Olga Basso1,2. 1. Department of Obstetrics and Gynecology, McGill University, Montreal, QC, Canada. 2. Department of Epidemiology, Biostatistics, and Occupational Health, McGill University, Montreal, QC, Canada.
Abstract
BACKGROUND: Gestational-age-specific rates of postnatal endpoints are sometimes estimated with denominators based on fetuses-at-risk (FAR), rather than live births. However, as infants can only be included in the numerator after they are born alive, interpretation of such rates is problematic. METHODS: Using simple algebra it can be shown that, at each gestational week, FAR rates of postnatal endpoints are the product of the conventional risk of outcome among live births and the probability of live birth, which increases from near zero early in gestation to close to one in the final weeks. The consequences of such a pattern of live birth on FAR rates are further illustrated in hypothetical scenarios with known conditions. RESULTS: FAR rates of postnatal endpoints will generally increase towards the end of pregnancy due to the rising probability of live birth, regardless of the 'true' effect of immaturity on risk. In the presence of an exposure that increases the probability of early birth, the same mechanism will cause FAR rates to be higher in the exposed group, even if the exposure has no effect. CONCLUSIONS: Gestational-age-specific FAR rates of postnatal outcomes strongly depend on the probability of live birth. Thus, they reflect neither the causal effect of gestational length, nor that of a given exposure. Indeed, if an exposure shortens gestation, FAR rates will be higher in exposed infants even when the exposure has no impact on the outcome under study. These intrinsic limitations should be taken into account when applying FAR analyses to postnatal endpoints.
BACKGROUND: Gestational-age-specific rates of postnatal endpoints are sometimes estimated with denominators based on fetuses-at-risk (FAR), rather than live births. However, as infants can only be included in the numerator after they are born alive, interpretation of such rates is problematic. METHODS: Using simple algebra it can be shown that, at each gestational week, FAR rates of postnatal endpoints are the product of the conventional risk of outcome among live births and the probability of live birth, which increases from near zero early in gestation to close to one in the final weeks. The consequences of such a pattern of live birth on FAR rates are further illustrated in hypothetical scenarios with known conditions. RESULTS: FAR rates of postnatal endpoints will generally increase towards the end of pregnancy due to the rising probability of live birth, regardless of the 'true' effect of immaturity on risk. In the presence of an exposure that increases the probability of early birth, the same mechanism will cause FAR rates to be higher in the exposed group, even if the exposure has no effect. CONCLUSIONS: Gestational-age-specific FAR rates of postnatal outcomes strongly depend on the probability of live birth. Thus, they reflect neither the causal effect of gestational length, nor that of a given exposure. Indeed, if an exposure shortens gestation, FAR rates will be higher in exposed infants even when the exposure has no impact on the outcome under study. These intrinsic limitations should be taken into account when applying FAR analyses to postnatal endpoints.
Authors: Carmen Messerlian; Joseph M Braun; Lidia Mínguez-Alarcón; Paige L Williams; Jennifer B Ford; Vicente Mustieles; Antonia M Calafat; Irene Souter; Thomas Toth; Russ Hauser Journal: Environ Int Date: 2017-06-27 Impact factor: 9.621
Authors: Ifeoma Muoto; Blair G Darney; Bernard Lau; Yvonne W Cheng; Mark W Tomlinson; Duncan R Neilson; Steven A Friedman; Joanne Rogovoy; Aaron B Caughey; Jonathan M Snowden Journal: Health Serv Res Date: 2017-11-12 Impact factor: 3.402
Authors: Yu-Han Chiu; Mats J Stensrud; Issa J Dahabreh; Paolo Rinaudo; Michael P Diamond; John Hsu; Sonia Hernández-Díaz; Miguel A Hernán Journal: Epidemiology Date: 2020-09 Impact factor: 4.860
Authors: Amélie Boutin; Sarka Lisonkova; Giulia M Muraca; Neda Razaz; Shiliang Liu; Michael S Kramer; K S Joseph Journal: PLoS One Date: 2021-06-30 Impact factor: 3.240
Authors: Tim A Bruckner; Samantha Gailey; Abhery Das; Alison Gemmill; Joan A Casey; Ralph Catalano; Gary M Shaw; Jennifer Zeitlin Journal: BMC Pregnancy Childbirth Date: 2021-07-02 Impact factor: 3.007