To the Editor.—We read with interest the article by Torres et al,1 published in the January 1992 issue of the Archives, studying the effect of somatostatin on healing of gastrointestinal fistulas. Torres and colleagues studied two populations of 20 patients each with gastrointestinal fistulas, treating one group with total parenteral nutrition (TPN) alone and the other with TPN and intravenous somatostatin infusion. They concluded that there was no significant difference in the percentage of fistulas that closed, but patients treated with TPN and somatostatin experienced closure of fistulas within a significantly shorter period than patients treated with TPN alone.
However, during the performance of their study, the investigators transferred four patients from the TPN-alone group to the TPN plus somatostatin group. These four randomized patients, therefore, were excluded from analysis. We believe that this error impaired the credibility of the trial, and if these excluded patients had remained