’Review article Consumption of vegetables and fruit and the risk of inflammatory bowel disease: a meta-analysis Fang Li, Xiaoqin Liu, Weijing Wang and Dongfeng Zhang To date, associations between consumption of vegetables and fruit and the risk of inflammatory bowel disease have been a controversial subject. Therefore, we carried out a meta-analysis to evaluate the associations. A comprehensive search was performed in PubMed, Embase, Web of Science, and the China National Knowledge Infrastructure to identify all relevant studies. Pooled odds ratios (ORs) with 95% confidence intervals (CIs) from random-effects or fixed-effects models were calculated. Publication bias was estimated using Egger’s test and the funnel plot. A total of 14 case–control studies were included in this meta-analysis. On the basis of the highest versus the lowest analysis, consumption of vegetables was associated inversely with the risk of ulcerative colitis (UC) (OR = 0.71, 95% CI 0.58–0.88, n = 9 studies), but not with Crohn’s disease (CD) (OR = 0.66, 95% CI 0.40–1.09, n = 8 studies). Higher consumption of fruit was associated inversely with the risk of UC (OR = 0.69, 95% CI 0.49–0.96, n = 8 studies) and CD (OR = 0.57, 95% CI 0.44–0.74, n = 10 studies). For intake of vegetables and the risk of CD, subgroup analysis showed a significant association for studies carried out in Europe (OR = 0.36, 95% CI 0.23–0.57), but not in Asia (OR = 1.00, 95% CI 0.50–2.03). No significant publication bias was found for the analysis of intake of vegetables and the risk of UC, intake of fruit and the risk of UC, and intake of vegetables and the risk of CD. This meta-analysis indicates that consumption of vegetables and fruit might be associated inversely with the risk of UC and CD, and the results need to be further confirmed. Eur J Gastroenterol Hepatol 27:623–630 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Department of Epidemiology and Health Statistics, the Medical College of Qingdao University, Qingdao, Shandong, People’s Republic of China
of diet in the development of IBD is unclear [15,16]. Many foods or food components have been studied as potential etiological factors for IBD. A recent systematic review reported a protective effect of ω-3 polyunsaturated acids on UC, whereas animal protein was found to be associated with an increased risk of IBD [17]. Vegetables and fruit, rich in multiple beneficial nutrients, including monounsaturated and polyunsaturated fatty acids, n-3 fatty acids, fiber, vitamins, minerals, and phytochemicals, have been hypothesized to protect against chronic diseases generally [18], such as cardiovascular disease and cancer. Several epidemiological studies have investigated associations between intake of vegetables and fruit and the risk of IBD, but the findings are inconsistent: one study carried out by Hansen et al. [19] reported that the consumption of both vegetables and fruit was associated significantly with a decreased risk of UC and CD, whereas another study carried out by Ng et al. [20] showed no significant association between them. Therefore, we carried out a comprehensive metaanalysis by combining the results from all available observational studies to (i) assess the risk of UC and CD for highest versus lowest consumption of vegetables and fruit separately and (ii) explore the potential betweenstudy heterogeneity and publication bias.
Correspondence to Dongfeng Zhang, MS, Department of Epidemiology and Health Statistics, the Medical College of Qingdao University, 38 Dengzhou Road, Qingdao, Shandong 266021, People’s Republic of China
Materials and methods
Introduction
Inflammatory bowel disease (IBD), including ulcerative colitis (UC) and Crohn’s disease (CD), has become a global disease [1]. The incidence of IBD increased steadily from the 19th century and remained stable until the middle and latter part of the 20th century in North Europe and America [2,3]. In recent years, the incidence of IBD has been increasing in most developing countries, such as China, Japan, India, Iran, Thailand, etc. [4–6]. This geographical variation of IBD has been associated with a westernized lifestyle and environmental factors [1,7]. Smoking and appendectomy have been identified as risk factors for IBD [8–10]. In addition, urbanization of societies, antibiotic use, use of oral contraceptives, and changes in diet are also potential risk factors for IBD [11–14]. The association between diet and the risk of IBD was proposed as early as 50 years ago; however, the exact role European Journal of Gastroenterology & Hepatology 2015, 27:623–630 Keywords: Crohn’s disease, fruit, inflammatory bowel disease, ulcerative colitis, vegetables
Tel: + 86 532 8299 1712; fax: + 86 532 8380 1449; e-mails: [email protected], [email protected] Received 4 November 2014 Accepted 3 February 2015 Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's website (www.eurojgh.com).
Search strategy
To identify all relevant studies, we performed a computerbased search up to 15 September 2014 from the following databases: (i) PubMed, (ii) Embase, (iii) Web of Science (ISI), and (iv) the China National Knowledge Infrastructure
0954-691X Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
DOI: 10.1097/MEG.0000000000000330
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
623
624
June 2015 • Volume 27 • Number 6
European Journal of Gastroenterology & Hepatology
4796 articles from PubMed 7378 articles from Embase 6371 articles from ISI 2851 articles from CNKI
14 513 articles screened after excluding duplicates
14 410 articles excluded after review of title or abstract
103 full-text articles reviewed
89 articles excluded because: 59 did not evaluate this association of interest 11 review 7 lacking OR (RR) or corresponding 95% CI 6 not English or Chinese 2 vegetables and fruit combined 4 duplicate
14 articles included in the meta-analysis Fig. 1. Flowchart of the selection of studies included in the meta-analysis. CI, confidence internal; CNKI, China National Knowledge Infrastructure; ISI, Web of Science; OR, odds ratio; RR, relative risk.
(CNKI) using the key words or Mesh terms ‘fruit’, ‘vegetables’, ‘lifestyle’, ‘diet’, ‘nutrition’ in combination with ‘inflammatory bowel disease’, ‘ulcerative colitis’, and ‘Crohn’s disease’. Moreover, we reviewed the reference lists of all retrieved articles to search for undetected relevant studies. The flow diagram of the literature search is shown in Fig. 1. Inclusion criteria
All the studies identified were reviewed independently by two investigators and studies were included if they fulfilled the following criteria: (i) observational studies published originally; (ii) the topic of interest was consumption of vegetables and/or fruit; (iii) the outcome was UC and/or CD; (iv) the odds ratios (ORs) or relative risks with corresponding 95% confidence intervals (CIs) were reported or could be calculated from the data presented in articles; and (v) studies were reported in English or Chinese. If data were duplicated in more than one study, the most recent study was included in this meta-analysis. Data extraction
Data were extracted independently by two investigators who reached a consensus on all of the items. Information extracted from each study was as follows: name of the first author, publication year, country in which the study was
carried out, sex and age of the participants, number of cases, the ORs with corresponding 95% CIs for the highest versus lowest consumption, variables adjusted in the analysis, and control resource. Statistical analysis
We weighted the study-specific log ORs by the inverse of their variance to calculate pooled ORs with corresponding 95% CIs of the associations between consumption of vegetables and fruit and the risk of UC and CD. I2 of Higgins and Thompson [21] was used to assess heterogeneity among studies and I2 values of 0, 25, 50, and 75% represent no, low, moderate, and high heterogeneity, respectively. The DerSimonian and Laird random-effects model (REM) was selected as the pooling method if considerable heterogeneity was present (I2 > 50%) [22]; otherwise, the fixed-effects model (FEM) was used. Metaregression with restricted maximum likelihood estimation was used to explore the potential important covariates that might exert a major impact on between-study heterogeneity [23]. Subgroup analysis was carried out by the continent (Asia and Europe) and the status (yes or no) of adjusting for smoking. The leave-one-out sensitivity analysis was carried out to evaluate the key studies that had a major impact on between-study heterogeneity [24].
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
CD CD
42/51
< 20
The Netherlands
China
Japan
Sweden Denmark
China
England
Canada
Italy
Denmark
Chen et al. [36]
Sakamoto et al. [28]
Halfvarson et al. [29]
Shi et al. [35]
Abubakar et al. [31] Amre et al. [30]
Maconi et al. [32]
Hansen et al. [19]
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
38.5/40.5
37.5/40.4
33.10/32.90
5–79 11–79
15–34
22–75
39/37
UC CD
UC CD
CD
UC CD
UC CD
UC
UC CD
UC CD
UC
Disease type
Russel et al. [34]
30.14/29.45 29.12/29.45
15–79
Age range/mean age (case/control)
Israel
Sweden
Country
Persson et al. [26] Reif et al. [27]
References
144 123
26 25
130
218
51
125 102
108 126
96
398 290
54 33
145
Cases
Daily vs. rarer intake
High vs. low
Quartile 4 vs. quartile 1
High vs. low
Daily vs. less frequent
Daily vs. less frequent
Quartile 4 vs. quartile 1
High vs. low
> 5 per week vs. none or 1 per week
High vs. low
Daily vs. less frequent
Vegetable or fruit consumption 0.7 (0.4–1) for vegetable and UC 1.28 (0.55–2.95) for vegetable and UC 0.65 (0.25–1.65) for fruit and UC 0.32 (0.09–1.12) for vegetable and CD 0.42 (0.13–1.38) for fruit and CD 0.5 (0.4–0.0.8) for fruit and UC 0.5 (0.3–0.7) for fruit and CD 0.42 (0.19–0.93) for vegetable and UC 0.75 (0.35–1.62) for vegetable and UC 0.62 (0.29–1.32) for fruit and UC 1.55 (0.76–3.17) for vegetable and CD 0.8 (0.38–1.66) for fruit and CD 2.9 (0.9–9.4) for fruit and UC 0.2 (0.1–0.9) for fruit and CD 2.34 (0.9–6.09) for vegetable and CD 0.39 (0.19–0.74) for fruit and CD 0.78 (0.7–0.88) for fruit and CD 0.69 (0.33–1.44) for vegetable and CD 0.37 (0.16–0.86) for fruit and CD 0.37 (0.11–1.21) for vegetable and UC 0.41 (0.13–1.35) for fruit and UC 0.21 (0.05–0.78) for vegetable and CD 0.76 (0.26–2.21) for fruit and CD 0.51 (0.31–0.84) for vegetable and UC 0.56 (0.33–0.95) for fruit and UC 0.41 (0.24–0.71) for vegetable and CD 0.39 (0.22–0.7) for fruit and CD
OR (95% CI)
Smoking status, appendectomy, tonsillectomy, usage of oral contraceptives, consumption of fiber, sugar, coffee, and eggs
Age, sex, years of education, smoking, BMI
Age, sex, and BMI, total energy intake
Age, sex, study center
Not provided
Not provided
Age, sex, study area, education, and smoking habit
Not provided
Smoking, sex, age, educational level, and selected nutritional factors
Hospital based
Hospital based
Population and hospital based
Population based
Hospital based
Population based
Hospital based
Population based
Population based
Population and hospital based
Energy intake
Control resource Population based
Adjustment for covariates Age and sex
Table 1. Characteristics of the studies included on consumption of vegetables and fruit and the risk of ulcerative colitis and Crohn’s disease
Vegetables and fruit and IBD Li et al. www.eurojgh.com 625
Population based
Population based
Age, sex, ethnicity, area of residence and socioeconomic status
Sex, age and country income based on GNI
Population based
The influence analysis was carried out with the exclusion of one study at a time to assess whether the results could have been affected markedly by a single study. Publication bias was evaluated by visual inspection of the funnel plot and Egger’s test [25]. All statistical analyses were carried out using STATA Version 10 (StataCorp, College Station, Texas, USA). All reported P values were two-sided, with P value less than 0.05 considered statistically significant. Results Search results and characteristics of the included studies
The search strategy identified 4796 articles from PubMed, 6371 articles from ISI, 7378 articles from Embase, and 2851 articles from CNKI. After reviewing the title/abstract, 103 articles were retrieved. Then, 89 articles were subsequently excluded from the meta-analysis for various reasons. As a result, 14 articles fulfilled the inclusion criteria and were included in the meta-analysis [19,20,26–37]. Of these 14 articles, seven were from Europe [19,26,29,30,32–34], one was from America [31], and six were from Asia [20,27,28, 35–37] (the study carried out by Ng et al. [20] investigated eight Asia countries and Australia, respectively). All the included studies were case–control studies. The baseline characteristics of the studies are shown in Table 1. Quantitative synthesis Vegetables and ulcerative colitis
CD, Crohn’s disease; CI, confidence interval; GNI, gross national income; OR, odds ratio; UC, ulcerative colitis.
Daily vs. weekly or less frequent UC CD Nine countries Ng et al. [20]
38/39
256 186
Daily or > once weekly vs. 1 time weekly or less 56 59 UC CD 3–15 Denmark Jakobsen et al. [33]
UC China Zhou and Liu [37]
20–50
11
High vs. low ≥ 7 times vs. ≤ 2 times per week
1.72 (0.41–7.24) for vegetable and UC 0.57 (0.05–6.94) for fruit and UC 0.3 (0.1–0.8) for vegetable and UC 0.3 (0.1–1) for vegetable and CD 1.03 (0.68–1.54) for vegetable and UC 0.92 (0.71–1.28) for fruit and UC 0.81 (0.51–1.29) for vegetable and CD 0.92 (0.64–1.31) for fruit and CD
Unadjusted
Adjustment for covariates OR (95% CI) Vegetable or fruit consumption Cases Disease type Age range/mean age (case/control) Country References
Table 1. (Continued)
June 2015 • Volume 27 • Number 6
European Journal of Gastroenterology & Hepatology
Control resource
626
Nine studies [19,20,26–28,32,33,36,37] including 896 cases investigated the association between intake of vegetables and the risk of UC. Three of the nine studies showed a significant association, whereas the other six studies indicated no relation. The pooled OR of UC for the highest versus the lowest consumption of vegetables was 0.71 (95% CI 0.58–0.88, I2 = 41.6%, FEM, Pheterogeneity = 0.090; Fig. 2). Subgroup analysis by continent (Asia, Europe) and status of adjusting for smoking was carried out, and the results showed that higher intake of vegetables was associated inversely with the risk of UC for studies carried out in Europe (OR = 0.55, 95% CI 0.40–0.75, I2 = 0%, FEM), but not in Asia (OR = 0.89, 95% CI 0.66–1.21, I2 = 26.7%, FEM). The pooled ORs were 0.55 (95% CI 0.37–0.81, I2 = 0%, FEM) among studies that adjusted for smoking and 0.86 (95% CI 0.66–1.13, I2 = 44.6%, FEM) among studies that did not adjust for smoking (Table 2). Fruit and ulcerative colitis
Eight studies [19,20,27–29,32,34,37] including 1122 cases investigated the association between intake of fruit and the risk of UC. Two of the eight studies showed a significant association, whereas the other six studies indicated no relation. The pooled OR of UC for the highest versus the lowest consumption of fruit was 0.69 (95% CI 0.49–0.96, I2 = 50.7%, REM, Pheterogeneity = 0.048; Supplementary Fig. 1, Supplemental digital content 1, http://links.lww.com/ EJGH/A13). The results from subgroup analysis showed that the pooled ORs were 0.66 (95% CI 0.37–1.16, I2 = 63.6%, REM) for studies carried out in Europe and 0.82 (95% CI 0.63–1.07, I2 = 0%, FEM) in Asia. The pooled ORs were 0.52 (95% CI 0.40–0.68, I2 = 0%, FEM)
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
Vegetables and fruit and IBD Li et al.
www.eurojgh.com
References
OR (95% CI)
Weight (%)
Persson et al. [26]
0.70 (0.40–1.00)
22.07
Reif et al. [27]
1.28 (0.55–2.95)
6.57
Chen et al. [36]
0.42 (0.19–0.93)
7.35
Sakamoto et al. [28]
0.75 (0.35–1.62)
7.89
Maconi et al. [32]
0.37 (0.11–1.21)
3.22
Hansen et al. [19]
0.51 (0.31–0.84)
18.65
Zhou and Liu [37]
1.72 (0.41–7.24)
2.25
Jakobsen et al. [33]
0.30 (0.10–0.80)
4.28
Ng et al. [20]
1.03 (0.68–1.54)
27.73
Overall (I = 41.6%, P = 0.090)
0.71 (0.58–0.88)
100.00
2
0.1
0.2
0.5
1
2
5
627
10
Fig. 2. Meta-analysis of the association between intake of vegetables and the risk of ulcerative colitis. The size of the gray box is positively proportional to the weight assigned to each study, which is inversely proportional to the standard error of the odds ratio (OR), and horizontal lines represent the 95% confidence interval (CI).
Table 2. Subgroup analyses of consumption of vegetables and fruit and the risk of ulcerative colitis and Crohn’s disease Subgroup Vegetables and UC
Fruit and UC
Vegetables and CD
Fruit and CD
Continent Asia Europe Adjustment Yes No Continent Asia Europe Adjustment Yes No Continent Asia Europe Adjustment Yes No Continent Asia Europe Adjustment Yes No
Number of studies
OR (95% CI)
Effect model
I2 (%)
P value
Article included
5 4
0.89 (0.66–1.21) 0.55 (0.40–0.75)
FEM FEM
26.7 0
0.243 0.411
[20,27,28,36,37] [19,26,32,33]
3 5
0.55 (0.37–0.81) 0.86 (0.66–1.13)
FEM FEM
0 44.6
0.567 0.125
[19,28,32] [20,26,27,33,37]
4 4
0.82 (0.63–1.07) 0.66 (0.37–1.16)
FEM REM
0 63.6
0.788 0.041
[20,27,28,37] [19,29,32,34]
4 3
0.52 (0.40–0.68) 0.89 (0.67–1.17)
FEM FEM
0 0
0.921 0.741
[19,28,32,34] [20,27,37]
4 3
1.00 (0.50–2.03) 0.36 (0.23–0.57)
REM FEM
68.6 0
0.023 0.635
[20,27,28,35] [19,32,33]
3 4
0.56 (0.19–1.66) 0.66 (0.46–0.94)
REM FEM
81.7 21.7
0.004 0.280
[19,28,32] [20,27,31,33]
4 5
0.70 (0.52–0.94) 0.52 (0.34–0.80)
FEM REM
37.5 71.8
0.187 0.007
[20,27,28,35] [19,29,30,32,34]
4 4
0.52 (0.39–0.70) 0.78 (0.70–0.87)
FEM FEM
0 38.7
0.427 0.180
[19,32,34,35] [20,27,30,31]
for smoking
for smoking
for smoking
for smoking
CI, confidence interval; CD, Crohn’s disease; FEM, fixed-effects model; OR, odds ratio; REM, DerSimonian and Laird random-effects model; UC, ulcerative colitis.
among studies that adjusted for smoking and 0.89 (95% CI 0.67–1.17, I2 = 0%, FEM) among studies that did not adjust for smoking (Table 2). One study [20] was found to contribute toward between-study heterogeneity by the leave-one-out sensitivity analysis. After further excluding the study, low heterogeneity (I2 = 28.3%) was found and the pooled OR was 0.61 (95% CI 0.44–0.86). Vegetables and Crohn’s disease
Eight studies [19,20,27,28,30,32,33,35] including 733 cases investigated the association between intake of vegetables and the risk of CD. Two of the eight studies showed a significant
association, whereas the other six studies indicated no relation. The pooled OR of CD for the highest versus the lowest consumption of vegetables was 0.66 (95% CI 0.40–1.09, I2 = 67.5%, REM, Pheterogeneity = 0.003; Supplementary Fig. 2, Supplemental digital content 2, http://links.lww.com/ EJGH/A14). The results from subgroup analysis showed that higher intake of vegetables was associated inversely with the risk of CD for studies carried out in Europe (OR = 0.36, 95% CI 0.23–0.57, I2 = 0%, FEM), but not in Asia (OR = 1.00, 95% CI 0.50–2.03, I2 = 68.6%, REM). The pooled ORs were 0.56 (95% CI 0.19–1.66, I2 = 81.7%, REM) among studies that adjusted for smoking and 0.66 (95% CI 0.46–0.94, I2 = 21.7%, FEM) among studies did not adjust for smoking (Table 2).
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
June 2015 • Volume 27 • Number 6
European Journal of Gastroenterology & Hepatology
Two studies [28,35] were found to contribute toward between-study heterogeneity by the leave-one-out sensitivity analysis. After further excluding these two studies, low heterogeneity (I2 = 36.1%) was found and the pooled OR was 0.5 (95% CI 0.33–0.75). Fruit and Crohn’s disease
Ten studies [19,20,27–32,34,35] including 1284 cases investigated the association between intake of fruit and the risk of CD. Six of the 10 studies showed a significant association, whereas the other four studies indicated no relation. The pooled OR of CD for the highest versus the lowest consumption of fruit was 0.57 (95% CI 0.44–0.74, I2 = 59.5%, REM, Pheterogeneity = 0.008; Supplementary Fig. 3, Supplemental digital content 3, http://links.lww. com/EJGH/A15). The results from subgroup analysis showed that the pooled ORs were 0.52 (95% CI 0.34–0.80, I2 = 71.8%, REM) for studies carried out in Europe and 0.70 (95% CI 0.52–0.94, I2 = 37.5%, FEM) in Asia. The pooled ORs were 0.52 (95% CI 0.39–0.70, I2 = 0%, FEM) among studies that adjusted for smoking and 0.78 (95% CI 0.70–0.87, I2 = 38.7%, FEM) among studies that did not adjust for smoking (Table 2). Two studies [20,31] were found to contribute toward between-study heterogeneity by the leave-one-out sensitivity analysis. After further excluding these two studies, the heterogeneity (I2 = 0%) was decreased and the pooled OR was 0.46 (95% CI 0.36–0.59). Meta-regression
To explore the moderate to high between-study heterogeneity in these above-mentioned analyses, meta-regression with the covariates of year, number of cases, control resource, continent, and status of adjusting for smoking was carried out. The status of adjusting for smoking was found to contribute toward the heterogeneity in the analysis of intake of fruit and the risk of UC (P = 0.043). No significant findings were found for other analyses. Influence analysis and publication bias
Influence analysis showed that no individual study had an excessive influence on the above-mentioned pooled ORs. The funnel plot and Egger’s test showed no evidence of significant publication bias for the analysis of consumption of vegetables and the risk of UC (P = 0.558; Fig. 3), consumption of fruit and the risk of UC (P = 0.980; Supplementary Fig. 4, Supplemental digital content 4, http:// links.lww.com/EJGH/A16), and consumption of vegetables and the risk of CD (P = 0.601; Supplementary Fig. 5, Supplemental digital content 5, http://links.lww.com/EJGH/ A17). For consumption of fruit and the risk of CD, publication bias was found (P = 0.024; Supplementary Fig. 6, Supplemental digital content 6, http://links.lww.com/EJGH/ A18); however, after excluding two studies that had a strong effect on heterogeneity, no significant publication bias was found (P = 0.677; Supplementary Fig. 7, Supplemental digital content 7, http://links.lww.com/EJGH/A19) and the result remained significant.
0
0.2 SE of log OR
628
0.4
0.6
0.8 1 OR (log scale)
2
3
Fig. 3. Funnel plot with pseudo 95% confidence limits for the analysis of intake of vegetable and the risk of ulcerative colitis. OR, odds ratio.
Discussion
This meta-analysis assessed associations between intake of vegetables and fruit and the risk of UC and CD, respectively. Findings showed inverse associations between intake of vegetables and the risk of UC, intake of fruit and the risk of UC, and intake of fruit and the risk of CD overall. However, the inverse association between intake of vegetables and the risk of CD was only observed for studies carried out in Europe. Some potential mechanisms may explain the inverse associations in this meta-analysis. Vegetables and fruit are abundant in fiber, micronutrients (such as vitamins C and E and folate), and phytochemicals (such as carotenoids, phenolic, isoflavones, and indoles). First, butyrate, the major anion produced by the bacterial fermentation of dietary fiber in the colon, can reduce mucosal inflammation by reducing nuclear factor-κB activity in colon cells [38]. Second, plant fiber may influence the translocation of microbes across gut mucosa [39]. Third, vitamin A has been found to be necessary for the expression of surface markers (such as α4β7 and CCR9), which can correct homing of the cells to the gut [40]. Finally, flavonoids seem to be involved in maintenance of the intercellular tight junctions, which is one of the major determinants of the intestinal barrier function [41], and impairment in intestinal barrier function has been associated with IBD [42]. In our meta-analysis, moderate heterogeneity was found in the analysis of intake of fruit and the risk of UC, intake of vegetables and the risk of CD, and intake of fruit and the risk of CD. Therefore, the leave-one-out sensitivity analysis was used to reduce the heterogeneity by excluding studies that were found to contribute toward the between-study heterogeneity. The results remained significant after reducing the heterogeneity in the analysis of intake of fruit and the risk of UC, and intake of fruit and the risk of CD. To further explore the potential sources of between-study heterogeneity, meta-regression with covariates of year, number of cases, control resource, continent, and status of adjusting for smoking was carried out. Findings showed that status of adjusting for smoking contributed toward the between-study heterogeneity (P = 0.043) in the analysis of consumption of fruit and the risk of UC. The heterogeneity decreased to 0% (Pheterogeneity = 0.741) among studies that
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
Vegetables and fruit and IBD Li et al.
adjusted for smoking and 0% (Pheterogeneity =0.921) among studies that did not adjust for smoking. There are some advantages in our meta-analysis. First, to our knowledge, this is the first meta-analysis to explore associations between intake of vegetables and fruit and the risk of UC and CD, and the associations observed remained significant in the leave-one-out sensitivity analysis, indicating that our results were stable and reliable. Second, a relatively larger number of participants included, with a reduction in sampling error to a huge extent, enabled a much greater possibility of reaching reasonable conclusions. Third, no publication bias was observed for the analysis of consumption of vegetables and the risk of UC, consumption of fruit and the risk of UC, and consumption of vegetables and the risk of CD. However, several limitations should be acknowledged in our meta-analysis. First, the findings based on case–control studies were prone to recall bias and interviewer bias. Second, confounders adjusted in studies were different, which could affect associations between intake of vegetables and fruit and the risk of IBD. For example, some studies adjusted for smoking, whereas others did not. Third, the diet assessment methods and the retrospective time were different in the original studies, which led to incomparability in the results to some extent. Some studies used food frequency questionnaires and others used dietary history questionnaires. Fourth, the results of subgroup analysis were based on a limited number of studies. In summary, the results from this meta-analysis suggest that consumption of fruit reduces the risk of both UC and CD, and consumption of vegetables reduces the risk of UC. These findings need to be confirmed further by cohort studies and randomized controlled trials.
www.eurojgh.com
10 11
12
13
14
15
16
17 18
19
20
21 22 23
Acknowledgements
24
Conflicts of interest
There are no conflicts of interest.
25 26
References 1
2 3 4
5 6
7
8
9
Ng SC, Bernstein CN, Vatn MH, Lakatos PL, Loftus EV Jr, Tysk C, et al. Epidemiology and Natural History Task Force of the International Organization of Inflammatory Bowel Disease (IOIBD). Geographical variability and environmental risk factors in inflammatory bowel disease. Gut 2013; 62:630–649. M’Koma AE. Inflammatory bowel disease: an expanding global health problem. Clin Med Insights Gastroenterol 2013; 6:33–47. Burisch J, Jess T, Martinato M, Lakatos PL. The burden of inflammatory bowel disease in Europe. J Crohns Colitis 2013; 7:322–337. Leong RW, Lau JY, Sung JJ. The epidemiology and phenotype of Crohn’s disease in the Chinese population. Inflamm Bowel Dis 2004; 10:646–651. Makharia GK. Rising incidence and prevalence of Crohn’s disease in Asia: is it apparent or real? J Gastroenterol Hepatol 2006; 21:929–931. Thia KT, Loftus EV Jr, Sandborn WJ, Yang SK. An update on the epidemiology of inflammatory bowel disease in Asia. Am J Gastroenterol 2008; 103:3167–3182. Karlinger K, Györke T, Makö E, Mester A, Tarján Z. The epidemiology and the pathogenesis of inflammatory bowel disease. Eur J Radiol 2000; 35:154–167. Kaplan GG, Jackson T, Sands BE, Frisch M, Andersson RE, Korzenik J. The risk of developing Crohn’s disease after an appendectomy: a metaanalysis. Am J Gastroenterol 2008; 103:2925–2931. Lakatos PL, Vegh Z, Lovasz BD, David G, Pandur T, Erdelyi Z, et al. Is current smoking still an important environmental factor in inflammatory
27 28
29
30
31
32
33
629
bowel diseases? Results from a population-based incident cohort. Inflamm Bowel Dis 2013; 19:1010–1017. Danese S, Fiocchi C. Etiopathogenesis of inflammatory bowel diseases. World J Gastroenterol 2006; 12:4807–4812. Spehlmann ME, Begun AZ, Saroglou E, Hinrichs F, Tiemann U, Raedler A, Schreiber S. Risk factors in German twins with inflammatory bowel disease: results of a questionnaire-based survey. J Crohns Colitis 2012; 6:29–42. Loftus EV Jr. Clinical epidemiology of inflammatory bowel disease: incidence, prevalence, and environmental influences. Gastroenterology 2004; 126:1504–1517. Tragnone A, Valpiani D, Miglio F, Elmi G, Bazzocchi G, Pipitone E, Lanfranchi GA. Dietary habits as risk factors for inflammatory bowel disease. Eur J Gastroenterol Hepatol 1995; 7:47–51. Castiglione F, Diaferia M, Morace F, Labianca O, Meucci C, Cuomo A, et al. Risk factors for inflammatory bowel diseases according to the ‘hygiene hypothesis’: a case–control, multi-centre, prospective study in Southern Italy. J Crohns Colitis 2012; 6:324–329. Cashman KD, Shanahan F. Is nutrition an aetiological factor for inflammatory bowel disease? Eur J Gastroenterol Hepatol 2003; 15:607–613. Manzel A, Muller DN, Hafler DA, Erdman SE, Linker RA, Kleinewietfeld M. Role of ‘Western diet’ in inflammatory autoimmune diseases. Curr Allergy Asthma Rep 2014; 14:404. Andersen V, Olsen A, Carbonnel F, Tjønneland A, Vogel U. Diet and risk of inflammatory bowel disease. Dig Liver Dis 2012; 44:185–194. Boeing H, Bechthold A, Bub A, Ellinger S, Haller D, Kroke A, et al. Critical review: vegetables and fruit in the prevention of chronic diseases. Eur J Nutr 2012; 51:637–663. Hansen TS, Jess T, Vind I, Elkjaer M, Nielsen MF, Gamborg M, Munkholm P. Environmental factors in inflammatory bowel disease: a case–control study based on a Danish inception cohort. J Crohns Colitis 2011; 5:577–584. Ng SC, Leong RW, Ko Y, Studd C, Niewiadomski O, Bell S, et al. Environmental risk factors in inflammatory bowel disease in asia-pacific: a population-based case–control study. Gastroenterology 2014; 146: S771–S772. Higgins JP, Thompson SG. Quantifying heterogeneity in a metaanalysis. Stat Med 2002; 21:1539–1558. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003; 327:557–560. Higgins JP, Thompson SG. Controlling the risk of spurious findings from meta-regression. Stat Med 2004; 23:1663–1682. Patsopoulos NA, Evangelou E, Ioannidis JP. Sensitivity of betweenstudy heterogeneity in meta-analysis: proposed metrics and empirical evaluation. Int J Epidemiol 2008; 37:1148–1157. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315:629–634. Persson PG, Ahlbom A, Hellers G. Diet and inflammatory bowel disease: a case–control study. Epidemiology 1992; 3:47–52. Reif S, Klein I, Lubin F, Farbstein M, Hallak A, Gilat T. Pre-illness dietary factors in inflammatory bowel disease. Gut 1997; 40:754–760. Sakamoto N, Kono S, Wakai K, Fukuda Y, Satomi M, Shimoyama T, et al. Dietary risk factors for inflammatory bowel disease: a multicenter case–control study in Japan. Inflamm Bowel Dis 2005; 11:154–163. Halfvarson J, Jess T, Magnuson A, Montgomery SM, Orholm M, Tysk C, et al. Environmental factors in inflammatory bowel disease: a co-twin control study of a Swedish–Danish twin population. Inflamm Bowel Dis 2006; 12:925–933. Amre DK, D’Souza S, Morgan K, Seidman G, Lambrette P, Grimard G, et al. Imbalances in dietary consumption of fatty acids, vegetables, and fruits are associated with risk for Crohn’s disease in children. Am J Gastroenterol 2007; 102:2016–2025. Abubakar I, Myhill DJ, Hart AR, Lake IR, Harvey I, Rhodes JM, et al. A case–control study of drinking water and dairy products in Crohn’s Disease – further investigation of the possible role of Mycobacterium avium paratuberculosis. Am J Epidemiol 2007; 165:776–783. Maconi G, Ardizzone S, Cucino C, Bezzio C, Russo AG, Bianchi Porro G. Pre-illness changes in dietary habits and diet as a risk factor for inflammatory bowel disease: a case–control study. World J Gastroenterol 2010; 16:4297–4304. Jakobsen C, Paerregaard A, Munkholm P, Wewer V. Environmental factors and risk of developing paediatric inflammatory bowel disease – a population based study 2007–2009. J Crohns Colitis 2013; 7:79–88.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
630
June 2015 • Volume 27 • Number 6
European Journal of Gastroenterology & Hepatology
34 Russel MG, Engels LG, Muris JW, Limonard CB, Volovics A, Brummer RJM, et al. ‘Modern life’ in the epidemiology of inflammatory bowel disease: a case–control study with special emphasis on nutritional factors. Eur J Gastroenterol Hepatol 1998; 10:243–249. 35 Shi XH, Guo ZR, Zheng JJ, Chen FM, Wang Z. Research on diet in Crohn’s diseases. Chin J Gastroenterol Hepatol 2007; 16:434–436. 36 Chen YD, Wang Y, Dang T, Zhang J, Wang Xl, Guo L. Logistic regression analysis of risk factors for ulcerative colitis. Chin Prev Med 2004; 5:199–201. 37 Zhou MH, Liu XD. A case–control study on the risk factors of ulcerative colitis. Changchun: Jilin University; 2012. 38 Lührs H, Gerke T, Müller JG, Melcher R, Schauber J, Boxberge F, et al. Butyrate inhibits NF-kappaB activation in lamina propria macrophages
39
40 41 42
of patients with ulcerative colitis. Scand J Gastroenterol 2002; 37:458–466. Roberts CL, Keita AV, Duncan SH, O’Kennedy N, Soderholm JD, Rhodes JM, et al. Translocation of Crohn’s disease Escherichia coli across M-cells: contrasting effects of soluble plant fibres and emulsifiers. Gut 2010; 59:1331–1339. Mora JR, von Andrian UH. Differentiation and homing of IgAsecreting cells. Mucosal Immunol 2008; 1:96–109. Suzuki T, Hara H. Role of flavonoids in intestinal tight junction regulation. J Nutr Biochem 2011; 22:401–408. Meddings J. The significance of the gut barrier in disease. Gut 2008; 57:438–440.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
Department of Epidemiology and Health Statistics, the Medical College of Qingdao University, Qingdao, Shandong, People’s Republic of China
of diet in the development of IBD is unclear [15,16]. Many foods or food components have been studied as potential etiological factors for IBD. A recent systematic review reported a protective effect of ω-3 polyunsaturated acids on UC, whereas animal protein was found to be associated with an increased risk of IBD [17]. Vegetables and fruit, rich in multiple beneficial nutrients, including monounsaturated and polyunsaturated fatty acids, n-3 fatty acids, fiber, vitamins, minerals, and phytochemicals, have been hypothesized to protect against chronic diseases generally [18], such as cardiovascular disease and cancer. Several epidemiological studies have investigated associations between intake of vegetables and fruit and the risk of IBD, but the findings are inconsistent: one study carried out by Hansen et al. [19] reported that the consumption of both vegetables and fruit was associated significantly with a decreased risk of UC and CD, whereas another study carried out by Ng et al. [20] showed no significant association between them. Therefore, we carried out a comprehensive metaanalysis by combining the results from all available observational studies to (i) assess the risk of UC and CD for highest versus lowest consumption of vegetables and fruit separately and (ii) explore the potential betweenstudy heterogeneity and publication bias.
Correspondence to Dongfeng Zhang, MS, Department of Epidemiology and Health Statistics, the Medical College of Qingdao University, 38 Dengzhou Road, Qingdao, Shandong 266021, People’s Republic of China
Materials and methods
Introduction
Inflammatory bowel disease (IBD), including ulcerative colitis (UC) and Crohn’s disease (CD), has become a global disease [1]. The incidence of IBD increased steadily from the 19th century and remained stable until the middle and latter part of the 20th century in North Europe and America [2,3]. In recent years, the incidence of IBD has been increasing in most developing countries, such as China, Japan, India, Iran, Thailand, etc. [4–6]. This geographical variation of IBD has been associated with a westernized lifestyle and environmental factors [1,7]. Smoking and appendectomy have been identified as risk factors for IBD [8–10]. In addition, urbanization of societies, antibiotic use, use of oral contraceptives, and changes in diet are also potential risk factors for IBD [11–14]. The association between diet and the risk of IBD was proposed as early as 50 years ago; however, the exact role European Journal of Gastroenterology & Hepatology 2015, 27:623–630 Keywords: Crohn’s disease, fruit, inflammatory bowel disease, ulcerative colitis, vegetables
Tel: + 86 532 8299 1712; fax: + 86 532 8380 1449; e-mails: [email protected], [email protected] Received 4 November 2014 Accepted 3 February 2015 Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's website (www.eurojgh.com).
Search strategy
To identify all relevant studies, we performed a computerbased search up to 15 September 2014 from the following databases: (i) PubMed, (ii) Embase, (iii) Web of Science (ISI), and (iv) the China National Knowledge Infrastructure
0954-691X Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
DOI: 10.1097/MEG.0000000000000330
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
623
624
June 2015 • Volume 27 • Number 6
European Journal of Gastroenterology & Hepatology
4796 articles from PubMed 7378 articles from Embase 6371 articles from ISI 2851 articles from CNKI
14 513 articles screened after excluding duplicates
14 410 articles excluded after review of title or abstract
103 full-text articles reviewed
89 articles excluded because: 59 did not evaluate this association of interest 11 review 7 lacking OR (RR) or corresponding 95% CI 6 not English or Chinese 2 vegetables and fruit combined 4 duplicate
14 articles included in the meta-analysis Fig. 1. Flowchart of the selection of studies included in the meta-analysis. CI, confidence internal; CNKI, China National Knowledge Infrastructure; ISI, Web of Science; OR, odds ratio; RR, relative risk.
(CNKI) using the key words or Mesh terms ‘fruit’, ‘vegetables’, ‘lifestyle’, ‘diet’, ‘nutrition’ in combination with ‘inflammatory bowel disease’, ‘ulcerative colitis’, and ‘Crohn’s disease’. Moreover, we reviewed the reference lists of all retrieved articles to search for undetected relevant studies. The flow diagram of the literature search is shown in Fig. 1. Inclusion criteria
All the studies identified were reviewed independently by two investigators and studies were included if they fulfilled the following criteria: (i) observational studies published originally; (ii) the topic of interest was consumption of vegetables and/or fruit; (iii) the outcome was UC and/or CD; (iv) the odds ratios (ORs) or relative risks with corresponding 95% confidence intervals (CIs) were reported or could be calculated from the data presented in articles; and (v) studies were reported in English or Chinese. If data were duplicated in more than one study, the most recent study was included in this meta-analysis. Data extraction
Data were extracted independently by two investigators who reached a consensus on all of the items. Information extracted from each study was as follows: name of the first author, publication year, country in which the study was
carried out, sex and age of the participants, number of cases, the ORs with corresponding 95% CIs for the highest versus lowest consumption, variables adjusted in the analysis, and control resource. Statistical analysis
We weighted the study-specific log ORs by the inverse of their variance to calculate pooled ORs with corresponding 95% CIs of the associations between consumption of vegetables and fruit and the risk of UC and CD. I2 of Higgins and Thompson [21] was used to assess heterogeneity among studies and I2 values of 0, 25, 50, and 75% represent no, low, moderate, and high heterogeneity, respectively. The DerSimonian and Laird random-effects model (REM) was selected as the pooling method if considerable heterogeneity was present (I2 > 50%) [22]; otherwise, the fixed-effects model (FEM) was used. Metaregression with restricted maximum likelihood estimation was used to explore the potential important covariates that might exert a major impact on between-study heterogeneity [23]. Subgroup analysis was carried out by the continent (Asia and Europe) and the status (yes or no) of adjusting for smoking. The leave-one-out sensitivity analysis was carried out to evaluate the key studies that had a major impact on between-study heterogeneity [24].
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
CD CD
42/51
< 20
The Netherlands
China
Japan
Sweden Denmark
China
England
Canada
Italy
Denmark
Chen et al. [36]
Sakamoto et al. [28]
Halfvarson et al. [29]
Shi et al. [35]
Abubakar et al. [31] Amre et al. [30]
Maconi et al. [32]
Hansen et al. [19]
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
38.5/40.5
37.5/40.4
33.10/32.90
5–79 11–79
15–34
22–75
39/37
UC CD
UC CD
CD
UC CD
UC CD
UC
UC CD
UC CD
UC
Disease type
Russel et al. [34]
30.14/29.45 29.12/29.45
15–79
Age range/mean age (case/control)
Israel
Sweden
Country
Persson et al. [26] Reif et al. [27]
References
144 123
26 25
130
218
51
125 102
108 126
96
398 290
54 33
145
Cases
Daily vs. rarer intake
High vs. low
Quartile 4 vs. quartile 1
High vs. low
Daily vs. less frequent
Daily vs. less frequent
Quartile 4 vs. quartile 1
High vs. low
> 5 per week vs. none or 1 per week
High vs. low
Daily vs. less frequent
Vegetable or fruit consumption 0.7 (0.4–1) for vegetable and UC 1.28 (0.55–2.95) for vegetable and UC 0.65 (0.25–1.65) for fruit and UC 0.32 (0.09–1.12) for vegetable and CD 0.42 (0.13–1.38) for fruit and CD 0.5 (0.4–0.0.8) for fruit and UC 0.5 (0.3–0.7) for fruit and CD 0.42 (0.19–0.93) for vegetable and UC 0.75 (0.35–1.62) for vegetable and UC 0.62 (0.29–1.32) for fruit and UC 1.55 (0.76–3.17) for vegetable and CD 0.8 (0.38–1.66) for fruit and CD 2.9 (0.9–9.4) for fruit and UC 0.2 (0.1–0.9) for fruit and CD 2.34 (0.9–6.09) for vegetable and CD 0.39 (0.19–0.74) for fruit and CD 0.78 (0.7–0.88) for fruit and CD 0.69 (0.33–1.44) for vegetable and CD 0.37 (0.16–0.86) for fruit and CD 0.37 (0.11–1.21) for vegetable and UC 0.41 (0.13–1.35) for fruit and UC 0.21 (0.05–0.78) for vegetable and CD 0.76 (0.26–2.21) for fruit and CD 0.51 (0.31–0.84) for vegetable and UC 0.56 (0.33–0.95) for fruit and UC 0.41 (0.24–0.71) for vegetable and CD 0.39 (0.22–0.7) for fruit and CD
OR (95% CI)
Smoking status, appendectomy, tonsillectomy, usage of oral contraceptives, consumption of fiber, sugar, coffee, and eggs
Age, sex, years of education, smoking, BMI
Age, sex, and BMI, total energy intake
Age, sex, study center
Not provided
Not provided
Age, sex, study area, education, and smoking habit
Not provided
Smoking, sex, age, educational level, and selected nutritional factors
Hospital based
Hospital based
Population and hospital based
Population based
Hospital based
Population based
Hospital based
Population based
Population based
Population and hospital based
Energy intake
Control resource Population based
Adjustment for covariates Age and sex
Table 1. Characteristics of the studies included on consumption of vegetables and fruit and the risk of ulcerative colitis and Crohn’s disease
Vegetables and fruit and IBD Li et al. www.eurojgh.com 625
Population based
Population based
Age, sex, ethnicity, area of residence and socioeconomic status
Sex, age and country income based on GNI
Population based
The influence analysis was carried out with the exclusion of one study at a time to assess whether the results could have been affected markedly by a single study. Publication bias was evaluated by visual inspection of the funnel plot and Egger’s test [25]. All statistical analyses were carried out using STATA Version 10 (StataCorp, College Station, Texas, USA). All reported P values were two-sided, with P value less than 0.05 considered statistically significant. Results Search results and characteristics of the included studies
The search strategy identified 4796 articles from PubMed, 6371 articles from ISI, 7378 articles from Embase, and 2851 articles from CNKI. After reviewing the title/abstract, 103 articles were retrieved. Then, 89 articles were subsequently excluded from the meta-analysis for various reasons. As a result, 14 articles fulfilled the inclusion criteria and were included in the meta-analysis [19,20,26–37]. Of these 14 articles, seven were from Europe [19,26,29,30,32–34], one was from America [31], and six were from Asia [20,27,28, 35–37] (the study carried out by Ng et al. [20] investigated eight Asia countries and Australia, respectively). All the included studies were case–control studies. The baseline characteristics of the studies are shown in Table 1. Quantitative synthesis Vegetables and ulcerative colitis
CD, Crohn’s disease; CI, confidence interval; GNI, gross national income; OR, odds ratio; UC, ulcerative colitis.
Daily vs. weekly or less frequent UC CD Nine countries Ng et al. [20]
38/39
256 186
Daily or > once weekly vs. 1 time weekly or less 56 59 UC CD 3–15 Denmark Jakobsen et al. [33]
UC China Zhou and Liu [37]
20–50
11
High vs. low ≥ 7 times vs. ≤ 2 times per week
1.72 (0.41–7.24) for vegetable and UC 0.57 (0.05–6.94) for fruit and UC 0.3 (0.1–0.8) for vegetable and UC 0.3 (0.1–1) for vegetable and CD 1.03 (0.68–1.54) for vegetable and UC 0.92 (0.71–1.28) for fruit and UC 0.81 (0.51–1.29) for vegetable and CD 0.92 (0.64–1.31) for fruit and CD
Unadjusted
Adjustment for covariates OR (95% CI) Vegetable or fruit consumption Cases Disease type Age range/mean age (case/control) Country References
Table 1. (Continued)
June 2015 • Volume 27 • Number 6
European Journal of Gastroenterology & Hepatology
Control resource
626
Nine studies [19,20,26–28,32,33,36,37] including 896 cases investigated the association between intake of vegetables and the risk of UC. Three of the nine studies showed a significant association, whereas the other six studies indicated no relation. The pooled OR of UC for the highest versus the lowest consumption of vegetables was 0.71 (95% CI 0.58–0.88, I2 = 41.6%, FEM, Pheterogeneity = 0.090; Fig. 2). Subgroup analysis by continent (Asia, Europe) and status of adjusting for smoking was carried out, and the results showed that higher intake of vegetables was associated inversely with the risk of UC for studies carried out in Europe (OR = 0.55, 95% CI 0.40–0.75, I2 = 0%, FEM), but not in Asia (OR = 0.89, 95% CI 0.66–1.21, I2 = 26.7%, FEM). The pooled ORs were 0.55 (95% CI 0.37–0.81, I2 = 0%, FEM) among studies that adjusted for smoking and 0.86 (95% CI 0.66–1.13, I2 = 44.6%, FEM) among studies that did not adjust for smoking (Table 2). Fruit and ulcerative colitis
Eight studies [19,20,27–29,32,34,37] including 1122 cases investigated the association between intake of fruit and the risk of UC. Two of the eight studies showed a significant association, whereas the other six studies indicated no relation. The pooled OR of UC for the highest versus the lowest consumption of fruit was 0.69 (95% CI 0.49–0.96, I2 = 50.7%, REM, Pheterogeneity = 0.048; Supplementary Fig. 1, Supplemental digital content 1, http://links.lww.com/ EJGH/A13). The results from subgroup analysis showed that the pooled ORs were 0.66 (95% CI 0.37–1.16, I2 = 63.6%, REM) for studies carried out in Europe and 0.82 (95% CI 0.63–1.07, I2 = 0%, FEM) in Asia. The pooled ORs were 0.52 (95% CI 0.40–0.68, I2 = 0%, FEM)
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
Vegetables and fruit and IBD Li et al.
www.eurojgh.com
References
OR (95% CI)
Weight (%)
Persson et al. [26]
0.70 (0.40–1.00)
22.07
Reif et al. [27]
1.28 (0.55–2.95)
6.57
Chen et al. [36]
0.42 (0.19–0.93)
7.35
Sakamoto et al. [28]
0.75 (0.35–1.62)
7.89
Maconi et al. [32]
0.37 (0.11–1.21)
3.22
Hansen et al. [19]
0.51 (0.31–0.84)
18.65
Zhou and Liu [37]
1.72 (0.41–7.24)
2.25
Jakobsen et al. [33]
0.30 (0.10–0.80)
4.28
Ng et al. [20]
1.03 (0.68–1.54)
27.73
Overall (I = 41.6%, P = 0.090)
0.71 (0.58–0.88)
100.00
2
0.1
0.2
0.5
1
2
5
627
10
Fig. 2. Meta-analysis of the association between intake of vegetables and the risk of ulcerative colitis. The size of the gray box is positively proportional to the weight assigned to each study, which is inversely proportional to the standard error of the odds ratio (OR), and horizontal lines represent the 95% confidence interval (CI).
Table 2. Subgroup analyses of consumption of vegetables and fruit and the risk of ulcerative colitis and Crohn’s disease Subgroup Vegetables and UC
Fruit and UC
Vegetables and CD
Fruit and CD
Continent Asia Europe Adjustment Yes No Continent Asia Europe Adjustment Yes No Continent Asia Europe Adjustment Yes No Continent Asia Europe Adjustment Yes No
Number of studies
OR (95% CI)
Effect model
I2 (%)
P value
Article included
5 4
0.89 (0.66–1.21) 0.55 (0.40–0.75)
FEM FEM
26.7 0
0.243 0.411
[20,27,28,36,37] [19,26,32,33]
3 5
0.55 (0.37–0.81) 0.86 (0.66–1.13)
FEM FEM
0 44.6
0.567 0.125
[19,28,32] [20,26,27,33,37]
4 4
0.82 (0.63–1.07) 0.66 (0.37–1.16)
FEM REM
0 63.6
0.788 0.041
[20,27,28,37] [19,29,32,34]
4 3
0.52 (0.40–0.68) 0.89 (0.67–1.17)
FEM FEM
0 0
0.921 0.741
[19,28,32,34] [20,27,37]
4 3
1.00 (0.50–2.03) 0.36 (0.23–0.57)
REM FEM
68.6 0
0.023 0.635
[20,27,28,35] [19,32,33]
3 4
0.56 (0.19–1.66) 0.66 (0.46–0.94)
REM FEM
81.7 21.7
0.004 0.280
[19,28,32] [20,27,31,33]
4 5
0.70 (0.52–0.94) 0.52 (0.34–0.80)
FEM REM
37.5 71.8
0.187 0.007
[20,27,28,35] [19,29,30,32,34]
4 4
0.52 (0.39–0.70) 0.78 (0.70–0.87)
FEM FEM
0 38.7
0.427 0.180
[19,32,34,35] [20,27,30,31]
for smoking
for smoking
for smoking
for smoking
CI, confidence interval; CD, Crohn’s disease; FEM, fixed-effects model; OR, odds ratio; REM, DerSimonian and Laird random-effects model; UC, ulcerative colitis.
among studies that adjusted for smoking and 0.89 (95% CI 0.67–1.17, I2 = 0%, FEM) among studies that did not adjust for smoking (Table 2). One study [20] was found to contribute toward between-study heterogeneity by the leave-one-out sensitivity analysis. After further excluding the study, low heterogeneity (I2 = 28.3%) was found and the pooled OR was 0.61 (95% CI 0.44–0.86). Vegetables and Crohn’s disease
Eight studies [19,20,27,28,30,32,33,35] including 733 cases investigated the association between intake of vegetables and the risk of CD. Two of the eight studies showed a significant
association, whereas the other six studies indicated no relation. The pooled OR of CD for the highest versus the lowest consumption of vegetables was 0.66 (95% CI 0.40–1.09, I2 = 67.5%, REM, Pheterogeneity = 0.003; Supplementary Fig. 2, Supplemental digital content 2, http://links.lww.com/ EJGH/A14). The results from subgroup analysis showed that higher intake of vegetables was associated inversely with the risk of CD for studies carried out in Europe (OR = 0.36, 95% CI 0.23–0.57, I2 = 0%, FEM), but not in Asia (OR = 1.00, 95% CI 0.50–2.03, I2 = 68.6%, REM). The pooled ORs were 0.56 (95% CI 0.19–1.66, I2 = 81.7%, REM) among studies that adjusted for smoking and 0.66 (95% CI 0.46–0.94, I2 = 21.7%, FEM) among studies did not adjust for smoking (Table 2).
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
June 2015 • Volume 27 • Number 6
European Journal of Gastroenterology & Hepatology
Two studies [28,35] were found to contribute toward between-study heterogeneity by the leave-one-out sensitivity analysis. After further excluding these two studies, low heterogeneity (I2 = 36.1%) was found and the pooled OR was 0.5 (95% CI 0.33–0.75). Fruit and Crohn’s disease
Ten studies [19,20,27–32,34,35] including 1284 cases investigated the association between intake of fruit and the risk of CD. Six of the 10 studies showed a significant association, whereas the other four studies indicated no relation. The pooled OR of CD for the highest versus the lowest consumption of fruit was 0.57 (95% CI 0.44–0.74, I2 = 59.5%, REM, Pheterogeneity = 0.008; Supplementary Fig. 3, Supplemental digital content 3, http://links.lww. com/EJGH/A15). The results from subgroup analysis showed that the pooled ORs were 0.52 (95% CI 0.34–0.80, I2 = 71.8%, REM) for studies carried out in Europe and 0.70 (95% CI 0.52–0.94, I2 = 37.5%, FEM) in Asia. The pooled ORs were 0.52 (95% CI 0.39–0.70, I2 = 0%, FEM) among studies that adjusted for smoking and 0.78 (95% CI 0.70–0.87, I2 = 38.7%, FEM) among studies that did not adjust for smoking (Table 2). Two studies [20,31] were found to contribute toward between-study heterogeneity by the leave-one-out sensitivity analysis. After further excluding these two studies, the heterogeneity (I2 = 0%) was decreased and the pooled OR was 0.46 (95% CI 0.36–0.59). Meta-regression
To explore the moderate to high between-study heterogeneity in these above-mentioned analyses, meta-regression with the covariates of year, number of cases, control resource, continent, and status of adjusting for smoking was carried out. The status of adjusting for smoking was found to contribute toward the heterogeneity in the analysis of intake of fruit and the risk of UC (P = 0.043). No significant findings were found for other analyses. Influence analysis and publication bias
Influence analysis showed that no individual study had an excessive influence on the above-mentioned pooled ORs. The funnel plot and Egger’s test showed no evidence of significant publication bias for the analysis of consumption of vegetables and the risk of UC (P = 0.558; Fig. 3), consumption of fruit and the risk of UC (P = 0.980; Supplementary Fig. 4, Supplemental digital content 4, http:// links.lww.com/EJGH/A16), and consumption of vegetables and the risk of CD (P = 0.601; Supplementary Fig. 5, Supplemental digital content 5, http://links.lww.com/EJGH/ A17). For consumption of fruit and the risk of CD, publication bias was found (P = 0.024; Supplementary Fig. 6, Supplemental digital content 6, http://links.lww.com/EJGH/ A18); however, after excluding two studies that had a strong effect on heterogeneity, no significant publication bias was found (P = 0.677; Supplementary Fig. 7, Supplemental digital content 7, http://links.lww.com/EJGH/A19) and the result remained significant.
0
0.2 SE of log OR
628
0.4
0.6
0.8 1 OR (log scale)
2
3
Fig. 3. Funnel plot with pseudo 95% confidence limits for the analysis of intake of vegetable and the risk of ulcerative colitis. OR, odds ratio.
Discussion
This meta-analysis assessed associations between intake of vegetables and fruit and the risk of UC and CD, respectively. Findings showed inverse associations between intake of vegetables and the risk of UC, intake of fruit and the risk of UC, and intake of fruit and the risk of CD overall. However, the inverse association between intake of vegetables and the risk of CD was only observed for studies carried out in Europe. Some potential mechanisms may explain the inverse associations in this meta-analysis. Vegetables and fruit are abundant in fiber, micronutrients (such as vitamins C and E and folate), and phytochemicals (such as carotenoids, phenolic, isoflavones, and indoles). First, butyrate, the major anion produced by the bacterial fermentation of dietary fiber in the colon, can reduce mucosal inflammation by reducing nuclear factor-κB activity in colon cells [38]. Second, plant fiber may influence the translocation of microbes across gut mucosa [39]. Third, vitamin A has been found to be necessary for the expression of surface markers (such as α4β7 and CCR9), which can correct homing of the cells to the gut [40]. Finally, flavonoids seem to be involved in maintenance of the intercellular tight junctions, which is one of the major determinants of the intestinal barrier function [41], and impairment in intestinal barrier function has been associated with IBD [42]. In our meta-analysis, moderate heterogeneity was found in the analysis of intake of fruit and the risk of UC, intake of vegetables and the risk of CD, and intake of fruit and the risk of CD. Therefore, the leave-one-out sensitivity analysis was used to reduce the heterogeneity by excluding studies that were found to contribute toward the between-study heterogeneity. The results remained significant after reducing the heterogeneity in the analysis of intake of fruit and the risk of UC, and intake of fruit and the risk of CD. To further explore the potential sources of between-study heterogeneity, meta-regression with covariates of year, number of cases, control resource, continent, and status of adjusting for smoking was carried out. Findings showed that status of adjusting for smoking contributed toward the between-study heterogeneity (P = 0.043) in the analysis of consumption of fruit and the risk of UC. The heterogeneity decreased to 0% (Pheterogeneity = 0.741) among studies that
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
Vegetables and fruit and IBD Li et al.
adjusted for smoking and 0% (Pheterogeneity =0.921) among studies that did not adjust for smoking. There are some advantages in our meta-analysis. First, to our knowledge, this is the first meta-analysis to explore associations between intake of vegetables and fruit and the risk of UC and CD, and the associations observed remained significant in the leave-one-out sensitivity analysis, indicating that our results were stable and reliable. Second, a relatively larger number of participants included, with a reduction in sampling error to a huge extent, enabled a much greater possibility of reaching reasonable conclusions. Third, no publication bias was observed for the analysis of consumption of vegetables and the risk of UC, consumption of fruit and the risk of UC, and consumption of vegetables and the risk of CD. However, several limitations should be acknowledged in our meta-analysis. First, the findings based on case–control studies were prone to recall bias and interviewer bias. Second, confounders adjusted in studies were different, which could affect associations between intake of vegetables and fruit and the risk of IBD. For example, some studies adjusted for smoking, whereas others did not. Third, the diet assessment methods and the retrospective time were different in the original studies, which led to incomparability in the results to some extent. Some studies used food frequency questionnaires and others used dietary history questionnaires. Fourth, the results of subgroup analysis were based on a limited number of studies. In summary, the results from this meta-analysis suggest that consumption of fruit reduces the risk of both UC and CD, and consumption of vegetables reduces the risk of UC. These findings need to be confirmed further by cohort studies and randomized controlled trials.
www.eurojgh.com
10 11
12
13
14
15
16
17 18
19
20
21 22 23
Acknowledgements
24
Conflicts of interest
There are no conflicts of interest.
25 26
References 1
2 3 4
5 6
7
8
9
Ng SC, Bernstein CN, Vatn MH, Lakatos PL, Loftus EV Jr, Tysk C, et al. Epidemiology and Natural History Task Force of the International Organization of Inflammatory Bowel Disease (IOIBD). Geographical variability and environmental risk factors in inflammatory bowel disease. Gut 2013; 62:630–649. M’Koma AE. Inflammatory bowel disease: an expanding global health problem. Clin Med Insights Gastroenterol 2013; 6:33–47. Burisch J, Jess T, Martinato M, Lakatos PL. The burden of inflammatory bowel disease in Europe. J Crohns Colitis 2013; 7:322–337. Leong RW, Lau JY, Sung JJ. The epidemiology and phenotype of Crohn’s disease in the Chinese population. Inflamm Bowel Dis 2004; 10:646–651. Makharia GK. Rising incidence and prevalence of Crohn’s disease in Asia: is it apparent or real? J Gastroenterol Hepatol 2006; 21:929–931. Thia KT, Loftus EV Jr, Sandborn WJ, Yang SK. An update on the epidemiology of inflammatory bowel disease in Asia. Am J Gastroenterol 2008; 103:3167–3182. Karlinger K, Györke T, Makö E, Mester A, Tarján Z. The epidemiology and the pathogenesis of inflammatory bowel disease. Eur J Radiol 2000; 35:154–167. Kaplan GG, Jackson T, Sands BE, Frisch M, Andersson RE, Korzenik J. The risk of developing Crohn’s disease after an appendectomy: a metaanalysis. Am J Gastroenterol 2008; 103:2925–2931. Lakatos PL, Vegh Z, Lovasz BD, David G, Pandur T, Erdelyi Z, et al. Is current smoking still an important environmental factor in inflammatory
27 28
29
30
31
32
33
629
bowel diseases? Results from a population-based incident cohort. Inflamm Bowel Dis 2013; 19:1010–1017. Danese S, Fiocchi C. Etiopathogenesis of inflammatory bowel diseases. World J Gastroenterol 2006; 12:4807–4812. Spehlmann ME, Begun AZ, Saroglou E, Hinrichs F, Tiemann U, Raedler A, Schreiber S. Risk factors in German twins with inflammatory bowel disease: results of a questionnaire-based survey. J Crohns Colitis 2012; 6:29–42. Loftus EV Jr. Clinical epidemiology of inflammatory bowel disease: incidence, prevalence, and environmental influences. Gastroenterology 2004; 126:1504–1517. Tragnone A, Valpiani D, Miglio F, Elmi G, Bazzocchi G, Pipitone E, Lanfranchi GA. Dietary habits as risk factors for inflammatory bowel disease. Eur J Gastroenterol Hepatol 1995; 7:47–51. Castiglione F, Diaferia M, Morace F, Labianca O, Meucci C, Cuomo A, et al. Risk factors for inflammatory bowel diseases according to the ‘hygiene hypothesis’: a case–control, multi-centre, prospective study in Southern Italy. J Crohns Colitis 2012; 6:324–329. Cashman KD, Shanahan F. Is nutrition an aetiological factor for inflammatory bowel disease? Eur J Gastroenterol Hepatol 2003; 15:607–613. Manzel A, Muller DN, Hafler DA, Erdman SE, Linker RA, Kleinewietfeld M. Role of ‘Western diet’ in inflammatory autoimmune diseases. Curr Allergy Asthma Rep 2014; 14:404. Andersen V, Olsen A, Carbonnel F, Tjønneland A, Vogel U. Diet and risk of inflammatory bowel disease. Dig Liver Dis 2012; 44:185–194. Boeing H, Bechthold A, Bub A, Ellinger S, Haller D, Kroke A, et al. Critical review: vegetables and fruit in the prevention of chronic diseases. Eur J Nutr 2012; 51:637–663. Hansen TS, Jess T, Vind I, Elkjaer M, Nielsen MF, Gamborg M, Munkholm P. Environmental factors in inflammatory bowel disease: a case–control study based on a Danish inception cohort. J Crohns Colitis 2011; 5:577–584. Ng SC, Leong RW, Ko Y, Studd C, Niewiadomski O, Bell S, et al. Environmental risk factors in inflammatory bowel disease in asia-pacific: a population-based case–control study. Gastroenterology 2014; 146: S771–S772. Higgins JP, Thompson SG. Quantifying heterogeneity in a metaanalysis. Stat Med 2002; 21:1539–1558. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003; 327:557–560. Higgins JP, Thompson SG. Controlling the risk of spurious findings from meta-regression. Stat Med 2004; 23:1663–1682. Patsopoulos NA, Evangelou E, Ioannidis JP. Sensitivity of betweenstudy heterogeneity in meta-analysis: proposed metrics and empirical evaluation. Int J Epidemiol 2008; 37:1148–1157. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315:629–634. Persson PG, Ahlbom A, Hellers G. Diet and inflammatory bowel disease: a case–control study. Epidemiology 1992; 3:47–52. Reif S, Klein I, Lubin F, Farbstein M, Hallak A, Gilat T. Pre-illness dietary factors in inflammatory bowel disease. Gut 1997; 40:754–760. Sakamoto N, Kono S, Wakai K, Fukuda Y, Satomi M, Shimoyama T, et al. Dietary risk factors for inflammatory bowel disease: a multicenter case–control study in Japan. Inflamm Bowel Dis 2005; 11:154–163. Halfvarson J, Jess T, Magnuson A, Montgomery SM, Orholm M, Tysk C, et al. Environmental factors in inflammatory bowel disease: a co-twin control study of a Swedish–Danish twin population. Inflamm Bowel Dis 2006; 12:925–933. Amre DK, D’Souza S, Morgan K, Seidman G, Lambrette P, Grimard G, et al. Imbalances in dietary consumption of fatty acids, vegetables, and fruits are associated with risk for Crohn’s disease in children. Am J Gastroenterol 2007; 102:2016–2025. Abubakar I, Myhill DJ, Hart AR, Lake IR, Harvey I, Rhodes JM, et al. A case–control study of drinking water and dairy products in Crohn’s Disease – further investigation of the possible role of Mycobacterium avium paratuberculosis. Am J Epidemiol 2007; 165:776–783. Maconi G, Ardizzone S, Cucino C, Bezzio C, Russo AG, Bianchi Porro G. Pre-illness changes in dietary habits and diet as a risk factor for inflammatory bowel disease: a case–control study. World J Gastroenterol 2010; 16:4297–4304. Jakobsen C, Paerregaard A, Munkholm P, Wewer V. Environmental factors and risk of developing paediatric inflammatory bowel disease – a population based study 2007–2009. J Crohns Colitis 2013; 7:79–88.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
630
June 2015 • Volume 27 • Number 6
European Journal of Gastroenterology & Hepatology
34 Russel MG, Engels LG, Muris JW, Limonard CB, Volovics A, Brummer RJM, et al. ‘Modern life’ in the epidemiology of inflammatory bowel disease: a case–control study with special emphasis on nutritional factors. Eur J Gastroenterol Hepatol 1998; 10:243–249. 35 Shi XH, Guo ZR, Zheng JJ, Chen FM, Wang Z. Research on diet in Crohn’s diseases. Chin J Gastroenterol Hepatol 2007; 16:434–436. 36 Chen YD, Wang Y, Dang T, Zhang J, Wang Xl, Guo L. Logistic regression analysis of risk factors for ulcerative colitis. Chin Prev Med 2004; 5:199–201. 37 Zhou MH, Liu XD. A case–control study on the risk factors of ulcerative colitis. Changchun: Jilin University; 2012. 38 Lührs H, Gerke T, Müller JG, Melcher R, Schauber J, Boxberge F, et al. Butyrate inhibits NF-kappaB activation in lamina propria macrophages
39
40 41 42
of patients with ulcerative colitis. Scand J Gastroenterol 2002; 37:458–466. Roberts CL, Keita AV, Duncan SH, O’Kennedy N, Soderholm JD, Rhodes JM, et al. Translocation of Crohn’s disease Escherichia coli across M-cells: contrasting effects of soluble plant fibres and emulsifiers. Gut 2010; 59:1331–1339. Mora JR, von Andrian UH. Differentiation and homing of IgAsecreting cells. Mucosal Immunol 2008; 1:96–109. Suzuki T, Hara H. Role of flavonoids in intestinal tight junction regulation. J Nutr Biochem 2011; 22:401–408. Meddings J. The significance of the gut barrier in disease. Gut 2008; 57:438–440.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
