A PHS 795 student draws our attention to a quasi-experimental study examining the impact of England’s comprehensive public smoking ban on infant birth weight and mortality. Although it is true that an interrupted time series design is relatively weak, the magnitude of this intervention as a nationwide policy and the strength of the biological mechanisms connecting maternal smoking to these outcomes gives a whole lot of weight to its conclusion (why it was published in a high impact journal like Nature). Our student writes:
In July 2007, England passed a nationwide, comprehensive smoking ban. Since then, virtually all work and public places have been smoke-free. A few studies have looked at the positive impacts of this on adult health, in terms of direct smoking as well as second-hand smoke.
This research article from 2015 uses a quasi-experimental design to look at the effect of England’s smoke-free legislation on perinatal survival. As we’ve learned in this class, exposures during early stages of the life course can have the biggest impacts on lifetime health. This study looked at death certificates for all births between 1995 and 2011: over 52,000 stillbirths and over 10 million live-births were examined. They found the smoking ban caused a near-immediate reduction in stillbirths, low birth weight, and neonatal mortality for the study group.
Despite being one of the largest studies to investigate this topic, there are some inherent limitations to this type of analysis. How does the experimental design used here relate to the quasi experimental methods discussed with Dr. Remington? Are there other experimental designs that might be reasonable to examine the relationship between smoking bans and health benefits?