Atypical antipsychotic drugs such as clozapine and sertindole appear to be no more effective than conventional drugs, according to a report by UK researchers.

The metaanalysis, by Dr John Geddes and colleagues at the University of Oxford, involved data from 12,649 patients across 52 randomised trials comparing amisulpride, clozapine, olanzapine, quetiapine, risperidone and sertindole with conventional drugs such as haloperidol or chlorpromazine.

It has long been assumed that the newer atypical antipsychotics are better than the conventional ones. But, unlike atypical antipsychotics, doctors tend to prescribe the old drugs at very high doses, which produce more adverse effects without corresponding benefits to the patient.

In this analysis, the atypical antipsychotic drugs were slightly more effective and tolerable than the conventional agents, and produced fewer extrapyramidal side effects.

But, overall, the researchers found that when the dose was 12 mg daily or less of haloperidol (or an equivalent), the results were similar for all of the drugs examined. They therefore advise that conventional drugs should be the first line of treatment.

On the face of it, this seems like good news and might even simplify guidelines for prescribing antipsychotics. But, in an editorial accompanying the study, Drs Shitij Kapur and Gary Remington, from CAMH, Toronto, Canada, commented that most of their patients, who don't have to pay for medications, prefer the atypical antipsychotics because of the lower incidence of adverse effects.

They added that their own clinical experience plus the findings by Dr Geddes and colleagues "leave the clinician on a tightrope act between the persuasiveness of the marketing claims, the precise but somewhat myopic results of idealised clinical trials and the complex realities of clinical practice".

No mention of where this leaves the patient, who might feel better on one type of medication, but be unable to persuade his or her doctor that the evidence of their experience is as valid as the evidence of clinical trials (BMJ, 2000; 321: 1360-1, 1371-6).