Physical Activity is Associated with a Decreased Risk of Developing Inflammatory Bowel Disease: A Systematic Review and Meta-analysis

Abstract

Background and Aims

Modifiable risk factors in inflammatory bowel disease [IBD], such as physical activity, may be used as prevention strategies. However, the findings of previous studies on the association between physical activity and IBD risk have been inconsistent. We aimed to perform a systematic review and meta-analysis to estimate the effect of physical activity on IBD risk.

Methods

A search was conducted for relevant studies published before April 2023 that assessed the effect of pre-IBD diagnosis levels of physical activity on IBD incidence. Individual summary statistics [relative risks; RR], and confidence intervals [CI] were extracted with forest plots generated. We used the Grading of Recommendations Assessment, Development and Evaluation [GRADE] approach to assess the quality of evidence.

Results

Ten observational studies were included. For cohort studies, there were 1182 Crohn’s disease [CD] and 2361 ulcerative colitis [UC] patients, with 860 992 participants without IBD. For case-control studies, there were 781 CD to 2636 controls, and 1127 UC to 3752 controls. Compared with individuals with low physical activity levels, the RRs of CD in individuals with high physical activity levels for cohort and case-control studies were 0.78 [95% CI 0.68-0.88, p = 0.0001] and 0.87 [95% CI 0.79-0.95, p = 0.003], respectively. For UC, the RRs were 0.62 [95% CI 0.43-0.88, p = 0.008] and 0.74 [95% CI 0.51-1.07, p = 0.11].

Conclusion

This meta-analysis suggests that physical activity is inversely associated with the risk of developing IBD, more so in CD than in UC.

Physical activity, environmental protective factors, inflammatory bowel disease

1. Introduction

Inflammatory bowel disease [IBD] is a chronic inflammatory disorder, comprising Crohn’s disease [CD] and ulcerative colitis [UC].¹ It primarily affects the gastrointestinal tract but can also have extra-intestinal manifestations.² The exact aetiology of IBD is unknown, but it is hypothesised to be due to a complex interaction between genetic predisposition, the gut microbiota, a dysregulated immune response, and environmental factors.^3,4

IBD is a global health problem posing a significant disease burden and financial cost on patients, their families, and health care systems.^5,6 Therefore, it is important to better characterise the role of modifiable risk factors in IBD. Modifiable risk factors have the potential to be used as preventative therapies through effective public health interventions. These strategies can potentially benefit individuals at high risk of developing IBD, for example, those with a strong family history. Multiple modifiable environmental factors that could alter the risk of IBD have previously been identified. These include cigarette smoking, an unhealthy diet, obesity, and possibly physical activity level.^3,7

Physical activity has gained increasing interest with evidence of its benefit in non-communicable diseases. This includes cardiovascular disease,^7,8 as well as other autoimmune diseases such as rheumatoid arthritis.⁸ A recent consensus statement had encouraged regular physical activity in patients with IBD, due to its potential beneficial effect on a range of health outcomes.⁹ Animal models have provided evidence to suggest that exercise has an anti-inflammatory effect on the gastrointestinal tract.^10,11 In addition, in vivo studies have shown that physical activity may activate autophagy, a lysosomal degradation pathway that has been shown to protect against a range of chronic conditions, including inflammatory disorders.¹² These studies provide possible mechanisms for the possible inverse association between physical activity and the risk of IBD.

Previous observational studies investigating the association between physical activity and a protective effect on IBD risk have yielded a wide range of results and conclusions. In addition, the studies had been conducted in different demographic groups [eg, females¹³ or males¹⁴ only, older adults,¹⁵ or paediatrics,^16] and across different regions of the world.^13,16,17 There has only been one meta-analysis assessing the association between physical activity and IBD risk, but this was published almost 10 years ago,¹⁸ and new studies have been published since then. Therefore, we aimed to perform a systematic review and meta-analysis to estimate the aggregate effect of physical activity on IBD risk across all demographics.

2. Method

The Preferred Reporting Items for Systematic reviews and Meta-analyses [PRISMA] guided the design and execution of this meta-analysis.¹⁹

2.1. Literature search

A search was conducted in Medline [OVID], Embase, the Cochrane Library, and CINAHL for relevant studies published up to April 2023. Two combinations of search terms with appropriate Boolean operators were used: 1] Epidemiological studies OR Incidence OR Multivariate analysis OR Risk factors OR Protective factors AND Inflammatory Bowel Disease OR Crohn Disease OR Ulcerative colitis AND Exp physical exertion OR Lifestyle or sSdentary behaviour; 2] Exercise AND Inflammatory bowel disease Or Crohn’s disease OR Ulcerative colitis AND Risk factors.

2.2. Inclusion and exclusion criteria

Studies were included if: 1] they were cohort or case-control observational studies; 2] they measured physical activity level [of any form] as an exposure 3] the risk of developing new diagnosis of IBD was included as an outcome of interest; 4] clear description of how the confirmed new IBD cases were identified; 5] outcomes included odds ratios [ORs] or relative risk [RRs].

Studies were excluded if: 1] they were review articles; 2] the corresponding author did not respond to requests for missing information or clarification after at least two attempts 4 weeks apart, or if the corresponding author had indicated that they were unable to provide the necessary information.

2.3. Data collection

The included studies were reviewed in detail to extract study characteristics and summary statistics. These included the year of study, details of the database, geographical region, inclusion and exclusion criteria, duration of follow-up, sample size of cases and controls, the assessment and quantification of exposure, adjustment for confounders, and odds ratios and confidence intervals for the risk of IBD.

2.4. Quality assessment

We used the Grading of Recommendations Assessment, Development and Evaluation [GRADE]^20,21 approach to assess and rate the quality of evidence. In brief, the overall quality of evidence in relation to an outcome of interest is determined by five domains: risk of bias, imprecision [sample size adequacy], inconsistency [heterogeneity], indirectness [external validity], and publication bias. For each domain, assessors have the option of decreasing [and in rare cases, increasing] the level of certainty. Based on the grade of these domains, the overall level of evidence, or certainty in the quality of evidence, can be determined. This comes in four levels: ‘very low’, ‘low’, ‘moderate, or ‘high’. Because of residual confounding, evidence that includes observational data starts at ‘low’ quality. This contrasts with evidence that only consists of randomised controlled trials, which starts at ‘high’ quality.

To systematically assess the risk of bias domain, we used a risk of bias assessment tool devised by the CLARITY Group at McMaster University,²² in line with the methodology described in the GRADE Handbook.²¹ The key criteria for determining the risk of bias are: the appropriateness of eligibility criteria application [appropriate inclusion of the control population]; the appropriateness of measurement of exposure and outcome; the adequacy of confounder control; and the adequacy of follow-up [in cohort studies].

Two researchers independently assessed the studies’ eligibility, extracted data, and assessed the quality of the studies with the methodology described above. A third reviewer assisted in mediating any disagreements between the two researchers.

2.5. Statistical analysis

For cohort studies, we used the studies’ reported adjusted hazard ratios [HRs] along with confidence intervals as their summary statistics. For case-control studies, few studies reported adjusted ORs, so we estimated the ORs using the numerator and denominator data as presented. In instances where the numerator and denominator data were not reported in the study, we contacted the author for clarifications. If different levels of physical activity were reported as exposure, we used the highest level of physical activity to represent the physical activity group.

The meta-analysis and forest plots summarising the associations between physical activity and IBD incidence were generated using the RevMan 5.4 software. Random effects models were used and the heterogeneity in the strength of the association between studies was assessed using the Q-test and summarised as I² values.

3. Results

3.1. Literature search

We retrieved a total of 922 abstracts [Medline OVID 215, Embase 82, Cochrane 80, and CINAHL 545] from our search strategy [Figure 1]. After reviewing the abstracts, we excluded 904 duplicates and studies that did not fulfil the inclusion criteria. The full articles of the remaining 17 studies were retrieved for detailed review. Of these 17 studies, the following studies were excluded: four studies^23–26 did not report any measures of physical activity as an exposure for the risk of IBD; one study²⁷ was not a cohort or case-control study; one study²⁸ had missing information that was required for the meta-analysis and the author was unable to provide the data due to a limited period of 2-year data storage with their local data authority; and one study²⁹ did not respond to requests for clarifications. In the end, 10 studies were included in our analysis.

Figure 1

Flowchart of study selection

3.2. Studies characteristics

The study characteristics are summarised in Table 1 and Table 2. Three of these studies were large population cohort studies with long follow-up durations,^13–15 and the remaining seven were a mixture of small and large case-control studies.^{16,17,30–34} For cohort studies, there were totals of 1182 CD and 2361 UC, with 860 992 participants without IBD, contributing to a total of 9 122 260 person-years for the final statistical analysis. For the case-control studies, there were 781 CD cases to 2636 controls, and 1127 UC cases to 3752 controls. The study populations originate from a mixture of large population databases and tertiary hospital cohorts from different regions of the world, including Asia-Pacific, Europe, and North America. One of the case-control studies¹⁶ recruited a paediatric population [aged 16 years and under], and the other studies recruited adult populations, including older adults up to the age of 80 years. All studies included both male and female participants, except for the cohort from the Nurses’ Health Study¹³ [female-only cohort] and a Swedish Military conscription register¹⁴ [male-only cohort]. Four studies excluded patients with IBD diagnosed during the 1st year to 4th year of follow-up,^14,15,30,31 to reduce the risk of bias introduced through reverse causation.

Table 1

Study characteristics of the included cohort studies.

First author/year/database	Age range of cohort during enrolment	Measure of PA level	Highest level of PA	Control	Person-years	CD/UC cases	Adjusted confounders
Sun, 2023 [UK Biobank]	40–69	Regular vs irregular PAa	Regular PA	Irregular PA	5 036 481	707/1576	Age, gender, education, Townsend Deprivation Index, Charlson Comorbidities Index, presence of obesity, sleep adequacy, BMI, healthy diet and smoking status
Melinder, 2015 [Swedish Military Service Conscription Register]	17–18	Physical fitness [WMAX 6 min]	Highest fifth	Lowest fifth	2 724 118b	986/1878	Age, BMI, appendectomy, geographical latitude, inflammatory markers, parental socioeconomic index
Khalili, 2013 [NHS and NHS II, USA]	30–55 [NHS] and 25–42 [NHS II]	MET h/week	Highest fifth	Lowest fifth	1 355 805b	363/284	Age, smoking, BMI, oral contraceptives, hormone therapy, appendectomy, geographical latitude, NSAID use, NHS cohort, social economic status

NHS, Nurses Health Study; MET, metabolic equivalent of task; WMAX 6 min, maximum load sustainable at 6 min; PA: physical activity; NSAID: non-steroidal anti-inflammatory drugs; CD, Crohn’s disease; UC, ulcerative colitis; BMI, body mass index..

^aComposite measure of self-rated physical activity level and physical inactivity [based on reported common sedentary activities].

^bWe included only the contribution from the highest fifth and lowest fifth physical activity level included, therefore the stated person-years are a fraction of the total person-years quoted in the original article.

Table 2

Study characteristics of the included case-control studies.

First author/year/database	Age range of cohort during enrolment	Measure of PA level	Highest level of PA	Lowest level of PA	CD/UC cases [total = 781; 1127]	Control [total = 6388]
Hasosah, 2022 [cohort from three tertiary hospitals, Saudi Arabia]	0–16	Frequency of PA	>3/week	None/week	93;74	168
Furuya, 2022 [cohort from 34 hospitals, Japan]	40–69	Occupational PA [OPA] levela	High OPA level	Low OPA level	150;355	569 [CD]; 1277 [UC]
Salih, 2018 [VIP, Sweden]	40	Active v.s. inactiveb	Active	Inactive	26; 72	412
Ng, 2015 [ACCESS, Asia-Pacific]	Not specified	Frequency of PA > 30 minutes/week	Daily	<Once/week	156; 96	529
Chan, 2013 [EPIC, Europe]	20–80	Hours of recreational activity/day	>1 h/day	None/day	75; 177	300 [CD]; 708 [UC]
Hlavaty, 2013 [single tertiary IBD centre cohort, Slovakia]	Not specified	Frequency of sporting activity	≥2/week	<Twice/week	190; 148	355
Persson, 1993 [single tertiary IBD centre cohort, Sweden]	15–79	Frequency of PA	>3/week	None/week	154; 145	303

ACCESS, Asia-Pacific Crohn’s and Colitis Epidemiology study; EPIC, European Prospective Investigation into Cancer and Nutrition Study; VIP, Västerbotten intervention Project; OPA, occupational physical activity.

^a

An occupational history was determined during hospital admission. Each occupation was categorised into three levels of predetermined OPA, as determined using accelerometer from a previous study.

^bPhysical activity determined from multiple questions transformed to a composite categorical variable reflecting overall physical activity [inactive, moderately inactive, moderately active, and active]. These categories were further divided into a dichotomous variable [active vs inactive].

All studies used questionnaires to assess levels of physical activity, except for two that measured the level directly [one with heart rate monitors¹⁵ and one with bicycle ergometers].¹⁴ The definition of high-intensity physical activity is heterogeneous [Tables 1 and 2], with only one study using a standardised definition with the metabolic equivalent of task [MET].¹³

All but one study identified incident cases of CD and UC by searching the appropriate International Classification of Diseases [ICD] Codes from hospital admission or IBD-specific registries; the other study’s case ascertainment was based on subjects’ self-reported diagnosis of IBD,¹³ which was then verified by two independent gastroenterologists after reviewing medical records.

The participants’ major baseline characteristics are summarised in Table 3 and Table 4. The baseline characteristics that were reported by different studies are variable. A list of adjusted variables for the cohort studies is shown in Table 1.

Table 3

Participants’ baseline characteristics of the included cohort studies.

Studies	Agea		Male gender		Never smoker		BMI ≥30
	Exposureb	Controlb	Exposureb	Controlb	Exposureb	Controlb	Exposureb	Controlb
Sun	57.1 ± 8.2 [CD]; 57.7 ± 8.0 [UC]	56.7 ± 8.0	45.4% [CD]; 42.3% [UC]	45.40%	45.8% [CD]; 40.2% [UC]	54.40%	31.7% [CD]; 27.3% [UC]	23.80%
Khalili	41.8 ± 11.3	42.4 ± 10.6	0%	0%	54.60%	56%	7.80%	19.90%
Melinder	Not reportedc	Not reportedc	100%	100%	Not reported	Not reported	6.4%[CD]; 6.8% [UC]	7.40%

IBD, imflammatory bowel bisease; UC, ulcerative colitis; CD, Crohn’s disease; BMI, body mass index.

^aMean [standard deviation] reported.

^bExposure refers to the highest 5th quintile of physical activity level; control refers to the lowest 5th quintile of physical activity level. This is except for Sun et al., where only data for participants with IBD and without IBD [control] were presented.

^cParticipants’ age were 18–19 during the time of conscription assessment, which was the point of inclusion into the research database.

Table 4

Participants’ baseline characteristics of the included case-control studies.

Studies	Agea		Male gender		Never smoker		BMIa
	Case	Control	Case	Control	Case	Control	Case	Control
Ng	38 [25–50]	39 [26–53]	58%	55%	60%	81%	Not reported	Not reported
Chanc	56.4 [24.0–78.7] [CD]; 58.1 [30.4–80.8] [UC]	49.8 [24.0–69.0] [CD]; 52.7 [22.0/77.2] [UC]	36% [CD]; 46% [UC]	36% [CD]; 46% [UC]	58% [CD]; 73% [UC]c	72% [CD]; 73% [UC]c	25.1 [3.8]; 25.4 [3.7] [UC]	25.1[3.9] [CD]; 25.8 [4.0 [UC]
Salih	52[44–61]	50 [40–60]	55%	51%	27% [CD]; 29% [UC]	46%	25.1 [23.2–27.6]	25.3 [23.1–27.8]
Hlavaty	24.7[4–63] [CD]; 28.2 [5–71] UC]	28 [16–81]	41% [CD]; 36% [UC]	47%	58% [CD]: 64% [UC]c	81%c	Not reported	Not reported
Persson	Not reportedd	Not reportedd	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
Hasosah	Not reported	Not reported	51.50%	51.50%	Not reportede	Not reportede	Not reported	Not reported
Furuya	49.6 ± 7.8 [CD]; 52.6 ± 8.3 [UC]	52.5 ± 8.2	69.8% [CD]; 58.4% [UC]	61.80%	37.8% [CD]; 41.8 %[UC]	39.90%	Not reported	Not reported

UC, ulcerative colitis; CD, Crohn’s disease; BMI, body mass index.

^aMedian/range reported, except Salih et al. where median/25th–75th percentile was reported.

^bChan et al. is the only study which had separate control groups for CD and UC [hence two numbers reported under the control group data].

^cOnly non- smoking status at recruitment was reported. The percentage of ex-smokers was not reported.

^dThis study’s age inclusion range was 15–79.

^eThis study has a paediatric population; hence smoking status is not applicable.

3.3. Associations between physical activity and Crohn’s disease

The forest plots of IBD RRs as influenced by physical activity level are shown in Figures 2 and 3. Compared with individuals with low physical activity levels, those with high physical activity levels have a statistically significant lower risk of developing CD. This association is seen in both cohort and case-control studies, with ORs of 0.78 [95% CI 0.68-0.88, p = 0.0001] and 0.62 [95% CI 0.43-0.88, p = 0.008], respectively [Figure 2]. There was no statistically significant heterogeneity between the studies in the cohort summary [I² = 0%, p = 0.49] but within the case-control studies there was evidence of heterogeneity [I² = 69%, p = 0.004] studies.

Figure 2

Forest plot in Crohn’s disease for cohort studies [top] and case-control studies [bottom].

Figure 3

Forest plot in ulcerative colitis for cohort studies [top] and case control studies [bottom].

3.4 Associations between physical activity and ulcerative colitis

For UC, there is a statistically lower risk of disease in cohort studies [OR = 0.87, 95% CI 0.79-0.95, p = 0.003] [Figure 3]. A reduction in the risk of UC in the physical activity group among the case-control studies did not reach statistical significance [OR = 0.74, 95% CI 0.51-1.07, p = 0.11]. There was no statistically significant heterogeneity between the studies in the cohort summary [I² = 0%, p = 0.51] but within the case-control studies there was evidence of significant heterogeneity [I² = 81%, p <0.001] studies.

3.5 Quality of evidence: GRADE approach

The quality of evidence was assessed separately for cohort and case-control studies.

For the cohort studies, most studies had low risk or some concern only across all risk-of-bias domains [Table 5]. Therefore, no serious limitations were found for the cohort studies. Similarly, we found no serious limitations in the other domains [indirectness, imprecision, inconsistency, and publication bias] under the GRADE approach [Table 7 and Supplementary Table 1]. Therefore, no downgrading took place for the cohort studies based on the findings from all five domains.

Table 5

Risk of bias assessment for cohort studies

Table 6

Risk of bias assessment for case-control studies

Table 7

Summary of findings table.

Outcome: risk of developing IBD		Relative effect [95% CI]	Number of studies	Cases [physically active/total casesd]	Controls [physically active/total controlsd]	Certainty of the evidence
Cohort studies	Crohn’s disease	OR 0.78 [0.68-0.88]	3	2056	860 992e	⨁⨁◯◯ LOWa
	Ulcerative colitis	OR 0.62 [0.43-0.88]		3738
Case control studies	Crohn’s disease	OR 0.87 [0.79-0.95]	7	267/781	1057/2636	⨁◯◯◯ VERY LOWb,c
	Ulcerative colitis	OR 0.74 [0.51-1.07]		347/1127	1224/3752

GRADE working group grades of evidence include the following. HIGH: a lot of confidence that the true effect lies close to that of the estimated effect; MODERATE: moderate confidence in the estimated effect, the true effect is likely to be close to the estimated effect, but there is a possibility that it is substantially different; LOW: limited confidence in the estimated effect, the true effect might be substantially different from the estimated effect; VERY LOW: very little confidence in the estimated effect, the true effect is likely to be substantially different from the estimated effect.

IBD, inflammatory bowel disease.

^aNo serious limitations were found across all five GRADE Domains. However, because the body of evidence comes from observational studies, the overall certainty of the evidence starts at ‘LOW’ quality.

^b4/7 studies were found to have a serious risk of bias due to their methodology for the exposure assessment domain, which leads to a high risk of recall bias.

^cSubstantial heterogeneity is present for CD [p = 0.008, I² = 69%] and UC [p <0.00001, I² = 81%].

^dThe data of physically active as a proportion to the total number of cases or control are presented for case-control studies only.

^eThe ‘controls’ in cohort studies refers to study participants without a diagnosis of IBD.

For the case-control studies, 4/7 studies were found to have a serious risk of bias due to their methodology for the exposure assessment domain [Table 6 and Supplementary Table 1]. The questionnaires for physical activity level were administered after a diagnosis of IBD had been made, in some cases up to a few years after diagnosis, which increases the risk of recall bias. Additionally, four of seven studies only adjusted or matched for a small number of prognostic variables [for instance, age and sex only]. The risk of bias therefore contributed to a downgrade of one level for the overall quality of evidence in case-control studies. There was also statistically significant heterogeneity between both CD [p = 0.008, I² = 69%] and UC [p <0.00001, I² = 81%] studies. Therefore, there is a serious risk of inconsistency, which contributes to a further downgrade of one level for the overall quality of evidence in case-control studies.

The GRADE guidelines dictate that the overall quality of a body of evidence that comes from observational studies starts at ‘low quality’. Therefore, although there are no serious limitations across all domains for the cohort studies, which leads to no downgrading, the overall quality of evidence is ‘low’ [Table 7 and Supplementary Table 1]. For the case-control studies, the presence of a serious risk of bias and serious inconsistency led to a downgrading of two levels. The overall quality of the evidence is therefore ‘very low’ [Table 7 and Supplementary Table 1].

4. Discussion

Physical activity has gained increasing interest with evidence of its benefit in the prevention and management of non-communicable diseases. This includes cardiovascular disease^7,8 as well as other autoimmune diseases such as rheumatoid arthritis.⁸ Modulation of autophagy, a lysosomal degradation pathway of skeletal and cardiac muscles, is a potential mechanism between exercise and decreased IBD risk. In vivo studies have shown that physical activity induces autophagy, which may offer protection against a range of chronic conditions, including inflammatory disorders.¹² CD susceptibility loci within autophagy pathways have also been identified in a genome-wide association study.³⁵ In addition, animal models suggest that physical activity has anti-inflammatory effects in the intestines.^10,11

In this systematic review and meta-analysis, we included three large cohort studies and seven case-control studies that assessed the effect of the level of physical activity on the risk of developing new-onset IBD. We found that a high level of physical activity is associated with a statistically significant decrease in the risk of CD by 22% and 38%, as shown in the cohort and case-control studies, respectively. In UC, there was also a statistically significant decrease in risk of 13% as shown in the cohort studies, but a risk reduction of 26% that did not reach statistical significance was shown in the case-control studies. On assessing the quality of evidence with the GRADE approach, there were no serious limitations noted in the cohort studies, but there were serious limitations found in the case-control studies due to the high risk of bias and the significant heterogeneity present.

To our knowledge, there has only been one meta-analysis, published almost 10 years ago, which assessed the association between physical activity and IBD risk.¹⁸ This analysis also showed an association between physical activity and lower CD risk, but only in the aggregate OR of both case-control and cohort studies and not in the subgroup of cohort-study design. In comparison, we showed a protective effect for CD in both cohort and case-control study designs. We also showed that there is a similar but smaller association in UC when restricted to cohort studies, which was not demonstrated in the previous meta-analysis by Wang et al.¹⁸ This is likely because our analysis included newer studies, including two additional large prospective cohort studies^14,15 with long follow-up periods, which allowed us to detect a relatively smaller protective effect in UC by virtue of a larger total sample size.

Our study has several strengths. First, our methodology is robust and in keeping with PRISMA guidelines. We used the GRADE approach to grade our quality of evidence, which is the most widely adopted tool to assess the quality of evidence and is known for its reproducible and transparent framework.³⁶ We separated the meta-analysis and grading of the quality of evidence for cohort and case-control studies to allow better insight into the protective effect of physical activity and the overall quality of evidence. Second, our study included a large sample size. In the cohort study, there were totals of 1182 CD and 2361 UC, with 860 992 participants without IBD, contributing to a total of 9 122 260 person-years. For the case-control studies, there were 781 CD cases to 2636 controls, and 1127 UC cases to 3752 controls. This has led to improved precision in our estimate, particularly for the cohort studies, which had narrow confidence intervals. Third, our findings have improved external validity as compared with the findings from the individual studies, as this meta-analysis combines studies with a variety of demographics [eg, female¹³ or male¹⁴ only, older adults,¹⁵ or paediatrics^16] across different regions of the world [eg, USA, Europe, and Asia-Pacific].

There are important limitations that require caution in interpreting the findings of our study. First, all the included studies are observational and therefore suffer from the risk of residual confounding. Even after a seemingly comprehensive adjustment of possible confounding factors [Table 1], it is difficult to know for certain that our identified association is entirely independent of other confounders known to be associated with high physical activity, such as a healthier diet,³⁷ non-smoking status,³⁸ better socioeconomic status,³⁹ to name a few. In addition, four of seven case control studies assessed physical activity level with questionnaires after a diagnosis of IBD had been made, which increased the risk of recall bias, reflected in the downgrading of the overall quality of evidence for case control studies [although this risk is absent for all the prospective cohort studies]. Second, there is a risk of reverse causality accounting for the apparent association, where subjects with pre-diagnosis symptoms of IBD are less physically active due to disease activity. However, two of the cohort studies and two of the case-control studies excluded IBD diagnosed at least 12 months from recruitment to minimise the risk of reverse causality. Third, for the case-control studies, we calculated the ORs using the available denominator and numerator data. However, these unadjusted ORs are very comparable to the adjusted ORs where they were reported. In addition, the cohort studies adjusted for important confounders such as socioeconomic status,^13–15 body mass index [BMI],^13–15 smoking status,^14,15 and healthy diet,¹⁵ and despite this, the risk reduction associated with physical activity was still evident and therefore not solely explained by confounders. Fourth, most of the studies ascertained levels of physical activity with questionnaire surveys, which could lead to misclassification of physical activity levels. However, for the studies where the physical activity data were collected before the development of IBD [all three cohort studies and three of seven case-control studies], it is likely that this misclassification is nondifferential for outcome, which will likely drive any associations toward the null hypothesis. Last, the definitions of the level of physical activity are heterogeneous, with only one study using a standardised definition with the metabolic equivalent of task [MET].¹³ This may explain some of the large heterogeneity present in the case-control aggregate analysis. However, despite this, the overall ORs from the cohort studies had no statistically significant heterogeneity, and one could argue that this can further strengthen the findings from the cohort studies. We intended to perform a subgroup analysis by demographic features [eg, age, gender, geography] to investigate the heterogeneity, but the number of studies we retrieved is too small to enable meaningful comparisons between these subgroups.

IBD is a global health problem posing a significant disease burden and financial cost on patients, their families, and health care systems.^5,6 Therefore, it is important to better characterise the role of modifiable risk factors in IBD. Modifiable risk factors have the potential to be used as preventative therapies through effective public health interventions. The association between higher physical activity level and lower risk of inflammatory bowel disease identified by our study suggests physical activity as a possible modifiable risk factor for IBD. There could be a role of physical activity as a prevention strategy against developing IBD. In addition to implementing public health interventions to increase physical activity levels, there may be a place for physicians to advise increased physical activity levels, particularly to individuals at high risk of developing IBD, such as those with a strong family history of IBD. Physical activity also has a number of non-IBD health benefits, which supports the importance of its recommendation to patients, including improved quality of life and wellbeing,⁴⁰ improved cardiovascular health,⁴¹ and reduced risk of some cancers and even premature death.⁴² However, further studies with more robust assessment of the effect of all major confounding factors on the association between physical activity and IBD risk, as well as studies that assess effect of physical activity on disease activity in patients with existing IBD, will better support the recommendation of physical activity in clinical practice.

In conclusion, in this systematic review and meta-analysis of observational studies, higher levels of physical activity are associated with a lower risk of Crohn’s disease. In ulcerative colitis, such an association was only demonstrated when restricted to cohort studies. This suggests a role for physical activity as a prevention strategy against developing IBD.

Supplementary Data

Supplementary data are available at ECCO-JCC online.

Funding

This study received no funding.

Conflict of Interest

HTT, DF, and CF declare no conflict of interests. AA received grant from the Helmsley Charitable Trust [National Institutes of Health]; and consulting fees from Geneoscopy. RG received consulting fees from AbbVie, Zespri, Janssen; and honoraria from AbbVie, Zespri, Janssen, and Takeda.

Author Contributions

Concept and design of the work was conducted by RG and HTT. The systematic review of the literature and data analysis was conducted by HTT and DF. The statistical analysis was conducted by CF. Each of the authors contributed to the drafting and revision of the manuscript and approved the final version of the manuscript submitted.

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding author. Conference presentation: New Zealand Society of Gastroenterology Annual Scientific Meeting, Rotorua, New Zealand, 2023.