No new trials were found. The reviewers' conclusions regarding the need for further trials has been reworded.
Mechanical ventilation with intermittent positive or negative pressure was introduced in the 1960s. It was compared with standard treatment in five trials for infants with very severe lung disease and resulted in a reduction in mortality. This effect was observed principally in infants with birth weights over two kilograms. Mechanical ventilation has become standard therapy for severe respiratory failure. There have been no trials in modern neonatal intensive care units so the magnitude of the benefits and harms in current practice are not known.
Any IVH at autopsy is not significantly different between the groups in any study or overall in four studies reporting on 202 infants who had an autopsy. Pneumothorax was reported in two studies of 275 infants and there is a non-significant trend towards an increase in the mechanical ventilation group [summary RR 2.75 (0.72, 10.45)].
The introduction of neonatal intensive care, including MV, during the 1960s and its widespread application in the 1970s was associated with increased survival of very low birthweight infants and was shown to be more cost effective in infants of 1 - 1.5 kg birthweight compared with infants of less than 1 kg (Boyle 1983). Although MV, usually using IPPV, is now a standard treatment (Greenough 1996; Wiswell 2001) it is not clear what the balance of benefits and harms was at the time it was introduced and how these should be interpreted in terms of modern neonatal intensive care. In earlier debates Reynolds 1970 suggested that infants with severe RDS could be managed just as well with excellent standard care, without use of MV. Given the cost of IPPV, the question still arises as to whether it should be introduced as part of neonatal care in resource poor settings, such as in many developing countries (Ho 1996).
The effects of MV have been reviewed previously (Bancalari 1992) but such effects require formal re-examination by repeating the search for randomised controlled trials, including those that may have appeared only in abstract form, to clarify any remaining research questions and to better describe the population of infants who entered these trials.
Pre-specified sub-group analyses were to be carried out according to:
1. RDS as cause of respiratory failure vs all other causes
2. Early vs late (rescue) treatment with MV
3. Type of MV - either IPPV or INPV
4. Gestation (cut-offs at about 28 and 32 weeks)
5. Birth weight (cut-offs at about 1000 and 1500 grams)
Although no trials comparing MV with head box oxygen are likely to have been conducted since the availability of artificial surfactant or the use of positive end-expiratory pressure, if trials utilizing these latter interventions were found, sub-group analyses were to be done according to the use, or not, of these therapies.
Two authors independently assessed eligibility of retrieved reports, and extracted data; the results were compared and any differences resolved by discussion and consensus.
The data were synthesised using the standard method of the Neonatal Review Group with use of relative risk and risk difference and their 95% confidence intervals. A fixed effect model was used for meta-analyses.
Five published trials (Reid 1967, Silverman 1967, Sinclair 1968, Llewellyn 1970, Murdock 1970) were found and included. All trials were carried out in the latter half of the 1960s. No unpublished trials or trials published only in abstract form were found.
Participants
All studies enrolled infants of more than one kilogram birth weight with
respiratory distress. In all but one study, the infants all had clinical
and radiological RDS. In the other study (Sinclair 1968) 50% had RDS. No gestational age limits were used and only one study (Sinclair 1968)
specified an upper limit for birth weight, at 2.5 kg. The severity of respiratory
failure based on oxygen requirements varied. From least to most severe, it
was oxygen saturation less than 80% in air (Silverman 1967), use of 40-100% oxygen (Reid 1967), PaO2 75 mm Hg or less in 50% oxygen (Sinclair 1968), PaO2 less than 100 mm Hg in more than 95% oxygen (Llewellyn 1970) and PaO2 less than 50 in 95% or more oxygen (Murdock 1970).
Interventions
Mechanical ventilation was provided in a wide variety of ways. Silverman 1967 and Sinclair 1968 used negative pressure Airsheilds ventilators, Reid 1967 used IPPV via a naso-tracheal tube and Murdock 1970 gave IPPV via a face-mask. Llewellyn 1970
had one group on INPV, one on IPPV with a pressure cycled ventilator and
another on IPPV with a volume cycled ventilator. The authors found no statistical
difference in outcomes between these different methods and so combined them
in the publication. The original data on outcomes for each type of MV is
no longer available for analysis (Swyer personal communication).
Outcomes
Mortality was reported in all studies but the period of follow up (during
the study period, neonatal or prior to discharge) was variable. Sinclair 1968 reported deaths during the seven day study and in the neonatal period. Silverman 1967
only reported deaths during the seven day study period. Published reports
of three trials did not indicate when the deaths had occurred. Dr Paul Swyer,
co-author of two of these trials (Murdock 1970, Llewellyn 1970) was contacted and he recalled that deaths were ascertained by hospital discharge. Dr David Reid (Reid 1967)
provided time of death information from his trial indicating that all deaths
occurred in the first seven days and that there were no late deaths before
discharge. The outcome 'any reported mortality' in this review presents data
over the longest period follow-up period in each study. All trials reported
IVH, but this was only reported for those with autopsies as ultrasound examination
was not available in the 1960s. Overall, autopsies were carried out in 79%
of the mechanical ventilation group deaths and 91% of the control group deaths.
All trials randomly allocated subjects to treatment or control groups and all five concealed the random allocation from clinicians caring for the infants. Neither the treatments nor the outcome assessments were blinded in any trials. Outcomes were ascertained in almost all subjects randomised in each trial.
The five eligible trials reported on a total of 359 infants with RDS.
Mortality
All trials reported mortality. Sinclair 1968
found a higher neonatal mortality in the mechanical ventilation group [7/10
vs 1/10; RR 7.00 (1.04, 46.95)]. Overall in the five trials any reported
mortality was less frequent in the mechanical ventilation group with the
upper 95% confidence limit of RR on 1.00 [summary RR 0.86 (0.74, 1.00), RD
-0.10 (-0.20, -0.01), NNT 10 (5, 100)]. For RR there is a non-significant
trend suggesting heterogeneity; for RD there is highly significant heterogeneity.
Two studies reported mortality by birth weight (Reid 1967, Murdock 1970). In infants with a birthweight of 1 - 2 kg no significant difference in mortality was found [summary RR 0.86 (0.70, 1.07)]. In infants with a birth weight of more than 2 kg, one study (Murdock 1970) reported a significant reduction in mortality in the MV group compared with control [RR 0.67 (0.51, 0.86)] and, overall, in the two trials there was a significant reduction in mortality with MV [summary RR 0.67 (0.52, 0.87), RD -0.27 (-0.45, -0.10), NNT 4 (2, 10)].
Intraventricular haemorrhage (IVH)
Any IVH at autopsy was not significantly different between the groups
in any study or overall in four studies reporting on 202 infants who had
an autopsy [summary RR 1.05 (0.79, 1.39)].
Pneumothorax
Pneumothorax was reported in two studies (Silverman 1967, Llewellyn 1970)
of 275 infants and there was a non-significant trend towards an increase
in the mechanical ventilation group [summary RR 2.75 (0.72, 10.45)].
Proven systemic infection was only reported in one trial (Murdock 1970) which found five cases of septicaemia in the ventilated group (45 survivors and 96 with postmortem examination) and none in the control group (eight survivors and 45 with postmortem examination).
Other prespecified outcomes could not be assessed.
Subgroup analysis by IPPV and INPV was not done because in the largest study (Llewellyn 1970) outcomes by these modalities were not available. Subgroup analyses by gestation, cause of respiratory distress, and early or late use of intervention could not be carried out.
The trials included in this review were carried out in the 1960s when neonatal intensive care was just being introduced and overall care was less sophisticated than current neonatal care. The equipment and methods used to apply mechanical ventilation cannot be compared to those used now (reviewed by Wiswell 2001). Intermittent negative pressure is now rarely used and ventilators for IPPV have been especially developed for use with newborn infants (Wiswell 2001). Mortality rates for infants with moderate or severe RDS, such as those entered in the studies in this review, were much higher (overall 67%) than observed in the 1980s (5%) (Greenough 1985). This is despite the relatively high birth weight of the infants in the included studies compared to current NICU infants. Furthermore, treatments such as antenatal corticosteroids and artificial surfactants were not available. For these reasons there are considerable limitations in applying the results of this review to current NICU practice.
When these trials were done the main question was whether mechanical ventilation could save the lives of infants with severe RDS. Apart from one of the trials, the results suggest that mortality is reduced. Prespecified subgroup analyses could not be carried out to explore the heterogeneity in the overall mortality analysis. Post-hoc examination showed that the trial with a higher mortality in the MV group (Sinclair 1968) had a much lower control group mortality rate (10%) than the other four trials (range 63 - 85%). Caution is also warranted in interpreting the results for 'any mortality' presented here because the outcome was ascertained over different time periods in some studies. As ultrasound and computerised tomographic imaging were not available when these studies were carried out, IVH was ascertained at autopsy and the incidence in survivors is not known.
In some settings, such as some developing countries where resources limit the ability to provide neonatal intensive care, the question of whether MV is worthwhile is still valid (Ho 1996). In such settings it is uncertain what additional benefits MV might provide over oxygen administration alone or other lower cost support such as continuous positive airways pressure (reviewed by Ho 2005a; Ho 2005b).
The results here and those examining the cost effectiveness of neonatal intensive care (Boyle 1983) suggest the infants of greater birthweight might benefit more in settings where MV is being introduced. Furthermore, more MV resources are used in treating the lowest birth weight infants (Doyle 1996).
Study | Methods | Participants | Interventions | Outcomes | Notes | Allocation concealment |
Llewellyn 1970 | Concealment of randomisation - yes; blinding of intervention - no; completeness of follow up - yes; blinding of outcome assessment - no | 44 infants of 127 admitted with RDS; PaO2 < 100 in > 95% | IPPV via face mask using pressure (Bird Mk VIII) or volume (Bourns Pediatric Respirator, model LS 104) cycled ventilator, maximum pressure of 20 cms H2O vs Standard treatment (servocontrolled ventilator, 4 - 6 hrly blood gases, PaO2 kept at 50 - 80, metabolic acidosis corrected with NaHCO3, 10% dextrose) | Mortality before discharge, intubation for IPPV, duration of oxygen therapy | All infants outborn May 1968 - March 1969. Dr Swyer provided additional information that randomisation was concealed, standard errors reported, deaths were before discharge. | A |
Murdock 1970 | Concealment of randomisation - yes; blinding of intervention - no; completeness of follow up - yes; blinding of outcome assessment - no | 221 with RDS with PaO2 < 50 mmHg in FiO2 > 0.95 or cyanosis despite such FiO2 and blood gas not available or apnea unresponsive to bag and mask ventilation | Mechanical ventilation (IPPV with pressure or volume cycled ventilators and endotracheal tubes, INPV preferably without intubation) vs standard care (FiO2 to keep PaO2 60 - 100 mmHg, 10% dextrose, servocontrolled incubator, NaHCO3 to keep pH > 7.25) | Mortality before discharge by birthweight, Pneumothorax, IVH at autopsy | All infants outborn November 1965 - February 1968. Dr Swyer provided additional information that randomisation was concealed, standard errors were reported and deaths were before discharge. | A |
Reid 1967 | Concealment of randomisation - yes (off-site by independant statistician); blinding of intervention - no; completeness of follow up - unclear; blinding of outcome assessment - no | 20 infants with clinical RDS, Silverman Score > 4, in 40 - 100% O2 and an initial capillary pH of < 7.20 Infants of < 1000gms birthweight excluded | IPPV vs standard treatment (10% dextrose plus NaHCO3, oxygen to prevent cyanosis, heated humidified incubators) | Mortality by birthweight, IVH at autopsy (autopsy rate 100%) | All infants given Ampicillin and Cloxacillin 1964 - 66. Dr Reid supplied additional information about the method of randomisation and details of the deaths. | A |
Silverman 1967 | Concealment of randomisation - yes, paired by outborn (18) and inborn (36) and in 500gm weight groups; blinding of intervention - no; completeness of follow up - yes; blinding of outcome assessment - no | 474 admissions, 420 did not meet criteria, 54 infants included at mean age of 8 hrs with clinical and radiological (independent assessment) diagnosis of RDS and cyanosis or capillary SaO2 < 80% in air or PCO2 > 50 mmHg and birthweight > 1kg | INPV in Airshields incubator, pressures -15 to -45 cms H2O vs standard care (dextrose and NaCO3 IV) | Mortality during the 7 day study period, IVH at autopsy, pneumothorax | February 1963 - December 1964 Author confirmed that developmental follow up not done | A |
Sinclair 1968 | Concealment of randomisation - yes; blinding of intervention - no; completeness of follow up - yes; blinding of outcome assessment - no | 20 infants (50% with RDS) with birthweight 1000 - 2500gms, < 24 hrs old, pH < 7.25 or PaO2 < 76 in FiO2 0.5 | Randomised
to 4 groups according to use of unlimited O2, rapid bicarbonate administration
and assisted ventilation with INPV (Airshields) In this review, the 2 groups which received INPV were compared with the 2 groups not receiving INPV. | Mortality (first week and neonatal), IVH at autopsy (any or massive) | April 1966 - January 1967 Author confirmed that developmental follow up not done | A |
Llewellyn MA, Tilak KS, Swyer PR. A controlled trial of assisted ventilation using an oro-nasal mask. Archives of Disease in Childhood 1970;45:453-9.
Murdock 1970 {published and unpublished data}
Murdock AI, Linsao L, Reid MMcM, Sutton MD, Tilak KS, Ulan OA, Swyer PR. Mechanical ventilation in the respiratory distress syndrome: a controlled trial. Archives of Disease in Childhood 1970;45:624-33.
Reid 1967 {published data only}
* Reid DHS, Tunstall ME, Mitchell RG. A controlled trial of artificial respiration in the respiratory-distress syndrome of the newborn. Lancet 1967;1:532-3.
Reid DHS. Perinatal repiratory problems with special emphasis on the treatment of respiratory failure in the newborn with intermittent positive-pressure respiration. MD Thesis, University of St Andrews (now held in University of Dundee), Scotland 1968.
Silverman 1967 {published and unpublished data}
Silverman WA, Sinclair JC, Gandy GM, Finster M, Bauman WA, Agate FJ. A controlled trial of management of respiratory distress syndrome in a body-enclosing respirator. 1. Evaluation of safety. Pediatrics 1967;39:740-8.
Sinclair 1968 {published and unpublished data}
Sinclair JC, Engel K, Silverman WA. Early correction of hypoxemia and acidemia in infants of low birth weight: a controlled trial of oxygen breathing, rapid alkali infusion, and assisted ventilation. Pediatrics 1968;42:565-89.
* indicates the primary reference for the study
Bancalari E, Sinclair JC. Mechanical ventilation. In: Sinclair JC, Bracken MB, editor(s). Effective Care of the Newborn Infant. Oxford: Oxford University Press, 1992:200-20.
Boyle MH, Torrance GW, Sinclair JC, Horwood MD. Economic evaluation of neonatal intensive care of very-low-birth-weight infants. New England Journal of Medicine 1983;308:1330-7.
Doyle LW, Davis P, Dharmalingham A, Bowman E. Assisted ventilation and survival of extremely low birthweight infants. Journal of Paediatrics and Child Health 1996;32:138-42.
Greenough A, Roberton NRC. Morbidity and survival in neonates ventilated for the respiratory distress syndrome. British Medical Journal 1985;290:597-600.
Greenough A, Roberton NRC. Respiratory distress syndrome. In: Greenough A, Milner AD, Roberton NRC, editor(s). Neonatal Respiratory Disorders. London: Arnold, 1996:238-79.
Ho NK. Priorities in neonatal care in developing countries. Singapore Medical Journal 1996;37:424-7.
Ho JJ, Subramaniam P, Henderson-Smart DJ, Davis PG. Continuous distending pressure for the respiratory distress syndrome in preterm infants. In: The Cochrane Database of Systematic Reviews, Issue 1, 2005.
Ho JJ, Henderson-Smart DJ, Davis PG. Early versus delayed initiation of continuous distending pressure for respiratory distress syndrome in preterm infants. In: The Cochrane Database of Systematic Reviews, Issue 1, 2005.
Reynolds REO. Indications for mechanical ventilation in infants with hyaline membrane disease. Pediatrics 1970;46:193-202.
Sinclair JC. Prevention and treatment of the respiratory distress syndrome. Pediatric Clinics of North America 1966;13:711-30.
Wiswell TE, Donn SM (eds). Mechanical ventilation and exogenous surfactant update. Clinics in Perinatology 2001;28.
Henderson-Smart DJ, Wilkinson A, Raynes-Greenow CH. Mechanical ventilation for newborn infants with respiratory failure due to pulmonary disease. In: The Cochrane Database of Systematic Reviews, Issue 4, 2002.
Comparison or outcome | Studies | Participants | Statistical method | Effect size |
---|---|---|---|---|
01 Mechanical ventilation vs control | ||||
01 Any reported mortality | 5 | 359 | RR (fixed), 95% CI | 0.86 [0.74, 1.00] |
02 Mortality 1 - 2 kg | 2 | 140 | RR (fixed), 95% CI | 0.86 [0.70, 1.07] |
03 Mortality more than 2 kg | 2 | 101 | RR (fixed), 95% CI | 0.67 [0.52, 0.87] |
04 Any IVH in infants with an autopsy | 5 | 202 | RR (fixed), 95% CI | 1.05 [0.79, 1.39] |
05 Pneumothorax | 2 | 275 | RR (fixed), 95% CI | 2.75 [0.72, 10.45] |
Dr Andrew Wilkinson
Head
Neonatal Medicine
John Radcliffe Hospital
Headington
Oxford
UK
OX3 9DU
E-mail: andrew.wilkinson@paediatrics.oxford.ac.uk
The review is published as a Cochrane review in The
Cochrane Library, Issue 3, 2005 (see http://www.thecochranelibrary.com for
information). Cochrane reviews are regularly updated as new evidence emerges
and in response to comments and criticisms, and The Cochrane Library should
be consulted for the most recent version of the Review. |