Atividades Mentais e Físicas na Demência

Prévia do material em texto

Mental and Physical Activities Delay
Cognitive Decline in Older Persons
with Dementia
Sheung-Tak Cheng, Ph.D., Pizza K. Chow, M.Sc., You-Qiang Song, Ph.D.,
Edwin C.S. Yu, M.H., F.H.K.A.M. (Psych.), Alfred C.M. Chan, Ph.D.,
Tatia M.C. Lee, Ph.D., John H.M. Lam, M.Sc.
Received Decem
Aging, Hong Ko
University of Ho
Send correspond
Education, 10 Lo
Supplement
ajgponline.org).
� 2014 Ame
http://dx.d
Am J Geriatr Ps
Objectives: To examine the effects of cognitive stimulation (mahjong) and phys-
ical exercise (tai chi [TC]) on cognitive performance in persons with dementia.
Design: Cluster-randomized open-label controlled design. Setting: Nursing homes.
Participants: Onehundred ten residents,most ofwhomwere cholinesterase-inhibitor
naive. Inclusion criteria were Mini-Mental State Examination (MMSE) ¼ 10e24 and
suffering from at least very mild dementia (Clinical Dementia Rating � 0.5). Exclu-
sion criteria were being bedbound, audio/visual impairment, regular activity
participation before study, or contraindications for physical or group activities.
Interventions: Homes were randomized into three conditions (mahjong, TC, and
simple handicrafts [control]). Activities were conducted three times weekly for
12 weeks. Measurements: Primary outcome was MMSE. Secondary outcomes were
immediate/delayed recall, categorical fluency, and digit span. Various biological
risk factors, including apolipoprotein E ε4 allele, were included as covariates.
Measures were collected at 0 (baseline), 3 (posttreatment), 6, and 9 months.
Results: Intent-to-treat analyses were performed using mixed-effects regression.
Mahjong’s effect varied by time for MMSE, delayed recall, and forward digit span. TC
had similar effects but not for delayed recall. The typical pattern was that control
participants deteriorated while mahjong and TC participants maintained their
abilities over time, leading to enlarged treatment effects as time progressed. By
9 months, mahjong and TC differed from control by 4.5 points (95% confidence
interval: 2.0e6.9; d ¼ 0.48) and 3.7 points (95% confidence interval: 1.4e6.0;
d ¼ 0.40), respectively, on MMSE. No treatment effects were observed for immediate
recall and backward digit span. Conclusions: Mahjong and TC can preserve
ber 2, 2011; accepted June 27, 2012. From the Department of Psychological Studies and Center for Psychosocial Health and
ng Institute of Education (STC, PKC, JHML), Hong Kong; Departments of Biochemistry (YQS) and Psychology (TMCL),
ng Kong, Hong Kong; Kwai Chung Hospital (ECSY), Hong Kong; and Lingnan University (ACMC), Hong Kong.
ence and reprint requests to Sheung-Tak Cheng, Ph.D., Department of Psychological Studies, Hong Kong Institute of
Ping Road, Tai Po, N.T., Hong Kong. e-mail: takcheng@ied.edu.hk
al digital content is available for this article in the HTML and PDF versions of this article on the journal’s Web site (www.
rican Association for Geriatric Psychiatry
oi.org/10.1016/j.jagp.2013.01.060
ychiatry 22:1, January 2014 63
mailto:takcheng@ied.edu.hk
http://www.ajgponline.org
http://www.ajgponline.org
http://dx.doi.org/10.1016/j.jagp.2013.01.060
64
Leisure Activities and Cognition
functioning or delay decline in certain cognitive domains, even in those with
significant cognitive impairment. (Am J Geriatr Psychiatry 2014; 22:63e74)
Key Words: Cognitive decline, dementia, leisure activities, cluster-randomized
controlled trial
ongitudinal studies show that leisure activities,
Lespecially mental and physical activities, miti-
gate cognitive decline. Although both physical and
mentally stimulating activities are associated with
cognitive performance, studies with up to 5 years’
follow-up consistently show that among individuals
who were cognitively intact at baseline, only mental
activity independently predicts cognitive decline or
incident dementia.1e3 It is important to note that
those who are physically active also tend to be
mentally active, and so, the effect of one activity may
be masked by the other in survey studies. A recent
randomized controlled trial of cognitive training
(learning to use the computer), aerobic exercise, and
a nil-treatment control did not reveal differences in
the cognitive effects of cognitive training and exercise
after the 6-month treatment,4 although the long-term
effects were not tested. The therapeutic effects of
these activities may be mediated through neuro-
plasticity,5,6 which may attenuate the effects of
existing brain pathology, like white matter lesions, on
cognitive performance.7
Despite encouraging results in healthy older adults,
training programs for those with already significant
cognitive decline have produced different results.
Whereas physical activity has been found to be
moderately effective for people with cognitive im-
pairment or dementia,8 effects of cognitive training on
such people have been disappointing.9 Nevertheless,
cognitive training should not be confused with
cognitive leisure activities. Cognitive training usually
involves tasks tailor-made to tap specific functions,
with limited generalizability to untrained domains.10
These tasks may be unfamiliar and uninteresting to
the individual, and participation is difficult for those
with low cognitive resources. In contrast, leisure
activities are inherently pleasurable andmore likely to
draw interest and compliance. A recent study showed
that active leisure pursuits in the cognitive (e.g.,
reading and solving crossword puzzles), but not
physical, domain in the initial year or so after the
diagnosis of Alzheimer disease was associated with
a slower cognitive decline, but activities in later years
after diagnosis had no effect.11 However, as with
correlational studies, this study could not demon-
strate causality and the fact that progressive dementia
reduced activity participation in later years11 made
interpretation of the findings difficult. Unfortunately,
no experimental study has been conducted on cogni-
tive leisure activities. However, unlike cognitive
training, due to the holistic nature of leisure activities,
it is difficult to specify exactly which cognitive
domains are being trained, and therefore, measures of
global cognitive function would be more appropriate
outcomes than specific cognitive abilities.
This study tested the effects of two leisure activi-
ties, mahjong (a game with 136e152 tiles that is
similar to playing cards) and tai chi (TC), against
a control in persons with very mild to moderate
dementia. An earlier uncontrolled study suggested
the efficacy of mahjong in improving the cognitive
functions of nursing home residents with dementia,12
but a randomized controlled trial is necessary for
conclusions about causal effects to be made. Moreover,
the cognitive effects of TC have never been investigated
in individuals with this level of impairment, although
a recent randomized controlled trial showed that it
helped maintain Clinical Dementia Rating sum-of-box
scores up to 1 year over a toning and stretching exer-
cise group in able-bodied older adults with amnestic
mild cognitive impairment or very mild dementia
only.13,14 Thus, thiswas thefirst randomized controlled
trial investigating the effects of these two activities
simultaneously in persons with dementia. Mahjong
and TC are popular mental and physical activities,
respectively, in Chinese/Asian societies. Most, if not
all, persons are well-exposed to these activities,
although the degrees to which they master the skills
vary. Due to space limitations, shortage of staff, and
most importantly, management philosophies that
favor maintenance and order but not therapeutic or
quality-of-life issues,15e17 activity levels of residents in
Hong Kong’s nursing homes are typically very low,18
making it possible to introduce activity programs and
Am J Geriatr Psychiatry 22:1, January 2014
Cheng et al.
study their effects. We hypothesized that mahjong and
TC would lead to enhanced cognitive performance
relative to simple handicraft as control.
Intervention effects were assessedafter partialing
out the effects of a number of potential risk factors,
including apolipoprotein E ε4 allele (APOE4; a genetic
predisposition for Alzheimer disease), lipids, blood
glucose, blood pressure (BP), peak expiratory flow
(PEF) rate, depression, chronic illnesses, and demo-
graphic factors.19e25
METHODS
Study Design, Setting, and Participants
A cluster-randomized open-label controlled trial
was conducted. One hundred ten residents recruited
from nine nursing homes were randomized by home
into three experimental conditions—mahjong, TC,
and simple handicrafts (control), with three homes in
each condition (Figure 1). Inclusion criteria were
Mini-Mental State Examination (MMSE)26 10 or
greater and 24 and less and suffering from at least
very mild dementia as indicated by Clinical
Dementia Rating 0.5 or more.27 Exclusion criteria
were being bedbound, audio/visual impairment,
regular participation in any of the three activities in
previous 3 months, or contraindication to participate
in exercise or group activities (e.g., arrhythmia,
moderate to severe parkinsonism, upper limb pa-
ralysis, agitation, and negativism). A cluster design
was deemed necessary to avoid treatment contami-
nation within homes. An open-label design was
inevitable because activities could not be masked and
it was not possible to prevent residents from talking
to interviewers about the activities. With four
repeated measurements (see below) and time-level
intraclass correlation ¼ 0.05, three clusters per treat-
ment condition and n ¼ 10 per cluster were sufficient
to detect a small treatment by time interaction effect
(Cohen’s d ¼ 0.30) at a ¼ 0.05 and power ¼ 0.80.28
Informed consent was obtained from family mem-
bers, with assent from the participants. Ethics approval
was obtained from the Ethics Subcommittee of the Re-
search Committee, City University of Hong Kong, and
the Central Research Committee, Hong Kong Institute
of Education. The trial was registered with the Chinese
Clinical Trial Registry (ChiCTRTRC- 09000374).
Am J Geriatr Psychiatry 22:1, January 2014
Interventions
In each condition, participants practiced the activi-
ties for 1 hour, three times a week for 12 consecutive
weeks. It has been suggested that 2e3 months of cog-
nitive exercisemaybe the optimaldose.10 This duration
also appeared tobeanappropriatedose forTC, judging
from previous trials on physical exercise.8 We adopted
the 136-tile version of mahjong, which is the most
popular one in Hong Kong. For TC, we used a seated
12-form Yang style that has been tailor-made for frail
individuals.29 For the handicraft group, participants
connected beads to create different shapes. This formof
group handicraft is appropriate as a control activity
because it is less mentally and physically demanding
than mahjong and the seated TC but nonetheless pro-
vides control for attention and social activity as well as
any effects on mood and subject expectancy. All act-
ivities were led by an appropriate instructor affiliated
with the research team, with assistance from student
helpers, for the purpose of standardization. Ratio of
student helpers to participants was approximately 1:3
across all groups. (Supplemental Digital Content 1;
available online, summarizes other aspects of the
intervention protocol.)
Measures and Procedure
Primary outcome was MMSE. Secondary outcomes
included forward and backward digit sequence as
well as digit span,30 15-word immediate and 30-
minute delayed recall,30 and categorical verbal
fluency (animals, fruits, and vegetables—1 minute
per category).31 Impaired word-list delayed recall
(tapping episodic memory) and categorical fluency
(semantic memory) are two hallmark features of
early-stage Alzheimer disease,32,33 although the other
cognitive tasks, with the exception of forward digit
span/sequence, are also impaired.31
The battery of cognitive assessments, together with
the 15-item Geriatric Depression Scale,34 was admin-
istered by trained interviewers. Total cholesterol, high-
density lipoprotein cholesterol, triglycerides, and
glucose were assessed around 7e8 A.M. on the day
of assessment, after 12-hour overnight fasting, by
a registered nurse, using calibrated portable devices.
BP and PEF (using a peak-flow meter) were taken at
the same time. Low-density lipoprotein cholesterol
was estimated using the Friedewald formula.35 APOE
65
FIGURE 1. Flow diagram of the study.
Leisure Activities and Cognition
genotyping was done by restriction fragment length
polymorphism approach according to a modified
method.36 In this sample, there was only one person
who was APOE4-homozygous, and so, all partici-
pants who carried the APOE4 allele were lumped into
a group (0¼ noAPOE4, 1¼with APOE4). In addition,
home staff provided information on cholinesterase
inhibitor use (0 ¼ none, 1 ¼ yes) and diagnosed coro-
nary heart disease, stroke, hypertension, diabetes, and
chronic obstructive pulmonary disease, all of which
have been implicated in cognitive decline.19,37,38 A
composite score (0e5) indicates the number of illnesses
66
diagnosed in the participant. All outcome measures
and covariates, except APOE4 and medical diagnoses,
were obtained at baseline and subsequently at 3
(posttreatment), 6, and 9 months.
Moreover, residents were free to participate in
activities. It was possible that they participated in
activities intended for another treatment group (e.g.,
a control participant playing mahjong in his or her
leisure time) as well as in other activities of a similar
nature (e.g., TC might not be the only physical
activity). Hence, we asked nursing home staff to
record on a daily log sheet their participation in
Am J Geriatr Psychiatry 22:1, January 2014
Cheng et al.
various leisure activities (not limited to mahjong, TC,
and handicrafts) throughout the entire study period;
up to three activities lasting 20 minutes or more were
recorded daily. During the 12-week intervention
period, only activities that were not assigned for the
trial were recorded. Finally, new falls and strokes
after the start of the intervention were recorded on
a weekly sheet.
Statistical Analysis
Analysis of variance and c2 tests were conducted
to examine whether the groups differed at baseline
on the outcome measures as well as potential con-
founding factors. For the main analyses, there were
three levels in the data structure: repeated measure-
ments (level 1), within participants (level 2), and
within homes (level 3). Home-level intraclass corre-
lations for all outcome measures were 0.00e0.02
(mean ¼ 0.01). Intent-to-treat analyses were per-
formed on the multilevel data structure with mixed-
effects regression, using full-information maximum
likelihood estimation in Stata version 11.1 (StataCorp,
College Station, Texas). The advantage of this pro-
cedure is that the missing data imputation is not
necessary because the conditional distribution of
missing data on the basis of the observed data is
incorporated into the estimation of the parameters.
Therefore, data available from all the participants
were included to provide unbiased estimates.39 In the
regression models, the intercepts and effects of time
were specified to vary randomly at home and
participant levels, whereas the regression coefficients
of the predictors were treated as fixed effects. Within-
person covariance over time was specified using an
unstructured model. All regressions were linear
models aside from those for immediate and delayed
word-list recall, forwhichPoissonmodeling for discrete
events was used as the words correctly recalled were
few.40 Measures were entered as time-varying predic-
tors when they were obtained at all time points.
(Supplemental Digital Content 2; available online,
provides a detailed explanation of the multilevel
equations.)
There were one primary outcome and seven
secondary outcome measures in this study, and hence,
a risk of Type I error inflation for the secondary out-
comes. The Bonferroni correction is overly conservative
when the number of tests is high,the sample size is
Am J Geriatr Psychiatry 22:1, January 2014
small, and the outcome measures are correlated.41 Not
surprisingly, the present set of secondary outcome
measures was moderately correlated across all time
points (average r ¼ 0.33). Following the recommenda-
tion by Bender and Lange,41 we treated the analyses
involving secondary outcomes in an exploratory
fashion and did not adjust the p value. This is appro-
priate, given the novelty of this type of research. Thus,
for all tests, a levels were set at 0.05 (two-tailed).
We performed a number of procedures to select
appropriate variables for the final analytical model.
First, we tested to see whether the effects of inter-
vention varied by time and APOE4 status. The latter
was included because of the potency of APOE4 on
cognitive decline19 and because of the increasing need
to understand factors that moderate the effectiveness
of psychosocial interventions for older people.42 As
there were three groups, two dummies were created,
with the control as the reference group, and product
terms were created between the two treatment groups
on the one hand and time (centered) and APOE4 on
the other. These product terms were included to
examine whether treatment effects varied by time and
APOE status. Only time showed moderating effects,
and product terms involving APOE were dropped.
Second, we explored the effects of other covariates,
including age, sex, education, chronic illnesses, BP,
lipids, glucose, PEF, depression, cholinesterase inhib-
itor use (time-varying), and new falls (no participant
experienced stroke during the study period). Bivariate
correlations showed that age, sex, cholinesterase
inhibitors, and new falls were unrelated to outcome
variables and were dropped. Low-density lipoprotein
cholesterol correlated at r w0.87 across all time points
with total cholesterol, and so, it was dropped to avoid
collinearity. We then regressed each outcome measure
on the remaining covariates, together with APOE4.
Only education, APOE4, chronic illnesses, PEF, and
depression had independent effects on the outcomes.
Finally, we recomputed the findings, using a final
model consisting of mahjong, TC, time, mahjong �
time, TC � time, education, APOE4, chronic illnesses,
diastolic BP, PEF, and depression (the last three were
entered as time-varying predictors). Diastolic BP was
also included as a covariate as there was a group
difference at baseline.
When interactions between groups and time were
significant, we probed at which time point treatment
effect occurred by calculating the adjusted mean
67
Leisure Activities and Cognition
difference between treatment and control. As
mahjong and TC were coded as dummies, this was
done by calculating the simple slopes at specific time
points. For linear regression, effect size was indicated
by Cohen’s d. For Poisson regression, effect size is the
exponential function of the simple slope43; the
difference from 1 represents percentage change in the
outcome measure due to treatment as compared with
control.
Finally, two supplementary analyses were per-
formed. First, as mentioned earlier, participants might
initiate activities prescribed for another experimental
condition, thus contaminating estimates of treatment
effects. We repeated the analyses for the subsample
without such crossover activities throughout the
9-month period; two control, three mahjong, and four
TC participants were excluded accordingly. Because
the sample size became smaller, we reduced the
number of covariates by including only those thatwere
significant in the main analyses. (We did not report
detailed data on activity participation after the inter-
vention because the frequencies were low and had no
impact on the outcomes beyond that of the experi-
mental assignment; see also another report18). Second,
as indicated previously, the treatment groups were
coded as dummies with control as the reference group;
thus, mahjong and TC were each compared with
control but not with each other. To ascertain whether
the effects of mahjong and TC were significantly
different from each other, we dropped the control
group while recoding TC ¼ 0 and mahjong ¼ 1, thus
allowing direct comparison between the latter two.
RESULTS
The three groups did not differ on any baseline
characteristic other than diastolic BP (Table 1), which
was included as a covariate in further analyses.Mixed-
effects regression (Table 2) showed a significant
negative effect of time on MMSE and digit forward
memory (both span and sequence), meaning that these
cognitive abilities declined with time. People with
APOE4 had lower verbal memory (word-list recall)
and MMSE. Higher education was associated with
better digit forward memory but not in delayed recall
and categorical fluency for which the relationships
were surprisingly reversed. PEF appeared to be a good
biological marker for cognitive decline, with low PEF
68
being consistently associated with lower MMSE, cate-
gorical fluency, and digit forward memory, although
its relationship with delayed recall was surprisingly
reversed. Depression and diastolic BP were also asso-
ciated with selected measures.
Mahjong’s effect varied by time for MMSE and
forward digit span and sequence. It also had main
effects on these measures as well as on categorical
fluency. TChad the same interactionswith time, but no
main effects on any variables. The mahjong/TC-by-
time interaction effects are displayed in Figure 2.
These interaction effects were driven by a gradual
decline in the performance of control participants,
whereas the performances of the mahjong and TC
groupsweremaintained or even improved, so that the
differences between treatment and control were en-
larged over time. In fact, significant difference between
groups usually appeared at 6 and 9months only,when
control participants had reached a certain degree of
decline. This pattern was rather robust, as when a
measure did not showprospective decline (i.e. effect of
time was nonsignificant or positive; Table 2), the
treatment by time interaction did not exist. By 6 and 9
months, the primary outcome measure of MMSE
showed a difference between mahjong and control
equal to 3.0 (95%CI: 0.9e5.0; d¼ 0.34) and 4.5 (95%CI:
2.0e6.9; d¼ 0.48) points, respectively, and a difference
between TC and control equal to 2.3 (95% CI: 0.4e4.2;
d ¼ 0.26) and 3.7 (95% CI: 1.4e6.0; d¼0.40) points,
respectively. In addition, over the 9-month period of
the study, the control group dropped 2.9 points (95%
CI: �4.2 to �1.7) on the MMSE, whereas the mahjong
group and the TC group gained 1.5 (95% CI: �0.0 to
3.0) and 1.3 (95% CI: �0.0 to 2.5) points, respectively.
For the secondary outcomes, at 6 and 9 months,
mahjong differed from control by (a) 1.11 (95% CI:
0.45e1.77; d ¼ 0.31) and 1.78 (95% CI: 0.86e2.71;
d ¼ 0.45) points, respectively, on forward digit span,
and (b) 1.02 (95%CI: 0.33e1.70; d¼ 0.27) and 1.68 (95%
CI: 0.78e2.57; d¼ 0.42) points, respectively, on forward
digit sequence.Bycomparison,TCdifferedsignificantly
from control at 9 months only (difference between
means: 0.98; 95%CI: 0.12e1.84; d¼ 0.25) on the forward
digit span measure. Despite the overall significant
interaction with time on forward digit sequence, the
individual means were not significantly different from
those of the control group at all time points.
For categoricalfluency, only amain effect bymahjong
(overall mean: 13.4, SD: 12.4; 95% CI: 9.3e17.5) over
Am J Geriatr Psychiatry 22:1, January 2014
TABLE 1. Baseline Sample Characteristics
Control (n [ 35) Mahjong (n [ 36) Tai Chi (n [ 39) F c2 df p
Demographics
Age, years 80.9 (7.2) 81.9 (6.2) 81.8 (7.4) 0.214 2, 107 0.808
Sex (women %) 66 64 64 0.031 2 0.985
Educational level (%) 6.619 4 0.157
No formal education 51 25 44
Primary 37 50 33
Secondary or above 12 25 23
Health measures and depression
Hypertension, % 57 64 72 1.741 2 0.419
Coronary heart disease, % 29 25 26 0.133 2 0.936
Stroke, % 26 39 33 1.409 2 0.494
Diabetes, % 31 25 31 0.438 2 0.803
COPD, % 9 6 15 2.1262 0.345
Systolic BP, mm Hg 153.6 (33.1) 140.0 (26.3) 144.6 (24.5) 2.186 2, 107 0.117
Diastolic BP, mm Hg 87.9 (18.7) 75.0 (13.5) 77.7 (11.7) 7.500 2, 107 0.001
Total cholesterol, mg/dL 225.4 (68.5) 212.1 (43.4) 218.1 (50.9) 0.512a 2, 89.6 0.601
HDL cholesterol, mg/dL 62.1 (21.4) 50.0 (15.1) 54.5 (25.3) 2.192a 2, 96.2 0.117
LDL cholesterol, mg/dL 140.7 (58.4) 139.3 (35.7) 137.1 (35.7) 0.063a 2, 82.8 0.939
Triglycerides, mg/dL 111.5 (77.8) 127.6 (58.0) 133.3 (80.2) 0.875 2, 107 0.420
Blood glucose, mmol/L 5.87 (1.00) 6.06 (1.17) 6.55 (2.11) 1.984 2, 107 0.143
PEF, L/min 152.6 (71.9) 156.4 (95.7) 167.9 (64.3) 0.393 2, 107 0.676
Geriatric Depression Scale 6.20 (3.98) 5.42 (3.44) 5.59 (3.87) 0.422 2, 107 0.657
Genetic predisposition
APOE4 (yes %) 26 22 23 0.130 2 0.937
Medication
Cholinesterase inhibitors (%) 9 8 0 3.489 2 0.175
Cognitive measures
Clinical Dementia Rating, % 1.267 4 0.867
0.5 49 50 41
1 34 39 41
2 17 11 18
MMSE 18.9 (4.1) 19.0 (3.2) 18.7 (3.9) 0.068 2, 107 0.935
Verbal immediate recall 1.60 (1.65) 1.22 (1.31) 1.31 (1.15) 0.721a 2, 94.3 0.489
30-min verbal delayed recall 1.20 (2.34) 0.72 (1.52) 1.33 (2.36) 0.852 2, 107 0.430
Categorical fluency 12.9 (7.6) 13.9 (5.3) 13.3 (5.3) 0.215a 2, 91.0 0.807
Digit forward span 6.11 (1.32) 6.11 (1.39) 6.33 (1.78) 0.267 2, 107 0.766
Digit forward sequence 5.91 (1.46) 5.72 (1.49) 5.95 (1.75) 0.220 2, 107 0.803
Digit backward span 2.77 (1.66) 2.81 (1.33) 3.05 (1.62) 0.367 2, 107 0.693
Digit backward sequence 2.29 (1.38) 2.36 (1.10) 2.41 (1.35) 0.087 2, 107 0.917
Notes: Values in parentheses represent SDs. p value <.05 is boldfaced. COPD, chronic obstructive pulmonary disease; HDL, high-density
lipoprotein; LDL, low-density lipoprotein.
aBrowneForsythe adjusted F.
Cheng et al.
control (overall mean: 9.9, SD: 12.7; 95%CI:¼ 5.7e14.2)
was observed. No effects for mahjong and TC were
observed for verbal memory (whether immediate or
delayed recall) and digit backward memory.
Supplementary Analysis
The first supplementary analysis by dropping
participants with crossover activities produced
essentially the same results. The second analysis
comparing mahjong against TC showed a significant
mahjong main effect for categorical fluency (Z ¼ 2.11,
p ¼ 0.034) and forward digit span (Z ¼ 2.06,
Am J Geriatr Psychiatry 22:1, January 2014
p ¼ 0.039), as well as a significant mahjong-by-
time interaction effect for forward digit sequence
(Z ¼ 1.96, p ¼ 0.050). However, there was no
significant difference between the two treatments on
the primary outcome MMSE. On the whole, there
was some evidence that mahjong might be better
than TC in preserving selected cognitive domains.
DISCUSSION
This study provided experimental support to the
effects of mental and physical activities in delaying
69
TABLE 2. Results of Mixed-Effects Regression
MMSE Verbal Immediate Recall Verbal Delayed Recall Categorical Fluency
B (95% CI) p B (95% CI) p B (95% CI) p B (95% CI) p
Mahjong 2.240 (0.361 to 4.118) 0.019 0.311 (�0.026 to 0.648) 0.071 0.210 (�0.961 to 1.381) 0.725 3.460 (1.022 to 5.897) 0.005
Tai chi 1.617 (�0.175 to 3.410) 0.077 0.076 (�0.247 to 0.399) 0.646 0.937 (�0.153 to 2.028) 0.092 1.224 (�1.108 to 3.557) 0.304
Time (centered) �0.978 (�1.393 to �0.564) <0.001 0.095 (�0.035 to 0.224) 0.152 0.024 (�0.237 to 0.285) 0.855 �0.446 (�1.395 to 0.503) 0.357
Mahjong � time 1.477 (0.829 to 2.124) <0.001 0.127 (�0.076 to 0.330) 0.219 0.369 (�0.031 to 0.770) 0.070 1.125 (�0.292 to 2.541) 0.120
Tai chi � time 1.395 (0.818 to 1.972) <0.001 0.095 (�0.088 to 0.277) 0.309 0.085 (�0.255 to 0.424) 0.625 0.663 (�0.725 to 2.051) 0.349
Education �0.743 (�1.629 to 0.143) 0.100 �0.131 (�0.309 to 0.048) 0.152 �0.761 (�1.419 to �0.103) 0.024 �2.564 (�3.842 to �1.286) <0.001
APOE4 �1.904 (�3.467 to �0.342) 0.017 �0.337 (�0.653 to �0.021) 0.036 �1.342 (�2.503 to �0.182) 0.023 �2.021 (�4.205 to 0.164) 0.070
Chronic illnesses �0.133 (�0.724 to 0.459) 0.661 �0.134 (�0.253 to �0.015) 0.027 �0.510 (�0.923 to �0.097) 0.015 �0.544 (�1.380 to 0.292) 0.202
Diastolic BP �0.001 (�0.024 to 0.021) 0.900 0.002 (�0.005 to 0.009) 0.563 0.014 (0.003 to 0.026) 0.012 0.048 (0.012 to 0.085) 0.009
PEF 0.007 (0.003 to 0.012) 0.001 0.001 (�0.000 to 0.002) 0.058 �0.003 (�0.005 to �0.0004) 0.023 0.015 (0.008 to 0.022) <0.001
Depression �0.058 (�0.158 to 0.042) 0.256 �0.013 (�0.041 to 0.014) 0.341 �0.005 (�0.045 to 0.056) 0.837 �0.182 (�0.337 to �0.028) 0.021
Digit Forward Span Digit Forward Sequence Digit Backward Span Digit Backward Sequence
B (95% CI) p B (95% CI) p B (95% CI) p B (95% CI) p
Mahjong 0.777 (0.189 to 1.364) 0.010 0.686 (0.064 to 1.308) 0.031 0.619 (�0.040 to 1.277) 0.066 0.471 (�0.037 to 0.979) 0.069
Tai chi 0.404 (�0.153 to 0.960) 0.155 0.149 (�0.440 to 0.737) 0.621 0.575 (�0.042 to 1.191) 0.068 0.468 (�0.016 to 0.952) 0.058
Time (centered) �0.226 (�0.452 to 0.000) 0.050 �0.283 (�0.489 to �0.078) 0.007 �0.076 (�0.289 to 0.138) 0.487 �0.038 (�0.217 to 0.141) 0.678
Mahjong � time 0.671 (0.277 to 1.064) 0.001 0.659 (0.335 to 0.984) <0.001 0.193 (�0.146 to 0.532) 0.265 0.080 (�0.194 to 0.354) 0.567
Tai chi � time 0.384 (0.014 to 0.755) 0.042 0.317 (0.031 to 0.603) 0.030 0.137 (�0.162 to 0.436) 0.369 0.169 (�0.080 to 0.418) 0.184
Education 0.344 (0.466 to 0.642) 0.023 0.342 (0.030 to 0.653) 0.032 �0.001 (�0.338 to 0.335) 0.994 0.137 (�0.124 to 0.398) 0.304
APOE4 �0.446 (�0.969 to 0.077) 0.095 �0.122 (�0.669 to 0.424) 0.661 �0.362 (�0.954 to 0.230) 0.231 �0.234 (�0.692 to 0.225) 0.318
Chronic illnesses 0.045 (�0.156 to 0.246) 0.659 0.015 (�0.196 to 0.225) 0.892 �0.024 (�0.249 to 0.201) 0.834 �0.026 (�0.201 to 0.149) 0.771
Diastolic BP 0.002 (�0.009 to 0.013) 0.740 0.002 (�0.009 to 0.013) 0.699 0.004 (�0.007 to 0.015) 0.499 0.003 (�0.006 to 0.012) 0.500
PEF 0.003 (0.001 to 0.005) 0.006 0.003 (0.001 to 0.005) 0.004 0.0003 (�0.002 to 0.002) 0.795 �0.0005 (�0.002 to 0.001) 0.581
Depression �0.046 (�0.091 to �0.001) 0.043 �0.067 (�0.113 to �0.022) 0.003 �0.030 (�0.077 to 0.018) 0.224 �0.017 (�0.053 to 0.020) 0.369
Notes: p values were based on Z tests; those <0.05 are boldfaced.
70
A
m
J
G
eriatr
P
sych
iatry
22:1,
Jan
u
ary
2014
L
e
isu
re
A
ctiv
itie
s
a
n
d
C
o
g
n
itio
n
FIGURE 2. Effects of treatment varied with time on [A] MMSE, [B] forward digit span, and [C] forward digit sequence. Scores
displayed are means adjusted for education, APOE4, chronic illnesses, diastolic blood pressure, peak expiratory flow,
and depression. Error bars denote 95% confidence intervals. Bars with letters a, b, and c indicate significant difference
from control group based on Z tests; *p <0.05, **p <0.01, ***p <0.001.
Am J Geriatr Psychiatry 22:1, January 2014 71
Cheng et al.
Leisure Activities and Cognition
cognitive decline in individuals with dementia and
extended a recent study that showed that mahjong,
but not TC, had short-term effects on reducing
depressive symptoms in nursing home residents with
dementia.18 The findings are important because,
although similar studies have been done on physical
activity,8 this study represented the first randomized
controlled trial on cognitive leisure activity in indi-
viduals with dementia, with a built-in activity-
control group and repeated assessments of cognitive
performance up to 9 months or 6 months after the
completion of treatment. Results supported the rela-
tively long-term benefits of both types of activities,
although there was preliminary evidence that mah-
jong might be more effective than TC in short-term
memory of numerical units. The relative superiority
of mahjong in short-term digit memory could not be
attributed to practice effects as all the three groups
were exposed to the same amount of practice over
time.
Not all mental activities are equally effective. It
appears that only cognitively demanding activities
(e.g., mahjong), or what some researchers call
“complex” cognitive activity,4,7 may benefit individ-
uals for whom decline is already under way, whereas
less-demanding activities (e.g., simple handicrafts)
may not do the same. Moreover,although TC tended
to have smaller effects than mahjong, our analysis
showed that they were not very different from each
other. The findings paralleled that of a recent study
showing that a computer course and a physical ex-
ercise program produced similar improvements in
episodic memory and executive control over a nil-
treatment control group in healthy older women.4
However, it should also be noted that TC is
different from most other physical exercises in that it
requires memorization of complex motor sequence
and may hence have additional benefits on memory
when compared with other forms of physical exercise
without such an element.13 To what extent the
present findings about TC can be generalized to other
exercise modalities in this population remains to be
researched.
Not all cognitive domains benefited from TC and
mahjong. As expected, there were positive effects on
global cognition (i.e., MMSE). Consistent with
a preliminary study,12 digit forward, but not digit
backward, memory benefited. No effects on verbal
memory were found. Moreover, mahjong had a main
72
effect but no interaction with time on categorical
fluency. Judging from the baseline scores and regres-
sion results favoring digit forwardmemory over other
cognitive tasks (aside fromMMSE), it appears that the
strongest effects of leisure activities are in domains (in
this case, short-term memory) that are relatively well-
preserved to begin with. (No normative data on digit
span exist for very old populations, but a study
involving 118 Chinese older adults age 72.8 years and
without dementia reported amean forward digit span
of 7.4 units.31) More research incorporating other
cognitive measures is needed to test the limits of such
interventions.
Study Limitations
We acknowledge three limitations in this study.
First, a major limitation was the lack of blinding, as
activities could not be masked and it was not possible
to prevent residents from talking to interviewers
about the activities. Nevertheless, interviewers were
blind to the hypothesis and all outcome measures
were objective tests. Second, participants were rec-
ruited after randomization, leading to a possible
selection bias, which is a common issue in cluster
designs. However, this is the only way to prevent
treatment contamination in nursing home settings.
Moreover, the groups did not differ on a large
number of potential confounds other than diastolic
BP, which was controlled in subsequent analysis.
Finally, as is true in this region, few participants were
on cholinesterase inhibitors. Thus, it is not clear
whether the results are specific to older adults with
dementia who are cholinesterase-inhibitor naive.
Future studies are needed to investigate whether
these psychosocial treatments would add to the
effects of medications to produce optimal treatment
effects.42 Despite these limitations, it is noteworthy
that all homes were committed to implementing the
condition to which they were assigned, with few
attritions over time. The longitudinal design, together
with a randomized control group in a population
with high risk for cognitive decline, provides strong
support for the therapeutic effects of selected leisure
activities in old age, even in those with significant
cognitive impairment. Although we do not yet have
data about how such activities work in conjunction
with medications, the fact that they can lead to
significant improvements in MMSE is an important
Am J Geriatr Psychiatry 22:1, January 2014
Cheng et al.
support for their therapeutic values, especially in
developing countries where prescriptions for cholin-
esterase inhibitors may be deterred by costs. For the
nursing home population in particular, policies and
practices that provide active support for leisure
activities of sufficient intensity are warranted in light
of the present findings.
The authors thank William Tsang for making the
sitting Yang-style Tai Chi available for the study, the
nursing homes for assistance in data collection, and Isa-
bella Chan for outstanding efforts in collecting biological
data. They also thank Sing-Kai Lo for his assistance in data
analysis and Kee-Lee Chou for his contribution to initial
conceptualization of the study.
Am J Geriatr Psychiatry 22:1, January 2014
This study was supported by Competitive Earmarked
Research Grant No. HKIEd141307 of the Research Grants
Council of Hong Kong to Sheung-Tak Cheng. The funding
source had no involvement in any part of the project.
Author contributions: STC is the principal investi-
gator and led study design, data analysis, and writing the
article. PKC and STC designed the study protocol and
managed the project together; in addition, PKC designed
certain instruments, collected data, and contributed to the
writing. YQS performed APOE genotyping and contrib-
uted to the writing. JHML assisted STC in data analysis and
writing of the results. ECSY and TMCL contributed to the
initial study design and to the training of research assistants
for administering certain assessment instruments. ECSY
and ACMC assisted in the recruitment of study settings.
References
1. Akbaraly TN, Portet F, Fustinoni S, et al: Leisure activities and the
risk of dementia in the elderly: results from the Three-City Study.
Neurology 2009; 73:854e861
2. Verghese J, Lipton RB, Katz MJ, et al: Leisure activities and the
risk of dementia in the elderly. N Engl J Med 2003; 348:
2508e2516
3. Wilson RS, Mendes dL, Barnes LL, et al: Participation in cogni-
tively stimulating activities and risk of incident Alzheimer disease.
JAMA 2002; 287:742e748
4. Klusmann V, Evers A, Schwarzer R, et al: Complex mental and
physical activity in older women and cognitive performance:
a 6-month randomized controlled trial. J Gerontol A Biol Sci Med
Sci 2010; 65A:680e688
5. Park DC, Bischof G: Neuroplasticity, aging, and cognitive func-
tion, in Handbook of the Psychology of Aging. 7th ed. Edited by
Schaie KW, Willis SL. San Diego, CA, Elsevier Academic Press,
2011, pp 109e119
6. Foster PP, Rosenblatt KP, Kulji�s RO: Exercise-induced cognitive
plasticity, implications for mild cognitive impairment and
Alzheimer’s disease. Front Neurol 2011; 2:28
7. Saczynski JS, Jonsdottir MK, Sigurdsson S, et al: White matter
lesions and cognitive performance: the role of cognitively complex
leisure activity. J Gerontol A Biol Sci Med Sci 2008; 63A:848e854
8. Heyn P, Abreu BC, Ottenbacher KJ: The effects of exercise training
on elderly persons with cognitive impairment and dementia:
a meta-analysis. Arch Phys Med Rehabil 2004; 85:1694e1704
9. Clare L, Woods RT: Cognitive rehabilitation and cognitive
training for early-stage Alzheimer’s disease and vascular
dementia. Cochrane Database Syst Rev 2003;(4):CD003260
10. ValenzuelaM, Sachdev P: Can cognitive exercise prevent the onset
of dementia? Systematic review of randomized clinical trials with
longitudinal follow-up. Am J Geriatr Psychiatry 2009; 17:179e187
11. Treiber KA, Carlson MC, Corcoran C, et al: Cognitive stimulation
and cognitive and functional decline in Alzheimer’s disease: the
Cache County Dementia Progression Study. J Gerontol B Psychol
Sci Soc Sci 2011; 66B:416e425
12. Cheng ST, Chan ACM, Yu ECS: An exploratory study of the effect
of mahjong on the cognitive functioning of persons with
dementia. Int J Geriatr Psychiatry 2006; 21:611e617
13. Lam LCW, Chau RCM, Wong BML, et al: Interim follow-up of
a randomized controlled trial comparing Chinese style mind body
(tai chi) and stretching exercises on cognitive function in
subjects at risk of progressive cognitive decline. Int J Geriatr
Psychiatry 2011; 26:733e740
14. Lam LCW, Chau RCM, Wong BML, et al: A 1-year randomized
controlled trial comparing mind body exercise (tai chi) with
stretching and toning exercise on cognitive function in older
Chinese adults at risk of cognitive decline. JAMDA 2012; 13(6):
568.e15e568.e20 [published online ahead of print]
15. Cheng ST, Chan ACM: Regulating quality of care in nursing
homes in Hong Kong: a social-ecological investigation. LawPolicy 2003; 25:403e423
16. Cheng ST: The social networks of nursing-home residents in
Hong Kong. Ageing Soc 2009; 29:163e178
17. Cheng ST, Lam LCW, Chow PK: Under-recognition of dementia in
long-term care homes in Hong Kong. Aging Ment Health 2012;
16:516e520
18. Cheng ST, Chow PK, Yu ECS, et al: Leisure activities alleviate
depressive symptoms in nursing home residents with very
mild or mild dementia. Am J Geriatr Psychiatry 2012; 20:
904e908
19. Martins IJ, Hone E, Foster JK, et al: Apolipoprotein E, cholesterol
metabolism, diabetes, and the convergence of risk factors for
Alzheimer’s disease and cardiovascular disease. Mol Psychiatry
2006; 11:721e736
20. Euser SM, van Bemmel T, Schram MT, et al: The effect of age on
the association between blood pressure and cognitive function
later in life. J Am Geriatr Soc 2009; 57:1232e1237
21. Albert MS, Jones K, Savage CR, et al: Predictors of cognitive
change in older persons: MacArthur Studies of Successful Aging.
Psychol Aging 1995; 10:578e589
22. Modrego PJ, Ferrández J: Depression in patients with mild
cognitive impairment increases the risk of developing dementia
of Alzheimer type. Arch Neurol 2004; 61:1290e1293
23. Yasuno F, Tanimukai S, Sasaki M, et al: Association between
cognitive function and plasma lipids of the elderly after control-
ling for apolipoprotein E genotype. AmJ Geriatr Psychiatry 2012;
20:574e583
73
Leisure Activities and Cognition
24. Kohler S, van Boxtel M, Jolles J, et al: Depressive symptoms and
risk for dementia: a 9-year follow-up of the Maastricht Aging
Study. Am J Geriatr Psychiatry 2011; 19:902e905
25. Wysocki M, Luo X, Schmeidler J, et al: Hypertension is associated
with cognitive decline in elderly people at high risk for dementia.
Am J Geriatr Psychiatry 2012; 20:179e187
26. Folstein MF, Folstein SE, McHugh PR: Mini-Mental State: a prac-
tical method for grading the cognitive state of patients for the
clinician. J Psychiatr Res 1975; 12:189e198
27. Morris JC: The Clinical Dementia Rating (CDR): current version
and scoring rules. Neurology 1993; 43:2412e2414
28. Heo M, Leon AC: Sample size requirements to detect an inter-
vention by time interaction in longitudinal cluster randomized
clinical trials. Stat Med 2009; 28:1017e1027
29. Lu X, Wang BB, Lee YT, et al: Sitting tai chi can improve the eye
hand coordination in frail older adults. Chin J Rehabil Med 2009;
24:236
30. Lee TMC, Yuen KSL, Chan CCH: Normative data for neuropsy-
chologicalmeasures of fluency, attention, andmemorymeasures for
Hong Kong Chinese. J Clin Exp Neuropsychol 2002; 24:615e632
31. Lam LCW, Ho P, Lui VWC, et al: Reduced semantic fluency as an
additional screening tool for subjects with questionable
dementia. Dement Geriatr Cogn Disord 2006; 22:159e164
32. Salmon DP, Thomas RG, Pay MM, et al: Alzheimer’s disease can
be accurately diagnosed in very mildly impaired individuals.
Neurology 2002; 59:1022e1028
33. Vliet EC, Manly J, Tang M, et al: The neuropsychological profiles
of mild Alzheimer’s disease and questionable dementia as
compared to age-related cognitive decline. J Int Neuropsychol
Soc 2003; 9:720e732
74
34. Sheikh JI, Yesavage JA: Geriatric Depression Scale (GDS): recent
evidence and development of a shorter version. Clin Gerontol
1986; 5:165e173
35. Branchi A, Rovellini A, Torri A, et al: Accuracy of calculated
serum low-density lipoprotein cholesterol for the assessment of
coronary heart disease risk in NIDDM patients. Diabetes Care
1998; 21:1397e1402
36. Song YQ, Rogaeva E, Premkumar S, et al: Absence of association
between Alzheimer disease and the �491 regulatory region
polymorphism of APOE. Neurosci Lett 1998; 250:189e192
37. Areza-Fegyveres R, Kairalla RA, Carvalho CRR, et al: Cognition
and chronic hypoxia in pulmonary diseases. Dement Neuro-
psychol 2010; 4:14e22
38. de Toledo FA, Ferreira LK, Wajngarten M, et al: Cardiac disorders
as risk factors for Alzheimer’s disease. J Alzheimers Dis 2010; 20:
749e763
39. Rabe-Hesketh S, Skrondal A: Multilevel and Longitudinal Modeling
Using Stata. College Station, TX, Stata Press, 2008
40. Whyte EM, Mulsant BH, Butters MA, et al: Cognitive and
behavioral correlates of low vitamin B12 levels in elderly patients
with progressive dementia. Am J Geriatr Psychiatry 2002; 10:
321e327
41. Bender R, Lange S: Adjusting for multiple testing—when and
how? J Clin Epidemiol 2001; 54:343e349
42. McKibbin C, Deacon B: Psychosocial interventions for mental
disorders in late life: are we making progress toward efficiency
and impact? Am J Geriatr Psychiatry 2011; 19:835e838
43. Faul F, Erdfelder E, Buchner A, et al: Statistical power analyses
using G*Power 3.1: tests for correlation and regression analyses.
Behav Res Methods 2009; 41:1149e1160
Am J Geriatr Psychiatry 22:1, January 2014
	Mental and Physical Activities Delay Cognitive Decline in Older Persons with Dementia
	Methods
	Study Design, Setting, and Participants
	Interventions
	Measures and Procedure
	Statistical Analysis
	Results
	Supplementary Analysis
	Discussion
	Study Limitations
	References