Elderly Patients with Moderate-To-Severe Ulcerative Colitis Are More Likely to Have Treatment Failure and Adverse Outcome

Inflammatory bowel disease (IBD) is traditionally considered to be a chronic intestinal inflammatory disease popular in young adulthood. But with population aging and increasing incidence of IBD, elderly patients with IBD are becoming more common [1]. A Sweden cohort study published in 2017 observed the incidence of elderly onset IBD of 35/100,000 person-years and approximately 24% of ulcerative colitis (UC) cases occurred in elderly individuals [2]. Similarly, with the influence of lifestyles, the incidence of UC among the elderly is gradually increasing in China as well.

Previous studies revealed that elderly onset UC may present more benign disease characteristics. A French population-based cohort conducted by Charpentier et al. [3] reported that clinical course was less aggressive in elderly onset IBD patients, while recent studies suggested that elderly onset and adult-onset UC manifested a similar disease phenotype [1, 4]. Meanwhile, clinical prognosis including UC-related surgery or death was not consistent as well. Two Canadian population-based cohort showed that elderly patients with IBD had significantly higher postoperative complications and mortality [5, 6]. Advanced age was associated with increased mortality of acute severe UC [7]. But a meta-analysis published in 2020 found that elderly onset IBD patients had a similar risk of colectomy as the young [7]. However, there have been very few reports on the disease characteristics and treatment options of elderly UC patients in Chinese mainland. In our center, therapy for UC is generally based on the disease severity and distribution, according to the third European Crohn’s and Colitis Organization consensus guideline [8] and Chinese consensus (2018, Beijing) [9]. Most patients with moderate-to-severe disease activity required to hospitalize for evaluation and treatment. 5-Aminosalicylic acids (5-ASAs) are the most commonly used medications for elderly UC patients as its perceived safety advantages [10]. So for initial onset patients, 5-ASAs are still usually given first, and meanwhile, disease assessment is performed, after which treatment escalation (steroids, immunomodulators, or biologics) is considered in our center. There may be distinctions in disease characteristics between eastern and western populations. Besides racial or genetic background, a different healthcare environment may affect disease development. Nowadays, the medical insurance coverage rate reached 96.8% in China, but biological agents were approved by State Food and Drug Administration in 2007. Due to the relatively low incidence of IBD, infliximab was included in medical insurance in 2019 and vedolizumab and ustekinumab were approved into medical insurance in recent 2 years, which might affect the availability of biologics and disease patterns of UC. So, it is extremely vital to summarize disease features and therapy conditions of Chinese elderly UC patients for further clinical decision-making.

Additionally, there were significant variations in drug prescription rates in elderly IBD patients from different regions [8], but the assessment of medication responses in elderly population is still insufficient. So, a better comparison of disease features and adverse outcomes between the elderly and nonelderly patients is strongly requested as a first step of standardizing management for elderly UC patients.

Therefore, considering our main focus on the UC patients with moderate-to-severe disease activity, we summarized a retrospective UC cohort in our center. The aim of this study was: (1) to evaluate the utilization of medical drugs and therapeutic responses in the elderly and adult patients diagnosed with UC; (2) to compare clinical outcomes including UC-related colectomy, death, severe complications, and infections between elderly and adult UC patients; and (3) to assess the impact of diagnostic age and other factors on UC-related colectomy and serious infections in UC patients.

As shown in online supplementary Figure 1 (for all online suppl. material, see www.karger.com/doi/10.1159/000522569), a retrospective cohort consisted of 1,006 consecutive cases was identified from the electronic medical record system of Peking Union Medical College Hospital (PUMCH) between January 2000 and March 2020, in which the first discharge diagnosis was UC. The diagnosis of UC was defined based on clinical, endoscopic, radiological, and pathological characteristics according to the third European Crohn’s and Colitis Organization consensus guideline [8]. Microbial testing and small bowel radiology were performed to exclude infectious enterocolitis, Crohn’s disease, or IBD unclassified. The disease surveillance in the follow-up period confirmed the diagnosis of UC. Then, we defined UC patients with the age of diagnosis over 60 years as the elderly group, which contained 89 cases [11]. Seventy-four elderly patients with moderate-to-severe disease activity were finally included based on Truelove and Witts criteria [12] at the first admission, who were eventually included in the study. Each elderly case with moderate-to-severe activity was randomly matched with 1 patient with the age of diagnosis between 16 and 59 years from the retrospective cohort as well, which formed the adult group. This matching was performed according to gender, duration (±15%), disease activity, and admission time (±1 year) without replacement by SPSS software. Patients aged below 16 years at admission, who were with incomplete medical data or had diagnostic doubts, were excluded. In our center, some UC patients had repeated hospitalizations. The elderly patients hospitalized ranged from 1 times to 8 times (median 1 time, interquartile ranges [IQR] 1–2 times), while the adult UC patients hospitalized ranged from 1 times to 23 times (median 1 time, IQR 1–2 times). Cohort entry was set as the time of first admission at our center. The study was approved by the Ethical Committee of PUMCH (S-K1142).

The included patients were followed up until the latest medical record in PUMCH or December 2019, subject to the latest time. The primary outcome was defined as UC-related colectomy or death. And the secondary outcome was the first occurrence time of severe complications and infections. Severe complications were defined as bowel obstruction, perforation, and perianal disease which needed surgical intervention, toxin megacolon, lower gastrointestinal bleeding (bleeding from a site distal to the ileocecal valve) with abnormal vital sign or urgent surgical intervention [13], and intraepithelial neoplasia. Severe infections included intestinal infections caused by bacteria, fungus, parasite, mycobacteria, Clostridioides difficile(C. difficile), cytomegalovirus, or perianal abscess, or extraintestinal infections and infectious shock.

The following data were collected mainly from medical records: (1) demographic and basic clinical information including gender, age at admission, age of onset and diagnosis, disease duration, Montreal classification (clinical phenotype, disease extent, disease activity), and body mass index (BMI) at the time of entry. (2) Clinical characteristics: initial clinical symptoms (diarrhea, hematochezia, abdominal pain, fever), extraintestinal manifestations (arthropathy, skin lesions, ophthalmopathy, and thrombosis), previous occurrence of severe complications and infections (as mentioned above), history of intestinal surgery, Charlson comorbidity index [14], and polypharmacy condition were recorded. Polypharmacy was defined as more than one kind of prescribed drugs for chronic diseases, which did not include their IBD medications. (3) Treatment information: previous and subsequent medications, including 5-ASA, systemic glucocorticoids (dosage: low: <0.5 mg/kg/day; medium: 0.5–1.0 mg/kg/day; high: >1.0 mg/kg/day), immunomodulators (including azathioprine, methotrexate, thalidomide, cyclosporine, tacrolimus), and biologics. (4) Prognostic outcomes: severe complications, infections, and UC-related colectomy during the follow-up period. Reasons for colectomy included severe activity that cannot be controlled by medications, or serious complications (bowel obstruction, perforation, toxin megacolon, severe lower gastrointestinal bleeding, and intraepithelial neoplasia).

The criteria of effectiveness were as follows: regarding that 5-ASA was widely used and there was always a combined drug therapy, it was difficult to evaluate the response to 5-ASA alone. So, we judged the effectiveness of 5-ASA at 8 weeks after the initial usage of 5-ASA according to the Toronto consensus [15]. Based on the previous studies, the effectiveness was evaluated by the partial Mayo Clinic score from their medical records [16]. Clinical remission was defined as the partial Mayo Clinic score ≤2 with every subscore ≤1 (including stool frequency and rectal bleeding). The partial response was defined as a decrease in the partial Mayo Clinic score of at least 3 points or 30% from the baseline but did not reach the criteria of remission. No change in symptoms before and after 5-ASA was regarded as no response. And as for steroids, steroid use after admission (including 1 month before admission) was selected to evaluate the response to steroid. Clinical activity index (CAI) according to Rachmilewitz index [17], C-reactive protein, and erythrocyte sedimentation rate (ESR) before and 4 weeks after steroid use and steroid dosage were recorded. CAI was generally used to evaluate the therapeutic effect of treatment, which contains 7 variables: stool frequency, percentage of blood in stools, degree of abdominal pain, general assessment of symptomatic state, temperature due to colitis, extraintestinal manifestations, and laboratory findings (ESR, hemoglobin). No change of CAI was defined as the steroid resistance, and a responsive steroid therapy meant a decrease in CAI at 4 weeks after steroid use. Steroid dependence represented that the steroid dosage cannot be reduced to 10 mg/day in 3 months or disease relapse within 3 months after discontinuation of steroid.

Continuous variables are expressed as mean ± standard if obey normal distribution, otherwise as median range and IQR, and were compared with the bilateral t test or the Wilcoxon-Mann-Whitney test. Categorical data were presented as frequencies (percentages) and compared with the χ² test or the Fisher exact test. We performed a Kaplan-Meier analysis to evaluate the association between the age of diagnosis and UC-related colectomy, severe complications, and infections. Additionally, to further evaluate indicators relevant with the risk of adverse events in the elderly patients with UC, we performed the survival analysis stratified by gender, disease duration before admission, and whether concomitant with severe complications or infections. To assess risk factors of colectomy or death, we conducted a multivariable Cox proportional hazard analysis with backward variable selection at the p value threshold of <0.10 in the univariate analysis. As partial severe infections occurred just during the first admission period, we analyzed the risk factors of infections by multivariable logistic regression. p < 0.05 was considered statistically significant. Statistical analyses were performed by using SPSS (version 25.0, IBM Corporation, Chicago, IL, USA) and R (version 3.6.2, R Foundation for Statistical Computing, Vienna, Austria).

From January 2000 to March 2020, a total of 1,006 inpatients were identified, of whom 89 (8.8%) were diagnosed as UC at an age over 60 years. Seventy-four cases in these 89 patients had moderate or severe disease activity at admission, who were further matched with 74 UC patients diagnosed between 16 and 59 years with approximately the same duration and disease activity in the same period. As there were some time intervals between the initial symptom onset and diagnosis, among these 74 elderly patients, 54 (73%) patients had the initial onset and diagnosis of UC after 60 years old, while the initial onset of UC in 20 (27%) patients appeared in their fifties who were afterward diagnosed as UC aged 60 years and over. The median age of onset in the elderly and nonelderly was also significantly different (62 [58–66] vss. 29 [22–39], p < 0.001). As shown in Table 1, at the first admission, disease duration was similar around 3 years in these 2 groups. Pancolitis accounted for most at the initial diagnosis in the elderly and adult groups (45.9% vs. 58.1%). At admission, the percentage of pancolitis increased to over 80% in both groups and adult UC patients showed a slightly more extensive involvement than the elderly patients (pancolitis: 82.4% vs. 81.1%, p = 0.05). Diarrhea (54.1% vs. 81.1%) and abdominal pain (13.5% vs. 35.1%) in the elderly patients were less common than those in adult patients rather than hematochezia. Nine (12.2%) patients in the elderly group had extraintestinal manifestations including arthropathy, ophthalmopathy, and thrombosis, whereas 3 adult patients had articular or thrombotic lesions. And BMI of the 2 group was comparable (21.48 ± 4.01 vs. 21.00 ± 3.34 kg/m²), suggesting their similar nutrient states. There was no significant difference in the prior history of severe complications or infections between these 2 groups. In detail, 7 elderly patients appeared complications previously including 1 with bowel obstruction, 1 with bowel perforation, and 1 with gastrointestinal bleeding. Two adult patients were concomitant with simple perianal disease and gastrointestinal bleeding, respectively. This patient with perianal disease was not recurrent or complex, and there was no diagnostic doubt of UC in the follow-up. Three patients in both groups had serious infections including perianal abscess, C. difficile, and cytomegalovirus infection. In terms of prior treatment, most patients used 5-ASA at the initial onset of UC (85.1% vs. 77.0%), 11 (14.9%) elderly, and 17 (23.0%) adult patients directly received systematic steroid induction therapy. Before admission in our hospital, there was no significant difference in the rate of 5-ASA (75.7% vs. 82.4%) and steroid (43.2% vs. 33.8%) usage between elderly and adult UC patients. About 52 (92.9%) elderly and 56 (91.8%) adult patients took 5-ASA with a dosage of 3 g/day and over. There was no remarkable association between clinical response and dosage. Among the 8 UC patients with a dosage less than 3 g/day, only 1 elderly patient showed no clinical response and other 7 patients with similar low dosage presented clinical remission.

As illustrated in Table 2, about 17 (23.0%) elderly and 12 (16.2%) adult patients were treated with 5-ASA after admission and subsequent follow-up period. All elderly and adult UC patients received a high dose of 3 g each day and over. Despite no significant difference, about 47.1% of elderly patients and 33.4% of adult patients showed relief in clinical presentations with different degrees to 5-ASA therapy, and meanwhile, no response accounted for 23.5% and 25% in elderly and adult groups, respectively. For steroid use, there were 56 (75.7%) patients who received medium- to high-dose systemic corticosteroids (prednisolone, ≥0.5 mg/kg/day) in both groups. 76.8% of elderly UC patients were responsive to systematic steroid, lower than that in the nonelderly group (85.7%). Compared with CAI before steroid use, a decreased level of CAI at 4 weeks after steroid induction was much less with significance (2.0 [−1.5 to 4.00] vss. 6.0 [3.0–8.0], p < 0.001). Similarly, the reduction of C-reactive protein and ESR level was also less obvious in the elderly group. More elderly UC patients manifested steroid dependence (48.8% vs. 22.9%) and resistance (23.3% vs. 14.3%). Moreover, the utilization of immunomodulators (18.9% vs. 32.4%) and biological agents (1.4% vs. 10.8%) was lower in the elderly groups than the adult group. For biologics usage, most patients (1 elderly and 6 adult patients) received infliximab. One patient received vedolizumab and 1 patient received infliximab, adalimumab, and vedolizumab.

The included UC patients were followed over median 37.5 months after cohort entry, with no significant difference between elderly and adult groups (30 [2.88–64.75] vs. 42 [20.0–60.75] months). Table 3 displayed clinical outcomes of enrolled UC patients. No significant differences were observed in the complications between the elderly and adult patients (13.5% vs. 10.8%, p = 0.62). But compared with adult patients (41 [9.5–60.75] months), complications occurred earlier in the elderly UC patients with median 20 months (0.5–61.25) since cohort entry. As for serious infection, elderly UC patients were more likely to be concurrent with serious infections than younger patients (55.4% vs. 35.1%, p = 0.013), mainly concentrating on bowel bacterial infection and extraintestinal infection. Furthermore, 21 (28.4%) elderly UC patients took UC-related colectomy, remarkably more than young patients (12.2%). The operation in elderly UC patients occurred after a median of 8 months since admission (IQR 0.5–44.75 months), significantly earlier than the median 39.5 months of adult UC patients (p = 0.001). There were 4 (5.4%) deaths in elderly UC patients and no death in the nonelderly. From the Kaplan-Meier analysis (Fig. 1), there was a significant difference between the elderly and adult groups in risks of UC-related colectomy or death (p < 0.001) and serious infections (p = 0.004) but not for severe complications (p = 0.4).

We performed univariate and multivariable Cox proportional hazard analyses to explore the risk factors of UC-related colectomy and death, which is displayed in Table 4. On multivariate regression, the older diagnostic age of UC (over 60 years) was independently associated with the increasing risk of UC-related colectomy with a hazard ratio (HR) of 2.868. Besides, male gender (HR 2.360, 95% confidence interval [CI]: 1.049–5.311, p = 0.038), steroid resistance (HR 3.039, 95% CI: 1.390–6.646, p = 0.005), prior history of complications (HR 3.091, 95% CI: 1.122–8.513, p = 0.029), and occurrence of complications in the follow-up period (HR 2.997, 95% CI: 1.317–6.823, p = 0.009) were also independently related to UC-related colectomy. There was no significant interaction effect between Charlson comorbidity index (HR 1.579, 95% CI: 0.717–3.477, p = 0.257), polypharmacy (HR 0.975, 95% CI: 0.466–2.042, p = 0.947), BMI (HR 1.057, 95% CI: 0.955–1.170, p = 0.287), and colectomy risks, respectively. Furthermore, from the Kaplan-Meier analysis stratified by disease duration, gender, complications, and infections shown in online supplementary Figure 2, age of diagnosis and complications were also validated to be associated with colectomy or death.

As for serious infections, by multivariate logistic regression (online suppl. Table 1), age of diagnosis above 60 years (odds ratio [OR] 2.774, 95% CI: 1.355–5.675, p = 0.015), extensive colon involvement (OR 2.891, 95% CI: 1.229–6.801, p = 0.015), steroid resistance (HR 3.029, 95% CI: 1.068–8.589, p = 0.037), and biologics usage (HR 4.406, 95% CI: 1.053–18.436, p = 0.042) were independently risk factors of severe infections. Also, BMI, comorbidities, and conditions of polypharmacy did not show a notable impact on infection risks. In our study, exposure to 5-ASA, steroids, and immunomodulators did not present a prominent impact on the risk of colectomy and infections.

In our 20-year retrospective UC cohort, elderly patients diagnosed with UC after 60 years old accounted for 8.8%, which was significantly lower than the adult UC patients. This proportion was comparable with elderly onset UC patients reported in recent studies from Hong Kong (11.2%) and Japan (9.9%) [18, 19]. We matched elderly patients diagnosed with UC at the age of 60 years or above to adult patients diagnosed between 16 and 59 years mainly based on gender, duration, and disease activity. Compared with adult UC patients, a similar proportion of elderly patients aged over 60 years at diagnosis received systematic steroid therapy, yet with less relief of clinical symptoms and less reduction of inflammatory indicators and higher ratio of steroid resistance. After a median follow-up time of around 3 years, the risk of UC-related colectomy, mortality, and coinfections in elderly UC patients was significantly increased. Aging was one of the independent risk factors for UC-related colectomy and severe infections.

For medical treatment, as an initial option, 5-ASA was always the most common treatment for both the elderly and adult UC patients [10]. In the patients who were treated with 5-ASAs at the beginning of admission, the response to 5-ASAs in the elderly did not present a significant difference compared to adult patients. Although this may be affected by the relatively sample size, we predicted that 5-ASAs can still be tried to initially treat moderate-to-severe elderly UC patients. But for them, 5-ASA monotherapy may have limited effectiveness and great tendency of therapeutic failure. Very few patients obtained and maintained clinical remission during our follow-up, which suggested a demand for therapy escalation and more disease monitoring for changes in symptoms.

Another commonly used drug in our elderly UC patients was systematic steroids, similarly as in the adult. Previous studies revealed that the use of systematic steroid in elderly UC patients displayed a great regional difference, ranging from 20% to 40% [2, 19‒22]. Although diarrhea and abdominal pain were less obvious in our elderly patients, actual mucosal lesion was moderately or severely active and manifested a similar progression in disease extent as in the adult patients, which led to an increased steroid demand. So, physicians should hold more active and vigilant attitude to making an earlier diagnosis for suspected IBD in elderly patients rather than take conservative therapy because of their insensitivity of subjective discomforts. However, it is worth noted that a less therapeutic response was observed in the elderly UC patients. Until now, there have been few literatures about the effectiveness comparison of systematic steroid therapy in elderly and adult IBD patients. The relatively insensitivity to the rapid induction therapy seems to be a general feature in the elderly population. Lobaton et al. [23] indicated that elderly patients receiving antitumor necrosis factor medications had less clinical response in the short-term period (after 10 weeks) but showed no difference after 6 months. Likewise, in the elderly, low clinical and biochemical response was shown in the routine time point of evaluation for steroid effectiveness (around 1 month after beginning of steroid usage), and adverse reactions will increase inevitably with a prolonged usage of systematic steroids. Therefore, steroid usage in the elderly UC patients should be more cautious, and its risk-benefit ratio needs to be fully evaluated. We can possibly try an alternative steroid sparing strategy and increase the use of immunomodulator or biologics in these geriatric IBD individuals to prolong colectomy-free survival. The utilization of immunomodulators and biologics was low in our study, which was supported by some previous studies [2, 24, 25]. In the elderly, the risk of concomitant infections significantly increased, and furthermore the advanced age was one of the risk factors of infections, which might explain the less common use of immunomodulators and biologics in the elderly UC. The relatively rapid progression of elderly UC patients led to demand for surgical intervention as immunomodulators took effect slowly. Besides, in China before 2019, its lack of medical insurance and high cost were another two important reasons for their restricted usage. Drug availability also impacted the acceptance of biologics for UC patients; for example, infliximab was mainly prescribed in large, tertiary hospitals. However, there is still great room for development of biologics in the elderly UC patients. Cheng et al. [26] found that antitumor necrosis factor medication therapy did not bring about more increased risk of adverse events in old UC patients aged 60 years and over, including severe infections, hospitalization, or death. Also, recent studies have reported that new biological agents have satisfied effectiveness and safety in the treatment of elderly UC, such as vedolizumab or ustekinumab [27, 28].

In terms of clinical outcomes, in our study, elderly UC patients presented a more aggressive and refractory course and significantly a higher risk of UC-related colectomy, death, and severe infections. There have been several studies from other Asian regions comparable with our results. A nationwide research in Japan published in 2018 found that elderly onset UC patients showed a more severe disease course and a higher proportion of UC-related surgery and hospitalization [19]. And another cohort study from Hong Kong showed more prevalent cytomegalovirus and C. difficile infections in elderly patients despite similar colectomy rates in elderly onset and nonelderly onset patients [29]. Additionally, there have been more literatures from Western countries yet with quite different conclusions as well. In a meta-analysis published in October 2020 [30], it was reported that patients with elderly onset UC had a similar risk of surgery as adult-onset UC patients at 1 year (relative risk [RR] 0.61, 95% CI: 0.29–1.27) and 5 years (RR 1.29, 95% CI: 0.79–2.11). And a recent retrospective study from the Italy observed elderly UC patients presented with similar clinical features but increased possibility of surgery and complications [22]. In this situation, we suggested that what was different from previous studies was our range of study population focusing on moderate or severe elderly patients and highlighted the difference in treatment response and prognosis between the elderly and the nonelderly with the same duration, activity, and close admission time. Besides, therapeutic options also played an indirect role in clinical outcomes. Biological therapies can alter the natural history of IBD. As biologics administration is increasing, the prevalence of IBD surgery was significantly declined [31, 32], and moreover, biologics decreased the incidence of age-related postoperative death in UC patients [33]. Thus, the lower utilization of biologics may partly explain the higher rate of colectomy in our study.

To further explore risk factors of colectomy and severe infections by multivariable regression, we revealed that aging had an independent effect on colectomy or infections. In addition, male, steroid resistance, and occurrence of complications were also independently associated with surgery risk. The emergence of complications typically indicated failed current medical therapy as some complications needed surgical interventions to resolve. But our data did not show a significant difference on complications between the elderly and the adult, which instead highlighted the decisive role of other factors including aging and response to steroid. Nonresponse to steroid often means an urgent demand for drug escalation or surgical intervention. A study by Festa et al. [34] about acute severe UC found that response to steroids was independently related to a lower risk of colectomy with median follow-up time of 44 months. Brassard et al. [35] found the significant association between systematic steroid therapy and increased risk of serious infections in patients with elderly onset IBD. Therefore, besides age itself, geriatric patients are also prone to poor response to steroids and remarkably increased infections, which can all worse clinical outcomes drastically.

This is a retrospective cohort study in which we summarized the hospitalized UC patients with the age of diagnosis above 60 years in a large, tertiary hospital during nearly 2 decades and matched with adult UC patients by gender, activity, duration, and close admission time. Then we performed a follow-up over 3 years. We believe that the enrolled criteria make the comparison of therapeutic response and long-term prognosis between the elderly and adult patients more reasonable and reliable. However, there were still some limitations in our study. First, this was a single-center study which only included hospitalized patients with a relatively limited sample size. There might be some hospitalization bias in our study as we did not include the UC outpatients, and the distinction of threshold for hospitalization between the elderly and adults might also have some potential impact on the results. Second, this study failed to analyze the effectiveness of immunomodulators and biologics due to relatively low utilization. Third, because of retrospective restrictions and complexity of polypharmacy, the effectiveness of 5-ASA was not so sufficient and requests more detailed study in the future. And the comparison of clinical characteristics among different age groups also needs to be investigated to further confirm the role of aging in disease development.

In summary, elderly patients with the age of diagnosis above 60 years were more likely to experience drug therapeutic failure such as medium- or high-dose systematic steroid therapy. In a long-term follow-up observation, as an independent risk factor, the elderly presented an increased risk of UC-related colectomy, mortality, and severe infections. Therefore, we should strategically pay sufficient attention to active surveillance and treatment. The evaluation for disease severity and therapeutic response should be more flexible according to the individual conditions. Considering the poor response to steroids in the elderly, we can try an alternative steroid sparing strategy and increase use of immunomodulator or biologics in these geriatric IBD individuals to prolong colectomy-free survival. And further studies are required to perform about more thorough comparison in therapeutic responses and safety of new biological agents to optimize management for elderly diagnosed IBD.

The study was approved by the Ethical Committee of Peking Union Medical College Hospital, Beijing, China (S-K1142), and the included subjects have given their written informed consent.

The authors have no conflict of interest.

This study was supported by National Natural Science Foundation of China (Nos. 81570505 and 81970495), Health Research & Special Projects Grant of China (Nos. 201002020 and 201502005), Natural Science Foundation of Beijing, China (No. 7202161), and CAMS Innovation Fund for Medical Sciences (Nos. 2016-I2M-3-001 and 2019-I2M-2-007).

M.Z.: study design, patient recruitment, data collection and analysis, and drafting the manuscript. H.L.: data collection and analysis support. H.Z. and X.B.: data collection. H.Y., J.Q.: study concept and design, data collection and analysis support, and critical revision of manuscript.

The data underlying this study are available from the corresponding author upon reasonable request.