Objectives To determine time to diagnosis in a paediatric inflammatory bowel disease (IBD) cohort and the relative contribution of the component intervals, and to identify factors associated with diagnostic delay.
Design Prospective cohort study
Setting Single-centre study including children with incident IBD at the Hospital for Sick Children diagnosed between December 2013 and December 2015.
Interventions Time to diagnosis and its subintervals were determined and patient, disease and institutional factors were tested for associations.
Results Among 111 children, the median overall time to diagnosis was 4.5 (IQR 2.1–8.8) months. Time to diagnosis was longer in Crohn’s disease (CD) than ulcerative colitis (UC) (median 6.8 (IQR 2.9–12.5) vs 2.4 (IQR 1.3–5.3) months) and patients with isolated small bowel disease. Twenty per cent of patients were diagnosed≥1 year after symptom onset (86% CD, 14% UC, p=0.003). Time from symptom onset to gastroenterology referral was the greatest contributor to overall time to diagnosis (median 2.9 (IQR 1.6–8.2) months). Height impairment was independently associated with diagnostic delay (OR 0.59, p=0.02, for height-for-age z-score (HAZ), signifying almost 70% increased odds of delay for every 1 SD decrease in HAZ). This height discrepancy persisted 1 year after diagnosis. Bloody diarrhoea was protective against delay (OR 0.28, p=0.02). The subinterval from referral to diagnosis was shorter in patients with laboratory abnormalities, particularly hypoalbuminaemia.
Conclusions Diagnostic delay was more common in CD and associated with height impairment that persisted 1 year after presentation. The greatest contributor to time to diagnosis was time from symptom onset to referral.
- general paediatrics
Statistics from Altmetric.com
What is already known on this topic?
Diagnostic delay is associated with an increased risk of adverse outcomes in paediatric inflammatory bowel disease (IBD), including growth impairment
Time to diagnosis has been shown to be longer in children with Crohn’s disease and small bowel involvement
Uncertainty exists regarding the relative contribution of the various time intervals that make up overall time to diagnosis
What this study adds?
Twenty per cent of children, particularly with Crohn’s disease, continue to experience delays greater than 1 year between symptom onset and IBD diagnosis
Linear growth impairment is observed in children with delayed IBD diagnosis even 1 year after presentation
Time from symptom onset to referral is the primary contributor to time to diagnosis and primary care physicians are therefore crucial to minimising diagnostic delay
Inflammatory bowel disease (IBD), including paediatric-onset disease, is increasing worldwide.1 2 Canada has one of the highest rates of IBD in the world,1 3 and the incidence is rising most rapidly in children.3 Timely diagnosis is central to optimal IBD care given the association between diagnostic delay and disease outcomes.4–7 The matter of diagnostic delay is even more salient in paediatric IBD given the unique potential for linear growth impairment.8 In light of the globally rising incidence of IBD and the changing age demographic, paediatricians have an increasingly critical role to play in optimising IBD outcomes through early disease detection.
In Canada, primary paediatric care is provided by general paediatricians and family doctors. Childhood-onset IBD is diagnosed endoscopically by a paediatric gastroenterologist (GI) following referral by a primary care physician. The overall time to diagnosis consists of distinct intervals. These include time from symptom onset to contact with a physician who initiates a GI referral, and time from referral to diagnostic endoscopy, which itself consists of two intervals, time from referral to GI consultation and time from consultation to diagnostic endoscopy. There is a paucity of data regarding the relative contribution of each of these intervals to the overall time to diagnosis in children and the factors that affect them. Furthermore, prior studies have focused on factors, such as IBD type and disease location, not known at the time of presentation and thus not amenable to intervention. No study to date has considered institutional factors in relation to diagnostic delay.
We sought to determine the overall time to diagnosis and the relative contribution of the subintervals using a well-characterised, prospectively collected incident paediatric IBD cohort. We further aimed to identify patient, disease and/or institutional factors associated with diagnostic delay.
Setting and participants
The Canadian Children Inflammatory Bowel Disease Network, a joint partnership of the Foundation for Children with Intestinal & Liver Disorders and the Canadian Institutes of Health Research, is a national IBD network that includes 12 major paediatric IBD centres. Children aged ≤17 years and 2 months are prospectively recruited at diagnosis and followed over time. This study included inception cohort participants diagnosed and recruited at the Hospital for Sick Children (Sick Kids) in Toronto, the Network’s central coordinating site, between 1 December 2013 and 31 December 2015. One hundred and eighty children were diagnosed with IBD over the study period. Thirty-three were not eligible for network inclusion, largely due to treatment having been previously initiated. Thirty-six patients declined participation. One hundred and eleven children were therefore included in this study.
Comprehensive baseline data are collected at first patient contact, including presenting symptoms and signs, and the date the patient was last well. At Sick Kids, data are elicited and recorded by one of three IBD physicians onto standardised forms and entered in a central data registry. Patient and disease characteristics for the study were retrieved from this database. Hospital data sources were queried to ascertain institution level factors.
The intervals examined in this study were overall time to diagnosis (time from symptom onset to diagnostic endoscopy), time from symptom onset to GI referral, time from referral to diagnosis, time from referral to GI consultation and time from consultation to diagnosis. Diagnostic delay was defined as an overall time to diagnosis greater than the upper quartile. Subintervals were similarly defined as prolonged if they exceeded the upper quartile. Patient factors studied included age, sex, medical comorbidities (categorised into immune mediated and non-immune mediated), family history of IBD (defined as IBD in a first degree relative), family income (based on neighbourhood, obtained by linking postal codes to 2006 Canadian census data),9 10 disease activity assessed using the weighted Paediatric Crohn’s Disease Activity Index (wPCDAI)11 or Paediatric Ulcerative Colitis Activity Index (PUCAI)12 (categorised as none, mild, moderate or severe per established cut-offs), presenting symptoms and signs (listed in table 1) and anthropometrics. Extraintestinal symptoms included mucocutaneous and joint involvement. Anthropometric parameters included weight loss, body mass index (BMI)-for-age z-score and height-for-age z-score (HAZ). Institutional factors, shown in table 1, reflected both the busyness of the GI service and the capacity for diagnostic procedures. These variables were tested for associations with prolonged overall time to diagnosis and prolonged subintervals. In addition, disease details, including IBD type, disease location and behaviour, as per the paediatric Paris classification,13 were collected.
Continuous variables were summarised with medians and IQR and compared with the Mann-Whitney U test. Categorical variables were summarised with frequencies and proportions and compared with the χ2 test or Fisher’s exact test, as appropriate. Associations were tested using univariate (UVA) and multivariable (MVA) logistic regression. Wald backward selection was applied to variables with a p value <0.2 in UVA to build MVA models. Results were presented as OR with 95% CI. Times to diagnosis were compared with Kaplan-Meier curves and the log-rank test. A two-sided p value <0.05 was considered statistically significant. All analyses were performed using SAS Studio V.3.4 (SAS Institute, North Carolina, USA). Study approval was obtained from the Sick Kids Research Ethics Board.
Time to diagnosis
The median overall time to diagnosis was 137 (IQR 63–268) days, or 4.5 (IQR 2.1–8.8) months (minimum 3 days and maximum 4 years). Forty-one per cent of children were diagnosed within 3 months, 20% between 3 and 6 months, 19% between 6 and 12 months, 17% between 1 and 3 years and 3% beyond 3 years from symptom onset. The greatest contributor to the overall time to diagnosis was by far the interval between symptom onset and referral to GI by a primary care physician, with a median duration of 89 (IQR 48–250) days compared with a median of only 15 (IQR 7–45) days between referral and diagnostic endoscopy. The latter interval was fairly evenly split between time from referral to GI consultation (median 6 (IQR 0–16) days) and time from consultation to diagnosis (median 7 (IQR 5–27) days).
Patient, disease and hospital characteristics are displayed in table 1. A quarter of patients were diagnosed beyond the upper quartile of 8.8 months and made up the delayed diagnosis group. CD and isolated small bowel disease were more common in the delayed diagnosis group (85% vs 50%, p=0.001 and 23% vs 8%, p=0.03, respectively). Time to diagnosis was longer in Crohn’s disease (CD) (median 207 (IQR 87–379) days) than in ulcerative colitis (UC)/IBD-unclassified (IBD-U) (median 73 (IQR 40–160) days), as illustrated in figure 1 (log-rank p<0.001). Of patients diagnosed beyond 1 year, 86% had CD and 14% had UC/IBD-U (p=0.003). Conversely, of patients diagnosed within 2 months, 73% had UC/IBD-U and 27% CD (p<0.001). All subintervals were significantly longer in CD than UC/IBD-U patients, except time from referral to GI consultation (p=0.19). Diarrhoea and blood per rectum were each present in greater than three quarters of patients diagnosed promptly, and bloody diarrhoea occurred in 65% of this group (table 1). Intestinal symptoms were less common in the delayed diagnosis group (bloody diarrhoea in 32%, p=0.003). Fever and weight loss were more common in children diagnosed promptly (28% vs 8%, p=0.04 and 76% vs 54%, p=0.04, respectively). Patients with diagnostic delay had greater linear growth impairment at presentation, as reflected by significantly lower HAZ (−0.5 [IQR −1.5−0.2) vs 0.1 (IQR −1.0–0.9], p=0.007). Importantly, a significant height discrepancy persisted between groups even 1 year after diagnosis (HAZ −0.6 (IQR −1.6–0.09) vs −0.1 (IQR −0.9–0.6), p=0.04). Patients with diagnostic delay had less active disease based on lower wPCDAI (36 (IQR 18–60) vs 55 (IQR 35–68), p=0.04) and PUCAI scores (20 (IQR 5–35) vs 40 (IQR 20–60), p=0.001) and a lower proportion requiring steroid (33% vs 56%) or anti-tumour necrosis factor (TNF) (7% vs 15%) induction (p=0.02). No differences were observed for comorbidities, family history of IBD, laboratory values or institutional factors.
Excluding GI consultation, 62% of children presented to Sick Kids for IBD-related symptoms at least once prior to diagnosis, generating a total of 100 hospital visits, 84% of which were to the emergency department.
Factors associated with diagnostic delay
The results of UVA and MVA logistic regression examining factors known at time of presentation in relation to overall time to diagnosis are shown in table 2. Diarrhoea (OR 0.29, 95% CI 0.11 to 0.74), blood per rectum (OR 0.33, 95% CI 0.13 to 0.85), bloody diarrhoea (OR 0.25, 95% CI 0.10 to 0.65) and weight loss (OR 0.38, 95% CI 0.15 to 0.99) were significantly associated with decreased odds of delayed diagnosis in UVA, while height impairment was associated with increased odds of diagnostic delay (OR 0.60, 95% CI 0.42 to 0.88 for HAZ, signifying almost 70% increased odds of delay for every one SD decrease in HAZ). In MVA, bloody diarrhoea (OR 0.28, 95% CI 0.10 to 0.84) and linear growth impairment (OR 0.59, 95% CI 0.38 to 0.92) remained independently associated with time to diagnosis.
A quarter of patients experienced a delay greater than 250 days between symptom onset and referral and comprised the prolonged time to referral group. Only height impairment was associated with delayed referral, in both UVA and MVA analysis (OR 0.48, 95% CI 0.31 to 0.75 for HAZ in multivariable analysis, signifying a twofold increased odds of delayed time to referral for every 1 SD decrease in HAZ) (table 3).
In 25% of cases, the time from GI referral to diagnosis exceeded 45 days and these patients comprised the delayed referral-to-diagnosis group. More active diseases (OR 0.45, 95% CI 0.28 to 0.71), intestinal symptoms (OR 0.26, 95% CI 0.10 to 0.68 for diarrhoea and OR 0.27, 95% CI 0.10 to 0.72 for bloody diarrhoea), abdominal pain (OR 0.34, 95% CI 0.12 to 0.94), fever (OR 0.11, 95% CI 0.01 to 0.87) and higher erythrocyte sedimentation rate (ESR) (OR 0.97, 95% CI 0.95 to 0.99) were associated with decreased odds of delay in UVA, while higher albumin (OR 1.22, 95% CI 1.11 to 1.34) and haemoglobin (OR 1.05, 95% CI 1.02 to 1.09) were associated with increased odds of delay. In MVA, only bloody diarrhoea (OR 0.13, 95% CI 0.04 to 0.48) and albumin (OR 1.26, 95% CI 1.11 to 1.42) retained a significant association (table 4).
In this prospective cohort of 111 paediatric IBD patients, the median time to diagnosis was 4.5 months. A fifth of children were diagnosed greater than 1 year after symptom onset. The time to diagnosis in our Canadian cohort is comparable to that in two large, more recent European paediatric studies, which reported median delays of 4 and 5 months.14 15 Consistent with previous reports,14–17 we found an independent association between delayed diagnosis and linear growth impairment. Importantly, the height discrepancy persisted even 1 year after diagnosis, suggesting that this loss in height is not quickly recovered and may in fact be irreversible. Early IBD diagnosis in children is therefore key to maximise growth potential. Diagnostic delay is also relevant given its association with adverse outcomes, such as bowel stenosis, fistulising disease and decreased quality of life.4–7
In our study, the interval between symptom onset and GI referral by a primary care physician was by far the greatest contributor to overall time to diagnosis. This highlights the pivotal role of paediatricians and family doctors in minimising diagnostic delay in children presenting with IBD. This contrasts with the only other study to date to examine intervals, which found time from gastroenterologist contact to diagnosis to account for the largest part of time to diagnosis.18 This discrepancy may stem from differences in the healthcare system in Germany and Austria, compared with Canada.
Our study is the first to restrict the analysis of potential modifiers of diagnostic delay to factors known at the time of presentation and the first to consider institutional factors. Reassuringly, time to diagnosis was not influenced by institutional factors, such as the busyness of the GI service or the functional capacity of the endoscopy suite. Bloody diarrhoea was independently associated with decreased odds of diagnostic delay, likely reflecting the rapidity with which such an overtly abnormal symptom is recognised.
Our study is also the first to test factors for associations with the individual subintervals making up time to diagnosis. Linear growth impairment was significantly associated with a prolonged time to referral. The subsequent interval from referral to diagnosis was more likely to be shorter in patients with bloody diarrhoea and lower albumin. This suggests that IBD physicians, the primary drivers of this interval, are more highly influenced by markers of acute, severe disease. No association was found between diagnostic delay and any of sex, age, comorbidities, family history of IBD and average family income. In contrast, several studies have reported a link between diagnostic delay and younger age.14 15 19 The lack of association in our study may relate to the increasing incidence and awareness of very early onset IBD in Canada. Between 1999 and 2008, the incidence of IBD in children younger than 10 years in Ontario, Canada rose by 10% per year.3 In the entire Canadian inception cohort of now close to 1000 patients, 25% are aged 10 years or younger at diagnosis.
We observed a longer time to diagnosis in CD (median 6.8 months) versus UC/IBD-U (median 2.4 months), and in patients with isolated small bowel disease. Strikingly, 86% of patients diagnosed beyond 1 year had CD. This is consistent with previous studies,14–16 18 and likely reflects the lower prevalence of overt symptoms like bloody diarrhoea, particularly in small bowel CD.
The strengths of our study include the completeness and thoroughness of data and its prospective nature with data collection at the time of IBD diagnosis by a small group of IBD physicians, minimising variability. This contrasts with the two largest paediatric diagnostic delay studies to date, of which one was survey based,15 and the other allowed patient inclusion up to 1 year after diagnosis.14
There are limitations as well, including the study’s relatively small size, although the congruence of several key findings with earlier, larger paediatric studies is reassuring. Additional limitations include the fact that dates of symptom onset were based on patient and caregiver recall. However, these were corroborated by objective data like growth charts, laboratory findings and referral information.
Improving time to diagnosis in children with IBD remains an important goal of care, particularly given the link between delayed diagnosis and linear growth impairment. It is important to increase awareness of the potential for paediatric IBD to present insidiously without overt intestinal symptoms with an isolated height deficit.
To help minimise diagnostic delay, practitioners should have a low threshold for investigating children with non-specific symptoms, including linear growth impairment, for possible IBD. Such patients should be investigated with (1) medical history, including family history of IBD or other immune mediated disorders; (2) physical exam, including attention to anthropometrics and the perianal region; and (3) investigations, including complete blood count, albumin, C reactive protein, ESR and stool culture.20 If available, faecal calprotectin, a stool biomarker of intestinal inflammation, should be obtained. Values >50 µg/g are highly sensitive, but less specific, for an intestinal inflammatory process and warrant further investigation. Values <50 µg/g are useful for ruling out IBD.21 Children with suspected IBD should be referred promptly to a paediatric gastroenterologist. Investigations performed and growth concerns should be explicitly conveyed in the referral.
Contributors AR: Study design; data collection; analysis and interpretation of data; drafting of the manuscript; critical revision of the manuscript. JF: Data collection; critical revision of the manuscript. DET: Data collection; critical revision of the manuscript. NC: Data collection; critical revision of the manuscript. EC: Data collection; critical revision of the manuscript. CP: Data collection; critical revision of the manuscript. AM: Data collection; critical revision of the manuscript. TDW: Data collection; interpretation of data; critical revision of the manuscript. AMG: Data collection; interpretation of data; critical revision of the manuscript. PCC: Study concept and design; data collection; analysis and interpretation of data; critical revision of the manuscript.
Funding The Canadian Children IBD Network inception cohort study is funded by CH.I.L.D Foundation and CIHR.
Competing interests None declared.
Ethics approval Hospital for Sick Children Research Ethics Board.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement All study data are available for sharing to investigators of the CIDsCaNN network (Canadian paediatric IBD network) through Red Cap. The participating CIDsCaNN sites include: BC Children’s Hospital, Alberta Children’s Hospital, Stollery Children’s Hospital, Children’s Hospital of Eastern Ontario, Health Sciences Centre Winnipeg, IWK Health Centre, London Health Sciences Centre Children’s Hospital, CHU Sainte-Justine, Montreal Children’s Hospital, McMaster Children’s Hospital, Janeway Children’s Hospital and the Hospital for Sick Children.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.