Dear Editor,

We are happy to make it clear that we have never suggested that hypoxia is "beneficial" to babies with chronic lung disease. Indeed in describing our own practice we said, quite specifically, that "babies who were at least 8 weeks old [and it should be remembered that all our babies were born more than 12 weeks early], and whose retinal vasculature was mature, received liberal oxygen supplementation." We would, however, remind Dr Primhak that those babies in the recent STOP-ROP trial who were given enough supplemental oxygen to maintain a saturation of 96-99% (to see if this reduced the severity of the retinopathy they had already developed) developed significantly more pulmonary problems than those only given enough oxygen to maintain a saturation of 89-94%.[1] The idea that oxygen is always a "good thing" dies hard. Iles and Edmunds[2] showed that babies with a saturation below 90% in air at discharge were more likely to have a frightening colour change, apnoeic episode and/or sudden change in muscle tone during the subsequent three month study period, but they did not show that that this risk was reduced by giving oxygen. There is equally little objective evidence that offering sustained supplemental oxygen actually does reduce the incidence of troublesome pulmonary hypertension.

Win Tin
Department of Paediatrics
South Cleveland Hospital
Middlesbrough TS4 3BW, UK

David Milligan
Philippa Pennefather
Royal Victoria Infirmary
Newcastle upon Tyne NE1 4LP, UK

References
(1) The STOP-ROP Multicenter Study Group. Supplemental therapeutic oxygen for prethreshold retinopathy of prematurity (STOP-ROP), a randomised, controlled trial. I: Primary outcomes. Pediatrics 2000;105:295 -310.

(2) Iles R, Edmunds A. Prediction of early outcome in resolving chronic lung disease of prematurity after discharge from hospital. Arch Dis Child 1996;74:304-8.

Dear Editor,

I am glad to have a chance to respond to Dr Roberton's assertion that the care of the babies nursed using oximeter settings of 70-90% was "negligent", since I was responsible for these children, but time and space does not allow a full response. Neither does space allow me to respond to the criticism implicit in your own introductory statement that such care "breaches BAPM guidelines".

Dr Roberton says the cerebral palsy rate is "irrelevant", but parents might not agree. Parents might also be glad that, while 4 children monitored using an oximeter alarm set at 88-98% went blind, no child in the other group went blind. They might also be glad that half were off the ventilator in 7 rather than 22 days, and out of oxygen in 4 rather than 10 weeks. The NHS might be equally grateful for the reduction in cost such an approach delivers. Post delivery growth in the conservatively managed group was only retarded half as much as in the comparator group, even though only a quarter ever received any parenteral nutrition. I am happy to leave parents to be the judge of whether this was "negligent" care.

Babies were not "kept at 3.3-6.0 kPa for days and weeks" - target saturation delivered an arterial partial pressure of 5-11 kPa, but alarm settings were more generous than this to discourage staff from adjusting the ventilator every time saturation transiently fell below 80%. Nor was blood pressure monitored by oscillometry (a technique that is known to be unreliable),[1] as a proper reading of the paper would reveal. Dr Roberton mistakenly calls our survival rate our mortality rate, compares survival for mostly black American with that of our white English children, compares survival to discharge with survival to one year, and says nothing about the reliability with which gestation was documented.[2] The same issue of your journal contains a better review of survival.[3] We have every intention of following these children, but felt it would be wrong to wait ten years before reporting the above findings.

Dr Roberton mentions the outcome of a study of 38 children offered corrective surgery for transposition 6 months to 5 years after birth.[4] Those operated on early had a better cognitive outcome, but Dr Roberton does not mention the fact that these children had a mean saturation of 68% before operation, and that 8 had a history of acquired central nervous system damage.

However the main thrust of Dr Roberton's letter is that lack of an arterial line subjected these babies to unnecessary pain. This overlooks the fact that morphine was given during early care, while early extubation greatly reduced the total number of blood samples eventually taken (as the differing transfusion needs confirm). Samples were not taken every 2-3 hours initially, but every 6-8 hours. In fact Dr Roberton and I are at one in agreeing that minimising pain is a very valid reason for inserting an umbilical artery line in babies as immature as this; the limited use of lines was only mentioned because these have been considered necessary in the past to minimise the risk of severe retinopathy - a belief for which there is absolutely no controlled trial evidence.

Finally Dr Roberton asks if this approach ever had ethics committee approval. It did not, because it was merely a continuation of the non-invasive approach initiated by my predecessor Dr Neligan in the mid 1970s, aided by the arrival of transcutaneous gas monitoring. Neither was the introduction of pancuronium in Dr Roberton's own unit[5] placed before an ethics committee.

I can only conclude that someone ought to cull the two sacred cows Dr Roberton has been worshipping (along with all the other animals currently being culled in the UK). In the absence of any other evidence based information, we need a proper controlled trial to address these issues.

Edmund Hey
Retired Consultant Paediatrician, Newcastle

References
(1) Northern Neonatal Nursing Initiative. Systolic blood pressure in babies of less than 32 weeks gestation in the first year of life. Arch Dis Child Fetal Neonatal Ed 1999;80:F38-42.

(2) Wariyar U, Tin W, Hey E. Gestational assessment assessed. Arch Dis Child Fetal Neonatal Ed 1997;77:F216-20.

(3) Evans DJ, Levene MI. Evidence of selection bias in preterm survival studies: a systematic review. Arch Dis Child Fetal Neonatal Ed 2001;84:F79-F84.

(4) Newberger JW, Silbert AR, Buckley LP, et al. Cognitive function and age at repair of transposition of the great arteries in childen. N Engl J Med 1984;310:1495-9.

(5) Greenough A, Wood S, Morley CJ, et al. Pancuronium prevents pneumothoraces in ventilated preterm babies who actively expire against positive pressure ventilation. Lancet 1984;i:1-4.

Dear Editor,

The observations of Tin et al have led them to suggest that babies may have better overall outcomes when unit policies aim at oxygen levels of 70-90%, much lower than current practice in most NICUs. While I would support their call for further well-designed research into this question, I have major concerns that this concept of beneficial hypoxia might creep into clinical practice, and even be extended to the older survivors. The authors are clearly aware of the limitations of their study. There are obviously many possible alternative reasons for the differences in outcome between the nurseries; Table 2 of the study suggests widely divergent policies on a number of issues apart from oximetry levels. There are no data supplied regarding the actual oximetry levels maintained in the nurseries, which makes conclusions about the safety of a saturation of 70% rather speculative.

My main concern is the potential risk to older babies with chronic lung disease who might once again be sunjected to chronic hypoxia. Since the more widespread acceptance that babies with chronic lung disease require similar oxygen levels to their more fortunate brethren we have largely abolished the high first year mortality in these babies, and the pulmonary hypertension which was previously seen. One observational study of differing oximetry levels within a single unit confirmed the high risk of even mild chronic hypoxia in this group of infants,[1] showing a high risk of apparently life threatening events in the hypoxic infants.

While there is continued uncertainty about the optimum oximetry levels in the early life of a preterm baby, there is no justification for maintaining subnormal levels of oxygen in babies beyond 34-36 weeks of age with chronic lung disease, and I trust that this paper will not encourage such practice.

References
(1) Iles R, Edmunds A. Prediction of early outcome in resolving chronic lung disease of prematurity after discharge from hospital. Arch Dis Child 1996;74:304-8.

Dear Editor,

I read the descriptive study of Tin at al[1] with considerable interest. In essence it challenges two sacred cows of neonatal intensive care, whether intra-arterial monitoring is necessary, and what is the appropriate PaO2 at which to nurse critically ill babies.

Arterial Monitoring
They do not give us accurate details of arterial catheter use. There is a hint that they are used for the first few days before resorting to SpO2 and capillary measurements. Nor do they tell us what analgesia is used for multiple capillary samples.

A fundamental principle of neonatal intensive care is minimal handling, and indwelling arterial catheters allow all samples to be taken with no or minimal disturbance, and if the catheters are umbilical, they can also be used safely for virtually all infusions including TPN. Furthermore, as opposed to oscillometric techniques, they allow accurate blood pressure recordings.

Surprisingly, the literature, and my own clinical experience is that the serious complications of UAC are much more common in term infants, and within 48-72 hours of insertion, so that 28/52 babies leaving UAC in situ for many days is unlikely to have a major impact on the putative complication rate of this procedure that induces anxiety in neonatologists.

Local analgesia for heel pricks is surprisingly ineffective even if applied for (impractically) long periods prior to puncture. If general analgesia with, say, morphine is being used the manipulation involved in capillary sampling is still likely to result in the changes in oxygenation (and thus intracerebral haemodynamics with potentially damaging sequelae), that were so graphically illustrated in many papers in the 1970s and 1980s when continuos PaO2 and tcPO2 monitoring became available.

To inflict frequent painful capillary sampling procedures on an unstable 25/52 800 g neonate in the first week of life where blood gas sampling may be necessary at least 2-3 hourly could at best be described as ill judged, at worst negligent.

Appropriate PaO2
As they rightly say, running babies at SpO2 levels of 70-90 (PaO2 approximately 25-45mmHg, 3.3-6 kPa) is an old idea, and I remember spirited arguments about it with the late great Sir Peter Tizzard during my training in the mid 1960s.

It is interesting, but not new, that if you keep babies cyanosed, ROP is rare. Many anecdotal papers from the late 1950s during the panic over oxygen therapy showed that rigid restriction of oxygen dramatically reduced ROP, but probably increased mortality.[2] Although the validity of papers reporting on an association between oxygen restriction in the late 1950s and 1960s and mortality have been challenged, it remains an anxiety.

It remains in this study with an overall mortality of 52%. We do not know the proportion of babies off 23/24 weeks in the study, but it is likely to be relatively small. Contemporary studies reported by Lorenz from the USA[3] give an overall mortality of 38% in babies of compatible gestation and year of birth, falling to 26% in those of 25, 26 and 27 weeks. The figures for Cambridge were virtually identical. Therefore, unless the Newcastle units are over-loaded with 23/24 week babies the overall mortality rate with "physiological" oxygenation (has Blairite spin even penetrated neonatology?) is worryingly high.

Reporting their cerebral palsy rate is irrelevant. Improved neonatal care has in general (depressingly) little effect on cerebral palsy rates; what changes is the number of survivors and the numbers dying.

Importantly, cerebral palsy is only one part of the problem of surviving ELBW. Equally worrying is their under performance at school in childhood and adolescence. In the past, when children with cyanotic congenital heart disease were either inoperable or operated on only in early childhood, prolonged early hypoxaemia of the level used by Tin et al was associated with subsequent cognitive defects.[4] The Newcastle group has a distinguished track record of long term follow up studies, and hopefully the babies in this study will be followed until adolescence. However, until such data show that babies kept at 3.3-6.0 kPa for days and weeks during a vulnerable period for brain development do as well as those kept adequately oxygenated, I would regard the practice as described by Tin et al as of unproven benefit and possibly dangerous.

Finally, the marked restriction of oxygen therapy was, at best, experimental. Was a protocol for this study presented to the local research ethics committee? If not, why not?

References
(1) Tin W, Mulligan DWA, Pennefather P, Hey E. Pulse oximetry, severe reitnopathy and outcome at one year of babies of less than 28 weeks gestation. Arch Dis Child Fetal Neonatal Ed 2001;84:F106-10.

(2) Bolton DPG, Cross KW. Further observations on cost of preventing retrolental fibroplasa. Lancet 1974;i:445-8.

(3) Lorenz JM. Survival of the extrmely preterm infant in North America in the 1990's. Clinics in Perinatology 2000;27:255-62.

(4) Newburger JW, Silbert AR, Buckley LP, Fyler DC. Cognitive funtion and age at repair of transposition of the great arteries in children. N Engl J Med 1984;310:1495-9.