Battaglia 1997.

Methods	Allocation: randomised. Blindness: double. Duration: 24 hours.
Participants	Diagnosis: mania (N = 13), psychoactive substance use (N = 16), psychosis not otherwise specified (N = 27), schizophrenia (N = 47), schizophreniform disorder (N = 1).* N = 100.** Age: range 18‐57 years. Sex: 73 males, 25 females. History: "psychosis and behavioral dyscontrol (agitated, aggressive, destructive, assaultive, or restless behavior." Excluded: "clinically obvious alcohol intoxication, central nervous system (CNS) depression, pregnancy, allergic hypersensitivity, delirium, neuroleptic malignant syndrome, airway obstruction, severe hypotension or hypertension, acute narrow angle glaucoma and treatment with a benzodiazepine or neuroleptic within the previous 24 hours (previous two weeks for fluphenazine decanoate and previous 4 weeks for haloperidol decanoate)." Setting: Emergency department, USA hospitals.
Interventions	1. Haloperidol IM: dose up to 6 injections of 5 mg (mean number of IMs 2.86). N = 35. 2. Lorazepam IM: dose up to 6 injections of 2 mg (mean number of IMs 2.87). N = 31. 3. Lorazepam IM: dose up to 6 injections of 2 mg + haloperidol 5 mg (mean number of IMs 2.41). N = 32. First 3 injections ‐ at least 1 hour apart, remainder 2 hours apart. Need for subsequent doses made by blinded evaluator.
Outcomes	Tranquillisation or asleep. Global state: number of additional injections.** Adverse events: EPS, hypertension, others in over ≥ 9% of people. Unable to use ‐ Global state: CGI (reported, out of necessity, only for those awake), Efficacy Index (no data reported). Agitation: ABS (mean, SE/SD not reported, P values of significant findings reported, overall F value). Mental state: MBPRS (modified, unpublished; mean, SE/SD not reported, P values of significant findings reported, overall F value), Alertness Scale (no mean, SE/SD not reported, P values of significant findings reported). Physiological: vital signs (not reported).
Notes	* 6 people had more than one diagnosis. ** 2 people excluded from efficacy analysis ‐ after enrolment they had received proscribed antipsychotic medication.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Computer‐generated table of random numbers".
Allocation concealment (selection bias)	Unclear risk	Not stated.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	"The ED [Emergency Department] psychiatrist who treated and rated the patient remained blinded to the identity of the patient's medication throughout treatment" (p.336). [italic added by reviewers] "Double blind" ‐ participant blinding not specifically mentioned.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	"The ED [Emergency Department] psychiatrist who treated and rated the patient remained blinded to the identity of the patient's medication throughout treatment" (p.336).[italic added by reviewers]. Not tested.
Incomplete outcome data (attrition bias) All outcomes	High risk	"two patients were excluded from the efficacy analysis. It was determined after enrolment that both received proscribed antipsychotic medication before entering the study."
Selective reporting (reporting bias)	High risk	More details on level of significance for outcomes with P < 0.05. Adverse effects for central nervous system reported only if present in ≥ 9%.
Other bias	High risk	Sponsored by drug company.