Content uploaded by Ricardo Uauy
Author content
All content in this area was uploaded by Ricardo Uauy
Content may be subject to copyright.
Effect of 2-y n23 long-chain polyunsaturated fatty acid
supplementation on cognitive function in older people:
a randomized, double-blind, controlled trial
1–3
Alan D Dangour, Elizabeth Allen, Diana Elbourne, Nicky Fasey, Astrid E Fletcher, Pollyanna Hardy,
Graham E Holder, Rosemary Knight, Louise Letley, Marcus Richards, and Ricardo Uauy
ABSTRACT
Background: Increased consumption of n23 (omega-3) long-chain
polyunsaturated fatty acids (LC PUFAs), especially eicosapentae-
noic acid (EPA) and docosahexaenoic acid (DHA), may maintain
cognitive function in later life.
Objective: We tested the hypothesis that n23 LC PUFA supple-
mentation would benefit cognitive function in cognitively healthy
older people.
Design: At total of 867 cognitively healthy adults, aged 70–79 y,
from 20 general practices in England and Wales were randomly
assigned into a double-blind controlled trial of daily capsules pro-
viding 200 mg EPA plus 500 mg DHA or olive oil for 24 mo.
Treatment-allocation codes were obtained from a central computer-
ized randomization service. Trained research nurses administered
a battery of cognitive tests, including the primary outcome, the
California Verbal Learning Test (CVLT), at baseline and 24 mo.
Intention-to-treat analysis of covariance, with adjustment for base-
line cognitive scores, age, sex, and age at leaving full-time educa-
tion, included 748 (86%) individuals who completed the study.
Results: The mean age of participants was 75 y; 55% of the par-
ticipants were men. Withdrawals and deaths were similar in active
(n= 49 and n= 9, respectively) and placebo (n= 53 and n=8,
respectively) arms. Mean (6SD) serum EPA and DHA concentra-
tions were significantly higher in the active arm than in the placebo
arm at 24 mo (49.9 62.7 mg EPA/L in the active arm compared
with 39.1 63.1 mg EPA/L in the placebo arm; 95.6 63.1 mg
DHA/L in the active arm compared with 70.7 62.9 mg DHA/L
in the placebo arm). There was no change in cognitive function
scores over 24 mo, and intention-to-treat analysis showed no sig-
nificant differences between trial arms at 24 mo in the CVLT or any
secondary cognitive outcome.
Conclusions: Cognitive function did not decline in either study arm
over 24 mo. The lack of decline in the control arm and the relatively
short intervention period may have limited our ability to detect any
potential beneficial effect of fish oil on cognitive function in this
study. The Older People And n–3 Long-chain polyunsaturated fatty
acids (OPAL) Study was registered at www.controlled-trials.com as
ISRCTN 72331636. Am J Clin Nutr 2010;91:1725–32.
INTRODUCTION
The number of individuals with cognitive impairment is rising
in the United Kingdom (1) and globally (2). Current estimates
suggest that .81 million individuals will have dementia by 2040;
.70% of these individuals will live in public health resource–
constrained low- and middle-income countries (2). The n23
(omega-3) long-chain polyunsaturated fatty acids (LC PUFAs)
eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA),
which are most commonly found in oily fish, are frequently
cited as potential nutrition-specific modifiers of age-related
cognitive decline (3, 4). The human cerebral cortex and retina,
a brain-derived neuronal network specialized for photosignal
transduction and processing, contain high concentrations of
DHA (5, 6), and mechanistic data suggest roles for n23LC
PUFAs in vascular and neuronal health (7, 8).
Several observational studies reported associations of high fish
intake with better cognition (9–11) or a lower risk of dementia
(12–17), although these findings are not uniform across studies
(18–21). A recent randomized controlled trial that included a total
of 302 cognitively healthy Dutch adults aged 65 y showed no
statistically significant benefit of 400 or 1800 mg n23 LC PUFA
supplementation daily over 6 mo (22), and current evidence of
a benefit for cognitive health is not convincing (23).
We investigated whether 700 mg daily oral n23 LC PUFAs
(200 mg EPA + 500 mg DHA) for 24 mo benefitted cognitive
function compared with an olive oil placebo in a large sample of
cognitively healthy older adults in the United Kingdom.
1
From the Nutrition and Public Health Intervention Research Unit
(ADD, RK, and RU), the Medical Statistics Unit (EA and DE), and the
Non-Communicable Disease Epidemiology Unit (AEF), London School of
Hygiene and Tropical Medicine, London, United Kingdom; the Medical
Research Council General Practice Research Framework, London, United
Kingdom (NF and LL); the Clinical Epidemiology and Biostatistics Unit,
Murdoch Children’s Research Institute, Murdoch Royal Children’s Hospital,
Victoria, Australia (PH); the Department of Electrophysiology, Moorfields
Eye Hospital, London, United Kingdom (GEH); and the Medical Research
Council Unit for Lifelong Health and Ageing, London, United Kingdom (MR).
2
Support for the Older People And n–3 Long-chain polyunsaturated fatty
acids (OPAL) Study was provided by the UK Food Standards Agency
(NO5053). The UK National Health Service Research and Development
provided service support costs.
3
Address correspondence to AD Dangour, Nutrition and Public Health
Intervention Research Unit, Department of Epidemiology and Population
Health, London School of Hygiene and Tropical Medicine, Keppel Street,
London WC1E 7HT, United Kingdom. E-mail: alan.dangour@lshtm.ac.uk.
Received December 21, 2009. Accepted for publication March 9, 2010.
First published online April 21, 2010; doi: 10.3945/ajcn.2009.29121.
Am J Clin Nutr 2010;91:1725–32. Printed in USA. Ó2010 American Society for Nutrition 1725
at PONTIFICIA UNIV CATOLICA CHILE on May 6, 2013ajcn.nutrition.orgDownloaded from
SUBJECTS AND METHODS
Participants
Full details of the protocol were published (24). Participants,
aged 70–79 y at baseline, were enrolled at 20 general practices
across England and Wales from the Medical Research Council
General Practice Research Framework. Lists of potential par-
ticipants, prescreened by using general practice records to ex-
clude those with a current diagnoses of diabetes or dementia,
were drawn from practice records and checked by relevant
general practitioners for further exclusions (eg, recent bereave-
ment or terminal illness). Individuals with diabetes have a raised
susceptibility to vascular and neural damage and may there-
fore have been less sensitive to the intervention. Eligible indi-
viduals, selected from lists in random blocks until sample-size
requirements were met, were sent invitations. Those individuals
reporting current daily use of fish-oil supplements were excluded.
Interested eligible individuals attended an appointment at their
general practice where research nurses clarified any queries, and
the Mini-Mental State Examination (MMSE) (25) dementia
screen was administered. Participants with an MMSE score ,24
were excluded and offered a referral to their general practitioner
for further monitoring. Remaining participants who gave full,
informed, written consent were enrolled in the trial. A sub-
sample of 53 participants was also enrolled in an eye-health
study that included a detailed electroretinographic assessment
(data not shown).
Ethics committee approvals
The Older People And n–3 Long-chain polyunsaturated fatty
acids (OPAL) Study received ethics committee approvals from
the National Research Ethics Service (04/mre05/31) and the
London School of Hygiene and Tropical Medicine (2038).
Procedures
Research nurses telephoned a central computerized random-
ization service to obtain treatment-allocation codes previously
generated by the trial statistician. Randomization was minimized
by age group (70–74 and 75–79 y) and general practice to ensure
a balance across trial arms. All project staff were unaware of
group assignments until the completion of the trial and after data
analysis. The dietary intervention consisted of two 650-mg dark-
brown-colored, vanilla-flavored, soft-gel capsules daily. In the
active arm, the capsules contained a total of 200 mg EPA and 500
mg DHA (20:50 ethyl ester; Ocean Nutrition Canada Ltd,
Canada), and in the placebo arm, the capsules contained olive oil
(rich in n29 fatty acids). There is no evidence that olive oil
intake at this amount has any benefits for cognitive function. The
intervention dose was based on United Kingdom recom-
mendations on oily fish consumption (26), typical dietary rec-
ommendations (27), and the recognized safe dose of n23LC
PUFAs. The daily n23 LC PUFA dose used could be achieved
by consuming ’250 g oily fish/wk (or 1.75 portions) (26). The
higher DHA than EPA content in the active arm was justified by
the study rationale, which prioritized the neuroprotective actions
of DHA over the vascular and antithrombotic actions of EPA (8).
Supplements were packaged into identical pots, each containing
180 capsules, and labeled by staff who were not involved in the
study.
At the baseline appointment, information on educational
history (28), habitual fish consumption [frequency of con-
sumption and most commonly consumed types of fish sub-
sequently categorized as oily or white (29)], and 5-y history of
stroke and myocardial infarction (MI) was collected. At baseline
and 24 mo after randomization, height (measured to the nearest
0.1 cm), weight (measured to the nearest 0.1 kg), and blood
pressure (the average of 2 seated measurements separated by
5 min by using digital sphygmomanometers) were measured, and a
30-item General Health Questionnaire (GHQ-30) (30), a measure
of psychological health, was administered (data not shown).
At baseline and after 24 mo, trained research nurses admin-
istered a range of cognitive tests. The primary outcome of the
OPAL study (31) was a test of memory of a 16-item word list (list
A) from the California Verbal Learning Test (CVLT) (32). The
sum of words recalled after each of 3 repeats of list A and the
number of words recalled after a 20-min delay formed the pri-
mary outcome. Word-learning tests were shown to be sensitive to
change in older people over intervention periods of 24 (33) and
36 mo (34). Other memory tests included immediate and delayed
recalls of a short story (Anna Thompson) from the Wechsler
Memory Scale (35) and a test of spatial memory in which par-
ticipants had to replace 10 images in the correct place on a grid
after a 30-s viewing/memorization period (36). Processing-
speed tests included a letter search and cancellation task (37),
a symbol letter modality test (38), and measures of simple and
choice reaction time (39). Executive-function tests included
digit span backward from the Wechsler adult intelligence scale
(40) and verbal fluency (animal naming) (41). Three tests of
prospective memory (42) and digit span forward (40) were also
conducted.
Standard protocols, including the use of tape recordings of the
CVLT word lists and the Anna Thompson story, were used for all
data collection, which was performed in quiet rooms in par-
ticipating general practices. In 96% of participants completing
the trial, baseline and 24-mo cognitive testing was conducted by
the same research nurse. Baseline and 24-mo appointments lasted
’75 min. Participants were provided with relevant contact de-
tails, and further information was provided on the study website
(43). Participants were invited to seven 3-mo follow-up ap-
pointments at which information on minor adverse events was
recorded, and returned capsules were counted as a measure of
adherence. Participant medical records were checked by re-
search nurses every 3 mo, and any deaths or hospital admissions
for stroke or MI over the intervening 3-mo period were re-
corded.
Adherence to randomized intervention
Adherence was measured as the median proportion of the
number of provided capsules that were not returned at follow-up
appointments (on the basis of 2 capsules for each day of follow-
up). We also aimed to measure n23 LC PUFAs in swabs of
buccal cells (44, 45) at baseline and 24 mo after randomization,
but the amount of total lipid on the swabs was shown to be very
small, and the measurement of n23 LC PUFAs was highly
variable and unreliable. Therefore, we added a further measure
of adherence on the basis of serum fatty acid concentrations
24 mo after randomization. Separate ethics committee approvals
were provided in May 2007 to collect a 10-mL fasting blood
1726 DANGOUR ET AL
at PONTIFICIA UNIV CATOLICA CHILE on May 6, 2013ajcn.nutrition.orgDownloaded from
sample from consenting participants 7 d after their 24-mo
appointment. Given the delayed start, blood samples were only
collected on a subsample of 242 individuals, and sera from 235
(98%) participants were available for analysis of fatty acid
composition. There were no important differences in any pri-
mary or secondary outcomes between participants who provided
blood samples and participants who did not provide blood
samples. Blood samples, drawn in 2 ·5-mL serum-separating
tubes, were posted overnight to a handling laboratory that pre-
pared and froze sera at 280°C. Serum fatty acid concentrations
were measured as fatty acid methyl esters by using gas liquid
chromatography (46).
Withdrawals
There were no preset criteria for participant withdrawal during
the trial. Participants who wished only to discontinue supple-
mentation were invited to an interview at 24 mo.
Statistical analyses
The sample size calculation was based on a 0.3-SD difference
between arms in delayed free recall of list A of the CVLT 24 mo
after randomization. To detect a 0.3-SD difference between trial
arms, with 90% power and 1% significance (2-sided), 332
individuals were required per treatment group. Allowing for
a 20% dropout over the 24 mo of intervention, the total sample
size required for the study was 798 individuals. A difference of
0.3 SD was relevant (47) and equated to a one-word difference in
recall between study arms (48).
Primary analysis was carried out on the basis of groups as
randomly assigned (intention-to-treat). Intention-to-treat analysis
included participants who discontinued treatment and completed
cognitive assessment at 24 mo. Results were presented as ap-
propriate effect sizes with a measure of precision (95% CIs).
Analysis of covariance models were used to adjust for baseline
cognitive function test scores and the covariates age, sex, and age
at leaving full-time education. Analyses were conducted in
STATA version 10 (2007; StataCorp LP, College Station, TX).
In agreement with similar studies (22, 34), baseline and 24-mo
data for secondary cognitive outcomes were pooled to calculate
grand means and SDs per test, and this information was used to
create a set of zscores “Z” that were grouped for analysis into
the following cognitive domains:
Global cognitive function = (Z
CVLT sum of words recalled
+Z
CVLT
delayed recall
+Z
prospective memory-test1
+Z
prospective memory-test2
+Z
prospective memory-test3-item
+Z
prospective memory-test3-location
+Z
story recall
+Z
story recall-delayed
+Z
verbal fluency
+Z
letter cancella-
tion
+Z
location memory
+Z
location memory-delayed
+Z
symbol-letter
substitution
+Z
digit span forward
+Z
digit span backward
+Z
simple reaction
time
+Z
choice reaction time
)/17.
Memory = (Z
CVLT sum of words recalled
+Z
CVLT delayed recall
+
Z
location memory
+Z
location memory-delayed
+Z
story recall
+Z
story
recall-delayed
)/6.
Processing speed = (Z
letter cancellation
+Z
simple reaction time
+Z
choice
reaction time
+Z
symbol-letter substitution
)/4.
Executive function = (Z
digit span backward
+Z
verbal fluency
)/2.
Global delay score = (Z
CVLT delayed recall
+Z
location memory delayed
recall
+Z
story recall delayed
)/3.
To avoid losing information from participants with missing
cognitive outcomes the zscores were weighted according to the
number of cognitive outcomes recorded. The cognitive function
primary outcome was missing for 3 (0.7%) participants and 1
(0.2%) participant in the intervention and placebo groups, re-
spectively, at baseline. An independent data monitoring com-
mittee assessed safety and ongoing adherence data.
RESULTS
Participants were randomly assigned between April 2005 and
March 2006. Nearly 14,000 potential participants were identified,
of whom 5309 were sent invitation letters. A total of 2304 (43%)
potential participants reported current daily use of fish-oil sup-
plements and were excluded. Of the remaining 3005 potential
participants, 948 baseline interviews (32%) were scheduled, and
867 individuals were randomly assigned into the study (Figure
1). A median of 42 participants were randomly assigned per
general practice [interquartile range (IQR): 36.5, 50.5]. With-
drawal from the trial (49 individuals in the fish-oil group com-
pared with 53 individuals in the placebo group) and deaths (9
individuals in the fish-oil group compared with 8 individuals in
the placebo group) over the 24 mo after randomization were
similar between trial arms. A further 9 and 17 participants in the
fish-oil and placebo arms, respectively, discontinued randomized
treatment but provided data at the end of the trial. At baseline,
treatment groups were similar for demographic, health, fish
consumption, and cognitive function characteristics (Tables 1
and 2). Of participants who withdrew from the trial, 49 (48%)
were men, and their mean (6SD) age was 74.5 62.7 y. In the
fish-oil and placebo arms, the respective mean baseline CVLT
scores were 22.8 66.3 and 24.0 66.3 total words recalled im-
mediately and 6.9 63.0 and 6.9 63.2 words for delayed recall.
Cognitive function data for 748 participants (86% of randomized
participants), including those from 26 individuals who dis-
continued treatment, were available after 24 mo of intervention.
Reporting of minor adverse events was similar between trial
arms (flatulence: 4.6% of participants in the fish-oil arm com-
pared with 3.0% of participants in the placebo arm; belching:
4.8% of participants in the fish-oil arm compared with 3.0% of
participants in the placebo arm; abdominal discomfort: 2.5% of
participants in the fish-oil arm compared with 3.9% of partic-
ipants in the placebo arm; loose stools: 3.5% of participants in the
fish-oil arm compared with 2.5% of participants in the placebo
arm; and other: 5.5% of participants in the fish-oil arm compared
with 4.6% of participants in the placebo arm) over the course of
the study. There was no difference between the trial arms in the
proportion of capsules provided that were not returned over the
course of the study (fish-oil median: 0.95; IQR: 0.82, 1.00;
placebo median: 0.95; IQR: 0.81, 1.00). Analysis of the 235
serum samples showed that 24 mo after the start of the in-
tervention, concentrations of the n23 LC PUFAs EPA and DHA
were significantly higher in participants in the intervention arm
than those in the placebo arm (Table 3). Concentrations of 2 n26
fatty acids, arachidonic acid (20:4n26) and 20:3n26, were sig-
nificantly higher in the placebo arm, as were 2 constituents of olive
oil, palmitoleic acid, (16:1) and oleic acid (18:1n29) (P= 0.06).
There was no change in primary outcome or cognitive function
domain zscores over the 24 mo after randomization in either
study arm (Table 4), and no difference between trial arms at the
FISH OIL AND COGNITIVE FUNCTION IN OLDER PEOPLE 1727
at PONTIFICIA UNIV CATOLICA CHILE on May 6, 2013ajcn.nutrition.orgDownloaded from
end of the study (Table 4). After adjustment for baseline cog-
nitive function scores, age, sex, and age at leaving full-time
education, the mean difference at 24 mo in the total number of
words recalled over 3 trials in the CVLT between the in-
tervention and control arms was 20.5 words (95% CI: 21.2,
0.2), and the mean difference in delayed recall of list A in the
CVLT was 0.1 words (20.2, 0.4).
There was no change in the secondary cognitive function
measures over the 24 mo after randomization or between arms at
the end of the trial (Table 5). There was no evidence of an effect
of general practice on cognitive outcome variables. Blood
pressure, body mass index, and incidence of hospitalization for
stroke (7 participants in the fish-oil arm compared with 8 par-
ticipants in the placebo arm) or MI (5 participants in the fish-oil
arm compared with 9 participants in the placebo arm) were
similar in the 2 arms at 24 mo.
DISCUSSION
There is considerable public and scientific interest in the
question whether increased n23 LC PUFA consumption is
beneficial for cognitive function in later life, but the available
evidence from randomized controlled trials is extremely limited
(23). We reported findings from the largest randomized con-
trolled trial to date investigating the effect of n23 LC PUFA
supplementation on cognitive function and conducted in 867
cognitively healthy older people in the United Kingdom. There
was no change in cognitive function over the 24 mo of sup-
plementation. Results from our study do not provide evidence to
reject the null hypothesis of no effect of fish-oil supplementation
on cognitive function in cognitively healthy older people.
The study had several strengths arising from its rigorous de-
sign. Randomization was conducted via a secure, automated,
telephone randomization service, and participant retention was
high, probably as a consequence of the good, repeated, and
systematic follow-up by research nurses. The total number of
withdrawals and reasons for withdrawals were similar in the 2
trial arms. Baseline CVLT scores in participants who withdrew
were similar across trial arms, but those who withdrew had
initially poorer cognitive function than those who remained in the
study. Adherence to the intervention, as measured by the pro-
portion of capsules returned, was high. Although no information
FIGURE 1. Participant flow in the Older People And n–3 Long-chain polyunsaturated fatty acids (OPAL) Study. MMSE, Mini-Mental State Examination.
1728 DANGOUR ET AL
at PONTIFICIA UNIV CATOLICA CHILE on May 6, 2013ajcn.nutrition.orgDownloaded from
was collected on serum fatty acid profiles at the study baseline,
there is no reason to believe that the values were different be-
tween trials arms at the start of the study. Significant differences
in fatty acid profiles between trial arms at the end of the study
suggest that the intervention resulted in measurable biological
changes. The cognitive performance tests selected assessed
a range of cognitive domains, and in agreement with recent expert
recommendations (31), we used a test of memory, the CVLT, as
our primary outcome. The CVLT has good reliability, internal
consistency, and validity (32). The cognitive function tests we
selected were similar to those used in other nutritional trials and
are sensitive to change over 2–3-y intervention periods (34, 50).
There were also some weaknesses in the study. It may be that
the population under investigation already consumed sufficient
n23 LC PUFAs; indeed, fatty acid analysis of sera at 24 mo
showed DHA sufficiency in both treatment groups as evidenced
by a relatively high n23 DHA:n26 DPA ratio (51). Study
participants were nondiabetic, and a median MMSE score at the
start of the study of 29 (of 30) showed high cognitive func-
tioning. Other trials investigating the effect of nutritional sup-
plements on cognitive function in older people have similarly
randomly assigned cognitively healthy older people. For ex-
ample, 2 published trials (34, 50) designed to determine whether
vitamin B supplementation could enhance cognitive function
also randomly assigned cognitively healthy older people with
baseline MMSE scores of 29, and one of these trials (34) showed
improvement in function after 36 mo of intervention.
The total number of deaths was 17 during the 2 y of the current
study compared with 55 deaths that would be expected for this
age and sex distribution on the basis of mortality rates in England
and Wales. It may be that the dose or mixture of n23 LC PUFAs
used in the study was insufficient, or that the study intervention
or follow-up periods were too short, and that effects of sup-
plementation may only have become evident after several more
years of follow-up. We designed the study to test the hypothesis
that n23 LC PUFA supplementation would slow cognitive de-
cline. The absence of decline in cognitive function over 24 mo
TABLE 1
Baseline demographic, health, and dietary characteristics of participants in
the Older People And n–3 Long-chain polyunsaturated fatty acids (OPAL)
Study by trial arm
1
Fish oil
(n= 434)
Placebo
(n= 433)
Men (%) 53.4 56.6
Age (y) 74.7 62.5
2
74.6 62.7
70–74 y (%) 56.5 58.3
75–79 y (%) 43.5 41.7
Education
Age at leaving full-time education (y) 16 63.0 16 62.6
No qualifications (%) 32.9 33.3
O level, clerical (%) 25.5 25.5
A level, university (%) 19.2 17.8
Other (%) 22.5 23.4
Vascular health (%)
MI in previous 5 y 3.5 4.4
Stroke in previous 5 y 2.3 2.1
Blood pressure
Systolic (mm Hg) 144 621 145 621
Diastolic (mm Hg) 76 611 77 611
Hypertensive, .140/95 mm Hg (%) 54.9 56.9
BMI (kg/m
2
) 26.5
(24.2, 29.4)
3
27.7
(24.9, 29.8)
,18.5 kg/m
2
(%) 0.7 0.9
.30 kg/m
2
(%) 22.2 24.3
Frequency and type of fish consumption (%)
Once a month or less 8.6 7.9
Once a week/fortnight, mainly white 34.3 32.9
More than once a week, mainly white 28.9 30.8
Once a week/fortnight, mainly oily 12.0 11.3
More than once a week, mainly oily 16.2 17.1
GHQ-30 score 5 (%)
4
20.1 16.7
1
MI, myocardial infarction; GHQ-30, 30-item General Health Ques-
tionnaire.
2
Mean 6SD (all such values).
3
Median; interquartile range in parentheses (all such values).
4
The 30 items in the GHQ-30 are assigned values of 0 and 1, with the 2
most-positive answers scoring 0 and the 2 least-positive answers scoring 1.
The GHQ-30 score is calculated by adding the 30 questions. A GHQ-30 5
is a well-recognized cutoff for possible common mental disorders (49).
TABLE 2
Baseline cognitive function characteristics of participants in the Older
People And n–3 Long-chain polyunsaturated fatty acids (OPAL) Study by
trial arm
1
Fish oil
(n= 433)
Placebo
(n= 433)
MMSE score 29 (28, 29)
2
29 (28, 30)
Memory
CVLT
3
Total words recalled in first 3 trials 23.9 66.1
4
23.9 65.8
Words recalled at delayed recall 7.4 62.8 7.4 62.9
Story recall
Immediate (number of story items
recalled)
10.9 64.0 10.6 63.9
Delayed (number of story items recalled) 8.8 63.9 8.7 63.8
Spatial memory
Immediate (number of correct images) 5.3 62.5 5.2 62.5
Delayed (number of correct images) 4.5 62.5 4.6 62.6
Processing speed
Letter search/cancellation
Number correct as percentage of total
attempts
78.7 612.6 78.5 613.5
Symbol letter modality
Number correct 42.5 610.0 43.9 69.9
Reaction time (s)
Simple 0.30 60.07 0.31 60.22
Choice 0.75 60.54 0.72 60.48
Executive function
Digit span
Forward (number of correct sequences) 8.1 62.4 8.2 62.4
Backward (number of correct sequences) 6.4 62.0 6.5 62.1
Verbal fluency (number of animals named) 19.5 65.2 19.9 65.3
Domain-specific zscores
Global cognitive function 20.02 60.51 0.01 60.49
Memory 20.01 60.7020.02 60.70
Processing 20.05 60.63 0.02 60.67
Executive function 20.01 60.74 0.06 60.78
Global delay 20.02 60.7220.02 60.74
1
MMSE, Mini-Mental State Examination; CVLT, California Verbal
Learning Test.
2
Median; interquartile range in parentheses (all such values).
3
Fish oil, n= 430; placebo, n= 432.
4
Mean 6SD (all such values).
FISH OIL AND COGNITIVE FUNCTION IN OLDER PEOPLE 1729
at PONTIFICIA UNIV CATOLICA CHILE on May 6, 2013ajcn.nutrition.orgDownloaded from
in the total sample suggests that a considerably longer follow-up
period and perhaps repeated cognitive testing would be required
to identify a real change in cognitive function. Long-term pro-
spective observational studies may help elucidate the effect of
exposures on diseases with long latency periods such as de-
mentia and cancer (52), although large long-term trials probably
remain the most reliable means to detect the likely small-to-
moderate effects of therapeutic interventions (53).
The effect of n23 LC PUFA supplementation on cognitively
healthy adults has previously been investigated in one study (22)
that randomly assigned a sample of 302 individuals aged 65 y
(mean MMSE: 28). This trial (22) showed no significant dif-
ference in tests of memory between low- or high-dose n23LC
PUFA arms and placebo over 26 wk of intervention. Our study is
larger, longer, and recruited individuals who were, on average,
5 y older at baseline. The largest trial to date in participants with
Alzheimer disease (n= 174) showed no effect of n23 LC PUFA
supplementation for 6 mo on the MMSE score (54). One rele-
vant ongoing 36-mo trial (55) is investigating the effect of fish
oil, in combination with other potentially important inter-
ventions such as physical exercise and cognitive training, on
cognitive function among frail older people.
Our trial results do not provide evidence of a beneficial effect
of fish-oil supplementation on cognitive function in cognitively
healthy older people in the United Kingdom. We showed no
evidence of harm from consumption of fish-oil supplements over
24 mo in this population on the basis of self-reported side effects
or documented major events. Fish-oil supplementation may be
beneficial for other health-related outcomes (26). Given the
paucity of evidence in this area, we recommend that further trials
be conducted. Such trials could be designed with longer treatment
or follow-up periods, in a population with lower amounts of fish
consumption, or among individuals with mild cognitive im-
pairment.
TABLE 3
Fasting serum fatty acid concentrations (mg/L) at 24 mo in 235 individuals who were randomly assigned into the Older People And n–3 Long-chain
polyunsaturated fatty acids (OPAL) Study
Fatty acid species chemical
nomenclature (common name)
Fish oil (n= 116) Placebo (n= 119)
P
1
Mean 6SD Median Mean 6SD Median
16:0 (palmitic acid) 688.1 615.5 671.2 713.2 616.4 671.3 0.27
16:1 (palmitoleic acid) 70.4 63.0 66.6 81.2 63.2 72.3 0.014
18:0 (stearic acid) 218.6 64.7 216.9 221.3 64.5 220.4 0.68
18:1n29 (oleic acid) 653.9 615.7 638.4 700.4 619.1 668.0 0.062
18:2n26 (linoleic acid) 823.0 619.7 815.2 814.0 621.1 782.9 0.76
18:3n23(a-linoleic acid) 21.5 60.8 19.4 22.0 60.9 20.1 0.71
20:3n26 45.2 61.2 44.5 50.4 61.4 48.1 0.004
20:4n26 (arachidonic acid) 178.4 64.2 176.4 203.0 64.8 200.1 ,0.001
20:5n23 (eicosapentaenoic acid) 49.9 62.7 45.6 39.1 63.1 30.7 0.009
22:4n26 15.7 60.6 14.8 15.7 60.5 14.2 0.96
22:5n26 32.0 61.4 30.3 31.8 61.4 27.6 0.92
22:5n23 18.9 60.5 17.9 19.3 60.6 17.1 0.65
22:6n23 (docosahexaenoic acid) 95.6 63.1 92.9 70.7 62.9 65.3 ,0.001
1
Values were derived by ttest.
TABLE 4
Changes within and between randomized treatment arms in cognitive function test scores between baseline and 24 mo
1
Fish oil (n= 375) Placebo (n= 369)
Difference
(fish oil 2placebo)
Baseline 24 mo 24-mo change Baseline 24 mo 24-mo change
Unadjusted
mean difference
(95% CI) P
2
CVLT
Total words correct
in first 3 trials
24.1 66.0
3
24.1 66.7 20.02 64.7 23.9 65.7 24.4 66.4 0.5 65.0 20.5 (21.2, 0.2) 0.14
Words recalled at
delayed recall
7.5 62.8 7.6 63.2 0.1 62.4 7.5 62.9 7.5 63.0 20.01 62.3 0.1 (20.2, 0.4) 0.46
Global cognitive
function zscore
0.00 60.50 20.01 60.55 20.01 60.31 0.03 60.48 0.03 60.54 20.00 60.33 20.01 (20.05,0.04) 0.77
Memory zscore 0.02 60.70 0.02 60.79 0.01 60.51 20.00 60.71 0.01 60.72 0.01 60.51 20.00 (20.07,0.07) 0.97
Processing zscore 20.03 60.62 20.02 60.66 0.01 60.45 0.05 60.64 0.05 60.71 0.00 60.59 20.01 (20.08,0.06) 0.79
Executive function zscore 0.04 60.73 20.05 60.78 20.09 60.62 0.07 60.77 20.01 60.78 20.07 60.60 20.02 (20.11,0.06) 0.58
Global delay zscore 0.00 60.72 0.03 60.81 0.03 60.54 0.01 60.74 0.01 60.74 20.00 60.55 0.03 (20.05,0.10) 0.46
1
Includes only participants with cognitive function test scores at baseline and 24 mo. CVLT, California Verbal Learning Test.
2
Values were derived by ANCOVA (adjusted only for baseline cognitive function).
3
Mean 6SD (all such values).
1730 DANGOUR ET AL
at PONTIFICIA UNIV CATOLICA CHILE on May 6, 2013ajcn.nutrition.orgDownloaded from
We thank all participants in the OPAL study. We gratefully acknowledge
the important contribution of the following Medical Research Council Gen-
eral Practice Research Framework nurses and regional trainers: Rika Adams,
Ruth Bastick, Pauline Caspell, Phyllis Church, Anne Ellis, Jane Elwood, Sue
Fox, Kay Foulger, Sharon Hallam, Lesley Hand, Joanne Heslop, Angela Hill,
Christine Holliday, Jackie Houghton, Monica Lloyd, Eileen Marshall, Joan
McArdle, Anne MacDonald, Cynthia McVey, Jo Mead, Susanna Robinson,
Jennifer Rolt, and Anna Williams. We also thank steering committee external
members John Grimley Evans (chair), Alan Bird, Yvonne Davidson, Martin
Prince, Ursula Shine, and Madge Vickers (from January 2005); data monitor-
ing committee external members Tom Sanders (chair), Gill Livingston, and
Graham Dunn; and project management group members Alan Dangour
(chair), Elizabeth Allen (November 2006 to June 2009), Felicity Clemens
(April 2005 to October 2006), Korotimi Diallo (April 2004 to March
2008), Diana Elbourne, Nicky Fasey, Astrid Fletcher, Pollyanna Hardy (April
2004 to April 2005), Graham Holder, Rosemary Knight (April 2004 to August
2007), Louise Letley, Carole Maguire (January 2005 to July 2008), Debbie
Piercy (April 2005 to June 2009), Marcus Richards, Ann Truesdale (April
2004 to December 2007), Ricardo Uauy, and Madge Vickers (April 2004
to December 2004). Finally, we thank Mike Bennett for database and website
development, Robert Gray for conducting the serum analysis at King’s Col-
lege London, and Felicia Huppert for collaboration in the design of the cog-
nitive function test booklet.
The authors’ responsibilities were as follows—ADD and RU: conceived
the study; ADD, DE, AEF, GEH, MR, and RU: applied for the funding;
RK: acted as study coordinator; EA: conducted final statistical analyses;
ADD and EA: wrote the first draft of the manuscript; ADD: acted as guarantor
for the study; and all authors: involved in designing and implementing the
study, contributed to subsequent drafts of the manuscript, and approved
the final version of the manuscript. The funders had no role in the implemen-
tation, data collection, management, analysis, or interpretation of the study or
in the preparation, review, and approval of the manuscript. The corresponding
author (ADD) had full access to all data and had final responsibility for the
decision to submit the manuscript for publication. None of the authors de-
clared a conflict of interest.
REFERENCES
1. Personal Social Services Research Unit. Dementia UK. London, United
Kingdom: Alzheimer’s Society, 2007.
2. Ferri CP, Prince M, Brayne C, et al. Global prevalence of dementia:
a Delphi consensus study. Lancet 2005;366:2112–7.
3. Gillette Guyonnet S, Abellan Van Kan G, Andrieu S, et al. IANA task
force on nutrition and cognitive decline with aging. J Nutr Health Aging
2007;11:132–52.
4. Luchsinger JA, Mayeux R. Dietary factors and Alzheimer’s disease.
Lancet Neurol 2004;3:579–87.
5. Futterman S, Andrews JS. The fatty acid composition of human retinal
vitamin a ester and the lipids of human retinal tissue. Invest Ophthalmol
1964;3:441–4.
6. Svennerholm L. Distribution and fatty acid composition of phospho-
glycerides in normal human brain. J Lipid Res 1968;9:570–9.
7. Bazan NG. Cell survival matters: docosahexaenoic acid signaling,
neuroprotection and photoreceptors. Trends Neurosci 2006;29:263–71.
8. Uauy R, Dangour AD. Nutrition in brain development and aging: role of
essential fatty acids. Nutr Rev 2006;64:S24–33; discussion S72–91.
9. Ortega RM, Requejo AM, Andres P, et al. Dietary intake and cognitive
function in a group of elderly people. Am J Clin Nutr 1997;66:803–9.
10. Kalmijn S, van Boxtel MP, Ocke M, Verschuren WM, Kromhout D,
Launer LJ. Dietary intake of fatty acids and fish in relation to cognitive
performance at middle age. Neurology 2004;62:275–80.
11. Feart C, Samieri C, Rondeau V, et al. Adherence to a Mediterranean
diet, cognitive decline, and risk of dementia. JAMA 2009;302:638–48.
12. Barberger-Gateau P, Letenneur L, Deschamps V, Peres K, Dartigues JF,
Renaud S. Fish, meat, and risk of dementia: cohort study. BMJ 2002;
325:932–3.
13. Morris MC, Evans DA, Bienias JL, et al. Consumption of fish and n23
fatty acids and risk of incident Alzheimer disease. Arch Neurol 2003;60:
940–6.
14. Morris MC, Evans DA, Tangney CC, Bienias JL, Wilson RS. Fish
consumption and cognitive decline with age in a large community study.
Arch Neurol 2005;62:1849–53.
15. Huang TL, Zandi PP, Tucker KL, et al. Benefits of fatty fish on dementia
risk are stronger for those without APOE epsilon4. Neurology 2005;65:
1409–14.
TABLE 5
Changes within randomized treatment arms in secondary cognitive function test scores between baseline and 24 mo
1
Fish oil Placebo
nBaseline 24 mo nBaseline 24 mo
Memory
Story recall
Immediate (number of story items recalled) 374 11.1 63.9
2
11.0 64.3 367 10.7 63.9 10.9 63.9
Delayed (number of story items recalled) 374 8.9 63.8 9.3 64.2 366 8.8 63.7 9.1 63.8
Spatial memory (number of correct images)
Immediate (number of correct images) 373 5.3 62.6 5.3 62.5 368 5.3 62.5 5.2 62.5
Delayed (number of correct images) 373 4.6 62.5 4.4 62.4 368 4.6 62.6 4.4 62.5
Processing speed
Letter search/cancellation (number correct as
a percentage of total attempts)
374 78.8 612.6 79.2 612.5 369 78.9 613.3 79.2 612.8
Symbol letter modality (number correct) 369 43.0 69.6 41.7 610.4 369 44.3 69.2 42.7 610.0
Reaction time (s)
Simple 363 0.29 60.07 0.29 60.08 357 0.31 60.24 0.29 60.15
Choice 357 0.75 60.58 0.71 60.35 344 0.72 60.52 0.68 60.10
Executive function
Digit span
Forward (number of correct sequences) 364 8.3 62.4 7.9 62.3 355 8.2 62.3 8.0 62.4
Backward (number of correct sequences) 361 6.5 62.0 6.4 62.0 355 6.5 62.1 6.4 62.0
Verbal fluency (number of animals named) 374 19.8 65.1 19.1 65.4 369 19.9 65.0 19.5 65.3
1
Includes only participants with cognitive function test scores at baseline and 24 mo. ANCOVA with adjustment for baseline cognitive function test
scores, age, sex and age at leaving full-time education showed no significant differences between trial arms at 24 mo for any secondary cognitive function
outcome.
2
Mean 6SD (all such values).
FISH OIL AND COGNITIVE FUNCTION IN OLDER PEOPLE 1731
at PONTIFICIA UNIV CATOLICA CHILE on May 6, 2013ajcn.nutrition.orgDownloaded from
16. van Gelder BM, Tijhuis M, Kalmijn S, Kromhout D. Fish consumption,
n23 fatty acids, and subsequent 5-y cognitive decline in elderly men:
the Zutphen Elderly Study. Am J Clin Nutr 2007;85:1142–7.
17. Albanese E, Dangour AD, Uauy R, et al. Dietary fish and meat intake
and dementia in Latin America, China, and India: a 10/66 Dementia
Research Group population-based study. Am J Clin Nutr 2009;90:
392–400.
18. Laurin D, Verreault R, Lindsay J, Dewailly E, Holub BJ. Omega-3 fatty
acids and risk of cognitive impairment and dementia. J Alzheimers Dis
2003;5:315–22.
19. Engelhart MJ, Geerlings MI, Ruitenberg A, et al. Diet and risk of de-
mentia: does fat matter?: the Rotterdam Study. Neurology 2002;59:
1915–21.
20. Devore EE, Grodstein F, van Rooij FJ, et al. Dietary intake of fish and
omega-3 fatty acids in relation to long-term dementia risk. Am J Clin
Nutr 2009;90:170–6.
21. Kroger E, Verreault R, Carmichael PH, et al. Omega-3 fatty acids and
risk of dementia: the Canadian Study of Health and Aging. Am J Clin
Nutr 2009;90:184–92.
22. van de Rest O, Geleijnse JM, Kok FJ, et al. Effect of fish oil on cognitive
performance in older subjects: a randomized, controlled trial. Neurology
2008;71:430–8.
23. Lim WS, Gammack JK, Van Niekerk J, Dangour AD. Omega 3 fatty
acid for the prevention of dementia. Cochrane Database Syst Rev 2006;
CD005379.
24. Dangour AD, Clemens F, Elbourne D, et al. A randomised controlled
trial investigating the effect of n23 long-chain polyunsaturated fatty
acid supplementation on cognitive and retinal function in cognitively
healthy older people: the Older People And n23 Long-chain poly-
unsaturated fatty acids (OPAL) study protocol [ISRCTN72331636].
Nutr J 2006;5:20.
25. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical
method for grading the cognitive state of patients for the clinician. J
Psychiatr Res 1975;12:189–98.
26. Scientific Advisory Committee on Nutrition. Advice on fish consumption:
benefits and risks. London, United Kingdom: The Stationary Office, 2004.
27. Kris-Etherton PM, Harris WS, Appel LJ. Fish consumption, fish oil,
omega-3 fatty acids, and cardiovascular disease. Circulation 2002;106:
2747–57.
28. Department of Education and Science. Burnham further education
committee grading courses. London, United Kingdom: HMSO, 1972.
29. Dangour AD, Allen E, Elbourne D, Fletcher A, Richards M, Uauy R.
Fish consumption and cognitive function among older people in the UK:
baseline data from the OPAL study. J Nutr Health Aging 2009;13:
198–202.
30. Goldberg D. The detection of psychiatric illness by questionnaire.
London, United Kingdom: Oxford University Press, 1972.
31. Vellas B, Andrieu S, Sampaio C, Coley N, Wilcock G. Endpoints for
trials in Alzheimer’s disease: a European task force consensus. Lancet
Neurol 2008;7:436–50.
32. Delis DC, Kramer JH, Kaplan E, Ober BA. California Verbal Learning
Test. New York, NY: Psychological Corporation, 1987.
33. Tierney MC, Oh P, Moineddin R, et al. A randomized double-blind trial
of the effects of hormone therapy on delayed verbal recall in older
women. Psychoneuroendocrinology 2009;34:1065–74.
34. Durga J, van Boxtel MP, Schouten EG, et al. Effect of 3-year folic acid
supplementation on cognitive function in older adults in the FACIT trial:
a randomised, double blind, controlled trial. Lancet 2007;369:208–16.
35. Wechsler D. Wechsler memory scale—revised manual. San Antonio,
TX: Psychological Corporation, 1987.
36. van Niekerk JK, Nielsen TA, Pasu S, Shore S, Huppert FA. The de-
velopment of new tests of source memory and a new approach to the
testing of equivalence of parallel versions. Aging Neuropsychol Cogn
2004;11:416–27.
37. Richards M, Kuh D, Hardy R, Wadsworth M. Lifetime cognitive func-
tion and timing of the natural menopause. Neurology 1999;53:308–14.
38. Smith A. Symbol letter modalities test. Revised manual. Los Angeles,
CA: Western Psychological Services, 1982.
39. Cox BD, Huppert FA, Whichelow MJ. The Health and Lifestyle Survey:
seven years on. Aldershot, United Kingdom: Dartmouth Pub Co, 1993.
40. Wechsler D. Wechsler adult intelligence scale—revised. New York, NY:
The Psychological Corporation, 1981.
41. Goodglass H, Kaplan E. Assessment of dysphasia and related disorders.
Philadelphia, PA: Lea and Febiger, 1983.
42. Maylor EA. Aging and forgetting in prospective and retrospective
memory tasks. Psychol Aging 1993;8:420–8.
43. London School of Hygiene and Tropical Medicine. The OPAL study.
Homepage. Available from: http://opal-study.org.uk (cited 15 September
2009).
44. Connor SL, Zhu N, Anderson GJ, et al. Cheek cell phospholipids in
human infants: a marker of docosahexaenoic and arachidonic acids in
the diet, plasma, and red blood cells. Am J Clin Nutr 2000;71:21–7.
45. Hoffman DR, Birch EE, Birch DG, Uauy R. Fatty acid profile of buccal
cheek cell phospholipids as an index for dietary intake of docosahex-
aenoic acid in preterm infants. Lipids 1999;34:337–42.
46. Lepage G, Roy CC. Direct transesterification of all classes of lipids in
a one-step reaction. J Lipid Res 1986;27:114–20.
47. Samsa G, Edelman D, Rothman ML, Williams GR, Lipscomb J,
Matchar D. Determining clinically important differences in health status
measures: a general approach with illustration to the Health Utilities
Index Mark II. Pharmacoeconomics 1999;15:141–55.
48. Norman MA, Evans JD, Miller WS, Heaton RK. Demographically
corrected norms for the California Verbal Learning Test. J Clin Exp
Neuropsychol 2000;22:80–94.
49. Goldberg DP, Gater R, Sartorius N, et al. The validity of two versions of
the GHQ in the WHO study of mental illness in general health care.
Psychol Med 1997;27:191–7.
50. McMahon JA, Green TJ, Skeaff CM, Knight RG, Mann JI, Williams
SM. A controlled trial of homocysteine lowering and cognitive perfor-
mance. N Engl J Med 2006;354:2764–72.
51. Neuringer M, Connor WE, Lin DS, Barstad L, Luck S. Biochemical and
functional effects of prenatal and postnatal omega 3 fatty acid deficiency
on retina and brain in rhesus monkeys. Proc Natl Acad Sci USA 1986;
83:4021–5.
52. World Cancer Research Fund/American Institute for Cancer Research.
Food, nutrition, physical activity, and the prevention of cancer: a global
perspective. Washington, DC: American Institute for Cancer Research,
2007.
53. Hennekens CH, Demets D. The need for large-scale randomized evi-
dence without undue emphasis on small trials, meta-analyses, or sub-
group analyses. JAMA 2009;302:2361–2.
54. Freund-Levi Y, Eriksdotter-Jonhagen M, Cederholm T, et al. Omega-3
fatty acid treatment in 174 patients with mild to moderate Alzheimer
disease: OmegAD study: a randomized double-blind trial. Arch Neurol
2006;63:1402–8.
55. Gillette-Guyonnet S, Andrieu S, Dantoine T, Dartigues JF, Touchon J,
Vellas B. Commentary on “A roadmap for the prevention of dementia II.
Leon Thal Symposium 2008.” The Multidomain Alzheimer Preventive
Trial (MAPT): a new approach to the prevention of Alzheimer’s disease.
Alzheimers Dement 2009;5:114–21.
1732 DANGOUR ET AL
at PONTIFICIA UNIV CATOLICA CHILE on May 6, 2013ajcn.nutrition.orgDownloaded from