Exploration of cord acid-base and glucose status of term and near-term infants, and oxygenation in preterm infants, with regard to mortality and clinically significant outcome

Abstract

This thesis comprises a series of studies initially focusing on the acid-base and glucose status of term and near-term infants at birth. It defines normal values and explores the influence on both short- and medium-term outcomes to elucidate prognosis and any interventions that might be warranted to improve this. The research then moves onto the oxygenation of preterm infants, firstly by examining the relationship between arterial partial pressure of oxygen (PaO2) and pulse oximeter oxygen saturations (SpO2), and then via case control studies examines the influence of oxygenation on important clinical outcomes (mortality and retinopathy of prematurity (ROP)).

Acid-base status of term and near-term infants and likelihood of death and cerebral palsy

The acid-base status of infants around birth can provide information about their past, current and future condition. Although umbilical cord blood pH less than 7.0 or base deficit equal to or greater than 12mmol/l is associated with increased risk of adverse outcome, there is uncertainty about the prognostic value of degree of acidosis as previous studies have used different variables, thresholds, outcomes, and populations.

We aimed to describe the relationship between pH and base deficit around the time of birth, the immediate clinical condition of the infants after birth, the risk of neonatal encephalopathy, and the risk of death or survival with permanent disability at the age of 2 years.

We conducted a retrospective review of routinely collected clinical data in all liveborn inborn infants of 35 weeks gestation or more delivered over an eight-year period at the Royal Infirmary of Edinburgh. Infants were included if their lowest recorded pH was less than 7 and/or highest base deficit equal to or greater than 12mmol/l on either umbilical cord blood and/or neonatal blood gas within one hour of birth. Two-year neurodevelopmental outcomes were collected from a targeted follow-up database.

56,574 infants were eligible. 507 infants (0.9%) met inclusion criteria. Poor condition at birth and all adverse outcomes increased with worsening acidosis. Combined outcome of death or cerebral palsy was 3%, 10% and 39% at lowest pH of 6.9-6.99, 6.8-6.89 and <6.8 respectively, and 8%, 14% and 59% at a base deficit of 12-15.9, 16-19.9, and 20mmol/l or more respectively. We found that there is a dose-dependent relationship between the degree of acidosis within an hour of delivery, and the likelihood of adverse outcome within 2 years in infants born at 35 weeks gestation or more. This information may help to facilitate the setting of entry criteria in clinical trials and better inform medicolegal opinion as to the likely outcome had delivery been expedited and the severity of acidosis been reduced.

Analysing the role of umbilical cord blood glucose in stratifying risk of impaired metabolic adaptation

The normal range of blood glucose concentration in umbilical cord blood in contemporary obstetric practice is uncertain. Defining normal and abnormal values is important because hypoglycaemia can require prompt recognition and treatment. It is therefore a priority to conduct research to fully elucidate threshold values for hypoglycaemia in healthy and at-risk infants in order that clinicians can implement practices that prevent harm that can result from unrecognised or untreated hypoglycaemia, whilst minimising unnecessary interventions and admissions to neonatal units.

We conducted a hypothesis generating exploratory study examining the frequency and associations of umbilical cord arterial glucose concentrations lower than the current operational thresholds defining neonatal hypoglycaemia.

The aims of this study were firstly to define normative values of cord arterial blood glucose with centile thresholds for infants greater than 35 weeks gestational age at birth without hypoxic ischaemic encephalopathy (HIE) grade II and III in a centre in which umbilical cord sampling was routine for all deliveries, and secondly to explore whether umbilical cord arterial glucose less than the 10th centile is associated with risk factors for impaired metabolic adaptation. We gathered data in the Royal Infirmary of Edinburgh over a fifteen-month period. All umbilical cord blood gas measurements performed in infants of 35 weeks gestation or more were included. Centiles were defined for arterial cord blood glucose concentration, and electronic patient records were interrogated for gestation, birth weight, mode of delivery, diagnosis of HIE grade II or III and hyperinsulinism, neonatal unit admission, repeat blood glucose measurement in the first four hours after birth, and the presence of risk factors for metabolic maladaptation (intrauterine growth restriction <2nd centile, maternal labetolol and maternal diabetes).

9793 umbilical cord blood samples were downloaded in infants greater than 35 weeks gestation. This represented 5612 infants, with 4181 infants having paired arterial and venous cord glucose values. The mean (SD) arterial and venous blood glucose values were 3.84 (1.22) and 4.53 (1.15) mmol/l respectively. In 4181 infants with both arterial and venous samples, the venous glucose level was a mean 0.75mmol/l higher than the arterial value (95% CI of the mean difference 0.73-0.77). 456 infants had arterial cord glucose below the 10th centile (2.5mmol), and 168 infants below the 5th centile (2.2mmol/l). 83 infants with arterial cord glucose below the 10th centile were admitted to the neonatal unit (18%). In infants with arterial cord glucose below the 10th centile, there were 130 repeat glucose samples within four hours (mean 2.586 mmol/l, SD 1.371, range 0-10.3). 68 infants (15%) with arterial cord glucose below the 10th centile had one or more risk factors for impaired metabolic adaptation. In infants diagnosed with HIE, there was a statistically significant difference in cord arterial glucose between those that died (mean 2.8mmol/l, SD 1.69) and those that survived to discharge (mean 5.1 mmol/l, SD 1.8) and between those diagnosed with grade II encephalopathy (mean 5.2 mmol/l, SD 1.90) compared to grade III (mean 3.3 mmol/l, SD 1.69). Infants delivered by Elective Caesarean section made up 52% of infants below the 10th centile for cord arterial glucose.

We concluded that in this large dataset drawn from an unselected population of term and near-term infants without a diagnosis of HIE, the 10th centile for arterial cord glucose is 2.5 mmol/l and the 5th centile is 2.2 mmol/l. Although cord arterial glucose poorly stratifies those at risk of impaired metabolic adaptation, we recommend that initial cord arterial glucose is treated as a valid sample measured on a reliable device and therefore is included in the BAPM Framework for Practice operation threshold approach.

Exploring the relationship between PaO2 and SpO2 in very low birth weight infants in the first week of life

Titrating supplemental oxygen to SpO2 in neonates evolved with little supporting evidence. Prior to the widespread introduction of SpO2 targeting, recommended practice was to target transcutaneous or arterial partial pressure of oxygen to 6.7-10.7kPa (50-80mmHg). It is not known which approach is safer. Recent evidence shows that SpO2 targets below 90% increase mortality and NEC and higher targets increase ROP treatment but not blindness or disability.

We aimed to explore the relationship between PaO2 and SpO2 in oxygen dependent preterm infants targeted to an SpO2 range 90-95% and to estimate the probability that individual SpO2 readings would allow a PaO2 outside the range 6.7-10.7kPa (50-80mmHg) previously recommended.

Paired SpO2 and PaO2 data were collected from oxygen dependent babies with birth weight less than 1500g during the first seven days of life over a four-year period. 3748 data pairs were obtained from 140 very low birth weight infants.

We found that at any SpO2 below 90% more than half of PO2 readings were below 6.7kPa (50mmHg). Low PaO2 (<6.7kPa, 50mmHg) was more likely than high PaO2 (>10.7kPa, 80mmHg) at all SpO2 values below 98%. At SpO2 92-97% around 70% of specimens had PaO2 within between 6.7-10.7kPa (50-80mmHg) and high PO2 was less likely than low PO2.

We concluded that an SpO2 target range of 90-95% permits many PaO2 values below 6.7kPa (50mmHg). High PaO2 readings are uncommon at SpO2 below 98%. We concluded that further studies examining an SpO2 target range of 92-97% may be beneficial as a higher range may confer additional survival advantage without significant hyperoxia, and this was incorporated into the recommendations for research as part of the NICE guideline NG124 Specialist neonatal respiratory care for babies born preterm.

Achieved oxygen saturation profiles associated with risk of mortality in extremely preterm infants in the BOOST II UK trial

In the Neonatal Oxygen Prospective Meta-analysis (NeOProM), mortality was higher in preterm infants randomised to target SpO2 85-89% than infants targeted to 91-95%.

We aimed to describe the achieved SpO2 distribution of low SpO2 group infants in the BOOST II UK trial who died and compare it with that of surviving infants in both randomisation groups.

A case control study using recorded SpO2 data was conducted. Infants randomised to the low SpO2 target who died were matched to a control survivor from each randomisation group. SpO2 values recorded from the day of birth until the time of death in the case infant were analysed. Data for control infants were analysed for the same duration of monitoring.

The data demonstrated the achieved oxygen profiles of the groups differed significantly from the targeted range. The low SpO2 target group infants had SpO2 values above their intended range whether they lived or died (median SpO2 90% in low target group infants who died, 91% in low target group survivors and 93% in high target group survivors). Low target group infants who died spent less time with SpO2 > 92% than survivors in either control group (33% versus 40% versus 57% respectively) but did not spend more time with very low SpO2 (<80%) than low target group survivors (7% versus 7%). Observations were similar whether infants died in the first week or later. The data demonstrated that increased mortality in low target group infants in BOOST II UK was associated with median SpO2 90% and not with increased deep hypoxia. Survival at all time points was associated with more time at SpO2 >92%.

We concluded that the oxygen profiles of surviving babies in the BOOST II UK trial were significantly higher than their intended targeted ranges and that clinicians should be cognisant of this before applying the recommended ranges using servo-control of inspired oxygen (FiO2). Future trials should investigate the risks and benefits of higher SpO2 targets than 90-95%.

Achieved oxygen saturation profiles associated with risk of retinopathy of prematurity in extremely preterm infants in the BOOST II UK trial

The pathophysiology of ROP is thought to involve initial hyperoxic arrest of vascular growth followed by hypoxia-induced hyperproliferation of the retinal vasculature. ROP risk increases with decreasing gestational age (GA), and an association between ROP and severity, time and variability of oxygen exposure is well established. A secondary analysis of the SUPPORT trial data demonstrated that infants with severe ROP spent more time at saturations of between 97–100% SpO2 compared to those without severe ROP. ROP proceeds in two phases (I: vasoconstrictive; II: vasoproliferative), during which similar oxygen saturations may have different effects.

We aimed to describe the achieved SpO2 distribution of high SpO2 target group infants in the BOOST II UK trial who did not have ROP requiring laser treatment and compare it with that of high SpO2 target group infants who did require laser therapy for ROP.

An exploratory cohort study analysis of BOOST II UK data was conducted, with index cases being surviving infants randomised to the higher SpO2 target range without ROP or with ROP that did not require laser treatment, matched by sex, gestation and birth weight to surviving infants in the higher SpO2 target range with ROP that required laser treatment. We demonstrated that infants targeted to an SpO2 range of 91-95% who developed ROP requiring laser therapy had a statistically significant increase in time at saturations greater than 98% than controls targeted to the same SpO2 target range.

We concluded that a further study using a different dataset may elucidate whether an oxygen targeting strategy of between 92-97% could optimise survival and minimise ROP.