Associations of Fish and Fish Oil Consumption With Incident Inflammatory Bowel Disease: A Population-Based Prospective Cohort Study

Abstract

Background

We examined the associations of fish and fish oil consumption with inflammatory bowel disease (IBD) incidence.

Patients and Methods

We conducted a longitudinal analysis based on the UK Biobank, a population-based prospective cohort. Dietary consumption of fish and fish oil was collected by questionnaire. IBD incident cases were identified through links to National Health Services datasets. Cox proportional hazards regression models were used to assess the associations between oily fish, nonoily fish, and fish oil intake and IBD incidence with adjustment for various confounding factors.

Results

A total of 265 839 participants free of IBD at baseline were included, and 1554 incident IBD cases were identified during an average follow-up of 11.8 years. In fully adjusted models, we found that compared with participants who never ate oily fish, those having <1 serving/wk, 1 serving/wk, and >1 serving/wk had 9% (hazard ratio [HR], 0.91; 95% confidence interval [CI], 0.77-1.08), 19% (HR, 0.81; 95% CI, 0.69-0.96), and 12% (HR, 0.88; 95% CI, 0.73-1.06) lower risks of IBD, respectively, albeit not all statistically significant. A significant association was found between fish oil intake and a reduced risk of IBD (HR, 0.84; 95% CI, 0.75-0.93). We found no significant associations for nonoily fish. In a subsample (n = 105 714) of participants with multiple subsequent dietary reviews, we also found a negative association between the frequency of fish oil intake over time and incident IBD (P trend < .05).

Conclusions

Our findings indicate that oily fish and fish oil supplements might be protective factors against IBD.

Introduction

Inflammatory bowel disease (IBD) includes a group of immune-mediated, chronic, and incurable intestinal diseases that usually relapse, among which Crohn’s disease (CD) and ulcerative colitis (UC) are the 2 most common subtypes.¹ Currently, the pathogenesis of IBD has not been completely understood, but implicating factors include genetic susceptibility, gut microbiota, antibiotic use, lifestyle, environmental exposures, and immune dysregulation.² Among lifestyle factors in particular, diet is a widely acknowledged factor that is crucial to the development of IBD.³

Numerous studies have associated dietary intake of fish with a decreased risk of IBD, although controversial results also exist.⁴ In addition to the overall intake of fish, recent studies started to look into certain nutrients such as ω-3 fatty acids (FAs), the major components of fish oil. It has been hypothesized that ω-3 FAs have the potential to prevent or delay the development of IBD by promoting an anti-inflammatory process by decreasing the production of inflammatory cytokines, replacing arachidonic acid as an eicosanoid substrate, and inhibiting arachidonic acid metabolism, and changing the expression of inflammatory genes.⁵ Indeed, some animal experiments showed that ω-3 FAs could reduce chemically induced colon damage and inflammation.^2,6,7 However, a recent animal study suggested that dietary ω-3 FAs could trigger endoplasmic reticulum stress, fueling enteritis in CD.⁸

In the past decades, an increasing number of randomized controlled trials (RCTs) have analyzed the association between ω-3 FA intake and IBD, predominately the recovery or recurrence of IBD. However, the results are still controversial.⁹ By contrast, only a limited number of general population-based epidemiological studies investigated the associations between fish and fish oil intake and the risk of developing IBD (ie, incidence), and the results are also inconsistent.^10-15 Moreover, most of these studies are retrospective (eg, case-control studies) with a small study sample, and some of them did not take into account some important confounders such as cigarette smoking¹⁶ and body mass index (BMI),¹⁷ which are well-recognized risk factors for IBD.

Therefore, we investigated the associations between fish and fish oil consumption and IBD incidence based on a large population-based prospective cohort, in an effort to provide more robust evidence on the health effects of fish and fish oil on IBD development.

Methods

Study design and population

We conducted a longitudinal analysis using the UK Biobank project, a prospective cohort initially established to assess the main causes of various chronic diseases. The detailed study protocol has been published elsewhere.¹⁸ Briefly, the UK Biobank is an ongoing prospective study which recruited >0.5 million participants 40 to 69 years of age in 2006 to 2010. Extensive data for health-related factors were collected by questionnaires, physical measurements, and biomedical sample assays. The UK North West Multicenter Research Ethics Committee authorized the project, and each participant gave their written informed consent prior to participation.

Selection process of study population for the present study is shown in the flowchart (Supplementary Figure 1). Participants with missing dietary information including fish and fish oil consumption at baseline survey (n = 6965) were excluded. Participants with missing information on key covariates (n = 229 521) or with IBD at baseline (n = 138) were further excluded. At last, 265 839 participants were included in main analysis. Several follow-up questionnaires, including 24-hour dietary recalls, were also completed in 2009 to 2012 by a portion of the participants in addition to the baseline survey. A total of 105 714 participants who participated in ≥1 follow-up surveys were included in a subsample analysis to further evaluate the possible changes in dietary behaviors over time (Supplementary Figure 2).

Dietary assessment

At baseline, the consumption of oily fish and nonoily fish was assessed by a self-administered semi-quantitative food frequency questionnaire. Participants were asked, “How often do you eat oily fish (e.g., sardines, salmon, mackerel, herring)?” and “How often do you eat non-oil fish (e.g., cod, tinned tuna, haddock)?” and could choose 1 answer from 6 options (never, less than once a week, once a week, 2-4 times a week, 5-6 times a week, or once or more daily). Another question (“Do you regularly take any of the following?”) was used to collect the information about habitual intake of a group of supplements including fish oil, with the answers “yes” or “no.” In addition, potential dietary changes were collected by asking “Does your diet vary much from week to week?” On the UK Biobank’s official website, further information regarding the inquiries and potential replies is provided (https://biobank.ndph.ox.ac.uk/showcase/label.cgi?id=100052).

Among the subsample of 105 714 participants who finished the follow-up surveys, dietary assessments were completed using the Oxford WebQ, a Web-based 24-hour dietary assessment tool that inquires about a person’s consumption of as many as 206 different kinds of food. The Oxford WebQ had moderate to strong correlations with an interviewer-administered 24-hour dietary recalls in capturing similar food items and estimating intake of similar nutrients (Spearman’s correlation coefficients range from 0.5 to 0.9).¹⁹

Ascertainment of IBD

Information on the onset of IBD was derived from the health administrative datasets of the National Health Service. IBD that occurred during follow-up (ie, incident cases) was identified using self-reports, hospital admissions, and inpatient records (Hospital Episode Statistics for England, Scottish Morbidity Record data for Scotland, and Patient Episode Database for Wales). We identified IBD cases using the 10th International Classification of Diseases codes (UC: K50; CD: K51). Participants were followed from the time of recruitment till the occurrence of IBD for the first time, death, loss of follow-up, or February 6, 2021 (the latest date available in the UK Biobank), whichever came first.

Covariates

Demographic and socioeconomic factors and other potential confounding factors were collected at baseline survey, including age, sex (male and female), race/ethnicity (White or others), educational level (college or university degree, or lower), employment status (paid job, retired, or unpaid job), household annual income (<£18 000, £18 000-£31 000, £31 000-£52 000, £52 000-£100 000, ≥£100 000, or unknown), BMI (underweight [<18.5 kg/m²], normal [18.5-25.0 kg/m²], overweight [25.0-30.0 kg/m²], or obese [≥30.0 kg/m²]), smoking status (never, former, or current smoker), alcohol drinking (never, occasional, moderate, or heavy), physical activity (assessed by International Physical Activity Questionnaire and classified into 3 levels: low, moderate, or high), fruit and vegetable intake (low, moderate, or high), anti-inflammatory medication use (aspirin or nonsteroidal anti-inflammatory drugs: yes or no), and Townsend deprivation index (an indicator of socioeconomic status at the area-level [national census output areas]). A higher Townsend deprivation index indicates a greater socioeconomic deprivation of the area.].

Statistical analysis

Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) of incident IBD in relation to oily/nonoily fish intake (the group with lowest intake served as the reference) or fish oil supplementation. Four incremental models were established, including model 0, the unadjusted model; model 1, adjusted by age, sex, and race/ethnicity; model 2, additionally adjusted by education, income, and employment status; and model 3, further adjusted by BMI, lifestyle (smoking, drinking, physical activity, and fruit and vegetable intake), anti-inflammatory medication use and Townsend deprivation index. We tested the proportional hazards assumption by plotting Schoenfeld residuals, and observed no clear evidence of violation (Supplementary Figure 6). We also generated Kaplan-Meier curves to demonstrate the estimated cumulative hazard of IBD for participants with different levels of fish and fish oil consumption. In addition to overall IBD, we also analyzed CD cases and UC cases separately.

In the subsample of the 105 714 participants who completed ≥1 Web-based 24-hour dietary recall during follow-up surveys, we combined baseline information with follow-up information to classify the participants into 5 groups: constant nonusers (who consumed fish oil neither at baseline nor during the follow-up survey), occasional users (who consumed fish oil at baseline only or during the follow-up survey only), and constant users (who consumed fish oil both at baseline and during follow-up survey, including 1 [modestly constant users], 2 [moderately constant users], and 3 or more [highly constant users] reports of 24-hour recalls).²⁰ Risks of developing IBD for occasional and constant users were compared with constant nonusers using the aforementioned Cox models.

We conducted a variety of sensitivity analyses to test the robustness of the results, including (1) that participants who experienced an IBD event during the first 2 years of follow-up were excluded to account for potential reverse causality; (2) that participants who stated they had significantly changed their diet in the previous 5 years were excluded; and (3) that fish intake and fish oil supplement were added into same models for mutual adjustment; and (4) additionally, we made adjustment for the consumption of processed meat, poultry, beef, mutton, pork, cereal, and energy to further reduce the impact of diet, among a small group of participants (n = 37 495) with relevant data.

R 4.1.2 (R Foundation for Statistical Computing) was used for all statistical analyses. Statistical significance was defined as a 2-tailed P value <.05.

Results

Characteristics of the study population

The participants’ average age at baseline was 57.2 ± 8.0 years. A total of 44.7% were male, and the majority were White in race/ethnicity (95.4%). More than half participants (53.2%) never smoked and 29.6% of the participants reported use of anti-inflammatory medication. Overall, 35.2% of the participants stated they regularly took fish oil supplements. Compared with participants who did not develop IBD during follow-up, those who did reported lower levels of education (percentage of attending the university of college: 26.1% vs 33.1%; P < .001), income (percentage of having income >£18 000: 25.2% vs 20.8%; P < .001), and paid employment rate (49.5% vs 54.7%; P < .001), and were more likely to smoke (percentage of current smokers: 13.0% vs 10.0%; P < .001) (Table 1). The analytical cohort and the original UK Biobank cohort were generally comparable in terms of the distribution of demographic characteristics (eg, age, sex, and race/ethnicity) (Supplementary Table 1). During an average follow-up of 11.8 ± 1.7 years, 1554 incident IBD cases were identified (0.58% of the cohort), including 501 CD cases and 1053 UC cases. The mean age of onset was 57.8 years for both conditions.

Oily fish and nonoily fish consumption and IBD and its subtypes

As shown in Table 2 and Supplementary Figure 2, a negative association between oily fish consumption and the risk of IBD was observed. After adjustment for the covariates, the trend became borderline significant (P = .06). In fully adjusted models, we found that compared with individuals who never consumed oily fish, those having <1 serving/wk, 1 serving/wk, and >1 serving/wk had 9% (HR, 0.91; 95% CI, 0.77-1.08), 19% (HR, 0.81; 95% CI, 0.69-0.96), and 12% (HR, 0.88; 95% CI, 0.73-1.06) lower risks of IBD, respectively. By contrast, the association between nonoily fish consumption and IBD incidence was not statistically significant. Consistently, the Kaplan-Meier curves showed that the probability of cumulative hazard of IBD was higher among participants who consumed less oily fish, but such pattern was not evident for nonoily fish (Supplementary Figure 5).

In fully adjusted models, we found that compared with individuals who never consumed oily fish, those having <1 serving/wk, 1 serving/wk, and >1 serving/wk had 26% (HR, 0.74; 95% CI, 0.56-0.97), 41% (HR, 0.59; 95% CI, 0.45-0.78), and 35% (HR, 0.65; 95% CI, 0.47-0.89) lower risks of CD, respectively. By contrast, nonoily fish intake was not significantly associated with CD incidence. As for UC, neither oily fish nor nonoily fish consumption was associated with its incidence (Table 3).

Fish oil supplement and IBD and its subtypes

In main analysis of the whole analytical cohort, compared with individuals who did not regularly consume fish oil, those who reported regular use had a 16% lower risk of IBD after full adjustment for covariates (HR, 0.84; 95% CI, 0.75-0.93). For IBD subtypes, intake of fish oil was associated with reduced risks of both CD (HR, 0.78; 95% CI, 0.64-0.94) and UC (HR, 0.86; 95% CI, 0.75-0.98). The Kaplan-Meier curves also showed a clear pattern that the cumulative hazard of IBD was lower in fish oil users than in nonusers (Supplementary Figure 5).

Consistently, among the subsample of 105 714 participants, an inverse relationship also existed between fish oil consumption at baseline and incident IBD. After adjusting for all covariates, those who took fish oil supplements had a 17% (HR, 0.83; 95% CI, 0.69-1.00) lower risk of IBD (Table 4). In addition, in the analysis using additional dietary data collected from multiple follow-up surveys, a negative association was found between the over-time frequency of fish oil consumption and the risk of IBD as well. Compared with the participants who never took fish oil supplements, the HRs of developing IBD were 1.05 (95% CI, 0.84-1.32) for the occasionally constant users, 0.95 (95% CI, 0.74-1.22) for the modestly constant users, 0.68 (95% CI,0.45-1.02) for the moderately constant users, and 0.71 (95% CI,0.47-1.08) for the highly constant users (Table 4). A significant trend was found between a higher frequency of fish oil supplementation and lower risks of IBD (P < .05).

Sensitivity analyses

After excluding participants experiencing an outcome event during the first 2 years of follow-up, the results did not alter significantly (Supplementary Table 4). When participants who responded that they had significantly changed their diet in the previous 5 years were excluded, neither oily nor nonoily fish showed a statistically significant association with IBD incidence. By contrast, fish oil remained statistically associated with a reduced risk of IBD (HR, 0.88; 95% CI, 0.76-1.02) (Supplementary Table 5). In models including fish and fish oil together for mutual adjustment, the risk of IBD was still dramatically lower among fish oil users, while the negative correlations between oily fish and IBD diminished (Supplementary Table 6). Last, among the subsample of participants with more detailed dietary data, the negative associations between the intake of oily fish and fish oil and IBD were also observed after additional adjustment for meat consumption and total energy intake, albeit most associations were not statistically significant (Supplementary Tables 7-9).

Discussion

In this large cohort study, we found that compared with individuals who never ate oily fish, those who regularly consumed oily fish had a decreased risk of IBD. By contrast, no significant associations were found between nonoily fish intake and IBD. In addition, fish oil supplementation was significantly associated with a reduced risk of IBD. Consistently, among a subsample with multiple dietary measurements, we found a negative association between the over-time frequency of fish oil intake and incident IBD.

Current evidence regarding the associations between fish intake and IBD is still conflicting.²¹ A case-control study in 2010 showed that a high intake of fish reduced the risks of both CD and UC. Moreover, the study found that tuna fish consumption was inversely correlated with the risk of CD but not UC.¹³ Tuna is an oily fish as defined in our study, which supports the need for separate analysis of oily and nonoily fish. The results of our study showed that oily fish could reduce the risk of IBD while nonoily fish did not, which may result from the different contents of active bioactive ingredients. For example, the amount of fat in oily and nonoily fish varies greatly. Studies have shown that 1 portion of cod contains around 0.3 g of eicosapentaenoic acid and docosahexaenoic acid (DHA), and 1 portion of salmon contains about 1.5 g, whereas 1 portion of mackerel contains about 3.0 g.²² Moreover, a diet high in oily fish would increase not only ω-3 FA intake, but also intake of iodine, protein, selenium, etc., which may lead to different health effects.⁹

Our findings that fish oil intake was negatively associated with IBD incidence are in line with some previous observational studies, albeit most of them measured the intake of ω-3 FAs, the main bioactive components of fish oil. For example, a cohort study from the UK found a statistically significant, dose-dependent, negative association between DHA, a major subtype of ω-3 FAs, and the risk of developing UC.²³ The Nurses’ Health Study also showed that higher intake of ω-3 FAs was significantly linked with a lower risk of UC but not with a lower risk of CD among the studied women.⁹ By comparison, Chan et al²⁴ found that the DHA intake was negatively related with the risk of CD. As for findings from experimental studies, particularly RCTs, current evidence to support a clear clinical benefit of marine ω-3 FAs against IBD is still weak.²⁵ A recent meta-analysis of 83 RCTs concluded that increasing ω-3 FAs intake may even increase the risk of developing IBD, but due to issues with compliance and limited trial sizes, the quality of the evidence was poor.⁹ More investigations are still required to enhance our understanding of the health effects of fish oil or ω-3 FAs on the development as well as the prognosis of IBD.

A principal methodological advantage of our study was that we were able to directly estimate the relative risk of IBD associated with fish and fish oil consumption because of the prospective cohort design. Second, owing to the large sample size and extensively collected data, we had enough power to take into consideration a wide range of potential confounding factors, and to test the robustness of our results by conducting a number of sensitivity analyses. Most importantly, unlike most previous studies that only measured dietary intake at baseline, we repeatedly assessed dietary behaviors during follow-up, which could better account for diet changes over time than a single measurement. To be noted, among the participants who completed multiple 24-hour dietary recalls, we also found a negative association between the frequency of fish oil supplementation and IBD, albeit not all groups showed statistically significant associations, which could be partially attributable to the reduced number of incident cases in each group.

There are also a few limitations in our study. First, we only measured the frequency of fish intake through simple questions (ie, how many servings were taken per week), whereas more details were lacking such as the exact portion size of each serving, the specific types of fish consumed, and the methods used to cook the fish. The absence of these factors precluded a more in-depth analysis of fish intake and IBD. Second, there was inevitably some selection bias in the UK Biobank such as healthy volunteer bias, but this should not seriously bias our results, as the internal validity was not severely damaged. Third, the study population mainly consisted of adults 40 to 69 years of age at baseline. However, in Western Europe, the average age of IBD onset was around 31 to 34 years. In both CD and UC, the incidence peaked in the age interval of 20 to 30 years of age.²⁶ Therefore, the incidence of IBD was relatively low in our study, which may have reduced the statistical power, as fewer cases could be observed. In addition, the risk of IBD is influenced by race/ethnicity. Our results could not be extrapolated to other ethnic groups. Furthermore, as we limited our study to people 40 to 69 years of age, our findings cannot be generalized to children and adolescents.

Conclusions

We observed significant associations between higher dietary intake of oily fish and fish oil and lower risks of incident IBD in a large cohort. Such associations were stable after accounting for a wide variety of potential confounders and in a number of sensitivity analyses. Our results suggest that fish oil and oily fish intake are potential protective factors against IBD.

Supplementary data

Supplementary data is available at Inflammatory Bowel Diseases online.

Acknowledgments

The authors thank all participants of the UK Biobank. This work uses data provided by patients and collected by the NHS as part of their care and support. This research used data assets made available by National Safe Haven as part of the Data and Connectivity National Core Study, led by Health Data Research UK in partnership with the Office for National Statistics and funded by UK Research and Innovation (research which commenced between 1st October 2020 – 31st March 2021 grant ref MC_PC_20029; 1st April 2021 -30th September 2023 grant ref MC_PC_20058).

Funding

Z.Z. was partially supported by the Hundred Talents Program of Sun Yat-sen University.

Conflict of Interest

The authors declare that they have no conflict of interest related to this manuscript.

Data Availability

All data used in this study were obtained from the UK Biobank under Application Number 69550. Detailed information about the UK Biobank including data access policies can be accessed at https://www.ukbiobank.ac.uk/.

References

Author notes