Systematic evaluation of risk factors for diagnostic delay in inflammatory bowel disease

Abstract

Background:

The diagnosis of inflammatory bowel disease (IBD), comprising Crohn's disease (CD) and ulcerative colitis (UC), continues to present difficulties due to unspecific symptoms and limited test accuracies. We aimed to determine the diagnostic delay (time from first symptoms to IBD diagnosis) and to identify associated risk factors.

Methods:

A total of 1591 IBD patients (932 CD, 625 UC, 34 indeterminate colitis) from the Swiss IBD cohort study (SIBDCS) were evaluated. The SIBDCS collects data on a large sample of IBD patients from hospitals and private practice across Switzerland through physician and patient questionnaires. The primary outcome measure was diagnostic delay.

Results:

Diagnostic delay in CD patients was significantly longer compared to UC patients (median 9 versus 4 months, P < 0.001). Seventy-five percent of CD patients were diagnosed within 24 months compared to 12 months for UC and 6 months for IC patients. Multivariate logistic regression identified age <40 years at diagnosis (odds ratio [OR] 2.15, P = 0.010) and ileal disease (OR 1.69, P = 0.025) as independent risk factors for long diagnostic delay in CD (>24 months). In UC patients, nonsteroidal antiinflammatory drug (NSAID intake (OR 1.75, P = 0.093) and male gender (OR 0.59, P = 0.079) were associated with long diagnostic delay (>12 months).

Conclusions:

Whereas the median delay for diagnosing CD, UC, and IC seems to be acceptable, there exists a long delay in a considerable proportion of CD patients. More public awareness work needs to be done in order to reduce patient and doctor delays in this target population. (Inflamm Bowel Dis 2012;)

Crohn's disease (CD), ulcerative colitis (UC), and indeterminate colitis (IC) represent the three subtypes of inflammatory bowel diseases (IBD). Diagnosing IBD can be a considerable challenge, especially in cases with mild clinical activity due to the overlap of symptoms with functional diseases. Indeed, symptoms similar to irritable bowel syndrome (IBS) are frequently reported in patients before IBD is diagnosed.1 Furthermore, the diagnostic value of the Manning and Rome criteria for diagnosing IBS is limited because of the moderate sensitivity.2, 3 The increasing number of publications evaluating various biomarkers to differentiate IBD from IBS is triggered by these diagnostic limitations.4 These diagnostic difficulties may partially explain the presence of a delay from first IBD symptom onset until the diagnosis is finally established. Obviously, diagnostic delay represents an issue in IBD patients in many countries, considering the literature reporting on patients suffering from symptoms for many years prior to IBD diagnosis.5–8 Diagnostic delay also has important clinical impact, as there is increasing evidence demonstrating that treatment success is increased in early disease.9–13

Diagnostic delay in IBD may potentially be attributed to patient characteristics or physician unawareness. In Switzerland, the vast majority of patients first address their complaints to the family physician and only a minority primarily consults a gastroenterologist (time interval from first symptoms to the physician visit). Once the patient is referred to the physician, the IBD diagnosis can be established (time interval from physician visit to diagnosis). Therefore, diagnostic delay should be examined at these two different stages in order to identify at which of them improvements might be possible (on the patients' or physicians' side).

Many publications have illustrated that criteria such as patient age, cigarette smoking status, or oral contraception use can modify the clinical presentation of IBD.14–16 Therefore, these disease-modifying items should also be analyzed when reporting on diagnostic delays in IBD.

This study aimed to evaluate the following questions. First, what is the diagnostic delay for CD and UC patients in Switzerland? Second, is there a difference between the time interval from first symptoms to physician visit (patient-related) and from physician visit to IBD diagnosis (physician-related)? And third, which factors are associated with diagnostic delay?

MATERIALS AND METHODS

Patients

Since 2006, IBD patients from all regions of Switzerland have been included in the nationwide clinical Swiss IBD Cohort Study that is supported by the Swiss National Science Foundation and approved by the local ethical committees.17 The data acquisition focuses on clinical, socioeconomic, and psychosocial data. Written informed consent is mandatory for inclusion in this cohort as well as permanent residence status in Switzerland or Swiss health insurance. At inclusion, patients undergo a thorough clinical and laboratory assessment. Detailed questionnaires focusing on physician-reported outcomes are completed by the IBD physicians, whereas patient-reported outcome questionnaires (assessing, e.g., quality of life, social impairment, symptom questionnaires) are completed by patients themselves. The following analysis evaluates all patients included in the SIBDCS from October 2006 until April 2009. Patients included in this study were recruited in the following setting of care: 62% university hospitals, 13% large nonuniversity hospitals, 6% county hospitals, 19% private practice.

In Switzerland, the vast majority of patients first address their family physicians with their complaints and later are referred to the gastroenterologist if IBD is suspected. Depending on the health insurance policy, patients may either be obliged to first consult the family physician or they are free to directly consult a gastroenterologist (which applies only to a minority of patients).

Methods

The data from patients and doctors were collected and validated at the datacenter at the Institute of Social and Preventive Medicine at the University of Lausanne. Rigorous rules were applied to ensure data quality.

Three relevant time intervals were assessed in patient questionnaires and cross-checked for consistency with the physician questionnaires. Figure 1 provides an overview of the study concept and illustrates these specific time spans.

Figure 1. Reporting of the time delays in the patient questionnaire. Recorded is interval 1, corresponding to the timespan from first IBD symptoms until consulting a physician (family physician or gastroenterologist), and interval 2, representing the timespan from physician visit to IBD diagnosis. Accordingly, the timespan from first IBD symptoms to IBD diagnosis can be counted. Disease onset was defined as the time when symptoms first appeared that were later confirmed to be related to IBD. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

The time intervals were defined as follows: 1) Time from first IBD symptoms to consultation with the physician (interval 1): This interval represents the time span between the first manifestations of IBD-related symptoms (patient-reported) and a consultation with the physician (family physician or gastroenterologist) specifically due to these IBD-related complaints. The length of this period is mainly dependent on the patient herself/himself. 2) Time from physician visit to IBD diagnosis (interval 2): This represents the time span between the IBD symptom-related consultation of the physician and the time IBD is diagnosed by a gastroenterologist. The length of this period is mainly dependent on the treating physicians (family physician and gastroenterologist). 3) Time from first IBD symptoms to IBD diagnosis (interval 1+2): This interval is calculated by the addition of interval 1 and interval 2.

Diagnostic delay was defined as the time span (in months) from first symptoms to IBD diagnosis. The following items in the patient questionnaires were assessed for the purpose of this study: “Before the IBD diagnosis, how long did you experience symptoms that are now attributed to IBD?” and “How long was the time interval between first symptoms and the first visit to a physician?” and “What was the time span from the first physician visit (due to these complaints) until IBD diagnosis was established?” Additionally, patients answered questions regarding smoking habits, intake of nonsteroidal antiinflammatory drugs (NSAIDs), or oral contraception at the time of diagnosis. The time intervals reported in patient questionnaires were compared to those reported by physicians. In case of discrepancies, issues were resolved by either chart review and/or contacting the patient and/or treating physician.

We further evaluated the relationship between disease-associated items (see below) and the diagnostic delay in IBD patients. The analyses were first performed by univariate and later multivariate logistic regression modeling. The dependent variable was diagnostic delay. As independent variables, we evaluated the following factors (including the binary description of the variable): gender (male versus female), age at IBD diagnosis (<40 versus ≥40 years), disease location (for CD: terminal ileum versus colon; for UC: rectum versus left-sided and extensive), IBD family history (present versus absent), extraintestinal manifestations (EIM) at IBD diagnosis (present versus absent), education level (basic versus higher, see additional explanations below), cigarette smoking at IBD diagnosis (yes versus no), intake of NSAIDs (yes versus no, intake defined as at least 1 tablet per week), oral contraception (yes versus no), and geographic area (urban versus rural). The following EIMs were assessed: peripheral arthritis, uveitis, pyoderma gangrenosum, erythema nodosum, aphthous stomatitis, ankylosing spondylitis, primary sclerosing cholangitis, and psoriasis. The presence of at least one EIM was considered EIM-positive.

The patient questionnaire recorded the following levels of education: 1 = no education at all, 2 = elementary school, 3 = high school, 4 = apprenticeship and college programs, 5 = university degree (undergraduate, masters, and higher degrees). As dichotomous outcome we defined 1+2 as “basic education” which we compared to items 3–5 defined as “higher education.”

Our hypotheses with respect to the diagnostic delay (from first symptoms to IBD diagnosis) were that IBD patients under 40 years of age have a longer diagnostic delay (frequent overlap with IBS symptoms) compared to older patients, that ileal CD has a longer diagnostic delay compared to colonic disease, and that a positive IBD family history as well as the presence of EIM are associated with shorter diagnostic delay. We further hypothesized that basic education level is associated with longer diagnostic delay, that cigarette smoking is associated with shorter diagnostic delay (due to more aggressive disease course), and that NSAID intake is associated with shorter diagnostic delay (more aggressive disease course expected). We expected that oral contraception is associated with shorter diagnostic delay in CD patients, and that rural area is associated with longer diagnostic delay.

Statistical Analysis

Clinical data were retrieved from the datacenter of the Swiss IBD Cohort Study at the University of Lausanne. The data were entered into a database (Microsoft, Access 2000). All statistical examinations were performed with a statistical package program (Stata, v. 9, College Station, TX). Data distribution was analyzed using Normal-QQ-Plots. Results of quantitative data are presented either as median plus interquartile ranges (for nonparametric data) or mean ± SD and range (parametric data). Categorical data were summarized as the percentage of the group total. The Kruskal–Wallis test was used to analyze nonparametric quantitative data and to evaluate whether there was a difference in diagnostic delay between the distinct IBD groups (CD, UC, IC). Time delays were then further evaluated by pairwise comparisons using the Wilcoxon rank sum test with Bonferroni adjustment. P < 0.05 was considered statistically significant. Logistic regression analysis evaluating the diagnostic delay with potential associated factors was performed applying stepwise regression modeling. The dependent variable was long diagnostic delay (defined according to the time interval when 75% of the CD or UC patients had been diagnosed); the independent variables are described in the Methods section. In a first step, the potential risk factors were each tested separately. In a second step, all risk factors with P < 0.2 were entered together in a multivariate logistic regression model.

RESULTS

Clinical Characteristics of the Study Population

The clinical characteristics of the study populations are shown in Table 1. A total of 1591 patients were included. Disease location followed the Montreal Classification for CD, UC, and indeterminate colitis.18 Disease activity at inclusion was evaluated in CD patients according to the Crohn's Disease Activity Index (CDAI) and in UC patients according to the modified Truelove and Witts Index (MTWSI). In CD patients, a score ≥150 and in UC patients ≥10 describes active disease.19, 20

Table 1. Clinical Characteristics of the Patient Cohort

	CD	UC	IC
CD location (according to the Montreal Classification): L1 = terminal ileum; L2 = colon; L3 = ileocolonic, L4 = isolated upper gastrointestinal tract disease. CD = Crohn's disease; UC = ulcerative colitis; IC = indeterminate colitis; SD = standard deviation; CDAI = Crohn's disease activity index; MTWSI = modified Truelove and Witts Score for UC and IC.
Number of patients	932	625	34
Female	496 (53%)	289 (46%)	15 (44%)
Age at enrolment: mean ± SD, range	41±15 (16-88)	42±14 (16-82)	40±14 (21-73)
Age at disease onset	30±13 (5-84)	33±13 (5-82)	35±14 (14-71)
CD location	L1: 202 (21.7%)	—	—
	L2: 234 (25.1%)
	L3: 444 (47.6%)
	L4: 52 (5.6%)
UC location	—	Rectum 94 (15%)	—
		Left sided 243 (38.9%)
		Extensive 288 (46.1%)
IC location	—	—	Rectum 2 (5.9%)
			Left sided 10 (29.4%)
			Extensive 22 (64.7%)
CDAI: mean ± SD, 95%-CI	85±69 (-14-463)		—
MTWSI: mean ± SD, 95%-CI	—	4±3.8 (0-19)	3.6±3.2 (0-13)
Smoking status	376/932 (40%)	157/625 (25%)	13/34 (38%)
NSAID intake	254/932 (27%)	132/625 (21%)	9/34 (26%)
Oral contraception (fem)	175/496 (35%)	87/289 (30%)	6/15 (40%)
Education level
Basic	499/932 (54%)	285/625 (46%)	16/34 (47%)
Higher	433/932 (46%)	340/625 (54%)	18/34 (53%)
Provenience
Urban	742/932 (80%)	491/625 (79%)	28/34 (82%)
Rural	190/932 (20%)	134/625 (21%)	6/34 (18%)

Diagnostic Delay Is Longer in CD Compared with UC and IC Patients

The analysis of the different time intervals is presented in Table 2. As a general observation, time delay from first symptoms to IBD diagnosis was significantly longer in CD patients compared to UC and IC patients. All assessed time intervals (from first symptoms to physician visit, from physician visit to diagnosis, and from first symptoms to IBD diagnosis) were significantly longer for CD compared to UC patients. We observed that when the time from first appearance of symptoms to the physician visit was shorter than from physician visit to IBD diagnosis, the difference was much more pronounced in CD (interval 2 nearly as twice as long as interval 1) compared to UC (interval 2 a little longer than interval 1).

Table 2. List of the Three Important Time Windows (in Months) According to the Diagnosis

	CD	UC	IC	P-value
Time from first IBD symptoms until visiting a physician (interval 1), time from physician visit to IBD diagnosis (interval 2) and the time span from first symptoms until IBD diagnosis (interval 1+2). As data are nonparametric, values are presented as median and interquartile range (IQR). The P-value indicates the significance levels of the time-windows according to the diagnosis (nonparametric tests). CD = Crohn's disease, UC = ulcerative colitis, IC = indeterminate colitis.
Number of patients	932	625	34
Time from first symptoms to IBD diagnosis (median, IQR)	9, 3-24	4, 1-12	3, 2-6	CD vs. UC: P < 0.001
				CD vs. IC: P < 0.001
				UC vs. IC: P = 0.441
Time from first symptoms to physician visit (median, IQR)	2, 0-6	1, 0-4	1, 0-2	CD vs. UC: P = 0.002
				CD vs. IC: P = 0.025
				UC vs. IC: P = 0.181
Time from physician visit to IBD diagnosis (median, IQR)	4, 0-18	1, 0-5	1, 0-4	CD vs. UC: P < 0.001
				CD vs. IC: P = 0.003
				UC vs. IC: P = 0.429

We further analyzed the data distribution stratified to diagnosis in order to define what percentage of IBD patients was diagnosed with the specific disease after symptom onset. These findings are further illustrated in Table 3. In total, 50% of CD patients were diagnosed within 9 months after symptom onset, compared to 4 months in UC patients and 3 months in IC patients. Seventy-five percent of CD patients were diagnosed within 24 months after symptom onset, compared to 12 months in UC patients and 6 months in IC patients. These data illustrate that the long time delay is mainly attributed to a smaller patient population with long to very long time delay between symptom onset and IBD diagnosis. Furthermore, there is hardly any physician delay in the majority of UC patients. This finding is in contrast to that observed in CD patients, where 50% are diagnosed within 4 months after first presentation to the primary physician, while the other 50% take considerably longer until CD diagnosis can be finally established.

Table 3. Time Delays (Months) in Patients with CD (n=932), UC (n=625), and IC (n=34)

Time Intervals (Months)	Time from First Symptoms to IBD Diagnosis			Time from First Symptoms to Physician Visit			Time from Physician Visit to IBD Diagnosis
Disease	CD	UC	IC	CD	UC	IC	CD	UC	IC
The listing of percentiles allows the readout of the percentage of patients diagnosed at specific time intervals.
Percentile
1%	0	0	0	0	0	0	0	0	0
5%	0	0	1	0	0	0	0	0	0
10%	1	1	1	0	0	0	0	0	0
25%	3	1	2	0	0	0	0	0	0
50%	9	4	3	2	1	1	4	1	1
75%	24	12	6	6	4	2	18	5	3
90%	96	36	12	18	12	6	60	23	6
95%	120	60	24	48	24	7	108	47	12
99%	252	120	36	180	72	24	240	114	35
Range	0-516	0-192	0-36	0-456	0-120	0-24	0-516	0-168	0-35

Impact of Disease-associated Factors on Diagnostic Delay: Definitions

The impact of disease-associated factors on diagnostic delay in different populations of IBD patients was further evaluated. Due to highly variable time intervals to IBD diagnosis (see Table 3), the definition of “long diagnostic delay” was chosen as the time interval when the 76th to 100th percentile of IBD patients were diagnosed. In CD patients, time intervals >24 months from symptoms to IBD diagnosis, >6 months from symptoms to physician visit, and >18 months from physician visit to IBD diagnosis were therefore defined as “long diagnostic delay.” In UC patients, time intervals >12 months from symptoms to IBD diagnosis, >4 months from symptoms to physician visit, and >5 months from physician visit to IBD diagnosis were defined accordingly as “long diagnostic delay.” Due to the small number of patients with indeterminate colitis, logistic regression analysis was performed in CD and UC patients only.

Impact of Disease-associated Items on Time Interval from First Symptoms to IBD Diagnosis in CD and UC Patients

The results on the univariate and multivariate logistic regression are further depicted in Table 4.

Table 4. Logistic Regression Modeling Evaluating “Long Diagnostic Delay” from First Symptoms to IBD Diagnosis (Defined in CD >24 Months and in UC >12 Months) with Disease-associated Items

Diagnostic Delay from First Symptoms to IBD Diagnosis
		Univariate Logistic Regression			Multivariate Logistic Regression
Item	Diagnosis	OR	95%-CI	P-value	OR	95%-CI	P-value
The underlined binary outcome denotes to what the odds ratio (OR) is referring to. Items with P < 0.2 in univariate analysis were entered into the multivariate model. EIM = extraintestinal manifestations at IBD diagnosis.
Gender
- male	CD	0.89	0.61-1.29	0.528
- female	UC	0.67	0.41-1.1	0.115	0.59	0.34-1.05	0.079
Age at diagnosis
- <40 years	CD	2.03	1.2-3.4	0.08	2.15	1.20-3.86	0.010
- ≥40 years	UC	0.40	0.12-1.34	0.14	0.44	0.1-1.92	0.273
Disease location
- ileum vs. colon (CD)	CD	1.69	1.08-2.66	0.022	1.69	1.06-2.68	0.025
- rectum vs. more extensive (UC)	UC	0.89	0.43-1.85	0.758
Positive IBD family history
- positive	CD	1.45	0.84-2.49	0.185	1.54	0.84-2.84	0.159
- negative	UC	0.64	0.22-1.89	0.423
EIM
- yes	CD	1.41	0.98-2.03	0.068	1.47	0.98-2.21	0.064
- no	UC	0.91	0.53-1.55	0.723
Education level
- basic	CD	1.13	0.72-1.76	0.603
- higher	UC	0.89	0.5-1.6	0.713
Cigarette smoking
- yes	CD	1.25	0.86-1.81	0.238
- no	UC	1.35	0.79-2.31	0.270
NSAID intake
- yes	CD	1.30	0.83-2.01	0.25
- no	UC	1.67	0.88-3.16	0.117	1.75	0.91-3.38	0.093
Oral anticonception
- yes	CD	0.89	0.59-1.34	0.582
- no	UC	1.34	0.72-2.49	0.345
Origin
- rural	CD	1.30	0.83-2.03	0.258
- urban	UC	0.48	0.23-0.96	0.039	0.66	0.29-1.43	0.292

In the univariate logistic regression, age at CD diagnosis <40 years and ileal disease were identified as risk factors significantly associated with long diagnostic delay. Furthermore, the presence of EIM was associated with long diagnostic delay too (P = 0.068). All items that reached P < 0.2 were subjected to multivariate logistic regression, which identified the following independent risk factors for a long diagnostic delay: age at diagnosis <40 years and ileal disease.

Rural area was the only factor in the univariate model that significantly (negatively) correlated with prolonged diagnostic delay (defined as time >12 months) in UC patients. No single item reached significance in UC patients in the multivariate model. A trend for prolonged diagnostic delay in UC patients was found with NSAID intake (odds ratio [OR] 1.75, P = 0.093), whereas male gender showed a trend for negative association with long diagnostic delay (OR 0.59, P = 0.079).

Of note, the time interval from first symptoms to IBD diagnosis is composed of the time intervals from first symptoms to the physician visit (patient-related) and from the physician visit to IBD diagnosis (physician-related). In order to define whether the delay could be attributed to either patients or physicians, we further evaluated these two intervals.

Impact of Disease-associated Items on Time Interval from First Symptoms to Physician Visit in CD and UC Patients

The results of the logistic regression for this patient-related time interval are further depicted in Table 5. For CD patients, none of the assessed items was significantly correlated with long diagnostic delay (defined as time >6 months). Using the univariate logistic regression, there was a trend for prolonged diagnostic delay in patients that either were <40 years of age at the time of diagnosis (P = 0.128), had a positive CD family history (P = 0.102), or presented with EIM at CD diagnosis (P = 0.140). None of the assessed items were found to be independently associated with prolonged diagnostic delay when data were subjected to multivariate logistic regression. However, a trend was observed for patients who were <40 years of age at CD diagnosis (P = 0.111), had a positive CD family history (P = 0.086), or presented with EIM at CD diagnosis (P = 0.182).

Table 5. Logistic Regression Modeling Evaluating “Long Diagnostic Delay” from First Symptoms to Physician Visit (Defined in CD >6 Months and in UC >4 Months) with Disease-associated Items

Diagnostic Delay from First Symptoms to Physician Visit
		Univariate Logistic Regression			Multivariate Logistic Regression
Item	Diagnosis	OR	95%-CI	P-value	OR	95%-CI	P-value
Items with P < 0.2 in univariate analysis were entered into the multivariate model. EIM = extraintestinal manifestations at IBD diagnosis.
Gender
- male	CD	0.80	0.54-1.19	0.282
- female	UC	0.82	0.53-1.28	0.381
Age at diagnosis
- <40 years	CD	1.59	0.88-2.89	0.128	1.63	0.89-2.98	0.111
- ≥40 years	UC	0.73	0.29-1.83	0.507
Disease location
- ileum vs. colon (CD)	CD	1.34	0.82-2.19	0.248
- rectum vs. more extensive (UC)	UC	1.15	0.63-2.13	0.644
Positive IBD family history
- positive	CD	1.63	0.91-2.92	0.102	1.68	0.93-3.05	0.086
- negative	UC	1.08	0.47-2.47	0.862
EIM
- yes	CD	1.35	0.91-1.99	0.140	1.32	0.88-1.97	0.182
- no	UC	1.10	0.69-1.77	0.685
Education level
- basic	CD	1.04	0.65-1.66	0.859
- higher	UC	1.55	0.896-2.69	0.116	1.35	0.69-2.63	0.384
Cigarette smoking
- yes	CD	1.21	0.81-1.80	0.351
- no	UC	1.30	0.79-2.13	0.297
NSAID intake
- yes	CD	1.09	0.67-1.76	0.736
- no	UC	1.55	0.88-2.73	0.133	2.37	1.13-4.97	0.022
Oral anticonception
- yes	CD	0.93	0.60-1.44	0.753
- no	UC	1.47	0.85-2.55	0.169	1.54	0.77-3.09	0.222
Origin
- rural	CD	1.09	0.67-1.79	0.708
- urban	UC	1.00	0.59-1.71	0.998

In UC patients no single factor was significantly associated with long diagnostic delay (>4 months) in univariate regression modeling. However, a trend for long diagnostic delay was observed for patients with a lower education level (P = 0.116), or those using NSAIDs (P = 0.133), or oral contraception (P = 0.169). The multivariate logistic regression identified NSAID intake (P = 0.022) as the only risk factor being significantly associated with long diagnostic delay in the interval from first symptoms to the primary physician visit in UC patients.

Impact of Disease-associated Items on Time Interval from Physician Visit to Diagnosis of CD and UC, Respectively

Lastly, we evaluated the disease-associated items in the physician-dependent time interval (from physician visit to IBD diagnosis). The results for the logistic regression are further shown in Table 6. For CD patients, ileal disease was significantly associated with long diagnostic delay (defined as time >18 months) in the univariate logistic regression model. A trend was found for long diagnostic delay for the following disease-associated items: age at diagnosis <40 years (P = 0.060), presence of EIM at CD diagnosis (P = 0.156), and NSAID intake at CD diagnosis (P = 0.067). The multivariate logistic regression identified ileal disease (P = 0.013) and age at diagnosis <40 years (P = 0.039) as statistically significant risk factors for long diagnostic delay.

Table 6. Logistic Regression Modeling Evaluating “Long Diagnostic Delay” from Physician Visit to IBD Diagnosis (Defined in CD >18 Months and in UC >5 Months) with Disease-associated Items

Diagnostic Delay from Physician Visit to IBD Diagnosis
		Univariate Logistic Regression			Multivariate Logistic Regression
Item	Diagnosis	OR	95%-CI	P-value	OR	95%-CI	P-value
Items with P < 0.2 in univariate analysis were entered into the multivariate model. EIM = extraintestinal manifestations at IBD diagnosis.
Gender
- male	CD	0.88	0.61-1.25	0.468
- female	UC	0.72	0.47-1.11	0.139	0.69	0.44-1.08	0.104
Age at diagnosis
- <40 years	CD	1.66	0.98-2.82	0.060	1.91	1.03-3.55	0.039
- ≥40 years	UC	0.27	0.08-0.89	0.032	0.27	0.08-0.92	0.036
Disease location
- ileum vs. colon (CD)	CD	1.89	1.22-2.91	0.004	1.84	1.14-2.96	0.013
- rectum vs. more extensive (UC)	UC	0.76	0.39-1.46	0.405
Positive IBD family history
- positive	CD	1.41	0.82-2.41	0.210
- negative	UC	0.79	0.34-1.89	0.612
EIM
- yes	CD	1.29	0.91-1.85	0.156	1.33	0.87-2.03	0.189
- no	UC	1.28	0.81-2.02	0.294
Education level
- basic	CD	1.28	0.82-1.99	0.284
- higher	UC	1.06	0.63-1.78	0.839
Cigarette smoking
- yes	CD	1.16	0.81-1.66	0.419
- no	UC	1.27	0.78-2.07	0.338
NSAID intake
- yes	CD	1.49	0.97-2.28	0.067	1.34	0.84-2.13	0.216
- no	UC	0.96	0.52-1.77	0.895
Oral anticonception
- yes	CD	1.15	0.77-1.72	0.488
- no	UC	1.69	0.98-2.91	0.061
Origin
- rural	CD	1.06	0.67-1.65	0.811
- urban	UC	0.62	0.35-1.10	0.102	0.61	0.33-1.10	0.101

In UC patients, age <40 years at diagnosis was the only factor that was significantly negatively associated with long diagnostic delay (P = 0.036) in the multivariate logistic regression model. Furthermore, a trend for long diagnostic delay was observed in female patients (P = 0.104) and those from urban areas (P = 0.101).

DISCUSSION

This is the first nationwide cohort study that analyzed the diagnostic delay in IBD patients and evaluated the delay-associated risk factors. The study also discriminates between a time delay that is patient-related (from first symptoms to physician visit) and one that can be attributed to physicians (from physician visit to IBD diagnosis).

In CD patients the median diagnostic delay from first onset of IBD-related symptoms to IBD diagnosis was 9 months, which is significantly longer than in UC patients (4 months). These data are in accordance with the findings of Pimentel et al8 observing a longer diagnostic delay in CD compared to UC patients (mean 7.7 ± 10.7 years versus 1.2 ± 1.8 years). Of interest, the authors in a second step excluded the patients meeting the Rome criteria for IBS; thereby, the duration of the diagnostic delay dropped to 6.9 ± 9.8 years in CD versus 0.8 ± 1.3 years in UC patients. These differences in the time interval from first symptoms to IBD diagnosis are probably related to the fact that the clinical manifestations of CD are much more variable than those of UC. Of note, the diagnostic delays in the Pimentel et al study are presented as mean ± SD, implying that these data were normally distributed, which was not the case in our cohort.

Fatigue, prolonged diarrhea with abdominal pain, weight loss, fever, with and without gross bleeding are the hallmarks of CD.21 In comparison, the frequent rectal bleeding in UC patients is probably more alarming and leads to earlier referral to a physician.22 Our data are further in accordance with that obtained by the Mayo group, describing a broad range in diagnostic delays in patients from Olmsted County.23, 24

We also separately assessed the time intervals from first symptoms to visiting the physician and from this visit to IBD diagnosis. The goal was to define the relationship between these two intervals. We found that for CD patients the time interval from physician visit to IBD diagnosis was much longer than the time interval from first symptoms to physician visit. These data demonstrate that there is considerable physician's delay, leaving room for improvement. In contrast to CD, the time intervals from physician visit to UC diagnosis are comparable to the time intervals from first symptoms to physician consultation. So far, no studies systematically assessed these two intervals. The analysis of the data distribution permitted us to define the percentage of patients being diagnosed at a certain timepoint with the specific disease. We found that 50% of CD patients were diagnosed within 10 months after initial symptom presentation compared to 4 months in UC patients and 3 months in IC patients. Accordingly, 75% of CD patients were diagnosed within 24 months compared to UC patients with 12 months and IC patients with 6 months. The data distribution shows that the mean diagnostic delay is strongly influenced by the 25% of IBD patients ranging in the 76th to 100th percentile.

We also assessed delay-associated items stratified according to CD and UC. Age <40 years at IBD diagnosis as well as ileal disease was independently associated with a long diagnostic delay in CD. The further analysis demonstrated that young age was not significantly associated with long diagnostic delay in the patient-dependent time interval (from first symptoms to physician visit), but was associated with long diagnostic delay in the physician-related time interval (from physician visit to IBD diagnosis). We conclude that younger CD patients are at risk for long diagnostic delay that is less patient- and more physician-dependent. So far, no studies systematically evaluated delay-associated risk factors. The increased risk for long diagnostic delay in younger CD patients may be partially explained by the overlap of CD symptoms with the characteristics of IBS, leading to an underestimation of underlying CD by the treating physician.25 Ileal CD was also significantly associated with prolonged diagnostic delay. While there was no relevant patient delay, a considerable physician-related delay was observed. This observation may be explained by less dramatic symptom presentation (pain but no bloody rectal discharge) in ileal compared to colonic CD.

Of interest, no single delay-associated item could be identified in the multivariate analysis to be associated independently with the time interval from first symptoms to IBD diagnosis in UC patients. NSAID intake was associated with long diagnostic delay in multivariate analysis evaluating the time interval from first symptoms to primary physician visit (patient-related). NSAID intake has been associated with flares in CD, as well as in UC patients.26, 27 Therefore, our initial hypothesis was that NSAID intake would be inversely correlated with a long diagnostic delay. However, the positive association with the patient-related long diagnostic delay may be indicative of the fact that patients on NSAID self-medication take longer to address their complaints to their physician.

The question may arise if it is at all clinically relevant to keep the diagnostic delay as short as possible. Solid evidence indicates that this should indeed be a primary goal, not only because this leads to improved quality of life of the IBD patients, but also because of the increased therapy success in early disease.9–13

Our study has several strengths and also limitations. We present the largest IBD population in which diagnostic delay was systematically assessed. Not only did we examine the time interval from first symptoms until established IBD diagnosis, but we also analyzed the separate time intervals from first symptoms to the physician visit and from physician visit to IBD diagnosis. The differential examination of these two time intervals allows defining the extent of patient- and physician-dependent diagnostic delay. These data are of key importance for future intervention strategies in IBD patients.

One limitation is that IBS symptoms are not systematically assessed in the Swiss IBD Cohort Study. Considering the overlap between IBD and IBS symptoms, it may be difficult to discriminate between the symptoms attributed purely to IBD and those that are attributed to IBS. However, Pimentel et al8 demonstrated that the duration of the diagnostic delay was only marginally reduced when the patients meeting the Rome criteria for IBS were excluded (for CD patients drop from 7.7 to 6.9 and for UC patients from 1.2 to 0.8 years). Furthermore, the concept of superimposed IBS in IBD patients probably has to be reconsidered, as this notion of IBS-like symptoms may also evolve due to persistent subclinical chronic inflammation in patients with clinically inactive disease. Another limitation may be that our national cohort is only partially population-based and that IBD patients recruited in hospitals (81%) are overrepresented to IBD patients recruited from gastroenterologists in private practice.

In summary, we demonstrate in a large nationwide Swiss cohort that the diagnostic delay in CD patients is significantly longer than in UC patients. Although the diagnostic delay is influenced by both patients and physicians, the impact of physician's delay is more pronounced in CD compared to UC patients. Multivariate logistic regression identified young CD patients and those with ileal disease to be at major risk for long diagnostic delay.

In conclusion, increased awareness among members of the general public and health practitioners is mandatory in order to decrease the diagnostic delay in IBD patients.

Acknowledgements

Members of the SIBDCS study group (all involved in the acquisition of data): Pierluigi Ballabeni, Peter Bauerfeind, Christoph Beglinger, Stefan Begré, José Bengoa, Janek Binek, Daniel Boller, Jan Borovicka, Christian Braegger, Bernard Burnand, Rafael Camara, Dominique Criblez, Philippe de Saussure, Lukas Degen, Joakim Delarive, Tobias Ehmann, Matthias Engelmann, Ali El Wafa, Christian Felley, Alain Frei, Remus Frei, Michael Fried, Michael Friedt, Florian Froehlich, Suzanne Gallot-Lavallée, Tilman Gerlach, Martin Geyer, Marc Girardin, Oliver Goetze, Horst Haack, Serge Hediger, Peter Hengstler, Klaas Heyland, Patrick Janiak, Pascal Juillerat, Vera Kessler Brondolo, Christoph Knoblauch, Gerd A. Kullak-Ublick, Michael Manz, Rémy Meier, Christa Meyenberger, Pierre Michetti, Christian Mottet, Christoph Müller, Beat Müllhaupt, Thierry Nicolet, Andreas Nydegger, Isabelle Pache, Franziska Piccoli, Julia Pilz, Valérie Pittet, Ronald Rentsch, Jean-Pierre Rey, Silvia Rihs, Daniela Rogler, Gerhard Rogler, Markus Sagmeister, Bernhard Sauter, Niklaus Schaub, Susanne Schibli, Alain M. Schoepfer, Frank Seibold, Johannes Spalinger, Philippe Stadler, Michael Steuerwald, Alex Straumann, Michael Sulz, Michela Tempia-Caliera, Joël Thorens, John-Paul Vader, Stephan R. Vavricka, Jürg Vögtlin, Roland Von Känel, Gert Wachter, Jürg Wermuth, Paul Wiesel.