Haloperidol versus placebo for schizophrenia

Haloperidol was developed in the late 1950s for use in the field of analgesia. Research subsequently demonstrated effects on hallucinations, delusions, aggressiveness, impulsiveness and states of excitement and led to the introduction of haloperidol as an antipsychotic. To evaluate the clinical effects of haloperidol for the management of schizophrenia and other similar serious mental illnesses compared to placebo. Electronic searches of Biological Abstracts (1985-1998), CINAHL (1982-1998), The Cochrane Library (1998, Issue 4), The Cochrane Schizophrenia Group's Register (December 2000), EMBASE (1980-1998), MEDLINE (1966-1998), PsycLIT (1974-1998), and SCISEARCH (January 1974-December 1998) were undertaken. References of all identified studies were searched for further trial citations. Authors of trials and pharmaceutical companies were contacted for further information and archive material. All relevant randomised controlled trials comparing use of haloperidol (any dose) with placebo for those with schizophrenia or other similar serious, non-affective psychotic illnesses (however diagnosed). The main outcomes of interest were death, loss to follow up, clinical and social response, relapse and severity of adverse effects. Reviewers evaluated data independently and analysed on an intention-to-treat basis, assuming that people who left the study early, or were lost to follow up, had no improvement. Where possible and appropriate, dichotomous data were analysed using relative risk (RR) and their 95% confidence intervals (CI) calculated. If appropriate, the number needed to treat (NNT) or number needed to harm (NNH) was estimated. For continuous data, weighted mean differences were calculated. Continuous data were excluded if loss to follow up was greater than 50%. All data were inspected for heterogeneity. Seventy-four trials were identified but only 20 included. More people allocated to haloperidol improved in the first six weeks of treatment than those given placebo (three trials, n=159, RR failing to produce a marked improvement 0.44 CI 0.3 to 0.6, NNT 3 CI 2 to 5). A further eight trials (n=313) also found a difference favouring haloperidol across the 6-24 week period (RR no marked global improvement 0.68 CI 0.6 to 0.8 NNT 3 CI 2.5 to 5) but this may be an overestimate of effect as small negative studies were not identified. About half of those entering studies failed to complete the short trials, although, at 0-6 weeks, 10 studies found a difference that