Lifestyle Risk Factors and Cognitive Outcomes from the Multidomain Dementia Risk Reduction Randomized Controlled Trial, Body Brain Life for Cognitive Decline (BBL-CD)

Abstract

BACKGROUND/OBJECTIVES: To evaluate the efficacy
of a multidomain intervention to reduce lifestyle risk factors
for Alzheimer’s disease (AD) and improve cognition in individuals
with subjective cognitive decline (SCD) or mild cognitive
impairment (MCI).
DESIGN: The study was an 8-week two-arm single-blind
proof-of-concept randomized controlled trial.
SETTING: Community-dwelling individuals living in Canberra,
Australia, and surrounding areas.
PARTICIPANTS: Participants were 119 individuals (intervention
n = 57; control n = 62) experiencing SCD or MCI.
INTERVENTION: The control condition involved four
educational modules covering dementia and lifestyle risk
factors, Mediterranean diet, physical activity, and cognitive
engagement. Participants were instructed to implement this
information into their own lifestyle. The intervention condition
included the same educational modules and additional
active components to assist with the implementation of this
information into participants’ lifestyles: dietitian sessions,
an exercise physiologist session, and online brain training.
MEASUREMENTS: Lifestyle risk factors for AD were
assessed using the Australian National University-Alzheimer’s
Disease Risk Index (ANU-ADRI), and cognition was assessed
using Alzheimer’s Disease Assessment Scale-Cognitive subscale,
Pfeffer Functional Activities Questionnaire, Symbol Digit
Modalities Test (SDMT), Trail Making Test-B, and Category
Fluency.
RESULTS: The primary analysis showed that the intervention
group had a significantly lower ANU-ADRI score (X2 =
10.84; df = 3; P = .013) and a significantly higher cognition
score (X2 = 7.28; df = 2; P = .026) than the control group. A
secondary analysis demonstrated that the changes in lifestyle
were driven by increases in protective lifestyle factors
(X2 = 12.02; df = 3; P = .007), rather than a reduction in
risk factors (X2 = 2.93; df = 3; P = .403), and cognitive
changes were only apparent for the SDMT (X2 = 6.46; df = 2;
P = .040). Results were robust to intention-to-treat analysis
controlling for missing data.
CONCLUSION: Results support the hypothesis that
improvements in lifestyle risk factors for dementia can lead
to improvements in cognition over a short time frame with a
population experiencing cognitive decline. Outcomes from
this trial support the conduct of a larger and longer trial with
this participant group.

Full text

CLINICAL INVESTIGATION
Lifestyle Risk Factors and Cognitive Outcomes from the
Multidomain Dementia Risk Reduction Randomized Controlled
Trial, Body Brain Life for Cognitive Decline (BBL-CD)
Mitchell McMaster, Psych. Hons.,* Sarang Kim, PhD,
†
Linda Clare, ScD,
‡
Susan J. Torres, PhD,
§
Nicolas Cherbuin, PhD,* Catherine DʼEste, PhD,
¶
∥
and Kaarin J. Anstey, PhD**
††
BACKGROUND/OBJECTIVES: To evaluate the efficacy
of a multidomain intervention to reduce lifestyle risk factors
for Alzheimer’s disease (AD) and improve cognition in indi-
viduals with subjective cognitive decline (SCD) or mild cog-
nitive impairment (MCI).
DESIGN: The study was an 8-week two-arm single-blind
proof-of-concept randomized controlled trial.
SETTING: Community-dwelling individuals living in Can-
berra, Australia, and surrounding areas.
PARTICIPANTS: Participants were 119 individuals (inter-
vention n = 57; control n = 62) experiencing SCD or MCI.
INTERVENTION: The control condition involved four
educational modules covering dementia and lifestyle risk
factors, Mediterranean diet, physical activity, and cognitive
engagement. Participants were instructed to implement this
information into their own lifestyle. The intervention condi-
tion included the same educational modules and additional
active components to assist with the implementation of this
information into participants’lifestyles: dietitian sessions,
an exercise physiologist session, and online brain training.
MEASUREMENTS: Lifestyle risk factors for AD were
assessed using the Australian National University-Alzheimer’s
Disease Risk Index (ANU-ADRI), and cognition was assessed
using Alzheimer’s Disease Assessment Scale-Cognitive sub-
scale, Pfeffer Functional Activities Questionnaire, Symbol Digit
Modalities Test (SDMT), Trail Making Test-B, and Category
Fluency.
RESULTS: The primary analysis showed that the interven-
tion group had a significantly lower ANU-ADRI score (χ
2
=
10.84; df =3;P= .013) and a significantly higher cognition
score (χ
2
= 7.28; df =2;P= .026) than the control group. A
secondary analysis demonstrated that the changes in life-
style were driven by increases in protective lifestyle factors
(χ
2
= 12.02; df =3;P= .007), rather than a reduction in
risk factors (χ
2
= 2.93; df =3;P= .403), and cognitive
changes were only apparent for the SDMT (χ
2
= 6.46; df =2;
P= .040). Results were robust to intention-to-treat analysis
controlling for missing data.
CONCLUSION: Results support the hypothesis that
improvements in lifestyle risk factors for dementia can lead
to improvements in cognition over a short time frame with a
population experiencing cognitive decline. Outcomes from
this trial support the conduct of a larger and longer trial with
this participant group. J Am Geriatr Soc 68:2629-
2637, 2020.
Keywords: dementia prevention; lifestyle risk reduction;
subjective cognitive decline; mild cognitive impairment;
nonpharmacological intervention
The number of people with dementia is expected to rise
to 82 million by 2030 and to more than 152 million
by 2050.
1
Research has shown that together lifestyle risk
factors are responsible for between one-third and one-half
of all Alzheimer’s disease (AD) cases.
2,3
It is imperative that
interventions to reduce risk are designed to limit the num-
ber of people developing dementia. This can be achieved
From the *Centre for Research on Ageing, Health and Wellbeing
(CRAHW), The Australian National University, Canberra, Australia;
†
Wicking Dementia Research and Education Centre, University of
Tasmania, Hobart, Australia;
‡
Centre for Research in Ageing and Cognitive
Health (REACH), University of Exeter, Exeter, UK;
§
School of Exercise and
Nutrition Sciences, Institute for Physical Activity and Nutrition, Deakin
University, Geelong, Victoria, Australia;
¶
National Centre for Epidemiology
and Population Health (NCEPH), The Australian National University,
Canberra, Australia;
∥
School of Medicine and Population Health, The
University of Newcastle, Newcastle, Australia; **Neuroscience Research
Australia (NeuRA), Sydney, Australia; and the
††
School of Psychology,
University of New South Wales, Randwick, NSW, Australia.
Address correspondence to Mitchell McMaster, Centre for Research on
Ageing, Health & Wellbeing, The Australian National University, Florey
Building (54), Mills Road, Acton, ACT 2601, Australia. E-mail: mitchell.
mcmaster@anu.edu.au
DOI: 10.1111/jgs.16762
JAGS 68:2629-2637, 2020
© 2020 The American Geriatrics Society 0002-8614/20/$15.00

through primary prevention, focusing on lowering the
dementia risk for cognitively normal individuals, and sec-
ondary prevention aimed at high-risk individuals beginning
to experience subjective cognitive decline (SCD) or mild
cognitive impairment (MCI).
4
SCD and MCI are considered
to be prodromes of dementia, and emerging evidence clearly
demonstrates mild forms of neuropathology related to
dementia in both conditions.
5
Although some evidence is available from large lifestyle
interventions for the primary prevention of dementia and
AD,
6
systematic reviews have noted the need to further
investigate secondary prevention.
7-9
A meta-analysis by
Bhome et al.
10
specifically noted a lack of research evaluat-
ing lifestyle interventions in those with SCD, and previous
randomized controlled trials (RCTs) in this population have
been underpowered to detect the full spectrum of potential
outcomes included.
8
To date, more research has been con-
ducted on individuals with MCI than with SCD. A system-
atic review
11
of lifestyle-focused RCTs in the MCI
population found that all multidomain interventions evalu-
ated were associated with significant improvements in at
least one cognitive domain, suggesting that these interven-
tions were more promising than interventions focusing on
single domains, such as physical activity or cognitive
engagement alone. The review noted heterogeneity in the
cognitive domains showing improvements and in interven-
tion and assessment methods, and it recommended further
well-designed and adequately powered RCTs.
Objectives
This proof-of-concept study adapts elements of previously
successful trials from other participant groups to the SCD
and MCI groups.
12-16
It is hypothesized that given individu-
alized support, people experiencing cognitive decline can
make meaningful lifestyle changes, and in this prodromal
stage of dementia the brain will still retain sufficient neuro-
plasticity to modify the trajectory of the disease. The aims
of the work were to evaluate the efficacy of the Body Brain
Life for Cognitive Decline (BBL-CD) study to reduce the
overall lifestyle risk of AD and other dementias and to pre-
vent further cognitive decline. Here we report on the pri-
mary outcomes of the BBL-CD study: lifestyle risk of AD
and cognition.
METHODS
Design
The full protocol of the study was published.
15
The study
was an 8-week two-arm single-blind RCT of a lifestyle
modification program to reduce dementia risk for people
experiencing cognitive decline. It was conducted in accor-
dance with the Consolidated Standards of Reporting Trials
(CONSORT) statement
17
and the CONSORT statement for
nonpharmacological interventions.
18
The trial was regis-
tered with the Australian and New Zealand Clinical Trial
Registry (ACTRN12617000792325), and ethical clearance
was provided by the ANU Human Research Ethic Commit-
tee (Protocol No. 2016/360). All participants gave written
informed consent to take part.
Participants
Participants were community-dwelling individuals from the
Canberra region, Australia, aged older than 65; owning a
computer with Internet access; having sufficient English
skills; willing to make lifestyle changes to improve health;
and having been diagnosed with MCI by a medical profes-
sional or reporting experiencing SCD. The Jessen criteria
were used for SCD: clinically normal on objective assess-
ment, self/informant-reported cognitive decline, and decline
not better accounted for by a major medical, neurological,
or psychiatric diagnosis.
15
Although potentially limiting in
this age group, a computer with Internet access was essen-
tial for certain intervention components; greater than 50%
of Australians aged 65 and older have access to the Inter-
net, and this figure has been increasing in recent years.
19,20
Exclusion criteria were any major neurological or psy-
chiatric disorder, or other chronic condition that would pre-
vent participation in a lifestyle behavior change program;
and currently participating in any other lifestyle change
interventions. No restrictions were placed on current levels
of or adherence to Mediterranean diet, physical activity, or
cognitive engagement.
Interventions
The active control group completed four online informa-
tional modules to reduce dementia risk. The modules cov-
ered dementia literacy and lifestyle risk, Mediterranean diet,
physical activity, and cognitive engagement. Following each
module, participants were given a week with no education
to allow them to implement the information into their own
lifestyle.
The intervention group completed the same online edu-
cational modules, but in the weeks between undertaking the
modules, the intervention group took part in practical activ-
ities including meeting with a dietitian and exercise physiol-
ogist and completing brain training. These practical
components were designed to assist the participants to
implement the information from the modules more effec-
tively into their lifestyle. Participants had an initial 1-hour
appointment with a dietitian (week 3) and two follow-up
30-minute appointments (weeks 10 and 21) to assist with
adhering to a Mediterranean diet; an initial 1-hour appoint-
ment with an exercise physiologist (week 7) to formulate an
exercise plan, and two follow-up 30-minute appointments
(weeks 10 and 21) to modify as required; and 2 hours
weekly of online brain training (beginning week 5) on the
Brain HQ platform.
21
Further details of the intervention are
provided in the published protocol.
15
However, there was
one deviation from this protocol. The exercise physiologist
was hospitalized on two occasions during the study, and no
suitable replacement could be identified; therefore the two
follow-up appointments were not conducted.
Outcomes
Lifestyle Risk of Alzheimer’s Disease
Lifestyle risk for AD was assessed using the Australian
National University-Alzheimer’s Disease Risk Index (ANU-
ADRI). The ANU-ADRI yields scores for 11 lifestyle risk
2630 MCMASTER ET AL. NOVEMBER 2020-VOL. 68, NO. 11 JAGS

factors (age [moderated by sex], low educational attain-
ment, body mass index, diabetes mellitus type II, depres-
sion, cholesterol, traumatic brain injuries, smoking status,
low social engagement, and exposure to pesticides), four
protective factors (alcohol intake, physical activity, cogni-
tive engagement, and fish intake), or an overall score com-
bining both factors.
22
For the ANU-ADRI, lower scores
indicate lower lifestyle risk for all three measures. Partici-
pants completed the ANU-ADRI via computer at the
research team’soffice at baseline (week 0), immediately fol-
lowing the intervention (week 9), at 3-month follow-up
(week 20), and at 6-month follow-up (week 32).
The largest score component of the ANU-ADRI is age
(0–41 points). To assess the effect of the intervention
adequately, ANU-ADRI scores were calculated at all time
points from the participants’age at baseline. This prevented
any scores from increasing as a result of participants mov-
ing to a higher risk bracket of age, hence obscuring changes
due to lifestyle alterations.
Cognition
The cognitive outcomes for the study were the measures
comprising the Alzheimer’s Disease Assessment Scale-Cog-
nitive Plus (ADAS-Cog Plus)
23
: the ADAS-Cog 11,
24
Pfeffer
Functional Activities Questionnaire (PFAQ),
25
Trail Mak-
ing Test-B (TMT-B),
26
Symbol Digit Modalities Test
(SDMT),
27
and Category Fluency for vegetables.
28
The
Contacted Research Team
(n = 199)
Baseline Testing
(n = 123)
Enrolled
(n = 135)
Not enrolled (n = 64)
Ineligible (n = 21)
Declined (n = 43)
Incomplete Baseline Testing
(n = 4)
Withdrew (n = 12)
Availability (n = 7)
Health (n = 5)
Withdrawals (n = 8)
Lost to follow-up (n = 5)
Availability (n = 2)
Group allocation (n = 1)
Withdrawals (n = 10)
Availability (n = 6)
Health (n = 3)
Group allocation (n = 1)
Withdrawals (n = 0) Withdrawals (n = 4)
Lost to follow-up (n = 2)
AD diagnosis (n = 1)
Availability (n = 1)
Intervention
(n = 49)
Withdrawals (n = 1)
AD diagnosis (n = 1)
Withdrawals (n = 0)
Immediate Follow-up
(Week 9)
Three-month Follow-up
(week 20)
Six-month Follow-up
(week 32)
Randomization
Control
(n = 52)
Control
(n = 62)
Intervention
(n = 57)
Intervention
(n = 49)
Control
(n = 48)
Control
(n = 48)
Intervention
(n = 48)
Figure 1. Participant flowchart for BBL-CD study. Lost to follow-up: These participants could not be contacted/did not respond.
Availability: These participants formally withdrew due to other commitments. Group allocation: These participants formally with-
drew due to the group they were randomized to. AD diagnosis: These participants were diagnosed with Alzheimer’s disease. [Color
figure can be viewed at wileyonlinelibrary.com]
JAGS NOVEMBER 2020-VOL. 68, NO. 11 BBL-CD: A LIFESTYLE AD RISK REDUCTION RCT 2631

ADAS-Cog Plus measures were selected because they are
sensitive to the deficits seen in early stages of cognitive
decline.
23
All cognitive measures were assessed face to face
at baseline (week 0), 3-month follow-up (week 20), and
6-month follow-up (week 32). Further details on the out-
comes can be found in the protocol article.
15
These cognitive measures were combined into a single
composite score by conversion to zscores. These were
calculated based on the baseline mean and standard devia-
tion (SD) of each measure. The zscores were then averaged
across the measures for each participant at each time point.
For the ADAS-Cog, PFAQ, and TMT-B, lower scores indi-
cate better cognitive function; for calculating the composite
score, these were reversed so that increases indicated better
cognitive function for all measures. This composite zscore
as well as the results from the individual cognitive measures
are reported.
Blinding
All testing was carried out by researchers who were blind
to group allocation. Due to the nature of the intervention,
blinding of the participants was not possible. Participants
were asked not to discuss the intervention with any
researchers; if they had any questions, they could discuss
these with the project manager in private.
Statistical Methods
At the final follow-up, a minimum sample size of 36 partici-
pants per arm was required to detect a difference between
groups of .70 SDs in the primary outcome measures.
Accounting for potential attrition of 10% per follow-up
period gave a baseline target sample size of 120 participants.
All participants were randomized to either the intervention
or control group in a 1:1 ratio, within strata defined by sex,
baseline cognition (above or below median ADAS-Cog
11 score), and baseline lifestyle risk of AD (above or below
median ANU-ADRI score) in permuted blocks of eight. The
randomization sequence was generated by an independent
researcher (R.B.) from www.sealedenvelope.com.
Linear mixed models were used to compare outcomes
between groups at each follow-up time. Each model
included group, time point, and the group ×time point
interaction, as well as stratification variables: sex, ANU-
ADRI strata, and ADAS-Cog 11 strata. The likelihood ratio
test (LRT) was used to assess statistical significance of the
Figure 2. Scores for Primary Outcomes: Lifestyle risk and Cognition. Adjusted outcomes for Australian National University-
Alzheimer’s Disease Risk Index (ANU-ADRI) (A) Group ×time interaction: χ
2
=10.84;df =3;P= .012 and a cognitive composite z
score (B) group ×time interaction: χ
2
=7.28;df =2;P= .026. All data points are least square means generated from regression
models after adjusting for strata variables that were not the dependent variable (eg, ANU-ADRI adjusted for sex and cognition strata
and vice versa). Lower ANU-ADRI scores indicate lower lifestyle risk. Higher cognitive composite scores indicate better cognitive
function. Blue dashed lines represent the intervention group, solid red lines represent the control group, and error bars represent 95%
confidence intervals. Between-group significance denoted by *P< .05 and **P<.01.
Table 1. Baseline Characteristics and Primary Outcome
Measures of the Two Groups
Intervention
(n = 57)
Control
(n = 62)
Age, y 72.8 (5.3) 73.3 (5.8)
Female (%) 35 (61.4) 38 (61.3)
Education, y 12.4 (5.3) 14.0 (5.9)
ANU-ADRI total, score
range = −14 to 73
8.3 (10.8) 10.3 (11.6)
Protective factors, score
range = 0 to −14
−9.1 (4.2) −8.9 (4.5)
Risk factors, score range = 0 to
73
17.4 (9.0) 19.2 (10.0)
Cognitive composite zscore .091 (.67) .095 (.56)
ADAS-Cog 11, score range = 0
to 70
7.5 (3.6) 7.0 (3.5)
PFAQ, score range = 0 to 15 .8 (1.6) .8 (1.5)
SDMT, score range = 0 to 110 42.1 (9.8) 41.7 (9.2)
TMT-B, s 97.3 (33.2) 99.3 (41.7)
Category fluency, score
range = 0 to ∞
14.7 (4.2) 14.5 (4.4)
Note: Cognitive composite zscores were created using baseline means and
SDs for all participants, then averaging across these zscores to form a com-
posite. Participants with missing data on one or more cognitive measures
were not included in the zscore average.
Abbreviations: ADAS-Cog, Alzheimer’s Disease Assessment Scale-Cognitive
Subscale; ANU-ADRI, Australian National University-Alzheimer’s Disease
Risk Index; PFAQ, Pfeffer Functional Activities Scale; SDMT, Symbol Digit
Modalities Test; TMT-B, Trail Making Task-B.
2632 MCMASTER ET AL. NOVEMBER 2020-VOL. 68, NO. 11 JAGS

main effects and interaction term.
29
A statistically signifi-
cant interaction term indicated that the between-group dif-
ferences changed over time. Difference in least square mean
outcomes between intervention groups is reported at each
follow-up, with 95% confidence intervals (CIs) and
Pvalues from between-group ttests. A secondary analysis
was undertaken to examine the effect of the intervention on
the two subcomponents of the ANU-ADRI, risk and protec-
tive factors, and each of the five cognitive measures.
The primary analysis was a complete case analysis.
Sensitivity analysis was undertaken as a full intention-to-
treat (ITT) analysis with missing data accounted for using
multivariate imputation by chained equations (MICE).
All preliminary analyses and descriptions of baseline
characteristics were carried out with SPSS v.26.0.
30
Linear
mixed modeling was undertaken in R v.3.6.0
31
using the
lme4,
32
lmerTest,
33
emmeans,
34
and MICE packages,
35
with
graphs creating using ggplot2.
36
RESULTS
Participant Characteristics
Recruitment of participants took place between July 2017
and November 2017; 199 individuals were screened, and
135 were recruited into the study. Of these, 119 participants
completed baseline testing (January 2018) and were ran-
domized into the intervention (n = 57) or control (n = 62)
groups. At the final 6-month follow-up (November 2018),
48 participants remained in each group of the study. The
full flowchart of participants can be found in Figure 1.
At baseline, participants had a mean age of 73.0 years
(SD = 5.5 years) and had 13.3 years (SD = 5.7 years) of
education, and 61% (n = 73) were female. Three (3%) par-
ticipants had a diagnosis of MCI, and all other participants
(n = 116 [97%]) met the criteria for SCD. Baseline charac-
teristics for both groups are shown in Table 1.
Mean differences with 95% CIs and significance levels
for between-group differences for all variables, at all
follow-up time points, for the primary and secondary ana-
lyses can be found in Supplementary Table S1.
Overall, participants were well able to adhere with
most of the intervention requirements. All participants from
both groups who remained in the intervention until the final
follow-up completed the four educational modules. For the
intervention group there was mostly strong adherence to
the active interventions. All participants who remained in
the intervention attended the three dietitian appointments
and the exercise physiologist appointment. However, adher-
ence to the brain training component was lower, with 20%
adherence (10.8 hours) of the specified 54 total hours
(27 weeks ×2 hours/week).
Primary Analyses
Lifestyle Risk of Alzheimer’s Disease (ANU-ADRI)
The LRT analysis showed a significant group ×time point
interaction (χ
2
=10.84;df =3;P= .013). The between-group
difference was not significant at immediate follow-up (T2
intervention = 9.80; control = 11.98; difference = −2.18;
t=−1.28; P= .204), it was significant at the 3-month
follow-up (T3 intervention = 7.27; control = 11.35; differ-
ence = −4.08; t=−2.36; P= .019), and it was no longer sig-
nificant at the 6-month follow-up (T4 intervention = 7.48;
control = 10.37; difference = −2.88; t=−1.66; P= .098)
(Figure 2A). When looking at within-group changes over the
course of the intervention, the control group showed only a
minor reduction in ANU-ADRI scores (−0.54) compared
with the larger reduction seen in the intervention group
(−2.46); reductions of 2.0 or more ANU-ADRI points are
considered clinically meaningful.
Cognitive Composite zScore
The LRT analyses found a significant group ×time point
interaction (χ
2
= 7.28; df =2;P= .026). For the between-
group differences, the intervention group had significantly
higher cognition scores at both follow-up periods (T3 inter-
vention = .159; control = −.117; difference = .276; t= 2.62;
P= .010; T4 intervention = .231; control = −.014; differ-
ence = .245; t= 2.33; P= .021) (Figure 2B).
Figure 3. Australian National University-Alzheimer’s Disease Risk Index (ANU-ADRI) risk and protective factors. Adjusted out-
comes for risk factors (A) group ×time point interaction: χ
2
= 2.93; df =3;P= .403 and protective factors (B) group ×time point
interaction: χ
2
= 12.02; df =3;P= .007. All data points are least square means generated from regression models after adjusting
for strata variables (sex and cognition). Lower scores indicate lower lifestyle risk. Blue dashed lines represent the intervention
group, solid red lines represent the control group, and error bars represent 95% confidence intervals. Between-group significance is
denoted by *P< .05 and ***P< .001. [Color figure can be viewed at wileyonlinelibrary.com]
JAGS NOVEMBER 2020-VOL. 68, NO. 11 BBL-CD: A LIFESTYLE AD RISK REDUCTION RCT 2633

Secondary Analyses: Drivers of Significant Change
Lifestyle Risk and Protective Factor for Alzheimer’s
Disease (ANU-ADRI)
The outcomes for the ANU-ADRI risk and protective fac-
tors are shown in Figure 3.
Lifestyle Risk Factors for Alzheimer’s Disease
The LRT analysis found no significant effects (χ
2
= 2.93;
df =3;P= .403). Although both groups experienced a
within-group reduction of risk factor scores, neither of these
was clinically meaningful or statistically significant (T2
intervention = 19.3; control = 19.4; difference = −.092;
t=.07;P= .945; T3 intervention = 18.5; control = 19.5;
difference = .962; t= .72; P= .472; T4 intervention = 18.6;
control = 18.9; difference = .339; t=.28;P= .783).
Lifestyle Protective Factors for Alzheimer’s Disease
The LRT analysis showed a statistically significant group ×
time point interaction (χ
2
= 12.02; df =3;P= .007). The
intervention group’s protective scores were significantly
lower than the control scores at all follow-up periods (T2
intervention = −9.38; control = −7.77; difference = −1.62;
Figure 4. Outcomes for Measures of Cognition. Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-Cog 11) (A)
Group ×time point interaction: χ
2
= .73; df =2;P= .696; Pfeffer Functional Activities Questionnaire (PFAQ) (B) χ
2
= .44; df =2;
P= .802; Trail Making Test-B (TMT-B) (C) χ
2
= .94, df =2;P= .625; Symbol Digit Modalities Test (SDMT) (D) χ
2
= 6.459; df =
2; P= .040; Category Fluency (E) χ
2
= 2.04; df =2;P= .360. All data points are least square means generated from regression
models after adjusting for strata variables (sex and Australian National University-Alzheimer’s Disease Risk Index [ANU-ADRI]).
Lower scores for ADAS-Cog 11, PFAQ, and TMT-B and higher scores for SDMT and Category Fluency indicate better cognitive
function. Blue dashed lines represent the intervention group, solid red lines represent the control group, and error bars represent
95% confidence intervals. [Color figure can be viewed at wileyonlinelibrary.com]
2634 MCMASTER ET AL. NOVEMBER 2020-VOL. 68, NO. 11 JAGS

t= 2.10; P= .037; T3 intervention = −11.21; con-
trol = −8.54; difference = 2.67; t= 3.40; P< .001; T4 inter-
vention = −11.05; control = −9.05; difference = 2.00;
t= 2.51; P= .012).
Cognitive Measures
The outcomes for the individual cognitive measures that
made up the cognitive composite zscore are shown in
Figure 4. The measures were ADAS-Cog (Figure 4A), PFAQ
(Figure 4B), TMT-B (Figure 4C), SDMT (Figure 4D), and
Category Fluency (Figure 4E).
When all cognitive measures were analyzed separately
using the LRT analysis, only SDMT showed a significant
group ×time point interaction (χ
2
= 6.46; df =2;P= .040);
however, neither of the between-group differences was sig-
nificant at follow-up (T3 intervention = 42.4; control = 39.4;
difference = −2.96; t=−1.77; P= .078; T4 interven-
tion = 43.6; control = 40.6; difference = −3.29;
t=−1.96; P= .052).
Sensitivity Analysis: Intention-to-Treat Analysis
In ITT, which used complete cases following missing data
imputation, outcomes were highly consistent with the pri-
mary and secondary analyses. All between-group differ-
ences at specific time points were retained.
A further analysis was undertaken to examine the impact
that attrition may have had on outcomes. Supplementary
Table S2 shows that only one variable, protective lifestyle fac-
tors, showed a significant difference for those who withdrew
and those who remained in the study. Although there was a
significant difference, when this variable was further separated
by intervention group, it showed the significant difference was
only in the control group (withdrew = −6.19; remained = −9.80;
difference = 3.62; t= 2.91; P= .005), not the intervention
group (withdrew = −8.64; remained = −9.17; difference = .54;
t= .38; P= .704).
DISCUSSION
BBL-CD was a proof-of-concept RCT that adapted a suc-
cessful primary prevention study
12
to the cognitive decline
group and included new components adapted from previ-
ously successful interventions conducted in other partici-
pant groups.
12-16
The main findings from this study were
that a multidomain lifestyle intervention was able to
decrease exposure to lifestyle risk factors for AD signifi-
cantly, and improve cognition in a group experiencing cog-
nitive decline significantly, relative to a control group. The
results lend support to the hypothesis that secondary pre-
vention interventions may be able to modify the course of
disease progression.
Adherence
The intervention mostly achieved strong adherence. The
lowest levels of adherence were for the brain training com-
ponent. The lower levels of adherence may have been due
to such a large dose of brain training (54 hours) over such
a long period (27 weeks). A meta-analysis of brain training
in the MCI population showed the average dose across the
studies included was 34.2 hours over an average of
14 weeks (according to the study protocols); however, the
meta-analysis provided no information on actual adherence
to the interventions.
37
The Advanced Cognitive Training
for Independent and Vital Elderly (ACTIVE) trial, which
administered 10 to 12.5 hours of speed of processing train-
ing over 5 to 6 weeks, resulted in lower rates of dementia in
a cognitively normal sample of participants aged 65 and
older at 10 years postintervention.
38
This dose was compa-
rable with the actual dose achieved in BBL-CD, albeit over
a longer time frame. Unfortunately, in BBL-CD no qualita-
tive data were collected to determine what the barriers to
higher levels of adherence may have been.
Lifestyle Risk of Alzheimer’s Disease
By the 3-month follow-up, the intervention group showed a
significantly lower ANU-ADRI than the control group, but
this difference was not retained at the final follow-up. The
decline in overall ANU-ADRI score for the intervention
group was 2.7 (T3) and 2.5 (T4) points; a 2-point change
in ANU-ADRI is considered to be clinically meaningful.
39
This demonstrates that clinically relevant lifestyle changes
are feasible over the short term in participants experiencing
cognitive decline. The significant effects in overall ANU-
ADRI scores were driven by higher levels of protective fac-
tors, rather than lower levels of risk factors. Similar effects
were seen in past multidomain lifestyle interventions.
12,40
The reduction in ANU-ADRI score through increased pro-
tective factors was 2.5 (T3) and 2.3 (T4). In real terms these
changes are similar to the amount of AD risk conferred
between low (0 points) and moderate levels (−2 points)
of exercise or the presence of diabetes mellitus type II (+3
points).
39
When missing data were controlled for in the
ITT, all the between-group differences found in previous
analyses were still present, showing that these are robust
findings. The only scores that may have been affected by
participants withdrawing were protective factors for the
control group. Because the control participants who with-
drew had higher protective scores (ie, less protective), this
may have artificially lowered the control group scores.
Given that the only significant effects were for the interven-
tion group, if anything this artificial reduction may have
reduced the magnitude of difference between the control
and intervention groups. This has minimal impact on the
main findings of this study.
Cognition
At the end of the study, the intervention group had a signifi-
cantly higher cognitive composite score, and a significant
group ×time point interaction effect was observed for both
follow-up periods. When this was investigated further to
determine the specific measures underlying the significant
effects, only SDMT showed a significant group ×time point
interaction; however, there were no between-group differ-
ences at specific time points. The most likely explanation
for this is a weak but consistent positive effect across mea-
sures. Greater statistical power through a larger sample size
would be required to assess actual effects on individual cog-
nitive measures. There appeared to be improvements over
time for all cognitive measures in the intervention group,
JAGS NOVEMBER 2020-VOL. 68, NO. 11 BBL-CD: A LIFESTYLE AD RISK REDUCTION RCT 2635

but only the ADAS-Cog and Category Fluency measures
showed improvement for the control group.
In the limited number of multidomain SCD interven-
tions that have taken place, significant effects have been
found for executive function (color-word Stroop task)
40
and verbal fluency (letter fluency)
41
but not in cognitive
composite scores. MCI lifestyle interventions show a very
mixed pattern of outcomes with positive and null effects
across executive function, memory, and global cognition.
11
Other multidomain lifestyle interventions in different partic-
ipant groups have found significant differences in cognitive
composite scores of a similar magnitude (eg, BBL-CD inter-
vention group, z= .25, vs Finnish Geriatric Intervention
Study to Prevent Cognitive Impairment and Disability [FIN-
GER] trial intervention group, z= .20).
6
Importantly, in the
present study, the ITT analysis produced results that were
highly consistent with the primary and secondary analyses,
which is suggestive that significant effects were not statisti-
cal artifacts due to participant attrition.
Taken together, the intervention group experienced a
statistically significant and clinically meaningful lower life-
style risk and significantly higher cognition, relative to the
control group. No significant differences in lifestyle risk or
cognition were found for the control group. These results
support the hypothesis that individuals in the early stages of
cognitive decline retain sufficient neuroplasticity to achieve
cognitive improvements in the short term. The ultimate goal
of research in this participant group is to demonstrate long-
term improvements in lifestyle risk, cognition, and ultimately
slower rates of cognitive decline and lower rates of conver-
sion to dementia. The outcomes achieved in this study war-
rant the conduct of a larger and longer study to test the
longer term sustainability of improvements in lifestyle and
cognition.
Limitations
The main limitation of this study was the limited follow-up
time of 6 months. Meta-analyses of nonpharmacological
interventions for SCD recommend a minimum follow-up
duration of a year or longer.
8
A second limitation was that
the study overestimated the magnitude of change that
would be observed (.70 SD); hence there may have been
insufficient power to detect small differences.
15
Both limita-
tions can be overcome in future studies.
Implications for Future Research
The pattern of results for lifestyle outcomes illustrate a few
potential areas for refinements for this intervention, as well
as other studies in this area. First, lifestyle interventions
commonly report improvement in protective factors, but
significant reductions in risk factors is an area where clear
improvements are possible. For this reason, it would be
beneficial for future research to look at the outcomes of
protective and risk factors separately.
Second, a plateauing of improvement between time
points 3 and 4 for ANU-ADRI (a period where no further
intervention was being implemented) is suggestive of the
need for “booster sessions”to maximize and sustain life-
style improvements.
In terms of cognition, future interventions in this area
do need to account for small effect sizes of cognitive out-
comes with adequate sample sizes, so as not to negatively
bias evidence in this developing area. Given the heterogene-
ity of cognitive domains that can be improved through non-
pharmacological interventions, future research should
choose a battery of cognitive tests to cover all potential
domains and also standardize and combine outcomes to
detect any subtle but consistent effects across measures.
More emphasis on measures of everyday function would
also show whether interventions are having an immediate
effect on real-world outcomes.
Given there is some heterogeneity in the SCD and MCI
groups in terms of stability, progression, and remission of
deficits, further characterization of participants through
genotyping, neuroimaging, and other biomarkers would be
beneficial for any future research.
In conclusion, the present proof-of-concept study
adds evidence to the argument that modifying the life-
style risk of those experiencing cognitive decline can
result in improvements in cognition. The results
obtained are supportive of a larger, longer trial to inves-
tigate the possibilities of sustained improvements in life-
style and cognition, clearly demonstrate cognitive
domains showing improvements, and long-term follow-
up with participants to track cognitive decline and
development of AD and other forms of dementia for a
number of years postintervention.
ACKNOWLEDGMENTS
We thank Richard Burns (R.B.) for his advice and assis-
tance with the randomization of the sample and Ranmalee
Eramudugolla for her advice on clinical diagnoses and
selection criteria. Mitchell McMaster’s PhD scholarship was
supported by Dementia Australia Research Foundation,
the Australian National University, and Dementia Collabo-
rative Research Centre. Additional research funding was
provided by the NHMRC Centre for Research Excellence
in Cognitive Health, the Australian National University;
Neuroscience Research Australia, University of New South
Wales; the 2017 Royal Commonwealth Society Phyllis
Montgomery Award; NHMRC Fellowship Funds (Kaarin
J. Anstey: APP1102694); and the Dementia Collaborative
Research Centre (funds for precursor study).
Conflict of Interest: The authors have declared no conflicts
of interest for this article.
Author Contributions: Study concept and design:
McMaster, Kim, Clare, Torres, D’Este, and Anstey. Acqui-
sition of data: McMaster. Analysis and interpretation of
data: McMaster, Cherbuin, and D’Este. Drafting of the
manuscript: McMaster. Critical revision of the manuscript
for important intellectual content: All authors.
Sponsor’s Role: None of the funding bodies played any
part in the study design, data collection, analysis, interpreta-
tion, writing, or decision to submit this article for publication.
REFERENCES
1. World Health Organization. Risk Reduction of Cognitive Decline and
Dementia: WHO Guidelines. Geneva, Switzerland: World Health Organiza-
tion; 2019.
2636 MCMASTER ET AL. NOVEMBER 2020-VOL. 68, NO. 11 JAGS

2. Livingstone G, Sommerlad A, Ortega V, et al. Dementia prevention, interven-
tion, and care. Lancet. 2017;390:2673-2734.
3. Barnes DE, Yaffe K. The projected effect of risk factor reduction on
Alzheimer’s disease prevalence. Lancet Neurol. 2011;10(9):819-828.
4. Fratiglioni L, Winblad B, von Strauss E. Prevention of Alzheimerʼs disease
and dementia. Major findings from the Kungsholmen Project. Physiol Behav.
2007;92(1):98-104.
5. Shokouhi S, Albert K. Subjective cognitive decline and biomarkers of preclin-
ical Alzheimerʼs disease. Curr Behav Neurosci Rep. 2019;6(4):219-226.
6. Ngandu T, Lehtisalo J, Solomon A, et al. A 2 year multidomain intervention
of diet, exercise, cognitive training, and vascular risk monitoring versus con-
trol to prevent cognitive decline in at-risk elderly people (FINGER): a
randomised controlled trial. Lancet. 2015;385(9984):2255-2263.
7. Horr T, Messinger-Rapport B, Pillai JA. Systematic review of strengths and
limitations of randomized controlled trials for non-pharmacological interven-
tions in mild cognitive impairment: focus on Alzheimerʼs disease. J Nutr
Health Aging. 2015;19(2):141-153.
8. Smart CM, Karr JE, Areshenkoff CN, et al. Non-pharmacologic interven-
tions for older adults with subjective cognitive decline: systematic review,
meta-analysis, and preliminary recommendations. Neuropsychol Rev. 2017;
27(3):245-257.
9. Williams JW, Plassman BL, Burke J, Holsinger T, Benjamin S. Prevention
Alzheimerʼs Disease and Cognitive Decline. Rockville, MD: Agency for
Healthcare Research and Quality; 2010.
10. Bhome R, Berry AJ, Huntley JD, Howard RJ. Interventions for subjective
cognitive decline: systematic review and meta-analysis. BMJ Open. 2018;8
(7):e021610.
11. Huckans M, Hutson L, Twamley E, Jak A, Kaye J, Storzbach D. Efficacy of
cognitive rehabilitation therapies for mild cognitive impairment (MCI) in
older adults: working toward a theoretical model and evidence-based inter-
ventions. Neuropsychol Rev. 2013;23(1):63-80.
12. Anstey KJ, Bahar-Fuchs A, Herath P, et al. Body brain life: a randomized con-
trolled trial of an online dementia risk reduction intervention in middle-aged
adults at risk of Alzheimer’s disease. Alzheimers Dement (N Y). 2015;1(1):72-80.
13. Estruch R, Ros E, Salas-Salvadó J, et al. Primary prevention of cardiovascular
disease with a Mediterranean diet. N Engl J Med. 2013;368(14):1279-1290.
14. Kim S, McMaster M, Torres S, et al. Protocol for a pragmatic randomised
controlled trial of body brain life—general practice and a lifestyle modifica-
tion programme to decrease dementia risk exposure in a primary care set-
ting. BMJ Open. 2018;8(3):e019329.
15. McMaster M, Kim S, Clare L, Torres SJ, DʼEste C, Anstey KJ. Body, brain,
life for cognitive decline (BBL-CD): protocol for a multidomain dementia
risk reduction randomized controlled trial for subjective cognitive decline
and mild cognitive impairment. Clin Interv Aging. 2018;13:2397-2406.
16. Rebok GW, Ball K, Guey LT, et al. Ten-year effects of the ACTIVE cognitive
training trial on cognition and everyday functioning in older adults. J Am
Geriatr Soc. 2014;62(1):16-24.
17. Schulz KF, Altman DG, Moher D. CONSORT 2010 statement: updated
guidelines for reporting parallel group randomised trials. BMC Med. 2010;8
(18):1-9.
18. Boutron I, Moher D, Altman DG, Schulz KF, Ravaud P. Methods and pro-
cesses of the CONSORT group: example of an extension for trials assessing
nonpharmacologic treatments. Ann Intern Med. 2008;148(4):W-60-W-66.
19. Australian Bureau of Statistics. Household Use of Information Technology,
Australia, 2016–17 (#8146.0). https://www.abs.gov.au/ausstats/abs@.nsf/mf/
8146.0. Published 2017. Accessed June 11, 2020.
20. McCallum, J., Rees, K. & Maccora, J. (2017) Bridging the Senior Digital
Divide. Brisbane, Australia: National Seniors Australia. https://
nationalseniors.com.au/uploads/12172826PAR_BridgingTheDigitalDivide_
FNWeb.pdf. Accessed June 11, 2020.
21. Posit Sceince. BrainHQ. http://www.brainhq.com. Accessed May 4, 2020.
22. Anstey KJ, Cherbuin N, Herath PM. Development of a new method for
assessing global risk of Alzheimerʼs disease for use in population health
approaches to prevention. Prev Sci. 2013;14(4):411-421.
23. Skinner J, Carvalho JO, Potter GG, et al. The Alzheimerʼs Disease Assessment
Scale-Cognitive-Plus (ADAS-Cog-Plus): an expansion of the ADAS-Cog to
improve responsiveness in MCI. Brain Imaging Behav. 2012;6(4):489-501.
24. Rosen WG, Mohs RC, Davis KL. A new rating scale for Alzheimerʼs disease.
Am J Psychiatry. 1984;141(11):1356-1364.
25. Pfeffer R, Kurosaki T, Harrah C Jr, Chance J, Filos S. Measurement of func-
tional activities in older adults in the community. J Gerontol. 1982;37(3):
323-329.
26. Reitan RM. Trail Making Test: Manual for Administration and Scoring.
Reitan Neuropsychology Laboratory; 1992.
27. Smith A. Symbol digit modalities test. Los Angeles, CA: Western Psychologi-
cal Services; 1982.
28. Lucas JA, Ivnik RJ, Smith GE, et al. Mayoʼs older Americans normative stud-
ies: category fluency norms. J Clin Exp Neuropsychol. 1998;20(2):194-200.
29. Winter B. Linear models and linear mixed effects models in R with lingui-
stic applications [arXiv:13085499]. https://arxivorg/ftp/arxiv/papers/1308/
13085499pdf.
30. IBM Corp. IBM SPSS Statistics for Windows, Version 26.0. Armonk, NY:
IBM Corp.; 2019.
31. R Core Team. R: A Language and Environment for Statistical Computing.
Vienna, Austria: R Foundation for Statistical Computing; 2017.
32. Bates D, Maechler M, Bolker B, Walker S. Fitting linear mixed-effects
models using lme4. J Stat Softw. 2015;67(1):1-48.
33. Kuznetsova A, Brockhoff PB, Christensen RHB. lmerTest package: tests in
linear mixed effects models. J Stat Softw. 2017;82(13):1-26.
34. Estimated Marginal Means, “emmeans”R package [computer program].
https://mran.microsoft.com/snapshot/2017-12-11/web/packages/emmeans/
emmeans.pdf. Accessed April 6, 2020.
35. van Buuren S, Groothuis-Oudshoorn K. Mice: multivariate imputation by
chained equations in R. J Stat Softw. 2011;45(3):1-67.
36. Wickham H. ggplot2: Elegant Graphics for Data Analysis. New York, NY:
Springer; 2016.
37. Hill NT, Mowszowski L, Naismith SL, Chadwick VL, Valenzuela M,
Lampit A. Computerized cognitive training in older adults with mild cogni-
tive impairment or dementia: a systematic review and meta-analysis.
Am J Psychiatry. 2017;174(4):329-340.
38. Edwards JD, Xu H, Clark DO, Guey LT, Ross LA, Unverzagt FW. Speed of
processing training results in lower risk of dementia. Alzheimers Dement
(N Y). 2017;3(4):603-611.
39. Anstey KJ, Cherbuin N, Herath PM, et al. A self-report risk index to predict
occurrence of dementia in three independent cohorts of older adults: the
ANU-ADRI. PLoS One. 2014;9(1):e86141.
40. Park J, Jeon S, Kim S, et al. A multidomain intervention for modifying lifestyle
habits reduces the dementia risk in community-dwelling older adults: a single-
blinded randomized controlled pilot study. J Alzheimer Dis. 2019;70(1):51-60.
41. Small GW, Silverman DH, Siddarth P, et al. Effects of a 14-day healthy lon-
gevity lifestyle program on cognition and brain function. Am J Geriatr Psy-
chiatry. 2006;14(6):538-545.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the
online version of this article.
Supplementary Table S1: Least square differences in
means adjusted for strata variables that were not the depen-
dent variable (eg, ANU-ADRI adjusted for sex and cognition
strata and vice versa). The Pvalues are for ttests for between-
group differences. Lower scores for ANU-ADRI, protective
factors, and risk factors all indicate lower levels of lifestyle
risk. Lower scores for ADAS-Cog 11, PFAQ, and TMT-B,
and higher scores for SDMT and Category Fluency indicate
better cognitive function. Values shown in bold are significant
at P< .05. ANU-ADRI, Australian National University-
Alzheimer’s Disease Risk Index; ADAS-Cog, Alzheimer’sDis-
ease Assessment Scale- Cognitive Subscale; PFAQ, Pfeffer
Functional Assessment Questionnaire; TMT-B, Trail Making
Test-B; SDMT, Symbol Digit Modality Test.
Supplementary Table S2: The Pvalues are for ttests for
between-group differences. Lower scores for ANU-ADRI, pro-
tective factors, and risk factors all indicate lower levels of life-
style risk. Lower scores for ADAS-Cog 11, PFAQ, and
TMT-B, and higher scores for SDMT and Category Fluency
indicate better cognitive function. Values shown in bold are
significant at P< .05. ANU-ADRI, Australian National Uni-
versity-Alzheimer’s Disease Risk Index; ADAS-Cog,
Alzheimer’s Disease Assessment Scale-Cognitive Subscale;
PFAQ, Pfeffer Functional Assessment Questionnaire; TMT-B,
Trail Making Test-B; SDMT, Symbol Digit Modalities Test.
JAGS NOVEMBER 2020-VOL. 68, NO. 11 BBL-CD: A LIFESTYLE AD RISK REDUCTION RCT 2637