Table 1.

Included studies

Main paper	A.D. Collaborative Group [28]	Thoonsen et al. [30]	Ferrari et al. [29]	Lee et al. [31]
Study type	AD2000 trial Open-label RCT 2 × 2 arms	Non-randomised data analysis from two trials (AD2000 and Richard)	Virtual cohort	Virtual cohort
Population	Multicentre trial recruited from memory clinics in the UK for a trial of donepezil and aspirin. Likely AD, with or without vascular component. Community-dwelling with a proxy informant. No indications for aspirin nor contraindications	331 people from the AD2000 trial and 123 from Richard et al.’s trial [32], a trial of enhanced vascular care for people with mild AD with vascular lesions on MRI	Patients with probable AD, or with possible AD with positive AD biomarkers, seen between 2009 and 2012 in a neurology clinic. Required at least a 2-year follow-up and genetic testing for ApoE	The Taiwan National Health Insurance Research Database, people with AD identified through diagnosis code or prescription, based on a random extract from the full database
Age	< 60 years, 5%; 60–69, 19%; 70–79, 51%; > 80 years, 25% Median 74 years	74 years AD2000, 76 years Richard	Mean 76 years 38% < 66 years	> 65 years, 93% in the anti-platelet group, 85% in the no anti-platelet group
Cardiovascular risk profile	Excluded those with indications for aspirin, which included hx of myocardial infarction, unstable angina, cerebrovascular accident, or transient ischaemic attack	AD2000: excluded those with indications for aspirin. Richard: not an exclusion. Required white matter lesion/s of vascular origin	All included. Vascular diseases were ascertained from medical documentation and testing done during assessment	Vascular risk factors extracted from claims history and used to create propensity score
Dementia subtype Dementia severity	Alzheimer’s dementia Mild-moderate	Alzheimer’s dementia Mild-moderate	Alzheimer’s dementia Any (average MMSE 22)	Alzheimer’s dementia Any (first recorded dementia event)
Intervention (n, %) Control (n, %)	Aspirin (156, 50%) Avoid aspirin (154, 50%)	Aspirin (156 + 65, 51%) Control (154 + 58, 49%)	Aspirin (73, 46%) No aspirin (87, 54%)	Anti-platelet (including aspirin 100 mg+) prescribed for > 3 months: 824 (with subgroup aspirin = 656) No more than one-off anti-platelet prescription after first dementia event: 824 matched
Length of intervention Background treatment	3 years (but sample size reduces after 1 year) Approximately half were also randomised to take donepezil	Mean time of follow-up: 29 months (AD2000), 22 months (Richard) Donepezil by around 50% (AD2000), 20% (Richard)	Prevalent use of aspirin (length unspecified) Any allowed. around 27% on statins, 44% on AChEI, and 29% on memantine	Prevalent or incident aspirin, for mean 22 months. Follow-up up to 12 years. Mean 4.8 years Any treatment except anti-coagulants allowed
Outcomes extracted (i–iv from methods)	Mortality: ascertained through follow-up and national records (24 months) Cognition: change in MMSE ascertained during follow-up assessments (36 months) Care-home entry: ascertained through follow-up (36 months) Adverse events: multiple side effects elicited by follow-up interviews and investigation of any hospitalisation or death (severe/severe bleeding)	Adverse: intracranial haemorrhage—ascertainment in AD2000 as for that cohort, unspecified for Richard	Cognition: “fast decline” classified through routine administration of MMSE at least 2 years apart. Fast decline if faster than median decline ≥ 4 pts. in 24 months (24 months)	Adverse: intracranial haemorrhage—ascertained through routine reporting in national insurance database

AD, dementia in Alzheimer’s disease; MMSE, mini-mental state exam out of 30; RCT, randomised controlled trial