To the Editor We read with interest the recent meta-analysis by Amitay and Keinan-Boker1 about breastfeeding and childhood leukemia incidence. The authors identified published case-control studies and reported an inverse association for breastfeeding for 6 months or more compared with breastfeeding for less than 6 months (summary odds ratio [OR] = 0.81; 95% CL: 0.73, 0.89).1 Nonetheless, we describe herein the effect of unmeasured confounding as an alternate explanation for the observed association.