Introduction Faecal calprotectin (FC) has become an established non-invasive, sensitive marker of intestinal inflammation in Crohn's disease (CD). A single measurement of FC is used in clinical practice to aid decision making in the management of CD patients. Data on the day to day variation of FC levels in CD patients is sparse. If there was significant intra-individual variation of FC within CD patients, this could impact on the interpretation of single values in isolation. A prospective study to evaluate the degree of variability of FC in three consecutive days stool samples in CD patients in clinical remission was performed.

Methods 143 patients with CD in remission (CDAI <150, no escalation of medical therapy within 3 months) who were attending for routine follow-up were identified and enrolled. Patients were excluded if they were taking NSAIDs or developed any change in symptoms over the 3-day collection period. After informed consent, patients were asked to provide stool samples from three consecutive days for analysis. FC was analysed using the Buhlman FC ELISA as per the manufacturer's instructions.

Results Of the 143 patients recruited 34 did not return any samples, 6 returned less than the required number of stool samples, 2 withdrew consent and 2 developed clinical evidence of a flare of disease during the collection period and were excluded from analysis. Therefore 98 complete sets of three stool samples were obtained and analysed. The intraclass correlation coefficient of the 3 FC values, log-transformed, was 0.842 with corresponding 95% CI 0.788 to 0.886 showing that consistency between the three log-transformed FC measurements is high. The reliability of a “normal” FC result of <50 and of <100 was assessed using the κ statistic for agreement between the three measurements. For a FC result of <50 κ was 0.648 (95% CI 0.508 to 0.770) and for a FC result of <100, κ was 0.603 (95% CI 0.483 to 0.712) demonstrating that there is moderately good and similar levels of agreement across the 3 samples for both measures. The reliability of the cut-off of 50 is slightly better than the cut-off of 100.

Conclusion The consistency of three faecal calprotectin samples over three consecutive days is high and the CI is narrow suggesting that in a larger population consistency would also be high. This indicates that there is little intra-individual variability of the test and a one off sample can indeed provide reliable information for use in clinical practice. Higher intrapatient variability of high FC values suggests that the log-transformed values are more reliable.

Competing interests None declared.