Oxygen monitoring in preterm babies: too high, too low?

Current variations in practice 

Babies grow perfectly well in utero with arterial blood that is only 70–80% saturated (Fig. 1).15 Why then do we persist in trying to keep their oxygen saturation after birth above 90%, even if this requires prolonged periods of supplementation or even ventilation? Children with cyanotic congenital heart disease make the transition to extra-uterine life without difficulty despite all we know about the effect that oxygen has on pulmonary vascular16 and ductal17 tone. So why do we fear that mild desaturation could dangerously compromise that transition? It is not as though there is a uniform approach to care. Views on what constitutes a safe maximum or minimum saturation for a small baby in the first few weeks of life vary widely. Figure 2 shows how practice currently varies in the UK;18 policies in America vary just as much.19

View Large Image Download to PowerPoint

Figure 1. Maternal and fetal arterial and venous oxygen saturation and partial pressure in late pregnancy.15

View Large Image Download to PowerPoint

Figure 2. Oxygen saturation monitoring policies in the UK. Results from a telephone survey of 100 units with three or more intensive care cots caring for babies of less than 28 weeks’ gestation in 2001. High (A) and low (B) oximeter alarm settings.18

Evidence from a recent observational study 

Policy has varied in the north of England over the past 10 years at least as dramatically as it has in any other part of the UK, and a prospective observational study of every baby born alive before 28 weeks’ gestation to a mother resident in the north of England in 1990–94 stated some fairly provocative findings.20 Data collection started just after the artificial surfactant colfosceril (Exosurf) became widely available in the UK as part of a major collaborative trial,21 and was linked to an epidemiological study of the long-term outcome for all babies born more than 8 weeks early.22., 23. Nasal continuous positive airway pressure (CPAP) was just coming into common use at that time but was not being employed in the initial post-delivery management of babies as immature as this. The babies were born in, or referred for care to, one of five neonatal units in which most care policies were similar, but policy towards the monitoring of oxygen saturation varied widely. The overall survival rate, and the survival rates without evidence of cerebral palsy at 18 months, was almost identical in the five units in the 295/568 babies of 23–27 weeks’ gestation (52%) still alive 1 year after birth (Table 1).20

Table 1. Outcome at 1 year in all babies of 23–27 weeks’ gestation born in 1990–94 and its relationship to minimum and maximum pulse oximeter alarm settings.20

	Oximter alarm settings (%)	Number of babies admitted	One year survivors (number and percentage)
			Number of survivors	Median number of days ventilated	Cerebral palsy	Threshold retinopathy
	88–98a	123	65 (52.8%)	21	11 (16.9%)	18 (27.7%)
	85–95	235	128 (54.5%)	16	20 (15.6%)	20 (15.6%)
	84–94	84	37 (44.0%)	15	6 (16.2%)	5 (13.5%)
	70–90	126	65 (51.6%)	7	10 (15.4%)	4 (6.2%)

a Nellcor pulse oximeter measurements (functional saturation). Other measurements are fractional saturation. Target saturation lay in the upper half of the accepted range.

In one unit, the target fractional oxygen saturation (see Appendix A) was 80–90% (the lower alarm limit set to operate only if the saturation fell below 70%) once the baby was more than 2–3hours old. Such a policy was sustained for all babies thought to need supplemental oxygen until retinal vascularisation was complete. In another unit, target functional oxygen saturation was 94–98% (the lower alarm set to operate at 88%); the other three units had intermediate policies. A careful, uniform ophthalmic review of all the survivors showed that retinopathy severe enough to merit treatment with cryotherapy24 occurred in 6.2% (95% CI 1.7–15.0%) of the babies in the first of these units and 27.7% (95% CI 17.3–40.2%) in the latter (Fig. 3). No child from the first unit, but four from the latter unit, became blind. The three units employing intermediate policies for target oxygen saturation had “threshold” retinopathy rates in the middle of this range.20

View Large Image Download to PowerPoint

Figure 3. Relationship between the limits within which oxygen saturation was allowed to vary and the proportion of 1 year survivors so nursed who later developed severe (grade 3+) retinopathy of prematurity: a comparison of four policies. Staff aimed to keep saturation in the upper half of the permitted range.20 (Reproduced from ref. 20 with permission from BMJ Publishing Group.)

In the unit where the target oxygen saturation was 80–90%, half of the 65 long-term survivors were managing without respiratory support (defined as any plastic tubing below the vocal cords) by 7 days of age and without supplemental oxygen by 30 days. In the unit in which target oxygen saturation was 94–98%, these milestones were achieved by half the survivors in 21 and 72 days, respectively. Growth to discharge was curtailed twice as much in the latter unit. Eleven survivors from the first unit and 29 from the second had a weight below the third centile at discharge (possibly because nutritional intake was, quite subconsciously, increased more cautiously in babies still requiring respiratory support). Neurodevelopmental outcome was similar at 18 months (no child having been lost to follow-up).20 A further review is currently underway to establish whether there are any subtle long-term adverse consequences associated with this more “permissive” approach to mild oxygen desaturation in babies of less than 28 weeks’ gestation in the first few weeks of life.

Effect of a change in policy 

When the results of this analysis first became known, the unit that had been maintaining its functionally calibrated oximeter alarm settings at 88% and 98% modified its practice and lowered the settings to 75% and 93%. In the years 1995–96, before the change in policy, half the surviving babies of 25–27 weeks’ gestation spent more than 13 days on ventilation. In the years 1998–2000 inclusive, half the babies were weaned from ventilatory support by 5 days of age even though there had been no other change in unit policy during this time (Fig. 4).18 This within-unit change over time is comparable to the between-unit difference seen in 1990–94, strengthening the suggestion that accepting a lower target saturation may make it possible to halve the time it takes to wean babies from ventilatory support.

View Large Image Download to PowerPoint

Figure 4. Length of time that babies of 24–27 weeks’ gestation were ventilated in 1995–96 (when the upper and lower oximeter alarms were set at 98% and 88%, respectively) and in 1998–2000 (when the alarms were set at 93% and 75%).18

Evidence from two recent randomised controlled studies 

Two trials have recently been conducted to see whether it is better to keep arterial oxygen saturation high in very preterm babies when they are more than a few weeks old. The American STOP-ROP trial,25 which recruited 649 babies with a mean gestation of 25.4 weeks and a mean post-menstrual age of 35 weeks between 1994 and 1999, showed that keeping the fractional oxygen saturation above 95% slightly reduced the number of babies with pre-threshold retinopathy who went on to develop disease severe enough to require retinal surgery. Benefit was, however, seen only in those without evidence of “plus disease” (dilated and tortuous vessels in at least two quadrants of the posterior pole) at recruitment (32% vs. 46%). More unexpectedly, the higher oxygenation target significantly increased the number in hospital, in oxygen and receiving diuretics at a post-menstrual age of 50 weeks. Significant pulmonary deterioration after recruitment (13.2% vs. 8.5%) was seen only in those with more than average evidence of chronic lung disease at trial entry. The higher oxygenation target did not improve growth or the eventual retinal outcome as assessed 3 months after the expected date of delivery.

The result of the Australian BOOST trial has just been reported.26 There was no evidence that the growth and developmental outcome of babies of less than 30 weeks’ gestation at birth who were still oxygen dependent at a post-menstrual age of 32 weeks was improved by keeping their functional oxygen saturation in the high 90%s. This randomised, double-blind, multi-centre study recruited 358 babies between 1996 and 2000, and a full report of the study is likely to be published shortly. Collaborating units had different policies with regard to optimum oxygenation in the period immediately after birth (as in the STOP-ROP trial) but all monitored saturation using a pre-specified Nellcor N-3000 pulse oximeter after recruitment for as long as supplemental oxygen was deemed necessary. Trial oximeters were specially modified to keep the functional saturation in the range 91–94% or 95–98%, depending on allocation at entry, while displaying a figure in the range 93–96%.

Upper limit 

There is unfortunately no controlled trial evidence with which to shape policy. The observational study from the north of England20 suggests that the risk of retinopathy increases when the oxygen saturation is allowed to exceed approximately 90% in the first few weeks of life, a figure in line with the earlier observational study14 proposing an increase when the transcutaneous partial pressure reached or exceeded 80mmHg (since studies show that fractional oximeter readings above 92% can be associated with partial pressures of 80mmHg or more,27., 28. as Fig. 5 makes clear).

View Large Image Download to PowerPoint

Figure 5. Relationship between fractional oxygen saturation measured with a pulse oximeter and arterial partial pressure in mmHg and kPa. The dashed line marks the transcutaneous PO2 above which there was an increased risk of retinopathy in the study reported by Flynn in 1992.14 The bars in (B) show the range within which 95% of all measures of partial pressure varied when the oximeter read 90%, 92%, 94%, 96% and 98% in the study reported by Brockway and Hay in 1998.28 Note that pulse oximeters calibrated to display functional saturation produce readings about 2% higher than those reflecting fractional saturation.29., 30. (Reproduced from Neonatal Formulary 3, p. 171, Figure 1, Figure 2, 2002, with permission from BMJ Books.)

The babies in the two polar arms of the north of England study had been very similar on admission. Antenatal steroid use and the rates of death during delivery and before admission to special care were comparable. Neither adjusting for the minor difference in the distribution of gestation, nor excluding post-delivery transfers, altered any of the conclusions. Nevertheless, only a controlled trial can ever establish whether some unrecognised factor other than the difference in target oxygen saturation was responsible for the observed difference in the incidence of retinopathy, the difference in the amount of ventilatory support required and the magnitude of the post-delivery growth deficit.

Lower limit 

If there is debate over what constitutes the safe upper limit for fractional oxygen saturation, there is even greater uncertainty over what constitutes the lower safe limit. Mortality went undocumented in the first formal trial of restricted oxygen administration,1 and the second trial was large enough only to rule out a major increase.2 The current observational study,20 like the early small trial by Usher,6 is also too small to address such an issue but it does suggest that a more permissive approach to minor hypoxia may, like a permissive approach to hypercapnia, reduce the need for ventilatory support, the consequential risk of chronic lung disease and the associated health-care costs. We should not forget that the levels being discussed here are all higher than the retina and the brain would normally experience during fetal life.