Teri 2000.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Study duration: 16 weeks Rescue medication: not allowed for NPS or EPS.
Participants	Number randomised: 70 (34 haloperidol, 36 placebo) Mean age: 75.5 years Sex (female): 62.9% Type of dementia: Alzheimer's disease Severity of dementia: moderate‐severe Indication: agitation Setting: community Country: USA
Interventions	Intervention characteristics Haloperidol dosage: 0.5 mg to 3.0mg/day Placebo Medication was provided in tablets of haloperidol 0.5 mg, treatment began with one tablet each day and increased at the next clinic visit by one tablet a day unless the subject was rated as moderately or markedly improved, significant adverse events were noted, or the maximum dose was reached (3 mg/day for haloperidol).Participants achieved a mean haloperidol dose of 1.861 mg/day (range, 0 to 3 mg/d)ay.
Outcomes	Agitation: Cohen‐Mansfield Agitation Inventory (CMAI) Number of responders for agitation Extrapyramidal symptoms (Notes: Parkinsonian gait, rigidity, tremor, and bradykinesia are forms of EPS. Rigidity showed the largest contrast to avoid underestimating the risk of EPS: 33% (11) and 13% (5)). Somnolence Death Discontinuation (any reason) Cognitive function: Mini‐Mental State Examination (MMSE) Functioning in activities of daily living (ADL): Instrumental Activities of Daily Living (IADL) Carer burden or carer quality of life: The Screen for Caregiver Burden (SCB)
Identification
Notes	Sponsorship source: supported by a grant from the National Institute of Aging (AG‐010483). Active study medications and corresponding placebos were provided by Purpac Pharmaceutical, Elizabeth, NJ
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information
Allocation concealment (selection bias)	Unclear risk	Not enough information provided.It was a "randomized" trial. "Treatments were assigned in randomized blocks of nine (for three arms) or 12 (for four arms)." Restriction methods could potentially affect allocation concealment.
Comparability of groups (selection bias)	Low risk	There were small differences between the treatment groups. The authors found that "None of the prestated potential confounding variables [patient age, patient gender, MMSE score, total CMAIscore, and caregiver gender] [...] was significantly related to primary outcome." And the trial was a negative trial.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Not completely blinded, because one arm (behaviour management techniques) could not be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	"Assessments were conducted [...] by interviewers blind to treatment assignment." "To insure that interviewers remained blind to treatment assignment, caregivers did not discuss any aspect of their treatment with the interviewer. In no instance was the blinding compromised.""Medication was provided in [...] identically appearing tablets." "The primary outcome measure was [...] completed by a trained clinician blind to treatment assignment."
Incomplete outcome data (attrition bias) All outcomes	Low risk	Overall drop‐out was high (36%).ITT‐analysis: "Missing scores at posttreatment were imputed with last observation carried forward from discontinuation visit scores or, if those were not available, from midpoint visit scores. If no ADCS‐CGIC score [primary outcome, rev] was available, a value of “worse” was assigned (n=12). The rationale for doing this was to assume the worst case scenario: subjects who dropped out did so because they became worse. In addition,we reviewed the reasons caregivers cited for dropping out of the study and in each instance our assignment of“worse” was confirmed. "
Selective reporting (reporting bias)	Low risk	All outcomes seem to be reported.
Other bias	High risk	Subjects receiving psychotropic medications were required to discontinue them at least 2 weeks before study enrolment.