W e congratulate Denny et al. 1 for their report on direct visual inspection for cervical carcinoma screening. However, we question their recommendation to link visual inspection with immediate ablative treatment, an experimental strategy that they refer to as 'screen and treat'.Because poor-quality cervical screening programs can do more harm than good, and because entropy is real, cervical screening programs require routine audits. [2][3][4] Definitions of harm and good will vary among different communities, but the requirement for audit is universal to assess that screening programs remain in compliance with community standards. The screen-and-treat algorithm, outside of controlled research settings, produces no physical evidence on which to base meaningful program audits.Proponents of this algorithm, when challenged, have failed to provide solutions to the quality-control paradox embedded in that screening algorithm. 5,6 Therefore, it appears that the screen-and-treat protocol is not compatible with the requirements of 'first do no harm.' Consequently, this protocol has been characterized as an inappropriate public health policy. 7 We ask Denny et al. 1 to justify their continued support of the screenand-treat algorithm in the face of the concerns described above.