INTRODUCTION
The role of medical nutrition therapy (MNT) in the modulation of intestinal health in individuals with inflammatory bowel disease (IBD) is supported by the effectiveness of exclusive enteral nutrition (EEN), the administration of a nutritionally complete, liquid formula (most often polymeric, i.e. composed of whole macronutrients) while excluding all other oral intake for 6 weeks or longer [
1,
2]. In mild to moderate Crohn’s Disease (CD), one type of IBD, EEN has been associated with a remission rate of approximately 80% [
3–
6]. Early relapse upon return to a free diet (food intake ad libitum) following EEN and the ineffectiveness of partial enteral nutrition (PEN; typically 25–75% of total energy requirements) with a free diet as an alternative MNT suggests that the success of EEN is related to exclusion of certain dietary components [
6–
11]. Recent research has implicated food additives, such as emulsifiers (e.g. carrageenan, carboxymethylcellulose) in impaired intestinal health, including inflammatory changes and degradation of the mucous layer, which separates the intestinal wall from pathogens and commensal microorganisms [
12–
14]. Other dietary components are thought to have a protective effect on intestinal health. For example, the fermentation of non-digestible carbohydrates (e.g. dietary fibre, resistant starch) produces short-chain fatty acids in the large intestine; these are involved in maintenance of the mucus layer, immune modulation, and decreased intestinal lumen pH, which can inhibit the growth of pathogens [
15–
17].
Literature on specific dietary components (i.e. nutrients, foods) and intestinal health provides potential insight into CD pathophysiology but relies heavily on animal models, and mechanisms are not fully understood [
12–
17]. Furthermore, nutrients and foods impact one another, making it difficult to study independent effects. This is relevant to literature interpretation, as individuals eat mixed meals, rather than isolated nutrients/foods [
18–
20]. Approaching the diet-disease link from the perspective of the overall diet (dietary pattern) can help address these limitations and offers a more practical approach, rationale that has been recognized by several clinical practice guidelines [
21–
23].
The Western and Prudent dietary patterns have been repeatedly identified and evaluated in the context of chronic disease prevention using
a posteriori analysis (e.g. factor analysis) [
24–
26]. The Western pattern is associated with an increased risk of chronic disease associated with an inflammatory response, including IBD [
27–
30]. It is characterized by a high intake of animal protein, fried and processed foods, additives, refined grains, high glycemic index carbohydrates, high-fat dairy products, and added sugar [
20,
27,
31,
32]. Conversely, the Prudent pattern is associated with a decreased risk of chronic disease and is consistent with MNT that have been found to have a positive impact on chronic disease risk and/or inflammatory markers [
31,
33–
38]. It emphasizes vegetables, fruits, legumes, whole grains, nuts, fish, poultry, low-fat dairy products, and is low in processed foods and often high in fibre and resistant starch [
20,
27,
31,
39]. Examples of MNT consistent with the Prudent pattern include the Mediterranean Diet and Dietary Approaches to Stop Hypertension.
The CD–dietary pattern relationship has not been fully described. A Canadian case-controlled study in pediatric patients with newly diagnosed CD (n = 149) identified four dietary patterns through factor analysis that explained approximately one-quarter of the variance in the dietary data [
31]. A Prudent pattern was identified and was significantly negatively associated with CD development. In girls, a Western pattern was significantly associated with CD development. While a partial Western pattern (lower in animal protein) was identified in boys, there was no significant relationship with CD. Factor analysis using retrospective data on adolescent dietary habits from the Nurses Health Study II had similar results [
33]. The Prudent pattern was associated with a decreased risk of CD and no significant relationship with the Western pattern was observed [
33]. The Crohn’s Disease Exclusion Diet (CDED) is a whole-food dietary treatment for pediatric CD [
6,
10,
40–
42]. It excludes or limits certain foods and food additives characteristic of the Western pattern and emphasizes lean protein sources, fibre, and resistant starch [
6,
10,
40]. In a 12-week randomized controlled trial in pediatric patients with mild to moderate CD (n = 74), the CDED with PEN had a remission rate comparable to EEN [
6]. Its utility as a mono- or co-first line or rescue therapy has also been described in a recently published case series [
40]. The CD treatment-with-eating diet, another novel dietary therapy, aims to recreate EEN using food by excluding certain dietary components and matching macronutrient and micronutrient intake [
43]. An open-label trial in pediatric patients with mild to moderate CD found that 60% of participants (n = 5) were in clinical remission after 9 weeks of treatment [
43]. Limitations of the literature described above include use of retrospective data and lack of standardized dietary intake assessment tools and protocols.
The observed relationships between dietary patterns and chronic disease and the success of novel dietary therapies in the treatment of pediatric CD highlight the importance of assessing dietary patterns at different points in CD disease course. Therefore, this work had two objectives: (1) develop and evaluate three dietary indices, and (2) using dietary pattern analysis (DPA), describe the dietary patterns of Canadian Maritime pediatric patients with CD after return to free diet post remission induction via EEN [
44,
45].
DISCUSSION
This paper describes the development of three novel dietary indices and presents findings from DPA in a sample of Canadian pediatric patients with CD, following return to free diet after remission induction via EEN. The results suggest that the sample was consuming foods from both the Western and Prudent pattern. The observed non-adherence to either pattern may be due to low intake of certain components. The average component scores from the indices reflected low intake of whole grains, vegetables, lean poultry, fatty fish, legumes and high-fat dairy products, and moderate intake of nuts and red and processed meat.
The sample’s intake may be similar to the general Canadian population, with the exception of dairy product intake [
30,
55–
58]. The average number of servings of milk and alternatives consumed per day by Canadians 2–18 years old is 2.10–3.30 [
59]. Fluid milk and fortified soy beverages are an important source of milk and alternatives for this age group, and about one quarter consumed are high in fat and/or sugar [
59]. Other high-fat and/or high-sugar dairy products (e.g. cheese, dairy-based desserts) are also major sources of this food group for this age [
57]. No participants consumed more than two servings of high-fat dairy products per day. This could indicate intake of mostly low-fat dairy products (not assessed) or an overall low intake of dairy products. Ongoing supplemental EN, as per local protocol, could also contribute to low dairy intake.
Low intake of a variety of foods has been described in individuals with IBD. Tsiountsioura et al. found that compared to healthy controls, Scottish children with CD had a lower daily intake of dairy products, fish, and fruits and vegetables [
60]. Maconi et al. found over one-third of adult patients with IBD intentionally changed their diet due to symptoms, including reducing milk, cheese, fat, and high-fibre food intake [
61]. Similarly, a review on disordered eating habits in gastrointestinal (GI) disorders found that patients consumed less food than healthy individuals and suggested that reducing food intake was a way to cope with/control symptoms [
62]. Evidence of food restriction in GI disorders and the sample’s low to moderate intake of the four food groups raises the question of what other foods are being consumed by this population.
It is possible that the Canadian pediatric CD population consumes a high amount of “other foods” (e.g., oils/fats, condiments, snack foods), like the general Canadian population [
56]. The previously mentioned study in Scottish children found that children with CD had a higher daily intake of jams, chips, and savoury snacks than controls [
60]. In Canadian children with CD, pre-diagnosis dietary patterns high in condiments, desserts, chips, and snack foods have been reported [
31]. Since the three indices did not capture intake of these foods, the sample may have had a high intake of “other foods”.
Another potential explanation for the observed index scores is high consumption of liquids. A prior analysis of the MAREEN dietary data found that three of the eleven participants were consuming less than 600 g of solid food per day [
63]. This suggests the participants had either inadequate intake or were consuming liquids as well. Since ongoing supplemental EN in addition to free diet is recommended as an adjunct therapy for remission maintenance after remission induction via EEN, and the mean scores for the PWDI high energy drinks component and PA2020-AHEI sugar-sweetened beverages component reflect low intake of these liquids (less than one standard pop can per day), it is possible the sample was consuming EN formula and other liquids not captured by the FFQ/indices [
64]. For example, a pre-diagnosis dietary pattern with higher liquid consumption has been described in Canadian girls with CD [
31].
Despite the observed significant negative correlation and relationship between the Western-to-Prudent ratio and the PA2010-AHEI, only half the variance in the PA2010-AHEI scores was explained by the ratio. This may be due to differences in the components of the three indices. The PA2010-AHEI does not capture four components included in the PWDI (high-fat dairy products, eggs, potatoes) and PPDI (poultry). It also assesses nut and sodium intake, which are not part of the other indices. Variability in the indices’ assessment of certain dietary intake (e.g., omega-3) may have impacted correlation. Additionally, the Western-to-Prudent ratio results were interpreted with the assumption that if an individual has a high PWDI score, they have a low PPDI score (and vice versa); this would be violated in the context of high or low intake of foods from both patterns.
The results of the analysis help describe the dietary intake of pediatric patients with CD; however, more research is needed in larger, diverse samples to develop a full understanding of this population’s dietary pattern following remission induction via EEN and at other points in disease course. Additional data could assist with development of a standard whole-food-based MNT for pediatric CD, which is an emerging area of research. More information on this population’s habitual dietary intake could also help with the implementation of MNT, as patients required to make dietary changes benefit from tailored, comprehensive nutrition education strategies [
65,
66].
Dietary indices can also be used to assess adherence to MNT for CD, such as the CDED. Using a validated tool to assess adherence to MNT can contribute to evidence for and the understanding of the therapy’s effectiveness. Integrating validated measurements of adherence into dietary intervention protocols would also support monitoring and allow for identification of patients who require additional education and/or support to achieve adherence. Since dietary indices do not capture all dietary intake, they should be used to assess/monitor adherence in conjunction with other dietary intake assessment methods and detailed review of intake by a trained clinician within an interdisciplinary setting [
67,
68].
There are several limitations to this work. The indices did not completely describe the sample’s intake, a limitation inherent to
a priori analysis and FFQs. Recall bias and completion of the FFQ at home without support from clinicians may have impacted data quality. The analysis used a small sub-sample of the MAREEN Study, and demographic information (except age and sex) and markers of nutrition status and disease activity were not part of the analysis protocol. Therefore, the results may not be representative of the entire MAREEN sample, the Canadian pediatric CD population, or different demographic groups within this population. Dietary intake, and therefore index scores, may vary depending on disease activity and nutrition status. It should be noted that small sample sizes are common in pediatric CD research and the FFQ response rate is similar to those reported in the literature [
6,
10,
69–
71]. However, seasonal activities may have impacted the response rate.