Diet has long been hypothesised to impact inflammatory bowel disease (IBD), both Crohn’s disease (CD) and ulcerative colitis (UC). The widespread perception is that an unhealthy Western lifestyle alters IBD risk. Westernised diets provide consumers with more choice and longer shelf life at low cost. A consequence of this convenience is that food is comparatively higher in total energy, total and saturated fat, sugar and food additives and lower in fibre.1 These dietary components that are believed to be key players in IBD development are supported by some epidemiological and preclinical studies, but not necessarily clinical trials.

There is a concept that dietary components causative of IBD development persist in their action and that their modification can be utilised to treat established IBD. However, this concept does not fit all chronic disease. For example, removing gluten, a causative factor for coeliac disease, induces remission in the majority, but the removal of asbestos does not treat mesothelioma. The causative versus therapeutic targets for diet in IBD have been ambiguous and often clouded with dogma.2 Consideration of diet in different stages of IBD is needed. The aims of this review are to examine the strength of evidence for using diet in pre-disease, active disease and quiescent disease.

Diet for IBD prevention

Identifying modifiable risk factors in the pathogenesis of IBD is appealing to prevent IBD in the general population and specifically in those at high risk, including first-degree relatives of patients with IBD. Establishing the role of diet in IBD prevention is determined via large prospective cohort studies that assess diet through dietary questionnaires in a healthy population that is followed forward. Associations are made between diet and disease development. From these data, there are clear differences between CD and UC, but results are often inconsistent in finding associations.

Specific nutrients and dietary components associated with development of CD or UC are summarised in table 1. For CD, development of disease was associated with protein intake double that of most dietary guidelines3 and sugar-sweetened beverages.4 Protective dietary factors included docosahexaenoic acid (DHA),5 dairy,6 total fibre, specifically from fruit or grains or both,7–9 polyphenols in wine, grapes and certain herbs,10 dietary potassium11 and zinc.12 13 For UC, meat intake, particularly red meat, was consistently associated with disease development,3 14 as was trans-unsaturated fats15 and n-6 polyunsaturated fatty acids (PUFA).16–19 Oleic acid17 and vitamin D20 were protective.

Table 1

Associations of nutrients, food components or dietary patterns with development of Crohn’s disease and ulcerative from prospective cohort studies

Another approach is to consider dietary patterns that may consider the complex interactions between foods. This may be assessed through dietary scores based on diet quality indicators, such as a Mediterranean diet (MED) or ultra-processed food (UPF), a marker of a Western diet. Alternatively, indices are developed through associations of inflammatory markers, as with the empirical dietary inflammatory pattern (EDIP). Table 1 summarises dietary patterns associated with CD and UC development and some described below.

MED and healthy lifestyle patterns

A MED, often characterised as plant-based eating, describes a diet rich in wholegrains, fruit, vegetables, legumes, nuts, olive oil, moderate in fish/seafood, dairy, wine and limited in meat, particularly red meat. Features of a MED seem to mimic the data indicating dietary risk of CD, specifically its positive association with protein and negative associations with fibre, DHA found in fish, dairy, and polyphenols in wine and herbs. This relationship has been recognised and developed MED scores assessed through the prospective cohort model in three unrelated studies with inconsistent findings. A study comprising two Swedish cohorts showed reduced risk of CD onset with an MED,21 but the other two studies found no association.22 23 One explanation for this inconsistency may be due to the differences in how the MED scores were calculated. Dairy contributed to the MED score of the inversely associated Swedish study but was deducted or not considered in the MED scores for the studies showing no association. MED was not associated with UC development.21 23 24

While not specified as a MED, another cohort assessed a healthy lifestyle score (HLS), which considered comparable dietary factors, high in fruit, vegetables, nuts and fish and low in red meat, and also the lifestyle factors body weight, smoking and physical activity. An HLS demonstrated a reduced risk of developing both CD and UC in the Nurses’ Health studies and subsequently validated in three other cohorts.25 These data indicate non-dietary lifestyle modifiers likely play a role in IBD development. Indeed, US Biobank data suggested that adhering to a healthy lifestyle that considers smoking, body weight, sleep, physical activity in addition to a healthy diet mitigates genetic risk. Individuals with a high genetic risk had a 50% reduction in developing CD and UC if following a healthy lifestyle.26 Conversely, diets considered unhealthy, such as the carnivorous diet, rich in meat and a diet high in discretionary food but low in fruit and vegetables, were associated with UC and CD, respectively.22

Empirical dietary inflammatory pattern

The EDIP index is based on 18 inflammatory and anti-inflammatory foods associated with C reactive protein (CRP) and circulating cytokines in a healthy population.27 The inflammatory foods not only included processed, red and organ meats but also unexpected foods, such as certain fish and vegetables. Beer and pizza were surprisingly ‘anti-inflammatory’.28 However, the predefined foods were curious and often ambiguous. For example, ‘pizza’ and ‘snacks’ have variable composition and it was unclear why pizza was selected over, for example, burgers or fried food. Additionally, food categories did not represent usual intake, with 8/39 categories representing beverages but only one for fruit. A subsequent evaluation of the EDIP score across seven countries found no association with CD or UC.29

Ultra-processed food

Food defined as UPF via the NOVA classification system typically contains ≥5 ingredients, extracted from food components and often contain food additives30 and may be considered to be a marker of diet quality beyond considering macronutrient and micronutrient intake. Meta-analysis of five cohort studies showed a consistent increased risk of up to 40% for developing CD from UPF, but not UC.31 However, there are some difficulties in data interpretation, with misclassifications of UPF likely to exist, even through various modelling scenarios that were applied to the food database.31 The definition for UPF is of low threshold and the nutritional value and composition of UPF vary hugely. For example, both flavoured tinned tuna and ice cream are considered UPF. Indeed, only when the extremes of UPF intake were compared, there was a risk for CD development identified.31 This should be considered in translating UPF recommendations to the community as complete avoidance is unrealistic for most people.32 Another limitation with assessing UPF is that it is not known as an accurate marker of poor diet quality. Diets higher in UPF are often also higher in fat, sugar and lower in fibre, but it may be that the ultraprocessing itself rather than the diet composition is associated with health consequences. Notably, IBD guidelines recommend limiting UPF33 and for offspring of IBD patients.34

Interpretation to clinical practice

Prospective epidemiological studies have presented evidence for the role of specific dietary components, diet pattern and lifestyle factors in the development of CD, and, to a lesser extent, UC. The general message that features of a MED that is heavily plant based, moderate in fish and dairy and low in meat and UPF is pleasingly in line with healthy eating guidelines and assurance that these default dietary recommendations are likely to reduce risk of CD development. While dietary associations for UC development are less apparent, with the exception of red meat consumption increasing risk, it is judicious to recommend the same for those at risk of UC.

Diet for IBD treatment

The first dietary intervention to induce remission was exclusive enteral nutrition (EEN) for CD, but it is difficult to maintaining beyond 6–8 weeks. The last decade has seen multiple novel diets emerging as other candidates for induction therapy. As with IBD prevention, the role of diet in CD and UC treatment seems distinct. Induction dietary therapies assessed for IBD and their composition, based on descriptions of the diets rather than food composition analyses, are described in figure 1. For CD, there are no common dietary themes, except that all treatments contain dairy protein and fat. For UC, the two proposed therapeutic diets appear similarly high in fibre, low in meat protein and dairy.

Figure 1

Content of fibre, meat protein, ultra-processed foods, dairy protein and fat and gluten in diets used for treatment of: (a) active Crohn’s disease; and (b) active ulcerative colitis. Schematic representations are based on diet descriptions. CDED, Crohn’s disease exclusion diet; EEN, exclusive enteral nutrition; MED, Mediterranean diet; PEN, partial enteral nutrition; SCD, specific carbohydrate diet; UCED, ulcerative colitis exclusion diet.

Exclusive enteral nutrition

Efficacy data for EEN, which replaces food with liquid nutritional supplements for 6–8 weeks, are strongest for paediatric patients with CD, with a Cochrane review showing 83% remission rates.35 This is comparable to clinical remission from corticosteroids35 36 and superior for mucosal healing (74% EEN vs 33% corticosteroids).37 Furthermore, EEN in newly diagnosed patients is associated with corticosteroid avoidance over 6 years.38 The better safety profile of EEN over corticosteroids certainly makes EEN the more appealing option and consensus guidelines recommend EEN as first-line therapy to induce remission in children with active luminal CD.39 Use of EEN in adults is becoming more popular, although evidence of efficacy is not as strong compared with corticosteroids for clinical remission.35 However, remission rates vary between 45% and 81% and studies with higher adherence rates show equal efficacy to paediatric patients.40–42 Consensus guidelines have recently acknowledged EEN usage within an adult IBD service.33 43 44

Mechanism of action for EEN is unknown but likely result from complex interactions between mucosal immune response and luminal environment.45 There are features of EEN composition that seem consistent and provide clues on how diet may be modulated for targeting inflammation. Curiously, most of the compositional features of enteral formula are greatly contrasting to a MED, seemingly protective of CD development (figure 1A). Like the MED, protein sources are casein, whey and soy protein, but unlike the MED, enteral formulas are free from lactose, gluten and fibre.46 Maltodextrin, sucrose, glucose syrup and corn starch are the most common carbohydrate sources, found in higher concentrations than diet, the MED, in particular.46 This is also conflicting with in vitro and animal studies suggesting that food additives such as maltodextrin trigger or deteriorate intestinal inflammation.47 Furthermore, enteral formula is arguably the most UPF available. This paradox supports the idea that causal factors for CD development may not be targets for treatment.

Partial enteral nutrition

Partial enteral nutrition (PEN) describes a therapy in which only a proportion of food is replaced by enteral nutrition and the remaining diet is unrestricted. There is less evidence for PEN than EEN as a CD induction treatment; however, the percentage of caloric requirements provided by enteral nutrition is likely of importance on remission rates. Most trials have evaluated 50% of caloric intake from PEN, with remission rates quoted between 15% and 50%.48 49 Rather than PEN monotherapy, emerging evidence proposes PEN combined not only with biologic agents can enhance efficacy and achieve improved primary disease induction outcomes for active CD50 but also with dose escalation for biologic loss of response.51 Given the nature of PEN therapy, it is thought that dietary composition of PEN is similar to that of EEN, particularly as the food proportion of diet is unaltered, with the exception of the Crohn’s disease exclusion diet (CDED).

Crohn’s disease exclusion diet

The CDED coupled with PEN remains the only diet therapy, containing food, with reproducible evidence of efficacy for managing active CD.52–54 In paediatric CD patients, CDED with PEN is better tolerated and as effective as EEN in inducing clinical remission and reducing faecal calprotectin.52 Similar results are seen in adults with and without PEN.54 The rationale of the diet is that it excludes dietary components implicated in the development of CD from epidemiology and animal studies, including animal fats, red meat, gluten, dairy, maltodextrin and UPF.47 One would expect that the diet mimics the composition of a CD preventative diet, such as the MED, yet, in reality, the composition of the diet is very different. CDED is mostly studied coupled with PEN in significant volumes (50% of energy intake), which contains food additives, milk fat and milk protein and lacks fibre (figure 1A). However, the food component, comprising approximately a dozen foods and including the mandatory intake of chicken, eggs, potatoes, apple and banana on daily basis, is also high in animal protein (mandatory daily chicken and eggs) and low in fibre (limiting wholegrains, legumes, nuts and most fruit and vegetables). It may be that the similarities between the composition of CDED to EEN explain its efficacy for treating disease. This is further supported by the stable levels of faecal short-chain fatty acids during therapy with CDED and PEN and the similarities in faecal microbial effects of EEN and CDED with PEN previously described, including the reduction in the abundance of fibre-fermenting bacteria such as Bifidobacterium and Prevotella.52 55

Other diets targeting treatment of CD

The specific carbohydrate diet (SCD) is a food-based diet that eliminates disaccharides and polysaccharides (lactose, sucrose, starches, grains and most legumes), thus it is relatively low fibre, gluten free and contains lactose-free dairy56 (figure 1A), similar to EEN. SCD efficacy for active CD has been studied in retrospective studies,57 58 but recently compared against the MED, two contrasting diets, in patients with symptomatic CD.59 Symptomatic response was improved with both diets, without difference between them.59 There were no differences between CRP and faecal calprotectin between the diets, nor did either diet change the inflammatory markers from baseline, although faecal calprotectin was raised in less than a quarter of participants at basal conditions,59 suggesting that both diets have an impact on symptoms but not necessarily inflammation.

In an attempt to mimic the effects of EEN, the CD-TREAT (Crohn’s Disease Treatment-with-EATing) diet was devised. CD-TREAT-induced EEN-alike microbial effects and improved gut inflammatory markers in an animal model of disease.60 In human trials, CD-TREAT reduced faecal calprotectin and induced clinical remission in a pilot study of five paediatric patients.60 In a larger cohort of 57 children and adults with relapsing CD, a CD-TREAT course induced 75% remission and reduced faecal calprotectin.61 Consistent with its rationale, the diet composition is similar to that of EEN, being low fibre, gluten and lactose free, high in milk protein and fat and contains UPF (figure 1A). These pilot open-label data require independent replication and further investigation within controlled trials.

Diets targeting treatment of UC

Therapeutic diets for UC have lagged behind CD, but data from two UC-specific diets are now emerging, with similar themes of dietary modification (figure 1B). Pilot data for the UC exclusion diet (UCED) from an open-label study in children with active UC have shown improved inflammatory activity index in 9/24 patients with diet alone and an additional four with combined antibiotic rescue therapy, six of whom maintained symptom response, although inflammatory markers did not improve.62 Beyond symptomatic improvement, a subsequent trial in patients receiving UCED alone induced endoscopic remission in 4/15, but no difference when combined with faecal microbiota transplantation (FMT).63 The UCED is a step-down diet that initially restricts many wholegrains and legumes, with daily limits on animal protein, dairy and wheat and promotes increased fruit, vegetables, including potatoes. Red meat and UPF are avoided.

In a similar fashion, a pilot open-label study of 28 adult patients with UC following the 4-SURE diet (4-strategies-to SUlfide REduction Diet) showed clinical and endoscopic response in 46% and 36%, respectively.64 The four central strategies adopted by the 4-SURE diet are a limit of total protein to 75–90 g/d, a limit of sulphur-containing amino acids to 1.5–2.0 g/d, avoidance of sulphite/sulphate, nitrite/nitrate and carrageenan food additives and inclusion of 10–15 g/d resistant starch and slowly fermentable non-starch polysaccharide (figure 1B). It is notable that there was no difference in sulphur-containing amino acids between the 4-SURE and habitual diet, thus data may be more reflective of changes in total protein, fibre and resistant starch. A well-powered study with controlled diet comparison is much needed to substantiate the hypothesised therapeutic components of the diet. Nonetheless, the concepts of both the UCED and 4-SURE diets to avoid red meat align with the evidence to prevent UC development. However, both diets further modify the diet with focus on including resistant starch and other components that modulate distal luminal microenvironment, with basis from pre-clinical models.65

Interpretation to clinical practice

To date, EEN and CDED with PEN are the only dietary therapies supported by high-quality trials to treat active CD and are becoming standard-of-care. However, these findings provide a contradiction to those seen of diet-preventing CD. Outwardly, it would seem that a diet low in fibre, gluten and lactose and inclusive of dairy protein and fat, maltodextrin and UPF treats CD. However, a lack of mechanistic studies must preclude specific recommendations of modulating these nutritional components outside of the context of these short-term diets, particularly with the potential risk of harm from diets lacking fibre and high in UPF. While less robust in evidence, diets for treating UC appear to fit with predisease models and generally in agreement with healthy eating guidelines that limit red meat and increase fibre.

Diet preventing IBD relapse

Cumulative data demonstrate the potential for dietary manipulation to maintain remission. This was indicated through the observation that resumption of eating after EEN-induced remission reactivates intestinal inflammation in patients with CD.66 What is not clearly understood is whether diet as a maintenance agent should follow the principles of treating disease or preventing disease, particularly when there is a discord between the diets, as with CD. A few studies assessing the impact of specific diets on relapse rates in patients with quiescent IBD provide some clues to this question.

Maintaining remission for CD

The role of diet in maintaining remission in patients with CD for at least 6 months has been assessed in a few studies. Three trials evaluating 50% PEN in adults compared with regular diet or thiopurines, reduced relapse rates over the measured 6–24 months.67–69 Retrospective paediatric studies also indicated benefit at 12 months.70 71 The studies that have not shown sustained remission from PEN used 20%–30% of caloric requirements66 72 and as with induction therapy, PEN for maintenance is likely to depend on>35% PEN of total energy.73

Considering food, an internet-based trial recommending or limiting red and processed meat in >200 patients with quiescent CD found no associated with symptomatic relapse over 49 weeks.74 Conversely, a semivegetarian diet, described as reduced animal protein and fat, reduced relapse risk in an observational trial of 16 patients.75 Perhaps the high fibre component of the diet was protective, which is consistent with a study where >23 g dietary fibre daily reduced flare risk by 40% over 6 months.76 This presents another discrepancy in CD management as PEN promotes a low fibre intake.

Maintaining remission for UC

Dietary patterns promoted as healthy have been investigated for the prevention of UC relapse, specifically the role of meat with or without the consideration of fibre. Two uncontrolled observational studies showed inconsistent results over 12 months, one demonstrating a fivefold risk of UC relapse with red and processed meat77 but the other showing no such link.78 The aforementioned semivegetarian diet, also high in fibre, resulted in reduced relapse rates at 1 and 5 years’ in UC patients if incorporated during induction therapy.79 This study was consistent with an interventional cross-over study, in which a diet low in red meat, total fat and high fibre, compared with a standard American diet, increased faecal markers considered protective of IBD relapse, including Faecalibacterium prausnitzii, and a trend for reduced serum amyloid A, a marker of mucosal inflammation.80 This theme of reduced red meat combined with high fibre, as with a MED, positively altered microbiota, predicting relapse compared with a usual Canadian diet.81 Finally, in UC patients following total proctocolectomy with ileal pouch-anal anastomosis, adherence to a MED was associated with lower inflammatory markers and lower rates of pouchitis.82

More complex diets assessed include an ‘anti-inflammatory diet (AID)’ that aimed to increase fibre, probiotics, antioxidants and n-3 PUFA and decrease red meat and added sugar. No difference in relapse rates were seen in UC patients when compared with Canada’s healthy eating guidelines,83 but there may be little compositional differences between the diets. Another diet of similar composition, labelled FMT-AID, resulted in sustained deep remission at 12 months compared with standard medical therapy, however, the dietary treatment was administered with FMT, which precluded evaluation of diet alone.84

Interpretation to clinical practice

Studies assessing diet to maintain remission for CD have not provided more clarity on whether a diet resembling EEN or MED should be used, with data for fibre, as a surrogate marker differentiating the diets, showing conflicting results, as has associations with red meat. Healthy eating guidelines should be supported in the absence of clear diet pathways. Conversely, the concept of healthy eating is a more promising strategy for retaining UC remission, although whether the effects are due to reduced red meat, high fibre or both are not clear.