Monitoreo continuo de glucosa en tiempo real en cuidados intensivos neonatales.

Kathryn Beardsall⁎ Department of Paediatrics, University of Cambridge, Box 116 Cambridge Biomedical

Campus, Cambridge CB2 0QQ, UK The Neonatal Unit, Rosie Hospital, Cambridge University Hospitals NHS

Foundation Trust, Cambridge, UK. Early Human Development University 0378-3782/ © 2019 Published by

Elsevier B.V.

Resúmen:

La hipoglucemia y la hiperglucemia son comunes en los lactantes que requieren cuidados

intensivos y se asocian con peores resultados clínicos.

Sin embargo, los niveles de glucosa se toman con poca frecuencia, y sigue habiendo controversia

con respecto al manejo óptimo.

En adultos y niños, la monitorización continua de glucosa (MCG) se ha establecido como un

complemento importante para el cuidado de pacientes con riesgo de disglucemia.

Esta tecnología también proporciona cada vez más información sobre la regulación de la glucosa

en el recién nacido, lo que demuestra períodos significativos de hipoglucemia e hiperglucemia

clínicamente silenciosas.

Estos datos de referencia serán importantes para permitir evaluar la importancia de la

desregulación de la glucosa en los resultados a largo plazo.

Pequeños estudios también han demostrado la posibilidad de que la MCG apoye de forma segura

la focalización del control de la glucosa en los recién nacidos prematuros, y se está realizando un

gran ensayo multicéntrico.

La tecnología actual no está diseñada específicamente para su uso en la UCIN, pero con rápidos

desarrollos tecnológicos, CGM promete el cuidado futuro de los bebés en la UCIN.

Contents lists available at ScienceDirect

Early Human Development

journal homepage: www.elsevier.com/locate/earlhumdev

Real time continuous glucose monitoring in neonatal intensive care

Kathryn Beardsall⁎

University Department of Paediatrics, University of Cambridge, Box 116 Cambridge Biomedical Campus, Cambridge CB2 0QQ, UKThe Neonatal Unit, Rosie Hospital, Cambridge University Hospitals NHS Foundation Trust, Cambridge, UK

A R T I C L E I N F O

Keywords:GlucoseHyperglycaemiaHypoglycaemiaIntensive careNeonatalPretermContinuous monitoring

A B S T R A C T

Hypoglycaemia and hyperglycaemia are common in infants requiring intensive care and are associated withworse clinical outcomes. However, glucose levels are taken infrequently, and there remains controversy re-garding optimal management. In adults and children continuous glucose monitoring (CGM) is now established asan important adjunct to caring for patients at risk from dysglycaemia. This technology is also increasinglyproviding insights into glucose regulation in the newborn, demonstrating significant periods of clinically silenthypoglycaemia and hyperglycaemia. This baseline data will be important to allow the significance of glucosedysregulation on long-term outcomes to be assessed. Small studies have also shown the potential for CGM tosafely support targeting of glucose control in preterm infants, and a large multicentre trial is ongoing. Currenttechnology is not specifically designed for use in NICU, but with rapid technological developments, CGM holdspromise for the future care of babies in NICU.

1. Introduction

Hypoglycaemia and hyperglycaemia are common in babies re-quiring neonatal intensive care (NICU), but diagnosis and managementof both remain controversial [1–4]. Both extremes have been associatedwith poor clinical outcomes [5], but data is conflicting and clinicallysignificant thresholds of hypoglycaemia and hyperglycaemia remain tobe determined [3,6]. Current follow up studies are dependent on in-frequent glucose, taken for clinical reasons and there is inherent asso-ciated bias when assessing outcomes.

Standard intermittent blood glucose (BG) measurement results inlong periods when glucose levels are unknown. In contrast other phy-siological parameters such as oxygen saturation, blood pressure andheart rate are all monitored continuously to allow interventions toprevent wide fluctuations. It is increasingly thought that fluctuations inglucose levels may also have a significant impact on long-term out-comes [7].

The factors that determine the methods used for measuring glucosein the newborn are predominantly those of utility and accuracy.Methodology needs to be practical for use by those caring for babies,provide a result within a short time frame to allow for efficient changesin clinical management, and also provide accurate data across the rangeof clinically relevant glucose levels.

2. Continuous glucose monitoring

Continuous glucose monitoring (CGM) is widely used in patientswith type I diabetes. It has recently been agreed that it should be madeavailable to all women with type 1 diabetes during pregnancy, andstudies in adult intensive care have shown accuracy and safety with theability to reduce the risk of hypoglycaemia. CGM has the potential toimprove our understanding of the pathophysiology of glucose dysre-gulation in the newborn, to improve clinical management and to de-termine optimal targets of glucose control for different babies requiringintensive care. However, the use of CGM in neonatology has remainedlimited despite the challenges of glucose instability in the newborn.Despite great advances in the technology current devices are not de-signed for use in babies and there are technical challenges with usingdevices in the newborn which include: i) insertion methods, ii) accu-racy, iii) clinical interpretation [8].

3. The technology

Early CGM models used microdialysis with a hollow fibre that actsas a semipermeable membrane, but devices were large, and have lim-ited practical clinical use for neonates [9,10]. CGM technology hassubsequently been transformed since first receiving FDA approval, forthe management of type I diabetes, in 1999. Current technology inclinical practice, uses a fine disposable filament sensor (oxidase-based

https://doi.org/10.1016/j.earlhumdev.2019.104844

⁎University Department of Paediatrics, University of Cambridge, Cambridge, UK.E-mail addresses: kb274@medschl.cam.ac.uk, kb274@cam.ac.uk.

Early Human Development xxx (xxxx) xxxx

0378-3782/ © 2019 Published by Elsevier B.V.

Please cite this article as: Kathryn Beardsall, Early Human Development, https://doi.org/10.1016/j.earlhumdev.2019.104844

platinum electrode), which lies under the skin. This catalyses interstitialglucose generating an electrical signal to the transmitter that lies on theskin surface. The transmitter then either stores this data for laterdownload, or transmits it to a monitor or blue tooth device, for realtime read out. Currently there are two device companies whose tech-nology has been used in the newborn: Medtronic (Northridge, CA,United States) Fig. 1 and Dexcom (San Diego, CA, United States).However, none of the current devices are licensed for use in the new-born.

Early CGM models collected data in real time but the data couldonly be downloaded retrospectively in 5minute epochs. This is usefulfor research and for stable patients, but not well suited to acute man-agement in intensive care. However, the use of such devices has de-monstrated significant periods of clinically silent hyperglycaemia andhypoglycaemia in the newborn [11,13]. Newer models provide data inreal time, and key advances in technology including: extended life,smaller sensors and better accuracy at lower glucose thresholds haveled to their use in intervention trials [14].

4. The use of masked CGM and understanding physiology inneonates

There remains controversy regarding what is a ‘normal’ glucoseprofile during healthy transition and for babies considered at risk fromneonatal hypoglycaemia, as evidence is currently based on cross sec-tional data and intermittent sampling. The use of masked CGM providesthe opportunity to gain further insights into understanding the phy-siology and pathophysiology of neonatal transition.

4.1. Healthy newborn

It is traditionally taught that healthy babies have a physiologicalnadir in glucose levels after birth, however recent studies have failed todemonstrate this. The use of CGM to better understand neonatal tran-sition is challenging as sensors require a minimum of 2 h ‘warm up’ andtend to be the least accurate in the first 24 h, which is the critical timeof interest. The soon to report GLOW (Glucose in well babies study) hassucceeded in collecting CGM data in 67 well term babies alongsidemeasurement of alternative fuels, and this will provide much insightinto what we might consider normal neonatal glucose transition.

4.2. Newborn infants at risk of hypoglycaemia

The CHYLD study was a large prospective study of term and latepreterm infants at risk from neonatal hypoglycaemia, and in whommasked CGM was undertaken for the first 48 h of life [13]. This studyshowed that despite regular BG testing and a clinical aim to maintainBG >2.6mmol/l, many babies were exposed to prolonged periodswhen the sensor glucose (SG) was <2.6mmol/l. Furthermore of thebabies who did develop hypoglycaemia (BG <2.6mmol/l) 25% spentat least 5 h with SG <2.6mmol/l, and nearly a quarter of babies whowere considered clinically to have normal BG levels had episodes ofclinically silent hypoglycaemia detected using CGM [13]. At 2 yearfollow up of the CHYLD study 404 children underwent neurodevelop-mental assessment, but there was no association with hypoglycaemia.However, those with neurodevelopmental impairment had higher SGconcentrations during the first 48 h of life, and a steeper rise posttreatment for hypoglycaemia. These studies suggested that glucosevariability with hyperglycaemia post hypoglycaemia may be harmful tothe developing brain, and are in keeping with animal studies wheresuch neuronal injury is associated with the generation of reactiveoxygen species.

In contrast by 4.5 year follow up these episodes of clinically silenthypoglycaemia were associated with a four-fold increased risk of im-paired executive function. In comparison those with clinically detectedhypoglycaemia (i.e. with a BG <2.6mmol/l), who had therefore beentreated, the increased risk was only two fold [15]. This highlights theimportance of clinically silent hypoglycaemia as well as the importanceof developmental follow up beyond 4 years of age to assess executivefunction and the potential impact of glucose dysregulation.

Further studies have been undertaken using CGM to describe theglucose profiles of the offspring of mothers with type I diabetes, in theearly neonatal period, and its association with maternal intrapartumglucose control. Sixteen women had a CGM sensor inserted 2–3 daysbefore delivery, and infants had a masked CGM immediately after de-livery. Fifteen infants had at least one BG concentration <47mg/dl(2.6mmol/l), and in the first 24 h after delivery 4 infants spent >50%time with SG <47mg/dl (2.6mmol/l). However, the relationship be-tween maternal intrapartum and neonatal glucose control was poor[16].

CGM will be important in exploring the physiology of hypogly-caemia in these at risk babies, and the effect of different treatment orfeeding strategies on glycaemic stability. Moreover, it is only with ro-bust data on actual exposure that we can interpret later outcome stu-dies.

4.3. Prematurity

Preterm babies are at risk of wide fluctuations of glucose levels andmasked CGM has been used to characterize this and demonstrate thehigh prevalence of undetected hypoglycaemia and hyperglycaemia. Thefirst large trial to use CGM in these babies was the NIRTURE study,which used blinded CGM in 389 preterm babies [12,17]. CGM was welltolerated even in babies of 23 and 24weeks gestation and with birthweights of 500 g. There were no concerns about infection or skin in-tegrity and the CGM did not interfere with nursing care. More than halfthe babies had SG >10mmol/l at some time and this was found to beassociated with gestational-age, birth weight and the use do inotropes[17].

Studies with CGM have also highlighted babies to be at risk fromclinically silent hypoglycaemia as intravenous support is weaned andduring surgery Fig. 2 [18]. Preterm babies who are presumed to bestable tolerating intermittent feeds when monitored with CGM alsodemonstrate significant dysglycaemia [19,20]. Infants born at<1000 g, studied at corrected gestation age of 32 ± 2 and33 ± 2weeks when clinically stable showed clinically silent SG<2.5mmol/l (40%) and >8.3mmol/l (83%) with rapid fluctuations in

Fig. 1. Continuous glucose monitoring (CGM) device in situ in a preterm babyand above demonstrating size of sensor and bedside monitor.

K. Beardsall Early Human Development xxx (xxxx) xxxx

2

levels [21].

5. Clinical applications

The wealth of data available from CGM makes it attractive for use inNICU where many physiological variables are measured continuously.Adult intensive care studies have demonstrated that the time within aphysiological target range is linked to less organ injury and bettersurvival but this remains to be determined for neonates. The benefits inNICU may be greater as BG sampling is much less frequent, in keepingwith the aims of minimal handling and limiting blood loss as well as theimportance of the potential impact of reducing dysglycaemia on thevulnerable developing brain.

5.1. Preterm infant

Up to 50% of preterm babies are exposed to glucose levels>10mmol/l [17,22]. Both hyperglycaemia and hypoglycaemia havebeen associated with increased mortality and multiple morbidities inthese babies including retinopathy of prematurity, necrotizing en-terocolitis, sepsis and long-term neurodevelopmental impairment [23].Animal studies suggest this association to be causal [24,25]. Themanagement of babies with hyperglycaemia however remains chal-lenging; balancing the desire to optimize nutrition against the risks ofinsulin [12,26].

A trial in very low birth weight babies (VLBW n=43) aiming to useCGM to detect silent hypoglycaemia, demonstrated that the CGM de-tected three times the number of hypoglycaemic episodes (SG<2.8mmol/l) compared to intermittent BG measurement. CGM alsoreduced the median duration of these episodes from 95 to 44min (50%)[27]. However, those with CGM had more dextrose infused as bolusesand this may cause glucose instability which has been linked to in-creased morbidity [7,28].

Alternative strategies using CGM to support the use of insulin haveshown that CGM can increase the time in target without increasing therisk of hypoglycaemia [29]. These feasibility studies have led to aninternational randomized controlled trial of the use of CGM in preterminfants. This will randomize babies to standard care with masked CGM(for comparative data collection), or the use of CGM to support thetargeting of glucose control with the support of insulin.

The concept of the ‘artificial pancreas’ and closed loop technology isgaining momentum in adult medicine. Computerized control can re-duce the frequency and duration of hypoglycaemia in diabetic patients

at home and adults in ITU [30]. Pilot studies of computer algorithmguided glucose control in NICU have shown promise. A trial in VLBWnewborns (n=50) used a computer-guided program to determineglucose infusion rate (GIR) with or without CGM. Neonates using CGMhad a greater percentage of time spent in a euglycaemic range(72–144mg/dl, 4–8mmol/l) compared with the blinded CGM group(median, 84% vs 68%, P< .001). Use of CGM also decreased glycaemicvariability [31]. The alternate strategy of computer guided insulin de-livery has been explored in a cohort of extremely preterm infants(n=20) and also showed an increase in time in target (4–8mmol/l)from 26% in the control group to 97% in those with ‘closed loop’ CGM.(Fig. 3) Further studies are needed and many regulatory hurdles willneed to be crossed before this can be introduced as a fully automatedsystem in clinical care.

5.2. Term hypoxic ischaemic encephalopathy (HIE)

Hyperglycaemia and hypoglycaemia are common in babies fol-lowing HIE, and both have been associated with poorer neurologicaloutcomes, although findings are not consistent [5,32,33,34]. Thepragmatic clinical and physiological approach to care is to aim tomaintain euglycaemia. CGM could be helpful in providing trend in-formation to guide clinical management but the accuracy and func-tioning of sensors during cooling remains to be determined. Pilot stu-dies are ongoing to validate their use in these high risk babies.

5.3. Prolonged hypoglycaemia and hyperinsulinism

A small group of babies have prolonged hypoglycaemia that can lastseveral weeks. Glucose dysregulation can be associated with in-trauterine growth restriction and with congenital hyperinsulinism.These infants can be challenging to care for with the need for centralaccess to avoid the risk of profound hypoglycaemia and long-termneurodevelopmental consequences. These infants require frequent BGsampling but we have found CGM can be an adjunct to BG monitoringto support their management and progress them on to enteral feeds andtherefore home. CGM also has a role in neonatal diabetes where thecombination of CGM and insulin pump therapy can allow better controland reduced the risk of hypoglycaemia [35].

6. Limitations of use in the newborn

Although the CGM has great potential to be of clinical benefit, it is

Fig. 2. Comparison of intermittent blood glucose monitoring (IGM) and continuous glucose monitoring (CGM) in a surgical preterm infant.

K. Beardsall Early Human Development xxx (xxxx) xxxx

3

important to understand the current limitations.

6.1. Practicalities of insertion

Sensors have a length of around 8.5–11mm, are designed for in-sertion into the subcutaneous tissue, and have varying insertion tech-niques and angles of insertion. Although insertion kits are available,they are not always appropriate for the preterm infant with limitedsubcutaneous tissue and therefore hand insertion may be required. Thisneeds appropriate training as the devices are not as yet designed forease of use in these babies.

6.2. Accuracy

The CGM does not measure BG it measures interstitial glucose andthere is a physiological lag in the diffusion of glucose between the twocompartments. As BG levels fall the time delay increases, so there canbe a delay in the detection of hypoglycaemia relative to BG levels.Accuracy assessed using the MARD will be higher where the differencebetween interstitial and BG is larger. This value often varies with glu-cose levels, being higher at low levels but similar levels have beenshown in the preterm population as in adults. The lack of point accu-racy the CGM means it is not a tool for diagnostic thresholds. However,the wealth of data can warn of incipient falls or increases in BG and ifguidelines are in place and trends in CGM are used it is the falling SGthat can prompt BG to be checked and earlier intervention.

6.3. Calibration

If calibrations are undertaken at a time of glucose instability thenthe effect of the lag between blood and interstitial glucose levels willmake calibrations more inaccurate. If there is a wide difference betweenthe sensor and a BG this may result in a request for a sensor to bechanged or for it to stop working. The rapid glucose fluctuations in thenewborn after birth may increase the risk of calibration failure, andwaiting for blood glucose levels to be stable before calibration is im-portant. The first calibration normally occurs at about 2 h but in somebabies sensors will not calibrate unless allowed a longer ‘wetting time’.

The CGM accuracy can be subject to drift, i.e. a change in sensoraccuracy over time, due to changes in the probe surface, related tobiofilm build up with different currents are generated by the same BGconcentration. If multiple calibration points are used by algorithmswhen there is significant drift then calibrations can be affected by BGlevels taken up to 24 h earlier and may lead to more inaccuracy. InNICU, where a blood gas analyser or an accurate point of care device isused for calibration, a single point calibration may be better. Nowsystems are moving towards factory calibration of sensors to avoid theneed for BG calibration, but how robust this calibration will be for theNICU population remains to be determined.

6.4. Clinical interpretation

Although the CGM provides a wealth of data that helps to guidemanagement, the point accuracy of single readings is currently not asgood as BG sampling. CGM can be an adjunct to support care providingearly warning of fluctuations in glucose levels. However, the clinicalsignificance of the dysglycaemia revealed by this technology still needsto be better understood. Responding inappropriately to these findingsmay unintentionally cause more harm than good.

7. Future technology

There is currently a technological race to develop novel sensors fornon-invasive CGM for patients with diabetes mellitus. Development insensor technology includes a contact lens with embedded microchips,optical sensors using near infrared spectroscopy and microneedles thatuse reverse iontophoresis. These hold promise for NICU as further de-velopments are required before CGM will be able to demonstrate its fullpotential in the newborn.

8. Conclusion

CGM is well established for the management of adults and childrenwho are at risk from glucose dysregulation. CGM in neonatology isincreasing our understanding of the unique physiology at this time.There remain technical issues with the use of the current devices in the

Fig. 3. Comparison of glucose controlduring closed loop and standard care.Median (IQR) of sensor glucose andinsulin infused in babies randomized toclosed loop management or continuousglucose monitoring with paper algo-rithm (control). The closed loop inter-vention period is denoted by the ver-tical lines, and the target glucose range4.0–8.0 mmol/l is denoted by hor-izontal lines.

K. Beardsall Early Human Development xxx (xxxx) xxxx

4

newborn but also clinical challenges in interpretation of the data andimplications for clinical practice. However, the rapid technologicaldevelopments in sensor technology, and closed loop systems, hold thepotential for us to optimize the management of glucose levels in thisvulnerable population.

Declaration of competing interests

The author has received equipment support from Medtronicalongside funding from the Evelyn Trust and a National Institute forHealth Research: Efficacy and Mechanism Evaluation (NIHR EME)program grant to fund research into the role of continuous glucosemonitoring in neonatal intensive care.

Acknowledgements

The author wishes to acknowledge and thank all the families thathave taken part in our clinical studies and the clinical team on theNeonatal Unit for their support. We would also like to acknowledge thesupport of the Evelyn Trust, the National Institute of Health ResearchEME Program and the National Institute of Health Research CambridgeBiomedical Research Centre without whom the studies would not havebeen possible. We further thank Medtronic for providing the continuousglucose monitoring system and sensors for use.

References

[1] A.L. Ogilvy-Stuart, K. Beardsall, Management of hyperglycaemia in the preterminfant, Arch. Dis. Child. Fetal Neonatal Ed. 95 (2) (2010) F126–F131.

[2] C. Morgan, The potential risks and benefits of insulin treatment in hyperglycaemicpreterm neonates, Early Hum. Dev. 91 (11) (2015) 655–659.

[3] M. Bottino, R.M. Cowett, J.C. Sinclair, Interventions for treatment of neonatal hy-perglycemia in very low birth weight infants, Cochrane Database Syst. Rev. 1(2009) CD007453.

[4] W.W. Hay Jr., P.J. Rozance, Neonatal hyperglycemia-causes, treatments, and cau-tions, J. Pediatr. 200 (2018) 6–8.

[5] S.K. Basu, J.R. Kaiser, D. Guffey, C.G. Minard, R. Guillet, A.J. Gunn, et al.,Hypoglycaemia and hyperglycaemia are associated with unfavourable outcome ininfants with hypoxic ischaemic encephalopathy: a post hoc analysis of the CoolCapStudy, Arch. Dis. Child. Fetal Neonatal Ed. 101 (2) (2016) F149–F155.

[6] A.L. Ogilvy-Stuart, J.E. Harding, K. Beardsall, Thresholds for hypoglycaemicscreening-a cause for concern? Arch. Dis. Child Educ. Pract. Ed. 104 (1) (2019)33–34.

[7] C.J. McKinlay, J.M. Alsweiler, J.M. Ansell, N.S. Anstice, J.G. Chase, G.D. Gamble,et al., Neonatal glycemia and neurodevelopmental outcomes at 2 years, N. Engl. J.Med. 373 (16) (2015) 1507–1518.

[8] T.L. Hernandez, W.W. Hay Jr., P.J. Rozance, Continuous glucose monitoring in theneonatal intensive care unit: not quite ready for ‘plug and play’, Arch. Dis. Child.Fetal Neonatal Ed. 104 (4) (2019) F344–F345.

[9] U. Hildingsson, H. Sellden, U. Ungerstedt, C. Marcus, Microdialysis for metabolicmonitoring in neonates after surgery, Acta Paediatr. 85 (5) (1996) 589–594.

[10] M. Horal, U. Ungerstedt, B. Persson, M. Westgren, C. Marcus, Metabolic adaptationin IUGR neonates determined with microdialysis—a pilot study, Early Hum. Dev. 42(1) (1995) 1–14.

[11] K. Beardsall, A.L. Ogilvy-Stuart, J. Ahluwalia, M. Thompson, D.B. Dunger, Thecontinuous glucose monitoring sensor in neonatal intensive care, Arch. Dis. Child.Fetal Neonatal Ed. 90 (4) (2005) F307–F310.

[12] K. Beardsall, S. Vanhaesebrouck, A.L. Ogilvy-Stuart, C. Vanhole, C.R. Palmer,M. van Weissenbruch, et al., Early insulin therapy in very-low-birth-weight infants,N. Engl. J. Med. 359 (18) (2008) 1873–1884.

[13] D.L. Harris, M.R. Battin, P.J. Weston, J.E. Harding, Continuous glucose monitoringin newborn babies at risk of hypoglycemia, J. Pediatr. 157 (2) (2010) 198–202 e1.

[14] K. Beardsall, L. Thomson, C. Guy, M.M. van Weissenbruch, I. Iglesias,P. Muthukumar, et al., Protocol of a randomised controlled trial of real-time con-tinuous glucose monitoring in neonatal intensive care ‘REACT’, BMJ Open 8 (6)(2018) e020816.

[15] C.J.D. McKinlay, J.M. Alsweiler, N.S. Anstice, N. Burakevych, A. Chakraborty,J.G. Chase, et al., Association of neonatal glycemia with neurodevelopmental out-comes at 4.5 years, JAMA Pediatr. 171 (10) (2017) 972–983.

[16] Z.A. Stewart, L. Thomson, H.R. Murphy, K. Beardsall, A feasibility study of pairedcontinuous glucose monitoring intrapartum and in the newborn in pregnanciescomplicated by type 1 diabetes, Diabetes Technol. Ther. 21 (1) (2019) 20–27.

[17] K. Beardsall, S. Vanhaesebrouck, A.L. Ogilvy-Stuart, C. Vanhole, C.R. Palmer,K. Ong, et al., Prevalence and determinants of hyperglycemia in very low birthweight infants: cohort analyses of the NIRTURE study, J. Pediatr. 157 (5) (2010)715–9-3.

[18] P. Saha, K. Beardsall, Perioperative continuous glucose monitoring in a preterminfant, BMJ Case Rep. 2018 (2018).

[19] H. Mizumoto, M. Kawai, S. Yamashita, D. Hata, Intraday glucose fluctuation iscommon in preterm infants receiving intermittent tube feeding, Pediatr. Int. 58 (5)(2016) 359–362.

[20] A. Pertierra-Cortada, M. Ramon-Krauel, M. Iriondo-Sanz, I. Iglesias-Platas,Instability of glucose values in very preterm babies at term postmenstrual age, J.Pediatr. 165 (6) (2014) 1146–1153 (e2).

[21] E. Mola-Schenzle, A. Staffler, M. Klemme, F. Pellegrini, G. Molinaro, K.G. Parhofer,et al., Clinically stable very low birthweight infants are at risk for recurrent tissueglucose fluctuations even after fully established enteral nutrition, Arch. Dis. Child.Fetal Neonatal Ed. 100 (2) (2015) F126–F131.

[22] I. Szymonska, M. Jagla, K. Starzec, K. Hrnciar, P. Kwinta, The incidence of hy-perglycaemia in very low birth weight preterm newborns. Results of a continuousglucose monitoring study—preliminary report, Dev Period Med. 19 (3 Pt 1) (2015)305–312.

[23] N.M. van der Lugt, V.E. Smits-Wintjens, P.H. van Zwieten, F.J. Walther, Short andlong term outcome of neonatal hyperglycemia in very preterm infants: a retro-spective follow-up study, BMC Pediatr. 10 (2010) 52.

[24] J.M. Alsweiler, J.E. Harding, F.H. Bloomfield, Neonatal hyperglycaemia increasesmortality and morbidity in preterm lambs, Neonatology 103 (2) (2013) 83–90.

[25] L. Mohsen, M. Abou-Alam, M. El-Dib, M. Labib, M. Elsada, H. Aly, A prospectivestudy on hyperglycemia and retinopathy of prematurity, J. Perinatol. 34 (6) (2014)453–457.

[26] J.M. Alsweiler, J.E. Harding, F.H. Bloomfield, Tight glycemic control with insulin inhyperglycemic preterm babies: a randomized controlled trial, Pediatrics 129 (4)(2012) 639–647.

[27] F. Uettwiller, A. Chemin, E. Bonnemaison, G. Favrais, E. Saliba, F. Labarthe, Real-time continuous glucose monitoring reduces the duration of hypoglycemia epi-sodes: a randomized trial in very low birth weight neonates, PLoS One 10 (1) (2015)e0116255.

[28] W. Oh, B.B. Poindexter, R. Perritt, J.A. Lemons, C.R. Bauer, R.A. Ehrenkranz, et al.,Association between fluid intake and weight loss during the first ten days of life andrisk of bronchopulmonary dysplasia in extremely low birth weight infants, J.Pediatr. 147 (6) (2005) 786–790.

[29] L. Thomson, D. Elleri, S. Bond, J. Howlett, D.B. Dunger, K. Beardsall, Targetingglucose control in preterm infants: pilot studies of continuous glucose monitoring,Arch. Dis. Child. Fetal Neonatal Ed. 104 (4) (2019) F353–F359.

[30] L. Leelarathna, S.W. English, H. Thabit, K. Caldwell, J.M. Allen, K. Kumareswaran,et al., Feasibility of fully automated closed-loop glucose control using continuoussubcutaneous glucose measurements in critical illness: a randomized controlledtrial, Crit. Care 17 (4) (2013) R159.

[31] A. Galderisi, A. Facchinetti, G.M. Steil, P. Ortiz-Rubio, F. Cavallin,W.V. Tamborlane, et al., Continuous glucose monitoring in very preterm infants: arandomized controlled trial, Pediatrics. 140 (4) (2017).

[32] M. Nadeem, D.M. Murray, G.B. Boylan, E.M. Dempsey, C.A. Ryan, Early bloodglucose profile and neurodevelopmental outcome at two years in neonatal hypoxic-ischaemic encephalopathy, BMC Pediatr. 11 (2011) 10.

[33] S.K. Basu, J.L. Salemi, A.J. Gunn, J.R. Kaiser, CoolCap Study G, Hyperglycaemia ininfants with hypoxic-ischaemic encephalopathy is associated with improved out-comes after therapeutic hypothermia: a post hoc analysis of the CoolCap Study,Arch. Dis. Child. Fetal Neonatal Ed. 102 (4) (2017) F299–F306.

[34] J.P. Boardman, J.M. Hawdon, Hypoglycaemia and hypoxic-ischaemic encephalo-pathy, Dev. Med. Child Neurol. 57 (Suppl. 3) (2015) 29–33.

[35] K. Beardsall, C.L. Pesterfield, C.L. Acerini, Neonatal diabetes and insulin pumptherapy, Arch. Dis. Child. Fetal Neonatal Ed. 96 (3) (2011) F223–F224.

K. Beardsall Early Human Development xxx (xxxx) xxxx

5