Effects of Hypothermia for Perinatal Asphyxia on Childhood Outcomes

Study Design

In the TOBY trial, 325 infants with a gestational age of at least 36 weeks who had moderate-to-severe asphyxial encephalopathy and abnormal results on amplitude-integrated electroencephalography (EEG) were randomly assigned within 6 hours after birth to receive standard care alone (control) or standard care with hypothermia to a rectal temperature of 33 to 34°C for 72 hours, followed by slow rewarming. Hypothermia was maintained by nursing the infant on a cooling blanket. The children were enrolled in the study from 2002 to 2006, and follow-up at 6 to 7 years was conducted from 2009 to 2013.

The National Research Ethics Service in the United Kingdom and the relevant ethics review board at each of the institutions outside the United Kingdom approved the TOBY protocol (available with the full text of this article at NEJM.org), and an independent trial steering committee oversaw the study. Written informed consent was obtained from the children's parents.

A letter of invitation, together with an information leaflet and consent form, was mailed to the parents of surviving children. After parental consent had been obtained, a psychologist and a pediatrician, both of whom were unaware of the study-group assignments, performed the assessments, usually at the child's school. The assessment comprised a neurologic examination and a neuropsychological assessment encompassing sensory function, cognition, memory, attention, and executive function, all areas that are likely to be affected by perinatal asphyxia. Questionnaire responses were requested from parents and teachers.

Neurologic Examination

A structured neurologic examination to detect signs of cerebral palsy and minor neurologic dysfunction was performed.^7,8 Neuromotor function was assessed with the use of the Gross Motor Function Classification System and the Manual Ability Classification System; scores on the two assessments range from 1 to 5, with higher scores indicating greater impairment.^9,10

Psychometric Assessment

We used the Wechsler Preschool and Primary Scale of Intelligence III (WPPSI-III) test or the Wechsler Intelligence Scale for Children IV (WISC-IV), standardized for children in the United Kingdom (2004) or as appropriate in other countries, to evaluate general cognitive performance.¹¹ Results on these two tests include a general measure of IQ (the primary outcome) with quotients for verbal and nonverbal performance and processing speed. All results are expressed as an age-standardized score, with a population mean of 100 and a standard deviation of 15.

Other Assessments

We also administered age-appropriate subtests selected from the following domains of the Developmental Neuropsychological Assessment II (NEPSY-II): attention and executive function, visuospatial processing, sensorimotor function, and memory and learning.¹² We report the mean scores for each domain, derived from the mean standardized subset scores.

Since deficits in working memory have been reported in children after neonatal encephalopathy and hypoxia,^13,14 we also assessed scores from the following three subtests of the Working Memory Test Battery for Children: block recall, to assess nonverbal short-term memory; digit recall, to assess verbal short-term memory; and backward digit recall, to assess central executive function.¹⁵

Parents and teachers, who were aware of study-group assignments, completed the Strengths and Difficulties Questionnaire (www.sdqinfo.com) (on a scale of 0 to 40, with higher scores indicating more difficulties) and the Attention Deficit–Hyperactivity Disorder (ADHD) Rating Scale IV (on a scale of 0 to 54, with higher scores indicating more severe symptoms).¹⁶ Appropriate translations were available for the European centers. Parents were also asked to complete questionnaires assessing their child's behavior, everyday memory, use of health care services, and demographic information and provided information about their child's health status with the use of the Health Utilities Index (HUI) as proxy respondents. We converted the responses into multi-attribute utility scores using published algorithms (HUI2 and HUI3); these range from −0.03 to 1.00 for HUI2 and from −0.36 to 1.00 for HUI3; the lowest score represents the utility of the worst possible state, with 0.00 indicating death and 1.00 perfect health, although negative scoring is possible. A change of 0.03 points in the mean overall score is considered to be clinically meaningful.^17,18

We obtained permission from parents to request information about educational attainment from the child's school. Teachers and parents then completed two detailed questionnaires — the Total Academic Achievement Score¹⁹ and the Strengths and Difficulties Questionnaire (with the latter including responses from both teachers and parents) — and provided information about any special educational needs.

Study Outcomes

The primary outcome was the frequency of survival with an IQ score of 85 or higher (1 SD below the standardized mean IQ score for the general population). Other prespecified outcomes were components of the primary outcome; the frequency of survival without neurologic abnormalities, which was defined as an IQ score of 85 or higher, a normal neurologic examination, normal vision, and normal hearing; the full-scale IQ score and subscale scores on the WPPSI-III or WISC-IV; NEPSY-II domain scores; overall memory score; score for total difficulties from the parental questionnaire; overall ADHD rating; prevalence of cerebral palsy; scores for gross motor function and manual ability; the grade of disability, which was categorized as mild disability (an IQ score of 70 to 84, level 1 gross motor function [is able to walk independently but may have some gait abnormalities], or abnormality in one or both eyes with normal or nearly normal vision), moderate disability (an IQ score of 55 to 69, level 2 or 3 gross motor function [has minimal ability to perform gross motor skills or requires assistance with walking], or moderately reduced vision), or severe disability (an IQ score of <55, level 4 or 5 gross motor function [needs adaptive seating or has severely limited mobility], or no useful vision); teacher's score for academic achievement; and score on the Health Utility Index (HUI2 and HUI3).

Partial Assessments and Missing Data

Two assessors who were unaware of the study-group assignments independently classified the children with missing or incomplete results on the WPPSI-III or WISC-IV into two IQ groups (<85 or ≥85) on the basis of additional information that was available. Children were classified into the group with an IQ score of 85 or higher if they had an IQ score of 85 or higher on any full subscale of the WPPSI-III and were unable to complete the other sections because of physical impairments. We included these children's adjudicated outcomes in the primary outcome analysis.

For any partially completed scores on the Strengths and Difficulties Questionnaire and ADHD ratings from parents and teachers, pro rata estimation was used if 10% or less of the items were missing (i.e., up to two items missing on each scale). Children for whom we had no information after the 18-month assessment were not included in the primary outcome analysis.

Sensitivity Analyses

We used two tests to explore the sensitivity of the primary outcome to missing data. In the first test, we assigned nonparticipants an IQ score of 85 or higher at the age of 6 to 7 years for those with a Mental Developmental Index of 85 or higher at 18 months and an IQ score of less than 85 for those with a Mental Developmental Index of less than 85 at 18 months. In the second test, we assigned participating children who were unable to complete the full-scale WPPSI-III or WISC-IV to the group with an IQ score lower than 85.

Statistical Analysis

We calculated relative risks and 95% confidence intervals on the basis of the proportion of children with an IQ score of 85 or higher and the proportions with secondary outcomes at the age of 6 or 7 years in the hypothermia group, as compared with the control group. We prespecified a test of interaction between treatment and abnormality on amplitude-integrated EEG, which provides an index of the severity of the asphyxial insult, and examined the effect of maternal education (classified according to whether the mother had graduated from high school) on the primary outcome.

For normally distributed continuous outcomes, we present means and standard deviations for each group and the mean between-group differences plus 95% confidence intervals. For non-Gaussian distributions, we present the median and interquartile range for each group.

A two-sided P value of less than 0.05 was considered to indicate statistical significance. No adjustment was made for multiple testing of the secondary outcome measures.

Patients

Table 1. Table 1. Demographic and Clinical Characteristics of the Children at Trial Entry.

Figure 1. Figure 1. Enrollment and Outcomes.

Outcome data were available for 280 children (184 survivors and 96 children who died before the assessment at 6 to 7 years of age). Of these children, 245 (88%) were from the United Kingdom, and the rest were from other countries. Parental educational level and socioeconomic status were similar in the two study groups, and the clinical characteristics of the children were similar at trial entry (Table 1; and Tables S1 through S4 in the Supplementary Appendix, available at NEJM.org). A total of 45 children did not participate in the follow-up study and were excluded from the primary analysis (Figure 1). As compared with participants, nonparticipants had a higher rectal temperature at the time of randomization, were significantly less likely to enter the study within 4 hours after birth, and had a higher frequency of severe abnormalities on amplitude-integrated EEG, although the last difference was not significant; at 18 months, nonparticipants had lower scores on the Mental Developmental Index.

Primary Outcome

Of the 184 surviving children who participated in the follow-up evaluation, 140 underwent full-scale IQ testing with the WPPSI-III or the WISC-IV. Another 41 children (18 in the hypothermia group and 23 in the control group) were unable to complete the test, in most cases because of physical impairment (37 children) or because they would not cooperate. Of the 41 children who did not undergo full-scale IQ testing, 35 (15 in the hypothermia group and 20 in the control group) were classified as having an IQ score lower than 85; 3 children in each group were classified as having an IQ score of 85 or higher by the expert assessors. For another 3 children, information was available from the parents only, and the IQ score could not be classified. Ultimately, 277 of 325 children in the TOBY population (85%) were included in the primary analysis. Scores on the Mental Developmental Index at 18 months were available for 39 children who were lost to follow-up at 6 to 7 years of age, and thus 316 children were included in the sensitivity analysis that relied on earlier data for children for whom a current IQ measurement was missing.

Table 2. Table 2. Primary Outcome and Its Components in Children 6 to 7 Years of Age.

The frequency of survival with an IQ score of 85 or higher was 52% (75 of 145 children) in the hypothermia group as compared with 39% (52 of 132 children) in the control group (relative risk, 1.31; 95% CI, 1.01 to 1.71; P=0.04) (Table 2). The number of children who would need to be treated with hypothermia in order to prevent 1 child from dying or having an IQ score lower than 85 was 8 (95% CI, 4 to 145). There was no significant interaction between treatment and grade of abnormality on amplitude-integrated EEG (P=0.97). The results of an analysis adjusted for maternal educational level were materially unchanged, as were the results of a sensitivity analysis that included data from nonparticipants (relative risk, 1.31; 95% CI, 1.02 to 1.67) and a sensitivity analysis that included children who were not able to complete IQ testing and were assigned to the group with an IQ score lower than 85 (relative risk, 1.37; 95% CI, 1.04 to 1.80) (Tables S5 through S8 in the Supplementary Appendix).

Other Outcomes

Table 3. Table 3. Pediatric Assessment in Survivors.

The rates of death did not differ significantly between the hypothermia group (in which 47 of 163 children [29%] died) and the control group (in which 49 of 162 children [30%] died) (Table 2). A total of 86 of the 96 deaths (90%) occurred before the 18-month assessment. A higher proportion of survivors in the hypothermia group than in the control group had an IQ score of 85 or higher (77% vs. 63%; relative risk, 1.22; 95% CI, 1.00 to 1.49; P=0.05) (Table 2). Significantly more children in the hypothermia group than in the control group survived without neurologic abnormalities (65 of 145 [45%] vs. 37 of 132 [28%]; relative risk, 1.60; 95% CI, 1.15 to 2.22) (Table S9 in the Supplementary Appendix). Among survivors, children in the hypothermia group had significantly reduced rates of cerebral palsy (21% vs. 36%, P=0.03) and moderate or severe disability (22% vs. 37%, P=0.03) and had significantly better scores for gross motor function and manual ability (Table 3). The rates of visual and hearing impairments did not differ significantly between the two groups, and head circumference and growth were similar in the two groups (Table S10 in the Supplementary Appendix).

Table 4. Table 4. Psychometric Assessment in Survivors.

Approximately 30% of children did not complete the psychometric tests. Among the children who completed the psychometric tests, there was no significant between-group difference with respect to IQ scores that were measured on a continuous scale and other scores, with the exception of 1 of the 11 scores compared (attention and executive function, P=0.03) (Table 4). There were also no significant differences between groups in mean scores on the index of health care status derived from parental assessments (HUI2 and HUI3) or in parental scores for children's strengths and difficulties and ADHD scores (Tables S9, S10, and S11 in the Supplementary Appendix).

The mean difference in the academic achievement score favored the hypothermia group but was not significant. The proportion of children who required use of special educational resources was lower in the hypothermia group than in the control group (8.2% vs. 26.9%; relative risk, 0.30; 95% CI, 0.12 to 0.79; P=0.01) (Table S12 in the Supplementary Appendix).

Children with asphyxial encephalopathy who were treated with hypothermia shortly after birth were significantly more likely to survive with an IQ score of 85 or higher at 6 to 7 years of age than were children who did not undergo such therapy. Similar proportions of children in the two groups died, but a higher proportion of survivors in the hypothermia group had an IQ score of 85 or higher, and the frequency of moderate-to-severe disability was lower in this group than in the control group.

Data are lacking with respect to longer-term outcomes after hypothermia therapy for perinatal asphyxia. The CoolCap study, which evaluated the efficacy of selective head cooling on the basis of questionnaire data from parents of 62 of 135 children (46%) at 7 to 8 years of age, showed that childhood outcomes broadly correlated with the 18-month assessments, although the study was underpowered to examine the effect of hypothermia on cognitive function at an older age.²⁰ In the Childhood Outcomes after Hypothermia for Neonatal Encephalopathy study, sponsored by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), there was a high follow-up rate among children 6 to 7 years of age, with children in the hypothermia group having a lower rate of death than those in the control group. However, the study showed no significant differences in rates of disability or cognitive outcomes, perhaps because of a lack of statistical power.⁵ In our study, in which we used the same psychometric tests as were used in the NICHD study but used a different cutoff point for the IQ score in our primary outcome, we found significant reductions in the rates of moderate-to-severe cognitive deficiency in the hypothermia group, as compared with the control group, but similar rates of mortality in the two groups.

Our study was limited by the lack of availability of primary outcome data for 15% of the original TOBY trial population (11% of the hypothermia group and 18% of the control group). In anticipation of dropout, we prespecified a sensitivity analysis in which we substituted the Mental Developmental Index score at 18 months for the missing IQ score in order to examine the robustness of the results. The high correlation for neurologic outcomes between earlier assessment and later follow-up that we observed among children who had undergone the two assessments, as well as correlation with the findings of the CoolCap and NICHD studies, provides support for our approach, and the close agreement of the results of the primary and sensitivity analyses increases our confidence in the conclusions.

Several children were unable to complete the full-scale WPPSI-III or WISC-IV test, in most cases because of physical impairments. Assessors who were unaware of the study-group assignments classified these children as having an IQ score below 85 or an IQ of 85 or higher, which allowed them to be included in the primary analysis. Almost all these children (all but six) were classified as having an IQ score that was lower than 85. The results of our primary outcome analysis were not materially altered when all the children without results on full-scale IQ testing were classified as having an IQ score of less than 85.

We found no significant between-group differences for many secondary outcomes assessed, including 10 of 11 psychometric test scores. Prioritizing assessment of IQ as the primary outcome meant that some other tests were curtailed if the assessor determined that the child could not maintain adequate performance. Thus, our study was not adequately powered for many of these comparisons. Most point estimates favored the hypothermia group, though the differences were modest.

In conclusion, our study provides evidence that the benefits of moderate hypothermia after perinatal asphyxia persist into middle childhood.