Objective Psychological factors are known to be associated with functional gastrointestinal disorders (FGIDs) including irritable bowel syndrome (IBS) and functional dyspepsia (FD). No prospective studies have evaluated whether it is the brain (eg, via anxiety) that drives gut symptoms, or whether gut dysfunction precipitates the central nervous system features such as anxiety. In a 12-year longitudinal, prospective, population-based study, we aimed to determine the directionality of the brain–gut mechanism in FGIDs.
Design Participants (n=1775) were a random population sample from Australia who responded to a survey on FGIDs in 1997 and agreed to be contacted for future research; 1002 completed the 12-year follow-up survey (response rate =60%), with 217, 82 and 45 people meeting Rome II for new onset FGIDs, IBS and FD, respectively. Anxiety and depression were measured using the Delusions Symptom States Inventory at baseline and follow-up.
Results Among people free of a FGID at baseline, higher levels of anxiety (OR 1.11; 95% CI 1.03 to 1.19, p=0.006) but not depression at baseline was a significant independent predictor of developing new onset FGIDs 12 years later. Among people who did not have elevated levels of anxiety and depression at baseline, those with a FGID at baseline had significantly higher levels of anxiety and depression at follow-up (mean difference coefficient 0.76, p<0.001 and 0.30, p=0.01 for anxiety and depression, respectively). In IBS higher levels of anxiety and depression at baseline were predictive of IBS at follow-up, while only depression was predictive of FD at follow-up.
Conclusions The central nervous system and gut interact bidirectionally in FGIDs.
- Irritable bowel syndrome
- functional dyspepsia
- brain-gut axis
- psychosomatic medicine
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
- Irritable bowel syndrome
- functional dyspepsia
- brain-gut axis
- psychosomatic medicine
Significance of this study
What is already known about this subject?
Irritable bowel syndrome (IBS) and functional dyspepsia (FD) are common disorders that persist lifelong with incidence rates up to 10 years between 4% and 15%.
There is a role for psychological distress in functional gastrointestinal disorders (FGIDS), including IBS and FD.
There is also some evidence that gut dysfunction (perhaps driven by low-grade inflammation secondary to increased permeability) is likely to be key in a subset of people with IBS and FD.
What are the new findings?
The central nervous dysfunction in FGIDs may be bidirectional.
In IBS and FD, it appears that the brain–gut pathway is dominant.
How might it impact on clinical practice in the foreseeable future?
It is clear that IBS and FD are chronic and while there is waxing and waning of symptoms they do not disappear on their own, making them a clear target for treatment.
In IBS and FD early identification and treatment of psychological distress in people presenting with abdominal pain like symptoms may help prevent the subsequent development of these conditions. In addition, adequate management and treatment of FGIDs may help to prevent subsequent development of anxiety and depression.
The functional gastrointestinal disorders (FGIDs) manifest as a characteristic combination of chronic or recurrent symptoms arising from the gastrointestinal (GI) tract.1 FGIDs are extremely common, affecting at least one-third of the population.2 Half of all patients with gastrointestinal complaints seen in primary care have FGIDs, and up to half of all referrals to specialist gastroenterologists are for FGIDs.3 ,4 FGIDs generally persist lifelong5–8 and significantly impair quality of life.9
The ‘little brain’ in the GI tract's enteric nervous system is intimately linked to the central nervous system, and this linkage may be key in the pathogenesis of FGIDs. Multiple pathophysiological disturbances have been identified in subsets with these disorders, including disordered motility, visceral hypersensitivity, abnormal gas handling, disturbed intestinal flora and low-grade intestinal inflammation (particularly mast cell and T-lymphocyte infiltration).10–17 A cytokine response has been documented in irritable bowel syndrome (IBS) and functional dyspepsia (FD), and tumour necrosis factor α (TNFα) levels have correlated with anxiety; cytokines have been postulated to drive psychological distress in the FGIDs.16 ,17 However, a unifying theory that explains all of the FGIDs has remained elusive, although there is emerging agreement that gut dysfunction (perhaps driven by low-grade inflammation secondary to increased permeability) is likely to be key in a subset of people.
Another promising unifying theory to explain FGIDs is primary central nervous system dysfunction, which clinically manifests as psychological distress and may drive GI distress. A plethora of literature supports a role for psychological factors, including anxiety, depression and somatisation in FGIDs.18–41 A number of studies have shown elevated psychological distress (anxiety, depression and somatisation) in outpatients and volunteers with FGIDs,19–21 and among non-consulters from the community with FGIDs.2 ,22–28 A causal link is supported by symptom improvement following psychological interventions such as cognitive behavioural therapy versus usual care or placebo.29 Randomised controlled trials have confirmed that antidepressants, both tricyclics and serotonin reuptake inhibitors, are superior to placebo in IBS, although their role in FD remains uncertain and their exact mechanism of action is unknown.30–32 Pathophysiological studies33–38 and longitudinal studies have shown an association of FGIDs with stress,39–41 although the majority of these studies only assessed IBS and FD, have been retrospective and have not been methodologically rigorous enough to ascertain if there is a causal connection. Acute infectious gastroenteritis can precipitate IBS and FD presumably in people genetically predisposed to these conditions,42 and in those who develop post-infectious IBS, the risk appears to be increased if there is pre-existing psychological distress.43–45
Thus, there is evidence that a brain to gut pathway may account for GI symptoms in some FGIDs, but other evidence points to the gut being the primary driver of symptoms (via gut to brain connections perhaps through cytokines). However, it is not known which of the various FGIDs, if any, is dominant. In this study, we aimed to determine whether it is likely that a brain–gut or a gut–brain pathway operates in the FGIDs. We also aimed to explore if the brain–gut or gut–brain pathways are the same for all FGIDs by looking at these pathways in IBS and FD separately. We hypothesised that anxiety and depression drive the new onset of all FGIDS, including IBS and FD.
Subjects (n=1775) aged 18 years and older who had participated in a random population survey of 4500 persons 12 years earlier and who agreed to be recontacted for future research were included in this study.2 The sampling frame for the initial study consisted of the 1996 electoral roll for the local government area of Penrith, which represents 3.6% of the Sydney population. Based on 1996 census data,46 the population of Penrith is representative of the Australian population as a whole in terms of its sociodemographics and ethnic composition, except that its inhabitants are slightly younger and have a slightly higher socioeconomic status.
Of those people recontacted, 1002 completed the 12-year follow-up survey. FGID cases were defined as those (n=376) who met Rome II criteria1 for at least one FGID diagnosis (globus, rumination syndrome, functional chest pain, functional heartburn, functional dysphagia, FD, aerophagia, IBS, functional abdominal bloating, functional constipation, functional diarrhoea, functional abdominal pain syndrome, gallbladder dysfunction, sphincter of oddi dysfunction, functional incontinence, functional anorectal pain, levator ani syndrome, proctalgia fugax, pelvic floor dyssynergia) at baseline. Healthy controls were people who did not meet criteria for any self-reported Rome II FGID (listed above) diagnosis at baseline (n=626).
Functional gastrointestinal disorders
Any FGID was defined by applying validated Rome II criteria.1 Subjects had to meet at least one FGID diagnosis (listed above). To meet criteria for a FGID at follow-up we also excluded self-reported organic diseases, including peptic and duodenal ulcer, pancreatitis, gastric, colon and oesophageal cancer, coeliac disease and inflammatory bowel disease.
Irritable bowel syndrome
IBS was defined as abdominal pain in the last 12 months that occurred for a period of 3 months or more and that was associated with two out of three of the following features: relieved with defecation, onset associated with a change in frequency of stool and/or onset associated with a change in form of stool.1 To meet criteria for IBS at follow-up we also excluded self-reported organic diseases, including coeliac disease, inflammatory bowel disease, colon cancer, peptic and duodenal ulcer, pancreatitis and gastric and oesophageal cancer. IBS controls were those who did not have IBS at baseline.
FD was defined as upper abdominal pain or discomfort in the last 12 months that occurred for a period of 3 months.1 To meet criteria for functional dyspepsia at follow-up we excluded the following self-reported organic diagnoses: peptic and duodenal ulcer, pancreatitis, and gastric and oesophageal cancer. FD controls were those who did not have FD at baseline.
The study was approved by the Sydney West Area Health Service Human Research Ethics Committee. The Australian Electoral Commission provided us with up-to-date addresses for subjects at the beginning of the study. We then sent subjects a participant information sheet, the follow-up survey and a reply paid envelope. The study was conducted in two parts. A pilot study (n=450) was conducted to test the effectiveness of including a lottery ticket with the survey versus sending participants a lottery ticket on return of a completed survey versus no lottery ticket. The results indicated that including a lottery ticket produced more completed surveys. Thus, the rest of the subjects (n=1325) received a lottery ticket with their survey. The follow-up protocol for non-responders included a reminder/thank-you letter sent to all subjects 1 week later, a replacement survey sent to non-responders only at week 4 and a thank-you/reminder letter sent to non-responders at week 5. Additional measures to maximise response rates included a personalised cover letter personally signed by the chief investigator, clear affiliation with the University of Sydney and Nepean Hospital, an easy to understand attractive coloured questionnaire booklet with probes to skip to the next section if questions do not apply, postage stamps on envelopes and inclusion of a reply-paid envelope.
The original survey
The original survey was 32 pages long and comprised a series of questions on functional GI symptoms. The questions were initially framed over the past 12 months, but include questions on the presence, frequency, duration and severity of symptoms over the past 3 months (‘Did this keep happening for a period of 3 months or more?’), which enabled a diagnosis of 19 FGIDS to be made based on slightly modified Rome II criteria. This questionnaire has been repeatedly tested and very carefully validated47 with little missing data observed for the individual questions (1.2–3%).2
The Delusion Symptom States Inventory (DSSI) was included to measure anxiety and depression.48 This scale contains seven items for anxiety and seven items for depression and was originally chosen for the 1997 survey because it has published clinical cutoffs and is a strong clinical measure that is well validated.48 ,49
Demographics in this questionnaire included gender, age, educational level and country of birth.
The follow-up survey
While minor changes were made to the follow-up survey to ensure that FGIDs can also be diagnosed using more recent versions of the Rome criteria, all the Rome II questions from the baseline survey were included. The questionnaire also contained the DSSI48 and basic demographic questions for cross-checking purposes.
Of the 2910 subjects who agreed to take part in the baseline study, 1775 agreed at that time to be recontacted for follow-up and all of these subjects were approached. A total of 1002 subjects returned the follow-up questionnaire yielding a response rate of 60% (taking into account deaths and return to senders which were taken out of the denominator).
Prevalence values are presented with exact 95% CIs. Given the 12-year interval between baseline and follow-up, an analysis was undertaken of response rate and potential responder bias. To assess the potential for responder bias we compared the n=1002 who returned the follow-up questionnaire with the n=2910 in the baseline sample with respect to key variables that are available. These comparisons are reported below.
Age and gender are controlled for in the primary analyses of both studies. All p values were two tailed and statistical significance was defined as p<0.05.
In the gut–brain pathway multiple regression was used in which the outcome was DSSI anxiety or depression at 12 years, in separate models, and the primary independent variable was baseline FGID, IBS or FD status. Age and gender were included as covariates to provide statistical control of these extraneous factors. Statistical inference (hypothesis tests) were based on log10-transformed DSSI measures due to violation of the normal assumption of this model. Subjects with demonstrable mood disorder, defined as mood score ≥4 for anxiety or depression, at baseline were excluded from this analysis.
In the brain–gut pathway unconditional logistic regression was used with FGID, IBS or FD status at 12 years as the outcome variable, in separate models, and DSSI anxiety or depression as the primary independent variable, also in separate models. Age and gender were included as covariates to provide statistical control of these extraneous factors. Subjects meeting criteria for FGID, IBS or FD at baseline were excluded from the respective model.
Responders versus non-responders
To investigate whether our sample contained any systematic bias we compared responders to the original survey sent to 4500 people (n=2910) and non-responders (n=1590) on gender and found slightly more women (51.9%) compared with men (45.9%). No other information was available for comparing responders and non-responders to the original survey.
We then compared responders to the 12-year survey (n=1002) with non-responders (n=1908), that is the remaining people from the original sample who responded to the original survey but not the 12-year survey. We found that responders to the 12-year follow-up survey were similar in age (mean age 45.1 vs 45.1 years), gender (women 51.4% vs 52.1%) and nationality (Australian born 76.6% vs 73.3%) compared with non-responders. However, we did find that significantly more responders to the 12-year survey had achieved a high school or better education (58.5% vs 54.0%) and were more likely to meet criteria for a FGID at baseline (37.4% vs 24.0%).
Overall there were no other major systematic differences for people who responded to the 12-year survey versus those who did not.
Baseline cases and controls were relatively similar in terms of age, nationality and educational level, although slightly more cases were women and had significantly higher levels of anxiety and depression compared with controls (table 1).
Incidence and disappearance of FGID, IBS and functional dyspepsia
We found 217 out of 626 (34.7%; 95% CI 30.9 to 38.5) controls developed a FGID diagnosis over the 12-year period. Of subjects free of a FGID at baseline, we found 8.8% (n=82) and 4.7% (n=45) developed IBS and FD, respectively, 12 years later (table 2). The number of subjects who developed new onset FGIDs, IBS or FD was similar to the number of subjects whose symptoms resolved over the 12-year period, and thus the prevalence of FGIDS was stable over the time frame (table 2).
Anxiety and depression as predictors of new onset FGIDs
Among subjects free of a FGID at baseline, a clinically elevated level of anxiety but not depression at baseline was a significant predictor of developing a new onset FGID 12 years later. This was unchanged after controlling for age and gender in the analysis (table 3).
For IBS, we also found higher levels of anxiety and depression at baseline to be predictive of developing IBS at 12-year follow-up among people free of IBS at baseline (table 3).
For FD, we found higher levels of depression but not anxiety at baseline to be predictive of developing FD at 12-year follow-up among people free of FD at baseline (table 3).
Correlation of baseline mood and follow-up GI symptoms
We did not find baseline anxiety or depression to be significantly correlated with any of the following follow-up FGID symptoms: upper abdominal pain severity (r=−0.01, p=0.9; r=0.06, p=0.5), lower abdominal pain severity (r=0.06, p=0.5; r=0.003, p=0.7), early satiety (r=−0.03, p=0.7; r=0.03, p=0.7), postprandial fullness (r=0.002, p=1.0; r=−0.01, p=0.9), bloating (r=0.1, p=0.2; r=0.02, p=0.8), hard/lumpy stools (r=0.03, p=0.6; r=0.04, p=0.4) or loose/water stools (r=0.09, p=0.08; r=0.09, p=0.08), respectively.
FGIDs as a predictor of new onset psychological distress
We found a FGID diagnosis at baseline to be significantly associated with higher levels of anxiety and depression at follow-up among subjects who did not have elevated levels of psychological distress at baseline (table 4).
For individual FGIDS, we did not find an IBS or FD diagnosis at baseline to be significantly associated with clinically elevated anxiety and depression at follow-up among subjects who did not have elevated levels of psychological distress at baseline (table 4).
Correlations of baseline symptoms with follow-up mood
As shown in figure 1, we found baseline abdominal symptoms to be significantly but modestly correlated with follow-up anxiety and depression, including lower abdominal pain severity and early satiety, respectively. We found upper abdominal pain to be significantly correlated with follow-up depression but not anxiety, although the magnitude of the correlation was only small. Similarly we found baseline postprandial fullness to be significantly but modestly correlated with depression but not anxiety. However, baseline bloating was significantly but modestly correlated with anxiety but not depression. None of the bowel symptoms were significantly correlated with anxiety or depression.
Here we report a novel, comprehensive, prospective longitudinal population-based study assessing the relationship between psychological factors and FGIDs. The primary aim of the study was to determine whether it is the brain that drives the onset of functional gut symptoms, or whether in FGIDs perhaps the gut drives the brain as manifest by psychological distress, particularly anxiety or depression. We found support for both the brain–gut and gut–brain hypothesis in FGIDs. For example, among subjects who did not report having a FGID at baseline, we observed that those who had higher levels of anxiety at baseline (but not depression) were significantly more likely to develop a FGID 12 years later. However, baseline anxiety and depression were not significantly correlated with any specific GI symptom at follow-up. We also found support for the alternative gut–brain hypothesis in FGIDs. That is, we observed among subjects who did not have clinically elevated levels of psychological distress at baseline, a FGID diagnosis at baseline was significantly associated with higher levels of subsequent anxiety and depression at follow-up. For specific GI symptoms, we observed that abdominal pain, postprandial fullness, early satiety and bloating, but not bowel symptoms, were significantly correlated with follow-up anxiety and depression, although the associations were all modest.
As part of the study we examined the incidence of any FGID and IBS and FD over a 12-year period. We found that 34% of subjects developed a new onset FGID over 12 years; a total of 9% developed new onset IBS. This onset rate for IBS was lower than reported by others,5–8 including Ford et al in a 10-year population-based follow-up study who found 15% developed new onset IBS; however, their study population was older and they used the less stringent Manning criteria for IBS.8 The incidence of FD was 4%, which is similar to previously published figures.5 ,6 We also showed that the number of people who did not meet criteria for a FGID at follow-up was similar to the new onset of these conditions, indicating that the prevalence of FGIDs is remarkably stable over time. This confirms previous longitudinal studies of up to 12 years, showing that the prevalence of IBS and FD is relatively stable over time in a population, although fluctuations in both the presence and severity of symptoms are common.5–8
These data provide some of the first direct evidence to indicate that the brain–gut pathway is bidirectional; that is, both brain–gut and gut–brain dysfunction may be occurring in FGIDs. While numerous studies have found an association between psychological distress and FGIDs,18–41 supporting a potential brain to gut mechanism, the majority of these studies were cross sectional and did not allow for true causal associations to be assessed. Of the few prospective studies that have been conducted, only IBS has been assessed. Gwee et al evaluated 75 patients admitted to hospital for gastroenteritis.43 Of the 22 patients who subsequently developed IBS symptoms, they reported higher scores for anxiety, depression, somatic distress and neuroticism at their hospital admission compared with those who returned to normal bowel function. These findings remained significant even after controlling for the confounding effects of having an acute illness. However, the inclusion of hospital patients with post-infectious IBS makes it difficult to generalise these findings to all people with IBS. Koloski et al prospectively evaluated 361 patients with unexplained abdominal pain and 120 controls with a questionnaire every 4 months over a 12-month period.39 Psychological distress was higher in controls who reported having GI symptoms compared with those who did not at the 4-month, 8-month and 12-month follow-ups, reaching significance at the 8-month and 12-month follow-ups. However, the short follow-up time, failure to include subjects meeting the threshold for IBS, and smallish sample sizes make it difficult to generalise the findings to FGIDs.
The gut–brain finding in FGIDs in this study is novel but deserves further replication. One possible explanation is low-grade intestinal inflammation with mast cell infiltration and activation sets off the release of cytokines and chemokines into the circulation that alter central nervous system functioning, inducing anxiety. Thus TNFα levels have been shown to correlate with anxiety in IBS, but more data are needed.16
We also aimed to explore whether the brain–gut mechanism is the same in two specific FGIDs, although here the sample sizes were smaller. In IBS, we found clear-cut support for the brain–gut hypothesis; that is, that psychological distress drives IBS symptoms. For example, among subjects who did not report having IBS at baseline, we observed that those who had higher levels of anxiety and depression at baseline were significantly more likely to develop IBS 12 years later. The sample size was too small to explore IBS subgroups. Gwee et al also showed that psychological distress predates post-infectious IBS.43 However, their sample was limited to hospital patients with IBS who had severe gastroenteritis at baseline.
We found similar results for FD. Among subjects who did not report having FD at baseline, we observed that those who had higher levels of depression at baseline were significantly more likely to develop FD 12 years later. However, this was not the case for anxiety. These findings are in line with an earlier prospective study by Koloski et al. They found that psychological distress was significantly higher in controls who reported having GI symptoms, including abdominal pain, compared with those who did not at the 8-month and 12-month follow-ups.39 However, this study was limited by the failure to include people meeting Rome criteria for FD and a short follow-up time. Our study extends these results by applying a longer 12-year follow-up and including people who met Rome II criteria for FD, although the diagnosis was not endoscopically confirmed. In contrast, there was no evidence for the alternative gut–brain hypothesis in IBS or FD. Thus it appears that higher levels of psychological distress may be a precursor to the subsequent development of IBS and FD symptoms, but not vice versa.
For FGIDs as a whole group, we observed that the brain–gut and gut–brain mechanisms appear to be in operation. It is possible that the gut–brain direction may be dominant in other individual FGIDS and this is driving the association; however, this needs further investigation.
This study was methodologically superior compared with previous longitudinal work. We prospectively followed up a community sample of healthy controls and cases to see if psychological factors measured at baseline predicted the subsequent development of FGIDS, including IBS and FD, over a 12-year period, and vice versa. The original sample was randomly selected from the Australian electoral roll, which by law requires that people aged 18 years and over be registered. Therefore, the results should be generalisable and unbiased.46 The original sample responded to the first questionnaire with a 72% response rate.2 We then achieved a 60% response rate to our follow-up questionnaire 12 years later. We did not find any major systematic differences between responders and non-responders, although responders to the 12-year follow-up were more likely to have a FGID diagnosis at baseline. We also used well validated measures of psychological symptoms, including the DSSI which provides published cutoffs for clinically elevated levels of distress.48 ,49 However, we do not know whether these findings would also apply to other psychological factors, particularly somatisation. FGIDS were diagnosed according to modified but well accepted Rome II criteria1 using a validated questionnaire assessing functional GI symptoms.47 Moreover, we excluded subjects who had self-reported organic diagnoses that may have been able to explain FGID symptoms, helping ensure that the FGIDs were not misclassified. We also found very low overlap rates (of 1%) between IBS and FD at baseline and follow-up in this particular cohort.
There are several implications resulting from this study. First, FGIDs in the community are chronic and while there is a waxing and waning of symptoms they do not usually disappear on their own, making them an important target for treatment. Second, in the FGIDs, including IBS and FD, the hypothesis that early identification and treatment of psychological distress in patients presenting with abdominal pain like symptoms will help prevent the subsequent development of persistent FGIDs now needs testing. In addition, adequate management and treatment of FGIDs may also help to prevent subsequent development of anxiety and depression.
In conclusion, it appears that central nervous dysfunction in the FGIDs may be bidirectional. In IBS and FD it also appears that the brain–gut pathway is dominant.
Competing interests None.
Ethics approval Ethics approval was provided by Sydney West Area Health Service Human Research Ethics Committee.
Provenance and peer review Not commissioned; externally peer reviewed.