Customize your JAMA Network experience by selecting one or more topics from the list below.
Solomon PR, Adams F, Silver A, Zimmer J, DeVeaux R. Ginkgo for Memory Enhancement: A Randomized Controlled Trial. JAMA. 2002;288(7):835–840. doi:10.1001/jama.288.7.835
Author Affiliations: Department of Psychology (Dr Solomon and Ms Zimmer), Program in Neuroscience (Dr Solomon and Mss Adams, Silver, and Zimmer), Department of Mathematics and Statistics (Dr DeVeaux), Williams College, Williamstown, Mass; and The Memory Clinic, Southwestern Vermont Medical Center, Bennington (Dr Solomon).
Context Several over-the-counter treatments are marketed as having the ability
to improve memory, attention, and related cognitive functions in as little
as 4 weeks. These claims, however, are generally not supported by well-controlled
Objective To evaluate whether ginkgo, an over-the-counter agent marketed as enhancing
memory, improves memory in elderly adults as measured by objective neuropsychological
tests and subjective ratings.
Design Six-week randomized, double-blind, placebo-controlled, parallel-group
Setting and Participants Community-dwelling volunteer men (n = 98) and women (n = 132) older
than 60 years with Mini-Mental State Examination scores greater than 26 and
in generally good health were recruited by a US academic center via newspaper
advertisements and enrolled over a 26-month period from July 1996 to September
Intervention Participants were randomly assigned to receive ginkgo, 40 mg 3 times
per day (n = 115), or matching placebo (n = 115).
Main Outcome Measures Standardized neuropsychological tests of verbal and nonverbal learning
and memory, attention and concentration, naming and expressive language, participant
self-report on a memory questionnaire, and caregiver clinical global impression
of change as completed by a companion.
Results Two hundred three participants (88%) completed the protocol. Analysis
of the modified intent-to-treat population (all 219 participants returning
for evaluation) indicated that there were no significant differences between
treatment groups on any outcome measure. Analysis of the fully evaluable population
(the 203 who complied with treatment and returned for evaluation) also indicated
no significant differences for any outcome measure.
Conclusions The results of this 6-week study indicate that ginkgo did not facilitate
performance on standard neuropsychological tests of learning, memory, attention,
and concentration or naming and verbal fluency in elderly adults without cognitive
impairment. The ginkgo group also did not differ from the control group in
terms of self-reported memory function or global rating by spouses, friends,
and relatives. These data suggest that when taken following the manufacturer's
instructions, ginkgo provides no measurable benefit in memory or related cognitive
function to adults with healthy cognitive function.
Some over-the-counter treatments are marketed as having the ability
to improve memory, attention, and related cognitive functions. These claims
are generally not supported by well-controlled clinical studies. Ginkoba claims
to "enhance mental focus and improve memory and concentration."1
Several published studies reported beneficial effects of ginkgo on cognition.
These studies, however, either report cognitive improvement in only 1 of many
memory tests administered2,3 or
report cognitive enhancement in cognitively impaired clinical populations
such as patients with cerebrovascular or Alzheimer disease.4,5
In contrast, advertising claims imply that the compound is broadly beneficial
to those both with and without clinically significant cognitive impairments.
Specific advertising claims cite more than 50 clinical trials that demonstrate
benefit centered around concentration and memory. These studies were conducted
for periods ranging from 14 days to 2 months. The manufacturer claims benefit
with "at least 4 weeks of uninterrupted use."6
The purpose of the present study was to evaluate ginkgo in healthy elderly
volunteers in a randomized, double-blind, placebo-controlled trial using standardized
tests of memory, learning, attention and concentration, and expressive language
as well as subjective ratings by participants and family.
Following approval by the Williams College institutional review board,
participants were recruited from newspaper advertisements that solicited individuals
who would participate in a study designed to improve memory. An initial telephone
interview was conducted to determine if the participant was likely to meet
entry criteria for the study. Those who passed the screen provided informed
consent and a medical history including current medications, neurologic or
psychiatric illness, and incidence of head trauma, stroke, mental illness,
mental retardation, or life-threatening illness over the last 5 years. Participants
were included in the study if they were community dwelling, older than 60
years, and could provide informed consent. They also needed to have a companion
who had contact with them on a regular basis (>4 times per week for ≥1
hour) and was willing to complete a questionnaire. The baseline Mini-Mental
State Examination7 score was required to be
greater than 26. All participants reported to be independent in instrumental
activities of daily living including shopping, transportation, and managing
finances. Participants were excluded if they had a history of psychiatric
or neurologic disorder or had a life-threatening illness in the last 5 years.
They were also excluded if they had taken antidepressant or other psychoactive
medications in the past 60 days. A total of 338 community-dwelling participants
were screened over a 26-month period from July 1996 to September 1998, and
230 participants (98 men and 132 women) aged 60 to 82 years were randomized
in the study.
A 6-week double-blind placebo-controlled study was conducted at a single
site. Figure 1 summarizes the study
participation. Participants were randomly assigned to 1 of 2 conditions: ginkgo
(Ginkoba, Boehringer Ingelheim Pharmaceuticals)1
or placebo control (1:1 ratio). Random assignment of participants to each
condition was determined by 1 of the investigators (P.R.S.) using a table
of random numbers.8 Medication was placed in
sealed envelopes by a research assistant and provided to the participants
by 1 of 3 other investigators (F.A., A.S., J.Z.). Dosages for ginkgo were
determined by following the manufacturer's label instructions: 1 tablet (40
mg) 3 times a day, with meals. The placebo group took lactose gelatin capsules
of similar appearance and on the same schedule as the ginkgo group. At the
beginning of the double-blind period, participants were provided with sealed
and dated envelopes, each containing medication for 1 day.
One day prior to taking ginkgo or placebo and again at the end of the
6-week double-blind period (while still taking ginkgo and within 3 days of
the end of the study), participants underwent neuropsychological evaluation
including tests of learning, memory, attention and concentration, and expressive
language. They also completed a questionnaire regarding subjective impressions
of their memory. Additionally, at the end of the 6 weeks of treatment, the
companion was asked to complete a global questionnaire designed to provide
an overall impression of change in memory for the participant. Evaluators
(F.A., A.S., J.Z.) were blinded to which randomized treatment the participants
Participants were contacted by telephone twice (at the end of weeks
2 and 4) during the 6-week period to evaluate compliance. They were excluded
from the study if they missed 6 doses in any 2-week period or did not take
3 consecutive doses. At this time, they were asked to stop taking study medication.
As an additional measure of compliance, participants were asked to return
all dated envelopes at the end of the study.
Outcome measures consisted of the following standardized tests of learning,
memory, attention and concentration, expressive language, and mental status.
Tests of learning and memory included the California Verbal Learning Test
(CVLT),9 in which the participant is asked
to learn a 16-item shopping list over 5 trials and then to later recall and
subsequently recognize the information; the Logical Memory subscale of the
Wechsler Memory Scale–Revised (WMS-R),10
in which the participant is asked to recall paragraphs both immediately after
hearing them and then after a 30-minute delay; and the Visual Reproduction
subscale, in which the participant is asked to draw designs both immediately
after seeing them and after a 30-minute delay.
Tests of attention and concentration included the Digit Symbol subscale
of the Wechsler Adult Intelligence Scale–Revised (WAIS-R),11
in which the participant must rapidly copy symbols that are paired with numbers;
the Stroop Test,12 which requires the participant
not to be distracted by extraneous aspects of stimuli; the Digit Span (WMS-R),
which requires the participant to repeat increasingly longer strings of numbers
immediately after hearing them; and Mental Control (WMS-R), in which the participant
must recite strings of numbers and letters.
Tests of expressive language included the Controlled Category Fluency
test,12 which requires the participant to name
members of a particular category (animals) over a 1-minute period; and the
Boston Naming Test,13 which requires the participant
to name pictures of items.
Additionally, the Memory Questionnaire14
as well as a global evaluation completed by a spouse, relative, or friend
with whom the patient had regular contact (at least 4 interactions per week)
was completed. The Memory Questionnaire consisted of 27 questions that asked
the participant to rate how often certain memory lapses occurred. The participant
answered on a 4-point scale with descriptors used as anchors: 1 indicating
very often, 2 indicating sometimes, 3 indicating rarely, and 4 indicating
not at all. The global evaluation was based on the Caregiver Global Impression
of Change rating scale.15 Informants were asked
to indicate the option that best described the change in memory over the preceding
6 weeks. The options included: (1) very much improved, (2) much improved,
(3) minimally improved, (4) no change, (5) minimally worse, (6) much worse,
or (7) very much worse.
All outcome measures, with the exception of the global evaluation, were
administered at both the beginning and end of the study. The global evaluation
was administered only at the end of the study. Participants who withdrew from
the study, or who were dropped because of noncompliance, were asked to return
at the end of the study for evaluation. Adverse events were not specifically
monitored in this study. Patients who experienced an adverse event were instructed
to discontinue study medication and to contact their primary care physician.
Analysis for efficacy was performed on 2 participant samples: the modified
intent-to-treat primary analysis and the fully evaluable population. The modified
intent-to-treat population included all participants who were randomized to
treatment, underwent baseline analysis, received at least 1 dose of study
drug, and returned for posttreatment evaluation. The fully evaluable population
was defined as participants who completed 6 weeks of double-blind treatment
and who complied with the standards for taking medication.
Differences in group means for all neuropsychological tests were assessed
using both individual t tests and repeated-measures
analysis of variance in which treatment condition served as the predictor
and the cognitive tests served as the repeated measures. The test by condition-interaction
term was then tested for statistical significance. Demographic variables were
analyzed using the individual t tests. Categorical
variables were analyzed using the χ2 test. Results were considered
statistically significant if differences reached the .05 level. Nonparametric
analyses were used to assess the changes from baseline to week 6 for the Caregiver
Global Impression of Change. We sought to detect differences of .05 SD with
a power of 90% (α = .05), requiring a sample size of 172 participants.16 JMP version 5.0 (SAS Institute Inc, Cary, NC) statistical
software was used for all analyses.
A total of 230 participants were enrolled in the study over a 26-month
period, with 203 participants (88%) completing the study (Figure 1). The percentage of participants who completed the study
did not differ significantly by treatment group. Of the 27 participants who
did not complete the study, 16 (7 ginkgo and 9 placebo) did not comply with
the medication dosage regimen and 11 (4 ginkgo and 7 placebo) withdrew consent.
All participants were requested to return at the end of week 6 for evaluation.
A total of 219 participants (111 ginkgo and 108 placebo) returned at
the end of the 6-week period for reevaluation. This included the 203 participants
who completed the protocol as well as 13 of 16 participants (6 ginkgo and
7 placebo) who were noncompliant and 3 of the 11 participants (2 ginkgo and
1 placebo) who withdrew consent. The remaining 11 participants (4 ginkgo and
7 placebo) did not return for evaluation and were excluded from the analysis.
There were no significant differences between the ginkgo and placebo groups
for any of the outcome measures. Neither demographic characteristics nor Mini-Mental
State Examination scores varied as a function of treatment condition at baseline
There were no significant differences between the ginkgo and placebo
groups on any of the objective neuropsychological tests. In general, participants
performed better during their second evaluation than during their first, but
there were no significant test-by-treatment condition interactions as tested
by a repeated-measures analysis of variance (F14,172 = 0.099, overall P = .31). Superior performance in all groups at the second
testing session was likely due to a practice effect.
When tested by individual t tests, measures
of attention and concentration, including the Digit Symbol subscale of the
WAIS-R, the Stroop Test, and the Mental Control and Digit Span (forward and
backward) subscales of the WMS-R, showed no significant differences between
the ginkgo and placebo groups (Table 2
and Figure 2). Similarly, tests
of verbal and nonverbal learning and memory, including the Logical Memory
(I and II) and Visual Reproduction (I and II) subscales of the WMS-R, and
the CVLT (initial acquisition, short and long delay, and recognition), also
showed no significant differences between the ginkgo and placebo groups. There
were no differences in tests of naming (Boston Naming Test) or verbal fluency
(Controlled Category Fluency) between the ginkgo and placebo groups. Finally,
self-report on the Memory Questionnaire was scored on a scale of 27 to 108
with higher scores indicating more difficulties. There was no difference in
the mean reported scores for participants in the ginkgo and placebo groups
(P = .26).
At the end of the second testing session, participants were asked if
they thought they had been taking ginkgo or placebo. Self-report in the ginkgo
group indicated that 79 participants (71%) thought they were takingT ginkgo,
and self-report in the placebo group indicated that 81 participants (75%)
thought they were taking ginkgo (P = .49). Informant
response to the global rating indicated no difference between the ginkgo and
placebo groups (P = .76). Table 3 shows the distribution of responses.
Figure 2 shows the 95% confidence
intervals (CIs) for differences (treatment group minus control) for performance
on each test in the modified intent-to-treat analysis. Each interval contains
a zero, indicating that none of the differences are statistically significant.
Moreover, 7 of the point estimates are positive (favoring ginkgo) and 7 are
negative (favoring placebo).
A total of 203 participants completed the protocol (fully evaluable
population). There were no significant differences between the ginkgo and
placebo groups for any outcome measure (Table 2).
The results of this 6-week study indicate that ginkgo, marketed over-the-counter
as a memory enhancer, did not enhance performance on standard neuropsychological
tests of learning, memory, naming and verbal fluency, or attention and concentration.
Moreover, there were no differences between ginkgo participants and placebo
controls on subjective self-report of memory function or on global rating
by spouses, friends, and relatives. These data suggest that when taken following
the manufacturer's instructions, this compound provides no measurable benefit
in cognitive function to elderly adults with intact cognitive function.
In total, 14 different measures of cognition were evaluated in the present
study. Seven of the measures were better in the placebo group, and 7 of the
measures were better in the ginkgo group. None of the differences between
the means of the 2 groups were statistically significant. The 95% CIs were
calculated for each mean difference. Even if one assumes that the true difference
between treatments is the upper limit of the 95% CI, it would still be difficult
to argue that meaningful benefit was derived from taking ginkgo. For example,
the Logical Memory portion of the WMS-R measures the participants' ability
to recall 2 paragraphs that they initially heard 30 minutes earlier. There
are 25 possible discrete items in each paragraph that the participant could
recall. The upper limit of the 95% CI for the mean difference between ginkgo
and placebo was 0.20 items (ie, participants in the ginkgo group remembered
less than 1 item more than participants in the placebo group). Similarly,
on the CVLT, participants learn a 16-item shopping list over 5 trials. A perfect
score is 80. The upper limit of the 95% CI for the mean difference between
ginkgo and placebo was 1.01 items. It would be difficult to argue that either
of these differences are of any clinical significance, even if they are real.
The results of the Caregiver Global Impression of Change rating scale further
support the failure of ginkgo to provide clinically significant improvement
in memory. In general, caregivers did not rate changes in memory over the
6-week trial any differently in participants randomized to ginkgo vs placebo
participants. Sixty-six percent of those randomized to placebo and 70% to
ginkgo were judged by caregivers as showing no change over 6 weeks. Thirty-three
percent of placebo and 28% of ginkgo participants were judged as minimally
improved, and 3 participants were judged to be much improved; 2 were in the
ginkgo group and 1 was in the placebo group (Table 3).
Ginkgo has been evaluated in several double-blind studies that have
reported beneficial effects, but these effects were not broad or consistent.
Wesnes et al3 conducted a 3-month double-blind,
randomized, placebo-controlled study in 54 patients. Patients were evaluated
at weeks 4, 8, and 12. Patients receiving Tanakan (ginkgo extract) performed
better on only 2 of 8 tests of memory (P = .03) and
attention and concentration (P = .05) and in each
case at only 1 evaluation point. There was not a consistent effect for any
outcome measure. Additionally, neither physicians nor patients could distinguish
between placebo and compound on an overall scale. Rai et al2
compared 12 ginkgo-treated with 15 placebo-treated participants who were classified
as having mild to moderate memory impairment in a double-blind study and reported
significant differences in favor of the gingko group only on the Kendrick
Digit Copying task, but not on tests of learning or memory. Rigney et al17 evaluated 31 participants and 4 doses of ginkgo in
a crossover design. They only reported improvement with 1 dose of ginkgo (120
mg), in only the oldest group of participants (50-59 years), and only in 1
of the multiple tests of memory administered. Other studies that have reported
positive effects in favor of ginkgo have also either studied small numbers
of participants in uncontrolled studies,18,19
have found benefit in one of many cognitive tasks administered,20
or have found changes in objective tests relative to controls but not in physician
ratings in clinical populations.4,5
Despite the manufacturer's claims of improved memory in healthy adults, we
were unable to identify any well-controlled studies that document this claim.
Recently, ginkgo was reported to be beneficial in a sample of patients
with dementia.4 Mildly to severely demented
patients characterized as having either Alzheimer disease or multi-infarct
dementia were given either ginkgo (120 mg/d) or placebo for 52 weeks in a
randomized double-blind study. The intent-to-treat analysis on 202 patients
indicated a 0.1-point decline on the Alzheimer Disease Assessment Scale–Cognitive
portion (ADAS-Cog) in the ginkgo group compared with a 1.48-point decline
in the placebo group. No subjective differences were reported by either family
members or physicians. While provocative, these differences on the ADAS-Cog
are significantly smaller than those reported for approved cholinesterase
inhibitors in treating patients with Alzheimer disease.15
Moreover, the failure to find any differences in either physician or family
rating raises the issue of whether the small difference on the ADAS-Cog is
Despite the paucity of well-controlled studies, ginkgo continues to
be marketed and widely used.21,22
Sales in the United States reached $240 million in 199723
and more than 5 million prescriptions are written each year in Germany primarily
for dementia, cerebral decline, and peripheral arterial insufficiency.18
Our study has limitations. It is certainly possible that higher doses
or longer periods of exposure than used in this study are necessary to detect
changes; however, we administered the compound following the manufacturer's
instructions. The manufacturer's label indicates that ginkgo should be administered
at a dose of 120 mg/d and that doses of greater than 120 mg show no additional
benefit.6 This is also the dose suggested by
the German Commission E.24 The daily dose in
the present study was 120 mg/d. The label also states that a noticeable benefit
should be apparent after 4 weeks of usage. The present study evaluated cognition
after a 6-week interval. Moreover, there was no indication of a statistical
trend toward significance for any of the compounds on any of the measures.
Nevertheless, it is possible that longer exposures could produce beneficial
We did not monitor adverse effects in the present study. Although ginkgo
is generally characterized as a benign compound,21
it is not without adverse effects. Reported adverse effects include bleeding,
mild gastrointestinal upset, and headache.25
None of the participants in the present study discontinued treatment due to
adverse effects and none spontaneously reported any adverse effects. This
finding is generally consistent with studies that did systematically monitor
The issue of quality control has also been raised as a potential source
of variance in studies using over-the-counter compounds.26
One limitation of the present study is that we did not analyze the content
of the ginkgo used in this study. However, the manufacturer claims that ginkgo
"is processed under strict guidelines . . . ensured through extensive quality
We recognize the possibility that ceiling effects may have contributed
to the nonsignificant findings in the present study. However, we selected
tests that are normalized for the age group that we studied and, as such,
have an appropriate range of scores. For example, in the Logical Memory WMS-R
scale (Logical Memory I), the potential range of scores is 0 to 50. The ginkgo
participants in the present study scored a mean of 20.49 (SD, 5.08) and the
placebo participants scored a mean of 23.61 (SD, 4.65). Each of these is well
below the maximum score of 50. In addition, none of the participants obtained
a maximum score on this scale or any of the other scales used in this study.
We also recognize that the method of blinding in this study could have
resulted in unblinding for some participants. However, the finding that participants
taking ginkgo as well as those taking placebo reported in equal proportions
taking the active compound ginkgo (71% vs 75%) mitigates this concern.
In summary, this study does not support the manufacturer's claims of
the benefits of gingko on learning and memory. Treatment over a 6-week period
following the manufacturer's dosing suggestions did not produce objective
benefit on any of 14 standard neuropsychological tests, nor were any benefits
detected in self-report by the participants or observation by a family member
Create a personal account or sign in to: