ORIGINAL INVESTIGATION
Prebiotic intake reduces the waking cortisol response
and alters emotional bias in healthy volunteers
Kristin Schmidt & Philip J. Cowen & Catherine J. Harmer &
George Tzortzis & Steven Errington & Philip W. J. Burnet
Received: 23 July 2014 /Accepted: 10 November 2014 /Published online: 3 December 2014
# The Author(s) 2014. This article is published with open access at Springerlink.com
Abstract
Rationale There is now compelling evidence for a link be-
tween enteric microbiota and brain function. The ingestion of
probiotics modulates the processing of information that is
strongly linked to anxiety and depression, and influences the
neuroendocrine stress response. We have recently demonstrat-
ed that prebiotics (soluble fibres that augment the growth of
indigenous microbiota) have significant neurobiological ef-
fects in rats, but their action in humans has not been reported.
Objectives The present study explored the effects of two
prebiotics on the secretion of the stress hormone, cortisol
and emotional processing in healthy volunteers.
Methods Forty-five healthy volunteers received one of two
prebiotics (fructooligosaccharides, FOS, or Bimuno®-galacto-
oligosaccharides, B-GOS) or a placebo (maltodextrin) daily for
3 weeks. The salivary cortisol awakening response was sam-
pled before and after prebiotic/placebo administration. On the
final day of treatment, participants completed a computerised
task battery assessing the processing of emotionally salient
information.
Results The salivary cortisol awakening response was signif-
icantly lower after B-GOS intake compared with placebo.
Participants also showed decreased attentional vigilance to
negative versus positive information in a dot-probe task after
B-GOS compared to placebo intake. No effects were found
after the administration of FOS.
Conclusion The suppression of the neuroendocrine stress
response and the increase in the processing of positive
versus negative attentional vigilance in subjects supple-
mented with B-GOS are consistent with previous find-
ings of endocrine and anxiolytic effects of microbiota
proliferation. Further studies are therefore needed to test
the utility of B-GOS supplementation in the treatment
of stress-related disorders.
Keywords Cortisol . Hypothalamic-pituitary-adrenal axis .
Gut microbiota . Prebiotics . Anxiety . Attention . Emotional
processing
Introduction
The adult human gut microbiota comprises over 1000
species and 7000 bacterial strains and is characterised
by a balanced compositional signature with moderate
inter-individual variability (Gareau et al. 2010; Cryan
and Dinan 2012). Probiotic strains, which have the
ability to confer beneficial effects upon the host, have
received renewed attention in recent years (e.g. Forsythe
and Kunze 2013). A particular focus has been put on
their ability to influence neural and endocrine systems
and behavioural phenotypes (Cryan and O’Mahony
2011; Dinan and Cryan 2012). Their potentia.
ORIGINAL INVESTIGATIONPrebiotic intake reduces the waking .docx
1. ORIGINAL INVESTIGATION
Prebiotic intake reduces the waking cortisol response
and alters emotional bias in healthy volunteers
Kristin Schmidt & Philip J. Cowen & Catherine J. Harmer &
George Tzortzis & Steven Errington & Philip W. J. Burnet
Received: 23 July 2014 /Accepted: 10 November 2014
/Published online: 3 December 2014
# The Author(s) 2014. This article is published with open access
at Springerlink.com
Abstract
Rationale There is now compelling evidence for a link be-
tween enteric microbiota and brain function. The ingestion of
probiotics modulates the processing of information that is
strongly linked to anxiety and depression, and influences the
neuroendocrine stress response. We have recently demonstrat-
ed that prebiotics (soluble fibres that augment the growth of
indigenous microbiota) have significant neurobiological ef-
fects in rats, but their action in humans has not been reported.
Objectives The present study explored the effects of two
prebiotics on the secretion of the stress hormone, cortisol
and emotional processing in healthy volunteers.
Methods Forty-five healthy volunteers received one of two
prebiotics (fructooligosaccharides, FOS, or Bimuno®-galacto-
oligosaccharides, B-GOS) or a placebo (maltodextrin) daily for
3 weeks. The salivary cortisol awakening response was sam-
pled before and after prebiotic/placebo administration. On the
final day of treatment, participants completed a computerised
task battery assessing the processing of emotionally salient
2. information.
Results The salivary cortisol awakening response was signif-
icantly lower after B-GOS intake compared with placebo.
Participants also showed decreased attentional vigilance to
negative versus positive information in a dot-probe task after
B-GOS compared to placebo intake. No effects were found
after the administration of FOS.
Conclusion The suppression of the neuroendocrine stress
response and the increase in the processing of positive
versus negative attentional vigilance in subjects supple-
mented with B-GOS are consistent with previous find-
ings of endocrine and anxiolytic effects of microbiota
proliferation. Further studies are therefore needed to test
the utility of B-GOS supplementation in the treatment
of stress-related disorders.
Keywords Cortisol . Hypothalamic-pituitary-adrenal axis .
Gut microbiota . Prebiotics . Anxiety . Attention . Emotional
processing
Introduction
The adult human gut microbiota comprises over 1000
species and 7000 bacterial strains and is characterised
by a balanced compositional signature with moderate
inter-individual variability (Gareau et al. 2010; Cryan
and Dinan 2012). Probiotic strains, which have the
ability to confer beneficial effects upon the host, have
received renewed attention in recent years (e.g. Forsythe
and Kunze 2013). A particular focus has been put on
their ability to influence neural and endocrine systems
and behavioural phenotypes (Cryan and O’Mahony
2011; Dinan and Cryan 2012). Their potential influence
on the mechanisms underlying stress-related disorders
3. such as irritable bowel syndrome (IBS), anxiety and
depression is also beginning to be elucidated (Dinan
et al. 2006; Rhee et al. 2009; Mayer 2011; Bravo
et al. 2012). We have recently demonstrated in rats that
prebiotics—oligosaccharides that promote the growth of
indigenous beneficial gut bacteria such as Lactobacilli
and Bifidobacteria—also have neurotropic effects
K. Schmidt: P. J. Cowen: C. J. Harmer: P. W. J. Burnet (*)
Department of Psychiatry, University of Oxford, Warneford
Hospital,
Oxford OX3 7JX, UK
e-mail: [email protected]
G. Tzortzis
Clasado Research Services Ltd, Reading RG6 6BZ, UK
S. Errington
Institute for Ageing and Health, Newcastle University, Campus
for
Ageing and Vitality, Newcastle upon Tyne NE4 5PL, UK
Psychopharmacology (2015) 232:1793–1801
DOI 10.1007/s00213-014-3810-0
(Savignac et al. 2013), but the central actions of these
compounds in humans have not been reported.
Convincing evidence now exists for a role of the gut
microbiota composition in the regulation of the stress
hormone corticosterone (cortisol in humans). Raised
levels of circulating corticosterone in germ-free rodents
(Crumeyrolle-Arias et al. 2014) are reduced following
the administration of probiotics (Sudo et al. 2004), an
effect replicated in mice subjected to a stress-inducing
4. behavioural paradigm designed to elevate corticosterone
levels (Bravo et al. 2011). However, there is a need to
further clarify the mechanisms involved in the complex
bidirectional relationship between the stress response and
the gut microbiota (Gareau et al. 2007; Dinan and Cryan
2012).
Most indications of a ‘microbiota-gut-brain axis’ in
humans have come from patients with gastrointestinal
disorders (Brenner et al. 2009; Gareau et al. 2010; Ken-
nedy et al. 2012). Additionally, there is now preliminary
evidence for reduced subjective feelings of anxiety and
improved aspects of well-being after probiotic intake (Rao
et al. 2009; Messaoudi et al. 2011). More recently, a
functional MRI investigation found that healthy subjects
who received a fermented milk product with probiotics
showed decreased BOLD activity to an emotional atten-
tion task using facial expressions in the insula and so-
matosensory regions (Tillisch et al. 2013). These areas
play a crucial role in the integration of visceral inputs
and the processing of emotional and interoceptive infor-
mation (Craig 2009). The study demonstrates that manip-
ulations of the gut microbiota can result in measurable
changes in emotional processing in the healthy brain.
Neural and behavioural biases in the processing of emo-
tional information, in particular increased processing of threat-
related and negatively valenced stimuli, are core functional
markers of anxiety and depression (Beck 2008; Cisler and
Koster 2010). These markers are evident in symptomatic
patients (Sheline et al. 2001) as well as high-risk (Chan et al.
2007) and remitted groups (Bhagwagar and Cowen 2007),
and are essential to our understanding of disease symptom-
atology and treatment efficacy (Harmer et al. 2009; Elliott
et al. 2011). Notably, the extent to which these biases can be
modulated by pharmacological therapies in patients has been
5. found to be indicative of treatment response (Sheline et al.
2001; Pizzagalli 2010), and assessing novel compounds on
their ability to target emotional biases may thus provide a first
line of assessing potential clinical utility.
Our study explored the effects of two commercially
available prebiotics (fructooligosaccharides [FOS] and
Bimuno®-galactooligosaccharides [B-GOS]) on the pro-
cessing of emotional information and hypothalamic-
pituitary-adrenal (HPA) axis activity in healthy human
volunteers.
Materials and methods
Participants
Forty-five volunteers (22 males, 23 females) recruited through
online and poster adverts completed the study. Inclusion
criteria were aged 18–45 years, fluent English speaker and
BMI range 18–25. Exclusion criteria were previous or current
neurological, psychiatric, gastrointestinal or endocrine disor-
ders, or other relevant medical history; current or recent
(<3 months) regular medication use; previous or current
substance/alcohol dependence or abuse within the last
3 months; regular tobacco use (>5 cigarettes/day); and partic-
ipation in research studies involving medication intake (within
3 months) or prior completion of the Emotional Test Battery
(ETB). To ensure that the enteric environment was consistent
in all volunteers, additional exclusion criteria were: no antibi-
otic use 3 months prior to the study, no regular use of pre- and
probiotics (and within 3 months prior to the study) and no
vegan diets. Finally, participants were asked to adhere to their
regular diets and avoid supplements or special diets. No
significant dietary variations were noted.
Participants were assessed with the Structured Clinical
6. Interview for DSM-IV (SCID; First et al. 1997) to confirm
the absence of DSM-IV axis I psychiatric conditions. The
study was approved by the Oxford Central University Re-
search Ethics Committee. All participants provided written
informed consent and were reimbursed for their time and
expenses.
Materials
Demographic and questionnaire measures
Participants completed the National Adult Reading Test
(NART; Nelson 1982) to provide an estimate of verbal IQ.
Self-report questionnaires assessing trait measures of person-
ality (Eysenck Personality Questionnaire, EPQ; Eysenck and
Eysenck 1975), stress responsivity (Perceived Stress Reactiv-
ity Scale, PSRS; Schlotz et al. 2011), subclinical symptoms of
depression (Beck Depression Inventory, BDI; Beck et al.
1961) and anxiety (State-Trait Anxiety Inventory, STAI-trait;
Spielberger et al. 1970) were also completed. Before and after
prebiotic/placebo intake (days 0 and 21, respectively), anxiety
(STAI-state) and measures of perceived stress (Perceived
Stress Scale, PSS; Cohen et al. 1983) and mood (Visual
Analogue Scales, VAS; Bond and Lader 1974; Positive and
Negative Affect Schedules, PANAS; Watson et al. 1988) were
measured using self-report questionnaires. The digit span
index of verbal working memory was used in order to monitor
group differences in executive functioning on the final day of
treatment.
1794 Psychopharmacology (2015) 232:1793–1801
Prebiotic supplements
7. The study was placebo controlled, and male/female partici-
pants were randomised to receive one of two prebiotics (fruc-
tooligosaccharides [N=15; 8 males, 7 females] or Bimuno®-
galactooligosaccharides [N=15; 7 males, 8 females]) or a
placebo (maltodextrin [N=15; 7 males, 8 females]). The use
of maltodextrin as a placebo compound in prebiotic trials is
well established (Vulevic et al. 2008). Preparations were pro-
vided by Clasado Research Services Ltd., Reading, UK. Par-
ticipants took the supplements (at 5.5 g per day) in powder
form orally with breakfast for 3 weeks. The study used a
double-blind randomised design with both the participant
and the experimenter being unaware of the group they had
been allocated to.
Salivary cortisol
HPA axis activity was assessed on the day before (day 0) and
on the final day of prebiotic/placebo administration (day 21),
using the salivary cortisol awakening response (CAR;
Pruessner et al. 1997). For each CAR measurement, partici-
pants were instructed to provide five saliva samples (using
Salivettes, Sarstedt Ltd., Nümbrecht, Germany) taken in their
own home immediately upon waking and subsequently every
15 min until 1 h post-waking. Saliva samples were stored at
4 °C prior to analysis. Cortisol was measured using a com-
mercial ELISA (Salimetrics Europe Ltd., Newmarket, UK),
within 7 days of sample collection.
Emotional processing tasks
On the final day of prebiotic/placebo intake (day 21), partic-
ipants completed a validated computerised test battery
assessing the processing of emotional stimuli (the ETB;
Harmer et al. 2004).
Attentional dot-probe task
8. Sixty negative and 60 positive words were paired with
neutral words matched for length. On each trial, a fixation
cross was presented for 500 ms in the centre of the screen,
followed by two words presented at the top and bottom of
the screen. In the unmasked condition, the words were
presented for 500 ms. In the masked condition, word pairs
were presented for 17 ms after which a mask was
displayed for 483 ms. Masks were constructed from
digits, letters and non-letter symbols and were matched
for word position and length. Words or masks were re-
placed by a probe of either one or two stars in the location
of one of the preceding stimuli (probes were presented at
the top or bottom of the screen with equal frequency).
Participants were instructed to indicate the number of
stars as quickly and accurately as possible using two
labelled keys. A key press terminated the probe presenta-
tion and trial. There were 180 trials in total (30 positive-
neutral, 30 negative-neutral, 30 neutral-neutral word pairs
each for masked and unmasked conditions), and emotion-
al words were presented at the top and the bottom location
with equal frequency. Masked and unmasked trials were
presented in random order. Reaction time and accuracy
scores were recorded, and attentional vigilance scores
were calculated for each participant by subtracting the
reaction time from trials when probes appeared in the
same position as the emotional word (congruent trials)
from those trials when probes appeared in the opposite
position to the emotional word (incongruent trials).
Facial expression recognition task
In the facial expression recognition task (FERT), the percep-
tion of six basic emotions (happiness, surprise, sadness, fear,
anger, disgust) or a neutral expression (taken from the Pictures
9. of Affect Series; Ekman and Friesen 1976) was assessed. Each
emotion was shown at 10 morphed intensity levels from
neutral to maximum emotional expression (Young et al.
1997) leading to a total of 250 randomly presented stimuli.
Each stimulus was presented for 500 ms and replaced by a
grey screen until the participant responded (as quickly and as
accurately as possible) by selecting a key corresponding to
one of the basic emotions. The outcome measures were clas-
sification accuracy, number of misclassifications and reaction
times.
Emotional categorisation and memory
Sixty words representing either disagreeable (N=30) or
agreeable (N=30) personality characteristics (from
Anderson 1968) and matched for meaningfulness, word
frequency and word length were presented on a computer
screen for 500 ms each. Participants were asked to cate-
gorise each word as quickly and accurately as possible
according to whether they would like or dislike to be
described by it. After completion, participants were
instructed to recall and write down as many words as they
could within a 2-min time limit. Subsequently, partici-
pants were instructed to categorise words presented on
the screen into those which were previously presented
(60 target words) or those which were novel words (60
matched distracter words). Outcome measures for the
emotional categorisation were classifications and reaction
times. For the memory recall, the number of correct
responses and false positives was recorded, and partici-
pants’ memory recognition was assessed using correct
responses, false positives and reaction times.
Psychopharmacology (2015) 232:1793–1801 1795
10. Statistical analysis
Demographic and questionnaire values were analysed using
one-way ANOVA with group as factor. Salivary cortisol
values (in nmol/l) were square root transformed and analysed
in a mixed design ANOVAwith time point of sampling (0, 15,
30, 45 and 60 min after waking) and day of sampling (pre- vs.
post-treatment) as repeated-measures variables and prebiotic
treatment group (placebo, FOS or B-GOS) as a between-
subjects variable. Raw cortisol values are presented for clarity.
The behavioural outcome variables of the ETB were also
analysed with mixed design ANOVAs, with prebiotic treat-
ment group as between-subjects factor and emotion and task
condition as within-subjects factors. Significant interactions
were followed up using main effects analyses. Greenhouse-
Geisser corrections were used where assumptions of spheric-
ity were not met.
Results
Baseline measures and compliance
There were no significant differences between groups for age,
trait measures of anxiety, stress reactivity, neuroticism or
cognitive status as assessed by digit span (see Table 1). This
suggests that groups were well matched between the two
prebiotic and placebo conditions.
Out of 48 participants, three participants did not complete
the full course of prebiotic/placebo intake and were excluded
from all analyses, leading to the final sample of 45 partici-
pants. Participants completed a checklist each day to monitor
prebiotic/placebo intake, and none reported missing more than
two intakes (0 missed=41, 1 missed=3, 2 missed=1).
11. Hormonal contraceptive use and menstrual cycle
Of the 23 female participants, 13 used hormonal methods of
contraception. Menstrual cycle phase was reported by 16
females, and an additional two female participants reported
no or very infrequent menses due to hormonal contraceptive
use—these were coded as a separate group. Cycle phase was
analysed according to the following phases (based on aver-
ages from Wolfram et al. 2011; Fehring et al. 2006): menses
(days 1–6), follicular phase (days 7–12), ovulation phase
(days 13–19), luteal phase (day 20–end of cycle). In order to
test the between-subjects effects of contraceptive use on the
waking cortisol response, repeated-measures ANOVAs were
performed on the cortisol level upon waking (first sample) and
the cortisol area under the curve with respect to ground (AUC-
g) including the factor supplement group to test for potential
interactions.
There was no significant difference in the number of
females who took hormonal contraceptives between the
groups (χ2=3.45, p>0.1). The effect of hormonal con-
traceptive on the first sample of salivary cortisol after
waking showed a trend for lower levels in females
taking contraceptives compared to those who were not
(mean (SD)=8.20 (3.42) vs. mean (SD)=10.65 (4.37),
F(1,17)=3.74, p=0.07); however, this difference was
stable across days of testing and treatment groups (all
interactions’ p values >0.7). The effect of hormonal
contraceptives on cortisol AUC-g was not significant
(F(1,17)=3.01, p>0.1), and there were no significant
interactions with testing day or treatment group
(all p>0.1).
The number of participants who were in a particular cycle
phase at the time of testing did not differ between the groups
(χ2=8.75, p>0.1). Due to insufficient power, the effects of
12. menstrual phase on cortisol or attention were not tested with
ANOVAs.
Table 1 Baseline characteristics
of participants by treatment group Measure Mean (SD) p
Placebo FOS B-GOS
Age, in years 23.27 (3.86) 24.53 (3.87) 23.27 (3.95) 0.31
NART score 116.50 (5.11) 115.71 (5.24) 112.62 (5.99) 0.18
EPQ, neuroticism 5.17 (4.61) 5.64 (4.22) 4.38 (3.36) 0.73
Perceived Stress Reactivity Scale 16.92 (10.06) 16.86 (6.95)
14.38 (5.66) 0.64
Beck Depression Inventory 2.58 (4.21) 2.14 (3.04) 2.54 (2.76)
0.93
Spielberger Anxiety Inventory, trait 32.00 (10.33) 32.86 (5.41)
31.69 (4.80) 0.91
Digit span, forward 10.08 (1.38) 8.43 (2.31) 8.54 (2.60) 0.12
Digit span, backward 8.50 (1.98) 6.79 (2.52) 7.69 (2.29) 0.18
1796 Psychopharmacology (2015) 232:1793–1801
Cortisol
Salivary cortisol did not differ significantly between groups at
baseline but was significantly lower following B-GOS com-
pared with placebo (Fig. 1). This was shown by an ANOVA
13. interaction effect of pre- versus post-treatment, group
(placebo, FOS or B-GOS) and sampling time point (in
minutes post-waking) on salivary cortisol levels, followed
up with separate group×time point ANOVAs for each day
of sampling (day 0: main effect of group F(2,41)=1.08, n.s.;
day 21: main effect of group F(2,41)=4.20, p<0.05, followed
up with Sidak-corrected contrasts: placebo vs. GOS, p=0.02,
all others p>0.1). Main effects analyses of the day×group×
time ANOVA confirmed that cortisol levels were increased
post-waking 15, 30, 45 and 60 min after waking (significant
main effect of time F(2.41,98.63)=58.61, p<0.001, planned
follow-up contrasts all significant at p<0.001). The main
effects of day of sampling and treatment group were not
significant (p>0.1). Gender was not entered as a factor of
interest due to insufficient power.
The lowered CAR in the B-GOS compared to the placebo
group on day 21 of supplement administration was also con-
firmed when analysing area under the curve with respect to
ground (Fig. 2; day×group ANOVA on square-root-
transformed salivary cortisol values: day×group interaction
[F(2,41)=3.52, p=0.039], followed up with separate group
ANOVAs for pre/day 0 [F(2,41)=1.24, n.s.] and post/day 21
[F(2,41)=4.12, p=0.023, follow-up contrasts: placebo vs. B-
GOS p=0.019, placebo vs. FOS p>0.1, FOS vs. B-GOS
p>0.1]).
Emotional Test Battery
Attentional dot-probe task
There was a significant group×emotion×masking condition
interaction in the visual dot-probe task (group×emotion×
masking condition [F(2,41)=3.14, p=0.05]). As can be seen
in Fig. 3, this effect was driven by decreased attentional
vigilance to negative versus positive information in the
14. unmasked condition (Fig. 3b), with no significant main effects
or interactions in the masked condition (Fig. 3a; valence×
group interaction in unmasked: F(2,41)=4.29, p=0.02;
masked: F(2,41)=0.85, p>0.1). Follow-up analyses with sep-
arate ANOVAs for prebiotic group compared with placebo in
the unmasked condition confirmed this effect as driven by
increased positive versus negative vigilance after B-GOS
compared to placebo, while the FOS group did not perform
differently to placebo (B-GOS vs. placebo: valence×group
F(1,27)=6.94, p=0.014, FOS vs. placebo: valence×group
F(1,27)=3.20, n.s.).
FERT
There were no significant effects of prebiotic treatment on
measures of accuracy (main effect of group: F(2,42)=1.71,
n.s., emotion×group interaction F(7.83,164.52)=0.67, n.s.).
Analysis of reaction time data revealed no significant interac-
tion between group and emotion (main effect of group:
F(2,42) = 0.53, n.s.; emotion × group interaction
F(8.16,773.38)=1.10, n.s.).
Emotional categorisation, recall and recognition
Participants responded faster to positive (mean reaction
time=1031 ms, SD=228 ms) compared to negative
(mean reaction time=1083 ms, SD=200 ms) self-
referential personality words in the emotional
categorisation task (valence×group ANOVA, main effect
of valence: F(1,42)=12.82, p<0.01) and in the emotion-
al word recognition task (mean reaction time to positive
words=1220.92, SD=263.03; mean reaction time to
negative words=1381.07, SD=348.13; main effect of
valence: F(1,41)=23.34, p<0.001); however, there was
no significant main effect of prebiotic treatment group
15. (F(2,42)=0.80, p>0.1) and the relative speeding for
positive words did not differ between groups (emo-
tion×group F(2,42)=0.35, p>0.1). Positive words were
also remembered more often than negative words in
both the surprise recall task (mean accuracy, positive
words = 7.41 (SD = 2.45), negative = 5.70 (2.29);
F(1,41)=16.16, p<0.001) and in the recognition task
Fig. 1 Cortisol awakening response before and after
administration of placebo, B-GOS or FOS. There were no
differences in the salivary CAR pre-
administration. Salivary cortisol awakening response was
significantly lower after 3 weeks of B-GOS intake, but not FOS
intake, compared with placebo
Psychopharmacology (2015) 232:1793–1801 1797
(mean correct recognition, positive words=25.67 (SD=
3.41), negative words=22.36 (SD=3.88); F(1,41)=
53.54, p<0.001), but these effect did not differ between
groups (p>0.1).
Self-report questionnaires
There were no significant effects of group on self-report
measures of state anxiety or perceived stress before or after
prebiotic/placebo administration (see Table 2). Furthermore,
there were no group differences in the overall cognitive status
as assessed by digit span on the day of psychological testing.
Correlational analyses
To test the hypothesis that cortisol levels were associated with
changes in attentional dot-probe performance in the B-GOS
16. group, we correlated difference scores of positive versus neg-
ative reaction times in the unmasked condition with absolute
cortisol levels upon waking on the day of testing and with the
difference in pre- versus post-prebiotics cortisol values (day
0–day 21). There were no associations of cortisol with atten-
tional performance (all p>0.1).
Discussion
The current study explored the neuroendocrine and affective
effects of two types of prebiotic supplements in healthy hu-
man volunteers, using salivary CAR and a validated test
battery of emotional processing. Results revealed that B-
GOS prebiotic intake was associated with decreased waking
salivary cortisol reactivity and altered attentional bias com-
pared to placebo. These results are consistent with previously
found anxiolytic-like effects of probiotics and reveal key
differences between two different prebiotic supplements.
Our findings of lowered cortisol awakening reactivity in
the group receiving B-GOS prebiotics compared to the place-
bo group indicate that prebiotic administration may modulate
HPA activity in a similar fashion as the administration of
probiotic strains directly seen in rodents (Sudo et al. 2004;
Gareau et al. 2007) and humans (Messaoudi et al. 2011). The
cortisol awakening response is a reliable marker of HPA axis
activity which has been found to be increased by work
stressors (Pruessner et al. 1997; Kunz-Ebrecht et al. 2004)
and in individuals at high risk of depression (Mannie et al.
2007). Insufficient or excessive cortisol reactivity may indi-
cate dysfunctional HPA axis feedback mechanisms, which
may provide useful targets for modulation by treatments in
certain vulnerability or disease states (Pariante and Lightman
2008; Dinan and Cryan 2012).
Participants receiving B-GOS supplements showed in-
17. creased attentional vigilance to positive versus negative stim-
uli on the dot-probe task. Our effects are similar to those seen
following administration of pharmacological agents such as
the selective serotonin reuptake inhibitor citalopram or the
benzodiazepine diazepam in healthy individuals (Browning
et al. 2006; Murphy et al. 2009a, b). These effects have been
interpreted as showing an early anxiolytic-like profile, where
threatening stimuli are less likely to be attended to (Harmer
2010). Interestingly, we found effects of the B-GOS prebiotic
administration on altered attentional processing only in the
unmasked condition (500 ms presentation) of the dot-probe
task. Attentional vigilance to brief, masked presentations of
threatening cues has primarily been interpreted as an
Fig. 2 Area under the curve (with respect to ground) of salivary
cortisol
awakening response pre- and post-prebiotic supplement/placebo
intake.
*p<0.05
Fig. 3 Vigilance reaction times in the attentional dot-probe task.
a
Attentional vigilance did not differ between groups during
masked trials
of the attentional dot-probe task. b Participants showed
decreased
attentional vigilance to negative versus positive words in the
unmasked
condition of the dot-probe task after B-GOS but not FOS intake
compared
to placebo
1798 Psychopharmacology (2015) 232:1793–1801
18. involuntary deployment of attention (Browning et al. 2010),
whereas at longer stimulus durations, it may further involve a
difficulty to disengage from salient emotional stimuli (Koster
et al. 2004; Cisler and Koster 2010). Research suggests that
while responses to masked stimuli may be particularly prev-
alent in anxiety disorders rather than depression, attentional
bias seen at longer exposure durations may also be of rele-
vance to depression. The increase of positive compared to
negative emotional information processing in the B-GOS
group provides initial evidence that behavioural effects of
probiotics in rodent models (Bravo et al. 2011) can be extend-
ed to affective processing in humans using prebiotics. These
results are also consistent with a recent fMRI study which
reported that a 3-week probiotic administration reduced neural
response in a network of areas (including the somatosensory
cortex, insula and parahippocampal gyrus) to angry and fear-
ful facial expressions (Tillisch et al. 2013).
Differences in attentional resource allocation to negative or
positive stimuli are associated with individual variability in
trait and state measures. Specifically, vigilance to cues of
threat or danger is greater in highly anxious individuals com-
pared with low-anxious individuals and healthy controls
(Mogg et al. 1994; Bradley et al. 1998; Koster et al. 2005),
and threat-related processing is believed to play a key role in
the symptomatology of anxiety and its modulation by anxio-
lytics (Beck and Clark 1997; Mogg and Bradley 1998). Phar-
macological anxiolytics and antidepressants that are clinically
effective in reducing symptoms have been found to affect
reductions in specific negative biases in neural correlates of
emotional information processing (Sheline et al. 2001; Fu
et al. 2004; Godlewska et al. 2012), and also modulate biases
when administered in healthy control and at-risk groups
(Harmer et al. 2003; Browning et al. 2006; Murphy et al.
2009b). Based on the initial results of gut microbiota inter-
ventions, it is now crucial to investigate the extent and spec-
19. ificity of information processing biases that may be targeted
by different gut microbiota manipulations and whether they
prove clinically beneficial.
Although we were unable to test the mechanisms of action
directly through characterisation of gut microbiota, a previous
characterisation of B-GOS and FOS prebiotics showed
pronounced increases in bifidobacteria in faecal pellets of rats
treated with B-GOS administration compared to placebo, with
more moderate effects following the FOS intervention
(Savignac et al. 2013). The specificity of galacto-
oligosaccharides affecting behavioural and endocrine changes
is thus in line with previous findings of galacto-
oligosaccharides as particularly effective in stimulating enteric
microbial growth (Abou Hachem et al. 2013). Of course, the
possibility of an additional, direct effect of B-GOS on the gut
mucosa [10] cannot be ruled out.
Given the lack of association between altered attentional
processes and a reduction in the salivary cortisol awakening
response (CAR), we were unable to confirm a moderating
effect of cortisol reactivity on behaviour. Although there is
strong evidence for a role of the gut microbiota in the regula-
tion of the HPA axis (see Dinan and Cryan 2012 for a review),
the exact mechanisms by which they interact with central
effects remain to be investigated. One potential mechanism
of action underlying these effects is via anti-inflammatory and
immune responses following probiotic proliferation (Lyte
2011; Ait-Belgnaoui et al. 2012), and central effects of
probiotics have been found to be vagus nerve dependent
(Bravo et al. 2011).
While initial results of the B-GOS prebiotic on the cortisol
awakening response are promising, we had insufficient power
to investigate gender effects, which have previously been
20. reported (Pruessner et al. 1997). Further, there were no effects
of either prebiotic on the remaining tasks of the ETB which
examine aspects of facial expression recognition, self-
referential processing and emotional memory. The current
results also differ from previous findings that indicate subjec-
tive anxiolytic effects of probiotics (Rao et al. 2009;
Messaoudi et al. 2011) as we found no effects of prebiotics
on subjective measures of subclinical anxiety or perceived
stress. One improvement may be to extend the administration
period of prebiotics as probiotics take several weeks to prolif-
erate. It is also possible that these results are in part due to our
study population of young healthy volunteers with low sub-
clinical scores and presumably healthy gut microbiota com-
positions even before treatment. Studying a population with a
potential deficiency in their gut microbiota compositions, for
Table 2 State measures of
anxiety and perceived stress
(mean, SD) before and after
prebiotic/placebo administration
and cognitive status on day of
assessment (day 21)
Measure Mean (SD) p
FOS Placebo B-GOS
Spielberger Anxiety Inventory, state Pre (day 0) 32.00 (7.13)
32.92 (10.03) 32.15 (9.11) 0.96
Post (day 21) 30.00 (5.53) 30.17 (7.55) 31.31 (5.23) 0.84
Perceived Stress Scale Pre (day 0) 9.86 (4.64) 9.92 (6.05) 10.77
(3.77) 0.87
Post (day 21) 9.57 (5.20) 8.92 (5.40) 10.46 (4.86) 0.75
21. Digit span (post, day 21) Forward 9.71 (2.40) 10.25 (1.96) 9.54
(2.44) 0.72
Digit span (post, day 21) Backward 7.50 (2.82) 8.25 (2.73) 8.62
(2.66) 0.56
Psychopharmacology (2015) 232:1793–1801 1799
example elderly individuals or IBS patients with comorbid
psychiatric symptoms, may improve power. The applicability
of our results to other populations—such as those with dete-
rioration in the health of their intestinal gut flora or HPA axis
abnormalities—is a valid next step for future study.
We found a selective modulation of attention to emotional
stimuli and HPA axis reactivity following B-GOS prebiotic
supplement in healthy participants, supporting a key role for
gut microbiota in the regulation of affective function. This, to
our knowledge, is the first study extending findings of the
central effects of probiotics to behavioural effects of prebiotics
in humans.
Acknowledgments We thank Mrs Li Chen for her technical
assistance
with the cortisol assays. This study was funded by BBSRC,
Clasado Ltd.,
Wellcome Trust Vacation Scholarship (SE) and MRC (KS).
Open Access This article is distributed under the terms of the
Creative
Commons Attribution License which permits any use,
distribution, and
reproduction in any medium, provided the original author(s) and
22. the
source are credited.
References
Abou Hachem M, Andersen JM, Barrangou R et al (2013)
Recent insight
into oligosaccharide uptake and metabolism in probiotic
bacteria.
Biocatal Biotransform 31:226–235. doi:10.3109/10242422.2013.
828048
Ait-Belgnaoui A, Durand H, Cartier C et al (2012) Prevention of
gut
leakiness by a probiotic treatment leads to attenuated HPA
response
to an acute psychological stress in rats.
Psychoneuroendocrinology
37:1885–1895. doi:10.1016/j.psyneuen.2012.03.024
Anderson N (1968) Likableness ratings of 555 personality trait
words. J
Pers Soc Psychol 9:272–279
Beck AT (2008) The evolution of the cognitive model of
depression and
its neurobiological correlates. Am J Psychiatr 165:969–977
Beck AT, Clark DA (1997) An information processing model of
anxiety:
automatic and strategic processes. Behav Res Ther 35:49–58
Beck AT, Ward C, Mendelson M et al (1961) An inventory for
measuring
depression. Arch Gen Psychiatry 4:561–571
23. Bhagwagar Z, Cowen PJ (2007) ‘It’s not over when it’s over’:
persistent
neurobiological abnormalities in recovered depressed patients.
Psychol Med. doi:10.1017/S0033291707001250
Bond A, Lader M (1974) The use of analogue scales in rating
subjective
feelings. Br J Med Psychol 47:211–218
Bradley BP, Mogg K, Falla SJ, Hamilton LR (1998) Attentional
bias for
threatening facial expressions in anxiety: manipulation of
stimulus
duration. Cogn Emot 12:737–753.
doi:10.1080/026999398379411
Bravo JA, Forsythe P, Chew MV et al (2011) Ingestion of
Lactobacillus
strain regulates emotional behavior and central GABA receptor
expression in a mouse via the vagus nerve. Proc Natl Acad Sci
108:16050–16055. doi:10.1073/pnas.1102999108
Bravo JA, Julio-Pieper M, Forsythe P et al (2012)
Communication
between gastrointestinal bacteria and the nervous system. Curr
Opin Pharmacol 12(6):667–672. doi:10.1016/j.coph.2012.09.010
Brenner DM, Moeller MJ, Chey WD, Schoenfeld PS (2009) The
utility of
probiotics in the treatment of irritable bowel syndrome: a
systematic
review. Am J Gastroenterol 104:1033–1049
Browning M, Reid C, Cowen PJ et al (2006) A single dose of
citalopram
increases fear recognition in healthy subjects. J
24. Psychopharmacol
21:684–690. doi:10.1177/0269881106074062
Browning M, Holmes EA, Harmer CJ (2010) The modification
of atten-
tional bias to emotional information: a review of the techniques,
mechanisms, and relevance to emotional disorders. Cogn Affect
Behav Neurosci 10:8–20. doi:10.3758/CABN.10.1.8
Chan SWY, Goodwin GM, Harmer CJ (2007) Highly neurotic
never-
depressed students have negative biases in information
processing.
Psychol Med 37:1281–1291. doi:10.1017/S0033291707000669
Cisler JM, Koster EHW (2010) Mechanisms of attentional
biases towards
threat in anxiety disorders: an integrative review. Clin Psychol
Rev
30:203–216. doi:10.1016/j.cpr.2009.11.003
Cohen S, Kamarck T, Mermelstein R (1983) A global measure
of per-
ceived stress. J Health Soc Behav 24:385–396
Craig AD (2009) How do you feel—now? The anterior insula
and human
awareness. Nat Rev Neurosci 10:59–70
Crumeyrolle-Arias M, Jaglin M, Bruneau A et al (2014)
Absence of the
gut microbiota enhances anxiety-like behavior and
neuroendocrine
response to acute stress in rats. Psychoneuroendocrinology
42:207–
217. doi:10.1016/j.psyneuen.2014.01.014
25. Cryan JF, Dinan TG (2012) Mind-altering microorganisms: the
impact of
the gut microbiota on brain and behaviour. Nat Rev Neurosci
13:
701–712. doi:10.1038/nrn3346
Cryan JF, O’Mahony SM (2011) The microbiome-gut-brain
axis: from
bowel to behavior. Neurogastroenterol Motil 23:187–192.
doi:10.
1111/j.1365-2982.2010.01664.x
Dinan TG, Cryan JF (2012) Regulation of the stress response by
the gut
microbiota: implications for psychoneuroendocrinology.
Psychoneuroendocrinology 37:1369–1378. doi:10.1016/j.
psyneuen.2012.03.007
Dinan TG, Quigley EMM, Ahmed SMM et al (2006)
Hypothalamic-
pituitary-gut axis dysregulation in irritable bowel syndrome:
plasma
cytokines as a potential biomarker? Gastroenterology 130:304–
311.
doi:10.1053/j.gastro.2005.11.033
Ekman P, Friesen WV (1976) Pictures of Facial Affect.
Consulting
Psychologists Press, Palo Alto
Elliott R, Zahn R, Deakin JFW, Anderson IM (2011) Affective
cognition and
its disruption in mood disorders. Neuropsychopharmacol Off
Publ Am
Coll Neuropsychopharmacol 36:153–182.
26. doi:10.1038/npp.2010.77
Eysenck H, Eysenck S (1975) Manual of the Eysenck
personality ques-
tionnaire. Hodder & Stoughton, London
Fehring RJ, Schneider M, Raviele K (2006) Variability in the
phases of
the menstrual cycle. J Obstet Gynecol Neonat Nurs 35(3):376–
384
First M, Spitzer R, Gibbon M, Williams J (1997) Structured
Clinical
Interview for DSM-IV axis I disorders (SCID-I)
Forsythe P, Kunze WA (2013) Voices from within: gut microbes
and the
CNS. Cell Mol Life Sci 70:55–69. doi:10.1007/s00018-012-
1028-z
Fu CHY, Williams SC, Cleare AJ et al (2004) Attenuation of the
neural
response to sad faces in major depression by antidepressant
treat-
ment: a prospective, event-related functional magnetic
resonance
imaging study. Arch Gen Psychiatry 61:877–889
Gareau MG, Jury J, MacQueen G et al (2007) Probiotic
treatment of rat
pups normalises corticosterone release and ameliorates colonic
dys-
function induced by maternal separation. Gut 56:1522–1528.
doi:
10.1136/gut.2006.117176
27. Gareau MG, Sherman PM, Walker WA (2010) Probiotics and
the gut
microbiota in intestinal health and disease. Nat Rev
Gastroenterol
Hepatol 7:503–514. doi:10.1038/nrgastro.2010.117
Godlewska BR, Norbury R, Selvaraj S et al (2012) Short-term
SSRI
treatment normalises amygdala hyperactivity in depressed
patients.
Psychol Med 42:2609–2617. doi:10.1017/S0033291712000591
Harmer CJ (2010) Antidepressant drug action: a
neuropsychological
perspective. Depress Anxiety 27:231–233.
doi:10.1002/da.20680
Harmer CJ, Bhagwagar Z, Perrett DI et al (2003) Acute SSRI
adminis-
tration affects the processing of social cues in healthy
volunteers.
1800 Psychopharmacology (2015) 232:1793–1801
http://dx.doi.org/10.3109/10242422.2013.828048
http://dx.doi.org/10.3109/10242422.2013.828048
http://dx.doi.org/10.1016/j.psyneuen.2012.03.024
http://dx.doi.org/10.1017/S0033291707001250
http://dx.doi.org/10.1080/026999398379411
http://dx.doi.org/10.1073/pnas.1102999108
http://dx.doi.org/10.1016/j.coph.2012.09.010
http://dx.doi.org/10.1177/0269881106074062
http://dx.doi.org/10.3758/CABN.10.1.8
http://dx.doi.org/10.1017/S0033291707000669
http://dx.doi.org/10.1016/j.cpr.2009.11.003
http://dx.doi.org/10.1016/j.psyneuen.2014.01.014
29. Koster EHW, Crombez G, Verschuere B, De Houwer J (2004)
Selective
attention to threat in the dot probe paradigm: differentiating
vigi-
lance and difficulty to disengage. Behav Res Ther 42:1183–
1192.
doi:10.1016/j.brat.2003.08.001
Koster EHW, Verschuere B, Crombez G, Van Damme S (2005)
Time-
course of attention for threatening pictures in high and low trait
anxiety. Behav Res Ther 43:1087–1098.
doi:10.1016/j.brat.2004.
08.004
Kunz-Ebrecht SR, Kirschbaum C, Marmot M, Steptoe A (2004)
Differences in cortisol awakening response on work days and
week-
ends in women and men from the Whitehall II cohort.
Psychoneuroendocrinology 29:516–528. doi:10.1016/S0306-
4530(03)00072-6
Lyte M (2011) Probiotics function mechanistically as delivery
vehicles
for neuroactive compounds: microbial endocrinology in the
design
and use of probiotics. Bioessays 33:574–581. doi:10.1002/bies.
201100024
Mannie ZN, Harmer CJ, Cowen PJ (2007) Increased waking
salivary
cortisol levels in young people at familial risk of depression.
Am J
Psychiatry 164:617–621. doi:10.1176/appi.ajp.164.4.617
30. Mayer EA (2011) Gut feelings: the emerging biology of gut–
brain com-
munication. Nat Rev Neurosci 12:453–466.
doi:10.1038/nrn3071
Messaoudi M, Lalonde R, Violle N et al (2011) Assessment of
psychotropic-like properties of a probiotic formulation
(Lactobacillus helveticus R0052 and Bifidobacterium longum
R0175) in rats and human subjects. Br J Nutr 105:755–764. doi:
10.1017/S0007114510004319
Mogg K, Bradley BP (1998) A cognitive-motivational analysis
of anxi-
ety. Behav Res Ther 36:809–848
Mogg K, Bradley BP, Hallowell N (1994) Attentional bias to
threat: roles
of trait anxiety, stressful events, and awareness. Q J Exp
Psychol
Sect A 47:841–864. doi:10.1080/14640749408401099
Murphy SE, Norbury R, O’Sullivan U et al (2009a) Effect of a
single dose
of citalopram on amygdala response to emotional faces. Br J
Psychiatry 194:535–540. doi:10.1192/bjp.bp.108.056093
Murphy SE, Yiend J, Lester KJ et al (2009b) Short-term
serotonergic but not
noradrenergic antidepressant administration reduces attentional
vigi-
lance to threat in healthy volunteers. Int J
Neuropsychopharmacol Off
Sci J Coll Int Neuropsychopharmacologicum (CINP) 12:169–
179. doi:
10.1017/S1461145708009164
31. Nelson H (1982) The National Adult Reading Test (NART): test
manual.
NFER-Nelson, Windsor
Pariante CM, Lightman SL (2008) The HPA axis in major
depression:
classical theories and new developments. Trends Neurosci
31:464–
468. doi:10.1016/j.tins.2008.06.006
Pizzagalli DA (2010) Frontocingulate dysfunction in
depression: toward
biomarkers of treatment response. Neuropsychopharmacology
36:
183–206
Pruessner JC, Wolf OT, Hellhammer DH et al (1997) Free
cortisol levels
after awakening: a reliable biological marker for the assessment
of
adrenocortical activity. Life Sci 61:2539–2549
Rao AV, Bested AC, Beaulne TM et al (2009) A randomized,
double-
blind, placebo-controlled pilot study of a probiotic in emotional
symptoms of chronic fatigue syndrome. Gut Pathog 1:6. doi:10.
1186/1757-4749-1-6
Rhee SH, Pothoulakis C, Mayer EA (2009) Principles and
clinical impli-
cations of the brain–gut–enteric microbiota axis. Nat Rev
Gastroenterol Hepatol 6:306–314. doi:10.1038/nrgastro.2009.35
Savignac HM, Corona G, Mills H et al (2013) Prebiotic feeding
elevates
central brain derived neurotrophic factor, N-methyl-d-aspartate
32. re-
ceptor subunits and d-serine. Neurochem Int 63:756–764.
doi:10.
1016/j.neuint.2013.10.006
Schlotz W, Yim IS, Zoccola PM et al (2011) The perceived
stress
reactivity scale: measurement invariance, stability, and validity
in
three countries. Psychol Assess 23:80–94.
doi:10.1037/a0021148
Sheline YI, Barch DM, Donnelly JM et al (2001) Increased
amygdala
response to masked emotional faces in depressed subjects
resolves
with antidepressant treatment: an fMRI study. Biol Psychiatry
50:
651–658
Spielberger C, Gorsuch R, Lushene R (1970) Manual for the
State-Trait
Anxiety Inventory (STAI). Consulting Psychologists Press, Palo
Alto
Sudo N, Chida Y, Aiba Y et al (2004) Postnatal microbial
colonization
programs the hypothalamic–pituitary–adrenal system for stress
re-
sponse in mice. J Physiol 558:263–275
Tillisch K, Labus J, Kilpatrick L et al (2013) Consumption of
fermented
milk product with probiotic modulates brain activity.
Gastroenterology 144:1394–1401.e4. doi:10.1053/j.gastro.2013.
02.043
33. Vulevic J, Drakoularakou A, Yaqoob P, et al (2008) Modulation
of the
fecal microflora profile and immune function by a novel trans-
galactooligosaccharide mixture (B-GOS) in healthy elderly
volun-
teers. Am J Clin Nutr 88:1438–1446
Watson D, Clark L, Tellegen A (1988) Development and
validation
of brief measures of positive and negative affect: the Positive
and Negative Affect Schedule scales. J Pers Soc Psychol 54:
1063–1070
Wolfram M, Bellingrath S, Kudielka BM (2011) The cortisol
awakening
response (CAR) across the female menstrual cycle.
Psychoneuroendocrinology 36(6):905–912
Young AW, Rowland D, Calder AJ, et al (1997) Facial
expression
megamix: tests of dimensional and category accounts of
emotion
recognition. Cognition 63:271–313
Psychopharmacology (2015) 232:1793–1801 1801
http://dx.doi.org/10.1038/sj.npp.1300004
http://dx.doi.org/10.1038/sj.npp.1300004
http://dx.doi.org/10.1192/bjp.bp.108.051193
http://dx.doi.org/10.1016/j.neubiorev.2011.07.001
http://dx.doi.org/10.1016/j.brat.2003.08.001
http://dx.doi.org/10.1016/j.brat.2004.08.004
http://dx.doi.org/10.1016/j.brat.2004.08.004
http://dx.doi.org/10.1016/S0306-4530(03)00072-6
http://dx.doi.org/10.1016/S0306-4530(03)00072-6
35. Gregory E. Simon, MD, MPH; Richard Gater, MRCPsych
Context.— There is little information on the extent of persistent
pain across cul-
tures. Even though pain is a common reason for seeking health
care, information
on the frequency and impacts of persistent pain among primary
care patients is in-
adequate.
Objective.— To assess the prevalence and impact of persistent
pain among pri-
mary care patients.
Design and Setting.— Survey data were collected from
representative samples
of primary care patients as part of the World Health
Organization Collaborative
Study of Psychological Problems in General Health Care,
conducted in 15 centers
in Asia, Africa, Europe, and the Americas.
Participants.— Consecutive primary care attendees between the
age of major-
ity (typically 18 years) and 65 years were screened (n = 25 916)
and stratified ran-
dom samples interviewed (n = 5438).
Main Outcome Measures.— Persistent pain, defined as pain
present most of the
time for a period of 6 months or more during the prior year, and
psychological ill-
ness were assessed by the Composite International Diagnostic
Interview. Disabil-
ity was assessed by the Groningen Social Disability Schedule
36. and by activity-
limitation days in the prior month.
Results.— Across all 15 centers, 22% of primary care patients
reported persis-
tent pain, but there was wide variation in prevalence rates
across centers (range,
5.5%-33.0%). Relative to patients without persistent pain, pain
sufferers were more
likely to have an anxiety or depressive disorder (adjusted odds
ratio [OR], 4.14; 95%
confidence interval [CI], 3.52-4.86), to experience significant
activity limitations
(adjusted OR, 1.63; 95% CI, 1.41-1.89), and to have
unfavorable health perceptions
(adjusted OR, 1.26; 95% CI, 1.07-1.49). The relationship
between psychological
disorder and persistent pain was observed in every center, while
the relationship
between disability and persistent pain was inconsistent across
centers.
Conclusions.— Persistent pain was a commonly reported health
problem
among primary care patients and was consistently associated
with psychological
illness across centers. Large variation in frequency and the
inconsistent relation-
ship between persistent pain and disability across centers
suggests caution in
drawing conclusions about the role of culture in shaping
responses to persistent
pain when comparisons are based on patient samples drawn
from a limited num-
ber of health care settings in each culture.
37. JAMA. 1998;280:147-151
PAIN is one of the most common1 and
among the most personally compelling
reasons for seeking medical attention.
People seek health care for pain not only
for diagnostic evaluation and symptom
relief, but also because pain interferes
with daily activities, causes worry and
emotional distress, and undermines con-
fidence in one’s health. When pain per-
sists for weeks or months, its broader
effects on well-being can be profound.
Psychological health and performance of
social responsibilities in work and family
life can be significantly impaired.2
Despite evidence that pain affects
well-being, little is known about how
common persistent pain is among pri-
mary care patients. There is evidence
that the effects of persistent pain on psy-
chological health and functional status
are similar for pain problems at different
anatomical sites.3 However, it is not
known whether impaired emotional
well-being and increased disability are
consistent correlates of persistent pain,
or whether the impacts of persistent
pain on well-being are consistent across
cultures. Several recent studies have
compared pain perceptions and coping
across cultures,4-6 but cross-cultural re-
search on pain has typically studied rela-
38. tively small numbers of patients in con-
venience samples. Comparison groups
of pain-free controls have often been
lacking.
This article reports data from a World
Health Organization (WHO) survey of
primary care patients, the WHO Col-
laborative Study of Psychological Prob-
lems in General Health Care.7 As part of
a broader assessment of health and men-
tal health status, this cross-national sur-
vey collected information on persistent
pain. This report estimates the preva-
lence of persistent pain among primary
care patients in different countries, and
determines the association of persistent
pain with health perceptions, psycho-
logical distress, and activity limitations.
(Persistent pain was defined as pain
present most of the time for a period of 6
months or more during the prior year.)
This article provides the first cross-
national data on the prevalence of per-
sistent pain among primary care pa-
tients, and is also the first large-scale
cross-national study to assess whether
persistent pain shows consistent rela-
tionships to impaired well-being and
functioning among primary care pa-
tients in many different countries. While
this study was not designed or intended
to test specific hypotheses about cross-
cultural differences in the prevalence or
impacts of persistent pain, it provides
new information on the frequency and
40. ing the relationship between psychologi-
cal illness and disability were previously
reported in this journal.10 The 15 partici-
pating centers in 14 different countries
were selected to represent broad diver-
sity of culture and socioeconomic devel-
opment. Centers were selected on the
basis of previous successful collabora-
tion with WHO, experience with re-
search in primary care settings, access
to primary care patient populations,
availability of appropriately skilled per-
sonnel to ensure full adherence to the
study protocol, and approval for the
study by local ethics committees. Each
center was required to identify health
care facilities that could be regarded as
prototypical of primary health care ser-
vices in that country.
The study population was consecutive
patients attending the participating pri-
mary care facilities, including both new
and returning patients. Patients were in-
cluded if they were between the age of
majority (typically 18 years) and 65
years. Eligible subjects were not too ill
to participate, had a fixed address, were
attending the clinic for a medical consul-
tation, and gave informed consent. In-
formation on the presenting problems of
patients enrolled in the study is present-
ed for each center elsewhere.7 The 12-
item General Health Questionnaire
(GHQ)11 was administered as a screen-
ing instrument to obtain a stratified
41. random sample in which patients who
were psychologically distressed were
sampled with higher probability than pa-
tients who were not distressed.
A total of 25 916 patients were suc-
cessfully screened. This represented a
response rate of 96%. Patients were se-
lected for the second-stage assessment
using stratified random sampling based
on their GHQ score. Using center-spe-
cific GHQ score norms determined from
a large pilot test in each center,9 patients
were placed in a low GHQ score stratum
(approximately 60% of consecutive pa-
tients in a particular center), a medium
GHQ score stratum (20% of patients), or
a high GHQ score stratum (20% of pa-
tients). The high GHQ score stratum cor-
responds to a moderate to severe level of
psychological distress, the medium GHQ
stratum to a mild level of distress, and
the low GHQ score stratum to a low level
of psychological distress. All high GHQ
scorers, 35% of medium GHQ scorers,
and 10% of low GHQ scorers were ran-
domly sampled for the second-stage as-
sessments. The analysis of data from this
stratified random sampling scheme was
weighted taking the sample selection
probabilities in each stratum into ac-
count (as explained below), so that un-
biased estimates were obtained for the
population of consecutive primary care
attendees in each center. Sampled pa-
42. tients were interviewed at a place of
their choice, commonly their home. Of
8729 eligible patients, 5447 completed
the second-stage assessments (average
response rate, 62%).
Assessment
Patients sampled for the second-stage
evaluation were assessed by highly
trained interviewers using the WHO
primary care version of the Composite
International Diagnostic Interview
(CIDI).12 This version assessed persis-
tent pain in addition to identifying psy-
chological disorders (eg, anxiety and de-
pressive disorders) defined according to
International Statistical Classification
of Diseases, 10th Revision (ICD-10)13 di-
agnostic criteria. Using this question-
naire, a pain problem was defined as cur-
rent and persistent if pain was present
most of the time for a period of 6 months
or more during the prior year. To elimi-
nate insignificant aches and pains, pa-
tients needed to report that at some time
during their lifetime they talked to ei-
ther a physician or other health profes-
sional about the pain, had taken medica-
tion for the pain more than once, or had
reported that the pain had interfered
with life or activities a lot. Although the
CIDI obtained ratings of whether per-
sistent pain was “medically explained”
or not, these ratings were ignored for
the purposes of this report. Review of
43. these ratings indicated that the ratings
of what conditions were medically ex-
plained were inconsistent across cen-
ters. Moreover, understanding the fre-
quency of persistent pain is clinically
important whether the pain is medically
explained or not.
Disability was assessed using the “Oc-
cupational Role” section of the Social
Disability Schedule (SDS).14 The SDS is
a semistructured interview that rates
disability on the basis of work role per-
formance relative to cultural expecta-
tions. Daily work activities (including
gainful employment, volunteer work, or
housekeeping), activities directed at se-
curing a job for individuals not employed
(study and job searching), and the struc-
turing of daily activities for retired indi-
viduals were assessed. Interviewer rat-
ings were made on a 4-point scale: 0 (no
disability), 1 (mild disability), 2 (moder-
ate disability), and 3 (severe disability).
Interviewer-observer reliability of the
SDS occupational role was assessed with
19 videotaped interviews circulated
across the centers. An overall k of 0.85
was obtained, with a range of 0.72 and
0.93 on items.9 In addition, each subject
was asked the number of days in the pre-
vious month they had been unable to
carry out their usual activities.15 Pa-
tients rated their overall health status
as excellent, very good, good, fair, or
44. poor.
The physician seeing each patient in
the sample completed an encounter form
that included a rating of the patient’s
physical health status at the time of the
visit. Patients were rated by their phy-
sicians as completely healthy, having
some symptoms but subclinical physical
illness, mild physical illness, moderate
physical illness, or severe physical ill-
ness. All participating physicians were
instructed in the use of the encounter
form in practice sessions with the local
investigators. These ratings were used
to control for severity of physical illness
in multivariate analyses.
At non–English-speaking centers,
questionnaires were translated by a
panel of local bilingual experts. Back-
translations to English were checked
centrally at WHO. At least 1 English-
speaking investigator from every center
participated in a 5-day joint training ses-
sion in the use of the instruments. In gen-
eral, the interviewers who assessed
study subjects had mental health train-
ing and experience.
Data Analysis
Because this study used a stratified
random sampling plan, the estimates we
report are based on weighted data.
Weighted data from the second-stage as-
46. pain vs those without persistent pain af-
ter controlling for center, sex, age, phy-
sician-rated physical health status, and
whether a CIDI-diagnosed anxiety or
depressive disorder was present in the
prior month. For these analyses, we re-
port the estimated ORs, CIs, and P val-
ues for all centers combined. In addition,
we report the percentage with each im-
pairment comparing patients with and
without persistent pain for each center,
and indicate whether the difference was
greater than expected by chance at the
.05 significance level for a 2-sided test.
These significance tests were based on
Wald statistics from logistic regression
models controlling for age, sex, physi-
cian-rated severity of physical disease,
and the presence of a depressive or anxi-
ety disorder.
RESULTS
Prevalence of Persistent Pain
Persistent pain was common among
primary care patients across a wide
range of settings in different countries.
The prevalence of persistent pain for all
centers combined was 21.5%, with
prevalence rates varying from 5% to
33%. Sex-specific prevalence rates are
shown in Table 1, with the centers or-
dered from the highest overall preva-
lence rate to the lowest. The difference
47. in prevalence rates across centers was
highly significant after adjusting for age
and sex (Wald statistic = 217.7, df = 14,
P,.001).
Among the European centers, Ath-
ens, Greece (12%), and Verona, Italy
(13%), had relatively low prevalence
rates, while the remaining centers in
Germany, France, the Netherlands, and
England were found to have persistent
pain prevalence rates in excess of 20%.
The 2 Asian centers (Nagasaki, Japan,
and Shanghai, China) had relatively low
prevalence rates of persistent pain (12%
and 13%, respectively), while the 2 South
American centers (Rio de Janeiro, Bra-
zil, and Santiago, Chile) had relatively
high prevalence rates (31% and 33%, re-
spectively). The center in Ibadan, Nige-
ria, had the lowest prevalence rates of
persistent pain of any center for both
men and women.
As shown in Table 1, persistent pain
was significantly more common among
women than men based on a pooled es-
timate for the 15 participating centers,
with 25% of women compared with 16%
of men reporting persistent pain. After
adjusting for age, the prevalence of per-
sistent pain was significantly higher
among women than men in 9 of the 15
centers.
48. Anatomical Site
As shown in Table 2, among patients
with persistent pain, the 3 most com-
monly reported anatomical pain sites (in
order of frequency) were back pain,
headache, and joint pain. The large ma-
jority (68%) of primary care patients
with persistent pain reported pain in at
least 2 anatomical sites (Table 2). Be-
cause pain was typically reported at mul-
tiple sites, the remaining analyses con-
cern persistent pain without differentia-
tion by anatomical site.
Persistent Pain and Well-being
Persons with persistent pain were
substantially more likely to have an anxi-
ety or depressive disorder meeting ICD-
10 diagnostic criteria than persons not
experiencing persistent pain (Table 3).
After adjusting for center, age, sex, and
physician-rated severity of physical dis-
ease, the odds of having a psychological
disorder meeting diagnostic criteria
among persons with persistent pain
showed a 4-fold increase over those not
affected by persistent pain. The associa-
tion of persistent pain was not specific to
depression, as both anxiety and depres-
sive disorders showed a comparable as-
sociation with persistent pain.16
For all 15 centers combined, the pres-
49. ence of persistent pain was associated
with a modest increase in the likelihood of
patients rating their overall health as fair
or poor (Table 3). Unfavorable health per-
ceptions were reported by 33% of those
with persistent pain compared with 21%
of those without persistent pain.
Work role disability was assessed by a
semistructured interview protocol tak-
ing cultural norms into account in deter-
mining the extent of disability.14 Across
the participating centers, 31% of those
with persistent pain were rated as having
moderate to severe work role interfer-
ence, compared with 13% among those
without persistent pain. After adjusting
for center, age, sex, psychological disor-
der status, and physician-rated severity
of physical disease, the odds of work dis-
ability showed a 2-fold increase among
those with persistent pain (Table 3). Simi-
Table 1.—Subjects With Persistent Pain by Sex, World Health
Organization Psychological Problems in General Health Care
Survey, 1991-1992 (Weighted Data)
Participating Center
(No. of Cases/No. of Subjects)
Men, %
(n = 1919)
Women, %
(n = 3519)
51. Ibadan, Nigeria (27/269) 6.2 5.3 5.5 0.92 (0.28-2.95) .88
All centers (1569/5438) 16.2 24.8 21.5 1.69 (1.47-1.95) ,.001
*The adjusted odds ratio (OR) measures the risk of having
persistent pain among women relative to the risk of persistent
pain among men, after adjusting for age. The all
centers OR was adjusted for age and center. CI indicates
confidence interval.
Table 2.—Subjects Reporting Current Pain at
Different Anatomical Sites and the Number of
Anatomical Sites With Pain Among Subjects With
Persistent Pain, World Health Organization
Psychological Problems in General Health Care
Survey, 1991-1992 (Weighted Data)
Variable
Subjects Reporting
Current Pain, %
Anatomical site
Back pain 47.8
Headache 45.2
Joint pain 41.7
Arms or legs 34.3
Chest 28.9
Abdominal pain 24.9
Pain elsewhere 11.7
No. of anatomical sites
1 32.1
2 27.5
3 22.8
53. tistically significant for only 5 of the 15
centers (significant differences are indi-
cated by numbers in boldface type in
Table 4). Interviewer-rated work dis-
ability was significantly more common
among those with persistent pain for 5 of
the 15 centers, and patients with persis-
tent pain were significantly more likely
to report 3 or more days of activity limi-
tation in the prior month for 6 of the 15
centers. For the centers with nonsignifi-
cant differences in work disability or in
activity-limitation days between those
with and without persistent pain, pa-
tients with persistent pain almost al-
ways had a higher percentage with ac-
tivity limitation than patients without
persistent pain.
COMMENT
This is the first large-scale cross-na-
tional study of persistent pain among
primary care patients in which standard
methods were applied to estimate its
prevalence and impacts in a wide range
of countries. Even though there was sub-
stantial variation in prevalence rates
across centers, persistent pain was a
common problem among patients con-
sulting primary care physicians in every
participating center. It should be noted
that the patients eligible for this study
were seeking professional health care,
54. and that the care settings were gener-
ally in urban areas. Persons seeking
health care are likely to have higher
prevalence rates of persistent pain than
a general population sample. In addition,
the patient populations studied may dif-
fer from those seeking services from tra-
ditional providers or from persons seek-
ing health care in rural areas. However,
the kinds of primary care settings in-
cluded in this study provide health care
services to large segments of the popu-
lation in each of the countries included in
this study.
This study was not designed to ex-
plain cross-cultural differences in the
prevalence or cross-cultural differences
in the impact of persistent pain. How-
ever, the large variation in rates of
occurrence of persistent pain across cen-
ters, the inconsistency in the relation-
ship between persistent pain and disabil-
ity, and the lack of a readily explainable
pattern for the variation in results
should give pause. This variability, and
the lack of any clear pattern to the varia-
tion across centers, suggests that it may
be difficult to draw meaningful conclu-
sions about cultural differences from
samples of patients drawn from a limited
number of health care settings in each
culture being studied. Prior cross-cul-
tural research on chronic pain has often
used samples of pain patients smaller
than the numbers available for the indi-
55. vidual centers participating in this
study. In this study, 10 of the 15 partici-
pating centers had over 100 patients
with persistent pain (Table 1). Most prior
cross-national studies of pain patients
Table 3.—Indicated Quality-of-Life Impairment by Persistent
Pain Status and the Adjusted Odds Ratio (OR)
for Impairment for Persons With vs Without Persistent Pain,
World Health Organization Psychological
Problems in General Health Care Survey, 1991-1992 (Weighted
Data)
Quality of Life Impairment
Persistent Pain
Present, %
Persistent Pain
Absent, %
Adjusted OR
(95% CI)* P
ICD-10 definition of anxiety
or depressive disorder†
33.7 10.1 4.14 (3.52-4.86) ,.001
Health status rated fair to poor‡ 33.4 20.9 1.26 (1.07-1.49) .006
Interviewer-rated interference
with work performance‡
31.4 13.0 2.12 (1.79-2.51) ,.001
$3 Activity-limitation days
56. in prior month‡
41.2 26.0 1.63 (1.41-1.89) ,.001
*The adjusted OR measures the risk of the indicated form of
impairment among persons with persistent pain
relative to those without persistent pain, after adjusting for
covariates. CI indicates confidence interval.
†The ORs were adjusted for center, age, sex, and physician-
rated severity of physical disease. ICD-10 indicates
International Statistical Classification of Diseases, 10th
Revision.
‡The ORs were adjusted for center, age, sex, physician-rated
severity of physical disease, and presence of an
anxiety or depressive disorder.
Table 4.—Indicated Quality-of-Life Impairment Comparing
Persons With and Without Persistent Pain, World Health
Organization Psychological Problems in Gen-
eral Health Care Survey, 1991-1992 (Weighted Data)*
Participating Center
Depressive or Anxiety
Disorder Health Rated Fair to Poor
Interviewer-Rated
Work Interference $3 Activity-Limitation Days
With Pain, % Without Pain, % With Pain, % Without Pain, %
With Pain, % Without Pain, % With Pain, % Without Pain, %
Santiago, Chile 60.1 27.9 31.5 13.3 25.5 13.1 9.0 12.0
57. Berlin, Germany 23.1 8.8 48.1 24.0 22.1 15.4 38.5 26.5
Rio de Janeiro, Brazil 52.5 20.1 11.6 11.9 15.2 13.2 38.7 28.4
Ankara, Turkey 29.4 5.3 46.9 27.2 13.1 3.1 39.3 22.8
Paris, France 34.7 15.8 22.8 12.8 19.4 13.6 27.4 22.6
Mainz, Germany 27.6 11.5 51.4 37.8 37.7 18.3 50.3 33.1
Groningen, the Netherlands 37.8 10.6 22.3 10.5 51.1 20.7 42.7
35.7
Manchester, England 41.0 14.5 36.7 11.3 100.0 8.6 71.8 29.9
Bangalore, India 36.0 9.3 43.1 29.5 33.5 11.9 61.5 37.9
Seattle, Wash 18.5 5.4 27.8 5.4 22.6 6.0 48.8 20.7
Verona, Italy 27.5 4.0 40.9 29.7 11.9 7.5 23.5 16.3
Shanghai, China 13.3 3.9 21.8 26.5 28.9 16.8 31.6 17.0
Athens, Greece 39.1 14.6 8.3 12.8 25.7 8.5 53.5 25.2
Nagasaki, Japan 12.8 5.8 44.1 38.1 37.0 9.3 49.6 21.7
Ibadan, Nigeria 26.7 5.2 6.9 13.3 21.4 26.6 46.7 40.6
All centers 33.7 10.1 33.4 20.9 31.4 13.0 41.2 26.0
*Percentages in boldface type indicate that the percentage with
the indicated quality-of-life impairment among persons with
persistent pain exceeds those without persistent
pain at P ,.05, for a 2-sided test. Significance of differences
between persons with and without persistent pain was tested by
59. ment procedures were used to reduce
variation due to study methods. How-
ever, it is difficult to guarantee uniform
application of study methods in a widely
dispersed multicenter study conducted
in many different languages. It was only
possible to study a limited number of care
settings in each locale, so differences due
to care setting are confounded with cul-
tural differences. For these reasons, our
results regarding differences in preva-
lence and impacts of persistent pain be-
tween countries are exploratory. While
the differences in prevalence rates across
centers were statistically significant af-
ter controlling for age and sex differ-
ences, this variation may be due to so-
ciodemographic, care setting, and/or
methodological differences rather than
culture.
The observation that women tended to
have elevated rates of persistent pain
relative to men has been reported by oth-
ers.17-20 This study does not shed light on
reasons for this sex difference, other than
to suggest that it is not unique to Western
societies. Prior research has suggested
sex differences in pain prevalence for
some anatomical sites and not for oth-
ers,19,20 but this study did not examine site-
specific prevalence rates by sex.
Overall, persistent pain was associated
with marked reductions in several differ-
60. ent indicators of well-being, particularly
psychological illness and interference
with activities. Differences in self-rated
health status were of smaller magnitude,
although they were statistically signifi-
cant for all centers combined. In the
pooled analysis, patients with persistent
pain were more likely to have impaired
work role functioning and to have missed
3 or more days from their usual activities
in the prior month. While patients with
persistent pain were more disabled than
those without persistent pain overall, this
association was not consistently statisti-
cally significant across the participating
centers. However, the trend was in the
same direction in almost every center for
both disability measures.
The commonly reported association of
persistent pain with psychological ill-
ness16,17,21,22 was confirmed by this study.
A significant association was found in ev-
ery participating center. This study does
not address the direction of causality
between persistent pain and affective
illness. Prior studies have yielded dif-
fering results on this question.16,23,24 The
results of this study indicate that psy-
chological disorder is a common corre-
late of persistent pain, and that this
association is observed in a wide range
of cultural settings.
In conclusion, persistent pain was com-
mon among primary care patients in many
61. different cultures. Across all centers, per-
sistent pain was associated with psycho-
logical disturbance and significant ac-
tivity limitations. Further research is
needed to better understand cross-na-
tional variation in the prevalence of per-
sistent pain, and variation in the effects of
persistent pain on well-being and func-
tioning. The results of this study point to
the difficulty in drawing conclusions about
cultural differences in the frequency or
the impacts of persistent pain from mod-
est samples of pain patients sampled from
a limited number of care settings in a par-
ticular culture. While further research is
needed, this study shows that persistent
pain is a common problem among primary
care patients in a wide range of cultural
settings.
Dr Von Korff’s work on this report was supported
in part by grant DE08773-10 from the National In-
stitutes of Health, Bethesda, Md.
The data reported in this article were collected as
part of a World Health Organization’s Psychological
Problems in General Health Care project. Partici-
pating investigators include O. Ozturk and M.
Rezaki, Ankara, Turkey; C. Stefanis and V.
Mavreas, Athens, Greece; S. M. Channabasavana
and T. G. Sriram, Bangalore, India; H. Helmchen
and M. Linden, Berlin, Germany; W. van der Brink
and B. Tiemens, Groningen, the Netherlands; M.
Olatawura, Ibadan, Nigeria; O. Benkert and W.
Maier, Mainz, Germany; S. Kisely, Manchester,
England; Y. Nakane and S. Michitsuji, Nagasaki,
62. Japan; Y. Lecrubier and P. Boyer, Paris, France; J.
Costa e Silva and L. Villano, Rio de Janeiro, Brazil;
R. Florenzano and J. Acuna, Santiago, Chile; G. E.
Simon, Seattle, Wash; Y. He-Quin and X. Shi Fu,
Shanghai, China; and M. Tansella and C. Bellan-
tuono, Verona, Italy. The study advisory group
include J. Costa e Silva, D. P. Goldberg, Y. Lecru-
bier, Michael Von Korff, and H-U Wittchen. Coor-
dinating staff at World Health Organization head-
quarters include N. Sartorius and T. B. Ustun.
References
1. Schappert SM. National Ambulatory Medical
Care Survey: 1989 summary. National Center for
Health Statistics. Vital Health Stat 13. 1992; No. 11.
2. Turk DC, Rudy TE. Toward an empirically de-
rived taxonomy of chronic pain patients: integration
of psychological assessment data. J Consult Clin
Psychol. 1988;56:233-238.
3. Von Korff M, Ormel J, Keefe FJ, Dworkin SF.
Grading the severity of chronic pain. Pain. 1992;50:
133-149.
4. Bates MS, Rankin-Hill L, Sanchez-Ayendez M, Men-
dez-Bryan R. A cross-cultural comparison of adapta-
tion to chronic pain among Anglo-Americans and na-
tive Puerto Ricans. Med Anthropol. 1995;16:141-173.
5. Moore R. Ethnographic assessment of pain cop-
ing perceptions. Psychosom Med. 1990;52:171-181.
6. Sanders SH, Brena SF, Spier CJ, Beltrutti D,
McConnell H, Quintero O. Chronic low back pain
patients around the world: cross-cultural similari-
ties and differences. Clin J Pain. 1992;8:317-323.
7. Ustun TB, Sartorius, N, eds. Mental Illness in
General Health Care: An International Study. New
York, NY: John Wiley & Sons Inc; 1995.
63. 8. Sartorius N, Ustun TB, Costa e Silva JA, et al. An
international study of psychological problems in pri-
mary care. Arch Gen Psychiatry. 1993;50:819-824.
9. Von Korff M, Ustun TB. Methods of the WHO
Collaborative Study on Psychological Problems in
General Health Care. In: Ustun TB, Sartorius N,
eds. Mental Illness in General Health Care: An In-
ternational Study. New York, NY: John Wiley &
Sons Inc; 1995.
10. Ormel J, Von Korff M, Ustun TB, Pini S, Korten
A, Oldehinkel T. Common mental disorders and dis-
ability across cultures. JAMA. 1994;272:1741-1748.
11. Goldberg D, Williams P. A Users’ Guide to the
General Health Questionnaire: GHQ. Windsor,
Berkshire, England: NFER-NELSON Publishing
Co Ltd; 1988.
12. World Health Organization. Composite Inter-
national Diagnostic Interview. Geneva, Switzer-
land: World Health Organization, Division of Men-
tal Health; 1989. Publication MNH/NAT/89.
13. World Health Organization. International Sta-
tistical Classification of Diseases, 10th Revision
(ICD-10). Geneva, Switzerland: World Health Or-
ganization; 1992.
14. Wiersma D, DeJong A, Ormel J. The Groningen
Social Disability Schedule: development, relation-
ship with ICIDH, and psychometric properties. Int
J Rehabil Res. 1988;11:213-224.
15. Von Korff M, Ustun TB, Ormel J, Kaplan I, Si-
mon G. Self-report disability in an international pri-
mary care study of psychological illness. J Clin Epi-
demiol. 1996;49:297-303.
16. Von Korff M, Simon G. The relationship of pain
and depression. Br J Psychiatry. 1996;168(suppl 30):
101-108.
64. 17. Magni G, Marchetti M, Moreschi C, Merskey H,
Luchini SR. Chronic musculoskeletal pain and de-
pressive symptoms in the National Health and Nu-
trition Examination, I: epidemiological follow-up
study. Pain. 1993;53:163-168.
18. Andersson HI, Ejlertsson G., Leden I, Rosen-
berg C. Chronic pain in a geographically defined
general population: studies of differences in age,
gender, social class, and pain localisation. Clin J
Pain. 1993;9:174-182.
19. Von Korff M, Dworkin SF, LeResche L, Kruger
A. An epidemiologic comparison of pain complaints.
Pain. 1988;32:173-183.
20. Le Resche L, Von Korff M. Epidemiology of
chronic pain. In: Block AR, Kremer EF, Fernandez
E, eds. Handbook of Pain Syndromes: Biopsycho-
social Perspectives. Mahwah, NJ: Lawrence Er-
baum Associates Inc. In press.
21. Roy R, Thomas M, Matas M. Chronic pain and
depression: a review. Compr Psychiatry. 1984;25:
96-105.
22. Romano JM, Turner JA. Chronic pain and de-
pression: does the evidence support a relationship?
Psychol Bull. 1985;97:18-34.
23. Polatin PB, Kinney RK, Gatchel RJ, Lillo E,
Mayer TG. Psychiatric illness and chronic low-back
pain: the mind and the spine—which goes first?
Spine. 1993;18:66-71.
24. Von Korff M, LeResche L, Dworkin SF. First
onset of common pain symptoms: a prospective
study of depression as a risk factor. Pain. 1993;55:
251-258.
JAMA, July 8, 1998—Vol 280, No. 2 Persistent Pain and Well-
being—Gureje et al 151
66. To the shame of the pharmaceutical industry, some compa-
nies have suppressed data from publication4 or published data
in a manner not consistent with that reviewed by the FDA.5
Such cases have become public embarrassments that received
widespread coverage in the popular press. These examples
are the exception, not the rule. Physicians and the public should
be assured of the validity of pharmaceutical clinical research
and of the integrity of those who conduct and oversee it.
Kenneth J. Gorelick, MD
DuPont Merck Pharmaceutical Company
Wilmington, Del
1. Barnes DE, Bero LA. Why review articles on the health
effects of passive smoking
reach different conclusions. JAMA. 1998;279:1566-1570.
2. Cho MK, Bero LA. The quality of drug studies published in
symposium proceed-
ings. Ann Intern Med. 1996;124:485-489.
3. Rochon PA, Gurwitz JH, Simms RW, et al. A study of
manufacturer-supported tri-
als of nonsteroidal anti-inflammatory drugs in the treatment of
arthritis. Arch Intern
Med. 1994;154:157-163.
4. Rennie D. Thyroid storm [published correction appears in
JAMA. 1997;277:1762].
JAMA. 1997;277:1238-1243.
5. Wenzel RP. Anti-endotoxin monoclonal antibodies—a second
look. N Engl J Med.
1992;326:1151.
In Reply.—Every independent scientific body that has re-
viewed the scientific evidence has concluded that exposure to
passive smoke is harmful to health.1 As the title of our article
suggests, the goal of our study was to determine why many
67. published review articles reach conclusions that differ from
these independent scientific bodies. We investigated several
factors in addition to quality and funding source that might be
associated with outcome, including peer review, date of pub-
lication, and topic of review. The only factor associated with
the conclusion of a review article was the affiliation of its au-
thor: 94% of reviews by tobacco industry–affiliated authors
concluded that passive smoking is not harmful compared
with 13% of reviews by authors without tobacco-industry
affiliations.
We did not define tobacco-industry affiliation “generously,”
as Dr Heck states. Rather, we used strict, well-defined crite-
ria: an author must have received funding from the tobacco
industry, submitted a statement on behalf of the tobacco in-
dustry regarding the Environmental Protection Agency’s risk
assessment on passive smoking, or participated in (not simply
attended, as misstated by Heck) at least 2 tobacco-industry–
sponsored symposia. Moreover, our data are not “testament to
the legitimate diversity of scientific opinion on the topic of
ETS,” as Heck suggests. Rather, our findings suggest that,
among scientists who are not affiliated with the tobacco in-
dustry, there is consensus that passive smoking is harmful.
The only diversity of opinion comes from the authors with
tobacco-industry affiliations.
Dr Gorelick takes issue with 2 articles we cited to support
our statement that original research articles sponsored by the
pharmaceutical industry tend to draw proindustry conclu-
sions. A careful reading of both articles reveals that they do
support our statement. The article by Cho and Bero2 did not
examine “only articles from symposia sponsored by single drug
companies.” It examined 127 articles from symposia (of which
39% were sponsored by a single drug company) and 45 articles
from peer-reviewed journals: 98% of the articles with drug
company support favored the drug of interest compared with
68. 79% of the articles without drug company support (P,.01).
In the article by Rochon et al,3 the analysis was limited to
manufacturer-associated trials due to “the scarcity of non-
manufacturer-associated trials.” However, the authors found
that the manufacturer-associated drug was reported as com-
parable or superior to the comparison drug in all cases and that
the claims often were not supported by trial data.
We do not suggest that industry-sponsored research is al-
ways biased. However, to our knowledge, every study that has
examined the relationship between sponsorship and outcomes
has found that industry-sponsored research is more likely to
draw proindustry conclusions than non–industry-sponsored
research. We recommend that financial interests always
should be disclosed and that readers of research articles should
consider these disclosures when deciding how to judge an ar-
ticle’s conclusions.
Lisa A. Bero, PhD
University of California, San Francisco
Deborah Barnes
University of California, Berkeley
1. Barnes DE, Bero LA. Why review articles on the health
effects of passive smoking
reach different conclusions. JAMA. 1998;279:1566-1570.
2. Cho MK, Bero LA. The quality of drug studies published in
symposium proceed-
ings. Ann Intern Med. 1996;124:485-489.
3. Rochon PA, Gurwitz JH, Simms RW, et al. A study of
manufacturer-supported tri-
als of nonsteroidal anti-inflammatory drugs in the treatment of
arthritis. Arch Intern
Med. 1994;154:157-163.