Background: There is variable attrition from randomised controlled trails (RCTs) evaluating treatments for people with schizophrenia. Although statistical assumptions can be made to compensate for attrition and the effects of these assumptions tested, it is unclear at what degree of loss from such studies clinicians, researchers and service users or their carers begin to consider results meaningless.
Objectives: To derive estimates for subjectively acceptable attrition rates in schizophrenia trials as viewed by three different groups of stakeholders (clinicians, researchers and service users/carers).
Methods: Postal/e-mail survey March 2007; Piloted questionnaire asking a single question regarding loss to follow up in a schizophrenia trial. Sampling frame; all 128 consultant psychiatrists in General Psychiatry from the Yorkshire Deanery; 100 trialists in schizophrenia whose e-mail is publicly available; 100 service users and carers chosen by RETHINK, a UK-based mental health charity and service provider.
Results: We think this study is the first of its kind in health care. We have sent out a first wave of questions. Response rate to the question varies. About 1/3 of clinicians have replied on first request, 1/3 of researchers and over 90% of service users. A second request is now due and we expect response rates to improve. Currently, from all three stakeholders the level of attrition at about 12 weeks beyond which data are simply not trusted is about 20-30%. If this is so, and full data will be available by October 2007, there are implications for the interpretation of reviews.