We read with interest the article by Tsai et al about a significant association between African genetic ancestry in African American women and preterm delivery (PTD). They conclude that “more intensive investigation of genetic admixture in African Americans” is needed to identify novel PTD-related susceptibility genes. Their conclusions, however, rely on inconsistent trends in ancestry levels and the use of potentially biased estimates of ancestry, and they neglect important alternative risk factors for PTD.

First, their conclusion that subjects with higher levels of African ancestry have a higher risk of PTD is not supported by the inconsistent trends they report across the 8 PTD traits. For some traits, the 2nd and 3rd quartiles of African ancestry are significantly associated; for others, the 4th quartile is significant. It is difficult to imagine what sort of biologic function that is related to PTD could be associated with intermediate levels, but not high levels, of African ancestry. Furthermore, the marginally significant associations, with probability values that range from .007 to .04, are likely to disappear after correction for multiple testing across the PTD-related traits.

Second, accurate estimation of ancestry is highly dependent on choosing a sufficient number of markers and accurate reference populations. Tsai et al relied on a small set of 57 markers and comparative populations of Yorubans from Nigeria and Mormons from Utah, which are unlikely to represent all West African and European diversity. They reported similar ancestry estimates when using fewer ancestry informative markers, but this approach did not test the need for more markers or alternative reference populations, which potentially could change their estimates and alter the association with PTD.

Third, Tsai et al dismissed sociodemographic and environmental factors as unable to explain the racial disparity in PTD. Although they measured some environmental and behavioral variables (eg, education, drug use), they did not test for significant differences between cases and control subjects, despite finding higher rates of smoking and illicit drug use in PTD cases. Furthermore, previous research has shown strong associations between gestational outcomes and sociocultural factors (such as poverty, social support, residential segregation, and discrimination). Including these sociocultural factors in the study by Tsai et al may have caused the modest association between ancestry and PTD to disappear, as shown recently in a study of genetic ancestry and blood pressure in Puerto Rico.

In the absence of specific evidence for race-specific susceptibility genes, it may be more productive to investigate potentially modifiable sociocultural and behavioral risk factors that clinicians and policymakers can address to reduce the risk for PTD.