INTRODUCTION

Bile acid diarrhoea (BAD) is a chronic bowel disorder according to the Rome IV criteria¹ with a population prevalence of at least 1%.² Gastrointestinal (GI) symptoms overlap with those of diarrhoea-predominant irritable bowel syndrome (IBS-D) and functional diarrhoea, characterised by urgent, frequent, watery bowel movements.³ In primary BAD, also known as Type 2 BAD, pathophysiology involves the overproduction of hepatic bile acids⁴ due to a deficiency in fibroblast factor 19 to inhibit biosynthesis.⁵ Bile acids conjugate with taurine and glycine in the liver to increase efficient facilitation of the digestion and absorption of dietary fats across the small intestine before actively returning to the liver. Reabsorption efficiency of less than 95% from the ileum leads to higher concentrations of bile salts entering the colon, resulting in a secretory response and induction of diarrhoea to prevent mucosal damage.⁶ Secondary BAD is caused by disruption to the entero-hepatic circulation of bile acids as a result of disease or resection in the ileum (Type 1 BAD) or other GI conditions (Type 3 BAD).

Diet is not linked to any diagnostic criteria for BAD.^{7, 8} However, amongst people with BAD food appears to be an important factor in symptom control. In a cross-sectional study undertaken by the UK national charity for BAD, specific foods to control symptoms were avoided by 85.7% of responders (78 out of 91 participants analysed amongst the first 100 responders).⁹ Dairy was cited as the most frequently avoided food. “High-fat foods” were the second most frequent “trigger food”. However, many dairy products such as milk, cream and cheese are also high-fat foods, which makes interpretation difficult and further insight is needed. Vegetables, and specifically onion and garlic, were judged as “a large trigger group of foods for BAD sufferers”.⁹ This suggests that non-digestible carbohydrates worsen symptoms of BAD, potentially as a result of colonic hypersensitivity to distention and an increase in small bowel water content.¹⁰ This was objectively evidenced in adults with IBS and bloating but not diarrhoea via magnetic resonance imaging of the small and large intestines.¹¹ However, the prevalence of BAD in patients with IBS-D and functional diarrhoea has been found to range from 16.9% to 35.3% (pooled rate 28.1%; 95% confidence interval (CI) = 22.6%–34%).¹² A systematic review of the effectiveness of diet therapy for management of BAD symptoms involving eight cohort studies and 192 adults identified very-low certainty evidence of reducing fat intake and modifying carbohydrate intake.¹³ With such paucity of evidence on the effect of dietary or nutritional components on BAD symptoms, further exploratory research is needed amongst those living with BAD.

The primary objective of the present study was therefore to identify the prevalence of food intolerance in UK adults with BAD. Secondary objectives were to investigate the prevalence of food items perceived to provoke BAD symptoms, diarrhoea, abdominal pain, bloating, flatulence and consequential food avoidance, as well as food group avoidance.

METHODS

RESULTS

Food intolerance by individual symptoms and consequential avoidance

The prevalences of provocation of diarrhoea, abdominal pain, bloating and flatulence in those with a clear pattern of BAD symptoms after food item ingestion as a proportion of the whole cohort are shown in Figure 1. The prevalence of perceived provocation of diarrhoea after food ingestion ranged between 3.7% (potato, 16/434; 95% CI = 2.1–5.9) and 41.0% (take-away food, 178/434; 95% CI = 36.4–45.8). For abdominal pain, prevalence ranged from 5.3% (potato, 23/434; 95% CI = 3.4–7.8) to 38.2% (larger meals, 166/434; 95% CI = 33.7–43.0). For bloating, prevalence ranged between 5.1% (potato, 22/434; 95% CI = 3.2–7.6) and 37.3% (larger meals, 162/434; 95% CI = 32.8–42.1). For flatulence, prevalence ranged from 5.1% (potato, 22/434; 95% CI = 3.2–7.6) to 33.2% (take-away food, 144/434; 95% CI = 28.8–37.8).

Consequential food avoidance ranged between 2.3% (potato, 10/434; 95% CI = 1.1–4.2) and 44.7% (creamy sauces, 194/434; 95% CI = 40.0–49.5) (Figure 2). The top 10 food intolerances by individual symptom and consequential avoidance are shown in the Supporting information (Table S2).

DISCUSSION

Almost 90% of 434 participants reported that food affected their BAD symptoms. The most prevalent intolerances were to take-aways and fish and chips, each affecting half of the whole cohort. Thereafter, in ranked descending order, larger meals, creamy sauces, a large quantity of fruit, pastry or pasties, pizza or garlic bread, cream, ice cream and full-fat milk provoked overall GI symptoms in 47%–34% of participants. Nine of these 10 items were higher-fat foods, of which five were meals and four were full-fat dairy products. Diarrhoea affected a greater proportion of participants for 13 of the 39 foods compared to abdominal pain, bloating and flatulence. However, frequency differences between these symptoms were generally small and overall, food intolerances involved all of these GI symptoms. A notable exception was larger meals, which ranked as the worst offender for abdominal pain and bloating and the second worst for flatulence. Intolerance to a large quantity of fruit was reported by 45% of participants. Amongst the other lower-fat food items that were in the top 10 for provoking individual symptoms, legumes triggered bloating most frequently, whereas no lower-fat food was highly prevalent for exacerbating abdominal pain. Fruit in a large quantity, Brussel sprouts, legumes and onion were deemed the most flatulogenic. Potato was the most tolerated food. The top 10 foods consequently avoided, by 24%–45% of participants, were almost all the same top 10 food intolerances. A concerning outcome was that 15% of participants avoided both fruit and vegetables.

Comparative studies specific to BAD, IBS-D or functional diarrhoea from a UK cohort or elsewhere do not exist. In an Irish study in 135 secondary care patients of which 40% had IBS-D and 91.1% were women, 89.6% perceived that food caused or worsened GI symptoms.¹⁶ A French online survey in 84 IBS patients, of which 33.8% had IBS-D, reported on food-triggered and food-worsened symptoms amongst 73.3% and 93.4%, respectively.³¹ In a Swedish study of 197 IBS patients, of which 44% had diarrhoea-predominant symptoms, 84% reported GI symptoms in response to at least one of 56 different foods surveyed.¹⁷ In an older Swedish study consisting of 113 adults with diarrhoea amongst 330 with IBS, meals and 35 different foods were associated with their GI symptoms at rates of 64% and 51%, respectively.¹⁵

The very high prevalence rate is likely to be an overestimate in a cohort of mostly women, with persisting moderate to severe diarrhoeal symptoms including urgency, and high rates of anxiety, depression and IBS. Self-reported food intolerance is associated with more severe global symptoms and reduced quality of life.¹⁷ Female sex and anxiety predict food-related symptoms.¹⁵ Furthermore, a nocebo effect, is highly prevalent in GI disorders,³² in which food ingestion is perceived to cause symptoms, and symptom anticipation then leads to experiencing symptoms after consumption. In the negative context of persisting, urgent, diarrhoeal symptoms, with 60% of this cohort reporting abdominal pain, nocebo hyperalgesia via visceral or somatic mechanisms could be a factor.^{10, 33} Physiological mechanisms in BAD to explain pain relate to the absorption of bile acids by diffusion along the colon activating neuromodulation and consequentially inducing contractions.³⁴

Comparative data on food and diarrhoea provocation are limited to two out of three identified studies. The first is a nationwide Dutch online study involving 1601 IBS participants fulfilling Rome IV criteria, of which 34.8% had IBS-D and frequencies for severe symptoms but not diarrhoea were reported.³⁵ The second is a Swedish study amongst 330 adults with IBS and moderate to very severe symptoms recruited from the community and hospital settings,¹⁵ of which one in five of the 113 patients with IBS-D could have had BAD.³⁶ In the third study, comprising an Irish study, small sample sizes and foods being grouped together (e.g., cereals) deterred comparison.¹⁶ In the Dutch study, highest frequencies were for greasy foods (39%), fried foods (29%) and milk (24%), whereas other high-fat foods affected a small minority (chocolate, 11%; crisps, 12%; cheese, 9%; nuts and egg, 6%). In the Swedish study, cream (11%) and pizza (8%) were amongst the most common foods for stool looseness and urgency, whereas deep-fat fried and fried foods was more commonly implicated for dyspepsia, abdominal pain and flatulence. When considering fat intake, lack of portion size quantification hinders food comparisons between studies.

There were some anomalies in the data with foods that were difficult to place in context with food groups or fat content. These data showed that higher-fat milk and dairy products were less well tolerated than lower-fat milk and dairy products by the majority of participants. This suggests that the fat content ingested governs intolerance to dairy items, although other factors such as volume ingested and flux through the GI tract cannot be excluded as factors influencing tolerance. For full-fat dairy items, cream with a fat content of 14 g was more frequently poorly tolerated than full-fat milk and cheese with a fat content of 8–10 g and 10 g per serving, respectively. Chocolate ranked closely to cheese and non-dairy items, such as egg and nuts, all with a fat content of approximately 10 g per serving. However, biscuits and crisps with similar fat contents per serving, were comparatively well tolerated. In an old feeding study amongst six healthy men comparing daily fat intakes of 62 versus 152 g, Cummings et al.³⁷ suggested that high intakes of fat beyond habitual intake may increase the bile acid pool size and lead to increased bile acid excretion. Amongst the UK population, common snacks consumed daily include crisps and biscuits, whereas many of the food items least tolerated are foods which might be less frequently consumed (e.g., fish and chips on a Friday). Other dietary components could be to blame for GI symptoms due to known mechanistic effects. Caffeine increases colonic motor activity similar to meals compared to decaffeinated coffee and water in healthy subjects.³⁸ However, out of the reported symptoms that included abdominal cramps and “stooling”, the most frequently reported one was flatulence. Ethanol acutely stimulates hepatic bile acid synthesis,³⁹ which implicates increased excess colonic bile acid spillover and consequential diarrhoea.

For comparative lower-fat foods, highest prevalence rates were for onion (32% and 56%) and for legumes (24% and 46%) in the Dutch and Swedish studies, respectively, causing most commonly flatulence, abdominal pain and dyspepsia or distension but not diarrhoea.¹⁵ Brussels sprouts, legumes and onion, and vegetables known to be high in fermentable oligosaccharides were reported to be associated with flatulence.^40-42 Visceral hypersensitivity to luminal distention of the colon in IBS is a key underpinning mechanism for fermentable carbohydrate-related symptoms.¹¹ Over one-third of our cohort reported co-existing IBS. Colonic bile acids, including chenodeoxycholic acid, a secretory bile acid, have been found to correlate positively with visceral hypersensitivity and abdominal pain intensity,⁴³ with the data taken from patients with IBS who did not have BAD determined by serum C4 and FGF19 levels. In this cohort, fruit in a large quantity in one go was the fifth most problematic food for diarrhoea. By contrast, apple, mango, pear, stone fruits and watermelon were well tolerated by the vast majority. This suggests that, in BAD without co-existing IBS, fruit and vegetables should be well tolerated. Furthermore, those high in polysaccharides with gut luminal viscous properties and bile acid-binding capacities, other than psyllium and oat bran,⁴⁴ and potato,⁴⁵ may be beneficial on diarrhoea. Moreover, kiwifruit has a high water-holding capacity.⁴⁶ Therefore, when considering that psyllium is used as anti-diarrhoeal medication for all three types of BAD,⁴⁷ although untested, kiwifruit in moderate quantities, may be favourable for diarrhoea in BAD. Across a day, achieving an intake that conforms to the recommendation given in Eatwell Guide should be feasible, to “eat at least 5 portions of a variety of fruit and vegetables a day”.⁴⁸ Therefore, in extrapolating the proportions avoiding these food groups in this survey, overly-restrictive dietary practices may exist for 30% of people with BAD. It would be a substantial challenge for people with these perceived intolerances to meet the recommendation by the World Health Organization for fruit and vegetables of at least 400 g per day to prevent chronic disease and micronutrient deficiencies.⁴⁹ Further research is warranted to understand barriers to the consumption of fruit and vegetables in BAD.

Two comparative studies selected similar post-prandial timings but with multiple groupings: immediate, within 30 min, 30–60 min and >1 h,¹⁶ and 0–15 min, 15–30 min, 30–60 min, 1–3 h and >3 h.¹⁵ Outcomes for individual foods were not sought in one study.¹⁶ In the other study, they were not presented by individual food item and, although 52% of IBS patients had symptom onset within 30 min, individual GI symptoms were not reported.¹⁵ The study design considered perceived timing of symptom provocation⁵⁰ by defining food item intolerance as a self-reported clear pattern of gut symptoms after food ingestion. Food item intolerance was split to identify responses within 30 min and after 30 min. This was sought because of the gastro-colonic response to eating in healthy adults,^51-53 and colonic motor activity has been found to increase within 30 min of food ingestion.⁵⁴ The provocation of diarrhoea, abdominal pain, bloating and flatulence within 30 min of food ingestion was more frequently reported than after 30 min of ingestion for almost all selected foods. This suggests that for those with rapid post-prandial symptoms there was an exaggerated sensory component of the gastrocolonic reflex⁵³ and dietary fat induced colonic hypersensitivity, as experienced in IBS.⁵⁵

Four in every 10 participants had no gallbladder, which was an unexpectedly high finding in this cohort. After cholecystectomy, with the loss of food-coordinated release of bile salts, bile flux in the small intestine is almost continuous and the enterohepatic circulation rate becomes more rapid with a consequential smaller bile acid pool.⁵⁶ There is a paucity of evidence on how this impacts on dietary fat tolerance or small intestinal flux of dietary components. One uncontrolled study involving 123 consecutive patients advocated a low-fat diet preoperatively for diarrhoea and postoperatively for its prevention for at least 1 week.⁵⁷ However, fat and other nutrient intakes were not quantified. Also, although not following a low-fat diet was a predictor for diarrhoea at 1 week postlaparoscopic cholecystectomy, this was not the case at 3 months. Furthermore, diarrhoea after cholecystectomy may not be due to excessive colonic bile acid exposure.⁵⁸ It is also possible that, in this cohort, chronic diarrhoea could be an as yet undiagnosed condition such as coeliac disease, microscopic colitis, post radiation diarrhoea, small bowel overgrowth, lactose intolerance or fat malabsorption.⁵⁹ Because a diagnosis of BAD in the UK necessitates specialist review,⁵⁹ and BAD is confirmed by response to bile acid sequestrant medication, this is unlikely. However, 7% of participants in this cohort were not taking any medicines used for treating BAD, which could suggest consideration of lactose intolerance or sensitivity to fat intake.

The present study has strengths and limitations related to data assessment and collection methods. Convenience sampling is likely to have led to an overestimation of the prevalence of food intolerance in BAD. A geographically diverse sample across the UK not limited to a particular healthcare setting was sought. Online data collection is suited for the UK considering that Great Britain's 2020 census data showed that 96% of households had internet access.⁶⁰ Digital exclusion would have led to under-representation amongst those with socio-economic or health issues such as low household incomes, isolation, worries about online safety and cognitive or visual impairment.⁶¹ This would also include individual lacking digital skills such as those over the age of 65 years, as well as those in rural locations with limited internet access.⁶¹ Representation from British white women was high. UK studies providing demographic data are limited to specific areas within England and in the secondary care setting. Women may be more likely than men to respond to surveys,⁶² as was found in the first online UK survey in self-reported BAD, in which nine of the 100 respondents were men.⁹ The sample size achieved is a strength of the present study.

The present study risks recall bias because participants self-reported on their medical histories and food restrictions. Participation could be anticipated to be greatest amongst those whose lives were most affected by BAD and symptoms, which would include those accessing the websites of BAD UK and NHS webinars. This appears likely considering the high proportion of participants that reported having severe or moderate BAD diagnosed by SeHCAT. Participants may have opted out as a result of inability or discomfort in completing the questionnaire digitally, although contact details were given for any technical difficulties. Data cleaning and preliminary analysis of 192 open comments were not suggestive of any duplicate participation or of any bogus responses.

The anonymous nature of online data collection excluded the possibility of selecting participants based on a confirmed diagnosis via their medical records. People were asked to participate only if they had been SeHCAT tested; thus, every effort was made to ensure accurate inclusion of eligible participants. SeHCAT is the most widely used test used by gastroenterologists in the UK⁶³ and internationally recognised.⁸ However, it is not readily available throughout the UK.⁶³ Some individuals may have been excluded who probably did have BAD determined by empirical treatment. It is also possible that people participated who did not have a SeHCAT determined diagnosis of BAD, but who nonetheless reported they did.

Disparities in how diarrhoea is defined was previously identified from systematic review of dietary interventions in BAD.¹³ The validated Bristol stool chart³⁰ was shown in the questionnaire to aid accurate self-assessment. The high prevalence of diarrhoea in eight out of 10 participants is consistent with the rate of 72% identified in a Danish survey involving 377 respondents with BAD determined via SeHCAT.⁴⁷ It is also consistent with the result on medication effectiveness and that six out of 10 were using colesevelam, the most potent BAS.⁶⁴ In a recent double-blind, randomised, placebo-controlled clinical trial involving 37 patients with primary BAD defined as moderate to severe by SeHCAT, efficacy was 59% defined by responder achieving diarrhoea remission.⁶⁵

Estimating food portion sizes to accurately quantify nutrient intake is difficult⁶⁶ but important for obtaining dietary data to understand symptom effect. A method to precisely measure portion sizes does not exist. Accuracy of estimates using digital pictures has been found to be approximately 10% with judgement on solid foods such as cake more accurate than amorphous food items and liquids.^{67, 68} Error of estimation, affected by an individual's perception, mental construct of a food item's quantity and memory, was reported to be as high as 37%.⁶⁹ This study involved 51 healthy adult volunteers and only six foods commonly consumed in the UK with angled photographs. There was underestimation of larger portion sizes compared with medium or smaller portion sizes and if BMI was ≥ 30 kg m⁻², and overestimation if female, of older age or retired. For cost-effectiveness, use of household measures may be as accurate as aerial images for some food forms.⁶⁸ In comparative studies, foods were not quantified.^{15-18, 35} In the present study, metric weights, household measurements, portion sizes used in packaged, manufactured foods and written descriptions were given to aid accurately representing food items. Participants on average had persisting symptoms for 3 years, suggesting people's dietary experiences for reporting on food intolerances were not naive. Food intolerance was defined as a non-allergic response of GI symptoms initiated by food ingestion.^{50, 70} The UK's official food guideline on what constitutes a healthy, balanced diet, the Eatwell Guide,⁴⁸ was used to group foods into lower-fat and higher-fat items. The cut-off for lower-fat and higher-fat foods of 8 g per eating session was modelled on an intake of 40 g of fat per day distributed as even doses of five eating sessions across a day consisting of three meals and two snacks between.⁷¹ Selection of foods took into account the popularity of groceries amongst UK consumers, including of out-of-home meals.^{72, 73} The fat content in take-away food in the UK has been shown to vary considerably.⁷⁴ Fish and chips was included as an item separate to take-away food because via the National Federation of Fish Friers, the UK's trade body representing the nation's fish and chip shops, a “perfect portion initiative” exists.⁷⁵ This was set up for guiding on portion control. The fat content in a standard portion is 52.3 g. Selection also included foods high in fermentable oligosaccharides, monosaccharides and polyols (FODMAPs),^{40, 41, 76-79} which are used to test for GI symptoms in the management of IBS-D.⁸⁰ Selection excluded foods related to pharmacological effects known to affect the GI-neuroendocrine system. For example, chilli, which contains capsaicin, was excluded. One study in IBS-D found abdominal pain induction but no worsening of diarrhoea within 2 h after ingestion of 2 g.⁸¹ Selection of food items also excluded those related to intolerances of known enzyme and transport defects, such as lactose and biogenic amines. The amount of detail collected on food and perceived diarrhoeal symptoms is a study strength in the context of limited research despite the fact that BAD is as common as coeliac disease⁸² and often very debilitating for sufferers.⁹

To our knowledge, this is the first study to investigate food intolerance related to GI symptoms amongst adults with BAD. It was exploratory and the data were described. Therefore, statistical testing to confirm any associations between food intolerances and diarrhoea or other GI symptoms was not justified as a result of this level of robustness within the data. These findings are not generalisable to the UK population living with BAD. Further demographic data on BAD are needed, particularly on prevalence rates amongst non-white groups. The study design cannot infer a causal relationship between food item ingestion and symptoms of BAD and should not be transferred to clinical dietetic practice. The study showed that food is an important factor in the management of BAD symptoms. Survey replication compared with a healthy, general population could strengthen these preliminary results. Further studies are needed on the tolerance to fat intake per serving, supporting the inclusion of lower-fat milk and dairy products, fruits and vegetables.

AUTHOR CONTRIBUTIONS

Yvonne A. McKenzie: conceived the idea for the study, generated the research questions, designed the survey and built it on REDCap. Sorrel Burden and Chris Todd: contributed to study design and survey development. Chloe French: contributed to developing the survey on REDCap. Yvonne A. McKenzie: managed recruitment and data collection. Sorrel Burden and Chloe French: contributed to data analysis and interpretation. Yvonne A. McKenzie: drafted the original manuscript. Sorrel Burden: critically reviewed and edited the manuscript. All authors reviewed and approved the final version of the manuscript submitted for publication.

CONFLICT OF INTEREST STATEMENT

The authors declare that there are no conflicts of interest.

TRANSPARENCY DECLARATION

The lead author affirms that this manuscript is an honest, accurate and transparent account of the study being reported. The reporting of this work is compliant with STROBE guidelines. The lead author affirms that no important aspects of the study have been omitted and that any discrepancies from the study as planned have been explained.