Powers et al raise 4 points in seeking to allay concerns I raised about tretinoin safety in light of the VATTC trial.1 None is convincing.
First, Powers et al are technically correct that an abstract reporting VATTC study results published in 20052 stated that excess mortality in the tretinoin group compared with the placebo group was not statistically significant after adjusting for confounders. But a more complete VATTC study report published in 20093 stated that excess mortality in the tretinoin group—122 of 566 participants (22%), compared with 90 of 565 (16%) participants in the placebo group—was statistically significant (P = .02), even after adjusting for confounders, and therefore cannot be attributed to imbalances in randomization. Moreover, the magnitude of the mortality difference is clinically significant. The number of VATTC study participants exposed to tretinoin per excess death—the “number needed to harm”—was only 18.
Kenneth A. Katz. Tretinoin: An Established Long-term Safety Profile—Reply. Arch Dermatol. 2009;145(9):1064–1065. doi:10.1001/archdermatol.2009.207