We agree with Cliff Prior's assertion (letters, 22 January) that the pharmaceutical industry's decades of research into drug treatments for schizophrenia have failed to answer which drugs cause fewer adverse effects and are more acceptable to users.

Our main concern about new antipsychotic drugs is that they have not yet been proven to cause less adverse effects or be more acceptable to users than the old antipsychotic drugs, nor have they been proven to be more clinically or costeffective.

In the absence of this evidence, the pharmaceutical industry expects the NHS to fund these drugs as a first-line treatment for schizophrenia.

We wonder which aspects of clinical services should be withdrawn to fund these drugs, given finite mental health resources.

It is good to hear of the National Schizophrenia Fellowship's investigation into quality of life among users of old and new antipsychotic drugs, and we agree that such research is long overdue.

We would, however, urge it to take great care over the design, wording and preparation of its proposed survey, as it is very easy to inadvertently introduce bias to this kind of research. We would not want to see this potentially exciting and important work rendered unusable by fundamental design faults.

We fully agree that an evidence-based approach should take account of users' experiences, but feel this is most reliably achieved by undertaking properly conducted, large, long-term randomised drug trials involving participants, interventions and primary outcome measures relevant to everyday practice.

Such randomised evidence should form the basis of mental health policy, but our review shows we cannot depend on the pharmaceutical industry to produce this.

Paul Wilson Anne-Marie Bagnall Simon Gilbody NHS Centre for Reviews and Dissemination York University