Discussion Questions

• ■
  

  What were the authors’ aims?

  
  
• ■
  

  What link might exist between carbon monoxide exposure and preeclampsia?

  
  
• ■
  

  What was the study design?

  
  
• ■
  

  What information is contained in the tables?

  
  
• ■
  

  What is misclassification bias?

  
  
• ■
  

  What direction should future research take?

When surveyed, nearly 22% of pregnant white women aged 15-44 years said they had smoked cigarettes in the previous month; 14.2% of black women and 6.5% of Hispanic women said the same. While cigarette smoking is always discouraged—and particularly during pregnancy—the practice has been shown to reduce the risk of preeclampsia in several populations. Carbon monoxide (CO) released by the burning cigarette appears to increase trophoblast invasion and uteroplacental blood flow, reduce hypoxia-induced apoptosis, and upregulate placental antioxidant systems. This month, Journal Club members discussed a new study by Zhai et al, which looked for a relationship between environmental CO and preeclampsia.

Environmental epidemiology

Because ecologic measurements are typically very expensive, investigators generally obtain fewer than they otherwise might. For this study, however, the authors had access to an incredibly large amount of data gathered by the Ontario Ministry of Environment. Measurements of ambient CO levels had been continuously collected at 23 sampling stations from January 2003 to December 2009. Researchers obtained the values from the Air Quality Ontario website, allowing them to accurately assess environmental CO exposure during gestation for pregnant women who lived within 10 km of a measuring site.

To accomplish this, Zhai and colleagues applied a common strategy in environmental epidemiology; they used postal codes to appraise subject exposure. The perinatal database, which was comprised of births that occurred between April 2004 and December 2009 and were recorded in Ontario’s provincial birth record system—the Better Outcomes Registry and Network (BORN) Ontario—was linked to the ambient CO database by the postal code of the mother’s residence. While this tactic might be very accurate in population-dense (urban) areas, it is relatively inaccurate in rural areas, because a postal code might encompass larger geographic areas with sporadic clusters of people. Exposure might then vary among all those who have been grouped together in this manner. However, the team limited the potential for error by only including women whose residence was within 10 km of a sampling site. This step reduced the error introduced by using a postal code to estimate proximity to the sampling site, although it could not be completely eliminated.