Legault 2011.

Methods	Design: 4‐arm randomised single‐blinded controlled trial with 2 × 2 factorial design Recruitment period: 2008 to 2009 No. of centres involved: 1 Unit of randomisation: individuals No. randomised: 73 Number of arms considered in this review: 4 Maximum trial duration: 4 months Funding by non‐profit organisation: Department of Health and Human Services, National Institutes of Health (1R01AG029285 ‐ 01A1), and the General Clinical Research Center of Wake Forest University Baptist Medical Center (M01‐RR07122) Funding by commercial organisation: none reported Publication status: full‐text report
Participants	Patient flow: 18 randomised, 18 described at baseline in experimental group; 18 randomised, 18 described at baseline in control group Number of females: 8 of 18 (44%) in experimental group 1; 7 of 18 (38%) in control group 1 Average age (SD): 76 (5.2) years in experimental group 1; 75 (4.8) years in control group 1 Education: experimental group 1: high school or less: 4 (22%), more than high school: 14 (78%); control group 1: high school or less: 5 (28%), more than high school: 13 (72%) Baseline cognitive function: experimental group 1: selection criteria on cognition overall: community‐dwelling persons, aged 70 to 85 years, who were at risk for cognitive decline but who did not have mild cognitive impairment Selection criteria on cognition in experimental group: community‐dwelling persons, aged 70 to 85 years, who were at risk for cognitive decline but who did not have mild cognitive impairment. Control group: community‐dwelling persons, aged 70 to 85 years, who were at risk for cognitive decline but who did not have mild cognitive impairment Ethnicity: experimental group: 17 white, 0 Indian, 0 Asian, 1 black, 0 other, 0 unclear; control group: 17 white, 0 Indian, 0 Asian, 1 black, 0 other, 0 unclear Genetic marker: experimental group: present: 2 (17%), absent: 10 (83%); control group: present: 3 (25%), absent: 9 (75%)
Interventions	Type of experimental intervention: computerised CT group; treatment duration of 17.2 weeks; intervention provided in small group format under supervision Details of experimental intervention: sessions were centre‐based, conducted via computer, carried out with small groups of no more than 6 individuals, and monitored by skilled trainers. For each session, participants studied a list of 30 words, followed by a recognition test consisting of 30 studied words and 30 new words with each new word repeated once, and were asked to respond "yes" to study words and "no" to new items both times they occurred Type of concomitant treatment provided: none in comparison 1; physical activity in comparison 2 Session duration: 10 to 12 minutes in experimental group Number of treatment sessions: 24 in experimental group Treatment frequency: training consisted of 4 consecutive 10‐ to 12‐minute sessions per day, administered 2 times per week for 2 months, which then tapered to 1 time per week for 2 additional months in experimental group Maximum treatment duration in weeks: 17.2 in experimental group Type of control intervention: inactive; treatment duration of 17.2 weeks; intervention provided in group format, under supervision Details of control intervention: Healthy Aging Education control intervention consisted of weekly lectures based on health education and was based on a programme developed originally at Stanford and adapted for the Lifestyle interventions and Independence for Elders pilot trial. Topics such as medications, foot care, travelling, and nutrition were covered Type of concomitant treatment provided: none in comparison 1; physical activity in comparison 2 Session duration: not reported in control group Number of treatment sessions: not reported in control group Treatment frequency: 1/week in control group Maximum treatment duration in weeks: 17.2 in control group
Outcomes	Cognitive functioning outcomes considered for both comparisons Episodic memory measured with Logical Memory task from the Wechsler Memory Scale‐III (LM2), Recall Total Score at 4 months, on a scale from 0 to not reported with higher values indicating benefit Executive functioning measured with Trails B Time‐Trails A Time at 4 months, on a scale from 0 to not reported with lower values indicating benefit Physical functioning outcome considered: none reported Quality of life outcome considered: none reported Safety outcome considered: none reported Depression outcome considered: none reported Other outcome data on cognitive functioning, not considered in our meta‐analyses Episodic memory measured with Logical Memory task from the Wechsler Memory Scale‐III (LM1) ‐ Supplemental Score, 1st Recall at 4 months, on a scale from 0 to not reported with higher values indicating benefit Episodic memory measured with Hopkins Verbal Learning Test (HVLT), immediate recall at 4 months, on a scale from 0 to not reported with higher values indicating benefit Episodic memory measured with Hopkins Verbal Learning Test (HVLT), delayed recall at 4 months, on a scale from 0 to not reported with higher values indicating benefit Executive functioning measured with Flanker Task, Incongruent‐Congruent RTs at 4 months, on a scale from 0 to not reported with higher values indicating benefit Executive functioning measured with Task Switching, Switch‐Non‐switch RTs at 4 months, on a scale from 0 to not reported with higher values indicating benefit Executive functioning measured with Self‐Ordered Pointing Task, % correct at 4 months, on a scale from 0 to not reported with higher values indicating benefit Executive functioning measured with 1‐Back, % Hits‐False Alarms at 4 months, on a scale from 0 to not reported with higher values indicating benefit Executive functioning measured with 2‐Back, % Hits‐False Alarms at 4 months, on a scale from 0 to not reported with higher values indicating benefit
Notes	As we pooled outcome data from the 2 comparisons within the trial, before pooling across trials, 50% of participants in the pooled experimental group and 50% in the pooled control group received standardised physical activity. Pooling was justified, as no interaction effect of physical activity was observed. Study authors stated: "Depending on the choice of outcome, two‐armed full‐scale trials may require fewer than 1000 participants (continuous outcome) or 2000 participants (categorical outcome)". One SAE occurred, but the trial authors did not report in which trial arm
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Judgement: method of random sequence generation not reported Quote(s): "following this, they were randomly assigned with equal probability among the four experimental conditions"
Allocation concealment (selection bias)	Unclear risk	Judgement: method of allocation concealment not reported Quote(s): "following this, they were randomly assigned with equal probability among the four experimental conditions"
Blinding of participants (performance bias)	High risk	Judgement: blinding not feasible
Blinding of personnel (performance bias)	High risk	Judgement: blinding not feasible
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Judgement: study described as "single‐blinded", and at clinicaltrials.gov, it is explicitly described that outcome assessors were blinded Quote(s): "clinicaltrials.gov ‐ Masking: Single Blind (Outcomes Assessor)"
Incomplete outcome data (attrition bias) All outcomes	Low risk	Judgement: for the outcome executive functioning: statistical analyses were reported to be done according to the intent‐to‐treat principle. Study authors described the analysis as being done according to the ITT principle, but we wonder if they referred only to the principle that the participant was analysed in the group to which he/she was randomised, regardless of cross‐over. We are unsure if the 2 participants with missing data in the experimental group and the 1 in the control group were included in the analyses. We treated them as not analysed in our meta‐analyses, in accordance with how the trial authors depicted their data in the tables
Selective reporting (reporting bias)	High risk	Judgement: all outcomes indicated in the methods section are adequately addressed in the results section, but at least 2 instruments (perceived cognitive functioning problems and quality of life) mentioned in NCT00688155 are not mentioned in the full publication
Other bias	High risk	Judgement: comparison 1: no other potential sources of bias detected; comparison 2: attendance rate in the combined CCI and physical activity group was statistically significantly better than in the physical activity only control group. The direction of bias would likely inflate CCI effects; we thus judged high risk of bias for comparison 2. Quote(s): "intervention attendance rates were higher in the CT and PACT groups: CT: 96%, PA: 76%, PACT: 90% (P = 0.004)"