• 
• 

Three Part Question

In term babies with HIE in a low resource setting [patient] does prophylactic phenobarbital [intervention] compared to standard treatment alone reduce mortality or disability [primary outcomes] or reduction in the occurrence of seizures [secondary outcome]

Clinical Scenario

You work in a resource poor country and cooling of infants with Hypoxic Ischaemic Encephalopathy (HIE) is not available. A term baby is born with low Apgar scores and admitted to the neonatal unit, where you confirm moderate to severe HIE. As cooling is not available you wonder if starting phenobarbital (PB) would reduce mortality or disability.

Search Strategy

Cochrane, Pubmed and Embase were searched independently (February 2015) using the search strings (MeSH terms in italics): (Hypoxia-Ischemia, Brain OR Hypoxia, Brain OR hypoxic ischaemic encephalopathy OR hypoxic ischemic encephalopathy OR Encephalopathies, Hypoxic-Ischemic OR HIE OR Asphyxia OR Anoxia OR hypoxia) AND (infant, newborn OR newborn OR Neonatal OR Neonate OR neonates OR newborns) AND (phenobarbital OR phenobarbitone).

These revealed 35, 107 and 377 articles respectively of which ten, six, and eight were relevant. Two systematic reviews have previous reviewed this question in 2000 and 2007. Having compared the two systematic reviews we excluded the earlier systematic review by Whitelaw et al as it was superseded by the Cochrane review. We reviewed the articles summated in the 2007 Cochrane review and have not included these three articles in our summary table. The reference lists of all relevant articles were searched and found one further paper. One relevant article was not available for interpretation.

Search Outcome

Once duplicate articles have been accounted seven articles are reviewed here.

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Evans et al 2007 UK	Cochrane review: Seven randomised or quasi-randomised controlled trials all include neonates, of 37 or more completed weeks of gestation, following perinatal asphyxia A meta-analysis comparing barbiturates with conventional therapy following perinatal asphyxia.	Intervention: Anticonvulsants administered in the early neonatal period (within the first seven days of life)	Risks of death	RR 1.13, 95% CI 0.59-2.17	All the studies reporting mortality and neurodevelopmental disability were small in number (n < 40). Only one study assessed neurodevelopmental outcome blind to allocated intervention. The definition of perinatal asphyxia varied between studies.
			Severe neurodevelopmental disability	RR 0.61, 95% CI 0.30- 1.22
			Combined outcome of death or severe neurodevelopmental disability	RR 0.78, 95% CI 0.49-1.23
			Seizures within neonatal period	RR 0.72, 95% CI .0.42-1.23.
Velaphi et al 2013 South Africa	Infants with a gestational age ≥ 34 weeks and/or weight 2 000g, with a Base-deficit of <16mmol/l on measurement of arterial blood gas within an hour of delivery and an Apgar score of <7 at 5 minutes, or required resuscitation for more than 5 minutes.	Randomised control trial. Patients were randomised to: 1. Phenobarbital 40mg/kg (n=50) 2. Normal saline placebo 1ml/kg (n=44).	Seizures	Of the patients who were given phenobarbital, 30.0% developed seizures, compared with 47.7% in the control group (RR 0.63; 95% CI 0.37 - 1.06; p=0.083).	Doctors and nurses in unit were blinded There were no significant differences in mean birth weight, gestation, 5-minute Apgar score, extent of resuscitation required, pH or base deficit Used ‘intention to treat’ principle for 6 control patients who’s code was broken due to on-going seizures not responsive to other anti-seizure medications. Electrographic monitoring was not performed so seizures could have been missed
			Mortality	Of all the 94 patients enrolled (i.e. HIE I, II or III), there was no difference in mortality rates between the two groups (14.0% v. 15.9%).
Gathwala 2011 India	Full term inborn babies with severe birth asphyxia. Tertiary neonatal unit. Ventilation available.	Randomised control trial. Patients were randomized to: 1. PB 40mg/kg (n=36) 2. Control group – standard unit protocol for management of HIE, no placebo given (n=36)	Seizures	Became passive at day three (median 52 hours, range 24-120 hours) in the phenobarbital group compared to day four (median 78 hours, range 24-160 hours) in the Control group (P<0.05).	No blinding Neurological follow up only at 1 month Unclear what anticonvulsant medications were used in control group for seizures
Avasiloaiei et al 2013 Romania	Term newborns with perinatal asphyxia	Randomised control trial. Patients were randomised to: 1. Control group - supportive treatment only (n=23) 2. Single dose of 40 mg/kg Phenobarbital within 4 hours of birth (n=22) 3. Three daily doses of 1000 IU/kg erythropoietin (n=22)	Mortality	The mortality rate was lower in the phenobarbital and erythropoietin groups (both 4.6%) than in the control group (17.4%; P = 0.0087, 95% confidence interval)	Unblinded allocation process Small group sizes once erythropoietin infants excluded.
			Long term neurological outcome	Long-term neurologic follow up showed a high incidence of sequelae in the control group compared to the phenobarbital group (no statistical analysis performed).
Meyn et al 2010 USA	Term infants who received whole body cooling for HIE.	Retrospective case-control study. Patients were allocated to: 1. 40mg/kg/day PB at the time of commencing cooling (n=20) 2. Treatment of seizures as they occurred (n=22)	Seizures	Fewer clinical seizures during their NICU course in cooled infants who received prophylactic phenobarbital (15% Vs 82% P < 0.0001).	No deaths in PB group compared to 3 in the control group (p=0.3) Infants in the prophylactic phenobarbital group achieved a body temperature of 33.5C two hours sooner than the control group (P = 0.03) – confounding. Allocation was not randomised, but left to the practice of attending neonatologists. The sample size was small. Data collected over 9 years.
			Neurodevelopmental impairment	No reduction in neurodevelopmental impairment (23% in prophylactic PB group Vs 45% in controls, P = 0.3).
Ajayi et al 1998 Nigeria	Term babies with Apgar scores of ≤5 at one and five minutes. Mechanical ventilation and blood gas analysis were not available.	Retrospective case-control study. Patients allocated to: 1. 10 mg/kg loading dose of phenobarbital within 1 hour following resuscitation and before the onset of signs of HIE (n=57), (with 5 mg/kg/day maintenance) 2. Phenobarbital if seizures observed (20 mg/kg loading dose and 5 mg/kg/day maintenance) (n=91)	Seizures	Early phenobarbital was associated with a threefold (P 0.025) increase in the incidence of subsequent seizures	Randomisation only done by on-call rota of 2 consultants with different practices (no patient selection bias), other bias may have been present No blinding
			Mortality	Two-fold increase in mortality in treatment group (12% vs. 6%; O.R 2.4 (0.7-8.0); P . 0.1),
Svenningsen et al 1982 Sweden	Term neonates with birth asphyxia.	Retrospective case-control study. Patients were randomized to care bundles: 1. ‘Group A’ treatment : ventilation, sodium bicarbonate, blood transfusion if Hb<14g/dl & diazepam (first line) only when seizures were observed (n=16) 2. ‘Group B’: early phenobarbital 10mg/kg (n=14), with ventilation, sodium bicarbonate, glucose bolus, daily fresh frozen plasma (3 days), transufion if Hb <15g/dl, daily bethamethasone and frusemide for 2-4 days.	Neurodevelopmental outcome	Incidence of normal survivors after early severe asphyxia was 71 % in PB group compared with 25% in controls, P<0.01.	Allocation to two groups in 3 year blocks. Other advances may have led to bias. No blinding. Multiple statistical analyses performed on small sample sizes. Therefore leading to questionable interpretation. Several independent interventions were variable between groups which could have accounted for results e.g. fresh frozen plasma, bethamethasone and furosemide were given to PB group.
			Mortality	Lower in PB group (14%) than in controls (50%), p<0.05.
			Neurodevelopmental outcomes	Neurodevelopmental handicaps in group B was (17%) compared with group A (50 %), p<0.05

Comment(s)

Every year nearly 41% of all under-five child deaths are among newborn infants with the majority of these being in resource poor countries. The three major causes of global neonatal deaths are infections (36%), pre-term birth (28%), and birth asphyxia (23%). Seizures are common following birth asphyxia. Treatment of these seizures with anticonvulsants including phenobarbital is widely practised and this has changed little over recent decades with little evidence for which anticonvulsant is most effective. In developed countries mechanical ventilation facilities are available and effective to support these infants. Recently, therapeutic hypothermia has also been shown to reduce mortality without increasing major disability in survivors and is now an established therapy for term and late preterm newborns with HIE. However, mechanical ventilation and cooling facilities are expensive, require intensive nursing support and are rarely available in low income settings. In this BET we therefore investigate whether phenobarbital is effective for reducing mortality and morbidity. This is an important question as phenobarbital is cheap and readily available agent. The studies we have looked at include: two studies from low-middle income countries (Nigeria and India), two from upper-middle-income countries (South Africa and Romania) and finally two papers from high-income nations (the USA and Sweden). None of the papers were undertaken in low-income nations, as described by the World Bank. One of the studies was undertaken in a facility where ventilation and blood gas analysis were not available and therefore may reflect practice in a low-income nation. Whilst we recognise the challenges of using research from high-income settings to answer this question, it is also important to note that facilities in low resource settings vary hugely, from well funded tertiary units in Capital cities to smaller rural clinics. Secondly, caring for an infant with HIE in a resource-poor country, such as the one described here, will be dictated by the lack of therapeutic cooling facilities. It is therefore worth noting that all but one of the papers included here are relevant in that cooling was not employed. We found seven relevant papers giving light on our outcome measures. The quality of the individual papers found was however mixed. One meta-analysis 1 and three randomised control trials were included, but only one of these used blinding. Two of the further papers included used on-call rotas for allocation whilst the final study compared two periods of three years following the introduction of multiple interventions. The definitions of birth asphyxia varied greatly between papers and this has a significant bearing on future clinical interpretation. For example Svenningsen included infants with severe hypotonia and insufficient respiration after 30 minutes of intensive resuscitation compared whilst Ajayi who used Apgar score of ≤ 5 at 1 and 5 minutes and therefore potentially including much milder cases of asphyxia. Our first primary outcome was related to mortality; a single meta-analysis combining results from five studies demonstrated no difference in risks of death. This finding was supported by three further studies. One of these papers actually found a two-fold increase in mortality, however this was non-significant. Two papers demonstrated a reduction in mortality rates. However the second of these had multiple confounders to explain the findings. Our second primary outcome was related to neurodisability; the meta-analysis combined three studies and showed no significant difference in neurodisability. This finding was supported by three further studies that found no significant benefits. One paper found a benefit in neurodisability but this paper had multiple confounders to explain this finding. A second paper also identified neurological improvements with phenobarbital prophylaxis but this was non-significant. Regarding seizures (our secondary outcome); the meta-analysis combined two studies and found no significant statistical difference in the incidence of seizures. Three further studies not included in this meta-analysis have shown a statistically significant reduction in seizures in neonates who received phenobarbital. Conversely, one study found that prophylactic phenobarbital was in fact associated with a three-fold increase in the incidence of subsequent seizures. The results here are therefore conflicting. However, as there appears to be no long-term beneficial effect in our primary outcomes of mortality and/or neurodisability, preventing seizures may only lead to modest short-term benefits that need to be weighed against possible side-effects. In facilities without mechanical ventilation, central venous access, adequate monitoring facilities and low nurse to infant ratios, the potential side effects (e.g. respiratory support and hypotension) of Phenobarbital could be of serious concern. These practical implications further support our Clinical Bottom Line. Therefore related to our specific PICO question we suggest that prophylactic phenobarbital should not be used in resource poor countries in newborns following birth asphyxia.

Clinical Bottom Line

1. Prophylactic phenobarbital should not be used in resource poor countries to improve survival or prevent neurodisability in newborns following birth asphyxia (GRADE B) 2. There is modest evidence that prophylactic phenobarbital can prevent seizures in the neonatal period in newborns following birth asphyxia (GRADE C). However this has no long-term beneficial effect in mortality and/or neurodisability.

References

1. Evans DJ, Levene MI, Tsakmakis M Anticonvulsants for preventing mortality and morbidity in full term newborns with perinatal asphyxia The Cochrane database of systematic reviews 2007(3):CD001240
2. Velaphi S, Mokhachane M, Mphahlele R, Beckh-Arnold E. . Effect of prophylactic phenobarbital on seizures, encephalopathy and mortality in neonates with perinatal asphyxia South African Journal of Child Health 2013;7(1):7-21
3. Gathwala G, Marwah A, Gahlaut V, Marwah P. Effect of high-dose phenobarbital on oxidative stress in perinatal asphyxia: an open label randomized controlled trial Indian pediatrics 2011;48(8):613-7
4. Avasiloaiei A, Dimitriu C, Moscalu M, Paduraru L, Stamatin M. High-dose phenobarbital or erythropoietin for the treatment of perinatal asphyxia in term newborns. Pediatrics international : official journal of the Japan Pediatric Society 2013;55(5):589-93
5. Meyn DF, Jr., Ness J, Ambalavanan N, Carlo WA Prophylactic phenobarbital and whole-body cooling for neonatal hypoxic-ischemic encephalopathy The Journal of pediatrics 2010;157(2):334-6
6. Ajayi OA, Oyaniyi OT, Chike-Obi UD Adverse effects of early phenobarbital administration in term newborns with perinatal asphyxia Tropical medicine & international health : TM & IH 1998;3(7):592-5
7. Svenningsen NW, Blennow G, Lindroth M, Gaddlin PO, Ahlstrom H Brain-orientated intensive care treatment in severe neonatal asphyxia. Effects of phenobarbitone protection. . Archives of disease in childhood 1982;57(3):176-83.