Presented in part at the 106th Annual Meeting of the American Gastroenterological Association, Chicago, Ill, May 14-19, 2005. Published in abstract form in Gastroenterology. 2005;128 (Suppl 2):A327.
A population-based study of the frequency of corticosteroid resistance and dependence in pediatric patients with Crohn's disease and ulcerative colitis†
Article first published online: 14 DEC 2006
Copyright © 2001 Crohn's & Colitis Foundation of America, Inc.
Inflammatory Bowel Diseases
Volume 12, Issue 12, pages 1093–1100, December 2006
How to Cite
Tung, J., Loftus, E. V., Freese, D. K., El-Youssef, M., Zinsmeister, A. R., Melton, L. J., Harmsen, W. S., Sandborn, W. J. and Faubion, W. A. (2006), A population-based study of the frequency of corticosteroid resistance and dependence in pediatric patients with Crohn's disease and ulcerative colitis. Inflamm Bowel Dis, 12: 1093–1100. doi: 10.1097/01.mib.0000235835.32176.85
- Issue published online: 14 DEC 2006
- Article first published online: 14 DEC 2006
- Manuscript Accepted: 4 JUL 2006
- Manuscript Received: 29 JUN 2006
- National Institutes of Health. Grant Number: AR30582
- inflammatory bowel disease;
- Crohn's disease;
- ulcerative colitis;
Background: The goal of this study was to examine the 1-year outcome after the first course of systemic corticosteroids in an inception cohort of pediatric patients with inflammatory bowel disease. Methods: All Olmsted County (Minnesota) residents diagnosed with Crohn's disease (n = 50) or ulcerative colitis (n = 36) before 19 years of age from 1940 to 2001 were identified. Outcomes at 30 days and 1 year after the initial course of corticosteroids were recorded. Results: Twenty-six patients with Crohn's disease (65%) and 14 with ulcerative colitis (44%) were treated with corticosteroids before age 19. Thirty-day outcomes for corticosteroid-treated Crohn's disease were complete remission in 16 (62%), partial remission in 7 (27%), and no response in 3 (12%), with 2 of these patients requiring surgery. Thirty-day outcomes for treated ulcerative colitis were complete remission in 7 (50%), partial remission in 4 (29%), and no response in 3 (21%). One-year outcomes for Crohn's disease were prolonged response in 11 (42%) and corticosteroid dependence in 8 (31%), whereas 7 (27%) were postsurgical. One-year outcomes for ulcerative colitis were prolonged response in 8 (57%) and corticosteroid dependence in 2 (14%), whereas 4 (29%) were postsurgical. Conclusions: Most pediatric patients with inflammatory bowel disease initially responded to corticosteroids. However, after 1 year, 58% of pediatric patients with Crohn's disease and 43% of pediatric patients with ulcerative colitis either were steroid dependent or required surgery. This finding emphasizes the need for early steroid-sparing medications in pediatric inflammatory bowel disease.
Therapeutic management of inflammatory bowel disease (IBD) has evolved rapidly over the past several decades. Glucocorticoids, however, remain the medical intervention of choice for the acute treatment of moderate to severe Crohn's disease or ulcerative colitis (UC).
To date, only 2 population-based studies of the natural history of patients with IBD receiving an initial course of corticosteroids have been fully published, and both have focused on the adult patient population. Munkholm and colleagues1 described the frequency of steroid resistance and dependence in a Danish cohort with Crohn's disease in 1994. In 2001, Faubion et al2 described the 1-year outcome after the first course of steroids in adult patients with Crohn's disease and UC in Olmsted County (Minnesota). Both studies concluded that, although immediate response to corticosteroids is common, long-term corticosteroid dependence and surgery are frequent complications.1,2 The impact of these observational studies has been a trend to earlier use of immunosuppressive therapy in patients with moderate to severe IBD.3,4 However, neither study focused on the pediatric population, and there are several reasons to determine whether the natural history of corticosteroid response differs in pediatric IBD. Younger children with IBD tend to have more extensive disease, and at least 2 studies have shown mucosal differences between adult and pediatric UC.5–7 Children also have fewer environmental confounders, so genetic factors are more likely to play a role in the pathogenesis of IBD in this group. In addition, evidence exists that although T-cell responses may be similar between children and adults, there is decreased production of interleukin-12 in healthy children even up to 12 years of age.8 In this study, we report the population-based, 1-year natural history among an inception cohort of pediatric patients with IBD from Olmsted County after a first course of corticosteroid therapy.
Study Setting and Design
Olmsted County, situated in southeastern Minnesota, had a population of 124,277 inhabitants in the 2000 U.S. Census. The majority of people reside in Rochester, which is the urban center of an otherwise rural county. Eighty-nine percent are non-Hispanic white, and a substantial portion is of northern European heritage. Although 25% of county residents are employed in the healthcare field (versus 8% nationwide) and the level of education is correspondingly higher (30% of adults have completed college versus 21% nationwide), the residents of Olmsted County are otherwise socioeconomically similar to the U.S. white population.9
Medical care is practically self-contained within the region and is provided mostly by the Mayo Clinic, Olmsted Medical Center, and their affiliated hospitals. Because of a unique medical record linkage system (the Rochester Epidemiology Project),9 it is possible to conduct population-based research on IBD in this geographic region. A total of 692 Olmsted County residents (378 UC, 314 Crohn's disease) were first diagnosed with IBD during the period of 1940 to 2001, according to previously established well-defined criteria.10–12 A second search of the centralized diagnostic index crossing diagnostic codes for IBD, Crohn's disease, and UC with Olmsted County residence also was used. The present study was limited to the 89 patients who were diagnosed with IBD before 19 years of age and were county residents at the time of diagnosis. However, only 86 patients authorized the use of their medical records for research in accordance with Minnesota privacy statutes.13 After approval was given by the institutional review boards of Mayo Foundation and Olmsted Medical Center, the medical records were reexamined by a single reviewer (J.T.), and demographic characteristics, extent of disease, and concomitant medications were recorded. Of the remaining 86 patients, 40 (47%) had a received a first course of corticosteroids by their 19th birthday. These 40 patients (26 UC patients, 14 Crohn's disease patients) make up the population for this study.
Corticosteroid treatment response was assessed at 30 days after induction of therapy and at 1 year after induction of the first course of steroid therapy. Assessment criteria at each time point have been defined previously.1,2 Although the 2 previous studies had assessed steroid treatment response at 30 days after discontinuation, the wide range in follow-up visit intervals precluded a meaningful assessment of this time point in the present study. Concomitant medications with each course of corticosteroids and at 1 year were recorded. Because of the retrospective nature of this study, the cumulative steroid dose was not calculated.
Assessment at 30 days after steroid induction was defined as complete remission, partial clinical remission, nonresponse, or postsurgical remission as follows:
Complete remission: total regression of clinical symptoms within 30 days with 2 bowel movements a day and no pain, fever, or weight loss.
Partial clinical remission: regression of symptoms within 30 days with 4 bowel movements a day, less abdominal pain, and no fever or weight loss.
Nonresponse: no improvement within 30 days.
Postsurgical remission: total regression of clinical symptoms achieved through surgical resection.
Assessment at 1 year after induction of the first course of steroids was defined as prolonged response, corticosteroid dependent, or postsurgical remission as follows:
Prolonged response: maintenance of complete remission or partial response after treatment was completed.
Corticosteroid dependent: relapse within 30 days of cessation of treatment or relapse when dose reduction was attempted.
Postsurgical remission: total regression of clinical symptoms achieved through surgical resection.
The proportion of patients (stratified by IBD subtype) responding at 30 days (complete remission, partial remission, or no response) and at 1 year (prolonged, dependent, and postsurgical) was estimated, and 95% confidence intervals (CIs) were calculated through the use of exact binomial distribution. Logistic regression models were used to assess the associations between predictor variables and 30-day response to an initial course of corticosteroids, defined as “no response or surgery” versus “partial or complete remission.” Predictor variables assessed included sex, age at diagnosis, concomitant medication, and interval from IBD diagnosis to start of first corticosteroid course. Each model included the subtype of IBD as a covariate, along with a single predictive variable, except when disease extent was assessed; then, the IBD subtype was assessed separately. These models are referred to as univariate models in the remainder of this article. A multiple variable logistic regression analysis was not undertaken because of the limited number of patients (n = 6) in the “no response” category at 30 days. One-year response was assessed in the same manner as the 30-day remission. Sustained response was defined as having no surgery by 1 year and not being treated with prednisone at 1 year.
The cumulative probability of surgery in the year after the initiation of the first course of corticosteroids was estimated separately for the 2 IBD subtypes with the Kaplan-Meier survival method. The cumulative probability of surgery also was estimated by stratifying patient response to their initial 30 days of corticosteroid treatment (complete, partial, and no response/surgery). For this assessment, follow-up was started on day 31 of the first corticosteroid course. The need for surgery within this first year was assessed for all patients with at least 1 year of follow-up and at a median duration of 329 days (range, 93 to 362 days) in the 13 patients (10 Crohn's disease, 3 UC) with <1 year of follow-up. A 2-sided α level of 0.05 was used for statistical significance.
Between 1940 and 2001, 86 Olmsted County residents were diagnosed with IBD (36 UC, 50 Crohn's disease) before 19 years of age and were available for study. In the same time period, a total of 692 Olmsted County residents were diagnosed with IBD (378 UC, 314 Crohn's disease). Thus, 12% of Olmsted County residents with IBD (10% of UC, 16% of Crohn's disease) were diagnosed as children (<19 years of age). Fourteen of the 36 patients with UC (39%) and 26 of 50 patients with Crohn's disease (52%) received their first treatment course with systemic corticosteroids before the age of 19. Of the initial 86 patients with IBD diagnosed before 19 years of age, 38 (44%) were older than 16 years of age at diagnosis. Of those who received corticosteroids before 19 years of age, 50% of the patients with ulcerative colitis (n = 7) and 38% of the patients with Crohn's disease (n = 10) were older than 16 years of age at diagnosis. Clinical and demographic characteristics of these patients, including sex, race, age at diagnosis, site of disease, disease duration at first corticosteroid therapy, and concomitant therapy, are shown in Table 1.
|UC (n = 14)||Crohn's Disease (n = 26)|
|Male sex,no.(%)||10 (71)||16 (62)|
|White,no.(%)||12 (86)*||26 (100)|
|Median age at diagnosis (range), y||16.2 (1.7–18.3)||15.2 (8.4–18.8)|
|Median duration of disease at first corticosteroid therapy (range), mo||4.7 (0–29.6)||0.18 (0–25.6)|
|Disease site, no.(%)|
|Concomitant medications, no.(%)|
|None||1 (7)||4 (15)|
|Sulfasalazine||9 (64)||12 (46)|
|Mesalamine||5 (33)||7 (27)|
|6-mercaptopurine||1 (7)||6 (23)|
|Antibiotics||3 (20)||9 (35)|
Immediate (30-day) outcomes for Crohn's disease were complete remission in 16 (62%; 95% CI, 41 to 80), partial remission in 7 (27%; 95% CI, 12 to 48), no response in 1 (4%; 95% CI, 0.1 to 20), and surgery in 2 (8%; 95% CI, 1 to 25) (Fig. 1). Immediate outcomes for UC were complete remission in 7 (50%; 95% CI, 23 to 77), partial remission in 4 (29%; 95% CI, 8 to 58), and no response in 3 (21%; 95% CI, 5 to 51). There was no significant association between type of IBD and immediate outcome classified as complete/partial remission versus no response (P = 0.41).
One-year outcomes for Crohn's disease were prolonged response in 11 (42%; 95% CI, 23 to 63), corticosteroid dependence in 8 (31%; 95% CI, 14 to 52), and surgical resection in 7 (27%; 95% CI, 12 to 48) (Fig. 2). One-year outcomes for UC were prolonged response in 8 (57%; 95% CI, 29 to 82), corticosteroid dependence in 2 (14%; 95% CI, 2 to 43), and surgical resection in 4 (29%; 95% CI, 8 to 58). One patient with ulcerative colitis required a second prolonged course of corticosteroids but was steroid free at year-end; this was classified as a prolonged response. There was no significant association between type of IBD and 1-year outcome classified as prolonged response versus corticosteroid dependence or surgical resection (P = 0.37).
Predictors of Response
The univariate associations of gender, age at diagnosis, disease duration, disease extent, and concomitant medication use with 30-day outcomes (“no response” versus “complete or partial” response) are reported in Table 2. None of these variables was significantly associated with remission. Disease extent was not assessed in this analysis because of the small number of nonresponse patients (n = 6). The same predictor variables, including disease extent, also were examined for association with nonresponse at 1 year, and again there were no significant associations (Table 2).
|Immediate Outcome*||1-Year Outcome$|
|Variables||P||OR (95%CI)||P||OR (95%CI)|
|Male||0.35||2.9 (0.3–28.3)||0.74||1.3 (0.3–4.7)|
|Female||1.0 (Reference)||1.0 (Reference)|
|Age at diagnosis†||0.33||1.2 (0.8–1.78)||0.12||1.19 (0.95–1.49)|
|Disease duration†||0.82||1.01 (0.90–1.14)||0.21||1.08 (0.96–1.20)|
|Any medications†||0.65||0.57 (0.1–6.6)||0.24||0.25 (0.03–2.55)|
Cumulative Risk of Surgery
By 1 year after initiation of corticosteroids, 7 Crohn's disease and 4 UC patients had undergone surgical resection. Two of the patients with Crohn's disease had undergone surgery within the first 30 days of corticosteroid induction. These patients were not included in the assessment of surgery within 1 year (“late surgery”), leaving a total of 5 patients with a “late” surgery. The rates of surgery classified by a patient's immediate outcome to corticosteroid therapy are given in Table 3. The actuarial risk of resection after the initial course of corticosteroids among Crohn's disease patients was 27% at 1 year (95% CI, 12 to 48) (Fig. 3); nonresponders and partial responders at 30 days seem to be somewhat more likely than the complete responders to require surgical resection by the end of 1 year (33%, 31%, and 13%, respectively; Table 3). For UC patients, the cumulative risk was 29% by 1 year (95% CI, 0.5 to 50). Among the UC patients, the nonresponders at 30 days seemed to be somewhat more likely to require surgical resection by the end of 1 year (67%) than did either the partial (25%) or complete (25%) responders.
|Crohn's Disease (n = 26)||UC (n = 14)|
|Surgeries no.||Cumulative, Probability, %(95% CI)||Surgeries, no.||Cumulative, Probability, %(95% CI)|
|All patients*||7||27(12–48)||4||9 (1–50)|
|Outcome at 30 d after first course of corticosteroids$|
|Complete response†||2||13 (0–32)||1||25 (0–61)|
|Partial response||2||31 (0–71)||1||25 (0–61)|
|No response||1||33 (0–70)||2||67 (0–93)|
In this population-based inception cohort of pediatric patients with IBD, 52% of children with Crohn's disease and 39% of those with UC received systemic corticosteroids before 19 years of age. Of those who received corticosteroids, 89% of patients with Crohn's disease and 79% of patients with UC initially had complete or partial remission. Only 12% of patients with Crohn's disease and 21% with UC had no response or required surgery within 30 days of starting corticosteroids. However, long-term rates of prolonged response were low. Within 1 year after starting systemic corticosteroids, 27% of patients with Crohn's disease and 29% of patients with UC required surgery, and an additional 31% of Crohn's disease and 14% of UC were steroid dependent.
The main strengths of our study are its population-based nature and the availability of detailed medical records to allow retrospective assessment of medication response. Our study has several weaknesses. First, although the study used a population-based inception cohort, the size of the cohort was small. Second, of the 89 pediatric patients with IBD who met criteria, 3 refused research authorization, and 46 had not received corticosteroids before age 19, leaving 40 patients who received their first course of corticosteroids during childhood. Thus, because of the small event rate and potential selection bias, the true rates of corticosteroid responsiveness, dependence, and resistance may differ. However, our data are generally consistent with the rates reported in population-based inception cohort studies in adults and those reported in cohort and registry studies from pediatric tertiary care centers (Table 4).
|30-d Outcome, %||1-y Outcome, %|
|Crohn's Disease||Required Steroids, %||Complete Response||Partial Response||Nonresponse||Prolonged Response||Steroid Dependent||Surgery, %|
|Olmsted County-pediatric (n = 26/40)||52||62||27||12||46||31||27|
|Copenhagen1 (n = 109/196)||56||48||32||20||48||39||21|
|Olmsted County-adult2 (n = 74/173)||43||58||26||16||32||28||38|
|Toronto Sick Children's16 (n = 100)||85 in 1 y||Assessed after 1 course of steroids: 14 had complete remission, 68 required no further steroids.||12 over 2 y|
|Pediatric IBD Registry17 (n = 109/*)||*||59$||29$||12$||7|
|Olmsted County-pediatric (n = 14/32)||39||50||29||21||57||14||29|
|Olmsted County-adult2 (n = 63/185)||34||54||30||16||49||22||29|
|Pediatric IBD Registry18 (n = 54/69)†||§||56||40||4||26|
At first glance, it may appear that pediatric-onset IBD in our regional cohort is infrequent (10% of all UC cases, 16% of all Crohn's disease cases); however, as a percentage of all IBD cases, our results are similar to other North American population-based cohorts. Kugathasan et al14 reported a pediatric-onset Crohn's disease incidence of 4.56 per 100,000 children in Wisconsin between 2000 and 2001. However, information regarding incidence or prevalence of adult IBD was not provided, so the age distribution of IBD incidence could not be calculated. Bernstein et al15 reported a Crohn's disease prevalence of 198.5 per 100,000 persons between 1989 and 1994 in a population-based cohort in Manitoba, Canada. In this cohort, only 4% (93 of 2268 total cases) of prevalent cases were children (<19 years of age). During the same time period, the average annual incidence of Crohn's disease in children was quite high at 13.4 per 100,000 person-years. However, children made up only 13% of the incident cases. Thus, the small numbers of pediatric-onset IBD cases in our study are a reflection of the small catchment area, not because the incidence or prevalence of pediatric-onset IBD differs from that reported elsewhere in North America.
Although no population-based data regarding the natural history of pediatric corticosteroid response exist, our data are similar to studies conducted at pediatric tertiary care centers (Table 4). The Hospital for Sick Children (Toronto, Canada) reported a referral center cohort of 100 consecutive, newly diagnosed children with Crohn's disease.16 Eighty-five percent of newly diagnosed patients received corticosteroids within a year of diagnosis. Although differences in definition do not allow direct comparison between the 2 studies, 26% of patients required at least 2 courses of corticosteroids (compared to our 31% steroid dependence rate), and the rate of surgery over 2 years was 12% (compared to our 27%).16 The Pediatric IBD Registry, an observational registry formed in 2000 at 20 centers in the United States and Canada, enrolled newly diagnosed patients with IBD and prospectively collected outcome data.17 The total number of patients with Crohn's disease enrolled in the registry was not reported. One hundred nine patients with Crohn's disease in the registry were treated with corticosteroids, of whom 12% failed to respond within 30 days and 7% underwent surgical resection.17 Sixty-nine patients with a diagnosis of UC in the registry were followed for 1 year after treatment with corticosteroids.18 Fifty-four (78%) required treatment with corticosteroids within 1 year, of whom 4% failed to respond within 30 days. The surgery rate was not reported.18 Our prevalence of steroid dependence in patients with UC at 1 year after initiation of corticosteroids (14%) is lower than that of the Pediatric IBD Registry (26%), despite some use of concomitant immunosuppressive therapy in the latter cohort. The differences between these studies may be attributable to differences in age, practice patterns, and cohort size.
These results are generally comparable to those reported in adult patients with Crohn's disease and UC in Olmsted County and in patients with Crohn's disease in Copenhagen County (Denmark) (Table 4).1,2 Our population-based study largely predates the introduction of immunosuppressive therapies, including azathioprine, 6-mercaptopurine, methotrexate, cyclosporine, and infliximab. Post hoc analysis of our study shows that of 6 Crohn's disease patients treated with azathioprine or 6-mercaptopurine during the year of initiation of corticosteroids (methotrexate, cyclosporine, and infliximab were not used in this cohort), none required surgery within 1 year of initiation of steroid therapy, 50% had prolonged response, and 50% were steroid dependent. Likewise, all 3 UC patients in our study treated with azathioprine or 6-mercaptopurine during the year of steroid therapy had prolonged responses at 1 year (none received an immunosuppressive agent with the initial course of steroids).
In conclusion, our population-based inception cohort study demonstrates that <60% of pediatric-onset patients with Crohn's disease and <40% of patients with early-onset UC require treatment with corticosteroids before adulthood. Among patients who required corticosteroids, 89% of patients with Crohn's disease and 79% of patients with UC had complete or partial remission initially. However, 1 year later, 58% of patients with Crohn's disease and 43% with UC were steroid dependent or required surgery. Thus, corticosteroid requirement in children is a predictor of relatively poor 1-year outcome. Follow-up studies are needed to determine whether concomitant immunosuppressive and biological therapy will improve 1-year outcomes.
The authors wish to thank Debra A. Jewell, RN, for her assistance in abstracting data.
- 3Ulcerative colitis practice guidelines in adults (update): American College of Gastroenterology, Practice Parameters Committee. Am J Gastroenterol. 2004; 99: 1371–1385., .Direct Link:
- 17Acute and 1 year outcome of corticosteroid therapy in newly diagnosed children with Crohn's disease: the multicenter experience of the Pediatric IBD Collaborative Research Group. Gastroenterology. 2005; 128: W1043. Abstract., , , et al.
- 18The natural history of corticosteroid therapy for ulcerative colitis in children. Gastroenterology. 2005; 128: W1157. Abstract., , , et al.