Therapeutic hypothermia in neonatal hypoxic encephalopathy: A systematic review and meta-analysis

BACKGROUND: Therapeutic hypothermia (TH) is regarded as the most efficacious therapy for neonatal hypoxic encephalopathy. However, limitations in previous systematic reviews and the publication of new data necessitate updating the evidence. We conducted this up-to-date systematic review to evaluate the effects of TH in neonatal encephalopathy on clinical outcomes. METHODS: In this systematic review and meta-analysis, we searched Medline, Cochrane Library, Embase, LIVIVO, Web of Science, Scopus, CINAHL, major trial registries, and grey literature (from inception to October 31, 2021), for randomized controlled trials (RCT) comparing TH vs normothermia in neonatal encephalopathy. We included RCTs enrolling neonates (gestation ≥35 weeks) with perinatal asphyxia and encephalopathy, who received either TH (temperature ≤34°C) initiated within 6 hours of birth for ≥48 hours, vs no cooling. We excluded non-RCTs, those with delayed cooling, or cooling to >34°C. Two authors independently appraised risk-of-bias and extracted data on mortality and neurologic disability at four time points: neonatal (from randomization to discharge/death), infancy (18-24 months), childhood (5-10 years), and long-term (>10 years). Other outcomes included seizures, EEG abnormalities, and MRI findings. Summary data from published RCTs were pooled through fixed-effect meta-analysis. RESULTS: We identified 36 863 citations and included 39 publications representing 29 RCTs with 2926 participants. Thirteen studies each had low, moderate, and high risk-of-bias. The pooled risk ratios (95% confidence interval, CI) were as follows: neonatal mortality: 0.87 (95% CI = 0.75, 1.00), n = 2434, I(2) = 38%; mortality at 18-24 months: 0.88 (95% CI = 0.78, 1.01), n = 2042, I(2) = 51%; mortality at 5-10 years: 0.81 (95% CI = 0.62, 1.04), n = 515, I(2) = 59%; disability at 18-24 months: 0.62 (95% CI = 0.52, 0.75), n = 1440, I(2) = 26%; disability at 5-10 years: 0.68 (95% CI = 0.52, 0.90), n = 442, I(2) = 3%; mortality or disability at 18-24 months: 0.78 (95% CI = 0.72, 0.86), n = 1914, I(2) = 54%; cerebral palsy at 18-24 months: 0.63 (95% CI = 0.50, 0.78), n = 1136, I(2) = 39%; and childhood cerebral palsy: 0.63 (95% CI = 0.46, 0.85), n = 449, I(2) = 0%. Some outcomes showed significant differences by study-setting; the risk ratio (95% CI) for mortality at 18-24 months was 0.79 (95% CI = 0.66,0.93), n = 1212, I(2) = 7% in high-income countries, 0.67 (95% CI = 0.41, 1.09), n = 276, I(2) = 0% in upper-middle-income countries, and 1.18 (95% CI = 0.94, 1.47), n = 554, I(2) = 75% in lower-middle-income countries. The corresponding pooled risk ratios for ‘mortality or disability at 18-24 months’ were 0.77 (95% CI = 0.69, 0.86), n = 1089, I(2) = 0%; 0.56 (95% CI = 0.41, 0.78), n = 276, I(2) = 30%; and 0.92 (95% CI = 0.77, 1.09), n = 549, I(2) = 86% respectively. Trials with low risk of bias showed risk ratio of 0.97 (95% CI = 0.80, 1.16, n = 1475, I(2) = 62%) for neonatal mortality, whereas trials with higher risk of bias showed 0.71 (95% CI = 0.55, 0.91), n = 959, I(2) = 0%. Likewise, risk ratio for mortality at 18-24 months was 0.96 (95% CI = 0.83, 1.13), n = 1336, I(2) = 58% among low risk-of-bias trials, but 0.72 (95% CI = 0.56, 0.92), n = 706, I(2) = 0%, among higher risk of bias trials. CONCLUSIONS: Therapeutic hypothermia for neonatal encephalopathy reduces neurologic disability and cerebral palsy, but its effect on neonatal, infantile and childhood mortality is uncertain. The setting where it is implemented affects the outcomes. Low(er) quality trials overestimated the potential benefit of TH.