Amieva 2016.

Study characteristics
Methods	A multicentre randomised, parallel‐group trial, with a two‐year follow‐up comparing group cognitive training, group reminiscence therapy, individual cognitive rehabilitation, and usual care
Participants	653 people (237 men) with mild‐to‐moderate AD (MMSE 16 to 26 and GDS 2 to 5), ≥ 50 years old, living at home, and with an identified care partner
Interventions	Individualised cognitive rehabilitation (individual sessions): participants received individual sessions (settings not specified) where meaningful activities (instrumental ADL or leisure) were selected for training in line with personal goals and then the training strategy was adjusted to match the person’s cognitive ability with errorless learning principles used where possible; goals/trained activity could be changed during the programme and could include personalised reminiscence activities Cognitive training (group sessions): structured programme of a set of standard tasks relating to and tapping on a particular activity of daily life, with two levels of difficulty to match the level of ability Reminiscence therapy (group sessions): participants were engaged in group discussions on different personal themes Usual care: participants received usual medical care excluding non‐drug therapy, and care partners received support group sessions once a week during the first 3 months and every 6 weeks afterwards All interventions involved 43.5 hours with a therapist over 24 months, with 14 weekly 90‐minute sessions in the first 3 months and 16 6‐weekly 90‐minute sessions thereafter, with parallel care partner support group sessions for the cognitive training and reminiscence therapy groups, and telephone support contact in the cognitive rehabilitation group Only the individualised cognitive rehabilitation and usual care control conditions were compared for the purposes of this review
Outcomes	Outcomes were collected at 3, 6, 12, 18, and 24 months, but only data at the 3‐ and 24‐month time points were published The primary outcome was the rate of patients alive and without moderately severe to severe dementia (MMSE < 15 or GDS = 5 to 6) at 2 years. Secondary outcomes were: institutionalisation, cognitive ability (ADAScog), behavioural symptoms (Neuropsychiatric Inventory; NPI), functional abilities (Disability Assessment for Dementia; DAD, and Grille d’Autonomie Gérontologique‐Groupes Iso‐Ressources; AGGIR), apathy (Apathy Inventory; AI), depressive symptoms (Montgomery‐Asberg Depression Rating Scale; MADRS), quality of life (Quality of Life ‐ Alzheimer’s Disease scale; QoL‐AD), caregiver’s burden (Zarit Burden Interview), and resource utilisation (RUD Lite) No benefits were observed for the primary outcome in any of the groups. No benefits were observed in the secondary outcomes in cognitive training and reminiscence groups. In the individualised cognitive rehabilitation group, participants had better functional ability scores and a 6‐month delay in institutionalisation at 2 years was observed.
Notes	Not clear what proportion of participants in the individualised cognitive rehabilitation chose personalised reminiscence activities
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The list of randomization was prepared by a statistician using permuted blocks, stratified by site."
Allocation concealment (selection bias)	Low risk	Quote: "The list of randomization was prepared by a statistician using permuted blocks, stratified by site." No reason to assume that allocation concealment was ineffective as a centralised randomisation procedure was used.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and therapist were aware of their group allocation. There was a scope for potential bias in the self‐reported subjective outcomes.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	All assessments were administered by blinded researchers. There is no reason to assume blinding was not effective, although no measure of blinding efficiency was reported.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Authors presented data with and without imputation of missing values (missing equals failure analysis vs analysis on available data), with both analyses providing comparable results.
Selective reporting (reporting bias)	High risk	Authors described that participants were followed at 3, 6, 12, 18, and 24 months, but reported data for 3 and 24 months time points only. There is no information about what measures were completed at 6, 12, and 18 months, and why these were not reported.
Other bias	Unclear risk	There is a potential risk of intervention contamination as it is unclear what proportion of participants in the CR group had instead received individual reminiscence therapy. No other significant sources of bias were identified.
Training of those delivering the intervention	Low risk	All therapists received a 3‐day training (no other details given) and were offered telephone support if needed.
Intervention manual	Unclear risk	A manual mentioned by authors but not made available to readers.