Gelenberg 1990.

Methods	Allocation: randomised, procedure conducted independently by trial statistician, stratified by maintenance antipsychotic drug therapy. Blindness: double, adequate. 1 blinded rater assessed TD and psychopathology. 1 open rater assessed side‐effects and distributed medication. Duration: 18‐20 weeks (4 weeks baseline, 8 weeks followed by 2‐4 weeks washout and then crossed to another 8 weeks). Design: cross‐over. Setting: patients recruited from mental health centres and private physicians, USA.
Participants	Diagnosis: schizophrenia (9), bipolar (6), major depression (3), generalised anxiety disorder (1), brief reactive psychosis (1), no psychiatric diagnosis (1). TD diagnosed by psychiatrist and neurologist using criteria. History: TD present 6 months ‐ 17 years (median 1.5 years). N = 21. Sex: 10 M, 11 F. Age: median 47 years, range 19‐70 years.
Interventions	1. Lecithin: containing PC 20 g/day. N = 5 (completers).* 2. Placebo. N = 9 (completers).* Antipsychotics stable during trial. No anticholinergics permitted. Patients took the following concomitant psychoactive medications during the trial: antipsychotic agents alone (N = 7), antipsychotic drugs plus lithium (N = 3), antipsychotic drugs plus trazodone (N = 1), antipsychotic drugs plus an antianxiety agent (N = 1), antianxiety drugs alone (N = 1), antianxiety drugs plus lithium (N = 3), and lithium alone (N = 1).
Outcomes	TD symptoms: AIMS. Unable to use ‐ Global impression: CGI (not reported). Movement disorders: TAKE (reported only final summary scores from both segments, after cross‐over). Mental state: BPRS, HAM‐D (reported only final summary scores from both segments, after cross‐over). Adverse effects (reported only final summary scores from both segments, after cross‐over). Leaving study early (reported only final summary scores from both segments, after cross‐over).
Notes	* No information given on how many were originally allocated to each group. 14 of 21 completed the trial. Sponsorship source: Funded by National Institute of Mental Health grant, the Arbour Research Foundation, and the Center for Brain Sciences. ITT analysis not performed for AIMS scores (results reported only for completers). Physiology (lab‐tests, ECG, serum choline) monitored during trial. No clinically important changes in lab variables or vital signs during study. Serum choline levels doubled during lecithin treatment. Authors contacted, awaiting further information. Details of allocation procedure from authors.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"random‐order," "patients were stratified by whether they were on maintenance antipsychotic drug therapy." Details of sequence generation not reported.
Allocation concealment (selection bias)	Low risk	Allocation concealment not reported, but procedure confirmed as adequate from study authors.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	"double‐blind". "Several lecithin preparations were used during the study. We started with frappes prepared with chunks of 55% PC. That preparation was succeeded by chunks, chicken soup, and granola bars that contained 80% to 100% PC. Placebo included corn oil in frappes, ground corn flakes, and matching chicken noodle soup and granola bars." Unclear if the lecithin and placebo preparations were identical (color, taste, smell...)."The 14 completers were asked to fill out a questionnaire in which they specified (l) which of the two medications they thought was most helpful, (2) what effects (if any) they noted on their mood, and (3) whether they could guess which of the two medications was lecithin and which was placebo. Seven of the 14 patients felt that one treatment was definitely more helpful than the other; of those, 6 indicated that lecithin was the more helpful treatment."
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	"We used two clinical raters, one blind rater who assessed TD and psychopathology and one open rater who rated side effects and distributed medication." "Both the blind rater and the patient completed Clinical Global Impressions and Improvement ratings at each visit and the blind rater assessed extrapyramidal effects with the Target Abnormal Kinetic Effects (TAKE) scale."
Incomplete outcome data (attrition bias) All outcomes	High risk	"Fourteen patients‐7 men and 7 women‐completed at least 3 visits on the second leg of the trial. Data from these 14 completers were used in the efficacy analyses." Number completed the first period and number completed the trial not reported. 14/21 participants were entered to the analyses (approximately 33% drop out).
Selective reporting (reporting bias)	High risk	Clinical Global Impressions and Improvement, Target Abnormal Kinetic Effects (TAKE) scale, Mental State (BPRS and HAM‐D), adverse effects, and leaving the study early not fully reported.
Other bias	Low risk	The study seems to be free from other sources of bias.