Insights From the Positive Association of Height With Incident Venous Thromboembolism
This article requires a subscription to view the full text. If you have a subscription you may use the login form below to view the article. Access to this article can also be purchased.
Height and Venous Thromboembolism
Zöller et al1 clearly show height positively associated with incident venous thromboembolism (VTE) in both men and women using an observational design and a cosibling design in a very large study, including a sizeable segment of the Swedish population, that is, all males born 1951 to 1992 without previous VTE and all first time pregnant women from 1982 to 2012. Despite the large size of the study,1 the observational design is open to bias from confounding by common causes of height and VTE. Inevitably, in a study making cost-effective use of invaluable Swedish registry data, Zöller et al1 were unable to adjust for all potential confounders leaving the study open to bias from residual confounding from unmeasured attributes affecting both height and VTE. For example, they used participant’s educational level as a confounder1 instead of all aspects of parental socioeconomic position which might not fully account for confounding by such an influential, multidimensional attribute as parental socioeconomic position. Residual confounding by parental socioeconomic position could in this instance partially obscure any harmful effects of greater height, meaning that the observational estimate might underestimate any harmful effects of height. Zöller et al1 also adjusted for family history of VTE and body mass index. Family history and body mass index are probably not confounders, because they may cause VTE, they are unlikely to determine height, although height might determine body mass index. As such, the adjusted estimates have also been adjusted for potential mediators, which might attenuate any harmful effects of height, meaning that the observational estimate might be an underestimate. In this situation, the cosibling design is particularly helpful because it makes different assumptions. The cosibling design essentially compares siblings, which automatically controls for measured and unmeasured confounders shared by siblings, such as parental socioeconomic position …