Table A.2.

Characteristics of UK Biobank dataset on 5,338 participants, who were excluded from the analysis due to missingness (*6289 participants for the cognition variable reaction time and years of education). There does not seem to be systematic differences between the two datasets, so the exclusions are not likely to have introduced some bias in the results of the analyses. Similar to Table 1 with an additional column summarising the number of missing values for each variable. Summaries calculated without missing values for each variable. The percentage values under each discrete variable in column 1 are (% of participants with risk factor present (level 1) in exclusions dataset (calculated from Table A.2) vs (% of participants with risk factor present in dataset used for the analyses (calculated from Table 1)), note for variables with more levels, vectors of incidences are presented. SD: standard deviation; WHR: waist-to-hip ratio; APOE: apolypoprotein-E; CVR: cerebrovascular risk; WMH: white matter hyperintensity.

Characteristics	Levels (N)	Mean (SD)	Median (range)	Missing
Age (years)	–	63.7 (7.5)	64.4 (46.6; 80.3)	0
Sex	Men (2,496), Women (2,842)	–	–	0
(53% vs 53% women%)
Head size	–	1.3 (0.12)	1.3 (0.9; 1.8)	0
Hypertension risk	0 (1,769), 1 (1,494)	–	–	2,075
(46% vs 47%)
Hypercholesterolemia	0 (4,020), 1 (1,107)	–	–	211
(22% vs 20%)
Diabetes	0 (4,932), 1 (241)	–	–	165
(5% vs 5%)
Smoking (score/pack years)	0 (1,827), 1 (340), 2 (25)	4.7 (11.3)	0 (0; 117.5)	3,146
(83%,16%,1%) vs (83%,16%,1%)
WHR (indicator/continuous)	0 (2,563), 1 (2,362)	0.9 (0.1)	0.9 (0.6; 1.2)	413
(48% vs 47%)
APOE-$\epsilon$ status	0 (3,622), 1 (1,092), 2 (110)	–	–	514
(75%,23%,2%) vs (75%,23%,2%)
WMH load (2 mm3 voxels)	–	593 (732.1)	356 (3; 11,450)	0
Reaction time* (milliseconds)	–	594.3 (109.9)	577 (150; 1,445)	1,505
Years of education	7 (468), 10 (900), 13 (330), 15 (704), 19 (946), 20 (2,752)	16.4 (4.5)	19 (7; 20)	189
(8%,15%,5%,12%,15%,45%) vs (6%,13%,6%,12%,16%,47%)