The Canadian Pediatric Society has just published new guidance for screening and treatment of infants at risk for neonatal hypoglycaemia. The older statement needed to be revised, in particular to include the use of oral glucose gel as an option, which had not been studied when the previous version was developed.

Prior to birth the infant receives a continuous supply of nutrients, including glucose, across the placenta. Immediately at birth, with interruption of this trans-placental supply, glucose concentrations fall and stimulate counter-regulatory mechanisms which take some time to affect glucose concentrations, and thus a fall in glucose concentrations during the first hours of life is a normal phenomenon.

Over 30% of newborn infants are considered to be “at-risk” as 10% of them are SGA, below the 10^th percentile, 10% are LGA, about 10% are preterm, and 16% are born to mothers with gestational or pre-gestational diabetes.  There is some overlap, of course, so the total number of babies who are “at-risk” isn’t immediately clear to me, but it’s a lot.

During this hypoglycaemic slump glucose concentrations often fall to levels below those which cause symptoms in adults and older children. During this period some term infants, and few preterm infants, produce ketone bodies, here is an older figure from an observational study among healthy singleton infants, excluding infants of diabetic mothers and the SGA, published in 1992 (Hawdon JM, et al. Patterns of metabolic adaptation for preterm and term infants in the first neonatal week. Arch Dis Child. 1992;67(4 Spec No):357-65.). Ketone bodies here refers to the sum of acetoacetate and β-hydroxybutyrate, as you can see among term infants with hypoglycaemia, many do not produce a lot of ketone bodies, and none of the preterm infants do either.

So the previous dogma, that hypoglycaemia is more dangerous among infants with hyperinsulinaemia because they don’t produce ketone bodies, is probably not true, babies without hyperinsulinemia don’t produce them reliably either! It may well be that other fuels are used by the neonatal brain during hypoglycemia, such as possibly lactate.

At what level does asymptomatic hypoglycemia become dangerous? The best evidence would come from RCTs (of which more in part 2) and absent such evidence then the second-best would be from scrupulous prospective cohort studies with prospectively determined glucose sampling, using reliable glucose measurement methodology (i.e. not bedside test strips), and surveillance for both short and long term impacts.

One observational study which is often quoted (Duvanel CB, et al. Long-term effects of neonatal hypoglycemia on brain growth and psychomotor development in small-for-gestational-age preterm infants. The Journal of Pediatrics. 1999;134(4):492-8) showed that multiple repeated episodes of a blood glucose under 2.6 mmol/L (on 6 occasions or more) were associated with smaller head circumference and motor delay among SGA preterm babies with an average birth weight of 1.1 kg, however, the majority of babies in that study with glucose <2.6 mmol/L mmol/L also had at least one blood glucose <1.6 mmol/L, and fewer episodes of glucose <2.6 were not significant. Other quoted studies, such as the classic study by Lucas in 1988 showed that persistently low blood glucose over several days may lead to problems, but his study, exclusively in preterm infants, showed a significant impact only when there were at least 5 days with values <2.6 mmol/L. In that study, the result was derived from a secondary analysis of blood glucose values collected for other reasons over the first several weeks of life, and in centres with variable treatment thresholds, which ranged between 1.5 and 2.5 mmol/L.

Another recent large cohort had routine, protocol-specified, glucose monitoring and was designed to prospectively evaluate a group of at-risk infants (McKinlay CJ, et al. Neonatal Glycemia and Neurodevelopmental Outcomes at 2 Years. NEJM. 2015;373;1507-18). The infants were followed to 2 years with tests of development (BSID-3) and executive and visual function. All the babies were ‘at-risk’ as defined by being over the 90^th%le, under the 10^th %le, an infant of a diabetic mother or preterm. and followed a protocol to intervene for low blood sugars at 2.6 mmol/L, but some were nevertheless found to have “severe hypoglycaemia” < 2.0, and some had multiple episodes of low sugar. That study did not show any additional risk of low blood sugars and they noted “Hypoglycemia… was not associated with an increased risk of the primary outcomes of neurosensory impairment (risk ratio, 0.95; 95% confidence interval [CI], 0.75 to 1.20; P=0.67) and processing difficulty, defined as an executive-function score or motion coherence threshold that was more than 1.5 SD from the mean (risk ratio, 0.92; 95% CI, 0.56 to 1.51; P=0.74). Risks were not increased among children with unrecognized hypoglycemia (a low interstitial glucose concentration only). The lowest blood glucose concentration, number of hypoglycemic episodes and events, and negative interstitial increment (area above the interstitial glucose concentration curve and below 47 mg per deciliter) also did not predict the outcome”  (47 mg/dl=2.6 mmol/L). In contrast, they did show an association of rapid correction of low glucose with adverse outcomes, babies whose blood glucose rose faster after they were treated for a blood glucose < 2.6 mmol/L had more neurosensory impairment. The range of blood sugars among these babies was a low as 0.5 mmol/L in the 1^st 12 hours, and as low as 1.0 thereafter.

It seems that this group of “at-risk” infants, who are at some increased risk of lower blood sugars (in particular those who are small for gestational age) have an increased prevalence of developmental difficulties. The same article, just referred to, showed that the at-risk infants in the study had substantially lower scores of Bayley testing than population norms, whether or not they had episodes of blood glucose <2.6. That association may be because of the impacts of growth restriction or maternal diabetes; there is no good evidence that mild to moderate transiently low blood sugars contribute to that risk, indeed I can’t find any reliable evidence of that.

In the pivotal ‘sugarbabies’ trial, (Harris DL, et al. Outcome at 2 Years after Dextrose Gel Treatment for Neonatal Hypoglycemia: Follow-Up of a Randomized Trial. J Pediatr. 2016;170:54-9 e1-2) at-risk babies whose blood sugar fell below 2.6 mmol/L were randomized to receive oral glucose gel immediately, or they received placebo and were treated later if the blood sugar remained below 2.6 mmol/L thirty minutes later. The babies in the two groups had identical outcomes at 2 years of age, including neurodevelopmental assessment even though the placebo group, by protocol design, had a longer period of time with a blood sugar less than 2.6 mmol/L.

There is, therefore, no good observational data to show what threshold of low glucose levels might lead to developmental problems. Severe, prolonged, or repeated low blood sugars in some studies seem to be associated with poorer outcomes, but some studies can’t even show that.

There is also the observation from the CHYLD cohort that low sugars which recover faster are associated with more long term problems.

Hypoglycaemia could be defined as a statistically low glucose, as a glucose which causes short term CNS dysfunction, or as a glucose level which causes permanent CNS injury, or as a blood glucose level below which treatment improves outcomes. It seems unlikely that there is a single threshold that applies to all these potential definitions. I think the most practical way to define a threshold for hypoglycemia would be to find a threshold below which intervention is required in order to reduce adverse outcomes. In other words, to perform RCTs (which should not be difficult in such a common problem affecting 1/3 of all newborn babies).