Neonatal glucose control continues to be challenging for intensivists, in particular in high-risk neonates. Extremely low birth weight infants (ELBW), infants <1000 grams, are at the highest risk of aberrant glucose control and can have long-term adverse consequences particularly affecting their neurodevelopment. We will be discussing the most common dilemmas encountered in the Neonatal Intensive Care Unit (NICU): What are the ranges for euglycemia?; Are there differences in glycemic response for preterm and term infants?; What about depending on day of life?; What is the best approach for glucose measurements/monitoring?; What are the best strategies to control parenteral glucose delivery?


According to the Pediatric Endocrine Society (PES), plasma glucose (PG) should be >50 mg/dL in high-risk neonates at <48 h of life (HOL) and >60 in high-risk neonates at >48 HOL [1]. These thresholds are derived from studies looking at the neuroendocrine response of newborns as well as risk factors for persistent hypoglycemia. A recent study (GLOW) showed that breastfed, term infants may have PG values as low as 40 mg/dL during the first 48 HOL and these values increase to about 60 mg/dL by 96 HOL [2]. In contrast, the 2011 AAP recommendations suggest neonates at risk who require immediate treatment with intravenous (IV) glucose are those with glucose levels <40 mg/dL and symptoms of hypoglycemia (jitteriness, apnea, cardiac arrest, cyanotic spells, hypothermia, lethargy, seizures, irritability) [3]. All of these recommendations are derived from studies in healthy term neonates; whether preterm, LGA/SGA babies, infants of diabetic mothers (IDM) and those who are critically ill have the same thresholds to prevent adverse neurological long-term effects remains unknown [3,4,5,6,7]. The physiology of glucose uptake in the neonate depends on many factors and should be taken into account as a predictor for euglycemia as shown in Table 1.

Table 1 Physiology unique to neonates as compared to older children and adults.

Methods for glucose measurement/monitoring

Each method of collection has differing attributes regarding accuracy of results due to the reporting error of the collection device. Neonatal blood glucose changes rapidly, and if the time between collection and result is delayed, the clinical implications can be significant. Initial screening should utilize methods of minimal blood sampling and be paired with venous/capillary sampling, even though neonatal blood sample validation has only been performed in arterial samples [8]. We need to consider arterial blood will have glucose levels that are 10–15% higher than venous samples whereas capillary samples are between these two values [9]. Other factors can also skew the accuracy including postprandial states, decreased perfusion, acidosis, and polycythemia; hematocrit has the largest impact of all variables measured [10,11,12,13]. Point of care (POC) glucose meters are frequently the first line in glucose measurement as results are available within seconds. Glucose meters were designed for adults with diabetes and therefore, the goal was to provide glucose ranges rather than absolute accuracy. There is a strong correlation of POC devices and laboratory equipment within the normoglycemic and hyperglycemic range, however there is significant deviation at hypoglycemic concentrations [14]. More recent glucometers use glucose oxidase or glucose dehydrogenase which is an enzyme reaction which generates a current that is measured by the meter. The size of the current is proportional to the amount of glucose in the blood sample and therefore polycythemia and anemia affects the accuracy of the results. In these cases, POC’s need to be measured in the laboratory via the glycose oxidase method to confirm accuracy of results [14]. Additionally, there is an anticipated 10–15% increase of plasma glucose when compared to whole blood glucose with variations from actual levels as much as 10 to 20 mg/dL (0.55 to 1.11 mmol/ L) [14,15,16]. Although this does limit the accuracy of results, a positive factor is the reduction of blood volume and time of sending a sample to the laboratory. A POC blood gas analyzer with glucose module works to minimize this inaccuracy (100 ul POC vs 300 ul glucose oxidase analyzer). This method retains the quick result time with the added benefit of improved result accuracy and minimal blood loss. These modules use a multilayer membrane with glucose oxidase which allows for a more accurate test similar to the method used in a clinical laboratory [9]. However, there is increased cost associated with these devices and they are rarely available in the NICU setting as they are only used in the research environment.

Plasma glucose oxidase measurement is the gold standard due to its high accuracy, but its utility is dependent on the speed of analysis. The difficulties are due to variations in clinical laboratory workflow processes such as turn-around time which can be 30 min (average), but can range up to 2 h depending on the laboratory [17,18,19,20,21]. Interventions based on a patient’s previous clinical state rather than the current state may compromise care, and one must decide when to act quickly and/or wait for accuracy. To assess the validity of the sample and therefore treatment decision, it is important to utilize both plasma and whole blood measurements and to follow each trend. Trending rapidly available samples allows an ability to avoid delayed laboratory results and delayed actions (i.e.: changing IV glucose rate and/or feeds) [17].

Monitoring, how often?

Glucose monitoring is dependent on the timing of treatment and trends of glucose measurements. A common approach is to check pre-prandial whole glucose every 3 h. If rapid glucose fluctuations are occurring (increasing/decreasing by 50–100 points) or the glucose concentration is in a potentially dangerous zone (<30 mg/dL), close monitoring every 30–60 min is recommended until glucose stabilizes, particularly after the first 4 h of life. Once glucose has stabilized, glucose checks can be safely spaced to every 3–6 h.

Another strategy for glucose monitoring is subcutaneous glucose monitoring, which allows for continuous monitoring as well as data trending. One challenge is that continuous glucose monitors (CGM) must be regularly calibrated to ensure accuracy, requiring continued blood samples from the neonate [22]. However, the ability to have early detection of hypoglycemia and the ability to prepare therapeutic interventions even earlier shows there is significant potential for this device in the future. Further study and evaluation of efficacy is needed as these devices require sufficient subcutaneous fat for measurement which is often limited in the neonate [9, 23].

Older studies comparing CGM with arterial sampling did not differ significantly from one another in terms of hypoglycemia or hyperglycemia and high sensitivity/specificity were demonstrated for CGM [24, 25]. However, newer studies showed that CGM improved glucose monitoring with potential to have increased control [26]. In one study, 33.33% of preterm infants experienced hypoglycemia that would not have been captured by intermittent arterial sampling within the first days of life [24]. One randomized trial compared real-time CGM with intermittent blood glucose monitoring in very low birthweight infants and found that CGM reduced the duration of hypoglycemic episodes by 50% and the number of capillary blood samples by 25% [27,28,29].

Despite the ability to study early trends in neonatal glycemia in greater detail, CGM does present some drawbacks. In addition to random errors, CGM devices exhibit a drift component when there is deviance of reported glucose concentrations from true values between calibration measurements [22]. Another limitation is the time lag that occurs waiting for glucose to diffuse from blood to the interstitial fluid. CGM cannot be relied upon for point accuracy, particularly when looking for outlying glucose measurements and needs further studies in several populations prior to implementation in NICU’s.

Treatment strategies

Parenteral glucose

If an infant is hypoglycemic and enteral intake cannot adequately treat their hypoglycemia, then the infant is treated with IV glucose. This is commonly encountered in the first few days after birth. If symptomatic hypoglycemia or significant asymptomatic hypoglycemia is found, the initial treatment consists of 200 mg/kg (2 mL/kg) of 10% dextrose in water administered over one minute. Recent data suggests more invasive diagnostic and treatment interventions may have deleterious effects, and therefore avoiding boluses in asymptomatic cases might be preferred, particularly if hypoglycemia is not severe [30]. Due to uncertainty in etiology (glucose variability versus severity of illness), there is limited guidance regarding need for administration of a bolus. For hypoglycemia, a continuous glucose infusion (GIR) should be administered at a minimum of 6–8 mg/kg/min and titrated up by 1–2 mg/kg/min every 4–6 h until euglycemia is achieved, after which titration down can occur [31]. Other interventions may be necessary if hypoglycemia persists despite a GIR exceeding 12–14 mg/kg/min. The rate at which GIR increases occur depends on the infant’s risk factors, medications, and projected needs/expenditures. If a neonate is hypoglycemic and unable to feed, the GIR should be increased until glucose stabilizes (there is no maximum number, the author has anecdotally supported hyperinsulinemic infants with GIR’s as high as 30 until other medications take effect or pancreatectomy is performed).

Hyperglycemia typically occurs within the first week of life. It is more common in infants who are low birth weight with the most profound effects on ELBW infants. ELBW infants may have insulin resistance (IR) from prematurity, therefore insulin infusions are typically unpredictable, if effective at all. The skeletal muscle and adipose tissue of preterm animals have impaired insulin signaling (decreased insulin receptor and phosphorylation of Akt) which in turn, causes peripheral IR and decreased insulin-stimulated glucose disposal [32, 33]. These developmental impairments improve after the first week of life and glucose titration can be somewhat liberalized as insulin sensitivity improves. However, careful titration of glucose infusions is necessary, especially during the first week of life, to avoid consequences of glucotoxicity to pancreatic, hepatic, muscular, retinal, neuronal, and other organ cell types that could be affected at critical periods of development [34,35,36,37]. Due to the potential deleterious effects of hyperglycemia and inability to increase glucose disposal into the muscle to promote growth, providers must limit glucose infusions, even though maintaining strict euglycemia is often quite challenging and labile. Due to peripheral IR, the infant’s skeletal muscle cells cannot utilize glucose to promote growth even if it is provided; Therefore providing additional calories may only promote end-organ damage but not growth. The ideal scenario is to provide glucose via enteral feeds. Enteral feeds stimulate incretin and insulin secretion naturally, promoting optimal growth that can be achieved with limited side effects; however, due to gut immaturity/dysmotility, some infants are unable to tolerate rapid feeding advancements.

Enteral feeds should be used first if possible, and hyperglycemia should be avoided regardless of glucose infusion rate. The necessary IV glucose infusion rate should be estimated based on factors such as gestational age (GA) and severity of illness. Intravenous amino acid infusions and intralipids must be added to the glucose solution via parenteral nutrition to avoid hypotonic solutions and improve nutrition [38]. Low concentrations of certain amino acid in IV solutions, particularly arginine and glutamine, had increased risk for hyperglycemia, potentially due to its role in insulin secretion [39]. The ideal scenario would be to add medications that would modulate insulin signaling such as AMPK activators, thiazolidinones and sulfonylureas. Unfortunately, these medications have not been tested in neonates, with current available research limited to animal models.

High plasma glucose has been shown to affect several systems of the body via glucotoxicity and increased oxidative stress, particularly the pulmonary, endocrine, and neurological systems. Hyperglycemia occurring within the first 24 h after birth was identified as a risk factor for reduction of white matter or death in preterm infants at term equivalent age [35]. Similarly, studies have identified morbidities associated with hyperglycemia such as retinopathy of prematurity are increased by as much as 4.5 fold in a predominantly Hispanic population of preterm infants [34]. If prematurity is coupled with hyperglycemia, there is an increased risk of oxidative stress in the premature kidney. The oxidative stress via the polyol pathway is thought to disrupt gene expression during neural tube formation. Exposures at critical periods of development disrupt kidney growth and even lead to apoptosis [36]. The kidney is at particular risk for disrupted development in the preterm neonate as nephrogenesis is still ongoing after birth [40]. The preterm pancreas is also at utmost risk of glucotoxic injury as it continues to have pluripotential cells as late as 2/3 gestation. The effects of early hyperglycemia and the differentiation of these cells into the beta cells of the endocrine pancreas remains unknown [41]. An example of a sliding scale to maintain euglycemia using the euglycemic and hyperglycemic clamp techniques in premature and term baboons is shown in Tables 2 and 3 [33]. These sliding scales may be an adjuvant to predict glucose concentrations, but a central line is imperative. If insulin is used for persistent severe hyperglycemia despite low glucose infusion rate strategy for 12-24 hrs, a sliding scale is provided from current experience (Table 3).

Table 2 Glucose sliding scale for acute hypo/hyperglycemia management.
Table 3 Insulin titration for hyperglycemia.

Infant Feeding Treatment

Feeding and neonatal euglycemia are closely linked; Timing, duration, and type of enteral feeding all affect neonatal glucose levels. Breastfeeding has been well established as a treatment for hypoglycemia and reduces the need for other forms of treatment [42]. In a study of prevention of hypoglycemia, the change in plasma glucose concentration after breastfeeding was similar when compared to the use of formula or dextrose containing gel. Independent of the initial blood glucose concentration, breastfeeding was associated with reduced need for a second treatment [43]. Additionally, certain at-risk populations may benefit the most from careful glycemic control. Breastfeeding may have a slower but more sustained effect on PG concentrations than either infant formula or DexGel. IDMs who were breastfed within the first 30 min of life had higher plasma glucose concentrations than IDMs who were not fed or received infant formula [43]. Additionally, those infants who were breastfed for >20 min had less hypoglycemia in the following 8 h [44].

Other pharmacological interventions

Other options beyond feeding and IV dextrose for the management of hypoglycemia and to support other therapies should be considered. Glucocorticoids such as hydrocortisone reduce peripheral utilization of glucose and enhance the effects of glucagon and stimulate gluconeogenesis in the liver and kidneys. Glucocorticoids can also be given as medications in dosage equivalents similar to the body’s own stress response and may stabilize hypoglycemia without causing hyperglycemia. Other medications also can have a significant role in the management of hypoglycemia with hyperinsulinemia. The best option in situations of hyperinsulinemia is diazoxide as it works by activation of potassium channels in the pancreatic cells, thereby suppressing insulin release [8]. Glucagon can be used in episodes of moderate to severe hypoglycemia in particular for rescue treatment; this therapy will not work in the presence of a glycogen storage disease. Octreotide can be used alone or in conjunction with glucagon therapy to suppress both insulin and glucagon output [8]. Additionally, in the extreme scenario of a patient who is unresponsive to hydrocortisone and/or diazoxide a trial dose of epinephrine can be used if the patient is borderline hypotensive. Epinephrine acts through multiple pathways to increase glucose levels including promoting glycogenolysis, gluconeogenesis, adipose lipolysis, and glucagon secretion. In a particular subset of infants with adenomatous pancreatic hyperplasia in which medical therapy fails, surgical intervention with a partial pancreatectomy may be the only option remaining for persistent hypoglycemia [45].

Dextrose gel

Multiple studies have examined the use of dextrose gel in the treatment of hypoglycemia. In 2016, a double-blind study was performed of at-risk late preterm and term infants. Infants were randomized to receive 200 mg of 40% DexGel versus placebo for hypoglycemia prophylaxis; NICU admission rates for hypoglycemia were decreased after administration of dextrose gel, however the diversity of the population was limited, consisting of only healthy term infants of white race with a low incidence of maternal diabetes [43]. Other studies have been performed assessing potential dosages to administer of dextrose gel. All neonates given dextrose gel developed hypoglycemia regardless of dose, however the onset was delayed by 1.4 h compared to infants who were given the placebo [46]. Administration of dextrose gel may also differ in at-risk populations and dextrose gel may cause a hyper-insulinemic response in some patients such as IDMs. In the U.S., a quasi-experimental study was performed with 236 asymptomatic, at-risk infants and showed that one dose of prophylactic dextrose gel (77% dextrose) did not result in differences in glucose concentrations nor NICU admissions [47]. A multicenter trial of at-risk infants, DexGel did not decrease NICU admission rate nor treatment for hypoglycemia; DexGel had a transient effect on blood glucose concentration at 2 h of life, however, infants receiving DexGel had a higher blood glucose concentration at baseline. Infants had similar rates of breastfeeding throughout their hospital stay [48]. Given the lack of benefit DexGel should not be used as it increases health care utilization and does not provide any benefit. Additionally, DexGel as a treatment in infants with hyperinsulinemia should be avoided as it might trigger further insulin secretion due to variation in glucose availability.


Glucose control in the neonatal period can be challenging. Understanding the mechanisms behind the pathophysiology of each neonatal condition is of utmost importance to be able to predict and adjust IV glucose. Utilizing the appropriate methods and choosing the right site and frequency for glucose monitoring are key to achieving and maintaining euglycemia. If possible, enteral nutrition should be utilized as an adjuvant for optimal glucose management. Other therapies are rarely needed unless faced with hyperinsulinemic conditions. Significant research gaps remain to understand the impact of asymptomatic hypo/hyperglycemia in high-risk populations.