Abstract
Introduction: Western diet pattern and its food components have been suggested to impact inflammatory bowel diseases (IBDs) clinical course. However, the importance of food processing level is uncertain. We aimed to evaluate whether the intake of foods with varying processing levels is associated with disease activity in IBD patients. Methods: This cross-sectional study was performed at a tertiary center between August 2019 and June 2022. Consecutive adult IBD patients were recruited. Clinical disease activity was defined using HBI (Crohn’s disease) and SCCAI (ulcerative colitis). Dietary intake was assessed using a food frequency questionnaire (FFQ) and a dedicated validated processed food questionnaire (PFQ) that categorizes dietary intake into three groups of processed food levels: unprocessed/minimally processed, processed, and ultra-processed. Adjusted odds ratios for active disease were determined using a multivariable logistic regression. Results: A total of 242 IBD patients (62.8% Crohn’s disease patients) were enrolled, of whom 73.1% were in clinical remission. A higher (upper tertile vs. lowest tertile) unprocessed/minimally processed foods consumption was negatively associated with active disease (OR = 0.38, 95% CI: 0.14–0.99), while high consumption of ultra-processed foods (UPFs) was positively associated with clinically active disease (OR = 3.82, 95% CI: 1.49–9.8). Consumption of UPF groups, almost invariably, was positively associated with clinically active disease, while consumption of the ultra-processed meats group had the strongest association (OR = 4.45, 95% CI: 2.07–9.79). Conclusion: Higher consumption of UPFs is positively associated with clinically active IBD, while higher consumption of unprocessed/minimally processed foods may be protective. Prospective studies are needed to confirm these associations.
Introduction
Inflammatory bowel disease (IBD) pathogenesis is suggested to result from a dysregulated immune response toward microbial antigens, among patients exposed to environmental triggers such as dietary factors [1, 2]. The increasing worldwide prevalence of IBD during the last decades, mostly in industrialized countries, suggests that the change to a Western diet is one of its main drivers [3].
A recent meta-analysis indicated that a Western dietary pattern was associated with risk of IBD (relative risk 1.92, 95% CI: 1.37–2.68) [4]. Furthermore, a prospective study among IBD patients showed that a Western dietary pattern was associated with a higher risk of disease flares [5]. Ultra-processed foods (UPFs) are considered a staple of the Western diet, characterized by high energy density, fat, sugar, salt, and additives, such as preservatives and emulsifiers [6], all of which are thought to be associated with the pathogenesis of IBD [7]. In animal model studies, processed foods and their specific components were associated with gut inflammation through direct impact on the epithelial barrier and immune response and indirectly through altering the microbiome composition and function [8, 9]. An alteration of the human gut microbiome and metabolome was shown to be mediated by a common emulsifier, carboxymethyl cellulose additive, in a randomized controlled feeding trial [10]. A recent meta-analysis that assessed prospective cohort studies, including more than one million participants, demonstrated that higher UPFs and lower non-processed/minimally processed food intake were associated with a higher risk of Crohn’s disease (CD) but not ulcerative colitis (UC) [11]. However, the association of dietary intake of foods, classified by level of processing, with disease activity in IBD patients has not yet been studied. For this purpose, we have developed and validated a processed food questionnaire (PFQ) dedicated to IBD patients [12].
The aim of this study was to assess the association between dietary intake of foods, categorized by processing level, and clinical disease activity, in patients with IBD, by using the PFQ. Our hypothesis was that in patients with IBD, the level of food processing is associated with disease activity.
Methods
Study Design and Population
This is a cross-sectional study conducted at the IBD clinic of the Tel-Aviv Medical Center, a tertiary medical center, between August 2019 and June 2022. The sample size was calculated to provide 80% power at a significance level of 5% with an effect size estimation of OR = 3.6 between the highest UPFs percentile intake to the lowest. Given that no previous studies assessed the impact of UPF on disease activity, our estimation was based on a previous cross-sectional study, which showed an association between processed meat and disease activity in IBD patients [13]. Accordingly, we aimed to recruit a minimum of 121 IBD patients, of whom at least 30 patients are clinically active. This study targeted adult patients with an established diagnosis of UC or CD, aged 18–75. Exclusion criteria included total colectomy, stoma, pregnancy, eating disorders, exclusive enteral nutrition or parenteral nutrition of ≥50% of total calories, or unstable systemic disease that might change eating behaviors. The study was approved by the Institutional Review Board (IRB) of Tel-Aviv Medical Center, and written informed consent was obtained from each recruited patient.
Data Collection and Definition of Outcomes
During visits, patients were queried about their age, diet type (omnivorous/vegetarian/vegan), family history of IBD, duration of their IBD, medications use, and dietary treatment. Weight and height were measured. The type of IBD and clinical data were obtained from the patients’ medical records of the IBD unit. Disease activity was assessed by interviewing the patients, using the Harvey Bradshaw Index (HBI) for CD and the Simple Clinical Colitis Activity Index (SCCAI) for UC [14, 15]. For the collection of nutritional intake data, the patients were asked to complete the PFQ and the food frequency questionnaire (FFQ) during their visit. The primary outcome was clinically active disease at the time of completion of the dietary questionnaires, defined according to HBI >4 for CD and SCCAI >2 for UC [14, 15].
Dietary Assessment
The PFQ has been developed and validated to evaluate the frequency and quantity of consumption of food items representing the various food processing levels according to the NOVA classification [16, 17]. The PFQ included 85 food items from all food groups and was characterized by three main processing levels as accepted (online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000541196). The PFQ was designed to be self-administered with instructions to indicate the frequency of consumption of each food item, according to typical Israeli serving sizes.
Furthermore, a semi-quantitative FFQ was obtained at the same time to assess dietary intake of total energy, macronutrients, and micronutrients. The nutrient components were obtained from the Israeli National Nutrient Database (BINAT), Ministry of Health. Each PFQ and FFQ was reviewed by a dietitian and missing data were completed by telephone. The completed PFQ and FFQ from each patient were encoded and formatted for analysis. Patients with unreasonable FFQ results (total daily calorie intake of less than 500 or 800 kcal or more than 3,500 or 4,000 kcal for women and men, respectively) [18] were excluded (N = 11). In a secondary analysis, we classified the UPFs into 5 food groups: (1) Bread pastries and starch: different types of industrial bread, crackers, ready-to-eat/-heat pastries or pizza, salty purchased snacks, ready-to-heat French fries or gnocchi; (2) oil and spreads: mayonnaise, margarine, ready-made sauces for cooking or seasoning, coconut cream, different types of spreads, different types of industrial ready-to-eat salads; (3) ultra-processed meat: processed red meat, processed poultry, hot dog, sausages, and pastrami; (4) sweet products and desserts: ready-made cake, sweetened yogurt, breakfast cereals, ice cream, energy bars, and ready-made sweet/chocolate bars; (5) Ultra-processed beverages: sweetened soft drinks, ready fruit juices, sweetened ready milk beverages, diet sweetened soft drinks or juices, energy drinks, alcoholic beer or wine beverages, and alcoholic liquors.
Statistical Analysis
Continuous variables were evaluated for normality distribution and reported as median (interquartile range) or mean (standard deviation) as appropriate. Nominal variables are summarized in frequency and are presented as n (%). Food consumption of each of the three processed food (PFQ) categories was divided into tertiles of the number of servings consumed per day. For the univariate analysis, differences between tertiles of UPFs intake of continuous variables were assessed by ANOVA test or the Kruskal-Wallis test, if nonparametric test was required. Pearson χ2 test was used to compare tertiles by nominal variables. The χ2 test of linear trend was used to examine a dose-response association between tertiles of the processed foods intake categories and presence of active disease. Multivariable logistic regressions were used to estimate the adjusted odds ratios (ORs) of active disease according to tertiles of food consumption from each processing level. Given that we excluded unreasonable FFQs, multivariable analyses were performed in two steps of adjustment: model A consisted non-dietary-related potential confounders (age, gender, and body mass index [BMI]), as well as potential confounders previously associated with the study outcome and found to be significantly different across the tertiles of UPFs intake. Model B included model A confounders, as well as dietary-related potential confounders or mediators based on the FFQ. The multivariable analysis was performed on the entire sample, and to avoid a potential bias, another analysis was performed on the entire sample excluding patients that were on elimination diets (N = 31). The differences between the OR for exacerbation of CD and UC were analyzed using the Mantel-Haenszel test of homogeneity. All statistical analyses were performed using the SPSS version 28 statistical analysis software (IBM, USA). All tests are two sided and were considered to be significant at p values <0.05.
Results
Study Population and UPFs Intake
A total of 242 enrolled IBD patients entered the final analysis after exclusion of 32 patients not meeting inclusion criteria or due to missing dietary data (shown in Fig. 1). Mean age was 37.2 ± 13.22 years, 49.2% were females, 62.8% had a diagnosis of CD and 37.2% of UC, with median disease duration of 9.3 (interquartile range: 4.9–17.1) years. In total, 26.9% suffered from a clinically active disease. The majority of the patients received biologics (Table 1). Patients with high UPFs intake had a higher caloric intake and higher proportional intake (% of total caloric intake) of carbohydrates and saturated fatty acids (Table 2).
Characteristic . | Total (n = 242) . | CD (n = 152) . | UC (n = 90) . |
---|---|---|---|
Female sex, n (%) | 119 (49.2) | 74 (48.7) | 45 (50.0) |
Age, median [IQR], years | 34.2 [26.3, 46.1] | 32.6 [25.5, 43.9] | 37.7 [27.7, 48.1] |
BMI, median [IQR] | 23.1 [20.8, 25.8] | 22.8 [20.7, 25.7] | 23.8 [21.1, 26.2] |
Disease duration, median [IQR], years | 9.3 [4.9, 17.1] | 9.3 [4.9, 17.4] | 9.6 [4.4, 17.0] |
Disease activitya | |||
Remission | 177 (73.1) | 108 (71.0) | 69 (76.7) |
Mild | 26 (10.7) | 18 (11.8) | 8 (8.9) |
Moderate | 37 (15.3) | 25 (16.4) | 12 (13.3) |
Severe | 2 (0.9) | 1 (0.6) | 1 (1.1) |
Family history of IBD, n (%) | 49 (20.2) | 33 (21.7) | 16 (17.8) |
Smoking status, n (%) | |||
Never smoker | 165 (68.2) | 101 (66.4) | 64 (71.1) |
Past smoker | 48 (19.8) | 31 (20.4) | 17 (18.9) |
Current smoker | 29 (12.0) | 20 (13.2) | 9 (10.0) |
Diet type, n (%) | |||
Omnivore | 225 (93.0) | 142 (93.4) | 83 (92.2) |
Vegetarian | 10 (4.1) | 6 (3.9) | 4 (4.4) |
Vegan | 7 (2.9) | 4 (2.6) | 3 (3.3) |
Disease location for UC, n (%) | |||
Extensive | - | - | 48 (53.3) |
Left-sided | - | - | 30 (33.3) |
Proctitis | - | - | 12 (13.3) |
Disease location for CD, n (%) | |||
Ileal | - | 62 (40.8) | - |
Colonic | - | 29 (19.1) | - |
Ileo-colonic | - | 61 (40.1) | - |
Disease behavior for CD, n (%) | |||
Non-stricturing, non-penetrating | - | 86 (56.6) | - |
Stricturing | - | 31 (20.4) | - |
Penetrating | - | 35 (23.0) | - |
Current medication treatment, n (%) | |||
Biologics | 169 (69.8) | 120 (78.9) | 49 (54.4) |
5-ASA | 85 (35.1) | 26 (17.1) | 59 (65.5) |
Immunosuppressants | 30 (12.4) | 25 (16.4) | 5 (5.5) |
Steroids | 15 (6.2) | 8 (5.3) | 7 (7.8) |
JAK inhibitors | 7 (2.9) | 1 (0.6) | 6 (6.7) |
No medication therapy | 14 (5.8) | 13 (8.6) | 1 (1.1) |
Current dietary treatment, n (%) | |||
Elimination diet and/or PEN | 31 (12.8) | 22 (14.5) | 9 (10.0) |
Characteristic . | Total (n = 242) . | CD (n = 152) . | UC (n = 90) . |
---|---|---|---|
Female sex, n (%) | 119 (49.2) | 74 (48.7) | 45 (50.0) |
Age, median [IQR], years | 34.2 [26.3, 46.1] | 32.6 [25.5, 43.9] | 37.7 [27.7, 48.1] |
BMI, median [IQR] | 23.1 [20.8, 25.8] | 22.8 [20.7, 25.7] | 23.8 [21.1, 26.2] |
Disease duration, median [IQR], years | 9.3 [4.9, 17.1] | 9.3 [4.9, 17.4] | 9.6 [4.4, 17.0] |
Disease activitya | |||
Remission | 177 (73.1) | 108 (71.0) | 69 (76.7) |
Mild | 26 (10.7) | 18 (11.8) | 8 (8.9) |
Moderate | 37 (15.3) | 25 (16.4) | 12 (13.3) |
Severe | 2 (0.9) | 1 (0.6) | 1 (1.1) |
Family history of IBD, n (%) | 49 (20.2) | 33 (21.7) | 16 (17.8) |
Smoking status, n (%) | |||
Never smoker | 165 (68.2) | 101 (66.4) | 64 (71.1) |
Past smoker | 48 (19.8) | 31 (20.4) | 17 (18.9) |
Current smoker | 29 (12.0) | 20 (13.2) | 9 (10.0) |
Diet type, n (%) | |||
Omnivore | 225 (93.0) | 142 (93.4) | 83 (92.2) |
Vegetarian | 10 (4.1) | 6 (3.9) | 4 (4.4) |
Vegan | 7 (2.9) | 4 (2.6) | 3 (3.3) |
Disease location for UC, n (%) | |||
Extensive | - | - | 48 (53.3) |
Left-sided | - | - | 30 (33.3) |
Proctitis | - | - | 12 (13.3) |
Disease location for CD, n (%) | |||
Ileal | - | 62 (40.8) | - |
Colonic | - | 29 (19.1) | - |
Ileo-colonic | - | 61 (40.1) | - |
Disease behavior for CD, n (%) | |||
Non-stricturing, non-penetrating | - | 86 (56.6) | - |
Stricturing | - | 31 (20.4) | - |
Penetrating | - | 35 (23.0) | - |
Current medication treatment, n (%) | |||
Biologics | 169 (69.8) | 120 (78.9) | 49 (54.4) |
5-ASA | 85 (35.1) | 26 (17.1) | 59 (65.5) |
Immunosuppressants | 30 (12.4) | 25 (16.4) | 5 (5.5) |
Steroids | 15 (6.2) | 8 (5.3) | 7 (7.8) |
JAK inhibitors | 7 (2.9) | 1 (0.6) | 6 (6.7) |
No medication therapy | 14 (5.8) | 13 (8.6) | 1 (1.1) |
Current dietary treatment, n (%) | |||
Elimination diet and/or PEN | 31 (12.8) | 22 (14.5) | 9 (10.0) |
IQR, interquartile range; IBD, inflammatory bowel disease; BMI, body mass index; UC, ulcerative colitis; CD, Crohn’s disease; 5-ASA, 5-aminosalicylic acid; PEN, partial enteral nutrition.
aDisease activity was classified as remission (HBI of <5; SCCAI of <3), mild (HBI of 5–7; SCCAI of 3–5), moderate (HBI of 8–16; SCCAI of 6–11), and severe (HBI of >16; SCCAI of 12–19).
Characteristic . | Tertile 1 (n = 82) . | Tertile 2 (n = 80) . | Tertile 3 (n = 80) . | p value . |
---|---|---|---|---|
Disease type UC/CD, n (%) | 30 (36.6)/52 (63.4) | 30 (37.5)/50 (62.5) | 30 (37.5)/50 (62.5) | 0.990 |
Female sex, n (%) | 46 (56.1) | 30 (37.5) | 43 (53.8) | 0.037 |
Age, median [IQR], years | 34.4 [27.4, 47.0] | 36.0 [26.5, 45.4] | 33.3 [25.0, 46.9] | 0.750 |
BMI, median [IQR], kg/m2 | 23.1 [20.3, 25.5] | 23.2 [21.0, 26.3] | 23.1 [20.5, 25.3] | 0.695 |
Disease duration, median [IQR], years | 8.7 [3.6, 16.2] | 11.8 [5.3, 19.1] | 8.1 [5.0–17.0] | 0.240 |
Clinically active diseasea, n (%) | 21 (25.6) | 16 (20.0) | 28 (35.0) | 0.096 |
Family history of IBD, n (%) | 19 (23.2) | 17 (21.3) | 13 (16.3) | 0.529 |
Smoking status, n (%) | ||||
Never smoker | 58 (70.7) | 56 (70.0) | 51 (63.7) | |
Past smoker | 18 (22.0) | 13 (16.3) | 17 (21.3) | 0.487 |
Current smoker | 6 (7.3) | 11 (13.8) | 12 (15.0) | |
Elimination diet and/or PEN, n (%) | 23 (28.0) | 3 (3.8) | 5 (6.3) | <0.001 |
No medication therapy, n (%) | 9 (11.0) | 2 (2.5) | 3 (3.8) | 0.070 |
Biologic treatment, n (%) | 57 (69.5) | 60 (75.0) | 52 (65.0) | 0.386 |
Meeting with IBD dietitian (last 6 months) | 23 (28.0) | 15 (18.8) | 12 (15.0) | 0.107 |
Disease location for UC, n (%) | ||||
Extensive | 14 (46.7) | 17 (56.7) | 17 (56.7) | |
Left-sided | 11 (36.7) | 9 (30.0) | 10 (33.3) | 0.898 |
Proctitis | 5 (16.7) | 4 (13.3) | 3 (10.0) | |
Disease location for CD, n (%) | ||||
Ileal | 24 (46.2) | 21 (42.0) | 17 (34.0) | |
Colonic | 6 (11.5) | 12 (24.0) | 11 (22.0) | 0.396 |
Ileo-colonic | 22 (42.3) | 17 (34.0) | 22 (44.0) | |
Disease behavior for CD, n (%) | ||||
Non-stricturing, non-penetrating | 34 (65.4) | 23 (46.0) | 29 (58.0) | |
Stricturing | 9 (17.3) | 14 (28.0) | 8 (16.0) | 0.290 |
Penetrating | 9 (17.3) | 13 (26.0) | 13 (26.0) | |
Nutritional data (n = 231; each tertile = 77) | ||||
Energy (mean±SD), kcal/day | 1,596.0 (577.2) | 1,923.3 (678.8) | 2,176.0 (698.4) | <0.001b |
Protein (% of total kcal) (mean±SD) | 19.5 (5.0) | 19.4 (3.8) | 18.4 (3.3) | 0.123 |
Fat (% of total kcal) (mean±SD) | 38.9 (9.1) | 38.4 (5.9) | 37.7 (6.3) | 0.619 |
Carbohydrates (% of total kcal) (mean±SD) | 38.8 (10.3) | 39.5 (6.8) | 41.4 (7.7) | 0.049c |
Saturated fatty acids (% of total kcal) (mean±SD) | 10.9 (4.3) | 11.7 (3.0) | 12.4 (3.3) | 0.010c |
Monounsaturated fatty acids (% of total kcal) (mean±SD) | 16.9 (4.4) | 15.5 (2.9) | 14.6 (3.1) | 0.001c |
Polyunsaturated fatty acids (% of total kcal) (mean±SD) | 8.6 (3.1) | 8.9 (2.9) | 8.5 (2.4) | 0.843 |
Fiber (mean±SD), g/day | 22.1 (10.9) | 24.1 (12.9) | 25.0 (13.4) | 0.305 |
Unprocessed/minimally processed foods (mean±SD), servings/day | 11.0 (5.1) | 10.4 (4.4) | 11.8 (6.5) | 0.502 |
Processed foods (mean±SD), servings/day | 2.3 (1.8) | 2.3 (1.4) | 3.1 (1.8) | 0.001c |
UPFs (mean±SD), servings/day | 1.7 (0.9) | 3.9 (0.5) | 7.3 (2.1) | <0.001 |
Characteristic . | Tertile 1 (n = 82) . | Tertile 2 (n = 80) . | Tertile 3 (n = 80) . | p value . |
---|---|---|---|---|
Disease type UC/CD, n (%) | 30 (36.6)/52 (63.4) | 30 (37.5)/50 (62.5) | 30 (37.5)/50 (62.5) | 0.990 |
Female sex, n (%) | 46 (56.1) | 30 (37.5) | 43 (53.8) | 0.037 |
Age, median [IQR], years | 34.4 [27.4, 47.0] | 36.0 [26.5, 45.4] | 33.3 [25.0, 46.9] | 0.750 |
BMI, median [IQR], kg/m2 | 23.1 [20.3, 25.5] | 23.2 [21.0, 26.3] | 23.1 [20.5, 25.3] | 0.695 |
Disease duration, median [IQR], years | 8.7 [3.6, 16.2] | 11.8 [5.3, 19.1] | 8.1 [5.0–17.0] | 0.240 |
Clinically active diseasea, n (%) | 21 (25.6) | 16 (20.0) | 28 (35.0) | 0.096 |
Family history of IBD, n (%) | 19 (23.2) | 17 (21.3) | 13 (16.3) | 0.529 |
Smoking status, n (%) | ||||
Never smoker | 58 (70.7) | 56 (70.0) | 51 (63.7) | |
Past smoker | 18 (22.0) | 13 (16.3) | 17 (21.3) | 0.487 |
Current smoker | 6 (7.3) | 11 (13.8) | 12 (15.0) | |
Elimination diet and/or PEN, n (%) | 23 (28.0) | 3 (3.8) | 5 (6.3) | <0.001 |
No medication therapy, n (%) | 9 (11.0) | 2 (2.5) | 3 (3.8) | 0.070 |
Biologic treatment, n (%) | 57 (69.5) | 60 (75.0) | 52 (65.0) | 0.386 |
Meeting with IBD dietitian (last 6 months) | 23 (28.0) | 15 (18.8) | 12 (15.0) | 0.107 |
Disease location for UC, n (%) | ||||
Extensive | 14 (46.7) | 17 (56.7) | 17 (56.7) | |
Left-sided | 11 (36.7) | 9 (30.0) | 10 (33.3) | 0.898 |
Proctitis | 5 (16.7) | 4 (13.3) | 3 (10.0) | |
Disease location for CD, n (%) | ||||
Ileal | 24 (46.2) | 21 (42.0) | 17 (34.0) | |
Colonic | 6 (11.5) | 12 (24.0) | 11 (22.0) | 0.396 |
Ileo-colonic | 22 (42.3) | 17 (34.0) | 22 (44.0) | |
Disease behavior for CD, n (%) | ||||
Non-stricturing, non-penetrating | 34 (65.4) | 23 (46.0) | 29 (58.0) | |
Stricturing | 9 (17.3) | 14 (28.0) | 8 (16.0) | 0.290 |
Penetrating | 9 (17.3) | 13 (26.0) | 13 (26.0) | |
Nutritional data (n = 231; each tertile = 77) | ||||
Energy (mean±SD), kcal/day | 1,596.0 (577.2) | 1,923.3 (678.8) | 2,176.0 (698.4) | <0.001b |
Protein (% of total kcal) (mean±SD) | 19.5 (5.0) | 19.4 (3.8) | 18.4 (3.3) | 0.123 |
Fat (% of total kcal) (mean±SD) | 38.9 (9.1) | 38.4 (5.9) | 37.7 (6.3) | 0.619 |
Carbohydrates (% of total kcal) (mean±SD) | 38.8 (10.3) | 39.5 (6.8) | 41.4 (7.7) | 0.049c |
Saturated fatty acids (% of total kcal) (mean±SD) | 10.9 (4.3) | 11.7 (3.0) | 12.4 (3.3) | 0.010c |
Monounsaturated fatty acids (% of total kcal) (mean±SD) | 16.9 (4.4) | 15.5 (2.9) | 14.6 (3.1) | 0.001c |
Polyunsaturated fatty acids (% of total kcal) (mean±SD) | 8.6 (3.1) | 8.9 (2.9) | 8.5 (2.4) | 0.843 |
Fiber (mean±SD), g/day | 22.1 (10.9) | 24.1 (12.9) | 25.0 (13.4) | 0.305 |
Unprocessed/minimally processed foods (mean±SD), servings/day | 11.0 (5.1) | 10.4 (4.4) | 11.8 (6.5) | 0.502 |
Processed foods (mean±SD), servings/day | 2.3 (1.8) | 2.3 (1.4) | 3.1 (1.8) | 0.001c |
UPFs (mean±SD), servings/day | 1.7 (0.9) | 3.9 (0.5) | 7.3 (2.1) | <0.001 |
First, second, and third tertiles of the number of UPFs servings per day were <3.0 servings/day, 3.0–5.1 servings/day, or >5.1 servings/day, respectively. Values of nutritional data based on the final analyzed FFQ.
UPFs, ultra-processed foods; UC, ulcerative colitis; CD, Crohn’s disease; IQR, interquartile range; IBD, inflammatory bowel disease; BMI, body mass index; PEN, partial enteral nutrition.
aActive disease was defined as HBI ≥5 for CD and SCCAI >2 for UC.bp value was significant only in the comparison between tertiles 1–2 and tertiles 1–3.
cp value was significant only in the comparison between tertiles 1–3.
Dietary Intake of Processed Foods Is Associated with Disease Activity
In the multivariable analysis, including all patients, a high intake (upper tertile vs. lowest tertile) of unprocessed/minimally processed foods (OR = 0.38, 95% CI: 0.14–0.99) and processed foods (OR = 0.45, 95% CI: 0.20–1.01) was negatively associated with an active disease, while high intake of UPFs was positively associated with an active disease (OR = 3.82, 95% CI: 1.49–9.8). We found similar associations in a sensitivity analysis, after excluding patients who were on elimination diets. There was a significant negative trend of the association between tertiles of unprocessed/minimally processed or processed foods intake categories and active disease (p trend = 0.005; p trend = 0.003, respectively). All associations between consumption of different levels of food processing and disease activity, as well as serving size of each tertile of the various categories, are presented in Table 3.
Food category . | Partially adjusted model Ab . | Fully adjusted model Bc . | ||||
---|---|---|---|---|---|---|
tertile 1 . | tertile 2 . | tertile 3 . | tertile 1 . | tertile 2 . | tertile 3 . | |
OR (95% CI), p value . | OR (95% CI), p value . | OR (95% CI), p value . | OR (95% CI), p value . | |||
Entire sample model A (N = 242) | Entire sample model B (N = 231) | |||||
Unprocessed/minimally processed | 1 (reference) | 0.84 (0.42–1.65), 0.609 | 0.29 (0.13–0.65), 0.003 | 1 (reference) | 1.09 (0.52–2.30), 1.097 | 0.38 (0.14–0.99), 0.048 |
Processed | 1 (reference) | 0.29 (0.14–0.62), 0.001 | 0.40 (0.19–0.82), 0.013 | 1 (reference) | 0.29 (0.13–0.66), 0.003 | 0.45 (0.20–1.01), 0.052 |
Ultra-processed | 1 (reference) | 1.10 (0.48–2.50), 0.820 | 2.24 (1.05–4.77), 0.037 | 1 (reference) | 1.91 (0.76–4.79), 0.167 | 3.82 (1.49–9.8), 0.005 |
No elimination diet sample model A (N = 211)d | No elimination diet sample model B (N = 201)d | |||||
Unprocessed/minimally processed | 1 (reference) | 0.89 (0.46–1.95), 0.890 | 0.27 (0.11–0.69), 0.006 | 1 (reference) | 1.19 (0.53–2.65), 0.671 | 0.26 (0.08–0.81), 0.021 |
Processed | 1 (reference) | 0.33 (0.15–0.75), 0.008 | 0.34 (0.15–0.75), 0.008 | 1 (reference) | 0.32 (0.13–0.76), 0.010 | 0.35 (0.14–0.88), 0.026 |
Ultra-processed | 1 (reference) | 0.97 (0.40–2.37), 0.955 | 2.16 (0.95–4.91), 0.065 | 1 (reference) | 1.67 (0.62–4.53), 0.312 | 3.42 (1.22–9.54), 0.019 |
Food category . | Partially adjusted model Ab . | Fully adjusted model Bc . | ||||
---|---|---|---|---|---|---|
tertile 1 . | tertile 2 . | tertile 3 . | tertile 1 . | tertile 2 . | tertile 3 . | |
OR (95% CI), p value . | OR (95% CI), p value . | OR (95% CI), p value . | OR (95% CI), p value . | |||
Entire sample model A (N = 242) | Entire sample model B (N = 231) | |||||
Unprocessed/minimally processed | 1 (reference) | 0.84 (0.42–1.65), 0.609 | 0.29 (0.13–0.65), 0.003 | 1 (reference) | 1.09 (0.52–2.30), 1.097 | 0.38 (0.14–0.99), 0.048 |
Processed | 1 (reference) | 0.29 (0.14–0.62), 0.001 | 0.40 (0.19–0.82), 0.013 | 1 (reference) | 0.29 (0.13–0.66), 0.003 | 0.45 (0.20–1.01), 0.052 |
Ultra-processed | 1 (reference) | 1.10 (0.48–2.50), 0.820 | 2.24 (1.05–4.77), 0.037 | 1 (reference) | 1.91 (0.76–4.79), 0.167 | 3.82 (1.49–9.8), 0.005 |
No elimination diet sample model A (N = 211)d | No elimination diet sample model B (N = 201)d | |||||
Unprocessed/minimally processed | 1 (reference) | 0.89 (0.46–1.95), 0.890 | 0.27 (0.11–0.69), 0.006 | 1 (reference) | 1.19 (0.53–2.65), 0.671 | 0.26 (0.08–0.81), 0.021 |
Processed | 1 (reference) | 0.33 (0.15–0.75), 0.008 | 0.34 (0.15–0.75), 0.008 | 1 (reference) | 0.32 (0.13–0.76), 0.010 | 0.35 (0.14–0.88), 0.026 |
Ultra-processed | 1 (reference) | 0.97 (0.40–2.37), 0.955 | 2.16 (0.95–4.91), 0.065 | 1 (reference) | 1.67 (0.62–4.53), 0.312 | 3.42 (1.22–9.54), 0.019 |
First, second, and third tertiles of the number of servings per day were unprocessed/minimally processed foods category <8.4 servings/day, 8.4–12.6 servings/day, and >12.6 servings/day, respectively; processed foods category <1.7 servings/day, 1.7–3.1 servings/day, and >3.1 servings/day, respectively; UPFs category <3.0 servings/day, 3.0–5.1 servings/day, and >5.1 servings/day, respectively. CI, confidence interval; OR, odds ratio; BMI, body mass index.
aActive disease was defined as HBI ≥5 for CD and SCCAI >2 for UC.
bFor partially adjusted model A, ORs are adjusted for sex, age (years), BMI (kg/m2), and being during an elimination diet or PEN (yes/no).
cFully adjusted model B further adjusted for dietary covariates: energy intake (kcal/day), saturated fatty acids (% of total kcal), and monounsaturated fatty acids (% of total kcal) (N = 231 for entire sample, N = 201 for no elimination diet sample).
dPatients during elimination diet treatment or PEN were excluded from the analysis (N = 31 for unadjusted and adjusted model, N = 30 for the fully adjusted model).
Processed Foods Consumption and Disease Activity among CD and UC
A separate analysis for patients with CD (N = 146) demonstrated similar directions of associations (Table 4). In a multivariable analysis of patients with CD, a high intake of unprocessed/minimally processed foods (OR: 0.47, 95% CI: 0.15–1.49) was not associated with disease activity, while high intake of UPFs was positively associated with active disease (OR: 3.36, 95% CI: 1.07–10.53). While the direction of the associations was similar for UC patients, the small sample size and low numbers of active patients (N = 18 in the fully adjusted model) limited the statistical assessment of the OR (online suppl. Table 2). No significant differences were found between the OR for exacerbation of CD and UC in both unprocessed/minimally processed foods and UPFs (highest tertile vs. lowest tertile, p = 0.641 and p = 0.524, respectively).
Food category . | Fully adjusted modelb . | ||
---|---|---|---|
tertile 1 . | tertile 2 . | tertile 3 . | |
OR (95% CI), p value . | OR (95% CI), p value . | ||
Patients with CD (N = 146) | |||
Unprocessed foods | 1 (reference) | 1.19 (0.47–3.03), 0.710 | 0.47 (0.15–1.49), 0.201 |
Processed foods | 1 (reference) | 0.56 (0.20–1.52), 0.254 | 0.81 (0.29–2.22), 0.677 |
UPFs | 1 (reference) | 1.22 (0.38–3.88), 0.737 | 3.36 (1.07–10.53), 0.037 |
Patients with CD – no elimination diet sample (N = 124)c | |||
Unprocessed foods | 1 (reference) | 1.88 (0.66–5.31), 0.234 | 0.42 (0.10–1.69), 0.220 |
Processed foods | 1 (reference) | 0.66 (0.22–1.97), 0.457 | 0.64 (0.20–2.07), 0.456 |
UPFs | 1 (reference) | 0.90 (0.25–3.17), 0.867 | 2.61 (0.74–9.17), 0.134 |
Food category . | Fully adjusted modelb . | ||
---|---|---|---|
tertile 1 . | tertile 2 . | tertile 3 . | |
OR (95% CI), p value . | OR (95% CI), p value . | ||
Patients with CD (N = 146) | |||
Unprocessed foods | 1 (reference) | 1.19 (0.47–3.03), 0.710 | 0.47 (0.15–1.49), 0.201 |
Processed foods | 1 (reference) | 0.56 (0.20–1.52), 0.254 | 0.81 (0.29–2.22), 0.677 |
UPFs | 1 (reference) | 1.22 (0.38–3.88), 0.737 | 3.36 (1.07–10.53), 0.037 |
Patients with CD – no elimination diet sample (N = 124)c | |||
Unprocessed foods | 1 (reference) | 1.88 (0.66–5.31), 0.234 | 0.42 (0.10–1.69), 0.220 |
Processed foods | 1 (reference) | 0.66 (0.22–1.97), 0.457 | 0.64 (0.20–2.07), 0.456 |
UPFs | 1 (reference) | 0.90 (0.25–3.17), 0.867 | 2.61 (0.74–9.17), 0.134 |
First, second, and third tertiles of the number of servings per day were unprocessed foods category <8.4 servings/day, 8.4–12.6 servings/day, and >12.6 servings/day, respectively; processed foods category <1.7 servings/day, 1.7–3.1 servings/day, and >3.1 servings/day, respectively; UPFs category <3.0 servings/day, 3.0–5.1 servings/day, and >5.1 servings/day, respectively. CI, confidence interval; OR, odds ratio; CD, Crohn’s disease.
aActive disease was defined as HBI ≥5.
bFor fully adjusted model, ORs are adjusted for sex, age (years), BMI (kg/m2), being during an elimination diet or PEN (yes/no), energy intake (kcal/day), saturated fatty acids (% of total kcal), and monounsaturated fatty acids (% of total kcal).
cPatients during elimination diet treatment or PEN were excluded from the analysis (N = 22 for patients with CD).
Most UPF Groups Are Associated with Disease Activity
The UPF groups, which were positively associated with higher odds of active disease (shown in Fig. 2), were bread pastries and starch, ultra-processed meat, oils and spreads, and ultra-processed beverages. Ultra-processed meat, including processed red meat, processed poultry, hot dog, sausages, and pastrami, had the strongest association (fully adjusted OR: 4.45, 95% CI: 2.07–9.79) with disease activity. High intake of ultra-processed beverages without alcoholic beverages also had a positive association with active disease, which did not reach statistical significance (partially adjusted and fully adjusted OR: 1.98, 95% CI: 0.95–4.13; 1.78, 95% CI: 0.78–4.07, respectively). Similarly, a high intake of ultra-processed beverages excluding only beer and wine showed a consistent association (partially adjusted OR: 1.94, 95% CI: 0.98–3.82; fully adjusted OR: 1.65, 95% CI: 0.78–3.51).
Discussion
This study examines the association between dietary intake of foods of various processing levels and clinical IBD activity, using a dedicated validated PFQ. Higher intake of UPFs was positively associated with active disease, while unprocessed/minimally processed and processed foods were negatively associated. These associations persisted after further adjustment for nutritional covariates, indicating that UPFs may be related to increased IBD activity through other mechanisms beyond nutrients’ composition.
These results are consistent with the cumulative data evaluating the role of diet and food processing in IBD development and in modulation of disease activity [19‒23]. Furthermore, recent studies using specific dietary interventions, while excluding UPFs and food additives, have shown positive clinical outcomes and mucosal healing in patients with active CD [24] and in patients with active UC [25, 26]. A recent prospective study among IBD patients indicated that high intake of UPFs was associated with adverse disease outcomes such as IBD-related surgery [23]. Similarly, large prospective epidemiological studies suggest that UPFs are associated with a higher IBD incidence and particularly with CD [20, 21, 23]. In the Nurses’ Health Study, a higher UPFs intake was associated with an increased risk of CD incidence but not UC [20]. These results are also consistent with a large prospective study conducted in the UK biobank, which indicated an association only with CD [23]. However, in the PURE cohort study, that examined the association of UPFs intake with the risk of IBD, there was an increased risk of UC [21], although the association was weaker than in CD and was not statistically significant after adjustment. Our analysis of patients with UC implied a similar association with UPFs intake but was probably underpowered to test the association with UC due to the lower number of patients.
A European cohort study that referred to different processing levels showed that a higher consumption of unprocessed/minimally processed foods was associated with lower risk of CD but not UC [27]. However, in that cohort, no associations were found between UPFs intake and CD or UC risk. In contrast, the Nurses’ Health Study did not demonstrate an association of unprocessed/minimally processed intake and lower risk of CD and UC incidence [20].
The association between UPF and IBD activity can be explained by several mechanisms. A mechanistic interplay between food components and disease activity has been suggested, specifically the harmful effect of food additives such as emulsifiers, thickeners, and preservatives present in UPFs on the inflammatory process [9, 10, 28, 29]. Different food additives such as carrageenan, carboxymethyl cellulose, and polysorbate 80 were shown to induce dysbiosis, increase epithelial permeability, and stimulate proinflammatory pathways [9, 10, 29]. In addition to food additives, ultra-processed products are also characterized with high amount of saturated fat, refined carbohydrate, and low content of fiber that could also be related to IBD pathogenesis [29]. There is some evidence from animal models that high fat-high animal protein diet adversely impacts the gut mucus layer and microbiome composition [30, 31] and in epidemiological prospective cohort studies such diets have been associated with increased risk of UC and disease exacerbation [32, 33]. Established evidence exists regarding the protective effect of dietary fibers that relate mostly to the production of short-chain fatty acids by the gut microbiome [34]. Short-chain fatty acids, which have anti-inflammatory properties, were found to be depleted in patients with IBD [34]. As opposed to the UPFs, unprocessed/minimally processed foods are rich in dietary fibers and have low content of saturated fat and animal protein [6, 16].
Our finding of a positive association of all the UPF groups with active disease implies that the processing level is related to the inflammatory process regardless of the nutrient content of a specific food group. These results are consistent with the PURE study, which demonstrated a higher IBD risk with consumption of different food groups of UPFs, including soft drinks, refined sweetened foods, salty snacks, and processed meat [21]. In contrast, in the Nurses’ Health Study cohort, there was a stronger association with specific subgroups, including breads and breakfast foods, ready-to-eat/-heat meals, sauces, cheeses, spreads, and gravies. It was suggested that these food groups are harmful due to a larger content of dietary additives and emulsifiers [20]. Notably, in our study, the strongest association of a food group, of more than four-fold increased odds, with an active disease, was with the ultra-processed meat food group suggesting that in addition to the additives in these processed foods, processing of animal protein and fat may further contribute to IBD activity [30‒33].
Our study has several strengths. First, we used a specific questionnaire, previously developed and validated by us as a tool to assess foods frequency consumption according to the different processing foods levels. The previous discussed epidemiological studies used FFQ or 24-h recalls, which are limited in defining consumption of UPFs intake. FFQ is limited to the food items within the questionnaire and was not designed to distinguish between the processing levels of the food items. Furthermore, a single 24-h recall questionnaire does not document the long-term frequency of foods consumed and is required to be repeated several days in order to reliably reflect the usual food consumption [35]. Second, we referred to the different levels of food processing and not only to the UPFs level. However, we acknowledge certain limitations in our study. Our outcomes were based on symptoms and we did not assess endoscopic inflammation or measure surrogate biomarkers of inflammation (e.g., C-reactive protein, fecal calprotectin). Thus, we could not distinguish if clinical disease activity reflected actual intestinal inflammation or functional disorders. Nevertheless, our results, that indicate an improvement in patients’ disease symptoms and quality of life, are significant for patient treatment on their own. In addition, the cross-sectional study design does not enable certainty in the direction of the association and causality inference; therefore, it only demonstrates an association between risk factor and outcome. However, a prospective cohort study following patients in clinical remission is underway in order to exclude the option that symptoms of an active disease have led to increased UPF consumption. Moreover, while the highest tertile of UPF consumption shows a higher calorie intake, the lack of difference in BMI among the groups suggests that other factors, such as metabolic rate, physical activity, and body composition, may influence energy balance. Further studies incorporating these variables are needed to fully understand the relationship between calorie intake and BMI in patients with IBD. Lastly, a report/recall bias in food intake may have occurred. To minimize this information bias, the PFQ was not transparent to the patients in terms of food processing classification.
In conclusion, in this cross-sectional study we showed that higher intake of UPFs in total and in various food groups of UPFs was positively associated with active disease while unprocessed/minimally processed foods consumption was negatively associated with an active disease in patients with an established diagnosis of CD or UC. Prospective studies are needed to confirm these associations incorporating different outcomes of clinical disease activity, inflammatory biomarkers, and endoscopic assessment.
Statement of Ethics
The study protocol was reviewed and approved by the Institutional Review Board (IRB) of Tel-Aviv Medical Center (Approved No. 0670-17-TLV; 0268-19-TLV). Written informed consent was obtained from each recruited patient.
Conflict of Interest Statement
C.S.S.: Wolfson Medical Center IP for Nestle Health Science and speaking fees from Nestle and Takeda. N.M.: speaking and/or consulting fees from Pfizer, Takeda, AbbVie, Lilly, Janssen, Ferring, BiomX, BMS, Nestle, and Trobix and grant support from Takeda, Janssen, Abbott, AbbVie, Pfizer, BMS, and Nestle. The remaining authors disclose no conflicts.
Funding Sources
The authors received no specific funding for this work.
Author Contributions
The authors’ responsibilities were as follows: C.S.S. conceived and designed the study, did the data collection, performed the statistical analysis, and wrote the manuscript; S.Z.S. conceived and designed the study, provided statistical input, helped with interpretation of the results, and reviewed the manuscript for important intellectual content; N.F.I., A.H., Y.R., R.A., A.B., and T.T. contributed to data collection; L.S.G. contributed to nutritional analysis; N.M. conceived and designed the study, supervised on data collection, helped with interpretation of the results, and reviewed the manuscript for important intellectual content; and all the authors read and approved the final manuscript.
Data Availability Statement
All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.