We have expressed qualitative data as percentage distributions and quantitative data using measures of central tendency and dispersion. We compared groups with the χ2 and Student t test or the applicable nonparametric test if the data did not fit a normal distribution (Wilcoxon or Mann-Whitney U test). Statistical significance was defined as a p-value of less than 0.05. The statistical analysis was performed with the software SPSS® Statistics version 20.

The study was approved by the research ethics committee of the coordinating hospital (CEIC 2298).

Of the 133 newborns in the final sample, 96 (72%) received a diagnosis of moderate HIE and 37 (28%) of severe HIE. Those with severe HIE had less favourable perinatal characteristics and a higher frequency of seizures, aEEG abnormalities and multiple organ involvement (P < .001). Of the 28 newborns that died, 22 (79%) had severe HIE, and of these 22, 86% died following withdrawal or withholding of life-sustaining treatment (Tables 1 and 2).

Fig. 3 summarizes the adherence to the main aspects of management documented in our study based on the recommendations of the SENeo guideline. We also took into account 2 aspects that were not considered in the guideline: monitoring of regional cerebral oxygen saturation (rScO2) and measurement of neuron-specific enolase (NSE) levels in cerebrospinal fluid (CSF).

In 84% of the newborns, the radiant warmer was turned off in the delivery room. Servo-controlled whole-body hypothermia (active cooling) was initiated a median of 5 h post birth (interquartile range [IQR], 3.3−6.3), with a median of 4 h (IQR, 1−6) in those with severe HIE vs 5 h (IQR, 4−7) in newborns with moderate HIE (P = .018). However, the target central temperature (33−34 °C) was achieved at age 3.5 h (IQR, 1−6), as 72% of newborns with severe HIE and 56% of newborns with moderate HIE achieved the target temperature with the initial passive cooling measures before initiation of active cooling (Table 3).

Fifty-seven percent of the patients were delivered in one of the hospitals that provided TH (inborn), and active cooling in these patients was initiated earlier compared to outborn patients at a median of 4 h post birth (IQR, 2–5) versus 6 h (IQR, 5−8) (P < .001). Adjusting for the severity of HIE, active cooling was initiated a mean of 3.3 h later in outborn infants compared to inborn infants (95% confidence interval [CI], 2.5–4.1). However, there were no differences between outborn and inborn infants in the median age at which the target central temperature was achieved: 3 h (IQR, 1–5) vs 4 h (IQR, 1−6) (P = .21).

Thirty-three patients were born off site and required transport. The median central temperature on arrival to the receiving hospital in those with severe HIE was 33 °C (IQR, 32.8−34) vs 34 °C (IQR, 33−34.7) in those with moderate HIE (P = .04). Overcooling (temperature < 33 °C) was more frequent in newborns with severe HIE compared to those with moderate HIE: 6/19 (32%) vs 7/37 (19%) (P = .33).

The TH protocol was completed in 125 (94%) patients, and rewarming lasted a median of 10 h (IQR, 8−12), with no differences based on the severity of HIE (P = .57). In 27 (20%) of the 133 patients, the target temperature had to be raised due to pulmonary hypertension, coagulopathy and/or severe haemodynamic shock, more frequently in patients with severe HIE (P = .008). In this subset, TH could be withdrawn completely in 8 of the 27 newborns before 72 h post birth (Table 3).

Two infants had subcutaneous fat necrosis. Antibiotherapy was used in all but one of the infants, and an aminoglycoside was included in the management of 114 patients (86%). Six patients (5%) had early-onset sepsis with a positive blood culture. During TH, 61% of the newborns received parenteral nutrition, and 10% enteral nutrition (enteral trophic feeding in 11/13).

Ninety-six percent of the patients received sedation/analgesia, 47% via infusion. The most frequent sedation regimen consisted of the combination of an opiate with midazolam (53%), followed by fentanyl as monotherapy (41%). Of the 5 infants that did not receive sedation/analgesia, 4 had severe HIE (coma). All patients but one (127/128) required respiratory support, and 121of them (94%) required mechanical ventilation (Table 3).

All patients were monitored with aEEG and 72% with monitoring of rScO2. Magnetic resonance imaging (MRI) was used in 80% of the patients, 94% of those with moderate HIE and 46% of those with severe HIE (P < .001). The median age at which the MRI scan was performed was 11 days (IQR, 8–15). Of the 22 infants with severe HIE who died, only 3 had been evaluated with MRI.

The levels of ENE in CSF were measured in 42% of the newborns during TH, and the results were available within 24 h in only 7 of these 56 patients.

The number of newborns managed with TH varied between hospitals (Appendix B, supplementary tables 1 and 2). While we did not find differences in the age at initiation of active cooling, the median age at which the target central temperature was achieved ranged from 1 h post birth (IQR, 1–4) in the HUBU and 6.5 h post birth (IQR, 1.3–7) in the CAULE (P = .001). In the latter hospital, most patients achieved the target temperature after initiation of active cooling (69%). In contrast, 77% of newborns in the HUBU achieved the target temperature with passive measures before initiation of active cooling (Fig. 2; Appendix B supplementary table 2). When we analysed the diagnostic measures, we found variability between centres in the use of rScO2 monitoring and the measurement of NSE levels (Fig. 3; Appendix B, supplementary table 2).

Figure 2.

Characteristics of initiation of hypothermia in inborn patients stratified by hospital.

The figure only shows inborn patients to avoid the bias resulting from the asymmetrical distribution of inborn and outborn patients between hospitals. We present the temperature at admission (median) to the neonatal unit once the newborn was transferred from the delivery room (thermometer symbol), the age at which the target central temperature of 33−34 °C was reached (median, interquartile range) (blue square) and the age at initiation of active cooling (purple circle). The bars represent the distribution of newborns by severity of HIE: moderate (solid grey) and severe (shaded grey). Hospitals: 1 (HUBU), 2 (HURH), 3 (HCUV), 4 (HCUS), 5 (CAULE).

(0.26MB).

Figure 3.

Characteristics of the management of patients treated with hypothermia included in the study based on the recommendations of the SENeo guideline, overall (A) and stratified by hospital (B).

aEEG, amplitude-integrated electroencephalogram; CSF, cerebrospinal fluid; Doppler US, cranial Doppler ultrasound; NSE, neuron-specific enolase; rScO2, regional cerebral oxygen saturation; temp, temperature; TH, therapeutic hypothermia.

The categories “duration of at least 72 h” and “rewarming over a minimum of 8 h” did not include patients who died before 72 h post birth or patients who died during “outpatient follow-up”.

(0.8MB).

A relevant finding of the study was the early time at which the target temperature of 33−34 °C was achieved. The median of 3.5 h post birth is lower compared to the time reported in early clinical trials and other case series in the literature, in which the target temperature was mostly achieved after 4 h post birth.5,7,21–23 Early achievement of the target, repeatedly recommended as a possible strategy to improve the outcomes of TH, is also facilitated by the early HIE detection programme in our region.14,22,24

The initiation of cooling measures in the delivery room was universal and in adherence with the recommendations of the International Liaison Committee on Resuscitation (ILCOR).25 This, combined with the initiation of passive cooling measures in the sending hospital and during transport, explains how outborn patients achieved the target temperature at a similar age as inborn patients, despite active cooling being initiated later at the receiving hospital.23 However, stricter temperature monitoring is needed, as evinced by the fact that 23% of transported newborns were overcooled at arrival to the receiving hospital, despite the existence of a protocol with recommendations for cooling during transport.15 In most regions in Spain, there are deficiencies in the availability of both specialised neonatal transport teams and servo-controlled cooling systems for transport.26

International recommendations, including the guidelines of the SENeo, emphasise the importance of protracted rewarming when a return to normal body temperature is required or at the end of the recommended 72 h of TH.12 Our study showed that rewarming is performed slowly, amply complying with the recommendation of rewarming at a rate of 0.5 °C or less per hour.12,25

Adherence to established care standards was very satisfactory when it came to the use of sedation and neuromonitoring.

Sedation/analgesia was used in practically every patient, and most required mechanical ventilation. Although in a high percentage of patients the need of respiratory support was associated with the underlying disease, and the need of intubation was singled out in a nationwide survey as the drawback of providing sedation to these patients, so that only 70% reported using it routinely.13 One aspect that needs resolving, therefore, is which drug or combination of drugs would be most appropriate based on its pharmacokinetics in the context of TH, with the aim of minimising discomfort while avoiding respiratory depression.27

On the other hand, neuromonitoring was widely used, mainly with aEEG and, to a lesser extent, rScO2 monitoring. Although there are no firm recommendations on the type of monitoring required for adequate management of these patients, in the case of aEEG there is ample evidence of its usefulness in guiding the identification of patients with HIE, detecting epileptic seizures and predicting outcomes in the first 3 days post birth.28,29 The usefulness of rScO2 monitoring in HIE remains to be established, although there is growing evidence that it contributes valuable information on the balance of regional cerebral oxygen availability and utilization and the haemodynamic status and outcomes.30,31

Magnetic resonance imaging contributes objective information on the severity of brain damage and has a high predictive value.32 Having this information is particularly important in the first days of life, when decisions regarding redirection of curative care to palliative care may be made in the most severely ill patients.33 Although performance of MRI was universal in patients who survived, an MRI scan was only performed in 3 of the 22 patients with severe HIE who died. Since in most of these patients (86%) there was redirection of care, it is clear that the MRI was not routinely included in the decision-making process. This is an opportunity for improvement, and in this context, measurement of biochemical markers of brain damage, such as the levels of NSE in CSF should also be contemplated.34 Both tools are particularly useful in the case of prognostic uncertainty, as they are not affected by sedative drugs, unlike clinical manifestations, the aEEG or the rScO2. Their drawbacks include the need of transport to the radiology room or performance of lumbar puncture in severely ill and unstable patients, in addition to the limitations in many hospitals when it comes to results becoming available within the necessary time window.13

In our sample, enteral feeding was only used in 10% of the patients, which stood in contrast to other studies.35 However, the current evidence on how to administer it and in which cases is scarce, although the practice appears safe and could help reduce the time elapsed to full oral feeding as well as the length of stay.35

Empirical antibiotherapy was provided routinely to these patients due to their increased risk of early-onset sepsis.36 This approach is controversial compared to the option of monitoring biochemical markers and initiating antibiotherapy once the findings suggest the development of sepsis, but it is important to take into account that biochemical marker levels may change as a result of asphyxia itself.37 On the other hand, the substantial use of aminoglycosides is alarming, as these agents should be used with caution due to the increased risk of toxicity in newborns undergoing TH.38

Only 6% of parents held their babies during treatment. Stress and suffering have been described in the families of newborns managed with TH, as well as anxiety at the thought of approaching the infant and participating in providing care.39 However, when parents provided routine care and comfort to newborns with TH they had positive feelings, including a connection to the child, reinforcement of their parental role and that these practices could be safe.39,40

There are limitations to our study. One of them is its ambidirectional design, which entails uncertainty as regards the precision of the data obtained from health records. In addition, the assessment of the severity of HIE may not have been homogeneous across facilities,16 although we believe that the existence of a programme for the early identification of HIE minimises this potential bias and is one of the strengths of the study.