Modal markers, specifically hedges, are frequent in medical discourse. Translation of modal markers is essential for the proper decoding of target language in the medical domain and poses the problem of conferring signals of mitigated claims, as scientific writing conventions differ between languages and cultures. I argue that, in the medical domain, biased translation of modal markers -- in particular by more affirmative choices -- can distort readers' interpretation of treatment effectiveness and, thus, interfere with the communicative purpose of the text, when these markers are included in specific lexico-grammatical patterns used in the mediation of medical knowledge. This paper explores distortions due to biased translation of modal markers in a corpus of Cochrane Systematic Review Abstracts translated from English into French. The results suggest that modal markers most frequently responsible for distortion in translation are evidential and auxiliary verbs, followed by modal adjectives and adverbs. Moreover, frequent instances of distortion with embedded and overlapping markers (e.g. modal auxiliaries plus change in tense) were observed. It is hoped these findings can benefit the development of domain-specific MT systems and the teaching of specialized translation.