Opposing or Complementary Perspectives? Perinatal Outcomes, Causality, and Time Zero

Published In

Paediatric And Perinatal Epidemiology

Document Type


Publication Date



Perinatal epidemiology has a denominator problem. How big of a problem is it, though? And what is the best way to address it? The issue arises in the analyses of perinatal complications when associations are estimated conditional on gestational length (eg at a given gestational age or within preterm births). There is consensus regarding the appropriate denominator for events that occur before the onset of labour, such as antepartum stillbirth: foetuses, or ongoing pregnancies—a term that reflects that many such outcomes, such as preeclampsia—pertain to the mother or the pregnancy rather than the foetus per se. Controversy arises when considering postnatal outcomes.- Including all ongoing pregnancies in gestation‐stratified analyses of such outcomes is argued to more clearly reflect the fact that aetiological process often begins before birth. But doing so results in the inclusion of study units that are never at risk for incidence of the eventual neonatal outcome (ie owing to a competing risk later in gestation, eg, a stillbirth)., , For this reason, a denominator of live births at the index gestational age is argued to be a better analytical choice. In truth, neither denominator perfectly accounts for these various analytical considerations. It is this challenge Dr Joseph addresses in his article in this edition of Paediatric and Perinatal Epidemiology. In this commentary, I respond to Joseph's contribution, offer critiques of the points made, and provide deeper engagement with the issues of outcome definition, time at risk, aetiology, and question formulation.


© 2019 John Wiley & Sons Ltd

Locate the Document




Persistent Identifier