Home > Health & Research > Health Education Campaigns & Programs > Cochrane Neonatal Review > Ethamsylate for the prevention of morbidity and mortality in preterm or very low birth weight infants

Ethamsylate for the prevention of morbidity and mortality in preterm or very low birth weight infants

Skip sharing on social media links
Share this:

Authors

Rod Hunt1, Edmund Hey2

Background - Methods - Results - Characteristics of Included Studies - References - Data Tables & Graphs


1Department of Neonatal Medicine, Murdoch Children's Research Institute, Parkville, Australia [top]
2Newcastle upon Tyne, UK [top]

Citation example: Hunt R, Hey E. Ethamsylate for the prevention of morbidity and mortality in preterm or very low birth weight infants. Cochrane Database of Systematic Reviews 2003, Issue 3. Art. No.: CD004343. DOI: 10.1002/14651858.CD004343.

Contact person

Rod Hunt

Department of Neonatal Medicine
Murdoch Children's Research Institute
The Royal Children's Hospital Melbourne
50 Flemington Road
Parkville
Victoria
3052
Australia

E-mail: rod.hunt@rch.org.au

Dates

Assessed as Up-to-date: 23 September 2009
Date of Search: 24 August 2009
Next Stage Expected: 24 August 2011
Protocol First Published: Issue 3, 2003
Review First Published: Not specified
Last Citation Issue: Issue 3, 2003

What's new

Date / Event Description

History

Date / Event Description

Abstract

Background

Ethamsylate decreases blood loss in certain clinical situations such as menorrhagia and following some surgical procedures. This potential to reduce bleeding has led to the hypothesis that it may have a role to play in reducing intraventricular haemorrhage in preterm infants.

Objectives

To determine if ethamsylate, when compared to placebo or no treatment, reduces morbidity and/or mortality in preterm infants.

Search methods

We searched the Cochrane Neonatal Group Trials Register (24 August 2009), the Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library 2009, Issue 2), MEDLINE and EMBASE (January 1966 to July 2009) and the Oxford Database of Perinatal Trials.

Selection criteria

Randomised controlled trials or quasi-randomised trials comparing ethamsylate with placebo or no treatment. The initial search for trials enrolling infants born less than 32 weeks gestation was subsequently expanded to include trials enrolling preterm infants < 35 weeks gestation or < 2000 grams birth weight. Studies were included if they reported on outcomes of all children until death or discharge home. Data from reports of neurodevelopmental follow-up were only included if at least 80% of participants were followed up.

Data collection and analysis

Both review authors independently assessed trial quality and extracted data. We calculated relative risk (RR) and risk difference (RD) together with 95% confidence intervals (CI) and used a fixed-effect model for meta-analysis.

Results

Eight studies were identified but only seven trials enrolling 1410 preterm infants were located. There was no significant difference detected in neonatal mortality or neurodevelopmental outcome at two years between infants treated with ethamsylate and controls. Infants treated with ethamsylate had significantly less intraventricular haemorrhage than controls at < 31 weeks (typical RR 0.63, 95% CI 0.47 to 0.86) and < 35 weeks gestation (typical RR 0.77, 0.65 to 0.92). There was also a significant reduction in grade 3 and 4 intraventricular haemorrhage when all infants < 35 weeks gestation (typical RR 0.67, 95% CI 0.49 to 0.94) were analysed as a single group, but not for the group of infants < 32 weeks alone. There was a reduction in symptomatic patent ductus arteriosus at < 31 weeks gestation (typical RR 0.32, 95% CI 0.12 to 0.87). There were no adverse effects of ethamsylate identified from this systematic review.

Authors' conclusions

Preterm infants treated with ethamsylate showed no reductions in mortality or neurodevelopmental impairment despite the reduction in any grade of intraventricular haemorrhage seen in infants < 35 weeks gestation.

Plain language summary

Ethamsylate for the prevention of morbidity and mortality in preterm or very low birth weight infants

Prematurely born infants are at risk of bleeding into the brain in the first few weeks of life. This is called intraventricular haemorrhage. The risk of this occurring is greatest to infants who are born less than 32 weeks gestation. Many potential therapies have been studied to determine if they might reduce the risk of this bleeding. One such therapy is a drug called Ethamsylate. It is not exactly known how this drug works, but it appears to reduce bleeding in other clinical situations, such as excessive menstrual bleeding and after some types of surgery.

A total of seven studies with 1410 preterm infants were included in this review. Most of these initial studies were conducted between 1980 and 1990. Preterm infants treated with ethamsylate had similar outcomes with respect to death and disability at the age of two years when compared to infants who were treated with a placebo. Infants born less than 35 weeks gestation appeared to have less intraventricular haemorrhage when treated with ethamsylate compared to controls, however this did not lead to improved developmental outcome in later childhood. There were no adverse effects noted with ethamsylate treatment.

Based on these results, routine use of ethamsylate for prematurely born infants to prevent intraventricular haemorrhage cannot be recommended. It is highly unlikely that any further trials will be conducted to explore this clinical question.

[top]

Background

Description of the condition

Intraventricular haemorrhage (IVH) remains a significant cause of morbidity and mortality for the prematurely born infant (Volpe 2002). The site of primary bleeding is the germinal matrix, which contains a fragile network of blood vessels lining the lateral ventricles. Four degrees of IVH have classically been identified by ultrasound examination. Grade 1 haemorrhage is bleeding into the germinal matrix with minimal intraventricular blood. Grade 2 haemorrhage is bleeding that breaks into the ventricle from the germinal matrix, but of a severity that does not cause ventricular dilatation. Grade 3 haemorrhage is intraventricular bleeding of germinal matrix origin which is severe enough to be classed as having caused posthaemorrhagic hydrocephalus. These effectively equate to Grades I, II and III IVH in the classification used by Papile 1978 (a classification originally developed to describe the appearance of the brain on computed tomography, but later widely utilised to describe ultrasound findings in the period immediately after birth). They also closely parallel the grading system recommended by Volpe 2002. Longer-term outcomes from grade I/II IVH are now generally accepted to be relatively benign, with grade III/IV IVH and consequent post-haemorrhagic ventricular dilatation being more consistently associated with impaired neurodevelopment. However, advances in neuroimaging have brought with them increased understanding of other mechanisms of cerebral insult in the preterm infant, shifting current day focus away from IVH as the major precursor for neurodevelopmental impairment.

The interpretation of what Papile 1978 originally termed grade IV IVH is more complex. It was initially thought to represent bleeding originating in the germinal matrix but extending into the parenchyma of the brain. However, it is now thought that the periventricular echo densities sometimes seen in the parenchyma of the brain soon after birth usually represent venous infarction rather than actual bleeding, and that such echo densities can occur even when there is no significant bleeding into the ventricles. Conversely, posthaemorrhagic hydrocephalus (grade III IVH) can occur with or without ultrasound features suggestive of parenchymal damage. These findings have very different prognostic implications. Most would now accept that the ischaemic changes associated with parenchymal damage cannot be described adequately using the classification used by Papile and may take some time to become detectable on ultrasound. Therefore, for the purpose of this review, the cases where there was evidence of altered parenchymal echogenicity or an increase in intraventricular volume will be considered together and labelled cases of grade III-IV IVH.

The limitations of ultrasound techniques available for assessing insults to the brain of perinatal origin in the 1980s need to be remembered when interpreting the outcomes of the trials done in that decade. Early ultrasound findings are mere surrogate measures for long-term morbidity, and often flawed measures at that. For this reason an assessment of functional outcome at least two years after birth is the only outcome of real consequence.

The exact aetiology of IVH in the human preterm infant remains uncertain. Animal studies using a beagle puppy model favour an ischaemia-reperfusion model for IVH (Goddard-Finegold 1982). The uncertainty in humans has confounded the search for an effective preventive agent for IVH. Over the past thirty years a number of possible therapies have been evaluated.

Description of the intervention

Ethamsylate (diethylammonium 1, 4-dihydroxy-3-benzene sulfonate) has been evaluated as a possible treatment to prevent IVH. The recommended International Non-proprietary Name for this drug is etamsylate, but the older name has been used in this review because this was the name used by all those who studied it before 2000. This non-steroidal drug was shown to be effective in reducing blood loss from menorrhagia (Harrison 1976) and following transurethral resection of the prostate (Symes 1975). The benefits of ethamsylate in reducing bleeding in these settings led to the postulate that it may be of benefit in reducing IVH.

The precise mechanism of action of ethamsylate is unknown. Ethamsylate has been shown to reduce bleeding time and blood loss from wounds (Vinazzer 1980). This appears to relate to increased platelet aggregation mediated by a thromboxane A2 or prostaglandin F2 alpha dependent mechanism (Kowacs 1981; Okuma 1982). It has also been associated with decreased levels of 6-ketoprostaglandin F1alpha, a metabolite of prostacyclin. Prostacyclin is a potent vasodilator (and may thus be implicated in reperfusion) and can disaggregate platelets (Rennie 1986). The prostaglandins may have a role in regulating cerebral blood flow, however, ethamsylate appears to have no effect on cerebral blood flow (Rennie 1989).

How the intervention might work

The potential benefits of ethamsylate include those associated with reduction in bleeding tendency, such as reduction in IVH and those associated with prostaglandin inhibition such as closure of the patent ductus arteriosus (Amato 1993; Rosti 1994).

The only adverse event reported in relation to the use of ethamsylate is transient hypotension (Langdon 1977; Watson 1977). Ethamsylate appears to have no anticyclooxygenase activity (Amato 1994) and, therefore, is not associated with the same side effect profile as other inhibitors of prostaglandin synthesis.

Why it is important to do this review

This review is important to summarise a number of randomised controlled trials that have investigated the potential beneficial effect of using ethamsylate to reduce intraventricular haemorrhage and in so doing improve neurodevelopmental outcome.

Objectives

To determine if ethamsylate when compared to placebo or no treatment reduces morbidity and/or mortality in preterm infants born < 35 weeks gestation or with birth weight < 2000 g.

[top]

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials or quasi-randomised trials comparing ethamsylate with placebo or no treatment.
Trials had to have adequate randomisation and > 80% follow-up of participants for outcome measures as described below. Where quasi-randomised trials were identified, meta-analysis was performed both including and excluding these trials.

Types of participants

Preterm infants born either before and including 34 weeks plus six days completed gestation or with a birth weight less than 2000 g.

Types of interventions

Ethamsylate administration at any dose compared with either no treatment or placebo.

Types of outcome measures

Primary outcomes

Primary outcome measures included any of the following:

  1. neonatal mortality (to 28 days post delivery) and mortality to hospital discharge;
  2. neurodevelopmental disability at two years of postnatal age defined as neurological abnormality, including cerebral palsy on clinical examination, developmental delay more than two standard deviations below population mean on any standard test of development, or blindness (visual acuity < 6/60), or deafness (any hearing impairment requiring amplification) at any time after two years corrected.
Secondary outcomes

Secondary outcome measures included any of the following:

  1. ultrasound evidence of IVH detected prior to term corrected described as any IVH (grades 1 to 4) and severe IVH (grades 3 and 4). Where the outcome IVH is being analysed, a subgroup analysis stratifying for the presence or not of IVH prior to treatment will be performed;
  2. symptomatic patent ductus arteriosus (PDA) requiring treatment with either indomethacin or surgical ligation;
  3. duration of respiratory support required including;
    • duration of endotracheal intubation (days)
    • duration of supplemental oxygen administration (days)
  4. evidence of adverse events from ethamsylate administration;
    • hypotension defined as mean blood pressure less than gestational age.

Search methods for identification of studies

Electronic searches

We searched the Neonatal Review Group Trials Register (24 August 2009) using their standard search strategy (see Neonatal Review Group details for more information). This was supplemented by additional searches of the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2009, Issue 2), MEDLINE and EMBASE (January 1966 to July 2009) and the Oxford Database of Perinatal Trials.

We searched MEDLINE and EMBASE (January 1966 to July 2009) including MeSH searches using the following terms (["infant, premature", "ethamsylate", "etamsylate"] and text searches using the terms "[ethamsylate and etamsylate]". Searches were limited to "clinical trials". No language restrictions were applied.

We contacted investigators of included studies to request specific information not included in their publications. Diane Elbourne provided information from the EC Trial Group 1994 regarding mortality of subjects to 28 days of age. Richard Cooke was able to provide long-term follow-up information on all children in who survived to discharge. A personal communication from Kishore Sanghvi confirmed that all hospital deaths reported in Sanghvi 1999 occurred prior to 28 days of age. Jane Schulte provided data on long-term outcomes for the study of Benson 1986.

Searching other resources

Additional searches included previous reviews plus cross references, abstracts, conference- and symposia proceedings (Perinatal Society of Australia and New Zealand 1998 to 2009 and Pediatric Academic Societies meetings 1998 to 2009). We contacted investigators about additional studies potentially eligible for inclusion. We also contacted manufacturers for information on any unpublished trials. Unpublished studies were eligible for review.

Data collection and analysis

This systematic review followed the Cochrane Collaboration methodology according to guidelines of the Neonatal Review Group.

Selection of studies

Both review authors independently identified the studies that were included.

Data extraction and management

Both review authors independently extracted the data.

Assessment of risk of bias in included studies

We used the standard methods of the Cochrane Neonatal Review Group. The methodological quality of the studies was assessed using the following key criteria: allocation concealment (blinding of randomisation), blinding of intervention, completeness of follow-up, and blinding of outcome measurement/assessment. For each criterion, assessment was yes, no, can't tell. Two review authors separately assessed each study. When necessary, additional information and clarification of published data was requested from the authors of individual trials. Any disagreement was resolved by discussion. This information was added to the table 'Characteristics of Included Studies'.

In addition, the following issues were evaluated and entered into the Risk of Bias Table:

  1. Sequence generation: Was the allocation sequence adequately generated?
  2. Allocation concealment: Was allocation adequately concealed?
  3. Blinding of participants, personnel and outcome assessors: Was knowledge of the allocated intervention adequately prevented during the study? At study entry? At the time of outcome assessment?
  4. Incomplete outcome data: Were incomplete outcome data adequately addressed?
  5. Selective outcome reporting: Were reports of the study free of suggestion of selective outcome reporting?
  6. Other sources of bias: Was the study apparently free of other problems that could put it at a high risk of bias?

Measures of treatment effect

Relative risk (RR) and risk difference (RD) were calculated for dichotomous data and weighted mean difference (WMD) for continuous data, with 95% confidence intervals (CI) for all analyses. The number needed to treat (NNT) and associated 95% CI were determined for a statistically significant change in the RD.

Assessment of heterogeneity

We estimated the treatment effects of individual trials and examined heterogeneity between trials by inspecting the forest plots and quantifying the impact of heterogeneity using the I2 statistic. If we detected statistical heterogeneity, we explored the possible causes (for example, differences in study quality, participants, intervention regimens, or outcome assessments) using post hoc subgroup analyses.

Data synthesis

Meta-analyses were performed using the fixed-effect model and heterogeneity tests were applied to ensure that pooling of data was valid.

Subgroup analysis and investigation of heterogeneity

Subgroup analysis was performed for a population of infants where the risk of IVH is more relevant based on current neonatal practice (infants born < 31 completed weeks or < 1500 g). Subgroup analysis was performed based on the methodology of the study, with randomised trials being analysed separately from analyses into which data from all trials is included. Further subgroup analysis was performed based on whether or not the subject had IVH detected on their initial ultrasound scan, i.e. after randomisation but prior to receiving either ethamsylate or placebo/control.

Sensitivity analysis

We performed sensitivity analyses to evaluate the effect of omitting less methodologically rigorous trials.

[top]

Results

Description of studies

Eight studies were identified by the search strategy. These studies were EC Trial Group 1994 (European Community Trial Group) with neurodevelopmental follow-up reported in Elbourne 2001, Amato 1993 which was also published as Amato 1992, Erdem 1988; Chen 1993 and Sanghvi 1999. Only two studies were eligible for inclusion in this subgroup analysis for participants < 31 completed weeks or birth weight < 1500 g. These studies were Morgan 1981 and Benson 1986.

Morgan 1981 recruited patients from Liverpool, UK and studied 73 infants with birth weight < 1500 g and excluded infants with lethal congenital malformation. Three infants were not considered in the analysis because they were found to have IVH before randomisation. Thirty-five infants were randomised to ethamsylate and the remaining 35 received placebo. Subjects received 0.1 ml/kg of either ethamsylate (equal to 12.5 mg/kg) or placebo as an intramuscular injection within two hours of birth and then six-hourly for four days. They reported short-term outcomes, specifically IVH, only. Cranial ultrasounds were performed prior to study entry and then daily for the first week of life. A description of the scanner used and limited description of the views recorded was provided, however, there are no data regarding the inter- or intra-observer reliability of the operators performing the scanning. Audit of the quality of scans was not undertaken as they were being carried out by the clinicians responsible for the patients. IVH was classified according to Papile 1978. Incidence and cause of neonatal death were also reported. Limited longer-term follow-up data were obtained directly from the authors.

Benson 1986 recruited patients from South Wales and the West of England and studied 366 infants with birth weight less than/or equal to 1500 g. Four of these infants with lethal congenital malformations were excluded. Two further infants were withdrawn from the study for logistic reasons (parental refusal and transfer to non-participating centre). Of the remaining 360 infants, 183 were randomised to receive ethamsylate and 177 placebo. Infants received 0.1ml/kg of either ethamsylate (12.5 mg/kg) or placebo (normal saline) within an hour of birth either i.v. or i.m., followed by six-hourly doses i.v. for four days. Only infants for whom there was no evidence of IVH on initial ultrasound scan were included in the analysis of incidence and severity of IVH. Data for 30 infants were, therefore, not reported for this outcome. We have dealt with this by subgroup analysis. Cranial ultrasound was performed soon after birth, and then on days three, seven and fourteen. There are no data to support the premise that cranial ultrasounds were standardised. IVH diagnosed by cranial ultrasound was classified using a modification of Papile 1978. Specifically, grade 1 referred to subependymal haemorrhage, grade 2 referred to intraventricular haemorrhage and grade 3 referred to parenchymal haemorrhage. This made it impossible to distinguish grades 2 and 3 under the classification of Papile 1978. Cause of neonatal death was reported, as were complications during the first 14 days of life - including hypotension and PDA, although it is not clear whether all PDA reported as a complication was treated with either indomethacin or surgical ligation. It is assumed for the purpose of the systematic review that those infants listed as having PDA were treated either medically or surgically.

The long-term neurodevelopmental follow-up for these infants is reported by Schulte 2005. Subjects were seen between 3.5 and 4.2 years of age and assessed with the McCarthy Scales of Children's Abilities, a full physical examination including vision test and audiology and a standardised neurological examination.

EC Trial Group 1994 involved collaborators from France, Greece and the United Kingdom. Collectively, they studied 334 infants who were deemed to be eligible based on gestational age of less than 33 weeks. For these infants, eligibility was further qualified by the statement that the treating physician had to be in a state of equipoise regarding the use of ethamsylate for such an infant. One hundred and sixty-five infants were randomised to receive ethamsylate and the remaining 169 were randomised to be controls. Sixty-nine per cent of treatment group infants and 70% of controls had a birth weight < 1500 g. Treatment subjects received 12.5 mg/kg ethamsylate either i.v. or i.m. within four hours of birth, followed by 15 further doses i.v. every six hours for four days. Control infants received nothing other than standard care. The authors report that one control infant received ethamsylate (number of doses unknown) and 82% of treated subjects received the full course of ethamsylate, with death being the major reason for non-completion of the treatment course. Subjects had a minimum of two cranial ultrasounds but the timing of these varied depending on the study centre and the way in which they were performed was not standardised. Reporting of cranial ultrasounds was by a modified Levene 1981 classification. Their definition of a major cerebral problem as "any lesion affecting the brain parenchyma, and/or hydrocephalus, and/or intraventricular haemorrhage complicated by ventricular dilatation" allows capture of severe IVH (grade 3 or 4). However, into this group there may also be included cases of periventricular leukomalacia (PVL) detectable at a three week cranial ultrasound scan, and PVL may develop independent of any grade of IVH. The major outcomes from this initial study were death, disability at three months of age and use of health resources.

The long-term neurodevelopmental follow-up for these infants is reported by Elbourne 2001. Subjects were seen at approximately two years of age. Neuromotor impairment was determined by an examining paediatrician and developmental delay was defined as mild (less than six months behind), moderate (6 to 12 months behind) or severe (> 12 months behind). Different approaches are known to have been used to assess developmental delay in the participating units, but because the randomisation process ensured a balance between active and control patients in each centre, this will have had little effect on the trial's ability to detect a difference in the number of children with developmental delay in the two trial groups.

Amato 1993 recruited patients from Berne, Switzerland and enrolled 20 preterm infants with severe RDS whose birth weight was between 700 g and 2000 g. Infants were excluded if there had been prolonged rupture of the membranes for greater than three weeks, grade 3 or 4 intraventricular haemorrhage on serial ultrasound scanning within 15 hours of birth, evidence of perinatal asphyxia (Apgar score < 4 at one minute of life or < 6 at five minutes of life) or major congenital anomalies. Ten infants were randomised to each group. The treatment group received their first dose of ethamsylate (12.5 mg/kg) intravenously within four hours of birth and thereafter received six-hourly doses for four days. The placebo group received 0.1 ml/kg normal saline at the same dosing interval. The primary outcome of this study was development of symptomatic PDA based on clinical criteria (increased heart rate > 150 beats per minute, characteristic murmur, hyperdynamic precordium with collapsing pulses) with supportive echocardiographic findings (increase in left atrium/aortic root dimension ratio > 1.30). It is assumed for the purpose of this review that those infants with PDA defined by these criteria were treated for such either medically or surgically. PIVH (grades 1/2 and 3/4) and mortality to hospital discharge were also reported. Frequency of cranial ultrasonography was not reported. Grade 3 or 4 IVH determined by serial ultrasound scan is cited as an exclusion criteria and then reported as an outcome. This trial was also reported in Amato 1992 (no mention of this publication was made in the subsequent paper). No long-term follow-up was reported.

Erdem 1988 recruited babies from Ankara, Turkey and studied 246 infants whose gestational age was < 34 weeks (excluding infants with lethal congenital malformations). The authors report that eligible infants were consecutively assigned a trial number and allocated to treatment with ethamsylate or no treatment. Exactly how they were allocated is not stated and 146 infants received ethamsylate while only 100 received no treatment, suggesting that sequential allocation was not employed. Ethamsylate was given unblinded according to a regimen similar to that employed by all other trials (12.5 mg/kg i.v. within an hour of birth followed by six-hourly doses over four days to maximum dose of 200 mg/kg). The only outcome reported was neonatal mortality, and mortality before seven days of age. Cranial ultrasonography was not available and so PIVH is not reported. No long-term follow-up was reported. Given the uncertain nature of randomisation, we conducted meta-analyses both including and excluding this trial.

Chen 1993 recruited babies from Taichung, Taiwan and enrolled 171 infants with birth weights < 1751 g into their randomised controlled trial. Babies with PIVH on initial ultrasound (performed within an hour of delivery) were excluded. Eighty-six infants were treated with ethamsylate, beginning within one hour of birth (same regimen as other studies) and control infants received 0.1 ml/kg normal saline i.v. at the same dosing interval. The primary outcome was detection of PIVH. Cranial ultrasonography was performed on days one, two, three, five, seven and fourteen and PIVH was classified according to the grading system of Papile 1978. No long-term follow-up was reported.

Sanghvi 1999 recruited babies from Mumbai, India and enrolled 200 infants born < 34 weeks gestation. They excluded infants with congenital malformations (eight), family history of bleeding disorders and Apgar score < 5 at five minutes. Eight infants died after randomisation (five due to sepsis and three due to respiratory distress), however, their allocated group was not stated and they were excluded from further analysis. Short term outcomes were reported on 192 infants. Treatment subjects received ethamsylate 12.5 mg/kg i.v. within one hour of birth and thereafter six hourly for four days. Control subjects received only routine medication. The primary outcome was PIVH and cranial ultrasonography was performed on days three to five, 10 to 14, 28 to 30 and then pre-discharge. Classification of PIVH was according to the grading system of Papile 1978. The authors reported that none of the infants exhibited any side effects of ethamsylate. They did report that hypotension requiring inotropic support was seen in six infants in the treatment group but that this was not attributed to ethamsylate; the timing of the hypotension or the presumed cause were not stated. No long-term follow-up was reported.

Two papers identified by the initial search strategy were not really eligible for inclusion. The findings reported by Cooke 1984 were not used because this report included outcome information on non-randomised as well as randomised children. This paper outlined two-year outcomes for babies in the Morgan 1981 trial but the data could not be used because it had been combined with data from other non-randomised children. Similarly, the findings reported by Rennie 1986 simply gave the 6-ketoPGF1 alpha levels seen in some of the children in the Benson 1986 trial (see table 'Characteristics of Excluded Studies').

The study report for one study Liu 1997 was not locatable. We intend to include this data if the report can be located.

Risk of bias in included studies

Randomisation

Morgan 1981

The method of randomisation was not stated, but it was performed by the drug company supplying the drug and placebo and the randomisation code was held by the drug manufacturer until the end of the trial. Benson 1986 stated that randomisation was by block randomisation tables generated by the manufacturer. EC Trial Group 1994 had valid randomisation of subjects - randomisation was within balanced blocks of variable size and stratified by centre. Amato 1993 stated that randomisation to treatment or control groups was by use of cards in sealed envelopes. Chen 1993 and Sanghvi 1999 claimed random allocation of subjects into two groups but the method of randomisation was not stated. Erdem 1988 conducted a quasi-randomised trial. The method of group allocation was not specified and they had 146 infants in their treatment group and only 100 infants in their control group. For subgroup analyses based on randomisation, we considered only Erdem 1988 to be quasi-randomised.

Allocation concealment:

Morgan 1981

Randomisation was performed by the manufacturer, independent of study centre. Benson 1986 had adequate allocation concealment, in that eligible infants were allocated the next consecutive trial number and trial numbers were obtained from block randomisation tables generated by the manufacturer of the drug. EC Trial Group 1994 had adequate allocation concealment - randomisation occurred when the clinician contacted an off-site randomisation centre. Amato 1993 had adequate allocation concealment - cards in sealed envelopes were used. In Erdem 1988 and Chen 1993 concealment of allocation was not stated. Sanghvi 1999 had adequate allocation concealment, using opaque envelopes to conduct randomisation.

Blinding of intervention:

Morgan 1981

Ethamsylate or placebo were dispensed by the manufacturer in identical ampoules. Benson 1986 stated that the ethamsylate and placebo were both provided by the manufacturer in identical packs. The treatment and placebo were visually indistinguishable. EC Trial Group 1994 did not have blinding of treatment - the control group did not receive any placebo. In Amato 1993 and Chen 1993 the control groups received saline solution rather than ethamsylate. The two substances were probably visually indistinguishable but whether or not the trialists were blinded to the intervention is not stated. In Erdem 1988 and Sanghvi 1999 the control groups received nothing other than routine therapy and the treatment groups received intravenous ethamsylate six hourly for the first four days.

Blinding of outcome measurement:

Morgan 1981

The cranial ultrasonographer was blinded to which arm of the trial the baby had been allocated. In Benson 1986, the cranial ultrasonographer was blinded to study arm by virtue that randomisation was generated by the manufacturer and ethamsylate and placebo were identical in appearance. In Schulte 2005, a single assessor performed all the long-term follow-up and was blind to the treatment arm of the subject. For the EC Trial Group 1994, early cranial ultrasound assessments were undertaken by staff who were unaware which arm of the trial the subject was in. Assessment at two years was reported as having been done blind to trial allocation by staff in the four French centres and the same was also 'usually' true in the six Greek centres. Software constraints mean that we have had to classify allocation concealment a little inadequately as "unclear". In Amato 1993 and Erdem 1988, blinding of outcome measurement was not stated. In Chen 1993 and Sanghvi 1999, the clinician evaluating the cranial ultrasounds was unaware of the treatment status of the patients.

Completeness of follow-up:

Schulte 2005

In the long-term follow-up of the subjects in Benson 1986, 97% of survivors were assessed. In the long-term follow-up of EC Trial Group 1994, 82% of survivors were assessed at two years of age (Elbourne 2001). Dr Cooke was able to provide follow-up information on all the children in Morgan 1981 who survived to discharge.

Effects of interventions

ETHAMSYLATE VERSUS PLACEBO (COMPARISON 1)

Results were calculated for one subgroup of infants born with gestation < 31 completed weeks or birth weight < 1500 g, as well as for the entire study population.

Neonatal mortality (to 28 days post delivery) - Outcome 1.1:
Analysis for entire study population, including infants < 35 completed weeks gestation or < 2000 grams (Outcome 1.1.1):

Data from five trials (Morgan 1981; Benson 1986; Erdem 1988; EC Trial Group 1994; Sanghvi 1999) were included to give 576 infants treated with ethamsylate and 626 controls. There was no significant difference in neonatal mortality when all trials that reported data for this outcome were considered (typical RR 1.01, 95% CI 0.83 to 1.23).

Analysis for entire study population (randomised trials only) (Outcome 1.1.2):

Erdem 1988 was a quasi-randomised study in which management and mortality of preterm infants was clearly very different to the other trials. When the Erdem 1988 trial was excluded there were 476 treated infants and 480 controls. There was no significant difference in neonatal mortality (typical RR 1.23, 95% CI 0.92 to 1.64). Sanghvi 1999 report in their publication mortality figures for death in hospital, however, subsequent personal communication (with thanks to Dr Kishore Sanghvi) confirmed that all of these deaths occurred prior to 28 days postnatal age.

Infants < 31 completed weeks gestation or < 1500 grams (Outcome 1.1.3):

Data were included from three trials which contributed 237 infants treated with ethamsylate and 242 control infants. The authors of Morgan 1981 were able to provide neonatal mortality figures and of the 35 infants treated with ethamsylate, 26 survived for 28 days. Of the 35 children in the control group, 28 survived to 28 days. Benson 1986 reported mortality to 14 days postnatal age rather than 28 days, however, in their follow-up paper (Schulte 2005) they give mortality figures to 28 days and this is included in this category. In the Erdem 1988 trial, neonatal mortality figures were provided and stratified for gestational ages 27 to 28, 29 to 30, 31 to 32 and 33 to 34 weeks. We have reported aata for the subgroups 27 to 28 and 29 to 30 days in this category. The split of patients that were 31 or 32 weeks is not currently available. There was no significant difference in neonatal mortality within this strata with typical RR 1.28 (95% CI 0.86 to 1.90).

Infants < 31 completed weeks gestation or < 1500 grams (randomised trials only) (Outcome 1.1.4):

Morgan 1981 and Benson 1986 contributed 218 treated infants and 212 controls. There was no significant difference in neonatal mortality (typical RR 1.28, 95% CI 0.86 to 1.90).

Mortality to hospital discharge - Outcome 1.2:

This analysis is an extension of the first outcome. It includes data from studies reporting neonatal mortality, assuming that those infants died in hospital under 28 days of age. Benson 1986 reported figures for death for up to 28 days and beyond 28 days. Similarly, Amato 1993 and Chen 1993Amato 1993 reported mortality figures without commenting when these deaths occurred. We have assumed that they occurred prior to hospital discharge as no longer-term data were reported for these studies and we have not been able to confirm when these deaths occurred. Again, this analysis has been run twice: The first time including the quasi-randomised Erdem 1988 trial and the second analysis excluding these data. Results for mortality to hospital discharge are as follows and none of them were statistically significant.

Analysis for entire study population (Outcome 1.2.1):

Data from Morgan 1981; Benson 1986; Erdem 1988; Amato 1993; Chen 1993; EC Trial Group 1994 and Sanghvi 1999 provided a total of 672 treated infants and 721 controls. Typical RR 0.97 (95% CI 0.80 to 1.17).

Analysis for entire study population (randomised trials only) (Outcome 1.2.2):

There were 572 treated infants and 575 controls. Typical RR 1.11 (95% CI 0.86 to 1.44).

Infants < 31 completed weeks gestation or < 1500 grams (Outcome 1.2.3):

Morgan 1981; Benson 1986; Erdem 1988 and Chen 1993 contributed 296 ethamsylate treated infants and 298 controls. Typical RR 1.19 (95% CI 0.83 to 1.71).

Infants < 31 completed weeks gestation or < 1500 grams (randomised trials only) (Outcome 1.2.4):

There were 277 ethamsylate treated infants and 268 controls. Typical RR 1.19 (95% CI 0.83 to 1.71).

Neurodevelopmental disability at 2 years of postnatal age in surviving children available for follow-up (Outcome 1.3):

The authors of Morgan 1981 were able to provide information beyond that which was published (with thanks to Richard Cooke). Of the 35 children treated with ethamsylate, 11 died by the age of three years; 2 of 24 survivors had spastic diplegia with cognitive delay and two others had significant cognitive delay without cerebral palsy. In the control group, seven died by the age of three years; 3 of 28 had developed cerebral palsy with some cognitive delay, one child with severe fetal growth retardation was dysmorphic with severe developmental delay and one child had trisomy 21.

Data from Benson 1986 were also included. Outcome at 3.5 to 4.2 years was known for all of the 137/183 surviving subjects in the treatment arm, and all 139/177 surviving control infants in their trial. Sixteen children developed cerebral palsy, nine had cognitive delay (general cognitive index < 70), three were registered blind and four had sensorineural hearing loss requiring aids. There was overlap between these categories such that 18 children in total had major impairment at the time of assessment. In the control arm, 14 children had cerebral palsy, 19 had cognitive delay (with general cognitive index < 70), one was blind and two were deaf - 22 control children in total were considered to have major impairment.

From the EC Trial Group 1994 study, after assessment at two years, information was available for 101 of 129 possible survivors in the treatment arm (28 out of 165 were lost to follow up; 83% retained), and 103 of a possible 134 control infants (31 out of 169 were lost to follow up; 82% retained). They reported figures for "any impairment or disability" in survivors assessed at two years and this is reported in our meta-analysis.
There was no long-term follow-up information available from any of the other trials.

Analysis for entire study population (Outcome 1.3.1):

Data from Morgan 1981; Benson 1986 and EC Trial Group 1994 provided 262 ethamsylate-treated survivors and 270 control survivors. There was no significant difference in neurodevelopmental disability at two years of age between these two groups (typical RR 0.79, 95% CI 0.53 to 1.17). No further data were available for this analysis.

Infants < 31 completed weeks gestation or < 1500 grams (Outcom 1.3.2):

Morgan 1981 and Benson 1986 contributed 161 ethamsylate-treated survivors and 167 control survivors with known outcome. There was no difference between the groups with respect to neurodevelopmental disability at two years of age (typical RR 0.57, 95% CI 0.30 to 1.07).

There were no long-term data available from the quasi-randomised study of Erdem 1988 so the above results reflect analysis of randomised controlled data only.

Death or disability at 2 years of postnatal age in children with known outcome (Outcome 1.4):

For this analysis we included all data for death at any stage and neurodevelopmental disability by at least two years of age. We ran the analysis twice; first with the quasi-randomised Erdem 1988 trial and then without. Data from the Morgan 1981; Benson 1986 and EC Trial Group 1994 were clarified by personal communication with the authors of these studies.

Analysis for entire study population (Outcome 1.4.1):

Data from seven trials (Morgan 1981; Benson 1986; Erdem 1988; Amato 1993; Benson 1986Chen 1993; EC Trial Group 1994; Erdem 1988Morgan 1981Sanghvi 1999) provided 644 treated infants and 690 control infants. Typical RR 0.96 (95% CI 0.82 to 1.11).

Analysis for entire study population (randomised trials only) (Outcome 1.4.2):

Excluding data from the Erdem 1988 trial provided equal numbers of treated and control infants (n = 544). Typical RR 1.04 (95% CI 0.87 to 1.25).

Infants < 31 completed weeks gestation or < 1500 grams (Outcome 1.4.3):

Morgan 1981; Benson 1986; Erdem 1988 and Chen 1993 contributed 296 ethamsylate-treated infants and 298 control infants. There was no significant difference in the combined outcome of death or disability at two years of postnatal age between these two groups (typical RR 1.04, 95% CI 0.81 to 1.34).

Infants < 31 completed weeks gestation or < 1500 grams (randomised trials only) (Outcome 1.4.4):

Morgan 1981; Benson 1986 and Chen 1993 contributed 277 ethamsylate-treated infants and 268 controls. Typical RR 1.04 (95% CI 0.81 to 1.34).

Cerebral palsy in surviving children available for follow-up - Outcome 1.5:

There were no long-term follow-up data available from Erdem 1988; Amato 1993; Chen 1993Erdem 1988 or Sanghvi 1999. This meant that only randomised controlled trials provided outcome data. Only cerebral palsy significant enough to cause moderate or severe impairment was included in this outcome. Mild changes in tone that did not result in disability were not counted. There was no significant difference in the rates of cerebral palsy between those infants who received ethamsylate and those who did not.

Analysis for entire study population (randomised trials only) (Outcome 1.5.1):

The EC Trial Group 1994 reported neuromotor impairment with moderate or severe disability in 12 of 101 treated infants and 7 of 103 controls. Benson 1986 reported the total number of children with cerebral palsy (16 of 137 treated and 14 of 139 controls for whom outcome was known). However, personal communication from Dr Jane Schulte clarified that only nine treated infants and 11 control infants had cerebral palsy serious enough to count as major impairment. Personal communication from the authors of Morgan 1981 clarified that 2 of 24 survivors who were treated with ethamsylate had spastic diplegia and of the 28 surviving control infants, two had spastic quadriplegia and one had spastic diplegia. In this meta-analysis there were 262 ethamsylate-treated infants and 270 control infants giving a typical RR of 1.13 (95% CI 0.64 to 2.00).

Infants < 31 completed weeks gestation or < 1500 grams (randomised trials only) (Outcome 1.5.2):

When we excluded data from the EC Trial Group 1994, meta-analysis of 161 treated infants and 167 control infants showed a typical RR 0.82 (95% CI 0.38 to 1.75).

Cognitive delay > 2 SD below mean in surviving children available for follow-up - Outcome 1.6:

Only Morgan 1981; Benson 1986 and EC Trial Group 1994 contributed data to this outcome, across two strata. There was no significant difference between the ethamsylate-treated infants and control groups.

Analysis for entire study population (randomised trials only) (Outcome 1.6.1):

There were 262 treated infants and 270 control infants in this analysis. Typical RR 0.71 (95% CI 0.42 to 1.19).

Infants < 31 completed weeks gestation or < 1500 grams (randomised trials only) (Outcome 1.6.2):

There were 161 ethamsylate-treated infants and 167 control infants in this analysis. Typical RR 0.57 (95% CI 0.30 to 1.07).

Blindness (visual acuity < 6/60) in surviving children available for follow-up (Outcome 1.7):

Only Benson 1986 and the EC Trial Group 1994 reported data for this outcome. Personal communication with Richard Cooke for the Morgan 1981 trial confirmed that no subjects from this trial suffered significant visual impairment. There was no significant difference between the ethamsylate treated infants and control groups.

Analysis for entire study population (Outcome 1.7.1):

Data were available for 262 treated infants and 270 controls. Typical RR 2.54 (95% CI 0.50 to 12.98).

Infants < 31 completed weeks gestation or < 1500 grams (Outcome 1.7.2):

Data were available for 161 ethamsylate treated infants and 167 controls. Typical RR 3.04 (95% CI 0.32 to 28.90).

Deafness requiring amplification in surviving children available for follow-up (Outcome 1.8):
Analysis for entire study population (Outcome 1.8.1):

Data were available for 262 treated infants and 270 controls. Personal communication with Richard Cooke provided information on this outcome from the Morgan 1981 trial. Whilst two children from the ethamsylate treated arm of the EC Trial Group 1994 had hearing loss, neither of these required hearing aids, and there was no hearing loss detected amongst the control infants in this study. Results of meta-analysis are typical RR 2.03 (95% CI 0.38 to 10.9)

Infants < 31 completed weeks gestation or < 1500 grams (Outcome 1.8.2):

Data were available for 161 ethamsylate treated infants and 167 controls.Typical RR 2.03 (95% CI 0.38 to 10.9).

Ultrasound evidence of any PIVH detected prior to full term (Outcome 1.9):

Meta-analysis of intraventricular haemorrhage from these studies is subgrouped according to whether or not infants had ultrasonographic evidence of PIVH prior to their enrolment into the study, and thus prior to the administration of ethamsylate or placebo. Subgroups are also stratified by maturity as with previous analyses. In Morgan 1981, 3 of 73 infants were found to have PIVH prior to treatment, one received ethamsylate and had a grade 1 haemorrhage that did not extend; two received placebo and had grade 1 PIVH, one of which extended to grade 3 at 12 hours. In Benson 1986, 21 of 183 treatment infants and 9 of 177 control infants had PIVH detected on initial scan.

Infants < 35 completed weeks or < 2000 grams with pre-existing PIVH (Outcome 1.9.1):

Twenty-two treatment infants and 11 control infants had pre-existing PIVH. No RR was calculable.

Infants < 35 completed weeks or < 2000 grams excluding pre-existing PIVH (Outcome 1.9.2):

Data were available from Morgan 1981; Benson 1986; Amato 1993; Chen 1993; EC Trial Group 1994 and Sanghvi 1999 providing data for meta-analysis on 551 ethamsylate-treated infants and 566 control infants. There was a significant reduction in the incidence of PIVH for infants who received ethamsylate (typical RR 0.77, 95% CI 0.65 to 0.92). The typical RD was -0.08 (95%CI -0.13 to -0.03) meaning that 13 infants < 35 weeks gestation needed to be treated with ethamsylate to prevent one episode of any grade of IVH.

Infants < 31 completed weeks or < 1500 grams with pre-existing IVH (Outcome 1.9.3):

Twenty-twotreatment infants and 11 control infants had pre-existing PIVH. No RR was calculable.

Infants < 31 completed weeks or < 1500 grams excluding pre-existing IVH (Outcome 1.9.4):

There were 197 treatment infants and 203 control infants included in this analysis. There was a significant reduction in the incidence of PIVH for infants who received ethamsylate (typical RR 0.63, 95% CI 0.47 to 0.86). The typical RD was -0.14 (95% CI -0.23 to -0.05).

Ultrasound evidence of grade 3/4 IVH detected prior to full term (Outcome 1.10):
Infants < 35 completed weeks or < 2000 grams with pre-existing IVH (Outcome 1.10.1):

Six of 22 ethamsylate treated and 5 of 11 control infants with pre-existing IVH had grade 3/4 IVH detected. In Morgan 1981, one of two control infants extended an IVH from grade 1 to grade 3 at 12 hours. In Benson 1986, 15 infants (10 treatment and 5 controls) had grade 1 IVH on initial scan. Six of ten treatment infants and 3 of 5 control infants extended their PIVH to a higher grade on subsequent scan. With a total of 22 ethamsylate treated infants and 11 control infants, there was no significant difference between treatment and control infants in the incidence of grade 3/4 IVH (typical RR 0.62, 95% CI 0.24 to 1.58).

Infants < 35 completed weeks or < 2000 grams excluding pre-existing IVH (Outcome 1.10.2):

Data from Morgan 1981; Benson 1986; Amato 1993; Chen 1993; EC Trial Group 1994 and Sanghvi 1999 was included in this analysis. A total of 551 ethamsylate treated infants and 566 control infants were analysed. There was a significant reduction in the incidence of grade 3/4 IVH amongst infants < 35 weeks gestation who were treated with ethamsylate compared to those treated with placebo (typical RR 0.68, 95% CI 0.48 to 0.97). The typical RD was -0.04 (95% CI -0.07 to 0.00).

When the results of those babies < 35 weeks gestation with pre-existing IVH were combined with those without pre-existing IVH, there was a total of 573 treated infants and 577 controls. There was a significant reduction in grade 3 or 4 IVH in ethamsylate treated infants compared to controls (typical RR 0.67 95% CI 0.49 to 0.94). The typical RD was -0.04 (95% CI -0.08 to -0.01) meaning that 25 infants < 35 weeks gestation would need to be treated to prevent one grade 3 or 4 IVH.

Infants < 31 completed weeks or < 1500 grams with pre-existing IVH (Outcome 1.10.3):

With a total of 22 ethamsylate treated infants and 11 control infants, there was no significant difference between treatment and control infants in the incidence of grade 3/4 IVH (typical RR 0.62, 95% CI 0.24 to 1.58).

Infants < 31 completed weeks or < 1500 grams excluding pre-existing IVH (Outcome 1.10.4):

A total of 197 treated infants and 203 control infants contributed to this meta-analysis. There was no significant difference between treatment and control infants in the incidence of grade 3/4 IVH (typical RR 0.58, 95% CI 0.30 to 1.11).

Symptomatic PDA (Outcome 1.11):
Analysis for entire study population (Outcome 1.11.1):

Data from the Benson 1986; Amato 1993; EC Trial Group 1994 and Sanghvi 1999 trials contributed to this analysis, giving a total of 451 treated infants and 455 controls. Only 2 of 10 treatment infants compared to 8 of 10 controls in Amato 1993 developed symptomatic PDA. After meta-analysis there was no significant difference between the two groups (typical RR 0.80, 95% CI 0.57 to 1.12).

Infants < 31 completed weeks gestation or < 1500 grams (Outcome 1.11.2):

Only data from the Benson 1986 trial were included here, providing 183 treated infants and 177 controls. The ethamsylate treated infants < 31 weeks gestation had significantly less symptomatic PDA than controls (typical RR 0.32, 95% CI 0.12 to 0.87). The typical RD was -0.06 (95% CI -0.10 to -0.01) meaning that 17 infants < 31 weeks need to be treated with ethamsylate to prevent one case of symptomatic PDA.

Duration of respiratory support

No data were available for analysis.

Hypotension (Outcome 1.12):

Only Benson 1986 and Sanghvi 1999 provided data relating to hypotension. Sanghvi 1999 reported that hypotension requiring inotropic support could not be attributed to use of ethamsylate, however, the justification for this was not cited, so we included the data here as a potential adverse side-effect.

Analysis for entire study population (Outcome 1.12.1):

Data from a total of 276 ethamsylate treated infants and 276 controls were available. There was no significant difference in the incidence of hypotension between the two groups (typical RR 0.95, 95% CI 0.56 to 1.63).

Infants < 31 completed weeks gestation or < 1500 grams (Outcome 1.12.2):

Only Benson 1986 contributed data here, with 183 ethamsylate treated infants and 177 controls. There was no significant difference in the incidence of hypotension between the two groups (typical RR 1.02, 95% CI 0.55 to 1.92).

Discussion

Ethamsylate has been studied in a number of randomised controlled trials to determine its role in neuroprotection for the preterm infant. Based on its mechanism of action and perceived benefit in other bleeding disorders (e.g. menorrhagia) it was postulated that ethamsylate may lead to a reduction in IVH. The results of this meta-analysis demonstrate that ethamsylate does indeed lead to a reduction in the incidence of any grade of IVH by 37% in infants < 31 weeks gestation or < 1500 grams (typical RR 0.63, 95% CI 0.47 to 0.86), by 22% in infants < 32 weeks gestation or < 1750 grams (typical RR 0.78, 95% CI 0.63 to 0.97) and by 23% in infants < 35 weeks gestation or < 2000 grams (typical RR 0.77, 95% CI 0.65 to 0.92). When high grade IVH is considered, there is a trend to reduction in the incidence of grade 3/4 IVH at < 31 and < 32 weeks gestation, and a statistically significant reduction of 33% in grade 3/4 IVH in infants < 35 weeks gestation (typical RR 0.67, 95% CI 0.49 to 0.94).

However, there are a number of problems with this finding. The first relates to the definitions of IVH used by the different studies. Benson 1986 used a modification of the classification first described by Papile 1978, simply separating intraventricular from parenchymal haemorrhage, and making it impossible to differentiate grade 2 IVH (without ventricular dilatation) from haemorrhage associated with ventricular dilatation. Morgan 1981 on the other hand used the grading system defined by Papile 1978 and the EC Trial Group 1994 used a different system again, recording information about IVH, ventricular dilatation and cystic white matter injury.

Availability and timing of ultrasound scans were also highly variable between studies making it very difficult to interpret results pertaining to the presence of IVH prior to trial entry (the EC Trial Group 1994 did not perform scans early enough to allow detection of pre-existing haemorrhage).

The primary studies reviewed here were performed as early as the late 1970s and as late as the early 1990s. These studies were, therefore, performed at times when standard perinatal practice would have varied widely. In the trial by Erdem 1988, routine cranial ultrasound was not available in clinical practice. This clearly has implications for the detection and management of IVH. Our finding of a significant reduction in the incidence of grade 3/4 IVH among infants born less than 35 weeks gestation is of little relevance when, in the scope of current practice, IVH is a rare event in infants > 30 weeks gestation, and routine screening ultrasound beyond 30 weeks is no longer recommended Harding 1998.

The strength of this review comes from the high quality of the long-term neurodevelopmental follow-up studies that relate to the larger randomised controlled trials (Morgan 1981; Benson 1986; EC Trial Group 1994). The investigators in these studies must be congratulated for persevering not only in obtaining the information from a high percentage of their cohort, but also for publishing this data despite the lapse of several years. Most notably, the use of ethamsylate at any gestation does not appear to confer any benefit on mortality (either neonatal or to hospital discharge) or on any component of neurodevelopment, including cognition, neuromotor performance (measured by evidence of cerebral palsy) or major sensory impairment. Any effect that ethamsylate may be observed to have on IVH is, therefore, really secondary and largely irrelevant if such effects do not translate into improvement in clinically meaningful longer-term outcomes. This finding is not unique. The systematic review of prophylactic indomethacin in preterm infants by Fowlie and Davis found a significant reduction in grade 3/4 IVH that was not coupled with any long-term neurodevelopmental benefit (Fowlie 2002).

It is also of interest that we found a significant reduction in the incidence of symptomatic PDA (68%, typical RR 0.32, 95% CI 0.12 to 0.87) among infants < 31 weeks gestation treated with ethamsylate. The potential role of the patent ductus on cerebral haemodynamics and subsequent cerebral injury makes this another interesting finding.However, without concomitant reduction in mortality or neurodevelopmental impairment it is of little consequence.

Our understanding of perinatal brain injury in preterm infants is evolving. With advances in neuroimaging and improved understanding of the pathogenesis of IVH and posthaemorrhagic ventricular dilatation, it is becoming increasingly apparent that IVH on its own is not the sole substrate of cerebral injury and subsequent neurodevelopmental impairment. Cranial ultrasound does not perform well in the detection of non-cystic white matter injury. The role of environment on influencing cerebral plasticity and mediating long-term outcome is still poorly understood. Despite the appeal of measuring IVH with cranial ultrasound as a short term outcome of clinical studies, it must be realised that clinical research must focus on clinically meaningful longer-term outcomes, and this will require funding bodies to appreciate the need for longer-term interest.

The only adverse side-effect of ethamsylate noted in previous clinical studies has been hypotension, and we could find no significant difference in the rate of hypotension between those infants treated with ethamsylate and those treated with placebo. Therefore, it would seem from this systematic review that, if nothing else, ethamsylate is safe when given to premature infants.

Authors' conclusions

Implications for practice

There is no evidence from randomised trials to support the routine use of ethamsylate in preterm infants as a therapy for improving mortality or neurodevelopmental outcome.

Implications for research

Ethamsylate was both studied and used clinically as recently as the mid-1990s. The randomised controlled trials investigating the use of ethamsylate have enrolled over 700 infants and have failed to find any clinically meaningful benefit in terms of mortality or neurodevelopmental outcome. Further studies of ethamsylate in the clinical setting of prematurity would be difficult to justify based on the results of this systematic review.

Acknowledgements

We are grateful to the authors of Morgan 1981; Benson 1986; EC Trial Group 1994Benson 1986 and Sanghvi 1999 for providing clarification of the methodology and results of their trials by correspondence during the preparation of this review. We would certainly include more data from the other trials in an update of this review if authors of Erdem 1988; Amato 1993 and Chen 1993Amato 1993were able to provide information regarding birth weight and gestation from the babies in their trials.

The Cochrane Neonatal Review Group has been funded in part with Federal funds from the Eunice Kennedy Shriver National Institute of Child Health and Human Development National Institutes of Health, Department of Health and Human Services, USA, under Contract No. HHSN267200603418C.

Contributions of authors

RH provided the first draft of the manuscript and made subsequent revisions.
EH provided advice on subsequent revisions.
Both review authors extracted the data for meta-analysis.

Declarations of interest

  • None noted.

Differences between protocol and review

Review has incorporated results of the randomised trials located in the search.

Additional tables

  • None noted.

Potential conflict of interest

  • None noted.

[top]

Characteristics of studies

Characteristics of Included Studies

Amato 1993

Methods

Randomised controlled trial - included post-facto because birth weight entry criteria do not meet initial protocol.

Participants

Recruited between 1989 and 1990. 20 preterm infants with severe RDS and symptomatic PDA with birth weight between 700 and 2000 grams. Babies were excluded if there was prolonged rupture of the membranes (> 3 weeks), severe intraventricular haemorrhage (grade 3 or 4) on serial ultrasound scanning within 15 hours of birth, perinatal asphyxia or major congenital anomalies.

Interventions

Ethamsylate (125 mg/ml in 0.1 ml/kg/dose, n = 10) or saline placebo (0.1 ml/kg, n = 10) given i.v. First dose given within 4 hours and then six hourly for four days to total dose of 200 mg/kg.

Outcomes

Primary outcome was prevention of symptomatic PDA. This was not defined by need for treatment but rather by clinical signs and echocardiographic findings.

Notes

Authors report exclusion of infants with grade 3/4 IVH on serial scans, and yet report grade 3/4 IVH as an outcome in Table 2.

Risk of bias table
Item Judgement Description
Adequate sequence generation? Unclear

Randomised controlled trial - allocation according to what was written on a card inside a sealed envelope.

Allocation concealment? Unclear

The control children received saline rather than ethamsylate from ampoules of the same size; it is not clear if investigators or treating staff knew who got what.

Blinding? Unclear

It is not clear whether the ultrasonologist responsible for assessing ductal closure and the incidence of IVH was aware of treatment allocation.

Incomplete outcome data addressed? Yes

Information on a range of secondary outcomes was collected prospectively and seems to be complete for all the 20 children in the trial.

Free of selective reporting? Yes
Free of other bias? Yes

Benson 1986

Methods

Multicentre randomised controlled trial. Manufacturer provided ampoules labelled with numbers from block randomisation tables.

Participants

Recruited between 1983 and 1986 and recruited 366 infants with birth weight less than/or equal to 1500 grams, excluding infants with lethal congenital malformation.

Interventions

Ethamsylate (12.5 mg/kg = 0.1 ml/kg, n = 162) or placebo (normal saline, 0.1 ml/kg, n = 168) given i.v. or i.m. within an hour of delivery, then six hourly i.v. for four days, to total dose of 200 mg/kg.

Outcomes

Primary outcome not stated but implied to be periventricular haemorrhage detected by serial ultrasound scan over the first two weeks of life. Neurodevelopmental outcome at 3.5 to 4.2 years of age, measured by McCarthy Scales, physical examination, vision and hearing testing.

Notes

Block randomisation tables employed to label treatment packs - ethamsylate and placebo visually indistinguishable.

Risk of bias table
Item Judgement Description
Adequate sequence generation? Yes

Multicentre randomised controlled trial. The manufacturer provided indistinguishable ampoules from a block randomisation table.

Allocation concealment? Yes

Adequate. Treatment vials were said to be indistinguishable.

Blinding? Yes

The treatment allocation was not known to anyone until all of the predischarge outcomes had been reported, and the outcome at three to four years was undertaken by a single assessor uninvolved in providing early care.

Incomplete outcome data addressed? Yes

Six children were excluded after recruitment for clearly specified reasons. Outcome to discharge and long-term outcome is known for every other child with 97% (260/268) of all the long term survivors assessed at 3.5 to 4.2 years, but by a single clinician unaware of the early trial allocation.

Free of selective reporting? Yes

Probably. The outcomes were not publicly pre-specified, but no outcome of obvious importance is missing.

Free of other bias? Yes

As far as can be assessed.

Chen 1993

Methods

Included post-facto. Subjects "randomly divided into two groups" - method of randomisation not stated.

Participants

Recruited between 1990 and 1992 and recruited 171 preterm infants with birth weight < 1751 g and with no IVH on initial ultrasound scan performed within an hour of delivery.

Interventions

Ethamsylate 12.5 mg/kg (= 0.1 ml/kg, n = 86) given i.v. during first hour of life followed by six-hourly doses for four days to total of 200 mg/kg. Control infants received 0.1 ml/kg (n = 85) normal saline i.v. at same time interval.

Outcomes

Primary outcome stated to be incidence and severity of IVH. Duration of mechanical ventilation or CPAP, and incidence of pneumothorax, birth asphyxia, sepsis and mortality to hospital discharge also reported.

Notes
Risk of bias table
Item Judgement Description
Adequate sequence generation? Unclear

Subjects "randomly divided into two groups" - method of randomisation not stated.

Allocation concealment? Unclear

Control children received saline rather than ethamsylate, but it is not clear how many people knew who got what.

Blinding? Yes

The ultrasonologist did not know which treatment group any child was in.

Incomplete outcome data addressed? Yes

IVH and its severity and predischarge death would seem to have been the only two planned trial outcomes.

Free of selective reporting? Yes

Outcome data is provided for all of the 171 babies enrolled.

Free of other bias? Yes

EC Trial Group 1994

Methods

Included post-facto because did not meet initial inclusion criteria regarding gestational age. Randomisation codes held in national blood bank. Group allocation advised by telephone call to randomisation centre.

Participants

Recruited between 1990 and 1991 and recruited 334 infants < 32 completed weeks gestation, randomised within 4 hours of birth, without known major malformations, if treating physician in equipoise regarding utility of ethamsylate.

Interventions

Ethamsylate (n = 165) vs no treatment (n = 169). Ethamsylate group received dose of 12.5 mg/kg i.v. or i.m., followed by 15 doses i.v. given every six hours for four days.

Outcomes

Primary outcomes were death, disability at 2 years of age, and use of healthcare resources.

Notes

Block randomisation stratified by centre. No blinding of treatment. Blinded outcome assessment in all but one centre.

Risk of bias table
Item Judgement Description
Adequate sequence generation? Yes

Randomisation codes held in national blood bank. Group allocation advised by telephone call to randomisation centre.

Allocation concealment? No

Treatment allocation could have been known to those responsible for neonatal care because the controls did not receive any 'placebo' injections.

Blinding? Yes

Mostly yes. The main early outcomes were objectively defined so blinding is probably not critical. The nature of early treatment was known to some of those doing the two year assessment in Greece, but not to any of those doing the two year assessment in France (where 181 of 334 babies were recruited).

Incomplete outcome data addressed? Unclear

Partially. Outcome data seems complete for all major events occurring in the first three months of life. 21% (59/279) of the neonatal survivors had been lost to follow up by two years. There is no reason to suppose that loss would have occurred differently in the two trial groups.

Free of selective reporting? Yes

All of the important predefined trial outcomes have been reported.

Free of other bias? Yes

Erdem 1988

Methods

Included post-facto. Quasi-randomised trial. Eligible infants assigned a trial number consecutively and either given ethamsylate or nothing.

Participants

Recruited between 1986 and 1987 and recruited 246 preterm infants with gestational age < 34 weeks, excluding infants with lethal congenital malformations.

Interventions

Treatment group (n = 146) received ethamsylate 12.5 mg/kg i.v. within an hour of birth, and thereafter given same dose six hourly for four days to total dose of 200 mg/kg. Control group n = 100.

Outcomes

Neonatal mortality was reported, stratified according to gestational age.

Notes

Despite implied alternate randomisation, study group ended up with 146 subjects and no treatment group ended up with only 100. Mortality for babies < 31 weeks gestation was much higher in this study than for other included studies, with 0/49 babies surviving.

Risk of bias table
Item Judgement Description
Adequate sequence generation? No

Quasi-randomised trial. Eligible infants assigned a trial number consecutively and either given ethamsylate or nothing. There is no explanation for why the study ended up with 146 treated babies and only 100 control babies.

Allocation concealment? No

Inadequate - treatment allocation could have been known to those responsible for neonatal care because controls did not receive any 'placebo' injections.

Blinding? Unclear

Given that early and late neonatal death is the only outcome reported for this trial, the lack of blinding is probably not critical.

Incomplete outcome data addressed? Yes
Free of selective reporting? Yes
Free of other bias? Yes

Morgan 1981

Methods

Randomised controlled trial.

Participants

Recruited subjects between 1980 and 1982 and included 70 infants with birth weight < 1500 grams plus a further three infants who already had an IVH when scanned at trial entry, excluding infants with lethal congenital malformation. Three babies with an IVH at entry into the trial were later excluded.

Interventions

Ethamsylate (n = 35) or placebo (n = 35). 12.5 mg/kg of ethamsylate or 0.1 ml/kg placebo i.m. within 2 hours of birth and thereafter six hourly for four days.

Outcomes

Primary outcome not stated. Implied outcome is development of IVH detected by ultrasound examination performed daily for first week of life.

Notes

Randomisation code held by manufacturer. Volume of drug or placebo identical. Nothing reported regarding ability to distinguish drug from placebo. Blinding uncertain.

Risk of bias table
Item Judgement Description
Adequate sequence generation? Yes

Randomised controlled trial. Ampoules provided from a block randomisation table.

Allocation concealment? Yes

Adequate. The vials were later reported to have been indistinguishable.

Blinding? Yes

Ultrasonologist was unaware of the treatment allocation.

Incomplete outcome data addressed? Yes

The outcome is known for the only two outcomes being looked at (the incidence and grade of IVH and death before discharge). Information on outcome at three years was later made available to these reviewers for all of the 54 babies who survived to discharge.

Free of selective reporting? Yes

Although three babies were later excluded from the main analysis because they were found to have an IVH at randomisation, their outcomes were fully reported.

Free of other bias? Yes

As far as can be addressed.

Sanghvi 1999

Methods

Included post-facto. Randomised controlled trial with randomisation using opaque envelopes.

Participants

Recruited between 1996 and 1998 and originally included 200 babies, however, 8 were later excluded because they had died within 48 hours of birth. Infants with gestational age < 34 weeks, excluding infants with congenital malformations, family history of bleeding disorders and with Apgar scores < 5 at 5 minutes. Birth weights of participants ranged from 760 to 1749 grams.

Interventions

Study subjects (n = 93) received ethamsylate 12.5 mg/kg i.v. within an hour of birth and thereafter six hourly for four days. Control subjects (n = 99) received nothing other than routine therapy.

Outcomes

Primary outcome was incidence and severity of IVH. Other outcomes reported included need for mechanical ventilation, pneumothorax, PDA, sepsis and mortality to hospital discharge.

Notes

Four infants with congenital heart disease, two with genito-urinary tract anomalies and two with intestinal obstruction were excluded after randomisation. Eight infants died before completing the drug schedule, five due to congenital sepsis and three due to respiratory distress syndrome. The group allocation of these exclusions is unknown.

Risk of bias table
Item Judgement Description
Adequate sequence generation? Yes

Randomised controlled trial with randomisation using opaque envelopes.

Allocation concealment? Yes

Adequate.

Blinding? Unclear

Partially. Ultrasound assessment was by a paediatric sinologist "who was blinded to the patient group". Most other reported outcomes were fairly objectively defined, and biased reporting is unlikely.

Incomplete outcome data addressed? No

208 babies were recruited into this trial, but the outcome for the four later found to have a congenital malformation and the eight who died before treatment with ethamsylate or placebo was completed, has not been provided.

Free of selective reporting? Yes

IVH and its severity and predischarge death would seem to have been the only two planned trial outcomes, but information has been provided for several other important predischarge outcomes.

Free of other bias? Yes

As far as can be assessed.

Characteristics of excluded studies

Cooke 1984

Reason for exclusion

Began as double blind, randomised, placebo controlled trial. However, then infants receiving no treatment or open label ethamsylate were added into analysis, based on evolving local practice. This report details 2 year outcome for the Morgan trial (Morgan 1981), but the data cannot be used because they have been combined with information from other non-randomised children.

Rennie 1986

Reason for exclusion

Blinded randomisation of infants < 1500 grams at birth. However, primary outcome was 6-ketoprostaglandin F1 alpha level, so data from 31/64 infants were excluded from analysis.
This was a study conducted on a sub-sample of the babies included into the Benson trial (Benson 1986).

Characteristics of studies awaiting classification

Liu 1997

Methods

(Information will be added should this trial report be located)

Participants
Interventions
Outcomes
Notes

[top]

References to studies

Included studies

Amato 1993

Amato M, Huppi P, Markus D. Prevention of symptomatic patent ductus arteriosus with ethamsylate in babies treated with exogenous surfactant. Journal of Perinatology 1993;13:2-7.

* Amato M, Huppi PS, Markus D. Prophylaxis of patent ductus arteriosus using ethamsylate in preterms treated with exogenous surfactant. Acta Paediatrica 1992;81:351-2.

Benson 1986

* Benson JW, Drayton MR, Hayward C, Murphy JF, Osborne JP, Rennie JM et al. Multicentre trial of ethamsylate for prevention of periventricular haemorrhage in very low birthweight infants. Lancet 1986;2:1297-1300.

Cooke RW. Effect of ethamsylate on cerebral blood flow. Lancet 1987;1:623-4.

Rennie JM, Doyle J, Cooke RW. Ethamsylate reduces immunoreactive prostacyclin metabolite in low birthweight infants with respiratory distress syndrome. Early Human Development 1986;14:239-44.

Schulte J, Osborne J, Benson JWT, Cooke R, Drayton M, Murphy J et al. Developmental outcome of the use of ethamsylate for prevention of periventricular haemorrhage in a randomised controlled trial. Archives of Disease in Childhood. Fetal and Neonatal Edition 2005;90:F31-5.

Chen 1993

Chen JY. Ethamsylate in the prevention of periventricular-intraventricular hemorrhage in premature infants. Journal of the Formosan Medical Association 1993;92:889-93.

EC Trial Group 1994

* EC Ethamsylate Trial Group. The EC randomised controlled trial of prophylactic ethamsylate for very preterm neonates: early mortality and morbidity. Archives of Disease in Childhood. Fetal and Neonatal Edition 1994;70:F201-5.

Elbourne D, Ayers S, Dellagrammaticas H, Johnston A, Leloup M, Lenoir-Piat S; EC Ethamsylate Trial Group. Randomised controlled trial of prophylactic etamsylate: follow up at 2 years of age. Archives of Disease in Childhood. Fetal and Neonatal Edition 2001;84:F183-7.

Erdem 1988

Erdem G, Yurdakok M, Tekinalp G, Hicsonmez G. The effect of ethamsylate on neonatal mortality in preterm infants. Turkish Journal of Pediatrics 1988;30:93-7.

Morgan 1981

Published and unpublished data

Cooke RW, Morgan ME. Prophylactic ethamsylate for periventricular haemorrhage. Archives of Disease in Childhood 1984;59:82-3.

* Morgan ME, Benson JW, Cooke RW. Ethamsylate reduces the incidence of periventricular haemorrhage in very low birth-weight babies. Lancet 1981;2:830-1.

Sanghvi 1999

Sanghvi KP, Merchant RH, Karnik A, Kulkarni A. Role of ethamsylate in preventing periventricular-intraventricular hemorrhage in premature infants below 34 weeks of gestation. Indian Pediatrics 1999;36:653-8.

Excluded studies

Cooke 1984

Cooke RW, Morgan ME. Prophylactic ethamsylate for periventricular haemorrhage. Archives of Disease in Childhood 1984;59:82-3.

Rennie 1986

Rennie JM, Doyle J, Cooke RW. Ethamsyate reduces immunoreactive prostacyclin metabolite in low birthweight infants with respiratory distress syndrome. Early Human Development 1986;14:239-44.

Studies awaiting classification

Liu 1997

Liu HC, He HL, He ZH. Randomised and controlled study on the preventive effects of ethamsylate on intracranial hemorrhage in newborn infants with asphyxia. Chinese Journal of Pharmacoepidemiology 1997;6:11-3.

Ongoing studies

Other references

Additional references

Amato 1992

Amato M, Huppi P, Markus D. Prevention of symptomatic patent ductus arteriosus with ethamsylate in babies treated with exogenous surfactant. Journal of Perinatology 1993;13:2-7.

Amato 1994

Amato M, Sun B, Robertson B. Ethamsylate and lung permeability in ventilated immature newborn rabbits. Biology of the Neonate 1994;65:103-7.

Elbourne 2001

Elbourne D, Ayers S, Dellagrammaticas H, Johnston A, Leloup M, Lenoir-Piat S; EC Ethamsylate Trial Group. Randomised controlled trial of prophylactic etamsylate: follow up at 2 years of age. Archives of Disease in Childhood. Fetal and Neonatal Edition 2001;84:F183-7.

Fowlie 2002

Fowlie PW, Davis PG. Prophylactic intravenous indomethacin for preventing mortality and morbidity in preterm infants. Cochrane Database of Systematic Reviews 2002, Issue 3. Art. No.: CD000174. DOI: 10.1002/14651858.CD000174 .

Goddard-Finegold 1982

Goddard-Finegold J, Armstrong D, Zeller RS. Intraventricular hemorrhage, following volume expansion after hypovolemic hypotension in the newborn beagle. Journal of Pediatrics 1982;100:796-9.

Harding 1998

Harding D, Kuschel C, Evans N. Should preterm infants born after 29 weeks' gestation be screened for intraventricular haemorrhage? Journal of Paediatrics and Child Health 1998;34:57-9.

Harrison 1976

Harrison RF, Campbell S. A double-blind trial of ethamsylate in the treatment of primary and intrauterine-induced menorrhagia. Lancet 1976;2:283-5.

Kowacs 1981

Kowacs L, Falkay G. Etamsylate as inhibitor of prostaglandin biosynthesis in pregnant human myometrium in vitro. Experientia 1981;37:1182-3.

Langdon 1977

Langdon L. Transient hypotension following intravenous ethamsylate (Dicynene). British Medical Journal 1977;1:1472.

Levene 1981

Levene MI. Measurement of the growth of the lateral ventricles in preterm infants with real-time ultrasound. Archives of Disease in Childhood 1981;56:900-4.

Okuma 1982

Okuma M, Takayama H, Sugiyama T, Sensaki S, Uchino H. Effects of etamsylate on platelet functions and arachidonic acid metabolism. Thrombosis and Haemostasis 1982;48:330-3.

Papile 1978

Papile LA, Burstein J, Burstein R, Koffler H. Incidence and evolution of subependymal and intraventricular hemorrhage: a study of infants with birthweight less than 1500 gm. Journal of Pediatrics 1978;92:529-34.

Rennie 1989

Rennie JM, Lam PK. Effects of ethamsylate on cerebral blood flow velocity in premature babies. Archives of Disease in Childhood 1989;64:46-7.

Rosti 1994

Rosti L, Piva D, Rosti D. Ethamsylate in the prevention of patent ductus arteriosus. Archives of Pediatrics and Adolescent Medicine 1994;148:1103-4.

Schulte 2005

Schulte J, Osborne J, Benson JWT, Cooke R, Drayton M, Murphy J et al. Developmental outcome of the use of ethamsylate for prevention of periventricular haemorrhage in a randomised controlled trial. Archives of Disease in Childhood. Fetal and Neonatal Edition 2005;90:F31-5.

Symes 1975

Symes JM. The effect of dicynene on blood loss during and after transurethral resection of the prostate. British Journal of Urology 1975;47:203.

Vinazzer 1980

Vinazzer H. Clinical and experimental studies on the action of ethamsylate on haemostasis and on platelet functions. Thrombosis Research 1980;19:783-91.

Volpe 2002

Volpe JJ. Neurology of the Newborn. Philadelphia: W.B. Saunders Company, 2002.

Watson 1977

Watson B. Transient hypotension following intravenous ethamsylate (dicynene). British Medical Journal 1977;1:1664.

Other published versions of this review

  • None noted.

Classification pending references

  • None noted.

[top]

Data and analyses

1 Ethamsylate versus placebo

Outcome or Subgroup Studies Participants Statistical Method Effect Estimate
1.1 Neonatal mortality (to 28 days postnatal age) 5 Risk Ratio (M-H, Fixed, 95% CI) Subtotals only
1.1.1 Infants < 35 completed weeks or < 2000 grams 5 1202 Risk Ratio (M-H, Fixed, 95% CI) 1.01 [0.83, 1.23]
1.1.2 Infants < 35 completed weeks or < 2000 g - randomised trials only 4 956 Risk Ratio (M-H, Fixed, 95% CI) 1.23 [0.92, 1.64]
1.1.3 Infants < 31 completed weeks or < 1500 grams 3 479 Risk Ratio (M-H, Fixed, 95% CI) 1.28 [0.86, 1.90]
1.1.4 Infants < 31 completed weeks or < 1500 g - randomised trials only 2 430 Risk Ratio (M-H, Fixed, 95% CI) 1.28 [0.86, 1.90]
1.2 Mortality to hospital discharge 7 Risk Ratio (M-H, Fixed, 95% CI) Subtotals only
1.2.1 Infants < 35 completed weeks or < 2000 grams 7 1393 Risk Ratio (M-H, Fixed, 95% CI) 0.97 [0.80, 1.17]
1.2.2 Infants < 35 completed weeks or < 2000 grams - randomised trials only 6 1147 Risk Ratio (M-H, Fixed, 95% CI) 1.11 [0.86, 1.44]
1.2.3 Infants < 31 completed weeks or < 1500 grams 4 594 Risk Ratio (M-H, Fixed, 95% CI) 1.19 [0.83, 1.71]
1.2.4 Infants < 31 completed weeks of < 1500 grams - randomised trials only 3 545 Risk Ratio (M-H, Fixed, 95% CI) 1.19 [0.83, 1.71]
1.3 Neurodevelopmental disability at 2 years of age in surviving children available for follow-up 3 Risk Ratio (M-H, Fixed, 95% CI) Subtotals only
1.3.1 Infants < 35 completed weeks or < 2000 grams - randomised trials only 3 532 Risk Ratio (M-H, Fixed, 95% CI) 0.79 [0.53, 1.17]
1.3.2 Infants < 31 completed weeks or < 1500 grams 2 328 Risk Ratio (M-H, Fixed, 95% CI) 0.57 [0.30, 1.07]
1.3.3 Infants < 31 completed weeks of < 1500 grams - randomised trials only 2 328 Risk Ratio (M-H, Fixed, 95% CI) 0.57 [0.30, 1.07]
1.4 Death or any disability by 2 years of age in children with known outcome at any point in time 7 Risk Ratio (M-H, Fixed, 95% CI) Subtotals only
1.4.1 Infants < 35 completed weeks or < 2000 grams 7 1334 Risk Ratio (M-H, Fixed, 95% CI) 0.96 [0.82, 1.11]
1.4.2 Infants < 35 completed weeks or < 2000 grams - randomised trials only 6 1088 Risk Ratio (M-H, Fixed, 95% CI) 1.04 [0.87, 1.25]
1.4.3 Infants < 31 completed weeks or < 1500 grams 4 594 Risk Ratio (M-H, Fixed, 95% CI) 1.04 [0.81, 1.34]
1.4.4 Infants < 31 completed weeks of < 1500 grams - randomised trials only 3 545 Risk Ratio (M-H, Fixed, 95% CI) 1.04 [0.81, 1.34]
1.5 Cerebral palsy in surviving children available for follow-up 3 Risk Ratio (M-H, Fixed, 95% CI) Subtotals only
1.5.1 Infants < 35 completed weeks or < 2000 grams 3 532 Risk Ratio (M-H, Fixed, 95% CI) 1.13 [0.64, 2.00]
1.5.2 Infants < 31 completed weeks or < 1500 grams 2 328 Risk Ratio (M-H, Fixed, 95% CI) 0.82 [0.38, 1.75]
1.6 Cognitive delay > 2SD below mean in surviving children available for follow-up 3 Risk Ratio (M-H, Fixed, 95% CI) Subtotals only
1.6.1 Infants < 35 completed weeks or < 2000 grams 3 532 Risk Ratio (M-H, Fixed, 95% CI) 0.71 [0.42, 1.19]
1.6.2 Infants < 31 completed weeks or < 1500 grams 2 328 Risk Ratio (M-H, Fixed, 95% CI) 0.57 [0.30, 1.07]
1.7 Blindness (VA < 6/60) in surviving children available for follow-up 3 Risk Ratio (M-H, Fixed, 95% CI) Subtotals only
1.7.1 Infants < 35 completed weeks or < 2000 grams 3 532 Risk Ratio (M-H, Fixed, 95% CI) 2.54 [0.50, 12.98]
1.7.2 Infants < 31 completed weeks or < 1500 grams 2 328 Risk Ratio (M-H, Fixed, 95% CI) 3.04 [0.32, 28.90]
1.8 Deafness requiring amplification in surviving children available for follow-up 3 Risk Ratio (M-H, Fixed, 95% CI) Subtotals only
1.8.1 Infants < 35 completed weeks or < 2000 grams 3 532 Risk Ratio (M-H, Fixed, 95% CI) 2.03 [0.38, 10.90]
1.8.2 Infants < 31 completed weeks or < 1500 grams 2 328 Risk Ratio (M-H, Fixed, 95% CI) 2.03 [0.38, 10.90]
1.9 Ultrasound evidence of any IVH detected prior to full term 6 1583 Risk Ratio (M-H, Fixed, 95% CI) 0.73 [0.63, 0.85]
1.9.1 Infants < 35 completed weeks or < 2000 grams - for infants with pre-existing IVH 2 33 Risk Ratio (M-H, Fixed, 95% CI) Not estimable
1.9.2 Infants < 35 completed weeks or < 2000 grams excluding pre-existing IVH 6 1117 Risk Ratio (M-H, Fixed, 95% CI) 0.77 [0.65, 0.92]
1.9.3 Infants < 31 completed weeks or < 1500 grams - for infants with pre-existing IVH 2 33 Risk Ratio (M-H, Fixed, 95% CI) Not estimable
1.9.4 Infants < 31 completed weeks or < 1500 grams excluding pre-existing IVH 2 400 Risk Ratio (M-H, Fixed, 95% CI) 0.63 [0.47, 0.86]
1.10 Ultrasound evidence of Gr 3/4 IVH detected prior to full term 6 1583 Risk Ratio (M-H, Fixed, 95% CI) 0.65 [0.49, 0.86]
1.10.1 Infants < 35 completed weeks or < 2000 grams - for infants with pre-existing IVH 2 33 Risk Ratio (M-H, Fixed, 95% CI) 0.62 [0.24, 1.58]
1.10.2 Infants < 35 completed weeks or < 2000 grams excluding pre-existing IVH 6 1117 Risk Ratio (M-H, Fixed, 95% CI) 0.68 [0.48, 0.97]
1.10.3 Infants < 31 completed weeks or < 1500 grams - for infants with pre-existing IVH 2 33 Risk Ratio (M-H, Fixed, 95% CI) 0.62 [0.24, 1.58]
1.10.4 Infants < 31 completed weeks or < 1500 grams excluding pre-existing IVH 2 400 Risk Ratio (M-H, Fixed, 95% CI) 0.58 [0.30, 1.11]
1.11 Symptomatic PDA 4 Risk Ratio (M-H, Fixed, 95% CI) Subtotals only
1.11.1 Infants < 35 completed weeks or < 2000 grams 4 906 Risk Ratio (M-H, Fixed, 95% CI) 0.80 [0.57, 1.12]
1.11.2 Infants < 31 completed weeks or < 1500 grams 1 360 Risk Ratio (M-H, Fixed, 95% CI) 0.32 [0.12, 0.87]
1.12 Adverse events - hypotension 2 Risk Ratio (M-H, Fixed, 95% CI) Subtotals only
1.12.1 Infants < 35 completed weeks and < 2000 grams 2 552 Risk Ratio (M-H, Fixed, 95% CI) 0.95 [0.56, 1.63]
1.12.2 Infants < 31 completed weeks and < 1500 grams 1 360 Risk Ratio (M-H, Fixed, 95% CI) 1.02 [0.55, 1.92]

[top]

Sources of support

Internal sources

  • Murdoch Children's Research Institute, Melbourne, Australia

External sources

  • NHMRC Grant ID 216757, Australia

This review is published as a Cochrane review in The Cochrane Library, Issue 1, 2010 (see http://www.thecochranelibrary.com External Web Site Policy for information). Cochrane reviews are regularly updated as new evidence emerges and in response to feedback. The Cochrane Library should be consulted for the most recent version of the review.