Intended for healthcare professionals

Education And Debate

Hypoxia in childhood pneumonia: better detection and more oxygen needed in developing countries

BMJ 1994; 308 doi: https://doi.org/10.1136/bmj.308.6921.119 (Published 08 January 1994) Cite this as: BMJ 1994;308:119
  1. T Dyke,
  2. N Brown
  1. Departments of Community Medicine and Paediatrics, Faculty of Medicine, University of Papua New Guinea, PO Box 5623, Boroko, NCD, Papua New Guinea.
  • Accepted 22 September 1993

Even though hypoxia is a major risk factor for death in children with acute respiratory infection in developing countries, oxygen is not part of first line treatment. Because oxygen is not readily available in developing countries it tends to be given to the most seriously ill children, whose outcome is poor. Oxygen might be useful if given earlier in the course of the disease. Clinical signs are not clear cut, however, though the presence of cyanosis and grunting together with a raised respiratory rate can significantly increase the detection of hypoxaemia. A simple oximeter would make detection easier, and oxygen concentrators are more cost effective than bottled oxygen. Ideally oxygen should be given to children in the early stages of clinical pneumonia to prevent deterioration.

Acute respiratory infection is a major killer of children in developing countries, especially of those aged less than 6 months.1 Although many cases of acute respiratory infection are initially caused by viruses, children are often secondarily infected with bacteria by the time they present to a health facility. The use of standard protocols for antibiotic use has been a major part of control programmes for acute respiratory infection throughout the world and is advocated by the World Health Organisation.2 Bacteraemia in acute respiratory infection has shown a significant association with hypoxaemia in terms of recorded cyanosis,3 but oxygen has not been considered as first line treatment in the same way as antibiotics.

Hypoxaemia has been recognised as a risk factor for death in children presenting with acute respiratory infection,4 but there have been no controlled trials in the developing world of the therapeutic value of administering oxygen. Paradoxically, those children who receive oxygen have a poorer outcome because they are more seriously ill when oxygen is started (Papua New Guinea Institute of Medical Research, unpublished data). Attempts to quantify the effect of oxygen in acute respiratory infection are likely to be considered unethical so we must develop a coherent strategy for the diagnosis of hypoxaemia and the use of oxygen in childhood pneumonia from clinical and pathological principles.

Mechanisms of hypoxia

The principal mechanism for the hypoxia of acute respiratory infection is a mismatch between ventilation and perfusion. The infectious organism, be it viral or bacterial, causes areas of pneumonic consolidation, which become inappropriately underoxygenated relative to their reactive hyperperfusion.5 The mismatch is not redressed by vascular redistribution to the unaffected parts of the lung as most pneumonia in children is of a bronchopneumonic distribution rather than showing the lobar pattern seen in adults. Moreover, lung compliance decreases as consolidation develops, leading to increased work required for ventilation. Dehydration from fever, panting, and inability to drink lead to haemoconcentration, peripheral underperfusion, and increasing metabolic acidosis and will cause a further deterioration in the general condition. The acidosis also leads to compensatory hyperventilation, which limits the usefulness of an elevated respiratory rate in assessing the degree of hypoxia despite its usefulness in gauging the degree of systemic disturbance. This has been confirmed in studies from the Gambia.6 The progressive deterioration raises the question of whether this course can be prevented by the earlier use of oxygen.

Are the indications for using oxygen clear and well understood?

Most health care in developing countries is provided by nurses and paramedical workers. Even in district hospitals the triage of new patients and their initial management is rarely done by doctors. It is therefore essential not only that changes in the use of oxygen are compatible with the resources available but also that the indications for its use are understood by the appropriate staff. The recently updated indications for the use of oxygen in the standard treatment manuals used in Papua New Guinea include cardiac failure, grunting, drowsiness, and apnoeic episodes in addition to cyanosis and restlessness.7

What clinical signs should be indications for use of oxygen?

As mentioned earlier, an elevated respiratory rate has limitations as an indication for the use of oxygen so other clinical signs must be used. Cyanosis is another obvious candidate, but, although highly specific, it has an unreasonably low sensitivity (T Dyke et al, annual symposium of Papua New Guinea Medical Society, 1991).4 Cyanosis is a late, probably terminal, sign by which to recognise hypoxia. It is also subtle and may be missed by an inexperienced observer, especially with dark skinned children. In those parts of the world where anaemia is common cyanosis becomes an even less sensitive sign. Grunting has been found to be positively related to hypoxaemia in studies in Kenya4 and Papua New Guinea (T Dyke et al, annual symposium of Papua New Guinea Medical Society). It is probably a response to hypoxia: the Valsalva manoeuvre of forcibly exhaling against a closed glottis raises intra-alveolar pressure and leads to a greater transfer of oxygen. Grunting probably also helps to prevent alveolar collapse. Cyanosis and grunting have previously been identified as independent risk factors for dying from pneumonia in children in the highlands of Papua New Guinea.3 The use of either of these signs together with drowsiness or a grossly elevated respiratory rate (more than 80 breaths per minute) can significantly increase the detection of hypoxaemia (T Dyke et al, annual symposium of Papua New Guinea Medical Society). The use of any clinical sign is limited, however, by poor correlation between different observers' assessments of the same clinical sign. Correlation for respiratory rate and retractions is particularly poor.8 We conclude that refining our understanding of the relation between clinical signs and hypoxaemia is unlikely to improve the detection of hypoxaemia.

Could modern technology improve detection of hypoxia?

If the cost of pulse oximeters was reduced decisions on the use of oxygen could possibly be directed by oximetery readings. But this is unlikely to be an option in the developing world for the foreseeable future except in a few well equipped centres. There is clearly a market for a simple oximeter that could be used in district health centres in the developing world. At an altitude of 1500 m in the highlands of Papua New Guinea oxygen saturation is never less than 93% among healthy children (Papua New Guinea Institute of Medical Research, unpublished data), but the lower limit of normal oxygen saturation decreases with increasing altitude.9 At higher altitudes supplementary oxygen given to all children with hypoxaemia (defined as a saturation of less than 90%) will include some children with only mild disease, but it is clearly preferable to include such children rather than to deny a beneficial treatment to other children by setting a lower saturation level such as 85%.

Can we increase the availability of oxygen?

Even if the detection of hypoxaemia could be improved a shortage of oxygen in most of the developing world would still force many health workers to make decisions on the basis of availability of oxygen rather than on the needs of children. Which children with pneumonia will benefit most from the limited oxygen that is available? A pragmatic approach is to give a test dose of oxygen and to monitor the child's response. If there is improvement treatment should continue, but the decision to stop may be very difficult without rational criteria.

The advent of reliable and appropriately designed oxygen concentrators, however, may make a substantial difference to the management of childhood pneumonia. Using oxygen concentrators is also likely to be much more cost effective than using conventional bottled oxygen.10 Although the unit cost of bottled oxygen is small, the high costs of transporting cylinders, often by plane, and of cylinder rental agreements make it expensive in many parts of the developing world. Three manufacturers (two of which are British) have now met the WHO's specifications with reliable oxygen concentrators for use in developing countries. However, even these concentrators depend on a steady supply of electricity (seldom available in rural health centres) and a small but essential degree of maintenance.

Where possible, oxygen should be administered to children with pneumonia by intranasal catheter. This route has distinct advantages over both face mask and head box methods: it is less wasteful, having been shown to give concentrations of 50% at low flow rates (0.5 l/min); it does not restrict breast feeding; and it is safer should the tubing become disconnected.11 A possible disadvantage is the minor degree of gastric distension that can result, which might lead to vomiting and aspiration.

Conclusion

Oxygen is being underused in childhood pneumonia in developing countries largely through not being available. The use of oxygen should perhaps not be thought of as a treatment for hypoxia but as a treatment for clinical pneumonia or bronchiolitis to prevent deterioration. Supplementary oxygen is probably most effective for children with pneumonia during the earlier stages of the disease before the child becomes exhausted and acidotic. A reasonable aim should be to give oxygen to all children with acute respiratory infection and moderate pneumonia (those who exhibit indrawing) who are currently admitted to a health centre rather than to the subgroup of children who currently qualify under the WHO guidelines. Where oxygen is scarce the decision to administer oxygen can be reviewed 24 hours later.

References

View Abstract