Systematic Review With Meta-analysis

Dietary Intake in Adults With Inflammatory Bowel Disease

Kelly Lambert; Daniel Pappas; Chiara Miglioretto; Arefeh Javadpour; Hannah Reveley; Laura Frank; Michael C. Grimm; Dorit Samocha-Bonet; Georgina L. Hold


Aliment Pharmacol Ther. 2021;54(6):742-754. 

In This Article



This systematic review was pre-registered on the PROSPERO database on 23 August 2020 (registration number CRD42020205690). Reporting of this review is in accordance with the Guidelines of Preferred Reporting Items for Systematic Reviews and Meta-analyses.[20]

Eligibility Criteria

Observational studies, including cohort, case–control and cross-sectional studies which reported the dietary intake of adults with existing IBD were included. Studies, where intake values were unable to be converted to estimated mean intake, were excluded. Intervention studies such as clinical trials were excluded because clinical trial populations are usually highly selected, and this would limit the generalisability of the findings. Studies of children and adolescents were also excluded because the nutritional requirements of children and adolescents differ from adults making the pooling of results illogical. To ensure the studies reflected contemporary medical management of IBD, studies were restricted to those published after 1 January 2000. Additional information regarding the inclusion and exclusion criteria are shown in Table 1. Abstracts, editorials, case reports, case series, study protocols, review articles and studies not in the English language were excluded.

Data Sources and Search Strategy

Studies were identified by a systematic search of four electronic databases and hand searching reference lists of all eligible studies. The four databases searched from 1 January 2000 to 25 September 2020 were: CINAHL, Embase, Medline and Scopus. These databases were searched using a search strategy adapted from previous work on systematic searches of dietary intake.[21] The search included MeSH terms (feeding behavior OR eating OR food intake OR diet* OR nutrition*) AND (inflammatory bowel disease OR Crohn's disease OR Ulcerative Colitis). The search strategy for Medline is shown in Table S1.

All references were downloaded into EndNote X9 (Thompson Reuters). Duplicates were removed. All titles and abstracts were screened against the eligibility criteria by two researchers (KL, and either DP, CM or AJ) and the full-text versions of eligible studies were obtained. Disagreements between reviewers were resolved via discussion with the main author (KL).

Data Extraction

Data were extracted from each eligible study according to the criteria outlined in Table 1. A standardised data extraction sheet was developed, and two reviewers independently extracted the data (KL and DP). Several reviewers then checked the accuracy of the extracted data (CM, LF and AJ). Where energy information was reported as kilojoules/megajoules per day it was converted to kilocalories. Similarly, values for food groups were converted from ounces to grams per day, or cups to grams per day. To calculate the estimated sample mean and SD intake when median and IQR were reported, the R package "estmeansd" ( was used. This was based on the Box-Cox transformation approach in studies where median, IQR and sample size were available.[22,23] Conversion to the estimated sample mean intake was not performed in studies with n < 25. Where sample size, median and range were reported, the estimated sample mean was calculated using[24] the online calculator at Studies, where estimated sample mean, were unable to be calculated by either method were excluded. The weighted mean was used to summarise data for each nutrient to account for the variation in sample sizes for included studies. Data were reported as the weighted mean intake for several prespecified subgroups to assist with analysis: All IBD; ulcerative colitis; Crohn's disease; males; females; remission (all, ulcerative colitis; Crohn's disease); Active disease (all, ulcerative colitis; Crohn's disease). Data were included only once in the analyses if multiple publications reported on the same dietary dataset (eg Vahid et al[25] and Rahmani et al;[26] Ueda et al[27] and Kawakami et al[28]). For one study, data from non-healthy controls were excluded.[29] Data from studies with implausible units were also excluded from the analysis of that nutrient, for example vitamin A intake,[30,31] energy intake.[32] Data for sugar intake were excluded if the differentiation of added from total sugar was unclear.[33] Results are reported as whole numbers where appropriate to assist with interpretation when comparing to nutrient reference values.

Two independent reviewers (KL and HR) assessed the risk of bias of individual studies using the Academy of Nutrition and Dietetics Criteria Checklist for primary research.[34] The tool contains 10 items evaluating study quality. The overall rating of study quality is scored as a positive, negative or neutral. Percentage agreement and Kappa statistic were calculated to check the agreement between reviewers and any differences were resolved by a third reviewer.


Studies of usual intake in adults with IBD were eligible for meta-analysis if more than two studies on the same outcome were available and (a) reported useable data in a compatible metric and (b) had a matched control group assessed at the same time. RevMan5 (Review Manager (RevMan) [Computer program]. Version 5.4.1, The Cochrane Collaboration, 2020) was used to conduct analyses. To account for heterogeneity between the studies, a random-effects model was used. Data synthesis was reported as standardised mean difference (SMD) (calculated as the difference in mean outcome between groups divided by the standard deviation of the outcome among participants). This approach was used to allow comparison across nutrients using the same scale, and to enable comparisons regarding the size of the effect in each nutrient category relative to the variability observed. The SMD was also preferred due to heterogeneity in data collection methods for collecting nutrient data, and the variations in standard deviation between studies were assumed to be a result of differences in dietary collection methods. Statistical significance was set at P < 0.05. Variance between studies was evaluated and reported as I 2, which indicates the degree of variance resulting from between study heterogeneity, where a high score closer to 100 indicates high heterogeneity between studies. No meta-regressions were performed due to a low number of available studies for individual nutrients.