Abha Khushu, Melanie J Powney

Psychotic disorders can lead some people to become agitated. Characterised by restlessness, excitability and irritability, this can result in verbal and physically aggressive behaviour - and both can be prolonged. Aggression within the psychiatric setting imposes a significant challenge to clinicians and risk to service users; it is a frequent cause for admission to inpatient facilities. If people continue to be aggressive it can lengthen hospitalisation. Haloperidol is used to treat people with long-term aggression. To examine whether haloperidol alone, administered orally, intramuscularly or intravenously, is an effective treatment for long-term/persistent aggression in psychosis. We searched the Cochrane Schizophrenia Group Trials Register (July 2011 and April 2015). We included randomised controlled trials (RCT) or double blind trials (implying randomisation) with useable data comparing haloperidol with another drug or placebo for people with psychosis and long-term/persistent aggression. One review author (AK) extracted data. For dichotomous data, one review author (AK) calculated risk ratios (RR) and their 95% confidence intervals (CI) on an intention-to-treat basis based on a fixed-effect model. One review author (AK) assessed risk of bias for included studies and created a 'Summary of findings' table using GRADE. We have no good-quality evidence of the absolute effectiveness of haloperidol for people with long-term aggression. One study randomising 110 chronically aggressive people to three different antipsychotic drugs met the inclusion criteria. When haloperidol was compared with olanzapine or clozapine, skewed data (n=83) at high risk of bias suggested some advantage in terms of scale scores of unclear clinical meaning for olanzapine/clozapine for 'total aggression'. Data were available for only one other outcome, leaving the study early. When compared with other antipsychotic drugs, people allocated to haloperidol were no more likely to leave the study (1 RCT, n=110, RR 1.37, CI 0.84 to 2.24, low-quality evidence). Although there were some data for the outcomes listed above, there were no data on most of the binary outcomes and none on service outcomes (use of hospital/police), satisfaction with treatment, acceptance of treatment, quality of life or economics. Only one study could be included and most data were heavily skewed, almost impossible to interpret and oflow quality. There were also some limitations in the study design with unclear description of allocation concealment and high risk of bias for selective reporting, so no firm conclusions can be made. This review shows how trials in this group of people are possible - albeit difficult. Further relevant trials are needed to evaluate use of haloperidol in treatment of long-term/persistent aggression in people living with psychosis.