Article Text

Download PDFPDF

Management of hyperglycaemia in the preterm infant
  1. A L Ogilvy-Stuart1,
  2. K Beardsall2
  1. 1Neonatal Unit, Rosie Hospital, Cambridge University Hospitals NHS Trust, Cambridge, UK
  2. 2Department of Paediatrics, University of Cambridge, Cambridge University Hospitals NHS Trust, Cambridge, UK
  1. Correspondence to Dr Amanda L Ogilvy-Stuart, Neonatal Unit, Rosie Hospital, Cambridge University Hospitals NHS Trust, Cambridge CB2 2SW, UK; amanda.ogilvy-stuart{at}


In the fetus, the predominant energy supply is glucose transported across the placenta from the mother. As pregnancy progresses, the amount of glucose transported increases, with glycogen and fat stores being laid down, principally in the third trimester. In the well-term baby, there is hormonal and metabolic adaptation in the perinatal period to ensure adequate fuel supply to the brain and other vital organs after delivery, but in the preterm infant, abnormalities of glucose homeostasis are common. After initial hypoglycaemia, due to limited glycogen and fat stores, preterm babies often become hyperglycaemic because of a combination of insulin resistance and relative insulin deficiency. Hyperglycaemia is associated with increased morbidity and mortality in preterm infants, but what should be considered optimal glucose control, and how best to achieve it, has yet to be defined in these infants.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

In the fetus, there is a constant infusion of glucose across the placenta, mediated by facilitated diffusion via insulin-independent glucose transporters with fetal glucose levels approximately 70% of maternal levels.1 Fetal glucose requirements increase during pregnancy, and increased placental transport of glucose is facilitated by both the increased maternal glucose production and the increased expression of the placental glucose transporters in late pregnancy. In a healthy pregnancy, the fetal liver does not produce glucose; the uptake from the placenta is matched by fetal utilisation. Forty per cent of glucose taken up by the fetus is converted to either glycogen in the liver and muscle or to lipid for storage. Liver glycogen and adipose tissue storage occurs mainly in the third trimester, with glycogen stores two to three times the adult levels and infants having a fat mass of 16% at term.2 Insulin does not cross the placenta, but fetal pancreatic insulin secretion is stimulated predominantly by fetal glucose and also by amino acid levels. Fetal insulin levels increase in the last trimester, promoting glucose uptake and protein synthesis and is critical for fetal growth in late gestation. Insulin also regulates insulin-like growth factor-I (IGF-I) and IGF-binding protein-1 levels with indirect effects on growth by increasing cellular availability of glucose.

When the umbilical cord is clamped after birth, there is abrupt interruption of the transplacental glucose infusion and the healthy newborn baby has to rapidly adapt to withstand an initial period of fasting when there is mobilisation of the stored glycogen and fat followed by “bolus” oral feeding, receiving a low-carbohydrate, fat-rich energy source in the form of milk. Immediately after delivery, there is a catabolic period when insulin levels fall and glucagon, cortisol and catecholamine levels rise. Growth hormone levels are high. Glycogenolysis is critical in maintaining glucose levels immediately after delivery, but stores are depleted after 12 h when gluconeogenesis becomes essential in maintaining glucose homeostasis if there is continued fasting. The low blood glucose levels and high cortisol levels stimulate hepatic glucose-6-phosphatase activity, and reversal of the insulin/glucagon ratio with reduction in insulin levels induces phosphoenolpyruvate carboxykinase (PEPCK), the rate-limiting enzyme in hepatic gluconeogenesis. Phosphoenolpyruvate carboxykinase levels rise steadily over the first 2 weeks of life, regardless of gestation. These adaptations increase hepatic glucose production, maintaining glucose supply to vital organs in the newborn.

In the newborn, growth hormone levels are high, with pulse characteristics similar to that seen during fasting, and it is thought to play a metabolic role with secretion related to insulin secretion and entrained on feeding; the growth hormone response to hyperglycaemia being a signal for protein synthesis. These secretory patterns may indicate a role for growth hormone in the regulation of lipolysis. With the introduction of oral feeds, the neonatal pancreas needs to respond with intermittent insulin secretion to maintain glucose homeostasis and to promote postnatal growth. In the fetus, there is a balance among β-cell replication, neogenesis and apoptosis. After delivery, there is a wave of β-cell apoptosis with development of a new population of β-cells. In adults, incretins play an important role in augmenting postprandial insulin secretion, and although the role of incretins has not been well studied in neonates, the secretion of incretins may be important in the development of this new cell population that is non-proliferative and better suited to postnatal metabolic control.3 Although glucose levels at term vary greatly compared with adults (ranging from 2.3 to 5.8 mmol/l),4 they do not usually rise to >7 mmol/l.5

Diagnosing hyperglycaemia

The optimal methods of measurement and thresholds for diagnosis of hyperglycaemia are still debated. Typically, blood glucose levels are monitored intermittently at variable intervals, with samples taken from capillary, venous and arterial sites, in babies thought to be at risk of abnormalities in glucose homeostasis. There are therefore often prolonged periods when glucose levels are unknown. It is also important that only accurate and validated methods that utilise an enzyme reaction such as glucose-6-phosphatase or glucose oxidase are used to measure glucose levels. Cot-side reagent strips and blood gas analysers provide rapid means of glucose measurements on small volumes of blood but show marked variability with laboratory levels particularly at high and low glucose levels. Whole blood samples sent to the laboratory must be collected into tubes containing fluoride and ideally on ice, and although considered the gold standard, if there are delays in processing samples, results can be inaccurately low. It must also be remembered that whole blood glucose levels are approximately 10% to 15% lower than plasma levels and are affected by the haematocrit.

The definition of hyperglycaemia remains controversial, as hyperglycaemia has historically been viewed as part of the normal physiological stress response. Glucose levels in utero and in the term newborn rarely rise to >7 mmol/l and some consider this should be the physiological threshold for the definition of hyperglycaemia, but most clinicians would not consider clinical intervention unless glucose levels were >10 mmol/l. It is, however, increasingly believed that the identification of hyperglycaemia is important as an early sign of sepsis and with implications for management of fluid balance and because of its association with increased mortality and morbidity.6,,10

The preterm infant1

In the preterm baby, there remains a lack of evidence for the definition of both hyperglycaemia and hypoglycaemia, with commonly used glucose values being extrapolated from limited data available from term babies. In the preterm infant, the normally accepted levels for hypoglycaemia (<2.6 mmol/l) and hyperglycaemia (>10 mmol/l) occur frequently, particularly in infants born <30 weeks or weighing <1500 g at birth (very low birth weight (VLBW)). Hypoglycaemia tends to occur immediately after birth, because of low or absent glycogen and fat stores, combined with high-energy demands due to respiratory distress, sepsis, hypoxia and the difficulties of maintaining normothermia with a relative large surface area. Maturation of gluconeogenic pathways may be delayed and counter-regulatory hormone responses are impaired. These babies are particularly vulnerable as the lack of fat results in increased proteolysis to provide substrate for glucose production. The lack of fat also limits ketogenesis. Hence, these babies have low blood glucose and they are unable to produce alternative metabolic fuels, and our arbitrary definition of hypoglycaemia of 2.6 mmol/l should probably be higher (possibly 3 or 3.5 mmol/l) if we are to ensure adequate energy supply to the preterm brain. However, once an exogenous glucose supply is established, hypoglycaemia is uncommon.

During the first week of life in the preterm infant, early hypoglycaemia is often followed by hyperglycaemia, with hyperglycaemia being 18 times more common in babies weighing <1000 g than in those weighing >2000 g.11 The prevalence of hyperglycaemia is between 20% and 86% depending on definition.12 The cause is multifactorial, the result of both insulin resistance and relative insulin deficiency. It may also result from sudden and excessive increases in the glucose infusion rate (often markedly in excess of 6 mg/kg/min). After birth, the preterm baby has higher plasma glucose and insulin levels than the term baby, suggesting insulin resistance. Hepatic insulin resistance results in failure to suppress endogenous glucose production despite high glucose and insulin levels. There is also less peripheral glucose uptake by insulin-sensitive tissues, which are less abundant than in the term baby. Chorioamnionitis may have been the precipitant to the preterm delivery, and these babies have elevated levels of cytokines. After delivery, sepsis and necrotising enterocolits result in raised levels of proinflammatory cytokines including tumour necrosis factor and interleukins 1 and 6, which in turn lead to changes in insulin receptor signalling, inducing both hepatic insulin resistance and peripheral insulin resistance in muscle and adipose tissue.13 The interventions of intensive care, such as the use of inotropes and corticosteroids, may also increase insulin resistance and suppress insulin secretion.

Insulin levels in the fetus increase steadily towards term and the preterm baby has relatively immature β-cells resulting in relative insulin deficiency. Many preterm babies are also compromised in utero, and growth restriction is associated with a reduced β-cell mass and relative insulin deficiency. Although the pancreatic cells are sensitive to hyperglycaemia, they produce proinsulin rather than insulin, which is 10 times less active. Insulin assays do not always differentiate insulin and proinsulin, which may explain the reported variability in insulin levels. The normal postnatal metabolic adaptation induced by intermittent enteral feeds is often delayed by prolonged, continuous intravenous infusions of glucose. With intravenous feeding, the preterm baby does not have the normal postnatal association of incretin stimulation with enteral feeds. In addition, preterm babies who are given enteral feeds do not mount a comparable incretin response to term babies.14 This relative insulin deficiency in preterm babies leads to hyperglycaemia and probably contributes to their often prolonged and sometimes profound, catabolic state in the first few days and weeks after birth. Insulin deficiency may also impact on IGF-I generation that in turn has important growth and metabolic roles.

Neonatal diabetes

Although exceptionally rare (incidence of 1:400 000 live births), neonatal diabetes should be considered in any baby in whom the hyperglycaemia persists beyond the first 2–3 weeks of life. Neonatal diabetes is defined as persistent hyperglycaemia presenting in the first 6 months of life and is either transient or permanent.15 Babies often have significant intrauterine growth retardation and, in addition to hyperglycaemia, often have pronounced glycosuria resulting in severe dehydration and sometimes metabolic acidosis but have minimal or absent ketonaemia or ketonuria. A genetic cause can be identified in most of the cases: transient neonatal diabetes is commonly associated with abnormalities in the imprinted region of chromosome 6q24, or mutations in genes KCNJ11 (Kir6.2) or ABCC8 (SUR1), which code for the different subunits of the KATP channel; permanent neonatal diabetes is less common than the transient form, but a wider range of mutations have been identified, including mutations in genes encoding the subunits of the KATP channel involved in insulin secretion, β-cell transcription factors resulting in abnormal pancreatic development often with other significant developmental anomalies, or associated with defects in glucose sensing, or accelerated β-cell destruction.

Insulin treatment is important for the initial management of all infants with neonatal diabetes but can be clinically challenging as they require very small doses of insulin. The prevention of chronic hyperglycaemia has to be balanced against the risks of hypoglycaemia. Identifying infants with mutations in the KATP channel (SUR1 or Kir6.2 subunits) is clinically important, as these infants may respond well to treatment with sulphonylureas and no longer require long-term insulin treatment.

Does hyperglycaemia matter?

Although hyperglycaemia is the “normal” response to stress, it is associated with increased morbidity and mortality in critically ill adults,16 children17 and preterm babies6,,10 receiving intensive care. Although a cause-and-effect relationship is not clear, there are a number of retrospective studies reporting an association between neonatal hyperglycaemia and increased mortality and morbidity, including intraventricular haemorrhage,18 retinopathy of prematurity19 20 and sepsis. In the newborn the effect of hyperglycaemia on clinical outcomes may be the result of the combination of hyperglycaemia itself as well as the impact of relative insulin deficiency and its effects on early metabolism and growth.

In the newborn, hyperglycaemia may result in a significant osmotic diuresis, electrolyte imbalance and is associated with intraventricular haemorrhage.18 The blood glucose level above which complications arise has not been clearly defined, but it has been suggested that significant osmotic changes may not occur with glucose levels <20 mmol/l.21 Other pathophysiological effects of hyperglycaemia have mainly been studied in adult intensive care, with a large randomised control trial in adult surgical intensive care patients reporting a dramatic reduction in mortality and morbidity in those with tightly controlled glucose levels. Mortality was reduced by 32%, bacteraemia rates halved and the number of patient requiring intensive care reduced.22 It is not clear whether the prevention of hyperglycaemia or the actions of insulin were responsible for these findings. However, the same investigators found no improvement in mortality in their medical intensive care unit patients except those requiring critical care for ≥3 days.23 Two more recent multicentre studies questioned the Leuven findings, with both reporting high rates of hypoglycaemia.24 25 Two meta-analyses reached contrasting conclusions.26 27 The recently published the Normoglycemia in Intensive Care Evaluation-Survival Using Glucose Algorithm Regulation (NICE-SUGAR) multicentre study of >6000 patients showed that intensive glucose control (4.5–6.0 mmol/l) compared with maintaining glucose levels between 8 and 10 mmol/l using intravenous insulin increased mortality in adult intensive care unit patients at 90 days with no significant difference between surgical or medical patients. However, all these studies have highlighted the problems of reducing hyperglycaemia without significantly increasing the risk of hypoglycaemia,28 and the control arm of the the Normoglycemia in Intensive Care Evaluation-Survival Using Glucose Algorithm Regulation (NICE-SUGAR) study maintained glucose levels <10 mmol/l.

The presence of hyperglycaemia in the preterm infant has similarities to that in seen in patients in adult intensive care, but the underlying clinical pathologies are very different and the implications of relative insulin deficiency may be much more important. In the preterm infant, early growth failure is related to both short-term survival and long-term neurodevelopmental outcomes, and relative insulin deficiency may therefore impact on short-term anabolism and longer-term health outcomes. The relative insulin deficiency may also contribute to low IGF-I levels, and hyperglycaemia and low IGF-I levels have been linked with retinopathy of prematurity.19 20 Low IGF-I levels may impede pancreatic development and β-cell mass.29 Therefore, the baby may be at increased risk for the development of diabetes in later life.30 While glucose is proinflammatory, an inducer of oxidative stress and inflammation, insulin is anti-inflammatory, suppressing three major proinflammatory transcription factors nuclear factor B, activator protein-1 and early growth response-1, which may be important in infants with sepsis.

Neonatal interventional studies

A number of retrospective and two small prospective studies suggest that insulin therapy in already hyperglycaemic preterm infants could improve glucose levels and weight gain.31,,34 A pilot study infusing insulin at a constant rate of 0.025 U/kg/h within the first 24 h of life in VLBW babies demonstrated improved glucose control, increased IGF-I bioactivity and improved lower-limb growth over the first week.35 This study led to a larger multicentre randomised controlled trial (Neonatal Insulin Replacement Therapy in Europe (NIRTURE)) to investigate the potential impact of insulin replacement on clinical outcomes.36 VLBW babies were randomised to receive either early insulin therapy, 0.05 U/kg/min titrated with 20% glucose to maintain glucose levels between 4 and 8 mmol/l, or standard care when either glucose infusion rate was reduced or insulin was commenced, only if the blood glucose level was >10 mmol/l on two occasions. The aim of the study was to determine whether early introduction of fixed-dose insulin replacement, with variable dextrose support, to maintain euglycaemia (4–8 mmol/l), compared with standard reactive management of glucose levels, would improve glycaemic control and thus reduce morbidity and mortality at the expected date of delivery. All babies had glucose levels assessed using continuous glucose monitoring to provide detailed data on glucose control throughout the study period. Babies receiving early elective insulin therapy had lower mean glucose levels and the prevalence of hyperglycaemia was less in the first week of life. They also had significantly higher energy intakes with less weight loss in the first week. However, the difference in mean glucose levels between study arms was small compared with the pilot study,35 and hypoglycaemia (blood glucose <2.6 mmol/l for >1 h) was more common in those babies receiving early insulin therapy particularly in babies weighing >1 kg. The significance of the latter can only be determined with long-term follow-up. There was no difference in mortality at expected date of delivery or in the secondary morbidity outcomes. This study aimed to use insulin replacement therapy to promote anabolism, but the intervention was undertaken for a very short period and 36% of infants in the control arm also received insulin treatment. The levels of protein intake in both arms of the study, although typical of clinical practice, were low if one is hoping to promote anabolism.

How should hyperglycaemia be managed?

There therefore remains controversy regarding the optimal management of hyperglycaemia in the preterm infant. The first step in management of hyperglycaemia is accurate diagnosis, and because the signs of both hyperglycaemia and hypoglycaemia are typically absent or non-specific, infants at risk need to have blood glucose levels monitored. Once hyperglycaemia has been noted, it is important to consider, diagnose and treat any underlying illness such as sepsis or pain and then consider management of the hyperglycaemia itself. Currently, the management of hyperglycaemia is to (A) do nothing, (B) reduce the amount of glucose given, (C) treat with insulin, or a combination of (B) and (C). Although there is no consensus as to what level of glucose should be treated, most neonatologists would intervene when the glucose level rises to >10–12 mmol/l. This level of hyperglycaemia should prompt calculation of the glucose infusion rate, although hyperglycaemia typically occurs even at low glucose infusion rates in the preterm infant. The stable extremely low-birth-weight baby requires 6 mg/kg/min with an additional 2–3 mg/kg/min to support protein anabolism,37 and the demands of the sick baby may be considerably higher. Hence, glucose restriction may result in significant catabolism and severe growth restriction during a critical period of development. The maximal glucose oxidative capacity in the newborn is about 12 mg/kg/min; greater levels resulting in energy inefficient conversion to fat.37 It would therefore seem appropriate to maintain the glucose infusion rate <12 mg/kg/min but >6 mg/kg/min. However, if hyperglycaemia is present and excess dextrose is not being infused, insulin should be considered if hyperglycaemia persists (table 1).

Table 1

Suggested management of hyperglycaemia in preterm babies

We believe that despite the essentially negative results of the NIRTURE study, we should not ignore the potential benefits of insulin. It is clear that hyperglycaemia is associated with poor clinical outcomes, but optimal target glucose levels need to be determined for the preterm infant and the role of insulin requires further investigation.38 Clinical practice is currently hampered by lack of reliable constant real-time glucose measurements and poor delivery systems for administration of insulin. Although there is a good argument for treating blood glucose >8–10 mmol/l in preterm babies, particularly in the presence of an osmotic diuresis, care must be taken to avoid hypoglycaemia. Until there are more effective methods of both continuous real-time glucose measurements and insulin delivery systems, we should not be aiming for blood glucose levels at the lower end of the normal range (4–7 mmol/l), except in the setting of a clinical trial, because the risks of potentially serious hypoglycaemia may outweigh the potential benefits.

Future developments in the management of hyperglycaemia in at-risk infants include the refinement of continuous glucose monitoring. Interstitial glucose levels, calibrated intermittently with blood glucose, can give detailed information about glucose control (figure 1).39 However, until now, glucose levels have not been able to be viewed in real time and hence these devices have only been useful as a research tool, but new monitors are being developed that display real-time glucose levels. Before they can be recommended for standard clinical use, these monitors need validation in the intensive care setting at both high and low glucose levels. Truly non-invasive glucose monitoring techniques, such as utilising near-infrared, photoacoustic or Raman spectroscopy and polarimetry, are not yet available for use in the neonatal population.

Figure 1

Continuous glucose monitoring profile from a very-low-birth-weight baby during the first week of life.

Future interventional studies, aiming to reduce hyperglycaemia, might benefit from using real-time glucose monitors to improve glucose control while reducing the risk of hypoglycaemia. The refinement of computer algorithms and of intelligent closed loop systems might support attempts to achieve tighter glucose control. The controversies regarding the role of insulin require further investigation, including follow-up of IGF-I levels and long-term neurodevelopmental outcomes in infants recruited to the NIRTURE trial. In addition, if insulin replacement is to be effective in these infants, optimal protein/energy balance will be needed to achieve healthy growth. A combined approach is likely to be needed to optimise clinical outcomes. In the future, there may be other ways in which we might improve metabolic adaptation and glycaemic control including the use of IGF-I. Determining the optimal perinatal management of hyperglycaemia in preterm babies is important because it is likely to have significant impact on both short and long health outcomes.


Hyperglycaemia in the preterm baby is common and associated with an increase in mortality and morbidity. Although the use of insulin may improve energy intake and reduce weight loss, there is a risk of hypoglycaemia. The optimal management is yet to be defined, but we suggest maintaining glucose infusion rates between 6 and 12 mg/kg/min and glucose levels <10 mmol/l, if necessary, with the use of insulin. If insulin is used, one should aim for glucose levels at the upper end of the normal range to prevent hypoglycaemia until the availability of reliable, real-time constant glucose monitoring systems as well as refinement in insulin delivery systems.



  • Competing interests None.

  • Provenance and peer review Commissioned; externally peer reviewed.

  • Contributors ALOS and KB were involved in the referenced pilot study35 and the NIRTURE study.36