Prediction of outcome methods assessing short- and long-term outcome after therapeutic hypothermia

Summary

Therapeutic hypothermia has significantly changed outcomes for newborns suffering neonatal encephalopathy. Outcome predictors established in the pre-cooling era may not automatically be transferred to the cooling era. This article reviews how the reliability of routinely used outcome predictors has changed. We summarize current knowledge about why this may be the case and when to best obtain and analyze different clinical, biochemical, and imaging outcome markers to predict outcome in cooled asphyxiated newborns.

Introduction

Therapeutic hypothermia (TH) for the treatment of neonatal encephalopathy (NE) of hypoxic–ischemic (HI) origin is one of the major improvements in neonatal medicine in the past 15 years. TH became standard treatment in 2010 following the large randomized controlled trials (RCTs) showing that TH significantly reduces death or severe disability in moderately asphyxiated newborns compared to normothermia (NT) treatment [1].

The prediction of neurodevelopmental outcome remains one of the major challenges for clinicians treating newborns with NE, and for parents this is a key question. Great progress had been made in establishing significant associations between early assessments and outcomes in non-cooled newborns suffering from NE, some aiming at prediction of outcome within the first days after birth and others providing information later in the first two weeks. However, there are questions about whether methods used to predict neurodevelopmental outcome in the pre-cooling era can be relied upon, now that cooling is standard care. There are many reasons why this may not be the case:

• Nowadays, there is unfortunately too little concern about applying TH more widely in units and to infants not eligible for the original RCTs. Thus, current cooled cohorts may differ from those previously included in studies assessing outcome predictors.

• Conversely, some of the pre-cooling data include newborns who would have not fulfilled all the entry criteria to the RCTs, hence prediction of outcome may be different from those applicable to the RCTs' strict recruitment criteria.

• Cooling is now initiated much earlier than during the RCTs, when the median time to start TH was 4.5 h [1]. Even though the neuroprotective potency of TH may be best when initiated very early after HI [2], it is not clear whether all affected newborns benefit from early cooling.

• Newborns with early mild encephalopathy (not included in the RCTs) are now often being cooled, some of whom would have spontaneously improved very quickly and would be expected to have a good outcome without TH.

• When newborns are cooled, less injury may occur, so very early markers of injury and outcome used previously are likely to be different.

• TH will induce changes in metabolism affecting biochemical markers. Therefore, the development of injury may be delayed, and hence the optimal timing of outcome markers may change.

This review focuses on methods suitable for use as outcome predictors in the neonatal period for infants with NE of presumed HI origin. The aim is to assess their accuracy and reliability for predicting outcomes up to early childhood, comparing data from the recent pre-cooling era and data from the non-cooling arms of RCTs, to data from infants undergoing TH.

This includes different degrees of encephalopathy and other neurological abnormalities persisting at least a week or more from birth, or death.

This is usually assessed using standardized tests [mainly the Bayley Scales of Infant Development, 2nd edn (BSID-II) or the Griffiths Mental Developmental Scales (GMDS)] that evaluate a child's cognitive and language skills and fine and gross motor development, most usually at 18–24 months. Severe outcome is generally taken as a developmental quotient (DQ) < 70 or severe cerebral palsy (CP; Gross Motor Function Classification System levels 3–5) [3]. Other severe outcomes are central visual difficulties not corrected with spectacles, hearing impairments requiring aids, and seizures requiring anti-epileptic medication. Sometimes secondary microcephaly with a fall of >2 SD from birth in head circumference is included. At the age at which most outcome assessments have been done following NE, behavioral and more subtle communication difficulties are still hard to assess and cognitive problems may well be underestimated. As yet, the longest follow-up of cooled infants is from three of the RCTs at 6–8 years [4], [5], [6].