To explore international differences in the classification of births at extremely low gestation and the subsequent impact on the calculation of survival rates.
We used national data on births at 22 to 25 weeks' gestation from the United States (2014; n = 11?144), Canada (2009-2014; n = 5668), the United Kingdom (2014-2015; n = 2992), Norway (2010-2014; n = 409), Finland (2010-2015; n = 348), Sweden (2011-2014; n = 489), and Japan (2014-2015; n = 2288) to compare neonatal survival rates using different denominators: all births, births alive at the onset of labor, live births, live births surviving to 1 hour, and live births surviving to 24 hours.
For births at 22 weeks' gestation, neonatal survival rates for which we used live births as the denominator varied from 3.7% to 56.7% among the 7 countries. This variation decreased when the denominator was changed to include stillbirths (ie, all births [1.8%-22.3%] and fetuses alive at the onset of labor [3.7%-38.2%]) or exclude early deaths and limited to births surviving at least 12 hours (50.0%-77.8%). Similar trends were seen for infants born at 23 weeks' gestation. Variation diminished considerably at 24 and 25 weeks' gestation.
International variation in neonatal survival rates at 22 to 23 weeks' gestation diminished considerably when including stillbirths in the denominator, revealing the variation arises in part from differences in the proportion of births reported as live births, which itself is closely connected to the provision of active care.