Simpson 1988.

Methods	Allocation: "randomly assigned". Blindness: double, identical‐appearing tablets. Duration: 20 weeks (6 weeks observation, 4 weeks dose finding, 6 weeks' treatment, 4 weeks follow‐up). Design: parallel. Setting: Inpatients from 2 chronic care institutions, USA. Raters: not reported.
Participants	Diagnosis: tardive dyskinesia in subjects treated with antipsychotics. N = 17. Sex: 8 M, 9 F. Age: mean 46 years, range 32 to 70 years. Duration of TD: no information.
Interventions	1. Carbidopa/levodopa: full dose: 50/350 mg/d (6 weeks of treatment, and 4 weeks of follow‐up after drug withdrawal). N = 9. 2. Placebo (6 weeks of treatment, and 4 weeks of follow‐up after drug withdrawal). N = 8. "When the appropriate dose was established in the dose finding period, patients received that dose for the next 6 weeks". Concomitant medication: no information.
Outcomes	TD symptoms: improvement and deterioration (AIMS and Simpson Abbreviated Dyskiesia Scale). Leaving the study early. Unable to use ‐ Treatment‐related side‐effects. Mental state: BPRS, SANS (F and P values only).
Notes	Sponsorship source: not reported. Medication and placebo supplied by Merck Sharp and Dohme, Rahway, NJ. (Unclear if medications were supplied free of charge).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"Patients were randomly assigned"; further details not reported.
Allocation concealment (selection bias)	Unclear risk	Allocation concealment not reported.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	"active (Sinemet) or placebo tablets (supplied by Merck Sharp and Dohme, Rahway, NJ). Both groups of patients received the same number of identical‐appearing tablets."
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	"double‐blind". Details not reported.
Incomplete outcome data (attrition bias) All outcomes	High risk	"Fifteen of the 17 patients completed the trial; there were two dropouts. A female patient experienced "seizures" and the blind was, therefore, broken; a male patient eloped from the hospital. Both patients were found to be in the placebo group." 25% dropped out from the placebo group versus 0% in the active medication group. According to the degrees of freedom in the F‐test, only completers must have been analysed.
Selective reporting (reporting bias)	High risk	"Because the AIMS and Simpson scale were very highly correlated, only data from the Simpson scale are presented." Also, mental state data (BPRS and SANS) unusable: reported as F and P values. Adverse Events (Treatment Emergent Side Effects Scale) outcome data not reported.
Other bias	Unclear risk	Insufficient information reported to make a judgement.