Zhang 2017.

Methods	RCT, parallel, (n‐3 DHA vs n‐6 LA), 12 months Summary risk of bias: moderate to high
Participants	Otherwise healthy elderly people with mild cognitive impairment. N: 120 intervention, 120 control (analysed, intervention: 110 control: 109) Level of risk for CVD: low Men: 35.8% intervention, 34.2% control Mean age in years (SD): 74.5 (2.65) intervention, 74.6 (3.31) control Age range: eligibility criteria were age 65‐85 years at trial start Smokers: 59.17% intervention, 61.67% control Hypertension: 9.17% intervention, 7.50% control Medications taken by at least 50% of those in the control group: not reported Medications taken by 20%‐49% of those in the control group: not reported Medications taken by some, but less than 20% of the control group: not reported Location: China Ethnicity: assumed Chinese
Interventions	Type: supplement (capsule) Comparison: DHA vs corn oil (n‐6) Intervention: 1 capsule twice a day, with meals, including 2 g algal DHA (45‐55% DHA by weight). Martek Biosciences, Columbia, MD. Dose: ˜1 g/d DHA Control: corn oil, orange‐flavoured and orange colour to protect the study blind Compliance: participants were asked to return any remaining tablets. Compliance was defined as a ratio (actually taken/should have taken). Achieved 97% for intervention, 95% for control. Serum levels of DHA also measured, DHA at 6 months barely higher in intervention than in controls Duration of intervention: 12 months
Outcomes	Main study outcome: cognitive function and hippocampal volume Dropouts: 10 intervention, 11 control Available outcomes: mortality, cognitive outcomes and cerebral volume measurements Response to contact: no reply to date
Notes	Study funding: Chinese Nutrition Society (CNS) Nutrition Research Foundation‐ DSM Research Fund
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated, also statistics analyst ignorant to this study used random number table
Allocation concealment (selection bias)	Unclear risk	Insufficient information
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Placebo capsules … identical in appearance. All capsules were orange‐flavoured and orange colour to protect the study blind . Packaged into identical pots, each containing 180 capsules, and labelled by staff who were not involved in the study. A blinding key linked each participant to his or her assigned treatment. This key was kept by an investigator not involved in any data collection or analyses, in a secure electronic file. The code was revealed at the completion of the trial following analyses of the main study aims.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	All project staff were unaware of group assignments until the completion of the trial and after data analysis
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	They did not describe how they imputed missing data (lost contact with patients, but called this an ITT analysis). Overall well matched and not high attrition.
Selective reporting (reporting bias)	Low risk	Registered trial prospectively. Outcomes match protocol
Attention	Low risk	"Adherence was encouraged and monitored throughout the trial by telephone assessment at 15 time points, and by blood assay at baseline" 6 months and 12 months. This and assessments were described as same for both arms.
Compliance	Unclear risk	Quote: "participants were requested to return any remaining tablets in order to measure compliance, together with the replenishment of capsules for the following month." Compliance … defined "as a ratio = actually taken/should have taken". "Adherence was encouraged and monitored throughout the trial by telephone assessment at 15 time points, and by blood assay at baseline" 6 months and 12 months On compliance tree, leads to "No, because no P values were supplied" therefore risk of compliance bias unclear
Other bias	Unclear risk	Although the register says single blind, the publication very clearly describes a double‐blind RCT