This retrospective cohort study aimed to determine incidence and timing of hypoglycemia in the first day of life in very premature infants in a tertiary NICU setting. We found that hypoglycemia is fairly common, occurring in nearly 1 out of 5 infants, and generally occurs early after birth. This study identified six risk factors for hypoglycemia in this high-risk population.
We found an incidence of hypoglycemia of 19%, which is relatively low as compared to previous studies reporting incidences up to 73% in preterm infants [1] [2] [3]. Differences in incidences may be explained by timing (time after birth) and route (peripheral or central) of IV glucose administration. Although previous studies provided no clear information on their policy, we speculate that our relatively low incidence may be explained by a protective effect of our local policy to start peripheral IV-glucose directly after stabilization in the delivery room. In our sample, lack of IV-glucose administration in the delivery room was associated with a 6-fold increase in the odds for hypoglycemia. In 8% of our patients, peripheral IV-access was not achieved in the delivery room. This likely illustrates practical difficulties of obtaining IV access in a subgroup of very preterm infants. We cannot rule out that infants in whom peripheral IV access was not achieved may have already been in a worse clinical condition, predisposing for or already including hypoglycaemia. We found no previous literature on timing of hypoglycemia in similar populations to compare our time-to-event of 126 minutes postpartum with.
With each week increase in gestational age we found an 1.3-fold increase in odds for hypoglycemia. This is in contrast to most studies in term and late preterm born infants, reporting that infants with the lowest GA are at highest risk [20]. We cautiously speculate that our contra-intuitive finding may be explained by the higher level of care provided to these infants. In our NICU all extremely preterm infants (i.e. below 26 weeks GA) are treated by experienced neonatologists and nurses, whilst the more mature infants are generally managed by care providers in training. As data on the care providers are lacking, we cannot rule out that these infants might be subjected to a less aggressive preventive care in terms of temperature management, respiratory support, IV-access etc., which might explain the increased risk for hypoglycemia.
Being small for gestational age was associated with 2.6 increased odds compared to those born AGA, which is comparable to the OR of 2.1 found in term born SGA infants in previous work [21]. On the other end of the scale, the LGA infants in our study showed a significantly increased risk in the univariable analysis, which did not persist in the multivariable model. This is in line with the results of several studies in term born infant showing that being LGA is not an independent risk factor for hypoglycemia [22, 23].
The fourth neonatal variable associated with hypoglycemia was respiratory support at NICU admission. Again counter-intuitive, we found a 2.3-fold increase in odds for hypoglycemia in infants without any form of respiratory support as compared to those on non-invasive respiratory support at NICU admission. To our knowledge, no other studies have investigated this relation in newborns. A possible explanation is that very preterm infants without respiratory support are not in favor by their better clinical state, but sooner exhaust their glycogen reserves due to increased work of breathing. Before drawing any conclusions, further research is needed, looking into metabolic and respiratory courses.
Insulin-dependent diabetes, but not non-insulin dependent diabetes, increased the odds for hypoglycemia nearly 3-fold as compared to non-diabetic mothers. This is explained by maternal hyperglycemia causing excess fetal and neonatal insulin production, which can lead to neonatal hypoglycemia [24]. In addition, prenatal corticosteroids within 72 hours before birth, increased the odds nearly 2-fold when compared to no corticosteroids. This is in line with previous studies in late preterms and support the hypothesis that antenatal corticosteroids administered shortly before delivery may increase neonatal insulin levels, leading to hypoglycemia [17].
To our knowledge, this study is unique in describing early hypoglycemia in very preterm infants and identifying risk factors for hypoglycemia. As compared to the current literature, we included a relatively large sample, with a rich dataset including frequent measurements and a broad spectrum of variables with very little missing values. This enabled us to run complex models whilst adhering to a minimum of 1:10 event-per-variable rate in a population, despite the relatively low incidence of hypoglycemia.
The main limitation of the current study is the retrospective design. In addition, because of the local interval-based screening protocol and possible variation in adherence to the protocol, it cannot be ruled out that short episodes of hypoglycemia in some patients may have been missed. Moreover, data were limited to the first 24 hours after birth and repeated hypoglycemic events were not studied. Outcome data on neurodevelopment or growth, to support the clinical relevance of early hypoglycemia, were outside the scope of this study. Finally, despite using broad inclusion criteria to increase generalizability to other NICU populations, the results from this mono-center study may be unique to this specific sample and local clinical practices. Although the current study is only descriptive, we cautiously suggest that lessons can be drawn from our data. First, it suggests a strong protective association of early intravenous access and continuous glucose infusion, which is a modifiable factor. Although causality cannot be inferred from our observational data, and comparative multicenter data are lacking, our data support the practice of starting continuous IV glucose infusion directly after cardiorespiratory stabilization, preferably before leaving the delivery room. Whether early central line insertion in the delivery room is preferred, in case peripheral access was unsuccessful, needs to be explored in a prospective clinical study. If prospective studies confirm our findings, the timing of IV access and glucose administration may deserve a place in the neonatal resuscitation algorithms. Second, more data are needed to explain the increased risk of hypoglycemia in infants with a higher gestational age and in those without respiratory support. These results are counter-intuitive, and may be explained by differences in level of care to the disadvantage of the more mature infants, as their generally better condition at birth did not seem to protect them from early hypoglycemia.
This explorative study provides the first step towards the development of models to predict hypoglycemia and improve clinical decisions-making. We suggest that these should cover a longer period after birth to include possible recurrence of hypoglycemia and should preferably be based on multi-center cohorts to explore and compare different policies. Prediction models and evidence based preventive measures are needed for new clinical guidelines on glucose screening in very preterm infants, with a better balance to reduce the toll of testing within safe limits.