Reply to Spada et al.

We read with great interest the comments by Spada et al., which we greatly appreciate. Their constructive criticism of the statistical analysis in reference to the specificity values in our study is well taken.

When the first feasibility study of the Colon 2 capsule was designed, colonoscopy was taken as the gold standard even though the literature has documented a miss rate of 10 % – 20 % for colonic polyps. Furthermore, polyp size assessment is of clinical importance. Given Imaging developed a novel research tool to estimate polyp size as visualized by the Colon 2 capsule. Our study attempted to carefully monitor the polyp size estimation algorithm of the Colon 2 capsule by comparing it to polyp size estimates obtained by standard colonoscopy. Even though polyp size estimation contains a significant subjective component, we accepted estimates obtained by standard colonoscopy as the gold standard. Our criteria for proof of principle of the capsule endoscopy algorithm for polyp size estimation were very rigorous.

The specificity was calculated on the basis of these very strict criteria. We could not agree more that these calculations do injustice to clinical relevance. It is obvious that a polyp estimated by one method to be 11 mm and by the alternate method to be 9 mm does not equate to a false-positive result, and it goes without saying that a polyp missed by standard colonoscopy is definitely not a false-positive result for capsule endoscopy if indeed a polyp is removed at a repeat colonoscopy.

In our paper we refer to this in the discussion when we say: ”The specificity has been calculated according to strict statistical rules but the clinical relevance of this data may be viewed differently, since polyps missed by colonoscopy or mismatch in polyp size were recorded as false-positive findings.”

We are therefore in full agreement with Spada et al. that polyp mismatch and the colonoscopy miss rate partially due to overinflation of the colonic lumen are invalid reasons to reduce the clinically relevant specificity of the Colon 2 capsule performance. Based on our experience we recommend that in future studies these limitations be rectified by designing a statistical algorithm which will allow the above deficiencies to be corrected. Clinical trials with complementary adjudication may be helpful too.