Discussion Questions

• ■
  

  Were the objectives of the study clearly defined?

  
  
• ■
  

  What were the key study findings?

  
  
• ■
  

  What are the strengths of the study?

  
  
• ■
  

  Can you identify potential sources of bias?

  
  
• ■
  

  Are the study results applicable to your clinical practice?

  
  
• ■
  

  How might you address the study question differently?

Introduction

Timing delivery so that the best possible outcome is achieved remains a critical issue in contemporary obstetrics. Preterm birth, a major cause of infant morbidity, is also responsible for over one-third of infant deaths in the United States, and its prevalence continues to climb. This trend is partly due to an increase in indicated or elective births beyond 32 weeks, especially late preterm births at 34 weeks’ gestation and above. Early preterm births, also on the rise, are particularly likely to result in adverse neonatal outcomes. Initiatives to optimize gestational age at delivery can contribute substantially to improved neonatal and infant outcomes.

Luisa A. Wetta, MD, Alan T. N. Tita, MD, PhD, and Todd R. Jenkins, MD, Associate Editor

Study Design

Tita: Welcome to the American Journal of Obstetrics & Gynecology Journal Club. Today, we are discussing an article by Donovan et al. Were the purpose and objectives of the study clearly defined?

Erwin: They were. Overall, the purpose was to generate data for evaluation of an Ohio public health initiative designed to reduce prematurity-related adverse health outcomes. The objective, more specifically, was to determine infant mortality rates adjusted by gestational age at delivery. So we’re really looking at how the rates vary across different gestational ages.

Tita: The study obviously focused on the population in Ohio. What were the key study findings, and were all relevant outcomes considered?

Doss: The study showed that adjusted infant mortality rates decreased as gestational age increased from 32 weeks to 40 weeks (12 per 1000 at 32–33 weeks vs 2 per 1000 at 39–40 weeks). This is not surprising—the lower the gestational age, the higher the infant mortality rate. The authors also assessed time to mortality, and the earlier the gestational age, the sooner infant mortality occurred. However, they did not look at fetal deaths or how that would affect the overall perinatal/infant mortality outcome. Although it’s likely a small number of fetal deaths occurred, it would be useful to know that reduced infant mortality associated with delivery at a later gestational age was not outweighed by fetal deaths. They also did not look at infant morbidity due to respiratory distress syndrome and other major infant morbidities.

Tita: In your view, aside from fetal deaths and neonatal morbidity, were there other relevant outcomes that were not considered?

Doss: No. Perhaps these were not available in their data source.

Tita: What are the strengths of this study?

Bates: First, the study was a large population-based study involving over 400,000 subjects. In addition, the authors were able to look at over 90% of the available target population, with very few subjects excluded.

Tita: Yes, they were unable to include about 0.6% of the population because age of death was unavailable. What about potential sources of bias and their likely impact on study results?

Bates: In my view, there were 2 major sources of bias. Of course, with a retrospective database study of this nature, information bias is likely for a range of variables, including gestational age. Bias in gestational age data could influence study results in either direction—probably the adjusted effects would be underestimated if such errors occurred randomly.

Confounding by additional risk factors for infant death and early delivery that had not been controlled for could also introduce bias. Also, infants with birth defects were included, and their risk for infant death was disproportionately higher—excluding them or perhaps stratifying results by absence of birth defects might have been preferable.

Tita: You are suggesting that, since mortality among infants with birth defects was so high compared with the rest of the study population, this group be looked at separately?

Bates: Yes.