Sustained Increase in Pediatric Inflammatory Bowel Disease Incidence Across the South West United Kingdom Over the Last 10 Years

Abstract

Background

Pediatric inflammatory bowel disease (pIBD) incidence has increased over the last 25 years. We aim to report contemporaneous trends across the South West United Kingdom.

Methods

Data were provided from centers covering the South West United Kingdom (Bristol, Oxford, Cardiff, Exeter, and Southampton), with a total area at-risk population (<18 years of age) of 2 947 534. Cases were retrieved from 2013 to 2022. Incident rates were reported per 100 000 at-risk population, with temporal trends analyzed through correlation. Subgroup analysis was undertaken for age groups (0-6, 6-11, and 12-17 years of age), sex, and disease subtype. Choropleth maps were created for local districts.

Results

In total, 2497 pIBD cases were diagnosed between 2013 and 2022, with a mean age of 12.6 years (38.7% female). Diagnosis numbers increased from 187 to 376, with corresponding incidence rates of 6.0 per 100 000 population per year (2013) to 12.4 per 100 000 population per year (2022) (b = 0.918, P < .01). Female rates increased from 5.1 per 100 000 population per year in 2013 to 11.0 per 100 000 population per year in 2022 (b = 0.865, P = .01). Male rates increased from 5.7 per 100 000 population per year to 14.4 per 100 000 population per year (b = 0.832, P = .03). Crohn’s disease incidence increased from 3.1 per 100 000 population per year to 6.3 per 100 000 population per year (b = 0.897, P < .01). Ulcerative colitis increased from 2.3 per 100 000 population per year to 4.3 per 100 000 population per year (b = 0.813, P = .04). Inflammatory bowel disease unclassified also increased, from 0.6 per 100 000 population per year to 1.8 per 100 000 population per year (b = 0.851, P = .02). Statistically significant increases were seen in those ≥12 to 17 years of age, from 11.2 per 100 000 population per year to 24.6 per 100 000 population per year (b = 0.912, P < .01), and the 7- to 11-year-old age group, with incidence rising from 4.4 per 100 000 population per year to 7.6 per 100 000 population per year (b = 0.878, P = .01). There was no statistically significant increase in very early onset inflammatory bowel disease (≤6 years of age) (b = 0.417, P = .231).

Conclusions

We demonstrate significant increases in pIBD incidence across a large geographical area including multiple referral centers. Increasing incidence has implications for service provision for services managing pIBD.

Introduction

Inflammatory bowel disease (IBD) comprises a chronic, relapsing, and remitting group of conditions, classically divided into Crohn’s disease (CD), ulcerative colitis (UC), and IBD unclassified (IBDU).¹ Global trends in pediatric-onset IBD incidence have demonstrated substantial increases over the last 25 years, across all age groups and disease subtypes.² Heterogeneity in incidence trends has been reported in international pediatric cohorts. In Canadian studies, the greatest increases have been observed in the population under 10 years of age.^3,4 The converse has been reported elsewhere, with incidence rates remaining stable in younger children, and increasing in adolescents in the United Kingdom,⁵ France,⁶ and Finland.⁷ A preponderance for CD has been noted globally, with ratios of 2:1 to 3:1 reported in meta-analysis.² In UK cohorts, increases in incidence have been highest among older males, though female rates continue to rise as well.^5,8 The drivers of increased pediatric incidence remain uncertain, although the interactions between genetic predisposition, Westernization of diet, and poor generation of immune tolerance to commensal microflora in childhood appear to be highly likely culprits.^9-11

Conversely, incidence in adult-onset IBD appears to have slowed or even plateaued over the last 10 years.^12-14 Despite this, due to the chronic nature of IBD, and its comparable mortality rates to the healthy population with relatively preserved life expectancy, total prevalence continues to rise.^15,16 This epidemiological pattern was modeled by Kaplan et al,¹⁷ and referred to as compounding prevalence. In this model, prevalence will continue to rise until mortality in the elderly IBD population and incident diagnoses reach an equilibrium.¹⁷ In this way, overall disease burden predominantly persists within the adult population, and it has been reported that only 1.5% of prevalent cases exist in patients ≤16 years of age. However, with increasing pediatric incident diagnoses, this healthcare burden could shift, with a greater prevalence of disease in the pediatric population.¹⁵ Disease burden remains high in this population, with pediatric IBD (pIBD) often presenting with a more severe phenotype, requiring higher surgical rates and escalated treatment regimens.¹⁸

Accurately assessing and updating epidemiological data is important for understanding patterns of disease to gain insight into pathogenesis. These data, in turn, need to be applied to tailoring and maintaining services for patients and by extension ensuring adequate workforce planning and resource allocation. The impact of changing patterns of disease on diagnostic services and treatment is considerable, and mapping incidence rates is beneficial in adjusting these services accordingly.^7,11,18

Local contemporaneous data have been reported from a number of centers across the United Kingdom.^8,12,19 However, better resolution of disease incidence often requires larger cohorts across geographical areas. Moreover, incidence rates subsequent to 2020 have been less well described. To this end, and to contribute to these areas, this study reports data from cohorts across the South West United Kingdom, from Cardiff, Oxford, Bristol, Exeter, and Southampton. We report data across a 10-year period, bookended by UK-wide census population records in 2011 and 2021, and describe age group–specific, demographic-specific, and disease subtype–specific IBD incidence. We present regional incidence with this cohort and provide data from across the COVID-19 pandemic.

Methods

pIBD Center Selection

pIBD is largely diagnosed and managed in specialist centers across the United Kingdom. Four specialist centers are included (Cardiff, Oxford, Bristol, and Southampton), covering 5 regions (with the fifth region being Exeter). Within the area covered by each specialist referral center, some diagnostic and management services are provided by local teams, with input from the specialist centers. For example, specialist pediatric gastroenterology services for the population diagnosed in the Exeter region, used throughout this work, are mostly provided by the team at Bristol Children’s Hospital. However, the Exeter region covers a large and geographically distinct area. The total Bristol referral area has been divided therefore, with Exeter including children and young people managed by services in the Royal Devon University Healthcare NHS Foundation Trust, as well as those managed and diagnosed primarily in Bristol with Devon and Cornwall post codes. The Bristol referral area is made up of all incident cases outside of this group managed by services from Bristol Children’s Hospital.

Similarly, the incident cases for South Wales are denoted by the Cardiff region, as this is the central referral center for the area. Data provided include Cardiff cases as well as patients from Swansea, another pediatric center in the region where cases have been diagnosed and managed previously.

These centers were included due to the similarities in pIBD cohort and services provided, as well as due to geographical location. Included centers are geographically continuous, allowing coverage of a composite region, covering the South West United Kingdom. This area houses a sizeable population, with an estimated total at-risk population 0 to 17 years of age of 2 947 534 in 2021.²⁰

pIBD Incident Case Identification

Data for incident cases were collected from locally derived databases at each IBD center. Data were mostly populated into local databases in a prospective manner, with some retrospective collection from case notes for subjects diagnosed prior to database establishment. All patients were diagnosed before their 18th birthday, in line with the modified Porto criteria (diagnostic criteria for pIBD integrating clinical presentation, and macroscopic and microscopic appearance from upper endoscopy and ileocolonoscopy as well as small bowel imaging).²¹ No patient identifiable data were extracted from the databases. We collated diagnosis (CD, UC, or IBDU), age at diagnosis, year of diagnosis, sex, first part of post code, and center of diagnosis into a single database for analyses. Duplicate cases identified were removed. Cases were omitted if noted to have a post code at diagnosis outside of the study area.

At-Risk Population Estimates

We utilized UK census data from 2011 and 2021 to derive at-risk population estimates in order to calculate disease incidence. We extrapolated population estimates for each year between these time points, and for 2022, by modeling the data and assuming a steady population growth over this time period. Population data are provided in the UK census by authority districts, denoting the number of residents within a defined area managed by a particular local government. The at-risk population was calculated by local authority area, corresponding to the post codes and geographic area provided by each region, and then summed to account for the total population of each IBD center. We summed the total at-risk population for the entire geographical region and calculated the overall incidence of disease for the collated South West United Kingdom.

Census data report population size by local authority in defined 5-year age brackets (0-4, 5-9, 10-14, and 15-19 years of age). To determine the at-risk pediatric population, we interpolated additional age cutoffs into the 15- to 19-year age bracket to restrict the population to those 17 years of age and below in each year of study, assuming a steady distribution of population within each age bracket. This strategy was also employed to calculate the incidence of very early onset IBD (VEOIBD) (<6 years of age), EOIBD (7-11 years of age), and pIBD (12-17 years of age).

Calculation of Incidence

Incidence was calculated per 100 000 of the at-risk population (patients <18 years of age) for each year of the study, including patients diagnosed between January 1, 2011, and December 31, 2022. This time frame (2011-2022) was selected for the potential to extrapolate the at-risk population from census data over this period, as well as to represent a modern epoch in diagnosis and treatment of these conditions, and for which granular data were available for each center. Incidence rates were calculated by center, sex, disease subtype (CD, UC, IBDU), and age at diagnosis (VEOIBD, EOIBD, and pediatric-onset IBD).

Incidence Maps

At-risk populations were calculated for local and unitary authority districts and incidence rates were calculated by year utilizing the first part of post codes provided. Where granularity of data was not available for the at-risk population, local authority districts were combined into the minimum larger area for which the population could be calculated. Data were plotted on choropleth maps for each year for the purpose of demonstrating overall size of increase in incidence, as well as for the purpose of highlighting areas of particular change, and resultant service need. Maps were created using online software from Datawrapper (https://www.datawrapper.de/).

Incidence by Center

Incidence rates were also calculated per 100 000 at-risk population for individual pediatric gastroenterology centers by referral area. A combination of the definition of referral area by pediatric gastroenterology teams at each site and the first part of the post code allowed the area background population to be calculated.

Statistical Analysis

Analyses were performed in SPSS (version 25.0; IBM). Incidence trends over the 10-year period were analyzed by linear regression for all IBD diagnoses, and for CD, UC, and IBDU. We also analyzed incidence trends by age at diagnosis and by sex using linear regression. Mean age of diagnosis by year was examined by analysis of variance.

Ethical Approval

The study used anonymized data only and was registered in each local center as a quality improvement study as required by the local governance teams. The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Results

Demographics and Results for the Total Cohort

The total number of pIBD cases diagnosed over the study period was 2497, with a mean age at diagnosis of 12.6 ± 3.2 years.

There were 966 females diagnosed with pIBD (38.7%). Total incident diagnoses increased from 187 in 2013 to 376 in 2022, with incidence rates per 100 000 at-risk population of 6.0 and 12.4, respectively. This increase demonstrated statistical significance (P < .01, Pearson’s correlation coefficient = 0.918). Data for incidence by year are visualized in Figures 1 and 2 with values for each year presented in (Table 1). There was no statistically significant difference in age at diagnosis (2013-2022) (P = .093).

Results by Disease Subtype

There were statistically significant increases in incident cases across the examined disease subtypes of CD, UC, and IBDU. Rates of CD increased from 3.1 per 100 000 at-risk population in 2013 to 6.3 per 100 000 at-risk population in 2022 (P < .01, Pearson’s correlation coefficient = 0.897) (Table 1). UC increased from 2.3 per 100 000 at-risk population in 2013 to 4.3 per 100 000 at-risk population in 2022 (P = .04, Pearson’s correlation coefficient = 0.813) (Table 1). IBDU also increased, with an incidence rate of 0.6 per 100 000 at-risk population in 2013, rising to 1.8 per 100 000 at-risk population in 2022 (P = .02, Pearson’s correlation coefficient = 0.851) (Table 1). These data can be visualized in Figure 1 and 2.

Results by Age Group

Data were also analyzed by age group bands (VEOIBD, EOIBD, and pIBD). There was a statistically significant increase in pIBD, with rates per 100 000 at-risk population increasing from 11.2 to 24.6 (2013-2022) (P < .01, Pearson’s correlation coefficient = 0.912) (Table 1). This was mirrored in EOIBD, with a statistically significant increase noted in incidence per 100 000 at-risk population, from 4.4 in 2013 to 7.6 in 2022 (P = .01, Pearson’s correlation coefficient = 0.878) (Table 1). There was no statistically significant increase in VEOIBD, with rates per 100 000 at-risk population of 1.1 in 2013 and 2.2 in 2022 (P = .231, Pearson’s correlation coefficient = 0.417) (Table 1).

Results by Sex

Females diagnosed with IBD increased significantly, with rates of 5.1 per 100 000 at-risk population in 2013 rising to 11.0 per 100 000 at-risk population in 2022 (P = .01, Pearson’s correlation coefficient = 0.865 (Table 2). This was mirrored across the male population examined, with rates increasing from 5.7 to 14.4 per 100 000 at-risk population (P = .03, Pearson’s correlation coefficient = 0.832 (Table 3). The overall trends in increasing incidence by disease subtype and age bands, as previously noted, were demonstrated in the female and male populations, and can be visualized in Tables 2 and 3, respectively.

Results by Referral Area

Data were analyzed by individual referral center, with each area noting increases in total incidence of IBD. In Cardiff, rates of total IBD rose from 4.0 per 100 000 at-risk population in 2013 to 11.3 per 100 000 at-risk population in 2022 (P < .01, Pearson’s correlation coefficient = 0.949) (Supplementary Table 1). For Oxford, rates of total IBD rose from 6.7 per 100 000 at-risk population in 2013 to 12.8 per 100 000 at-risk population in 2022 (P = .03, Pearson’s correlation coefficient = 0.831), and in Southampton, rates rose from 6.6 to 14.0 per 100 000 at-risk population (P = .01, Pearson’s correlation coefficient = 0.885) (Supplementary Table 1). There was an increase in incidence from 8.9 per 100 000 at-risk population in 2013 to 15.7 per 100 000 at-risk population in 2022 over the Exeter area (P = .034, Pearson’s correlation coefficient = 0.669) (Supplementary Table 1). There were also increases across the studied period for the Bristol area, with rates of 6.2 per 100 000 at-risk population in 2013, rising to 10.6 per 100 000 at-risk population in 2022; however, this did not achieve statistical significance (P = .51, Pearson’s correlation coefficient = 0.63) (Supplementary Table 1). Data were subdivided from each region by disease subtype, age band, and sex, with incidence rates calculated for each population. Figures demonstrating incidence for each area by subtype and age band are presented subsequently (Figure 2), and these data are included in the appendices as Supplementary Tables 1, 2, and 3. Choropleth maps for incidence per 100 000 at-risk population for each local authority or unitary district are presented in Figure 3.

Discussion

These data demonstrate increases in incident cases of pIBD in a large cohort over a duration of 10 years. We present a contemporary incidence rate of total pIBD of 12.4 per 100 000 at-risk population in 2022, with a 2× increase. The rate of increase is in keeping with other contemporary UK cohorts,¹² as well with international trends.² Data from the Wessex region have been reported since 2002 and up until 2021, noting an increase from 6.4 per 100 000 at-risk population to 12.1 per 100 000 at-risk population over this time frame.⁸ In Scotland, for 2015 to 2017, and in children <16 years of age, incident rates were reported of 12.0 per 100 000 at-risk population per year, a worldwide peak at the time of publication.²²

It has been well established that incident rates of pIBD have been increasing over the last 25 years.⁸ In their metanalysis of global cohorts, Kuenzig et al² hypothesized that during the mid to late 2010s these rates of increase had begun to plateau. However, we have demonstrated that incident figures continue to climb subsequent to this period. Multifactorial models are implicated in this ongoing increase, integrating genetic predisposition with Westernization of diet and abnormal immune response to intestinal microbiota.²³ The exact mechanism by which the environment triggers the onset of IBD is uncertain. Subsequent to 2019, the studied area saw various restrictions as a consequence of the COVID-19 pandemic, including school closures, and a decrease in the number of seasonal respiratory viral infections among the pediatric population.²⁴ The interplay between this period of decreased immune stimulation, and a subsequent potential effect on overexaggerated autoimmune response requires further investigation. As further incidence data are reported for pIBD, as well as for other chronic conditions, which are caused by a complex interaction between predisposition and the environment, as in type 1 diabetes, emerging trends may become apparent.²⁵ However, the role of delay to presentation and aversion to seeking healthcare services during this time cannot be understated in this discussion. Incident trends over the next 5 to 10 years will be of particular interest in exploring the effects of these factors, particularly if rates decrease to prepandemic levels.

Importantly, we do not report a statistically significant change in age of diagnosis over the time frame examined, making it more likely that these data represent a true rise in the number of cases seen in this population, as opposed to increasing numbers being a consequence of earlier pickup of adult- or later-onset disease. There is a risk that over the studied time frame incidence rates may be inflated by greater sensitivity in diagnostic tests and services, allowing detection of milder cases of IBD, which may have previously gone undiagnosed into adulthood, if diagnosed at all. Greater public awareness of these conditions could also have a role in promoting earlier health-seeking behaviors, particularly for those presenting with more vague or phenotypically mild symptoms. However, increases in the use of biologic medications over recent years may suggest that these cohorts are presenting with more phenotypically severe disease.²⁶

There has been heterogeneity among recent global studies. Previous UK cohorts have demonstrated a preponderance in male CD^8,27; however, work published internationally, for example in Japan, has noted higher rates of diagnosis of UC.²⁸ We report higher incident rates of CD than UC between 2013 and 2022, as well as greater numbers of presenting males than females. It is important to note, however, that there have been significant increases in both male and female populations, as well as in all subtypes of CD, UC, and IBDU; therefore, the overall incidence is not singularly driven by one sex or subtype. These data instead indicate that incident growth is consistent across sex and subtype. Moreover, within these increases we have demonstrated similar rates of increase between CD and UC (rising from 3.1 to 6.3 per 100 000 at-risk population and from 2.3 to 4.3 per 100 000 at-risk population, respectively). These data are also in opposition to epidemiological data from the adult population with IBD in the United Kingdom, in which a preponderance for females and those with UC is noted.^13,29

There is also variability in the age at presentation among international cohorts. For a Canadian cohort, rising rates of VEOIBD have been noted, with an increase of 7.2% per year in the 0- to 5-year-old population between 1999 and 2010.⁴ Moreover, other contemporary cohorts have reported consistently higher rates of VEOIBD within the pediatric population.³⁰ These findings are not supported by our data, however, with rates of VEOIBD reported between 0.6 and 2.2 for the study time frame, without a statistically significant increase noted. There is a higher likelihood of monogenic etiology and primary immunodeficiency in VEOIBD, which may provide rationale for why incidence rates in this age group have not been noted to increase, with genetic drift taking a far longer time frame than examined to affect the greater population.³¹ VEOIBD is rare, and management is an area of particular specialty—integrating genetics and immunology services. Significant increases in VEOIBD were not noted over study sites. Low total numbers, as well as year-on-year fluctuations in incidence within this cohort, are likely responsible for the lack of statistical correlation over time.

Implications for Services and Transition

These data have significant implications for service provision within pediatric and adult services managing IBD. It has been reported that pIBD makes up around 1.5% of total IBD cases, with the majority of prevalent cases managed by adult services.¹⁵ However, contemporary reports have noted stable or falling incidence of IBD among the adult population.^5,12 As a result, increasing numbers of cases diagnosed within pediatrics before transitioning to adult services could bring about a paradigm shift, wherein the division of diagnostic work is undertaken in the pediatric domain. Initial treatment, within the first months of diagnosis, is key in establishing remission and mucosal healing, making this initial period of particular importance.¹⁸ Total disease prevalence among our cohort is likely to be higher than other quoted works due to the inclusion of all patients <18 years in this dataset. Furthermore, while total prevalence may eventually stabilize due to plateauing rates of incidence within the adult population and mortality among the elderly IBD population, prevalence among the pediatric population will see significant increase. Therefore, our data have significant implications for pediatric services, wherein greater resources for diagnostics, including staff, equipment, endoscopy lists, and ward bed time are required. The time of transition to adult services is a period of particular vulnerability for young people, in which unmet health needs can lead to long-term consequences.³² With a growing number of patients undergoing this process, as incidence increases in the pediatric population, greater attention and resources will need to be allocated for this. This is also true regarding ongoing management, with a growing armamentarium of biologic medicines being increasingly used in this population, which require significant hospital contact as well as cost.²⁶ The associated costs of this care are likely to follow patients through their transition into adult care and throughout their lives.^33,34

Strengths

The strength of this work is the size of the cohort, from multiple sites and covering a vast at-risk population of nearly 3 000 000 young people. Data were mostly populated into local databases in a prospective manner, with some retrospective collection from hospital and endoscopy records prior to database establishment. This has culminated in a granular dataset, with age of diagnosis (to decimalized year), disease subtype, and postcode data available for all patients. Another strength of the methodology employed was the plotting of the first part of the post code data into local and unitary authority districts. This allowed for sense checking throughout the process, whereby it was ensured that the incident population denominator was calculated in correct relation to the present incident cases. Hence, a clearly defined background population could be calculated. It is important to note that within the National Health Service, services, particularly in specialist pediatrics, are not necessarily limited geographically, and children and young people can seek care out of their own area. We have only included patients for whom post code data correspond with the assumed referral area for each center for a more accurate estimate of the incident IBD population.

The size of this dataset, representing a significant at-risk population in the United Kingdom, increases the generalizability of these findings. These data are pertinent not only nationally, but also to other high- to middle-income countries. With incident rates of IBD increasing in countries adopting a more Western diet, these findings may continue to become more applicable in the future.

Weaknesses

While most databases were populated prospectively, there have been data retrospectively recalled. This does open up the possibility that some cases were missed. Furthermore, given the heterogeneity in referral pathways throughout, it may be possible that some cases, particularly those 16 years of age and above, will be diagnosed by local adult services. We expect that these rates of attrition will have been consistent for each site, and over time, although this was not formally tested.

While every effort was made to ensure that the at-risk population calculated was as accurate as possible, the 5-year age bands provided by the Office of National Statistics required manipulation. Population figures are provided for demographic groups of 0- to 4-year-old, 5- to 9-year-old, 10- to 14-year-old, and 15- to 19-year-old groups. The distribution of young people in each individual year was assumed to be consistent across age bands, for the purpose of calculating the total at-risk population as well as the at-risk population for each disease age group (VEOIBD, EOIBD, pIBD). The total fertility rate was relatively consistent, albeit with some decline, between 2013 and 2021, with 1.85 children per woman in 2011 and 1.61 in 2021.^35,36 Moreover, with adjustment by year for each 5-year band, the at-risk population figures used were calculated as accurately as possible without false manipulation. It is important to note that while every attempt has been made to ensure consistency across the time frame, interpolation between 2011 and 2021 would less accurate than if true population figures were available for each year. Furthermore, as slight decline in total fertility rate has been noted, assuming that consistency across the upper age band (15- to 19-year-olds) will have led to inclusion of a greater number of patients, and hence incidence figures for this older cohort will represent a slight underestimate.

While granular data were available for demographics and diagnoses, data were not recalled for factors that might contribute to the pathogenesis of IBD, such as family history, comorbidities, or socioeconomic status. Going forward, examining these risk factors alongside areas of high incidence may help to uncover causative factors in pathogenesis. Ongoing acquisition of incidence rates across contemporary and diverse cohorts will be instrumental in documenting trends and preparing services.

Conclusions

This contemporary dataset, collected over a 10-year time frame, and substantial geographical area demonstrate a significant increase in the incidence of pIBD in the South West United Kingdom. Significant increases have been noted across sex and disease subtype. Rates of IBD in those <6 years of age at diagnosis have remained stable over this time frame. These findings have considerable relevance to services managing and diagnosing IBD in pediatric and adult cohorts.

Supplementary data

Supplementary data is available at Inflammatory Bowel Diseases online.

Author Contributions

J.J.A., Z.G., and R.M.B. devised the study and its design. Data were acquired from local databases by all contributing authors and were analyzed by Z.G. and J.J.A. The article was drafted by Z.G. and J.J.A. and subsequently critically revised following comments and contributions from all authors. R.M.B. gave final approval of the submitted version.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Conflicts of Interest

J.J.A. reports that he has received payment for being on the scientific advisory board of Orchard Therapeutics and is funded by a National Institute for Health and Care Research advanced fellowship. The other authors have no conflicts of interest to declare.

Data Availability

The authors confirm that the data supporting the findings of this study are available within the article (and/or its supplementary materials). Data were generated as part of the routine care undertaken at the included institutions.

References

Author notes