Supplementary Methods
Detailed Screening protocol at wave 4
For each of the 1644 participants with interview data at Wave 4, assessment data were
screened for signs of decline based on the following criteria:
Criteria for Screen 1
A) previous PATH diagnosis of cognitive disorder (including dementia) at Wave 1, 2 or 3;
OR
B) either:
i) evidence of cognitive impairment: Mini-Mental State Exam (MMSE) ≤ 24 at Wave 4; or
ii) performance on one or more cognitive tests ≤ 6.7th percentile at Wave 4 (Immediate
recall, Delayed recall, SDMT, F words, A words, Boston Naming Test, Simple RT, Choice RT,
Pegboard dominant, Pegboard non-dominant, Pegboard both, Digits Back, Trails B, Stroop
Words, Stroop Colour-Word);
AND either:
iii) subjective decline: scores ≥ 25 on the Memory and Cognition Questionnaire (MAC-Q) (32);
or
iv) evidence of decline: >3 point decline in MMSE score since Wave 3, or
v) evidence of consistent cognitive impairment across time: MMSE ≤ 24 at Waves 3 and 4.
An Informant Interview was conducted with a consenting proxy designated by the
participant, with N=1438 (87%) completing this interview. The majority of
participants meeting criteria for any of the above (N=623) completed a detailed
Informant Interview (N=426, (68%)) with additional questions about medical and
psychiatric history, cognitive and behavioural changes.
All data from the health survey and cognitive testing as well as informant interview
were collated into a spreadsheet casefile for each participant. This casefile (Screen 2)
automatically screened each participant for meeting criteria for any one of the
following diagnoses: DSM 5 Major Neurocognitive Disorder, DSMIV Dementia,
DSM 5 Mild Neurocognitive Disorder, Mild Cognitive Impairment, Age Associated
Cognitive Decline, Age Associated Memory Impairment, DSMIV Amnestic Disorder
not otherwise specified, DSMIV Mild Neurocognitive Disorder, DSMIV Other
cognitive disorder.
General* Criteria for Screen 2
Major criteria for meeting most of the above diagnoses were operationalised as below:
1. Concern of self or informant of significant cognitive decline
MACQ>25 OR IQCODE>3.31 OR history of dementia diagnosis
2. Substantial impairment on at least one domain (relative to Wave 4 normative data)
(cut-offs: < -2SD for dementias, < -1.5SD for mild cognitive disorders)
Complex Attention: Mean z-score for SDMT, Trails A, Choice RT < cut-off.
Executive Function: Mean z-score for DigitsBack, Trails B, Stroop CW, Stroop Int < cut-off.
OR Mean z-score for Zoomap sequence and error, Dice safe choices < cut-off.
Memory: Mean z-score for CVLT immediate, delayed recall, recognition, BVRT-B < cut-off
OR Delayed recall z-score <cutoff AND decline in % of learnt words recalled at delay
between wave 3 to wave 4.
Language: Mean z-score for F words, A words, Boston Naming Test.
Social Cognition: Reading Mind in Eye Test< cut-off AND Informant report of social
behavioural issues.
3. Interference with independence and instrumental activities of daily living.
Self-reported IADL impairment OR Bayer IADLs>3.11 OR Informant reported everyday
cognitive difficulties.
4. Not exclusively during delirium
Cognitive changes since > 6 months, Onset of cognitive changes precede Informant report of
onset of delirium-like symptoms.
5. Not due to another co-existing disorder (e.g., depression, Schizophrenia etc).
PHQ9 < 9 AND No reported history of Schizophrenia or other psychosis.
*General criteria because these represent DSM5 NCD criteria, and slightly different criteria
are used for DSMIV, MCI, AACD, AAMI etc.
Those meeting criteria for one or more diagnoses (N=368) were screened for case file
review by a research neurologist. Diagnoses were made for N= 301 of these cases, of
which N=60 complex cases were selected for Diagnostic Consensus based on the
following criteria:
Criteria for inclusion in Consensus
1. Comorbid depression (moderate to severe)
2. Other comorbid psychiatric conditions.
3. Stroke.
4. DSM 5 Major Neurocognitive Disorder without memory impairment.
Following consensus diagnosis with a clinician specializing in Psychiatry, the final
diagnoses included 71 dementia/major NCD, 196 mild cognitive disorders (MCI/mild
NCD) and 34 other mild or medical related cognitive disorders.
Multistate-modelling
The central structure of the multistate models (MSMs) is the transition intensity
matrix Q, an n x n matrix that defines which transitions can occur.
−(𝑞12 + 𝑞13 )
𝑞12
𝑞13
𝑄= (
𝑞21
−(𝑞21 + 𝑞23 ) 𝑞23 )
0
0
0
Each element of the Q, denoted as qrs, represents the instantaneous risk of
transitioning from state r to state s (i.e., the hazard ratio [HR] from r to s, with a value
of 0 representing a position from which no transitions are possible [i.e., an absorbing
state]). Each row of the Q sums to 0. For a continuous-time homogeneous Markov
process with transition intensity matrix Q, it is possible to compute a transition
probability matrix P over a given period of time. The likelihood of transitioning from
state r to state s is conditional on the state currently occupied, but not on any state
prior and assumes that transition intensities are constant over time [44]. Transition
probabilities can be further modelled as a function of covariates via proportional
hazard models to estimate the HR from state r to state s in the presence of a
covariates. Within a multi-state model, misclassification error can be accounted for
using a hidden Markov model where the observed states are assumed to be
misclassifications of their true, hidden state [44]. Here the Markov process in Q
governs the progression through the hidden states and the observed states are
generated from the hidden states according to a misclassification probability matrix E,
an analogous n x n matrix to Q that determines the probability of a state been
erroneously classified as another state.
Supplementary Figure 1: Flow chart depicting the process of screening participants
for mild cognitive disorders.
3
Risk Score
30
20
2
10
1
0
ANU−ADRI
EV−GRS
Supplementary Figure 2: Distribution of the ANU-ADRI and EV-GRS scores
within the PATH cohort. The variable width of the violin plot indicates the
probability density and the box plot indicates the first, median and third quartile of the
ANU-ADRI scores.
Supplementary Table 1: LOAD risk SNPs used in this study
OR‡
Gene
SNP
Chromosome Alleles* MAF†
rs429358/rs7412
19
APOE 4
2/3/4 0.8/0.14 0.54/3.81
ABCA7
rs3764650
19
T/G
0.11
1.23
BIN1
rs744373
2
A/G
0.31
1.17
CD2AP
rs9296559
6
T/C
0.27
1.11
CD33
rs34813869
19
A/G
0.3
0.89
CLU
rs11136000
8
C/T
0.35
0.88
CR1
rs3818361
1
G/A
0.26
1.17
EPHA1
rs11767557
7
T/C
0.2
0.89
MS4A4A
rs4938933
11
T/C
0.5
0.88
MS4A4E
rs670139
11
G/T
0.34
1.08
MS4A6A
rs610932
11
T/G
0.45
0.90
PICALM
rs3851179
11
C/T
0.41
0.88
HLA-DRB5
rs9271100
6
C/T
0.31
1.11
HLA-DRB1
PTK2B
rs28834970
8
T/C
0.32
1.10
SORL1
rs11218343
11
T/C
0.03
0.77
SLC24A4rs10498633
14
G/T
0.19
0.91
RIN3
DSG2
rs8093731
18
C/T
0.01
0.73
INPP5D
rs35349669
2
C/T
0.44
1.08
MEF2C
rs304132
5
G/A
0.46
0.93
NME8
rs2718058
7
A/G
0.36
0.93
ZCWPW1
rs1476679
7
T/C
0.32
0.91
CELF1
rs7933019
11
G/C
0.34
1.08
FERMT2
rs17125944
14
T/C
0.08
1.14
CASS4
rs7274581
20
T/C
0.11
0.88
*
†
‡
Major/Minor Allele; Minor Allele Frequency: HapMap-CEU; Alzegene reported
OR for minor allele or OR reported in [24]
Supplementary Table 2: Differences between NC and individuals classified as MCI/Dementia or MCI-TB at any wave for the ANU-ADRI,
ANU-ADRI subindices and the EV-GRS.
Score
*
EV-GRS
ADRI*
Alcohol Intake†
Light - Mod
No intake
Age/Sex†
<65 years
65-70 males
65-70 females
Education†
> 11 years
8 - 11 years
< 8 years
Diabetes†
No Diabetes
Diabetes
Depression†
PHQ < 10
PHQ > 10
Traumatic Brain Injury†
MCI/Dementia
CN
Impaired
p
1.6 ± 0.41 1.7 ± 0.43
9.2 ± 5.8
11 ± 6.5
-3 1540 (81)
0 350 (19)
0 1884 (100)
1 2 (0.11)
5 4 (0.21)
0 1469 (78)
3 387 (20)
6 34 (1.8)
148 (79)
40 (21)
187 (99)
0 (0)
1 (0.53)
128 (68)
52 (28)
8 (4.3)
0 1759 (93)
3 131 (6.9)
166 (88)
22 (12)
0 1838 (97)
2 52 (2.8)
177 (94)
11 (5.9)
CN
0.1188 1.6 ± 0.42
0.0003 8.8 ± 5.6
0.38
1074 (82)
235 (18)
0.49
1306 (100)
1 (0.076)
2 (0.15)
0.004
999 (76)
297 (23)
13 (0.99)
0.027
1220 (93)
89 (6.8)
0.025
1283 (98)
26 (2)
0.61
MCI-TB
Impaired
1.6 ± 0.41
10 ± 6.3
p
0.5165
1.068e-08
0.22
614 (80)
155 (20)
0.45
765 (99)
1 (0.13)
3 (0.39)
1.70E-05
598 (78)
142 (18)
29 (3.8)
0.22
705 (92)
64 (8.3)
0.00048
732 (95)
37 (4.8)
0.41
No TBI
0 1793 (95)
TBI
4 97 (5.1)
†
Body Mass Index
BMI < 25
0 685 (36)
BMI 25 - 30
2 870 (46)
BMI > 30
5 335 (18)
†
Smoking
Never Smoker
0 989 (52)
Past Smoker
1 700 (37)
Current Smoker
4 201 (11)
†
Social Engagement
Highest
0 1162 (61)
Medium - Low
1 556 (29)
Low
6 172 (9.1)
†
Physical Activity
High
-3 237 (13)
Medium
-2 615 (33)
Low
0 1038 (55)
†
Cognitive Activity
Medium
-7 743 (39)
High
-6 689 (36)
Low
0 458 (24)
*
Unpaired two tailed t-test; †Fishers Exact test
177 (94)
11 (5.9)
1245 (95)
64 (4.9)
725 (94)
44 (5.7)
467 (36)
603 (46)
239 (18)
285 (37)
343 (45)
141 (18)
697 (53)
476 (36)
136 (10)
400 (52)
289 (38)
80 (10)
834 (64)
364 (28)
111 (8.5)
439 (57)
249 (32)
81 (11)
167 (13)
451 (34)
691 (53)
88 (11)
223 (29)
458 (60)
529 (40)
526 (40)
254 (19)
278 (36)
217 (28)
274 (36)
0.096
67 (36)
76 (40)
45 (24)
0.78
0.34
108 (57)
65 (35)
15 (8)
0.84
0.68
111 (59)
57 (30)
20 (11)
0.01
0.42
18 (9.6)
59 (31)
111 (59)
0.01
0.00072
64 (34)
54 (29)
70 (37)
4.60E-16