Conceived and designed the experiments: DA BS AH. Performed the experiments: DA BS AH LL EJ. Analyzed the data: DA BS LL. Wrote the paper: DA BS LL EJ JK PB DE.
The authors have declared that no competing interests exist.
Delay in implementing new treatments into clinical practice results in considerable health and economic opportunity costs. Data from the UK TOBY Cooling Register provides the opportunity to examine how one new effective therapy for newborn infants suspected of suffering asphyxial encephalopathy – therapeutic hypothermia- was implemented in the UK.
We analysed returned data forms from inception of the Register in December 2006 to the end of July 2011. Data forms were received for 1384 (67%) of the 2069 infants registered. The monthly rate of notifications increased from median {IQR} 18 {15–31} to 33 {30–39} after the announcement of the results of the recent TOBY trial, and to 50 {36–55} after their publication. This rate further increased to 70 {64–83} following official endorsement of the therapy, and is now close to the expected numbers of eligible infants. Cooling was started at 3.3 {1.5–5.5} hours after birth and the time taken to achieve the target 33–34°C rectal temperature was 1 {0–3} hours. The rectal temperature was in the target range in 83% of measurements. From 2006 to 2011 there was evidence of extension of treatment to slightly less severely affected infants. 278 of 1362 (20%) infants died at 2.9 {1.4–4.1} days of age. The rates of death fell slightly over the period of the Register and, at two years of age cerebral palsy was diagnosed in 22% of infants; half of these were spastic bilateral. Factors independently associated with adverse outcome were clinical seizures prior to cooling (p<0.001) and severely abnormal amplitude integrated EEG (p<0.001).
Therapeutic hypothermia was implemented appropriately within the UK, with significant benefit to patients and the health economy. This may be due in part to participation by neonatal units in clinical trials, the establishment of the national Register, and its endorsement by advisory bodies.
Delay in implementing new treatments into clinical practice once the evidence base has been established was identified as a ‘gap in translation’ by Sir David Cooksey in his review of UK Health Research, and results in considerable health and economic opportunity costs.
Extensive experimental and clinical research has been carried out into prolonged moderate hypothermia for perinatal asphyxial encephalopathy.
The UK TOBY Cooling Register was set up immediately following the conclusion of enrolment to the TOBY trial (ISRCTN8954757) a multicentre randomised controlled trial of whole body hypothermia for the treatment of perinatal asphyxial encephalopathy, supported by the Medical Research Council, UK, and predominantly carried out in the UK.
The horizontal lines indicate the range of expected eligible cases in the UK.
When the results of the TOBY trial together with the other trials confirmed the therapeutic efficacy of prolonged moderate cooling, the TOBY study group planned the systematic implementation of cooling therapy throughout the UK. Register personnel produced a clinical protocol and handbook, and delivered a programme of presentations about therapeutic hypothermia to UK Neonatal Networks and at study days throughout the country. These supported the organisation of local services for the provision of hypothermia and the development of local clinical protocols. Subsequently, the introduction of cooling therapy into clinical practice and the reporting of cases to the Register were endorsed by the UK National Institute for Heath and Clinical Excellence (NICE) and the British Association of Perinatal Medicine.
Shaded bars indicate hospitals that participated in the TOBY trial and clear bars hospitals that did not participate in the TOBY trial.
Three centres from Eire contributed data from 35 cases and these are included in this report.
Based on previous UK population studies
The UK TOBY Cooling Register has been briefly described before
The Register handbook gives guidance on which infants should be considered for treatment with cooling, with the aim to start cooling within six hours of birth. The indications are based on the criteria used in the TOBY trial, namely the presence in infants of at least 36 weeks gestation, of evidence of perinatal asphyxia together with signs of moderate or severe encephalopathy. Unlike in the TOBY trial, an abnormal amplitude integrated EEG (aEEG) recording is not a mandatory requirement, although clinicians are advised to perform aEEG recording whenever possible. This change from the TOBY trial was made primarily to avoid delay in initiating treatment with cooling when the aEEG is not available at the place of birth. The observation that patient characteristics and outcomes were similar amongst the randomised trials, yet only some included an abnormal aEEG as a trial entry criterion, suggests that it need not be a mandatory requirement.
Characteristic | n = 1384 |
Gestational age at birth, weeks (n = 1350) | |
Median {IQR} | 40 {38.4 to 41.1} |
[Range] | [34.0 to 44.4] |
Birth weight, grams (n = 1379) | |
Median {IQR} | 3340 {2900 to 3800} |
[Range] | [1530 to 5830] |
Age when cooling commenced, hours (n = 1331) | |
Median {IQR} | 3.3 {1.5 to 5.5} |
[Range] | [0 to 35.3] |
Cooling commenced after 12 hours of age, n (%) (n = 1331) | 29 (2.2) |
Cooled before gestational age of 36 weeks, n (%) (n = 1348) | 38 (2.8) |
Outcome at discharge, n (%) (n = 1362) | |
Discharged home | 912 (67) |
Transferred to a different hospital | 172 (13) |
Died before discharge | 278 (20) |
Centres that introduce treatment with cooling are asked to nominate a lead contact person. An information pack about the Register and training material are provided and the Register co-ordinator provides specific guidance as required. Details of the neonatal centres notifying cases to the Register are available on the Register website. A monthly reminder note asking for a list of cases cooled in that month is sent to each registered cooling centre. Cases can be notified to the Register by telephone or email and this can be done retrospectively.
Event | Frequency (%) |
Sepsis | 232 (17) |
Hypoglycaemia | 344 (25) |
Hypotension | 557 (40) |
Coagulopathy | 435 (31) |
Arrhythmia | 118 (9) |
Definitions:
Sepsis: Any evidence of infection requiring antibiotic therapy which is confirmed on culture.
Hypoglycaemia: Blood glucose below 2.6 mmol/litre.
Hypotension: Persistent mean blood pressure of <40 mmHg.
Coagulopathy: Any disorder requiring treatment in order to maintain or recover normal haemostasis.
Arrhythmia: Sinus bradycardia below 80 beats per minute and other arrhythmias identified on the electrocardiogram.
Clinical and treatment details are collected on a specific data record form that can be downloaded from the Register website. When the infant is approaching 2 years of age a request is sent to the notifying clinician asking for details of clinical outcomes at 2 years using a further downloadable record form; currently in the UK there is no statutory requirement for follow-up of infants who suffered neonatal encephalopathy or who received therapeutic hypothermia. Details sought are the findings on neurological examination, an assessment of neuromotor function using the Gross Motor Function Classification System (GMFCS)
Event | Frequency (%) |
Meconium aspiration | 144 (10) |
Pulmonary haemorrhage | 41 (3) |
Pulmonary hypertension | 99 (7) |
Pulmonary airleak | 63 (5) |
Pneumonia | 20 (1) |
Late onset sepsis | 30 (2) |
Renal failure treated with dialysis | 8 (0.6) |
Necrotising enterocolitis | 9 (0.7) |
Major cerebral anomaly on cranial ultrasound | 38 (3) |
Definitions:
Meconium aspiration: The presence of meconium stained liquor at birth and severe respiratory distress within 1 hour of birth and compatible X-ray changes.
Pulmonary haemorrhage: Copious bloody secretions with clinical deterioration requiring change(s) in ventilatory management.
Pulmonary hypertension: Severe hypoxaemia disproportionate to the severity of lung disease and evidence of a right to left shunt.
Pulmonary airleak: Any radiologically confirmed airleak serious enough to affect management (including pneumothorax, pulmonary interstitial emphysema, pneumopericardium, pneumoperitoneum and pneumomediastinum).
Late onset sepsis: Any evidence of infection after 72 hours from birth requiring antibiotic therapy which is confirmed on culture.
Necrotising enterocolitis: Infants with abdominal distension, gastric aspirate and/or blood in stools together with abdominal X-ray showing bowel oedema, pneumatosis or pneumoperitoneum, i.e. Bell’s staging 2 or 3.
Major cerebral anomaly on cranial ultrasound: Including evidence of parenchymal haemorrhage as determined by ultrasound, ventricular dilatation (defined as >97th centile for gestational age) or the presence of porencephalic cysts or cystic leukomalacia.
A chart of each infant’s temperature record is sent back to notifying clinicians if requested, following receipt of the data record forms and summary information about their cases is available to all registered cooling centres. Notification, use of Register material, feedback service and online and telephone support are provided freely to all clinicians and the entire Register website is publicly accessible.
Outcome | n | (%) |
Gross Motor Function Classification System (GMFCS), (n = 65) |
||
Level 1–2 | 38 | (58) |
Level 3–5 | 27 | (42) |
Manual Ability Classification System (MACS), (n = 65) |
||
Level 1–2 | 36 | (55) |
Level 3–5 | 29 | (45) |
Cerebral palsy (n = 251) | 56 | (22) |
Bilateral 2 limb | 4 | (2) |
Bilateral 4 limb | 29 | (12) |
Hemiplegic left | 6 | (2) |
Hemiplegic right | 3 | (1) |
Dystonic | 5 | (2) |
Choreoathetoid | 1 | (–) |
Ataxic | 3 | (1) |
Non-classifiable | 5 | (2) |
Occipital head circumference (cm), (n = 196) | ||
Mean (SD) | 48 | (2.5 ) |
15 children with GMFCS/MACS levels 1–2 were recorded as not having cerebral palsy and 5 children recorded as having cerebral palsy were not classified using these scales.
The numbers of hospitals registering their first case is shown in
We analysed returned data forms from inception of the Register in December 2006 to the end of July 2011. We report the individual and cumulative number of notifications and treatment centres contributing to the Register. Numbers (with percentages) are presented for binary and categorical variables, and means (with standard deviations) or medians (with lower and upper quartiles ({IQR}) for continuous variables as appropriate.
We used poisson regression to examine the difference in the rate of monthly registrations 6 months before and the 6 months after the results of the TOBY trial became available, their publication and the publication of the NICE guidelines. We analysed the temperature profiles and clinical details of notified cases, including severity of encephalopathy assessed by a modification of the Sarnat score
We investigated risk factors for adverse outcome at 2 years for those with available outcome data. Adverse outcome at 2 years was defined as death, cerebral palsy and/or a GMFCS or MACS score >2. First we examined the association of the following risk factors with adverse outcome in a bivariate analysis using Fisher’s exact test: clinical seizures prior to cooling, the aEEG, Apgar score at 10 minutes, cord gas pH, base excess and Thompson encephalopathy score prior to cooling. Significant factors in the bivariate analysis (p value<0.05) were included in a multivariate logistic regression analysis.
Two-sided 5% significance tests and 95% confidence intervals are reported throughout. Stata/SE version 11.1 was used for all data analysis.
The UK Central Office for Research Ethics Committees advised that Research Ethics Committee review was not required as the Register was considered a service evaluation. Subsequently the UK National Research Service also advised that since no patient identifiers are collected there is no statutory requirement for consent from the parents for data collection and consent was not sought. The Register guidelines advise clinicians to discuss offering treatment with cooling with the parents and to follow local practice as to whether formal consent for treatment should be sought, but we did not collect this information from registering centres.
2069 notifications were received by the end of July 2011.
Clinical characteristics after birth were consistent with moderate/severe asphyxial injury in most cases (
At the time of data analysis in July 2011, data forms had been received for 1384 (67%) of the 2069 infants registered and the key clinical characteristics of these infants are shown in
In 1331 cases with data available, cooling was started at a median {IQR} of 3.3 {1.5–5.5} hours after birth and the time to achieve the target 33–34°C rectal temperature was a median {IQR} of 1 {0–3} hours. The distribution of the age in hours at start of cooling is shown in
The proportion of infants with suspected clinical seizures before cooling started decreased over the time period of the Register (
Among 1121 babies with data recorded over the first four days, the proportion of infants with suspected clinical or aEEG seizures progressively fell from 62% at 24 hours of age to 13% on the 4th day after birth. There was a similar progressive improvement in the encephalopathy score during the same period. Among 1384 infants, respiratory support with mechanical ventilation or continuous positive airways pressure was required in 1144 (82%) on day 1, 842 (61%) on day 2, 673 (49%) on day 3 and 466 (34%) on day 4.
Thirty eight of 1350 (2.8%) cases were born before 36 weeks gestation; 4 were born at 34 weeks and the others were born at 35 weeks. The median Apgar scores were 1, 4 and 5 at 1, 5 and 10 minutes and the median pH and base excess were 6.82 and −19.1 mmol/L in the cord blood or first blood sample after birth. Suspected seizures occurred in 17/33 (52%) before cooling was started and placental abruption was reported in 18/38 (47%) cases.
Potential adverse events during the first four days after birth that were included in the Register data forms are shown in
We have recently reported the occurrence of subcutaneous fat necrosis in approximately 1% of registered infants.
Of 1362 infants with known outcomes during the first hospital admission, 278 (20%) died before discharge at a median {IQR} of 2.9 {1.4 to 4.1} days of age and 172 (13%) were transferred to a different hospital (
Outcome data at 2 years of age were received from 275/574 (48%) of requests (
Formal neurodevelopmental assessment was carried out in 135/249 (54%) of cases, but a developmental quotient was only obtained in 34 cases; in the other children it could not be obtained or the forms were incomplete, so it was not possible to explore factors associated with neurodevelopmental outcome. A further 8/275 (3%) died following discharge from hospital and before 2 years.
Risk factors significantly associated with adverse outcome at 2 years in the bivariate analyses were: clinical seizures prior to cooling, severely abnormal aEEG, Apgar score at 10 minutes, base excess and Thompson encephalopathy score prior to cooling. The Thompson encephalopathy score was not included in the multivariate model because the score was only available for 70 babies with 2 year outcome data. In the multivariate analysis of adverse outcome (n = 149), the factors that were independently associated with adverse outcome were clinical seizures prior to cooling (p<0.001) and severely abnormal aEEG (p<0.001).
Intervention with moderate cooling for neural rescue in newborns with hypoxic-ischaemic brain injury is the culmination of a series of research studies spanning decades. These proved the potential for neural rescue following perinatal asphyxia
The UK TOBY Cooling Register data indicate that there has been timely, systematic implementation of therapeutic hypothermia in the UK to a standard protocol, so successfully bridging the ‘gap in translation’ from research to clinical use.
Because of a lack of systematic recording, the incidence of neonatal encephalopathy in the UK is uncertain as are the outcomes following encephalopathy. The current rate of notifications to the UK TOBY Cooling Register is very close to our estimate that 750–1125 infants annually would be eligible for therapeutic hypothermia based on previous UK population studies and the term birth rate. We suggest that the rapid uptake and now consistently high rate of provision of therapeutic hypothermia was aided by two factors. First, several UK neonatal centres participated in the TOBY trial and so were ready to implement therapeutic hypothermia into clinical practice; this highlights the value to a healthcare system of active participation in clinical trials. Second, the guidance and the clinical feedback provided nationally by the Register, which demonstrates the value to patients and the UK Health Economy of continued research and development funding following the completion of pragmatic trials.
In a meta analysis of the clinical trials of cooling the mortality rate in the cooled groups was 25% and the rate of disabling cerebral palsy was 21% in survivors, which are very similar to our data from infants reported to the national Register of therapeutic hypothermia in the UK.
In the absence of any specific therapy for a devastating condition, and following encouraging if not definitive results of two previous randomised controlled trials
An initial concern expressed by regulatory authorities and specialist bodies is that therapeutic hypothermia may be applied outside the criteria used in the clinical trials- that is a ‘therapeutic drift’ to treatment with cooling in less severely affected newborns or in conditions other than suspected perinatal asphyxia encephalopathy.
It is to be expected that outside a clinical trial, clinicians will explore therapeutic options more widely on compassionate grounds. For any therapy, clinicians must consider how applicable the supporting evidence is to the individual case, the risk of side effects and the availability of other treatments; it is not sufficient to rely solely on the eligibility criteria of clinical trials, and clinicians may opt to treat patients that differ in some respects from those entered in the clinical trials.
It has been suggested that the severity of encephalopathy assessed clinically or by aEEG is less predictive of subsequent outcome when infants are treated with hypothermia.
Adverse events, such as biochemical and metabolic abnormalities and coagulopathy were frequent during the first four days after birth but it is likely that these were due to the severity of perinatal asphyxia rather than treatment with cooling. Culture proven sepsis was reported in 17% of cases in the first 72 hours following birth, and occurred in just 30/1384 (2%) after the end of cooling, suggesting that sepsis was unlikely to be related to treatment with cooling. Pneumonia has been associated with therapeutic hypothermia in adult studies but this complication was reported in less than 2% of cases notified to the Register. Apart from subcutaneous fat necrosis, which was reported in approximately 1% of cases, other adverse conditions occurred with a similar frequency to that reported in the clinical trials.
There are a number of limitations to the analysis of data from registers, such as: difficulty in ascertaining uptake of registration especially when the incidence of the disorder is not known; reporting bias; and incomplete or erroneous data. Therefore, caution is necessary when interpreting the data especially when comparing outcomes with those reported in clinical trials or with historical data. However, registers are ideal for monitoring practices and patient outcomes outside the narrow context of clinical trials. The UK TOBY Cooling Register was set up specifically to monitor the introduction of therapeutic hypothermia in the UK in infants with perinatal asphyxial encephalopathy. Registration of cases is entirely voluntary; we aimed to encourage participation in the Register by providing clinical guidance including individual advice if required, specific data forms that could be used as part of the clinical record and clinician feedback. The number of notifications during the last 12 months of the Register was close to the estimated number of infants eligible for treatment, which suggests that our data provide a representative account of the use of this therapy in the UK.
The Register data show that in the UK most infants are treated safely and appropriately, according to treatment criteria, within 6 hours of birth, with good temperature control and few complications related to cooling. The rates of death and severe disability are similar to those reported in the randomised controlled trials of hypothermia. These data indicate that therapeutic hypothermia was implemented appropriately within the UK, with significant benefit to patients and the UK health economy. This may be due in part to participation by UK neonatal units in clinical trials, the establishment of the national Register, and its endorsement by national advisory bodies.
Participants in the UK TOBY Cooling Register are listed on
We acknowledge the advice of Professor Neil Marlow on collection of the 2 year outcomes and the help of Mr James Griffiths with data entry and analysis.
We are grateful to the carers and parents of the infants notified to the Register.