Estimated Glomerular Filtration Rate and the Risk of Inflammatory Bowel Disease in Adults: A Swedish Population-Based Study

Abstract

Background

Kidney complications are common in patients with long-standing inflammatory bowel disease (IBD). Whether kidney complications, defined as low estimated glomerular filtration rate (eGFR), may predispose to later IBD is unknown.

Methods

We analyzed the association between eGFR and the risk of being subsequently diagnosed with IBD among 1 612 160 adults from Stockholm. The exposure was categories of eGFR, with 90 to 104 mL/min/1.73 m² as the reference. Cox regression models were used to investigate the association between eGFR, IBD, and IBD subtypes. Subgroup analyses included age strata, sex, education, and comorbidities. To explore the possibility of detection bias or reverse causation, we estimated IBD hazard ratios (HRs) after excluding cases and individuals censored during early years of follow-up.

Results

During a median of 9 years of follow-up, we detected 9663 cases of IBD (3299 Crohn’s disease, 5072 ulcerative colitis, 1292 IBD unclassified). Lower eGFR levels were associated with higher IBD risk (for eGFR 30-59 mL/min/1.73 m²: adjusted HR, 1.15; 95% confidence interval [CI], 1.01-1.33; and for eGFR <30 mL/min/1.73 m²: adjusted HR, 1.65; 95% CI, 1.16-2.37). This association was stronger in magnitude for Crohn’s disease (for eGFR 30-59 mL/min/1.73 m²: HR, 1.33, 95% CI, 1.04-1.72; and for eGFR <30 mL/min/1.73 m²: HR, 2.25; 95% CI, 1.26-3.99). Results were consistent across strata of age, comorbidities, and attained education but suggested the association between eGFR and IBD to be stronger in women (P for interaction <.05). Results attenuated but were robust to exclusion of early IBD cases.

Conclusions

We observed an association between reduced eGFR and the risk of developing IBD, which was stronger in magnitude for Crohn’s disease.

INTRODUCTION

Chronic kidney disease (CKD) is common, with an age-standardized prevalence of 8% to 16% in most developed countries,¹ and associated with a higher risk of comorbidities and complications, such as cardiovascular disease,² infections,³ anemia,⁴ or cancer.⁵ CKD inevitably results in disorders of gut microbiota,⁶ increased intestinal permeability, and small-bowel inflammation,^7,8 with clinical manifestations evident already at early stages.⁹

Inflammatory bowel disease (IBD) is a chronic inflammatory disease of largely unknown etiology, characterized by flares of inflammatory activity in the gastrointestinal tract and a higher risk of gut permeability.¹⁰ It is not uncommon for persons with established IBD to develop kidney complications,^11–13 but whether CKD-induced alterations in the gastrointestinal tract may lead to the development of IBD has not been well studied. Recent observational studies have reported a high prevalence of diagnosed IBD in patients undergoing kidney biopsy,¹⁴ and increased risk of IBD in persons with immunoglobulin A nephropathy (IgAN) compared with control subjects.^15,16 Case reports have also described a correlation between inflammatory activity in IBD and progression or regression of IgAN.^17,18

In this study, we test the hypothesis that loss of kidney function, regardless of underlying renal etiology, may lead to alterations in the gastrointestinal tract that in turn increase the risk of developing IBD. We thus set up to explore the risk of IBD across the full spectrum of estimated glomerular filtration rate (eGFR) in a large repository of healthcare data from the region of Stockholm, Sweden.

METHODS

Data source and study population

We used data from the Stockholm Creatinine Measurements (SCREAM) project, a healthcare utilization cohort from the region of Stockholm, Sweden, covering the period from 2006 to 2018.¹⁹ A single healthcare provider in the Stockholm region provides universal and tax-funded healthcare to 20% to 25% of the population of Sweden. Using unique personal identification numbers, SCREAM linked regional and national administrative databases that hold complete information on demographics, healthcare utilization, laboratory tests undertaken, dispensed drugs, diagnoses, and vital status until the end of 2019, without loss to follow-up. The Regional Ethical Review Board in Stockholm approved the study (reference 2017/793-31); informed consent was not deemed necessary because all data were de-identified at the Swedish Board of Health and Welfare.

For this study, we included all Stockholm citizens who underwent serum or plasma creatinine testing in outpatient care at an age ≥18 years. For each participant, we defined the study baseline as the time at which their first measurement of outpatient creatinine was available. Exclusion criteria were a recorded history of IBD, ongoing medication of 5-aminosalicylic acid (5-ASA), death on the same day as baseline, and missing information on age or sex. We excluded persons with ongoing 5-ASA prescription (during the period of 6 months before and 1 month after baseline) because it is the most specific first-line medication to control inflammation in IBD.²⁰

Study exposure

The study exposure was eGFR, calculated from the baseline creatinine test. All performed creatinine measurements were standardized to isotope dilution mass spectrometry standards. We considered creatinine tests with plausible concentrations within the range of 0.5 to 17.0 mg/dL and performed in connection with an outpatient healthcare encounter. We used the 2009 CKD Epidemiology Collaboration equation without correction for race,²¹ and categorized eGFR into 5 groups—<30, 30 to 59, 60 to 89, 90 to 104, and 105+ mL/min/1.73 m²—with 90-104 mL/min/1.73 m² being the reference group. This is consistent with previous research, showing that 90 to 104 mL/min/1.73 m² is an ideal reference as it allows risk to be assessed at higher and lower eGFR.^2,22

Study outcomes

The primary study outcome was diagnosis of IBD during follow-up, which was defined as having at least 2 relevant International Classification of Diseases–Tenth Revision diagnostic codes in either outpatient or inpatient specialist care (Supplementary Table 1). The date of the first diagnosis was considered the event date. The secondary study outcome was the type of IBD. The 2 most common types of IBD are Crohn’s disease (CD) and ulcerative colitis (UC).²³ Patients with the same IBD type in both diagnostic listings were considered as having either CD or UC, whereas patients with discordant IBD diagnostic listings or at least 1 listing of IBD unclassified were defined as having IBD unclassified.^24–26 Participants were censored at the end of follow-up (December 31^st, 2019), death, or emigration from the region of Stockholm, whichever occurred first. Death date was obtained from the National Board of Health and Welfare’s Cause-of-death register (http://www.socialstyrelsen.se). Emigration from the region was obtained from the regional census information.

Study covariates

Covariates were defined at baseline, including age, sex, highest attained education, substance abuse (alcohol and tobacco), and selected comorbidities and ongoing medications. Attained education was categorized into 3 levels: compulsory school (≤9 years), secondary school (10-12 years), and university (>12 years), and obtained from the LISA (longitudinal integrated database for health insurance and labour market studies) register.²⁷ Presence of comorbidities was ascertained using International Classification of Diseases (ICD)-10 codes registered in the regional database of healthcare consumption (VAL) any time prior to baseline and since 1997. Comorbidities included any history of diagnosed obesity, hypertension, diabetes mellitus, congestive heart failure, myocardial infarction, stroke, atrial fibrillation, and recent cancer (in the last 3 years), with specific definitions detailed in Supplementary Table 2.²⁸ Definitions of comorbid hypertension and diabetes were further enriched with information on recent dispensation of related medications. Tobacco and alcohol abuse were ascertained by relevant International Classification of Diseases–Tenth Revision diagnostic codes. Information on ongoing medications was ascertained using Anatomical Therapeutic Chemical codes registered in the national prescribed medication register, which collects all pharmacy dispensations of prescribed drugs. Medications were considered ongoing if a claim was detected at baseline or up to 6 months prior. We considered ongoing use of angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, β-blocking agents, calcium-channel blockers, nonsteroidal anti-inflammatory drugs, and statins (Supplementary Table 2).

Data analysis

Descriptive statistics for continuous variables were reported as mean ± SD and categorical variables as counts and percentages. We calculated incidence rates with 95% confidence intervals (CIs). Multivariable-adjusted Cox proportional hazards regression was used to study the association between eGFR categories and IBD. We performed sequential adjustment and adjusted for ongoing medications in a final step to explore whether medications commonly used to treat CKD explained the link between eGFR and IBD.

We performed several sensitivity analyses to test the robustness of results. First, we excluded individuals with IgAN diagnosis because a previous report suggested increased IBD risk in them,¹⁵ and we wanted to evaluate whether associations were observed in the remaining population with CKD. Second, to explore the possibility of detection bias and reverse causality, we excluded IBD cases and individuals who were censored within the first 1, 2, and 3 years of follow-up, to account for possible diagnostic delay during the work-up of an IBD case.^29,30 Third, we explored the association between eGFR categories and IBD risk in prespecified subgroups, including sex, age, attained education, and presence (vs absence) of selected comorbidities. The interaction between eGFR categories and the subgroups was checked by adding an interaction term in the statistical models. Finally, we explored specific CKD related etiologies in the subgroup of patients with eGFR <30 and 30-59 mL/min/1.73 m². We performed all analyses using R software (R version 4.0.5) (R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

Baseline characteristics

There were approximately 1.76 million adult Stockholm residents accessing health care and undergoing creatinine testing between 2006 and 2018. After applying exclusion criteria (Supplementary Figure 1), the study cohort consisted of 1 612 160 adults without IBD history at baseline. The mean age was 47 ± 18 years and 53% were women (Table 1). The mean eGFR was 97 ± 21 mL/min/1.73 m²; 37% of participants had an eGFR ≥105 mL/min/1.73 m², 29% had an eGFR of 90 to 104 mL/min/1.73 m², 29% had an eGFR of 60 to 89 mL/min/1.73 m², 4% had an eGFR of 30 to 59 mL/min/1.73 m², and 0.5% had an eGFR of <30 mL/min1.73 m². The most common comorbidities were hypertension (12.9%), recent cancer (10.0%), cardiovascular disease (6.5%), and diabetes mellitus (5.4%).

Incidence rate and hazard ratios of IBD across eGFR strata

Over a median follow-up of 9 years, 9663 new cases of IBD (0.6% of the study participants, 3299 CD cases, 5072 UC cases, 1292 IBD unclassified cases) were detected throughout 13 458 481 person-years (py). The overall incidence rate was 7.18 per 10 000 py. We observed higher incidence rates in participants with eGFR <30 (8.61 per 10 000 py) or ≥105 (9.50 per 10 000 py) mL/min/1.73 m². The most common IBD subtype was UC, representing 53% of all IBD cases, with incidence rate of 3.77 per 10 000 py (Table 2, Supplementary Table 6). In general, the incidence rates of single IBD types were higher at both eGFR extremes.

In multivariable-adjusted Cox regression models, compared with eGFR of 90 to 104 mL/min/1.73 m², lower eGFR strata were associated with higher IBD risk, with an adjusted hazard ratio (HR) of 1.15 (95% CI, 1.00-1.33) for eGFR of 30-59 mL/min and 1.65 (95% CI, 1.16-2.37) for eGFR <30 mL/min (Table 2). Table 2 also shows associations between eGFR strata and IBD subtypes: compared with eGFR of 90 to 104 mL/min/1.73 m², the HR for developing CD or IBD unclassified was higher across lower eGFR strata. The association between low eGFR categories, and the risk of UC remained less clear.

Sensitivity and subgroup analyses

We excluded 716 participants with a diagnosis of IgAN at baseline, which did not materially modify the results from our main analyses (not shown). Excluding IBD cases and individuals who were censored during the first 1 year of follow-up resulted in an attenuation of the associations observed, but we still observed an increasingly higher risk of IBD with lower eGFR, with an HR of 1.64 (95% CI, 1.07-2.50) for participants with eGFR <30 mL/min/1.73 m² compared with eGFR 90 to 104 mL/min/1.73 m² (Table 3, Supplementary Table 7). A similar pattern of association was observed when excluding IBD cases and individuals censored within the first 2 and 3 years of follow-up, or when evaluating single IBD subtypes (Table 3, Supplementary Table 7).

Subgroup analyses showed consistency across baseline age categories, attained education, presence of diabetes, cardiovascular diseases, hypertension, and cancer (Figure 1, Supplementary Table 4) but suggested that the association between low eGFR strata and IBD was stronger in magnitude for women compared with men (P values for interaction <.05).

Given the consistent association between low eGFR strata and IBD risk, we evaluated potential differences in the distribution of CKD etiologies (by clinical diagnoses) that might have driven the association. We failed to observe any major differences in the distribution of CKD diagnoses between persons with eGFR <30 or 30-59 mL/min/1.73 m² developing or not IBD (Supplementary Table 5).

DISCUSSION

In this large cohort study representative of the region of Stockholm, Sweden, we observed a modest association between low eGFR and the risk of being subsequently diagnosed with IBD. These findings are novel, and we were unable to find other studies in the literature to compare.

Our results offer support to the hypothesis that intestinal permeability and small-bowel inflammation of CKD^7,8 may link to subsequent IBD. These links have been established for IgAN,^15–18 but a high prevalence of diagnosed IBD has also been reported in patients undergoing renal biopsy for any clinical indication.¹⁴ The etiology of IBD remains unclear even today, and our observational analysis cannot determine whether CKD per se is a risk factor for IBD, whether they share genetic risk factors (such as immune-mediated inflammatory diseases, so that co-occurrence is probable)³¹ or whether yet unknown conditions (such as environmental factors)³² may predispose to both diseases and explain our findings. Regardless of which, our results indicate a link between both conditions, which warrants further investigation and may motivate greater surveillance for gastrointestinal diseases in people with progressive CKD.

IBD is not restricted to the gastrointestinal tract, and many patients experience the so-called extraintestinal manifestations.³³ These include, among others, kidney problems in up to 23% of IBD cases, largely explained by urological complications such as nephrolithiasis, or glomerular nephropathy, potentially aggravated by the use of medications to treat IBD such as 5-ASA and cyclosporine, which in turn are often associated with the development of tubulointerstitial nephritis.^12,13,17,33 Our results add to this knowledge and suggest the possibility of the association between these conditions to be bidirectional.

We recognize that the associations are modest in magnitude, and that despite our large sample size the outcome was rare (only 0.6% of study participants developed IBD during a median 9 years of follow-up), leading to broad CIs, particularly when evaluating IBD subtypes. Despite some of the estimated associations being nonsignificant, we note that the estimated associations for low eGFR levels remained elevated throughout all analyses, particularly for eGFR <30 mL/min/1.73 m². Thus, there is a clear and consistent pattern across analyses, indicating that low eGFR levels do represent a risk factor for IBD.

We observed higher IBD risk at both ends of the eGFR spectrum. Interpreting the risks of persons with eGFR >105 mL/min/1.73 m² is challenging because eGFR is an imperfect measure of kidney function affected by nonrenal factors and influenced by other diseases.³⁴ Thus, it is unknown if high eGFR in our study represents the individuals’ kidney function, glomerular hyperfiltration or presence of other conditions, such aging or frailty that result in lower muscle mass stores (hence lower creatinine in circulation). In support of this, adjustment for comorbidities largely attenuated the crude associations between IBD incident and the category of eGFR >105 mL/min/1.73 m². Finally, we observed a higher IBD risk for women in the lower eGFR strata compared with equal men, which could be attributed to the well-known difference in the age of IBD onset between sexes, particularly for CD.³⁵

Some strengths and limitations need to be considered when interpreting our results. To begin with, like all observational studies, there might have been unmeasured confounding. The criterion to enter in our cohort was to have a creatinine tested in connection to a healthcare encounter, and reasons for creatinine testing may be informative. IBD is often diagnosed at young age and in Sweden, the average age of IBD diagnosis is 39 years.²³ Routine creatinine testing in young otherwise healthy individuals may be less common, and may explain why the incidence of IBD in our study was slightly higher than that reported for the Nordic countries.^36,37 We used an algorithm to identify IBD cases of high predictive value.²⁵ However, it is based on clinical detection and diagnosis, and we cannot exclude the possibility that gastrointestinal disturbances in persons with low eGFR have been gone unnoticed, undiagnosed, or attributed to other diseases. Creatinine testing may be performed as part of the workout or investigations toward an IBD diagnosis. Although diagnostic delays in IBD vary across patients’ age, countries, and different settings of health care, studies have showed similar lags,^{29,30,38–40} and among these, a study showed that about 75% of CD patients receive their IBD diagnosis within 24 months and 90% of UC patients within 36 months.³⁰ Because our sensitivity analyses excluding IBD cases during the first 3 years of observation gave consistent results, we feel that our findings are consistent and robust.

To conclude, this study observed a modest association between reduced eGFR and the risk of developing IBD in the community that suggest bidirectionality between both conditions. Future studies should confirm this association and elucidate potential underlying mechanisms. As clinical implications, our findings are important in establishing and raising awareness of the plausible connection between both conditions. Until further studies are performed to evaluate the causality between this connection, clinicians may consider the possibility of IBD in patients with low kidney function reporting gastrointestinal problems. This may accelerate the screening and exploration of symptoms delaying the time to differential diagnosis and management.

Acknowledgments

We thank Roemer J. Janse, Nanbo Zhu, Faizan Mazhar, Alessandro Bosi, and Vivekananda Lanka for their generous support on this project. This study is based on population-based datasets, and General Data Protection Regulation was enacted and set out rules for personal data processing.

Funding

This study was supported by the Swedish Research Council (grant number 2019-01059), and the Martin Rind Foundation. O.O. was supported by grants from the Swedish Research Council and the Regional Agreement on Medical Training and Clinical Research between Stockholm County Council and Karolinska Institutet (ALF).

Conflict of Interest

O.O. has received research funding from, received speaker fees for, or participated on advisory boards for Janssen, Ferring, Takeda, AbbVie, Bristol Myers Squibb, Galapagos, and Pfizer for topics not related to the present study. J.J.C. has received research funding from, received speaker fees from, or participated on advisory boards for Vifor Pharma, AstraZeneca, Amgen, Astellas, Abbott, Bayer, Fresenius, Nestlé, and MSD for topics not related to the present study. J.F.L. has coordinated a study on behalf of the Swedish IBD quality register (SWIBREG), which received funding from Janssen Corporation. Y.Y. and A.S. have no conflicts of interest to declare.

REFERENCES