These power scenarios suggest there are many cases in which using public controls in GWAS would substantially improve power and justify the additional effort to obtain and use public controls. However, they do not take into account additional issues with use of public controls including differences in phenotype measurement, as well as potential systematic genetic and environmental differences between the public controls and study participants. Use of public controls requires either phenotype harmonization across contributing samples or, alternatively, use of population controls where the phenotype is not measured but is rare enough in the population that misclassification of true cases as controls is unlikely. Similarly, careful attention to population stratification across contributing public control datasets or between case and control datasets is also necessary to ensure that systematic biases are not introduced into the analyses of the combined datasets. For example, ascertaining African American study participants from one part of the United States and obtaining African American public controls ascertained from another location could introduce systematic differences and spurious findings due to differing types/levels of admixture or differences in environmental
