Investigators have followed a cohort of children who experienced a relatively common situation, neonatal hypoglycemia, in an attempt to quantify long-term outcomes. They shed some more light on this entity, but I'm always a bit skeptical when researchers rely heavily on post hoc analyses.
Source: McKinlay CJD, Alsweiler JM, Anstice NS, et al. Association of neonatal glycemia with neurodevelopmental outcomes at 4.5 Years. JAMA Pediatr. 2017; 171:972-983. doi: 10.1001/jamapediatrics.2017.1579. See AAP Grand Rounds commentary by Dr. Patricia Fechner (subscription required).
Neonatal hypoglycemia likely causes some degree of neurologic injury, but quantifying the risk has been difficult and has left practitioners with little agreement on how to manage such infants. A better understanding of outcomes could inform decisions such as at what glucose level to intervene, and then exactly how quickly to correct the hypoglycemia. Investigators in New Zealand originally reported no significant association of hypoglycemia with neurodevelopmental outcomes at 2 years of age, but now they have further follow-up data for us as the original cohort has reached 4.5 years of age. Again, they found no association of hypoglycemia (defined as blood glucose less than 47 mg/dL) with neurodevelopmental outcomes at this age. However, a post hoc analysis did show an association of hypoglycemia with impaired executive function and visual motor function. (Remember, a prospective cohort study can only demonstrate associations of various exposures to outcomes; it cannot prove causation.)
What's the deal with post hoc analysis? I've mentioned this occasionally in these pages, including a little over a year ago. It refers to data analyses not originally planned when the study was designed, and because it can be a form of cherry-picking the data that violates some assumptions underlying standard statistical testing, it should be interpreted with caution, and with further statistical analysis that can correct for the cherry-picking to some extent. In the current study, the authors performed a separate, post hoc analysis on subgroups of the infants, based on the severity and frequency of hypoglycemia they experienced as newborns. They did find some evidence that these factors were associated with worse outcomes, a form of dose-response effect that supports (but again, does not prove) a causal role. This effect is seen when the outcome of interest (such as executive function) is more common and/or more severe in children who experienced a "higher dose" of the toxin or adverse event, e.g. a greater degree of hypoglycemia as newborns. However, I had a hard time determining whether the authors subjected their post hoc analysis to more critical statistical analysis; it appears they did not.
The authors pointed out some other limitations of their study, probably most significantly the fact that only 78% of children originally in the study were available for testing at 4.5 years. It's hard to fault the authors for this, of course; it's pretty tough to keep a cohort together for multiple-year follow up. One hopes they will keep following these children, to let us know if this finding persists into the school-age years.