Crohn Disease: New Scoring System Predicts Mild Disease

June 18, 2013

Patients with mild Crohn disease could avoid expensive and potentially harmful treatment if they are identified early.

In this retrospective study of 162 patients with newly diagnosed Crohn disease (CD) in 14 German gastroenterology clinics, the investigators developed a 5-item scoring system, attempting to predict which patients will experience a mild course after initial diagnosis. Such a scoring system is relevant because of the continuing debate about whether professional societies and practice guidelines should advocate a “top-down” approach to CD, or the “step-up” approach generally favored in current guidelines.

If a scoring system could reliably predict those CD patients most likely to have a mild course, expensive and potentially harmful medications (biologics and other immunomodulators) could be avoided in favor of mesalazine or mesalazine in combination with a single short course of low-dose prednisone at the start of treatment. 

Two basic cohorts were retrospectively identified during the follow-up period (average 3.5 years) after being newly diagnosed:
• Mild CD: patients who required only mesalazine or mesalazine plus short-course, low-dose prednisone at the start of treatment.
• Moderate CD: patients who received other medical therapies or surgery.

A scoring system (Table 1) was developed based on existing risk factors for severe CD.

Table 1.



Age at first diagnosis

> 40


C-reactive protein, mL/L

2 to <4


Endoscopic score



Perianal lesions



Complications: >1, stenosis, fistula, extraintestinal manifestations, or fever (temperature >38ºC)



The accuracy of the scoring system for predicting a mild course at the time of initial diagnosis was expressed in terms of sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) using different cutoff points, seen in Table 2.

Table 2.





<1 vs >1





<2 vs >2





<3 vs >3






Cutoff Equals Trade-off
As always, manipulating a cutoff point introduces a trade-off: at lower cutoff points the specificity increased and the sensitivity decreased. The authors suggested using the middle cutoff ( ≤2) as a predictor of mild disease, saying that this would yield “sufficient” sensitivity (correctly identifying 72% of mild CD patients), with “acceptable” specificity (correctly identifying 54% of the CD patients who will not be mild). But that means that the algorithm has a false-negative rate of 28% (28% of the mild patients will be misidentified as moderate), and a false-positive rate of 46% (46% of moderates will be misidentified as mild). 

But this is similar to most medical tests, which simply decrease a fraction of the uncertainty, but do not provide a definitive answer. In combination with a rapid step-up strategy, if patients unexpectedly turn severe (see "Inflammatory Bowel Disease Treatment: Advance to Anti-TNFs?") this simple algorithm could be quite useful. 

Kruis W, Katalinic A, Klugmann T, et al. Predictive factors for an uncomplicated long-term course of Crohn's disease: a retrospective analysis. J Crohns Colitis. 2013;7:e263-e270.