts (i.e. no "intention to treat") analysis: this is a possible source of bias. The review could have been made more rigorous by arbitrarily assigning negative outcomes to non-completers.


This is a good systematic review, and does a great service in identifying 53 RCTs.

Although repeat analysis using different statistical procedures and addressing the above biases would be helpful, the authors seem justified in concluding cautiously that:
  1. Antidepressants are effective in the short term treatment of OCD.
  2. Clomipramine / SSRIs are more effective than non-serotonergic drugs.
  3. Concomitant depression is not necessary for effectiveness.

Practice Points

Clomipramine should be first line treatment in OCD as it is effective, well established and cheap.

SSRIs should be reserved for those with side effects or non-responders, as they are more expensive and newer. There was also a trend toward their being less effective than clomipramine.

Behavioural therapy and other psychological treatments are less readily available and more expensive in the short term. They may be tried when drugs have failed, and also in chronic cases, where the evidence for drug effectiveness is much weaker.

David Gill

Senior Registrar in Psychiatry, Oxford


M Piccinelli, S Pini, C Bellantuono, G Wilkinson. Efficacy of drug treatment in obsessive-compulsive disorder: A meta-analytic review. British Journal of Psychiatry 1995 166: 424-43.

previous or next story in this issue