While the Crohn's disease activity index (CDAI) is the gold standard for defining clinical endpoints in Crohn's disease (Crohn's) clinical trials, its ability to distinguish symptoms due to inflammation from those that are non-inflammatory has been questioned.
To compare CDAI scores in patients with Crohn's and those with Irritable Bowel Syndrome (IBS).
This was a prospective, cross-sectional cohort study of 91 patients with either Crohn's (n = 44) or IBS (n = 47). Total CDAI and individual component scores were recorded and comparisons were made between Crohn's and IBS patients.
Mean CDAI scores were higher in the IBS patients (183 vs. 157, P = 0.1). Sixty-two per cent (n = 29) of IBS patients had CDAI scores greater than 150. Mean CDAI haematocrit score (35.9 vs. 23.0, P = 0.02) and CRP level (6.8 vs. 2.0, P = 0.002) were higher in the Crohn's group. Analysis of CDAI sub-scores demonstrated that IBS patients had significantly higher pain (mean 1.7 vs. 0.8, P = 0.0007) and well-being scores (mean 1.2 vs. 0.8, P = 0.04) relative to patients with Crohn's. Specifically evaluating patients with CDAI greater than 150 (n = 51), IBS patients had higher pain sub-scores (mean 2.4 vs. 1.4, P = 0.002), whereas patients with Crohn's had higher CRP (mean 8.4 vs. 1.8, P = 0.001).
Our study demonstrates that the CDAI does not discriminate patients with symptoms due to active Crohn's from patients with IBS. Patients with IBS can have CDAI scores in the clinically meaningful range. Objective measures, such as CDAI haematocrit score and CRP, are more specific markers of inflammation.
The Crohn's disease activity index (CDAI) was developed and validated in the late 1970s as part of the National Crohn's Disease Co-operative Study.[1, 2] Based on a group of 18 physicians' assessment of 112 patients over 187 patient visits, the CDAI was originally intended for use as a clinical tool to allow an accurate, objective and reproducible assessment of disease activity in patients with Crohn's disease (Crohn's). The index was designed by multiple regression and stepwise deletion, beginning with eighteen independent variables that were thought to predict the physician's global assessment of Crohn's activity (PGA), which was the dependent variable. Eight variables were selected based on their correlation with PGA and remain the parameters that comprise the CDAI today. The main contributors to the index in the original study were number of loose stools, abdominal pain, well-being and haematocrit. The CDAI has been modified in several ways since its establishment and the recently developed short CDAI (sCDAI) has been shown to correlate well with the full index. Traditionally, for clinical trials, clinical remission is defined as a CDAI <150, clinical response is a decrease in the CDAI of 70 points, and more recently, 100 points. Mildly active Crohn's is defined as CDAI 150–220, moderate–severe Crohn's is typically a CDAI 220–450, and severe disease is defined as a CDAI >450.
The CDAI, although a validated[1, 2] and widely used index in clinical trials for Crohn's, has been criticised for being cumbersome, prone to inter-observer variability,[5, 6] poorly reproducible and not responsive to change over time, even when used amongst experienced gastrointestinal physicians.[5, 7] The CDAI and PGA have been shown to correlate poorly with more objective endpoints, such as laboratory markers, endoscopic and histological disease activity.[4, 6, 8, 9] Nonetheless, the index is the accepted gold standard for large clinical studies in Crohn's and is the benchmark by which severity of disease, clinical response and remission rates are measured. These indices also have been greatly relied upon in the approval process of most of the medical therapies available for Crohn's today. However, high placebo response rates are consistently observed in many Crohn's trials[10-12] and this phenomenon is thought to be due, at least in part, to the subjectivity of some of the CDAI variables and the lack of accurate discrimination between inflammatory and non-inflammatory symptoms, such as IBS, which coexists with Crohn's in 42–57% of patients.[14-16] A recent meta-analysis suggests that even for Crohn's patients judged to be in remission, the prevalence of IBS type symptoms is as high as 35%. Small intestinal bacterial overgrowth and bile salt malabsorption, which patients with Crohn's are at increased risk for, can also lead to similar symptoms. This lack of specificity is highlighted by the SONIC study in which 18% of Crohn's patients who entered the trial with a CDAI >220 did not have endoscopic evidence of active Crohn's.
Therefore, we sought to assess the utility of the CDAI in differentiating Crohn's from IBS and to isolate components within the CDAI that could be responsible for elevating the CDAI scores in IBS patients.
The study was a prospective, cross-sectional cohort study carried out over a 6-month period from September 2010 to March 2011. Patients were recruited from general GI, motility and inflammatory bowel disease out-patient clinics at the Division of Gastroenterology in a large university-affiliated teaching hospital.
The study was powered for comparison of 50 patients in each group, with 80% power to detect differences of 20% between groups. Subsequently, 50 patients with Crohn's and 50 patients with IBS were identified in out-patient clinics and interviewed. Patients with IBS were recruited from a combination of specialist motility/IBS clinics (20/47, 43%) and from general gastroenterology out-patient clinics (27/47, 57%). The diagnosis of IBS was made by the attending physicians responsible for their care utilising the Rome III criteria who have expertise in the diagnosis and treatment of functional bowel disorders. IBS patients were further categorised according to subtypes of IBS with diarrhoea, IBS with constipation, or IBS with mixed pattern. Patients with Crohn's were recruited from general gastroenterology (1/44, 2%) out-patient clinics and from the Center for Inflammatory Bowel Disease (43/44, 98%). The diagnosis of Crohn's was based on the clinical impression of the attending physicians responsible for their care who have expertise in the diagnosis and treatment of Inflammatory Bowel Diseases. All patients were recruited consecutively and thus represent a wide spectrum of disease activity for both Crohn's and IBS.
Prospective data collection
Patient interviews were conducted at the time of routine office visit. All attempts were made to sequentially enrol patients, regardless of the severity of symptomatology. All patients, no matter what the diagnosis, were administered the Crohn's Disease Activity Index (CDAI), Physician Global Assessment (PGA), BEST score (pain numeric rating scale for IBS) and IBS Module (IBS activity index).[18, 20]
Chart review and exclusions
Patients were excluded if they did not have an established diagnosis of Crohn's or IBS. Patients were also excluded if they had more than one gastrointestinal disease (e.g. Crohn's and coeliac disease) or if subsequent pathology was found that could account for their symptoms. Patients in the Crohn's group were classified according to the Montreal classification. Haematocrit was recorded when available within 60 days of enrolment in the study. Similarly, we documented C-reactive Protein (CRP) and Erythrocyte Sedimentation Rate (ESR) values available within 30 days of the interview for Crohn's and within 180 days for IBS. Blood draws were not performed as part of the study protocol and were only recorded where performed as part of routine clinical care. Retrospectively, endoscopy and radiology records were reviewed to confirm the diagnosis (Crohn's) or exclude underlying IBD (IBS patients).
Data were analysed comparing CDAI total scores, CDAI component scores, PGA score, BEST score and IBS Module in Crohn's and IBS patients. For each patient, we calculated short CDAI (sCDAI), which is a modified index including abdominal pain, stool frequency and well-being scores derived from the original CDAI and a constant. Additional comparisons were made between the Crohn's group and the different subsets of patients with IBS. All authors had access to the study data and reviewed and approved the final manuscript.
CDAI haematocrit score
We analysed differences in both absolute haematocrit and CDAI haematocrit score. The CDAI haematocrit score is the deviation of the patient's current haematocrit from the median normal value and is dependent on gender. The haematocrit score is higher in patients who are anaemic.
If no haematocrit was available in the prespecified time period (n = 8, 17%), the mean haematocrit for the overall group was substituted, as described previously. No substitution with mean values was made for CRP as it is not used in the calculation of the CDAI score.
Statistical analysis was carried out using STATA (StatCorp LP, College Station, TX, USA) and GraphPad Prism (GraphPad Software, San Diego, CA, USA) software packages using the student's t-test and Mann–Whitney for continuous parametric and nonparametric variables respectively and Fisher's exact test for categorical variables. anova/Kruskal–Wallis tests were used for comparing means between individual groups and Pearson and Spearman coefficients calculated correlations between parametric and nonparametric sets of data respectively. A P-value <0.05 was taken as significant for all analyses.
Full approval from the Institutional Review Board of Beth Israel Deaconess Medical Center was obtained before commencement of the study (IRB#2009P-000206). Written informed consent was obtained from each patient prior to enrolment.
Fifty patients with Crohn's and fifty patients with IBS were enrolled. Following chart review, 6 patients with Crohn's were excluded due to the co-existence of IBS (n = 1), concomitant diagnosis of coeliac disease (n = 1), diagnosis of indeterminate colitis (IBD-U, n = 2), and patients with ileal pouch-anal anastomosis and subsequent diagnosis of Crohn's (n = 2). Three IBS patients were excluded due to the subsequent discovery of ovarian neoplasm (n = 1), inability to definitively rule out Crohn's (n = 1) and history of major abdominal surgery felt to contribute to ongoing GI symptoms (n = 1).
Clinical, demographic and biochemical characteristics of the Crohn's (n = 44) and IBS (n = 47) patients are shown in Table 1. Forty-two (89%) IBS patients had undergone colonoscopy with biopsies taken in 41 patients (87%). Crohn's patients are characterised by Montreal Classification and IBS patients are sub-typed according to Rome Classification. Haematocrit was available in 44/44 (100%) Crohn's patients and 39/47 (83%) IBS patients. CRP was available in 38/44 (86%) of patients with Crohn's and 28/47 (60%) of patients with IBS.
Table 1. Clinical, demographic and biochemical differences between patients with Crohn's (n = 44) and those with IBS (n = 47)
Crohn's (n = 44)
IBS (n = 47)
Bold values indicate statistically significant.
Duration of disease and age at diagnosis were recordable in only 36 of 47 patients in the IBS group (77%).
Haematocrit was available in 39 of 47 IBS patients (83%). Substitution of median values, as previously reported, allowed calculation of CDAI haematocrit score for recording CDAI values in this group. However, comparison of haematocrit scores between Crohn's and IBS as presented here is based on actual values recorded in 100% Crohn's group and 83% IBS group, without substitution of median values.
Patients with Crohn's had a higher mean CRP (6.8 vs. 2, P = 0.002) and CDAI haematocrit score (35.9 vs. 23, P = 0.02) compared with IBS patients (Figure 1). There was no difference in absolute haematocrit value between Crohn's and IBS patients (P = 0.8, Table 1).
Mean CDAI scores in the IBS group trended towards a higher value than in the Crohn's patients (183 vs. 157, P = 0.1). The data are presented in Table 2. Short CDAI (sCDAI) values were significantly higher for the IBS patients than for the Crohn's patients (195 vs. 153, P = 0.01). Mean PGA score for IBS was 3.4 compared with 2.9 for Crohn's (P = 0.05). IBS Module scores were significantly higher in the IBS group (14.7 vs. 11, P = 0.02) as were BEST scores (53.4 vs. 29.6, P < 0.0001). CDAI, sCDAI, PGA IBS Module and BEST score distributions in both groups are shown in Figure S1.
Table 2. Clinical Disease Indices as Measured in Crohn's (n = 44) and IBS (n = 47) patients. All CDAI components were examined in both groups. Extra-intestinal manifestations are not represented, as there were none in the IBS group. There were no patients in either study group with a clinically detected abdominal mass or a recorded deviation from their ideal body weight. These parameters are not represented here
Crohn's (n = 44)
IBS (n = 47)
CDAI, Crohn's Disease Activity Index; sCDAI, Short Crohn's Disease Activity Index; PGA, Physicians Global Assessment; IBS Module, Irritable Bowel Syndrome Module (disease activity index); BEST Score, pain numeric rating scale for IBS.
Bold values indicate statistically significant.
CDAI (mean, s.d.)
sCDAI (mean, s.d.)
PGA score (mean, s.d.)
IBS Module (mean, s.d.)
BEST (mean, s.d.)
CDAI Sub-scores (n = 91)
n = 44
n = 47
Stool (mean, s.d.)
Pain (mean, s.d.)
Well-being (mean, s.d.)
CDAI haematocrit score (mean, s.d.)
CDAI >150 (%)
CDAI >220 (%)
CDAI >150 Sub-scores (n = 51)
n = 22
n = 29
Stool (mean, s.d.)
Pain (mean, s.d.)
Well-being (mean, s.d.)
Haematocrit score (mean, s.d.)
CRP (g/dL) (mean, s.d.)
ESR (mm/h) (mean, s.d.)
Sixty-two per cent of the IBS patients had a CDAI >150 (which would qualify as active disease if they had Crohn's) and 32% had a CDAI >220 (which meets the definition of moderate-severe Crohn's). By contrast, in the Crohn's patients, these higher scores were present in 50% and 15% of patients respectively. The higher scores in the IBS patients appear to have been driven by higher mean pain sub-scores (2.4 vs. 1.4, P = 0.002). Similar to the overall group, mean CRP was significantly higher in Crohn's patients with CDAI >150 compared with IBS (8.4 vs. 1.8, P = 0.001).
Differences between the Crohn's and IBS patients with respect to the individual components of the CDAI are included in Table 2. Mean abdominal pain score (1.7 vs. 0.8, P = 0.0007) and mean overall well-being score (1.2 vs. 0.8, P = 0.04) were significantly higher in IBS patients. The higher pain and well-being scores correlate with feeling worse. The predominance of pain and well-being scores at the higher end of the scale within the IBS group is shown in Figure 2.
Patients with IBS-D and IBS-M had trends towards higher mean CDAI scores (174.3 and 258.7 respectively) than those with IBS-C (158.1) and Crohn's (157), although this relationship was not statistically significant (P = 0.15, anova, Figure 3a). A sub-analysis looking at components of the CDAI showed higher average daily stool scores in Crohn's, IBS-D and IBS-M (3, 2.4 and 5.5 respectively) when compared with IBS-C (0.75) (P = 0.01, Figure 3b). There were no differences in other CDAI component scores between IBS sub-groups (data not shown). Patients with IBS-M (n = 7) had significantly higher CDAI scores compared with Crohn's (259 vs. 157, P = 0.03, Mann–Whitney), while there was no difference in mean CDAI values between the Crohn's and other IBS sub-types.
The correlation between physician assessment (PGA) and CDAI for the two groups is shown in Figure S2. The correlation was stronger and statistically significant in Crohn's (r = 0.51, P = 0.0005) than IBS (r = 0.28, P = 0.05).
To our knowledge, this is the first study to characterise how functional gastrointestinal symptoms can lead to significant elevations in the CDAI. We found that patients with IBS had higher median CDAI scores than patients with Crohn's. In fact, in our study population, 62% of patients with IBS had a CDAI score above 150 and 32% had a CDAI greater than 220, which represents moderate-to-severe Crohn's based on the CDAI (i.e. CDAI score between 220 and 450). These findings show that functional gastrointestinal symptoms, which are common in IBD, can affect CDAI scores, producing levels that could impact clinical trials and how we treat our patients.
Our study further demonstrates that abdominal pain and well-being are the predominant sub-scores resulting in elevation of the CDAI in patients with IBS. Additionally, it has previously been suggested that quantification of stool habits is over-used and can be misleading. Our findings confirm that, in patients with IBS-D and IBS-M, higher stool frequency can lead to elevated CDAI scores.
The assessment of more objective markers of disease activity in Crohn's patients may help increase the specificity for active inflammation. Sands et al. suggested that the inclusion of biomarkers such as CRP in a modified disease activity index would reduce the variability observed with the current index. These tests may include laboratory indices such as CRP or calprotectin or endoscopic indices such as mucosal inflammation and histological evaluation. We have shown that two parameters, CRP and CDAI haematocrit score, can more accurately differentiate patients with gut inflammation from those with IBS.
Haematocrit score is the deviation of the patient's current haematocrit from the median normal value and is dependent on gender. Patients who are anaemic will have higher CDAI haematocrit scores. The haematocrit value was no different between groups and this reflects the female predominance in the IBS group. As CDAI haematocrit score is adjusted for gender, this confounding is corrected and thus represents a more accurate representation of the difference between the Crohn's and IBS groups. Additionally, when clinicians are evaluating patients with Crohn's disease and on-going symptoms and are deciding whether to escalate therapy, it is important to ensure that the symptoms are driven by inflammation, as symptoms alone may not always be reliable. Utilisation of CRP, faecal calprotectin and endoscopy in addition to the CDAI will aid with this distinction. These objective markers should also enable clearer differentiation between inflammatory and non-inflammatory symptoms and facilitate reproducible assessment of disease activity in Crohn's for use in future clinical trials. C-reactive protein is already being used increasingly as an endpoint in some of the more recent clinical trials.[9, 23] Biomarkers such as CRP and faecal calprotectin are reproducible, allow easy statistical analysis and are a more consistent measure of treatment response.
We used CRP in our study as faecal calprotectin is not readily available for use at our institution. However, CRP does have its weaknesses as a biomarker in Crohn's. Up to 25% of patients with active inflammation due to IBD will not mount any significant CRP response. In addition, there is some evidence that in active Crohn's, CRP arises not from the intestinal mucosa but from the inflamed mesenteric fat and therefore may not be a good surrogate marker for intestinal inflammation. Faecal calprotectin is released from intestinal neutrophils, monocytes and macrophages and may be a more reliable surrogate for gut inflammation.[26, 27] Studies have shown it to correlate well with endoscopic disease activity and with response to TNF alpha treatments. However, it may be a better marker of inflammation in ulcerative colitis than in Crohn's.[30, 31] Additionally, calprotectin levels may be raised in IBS and in healthy controls and may reflect inflammation due to NSAID-related ulcers or diverticular disease Despite these limitations, many feel that it should be used both as an inclusion criterion for IBD clinical trials and as a marker of disease activity in clinical practice, perhaps in conjunction with CRP.
Other indices of disability in IBD, some of which include quality of life measurements[36, 37] and some of which incorporate endoscopic scores and cross-sectional imaging, have received attention in recent times and can expect to be used increasingly as outcome measures for future clinical trials.
Several attempts have already been made to modify the CDAI. Thia et al propose a short CDAI (sCDAI), which correlates well with the original index. Harvey et al proposed a simplified index in 1980, which again correlated well with the CDAI. However, neither of these indices contains any objective variables and thus are subject to the same limitations as the CDAI. In fact, the sCDAI is comprised of these same three problematic components (abdominal pain, well-being, and stool habit) and in our study, the median sCDAI scores were statistically significantly higher in the IBS group than in the Crohn's group in a similar way to CDAI scores.
Correlation between PGA and CDAI score was stronger in Crohn's than IBS, suggesting that physicians' interpretation of symptoms in Crohn's may be somewhat more reliable than in the IBS patient. Perhaps this signifies a tendency of physicians to under-rate symptoms in patients with functional disease, and certainly highlights practical difficulties in the objective assessment of the IBS patient, including patients with Crohn's and overlap IBS-type symptoms.
Our study did have a number of limitations. There were more female patients in the functional group, although this is not surprising as most trials of IBS patients suffer from high female-to-male ratio. Patients with Crohn's had a longer median duration of disease than patients with IBS. This may be due to the fact that Crohn's is easier to diagnose[40, 41] and physicians are less likely to label patients as IBS without investigation and a period of clinical monitoring. It must also be noted that we assessed duration of disease and did not specifically assess duration of symptoms. Interviewers were not blinded to the patients' diagnoses. However, the role of the interviewer was to simply ask the questions and hand out the indices, and thus there was little chance for bias to be introduced. Additionally, biochemical parameters were not available for all patients. CRP was noted in 38/44 (86%) of patients with Crohn's and 28/47 (60%) of patients with IBS. Haematocrits were available in all Crohn's and 39/47 (83%) of IBS patients. This discrepancy is a reflection of the nature of the study as laboratory tests were recorded at time of enrolment, but blood draws were only performed where clinically indicated and not as part of the study. To account for the patients without a haematocrit value, the mean from the group was used to calculate the CDAI as was as previously described by Best et al. Additionally, patients were recruited from a tertiary referral centre and it is therefore possible that IBS patients enrolled were at the severe end of the clinical spectrum. To offset this, we enroled patients not only from motility/IBS specialist out-patient clinics but also from a general gastroenterology clinic. In the CDAI >150 subgroup, the same sub-scores appear to be driving higher CDAI scores in both Crohn's and IBS. Furthermore, CRP was significantly higher in Crohn's patients with CDAI >150, which is similar to our observation for all patients, suggesting inflammatory and non-inflammatory aetiology for elevated CDAI scores in Crohn's and IBS patients respectively.
Our study demonstrates that the CDAI does not discriminate patients with symptoms due to active Crohn's from patients with IBS. Subjective measures, specifically abdominal pain, well-being and stool frequency, can elevate CDAI scores in patients who do not have Crohn's. Objective measures such as CDAI haematocrit score and CRP are more specific markers of gut inflammation.
Guarantor of article: A. S. Cheifetz.
Author contributions: AS Cheifetz: study concept and design, drafting of the manuscript, acquisition of data, analysis and interpretation of data; critical revision of the manuscript for important intellectual content of the manuscript. C Lahiff: acquisition of data, analysis and interpretation of data, drafting of the manuscript critical revision of the manuscript for important intellectual content of the manuscript. P Safaie: acquisition of data, analysis and interpretation of data, critical revision of the manuscript for important intellectual content of the manuscript. M Akbari: analysis and interpretation of data, drafting of the manuscript critical revision of the manuscript for important intellectual content of the manuscript. A Ahmed, S Sheth, L Gashin: acquisition of data, critical revision of the manuscript for important intellectual content of the manuscript. A Moss, A Lembo, D Leffler: study design, acquisition of data, analysis and interpretation of data, critical revision of the manuscript for important intellectual content of the manuscript. All authors have reviewed and approved the final version of the manuscript.
Declaration of personal interests: A. S. Cheifetz has served as a consultant for Janssen Pharmaceuticals, Abbott Labotatories, Prometheus Labs, Given Imaging and Warner Chilcott.