This study has identified numerous issues relating to oxygen therapy for paediatric patients from the perceptions of healthcare professionals involved in its’ supply and delivery. Across all sites, main barriers highlighted included poor electricity supplies, frequent concentrator malfunctions, low staffing levels and lack of training. At Atutur and Bududa DGHs, these barriers appeared compounded by lack of oxygen cylinders, lack of oxygen concentrators, lack of delivery equipment and no access to pulse oximetry, issues which were reflected in improvement proposals. Parents and carers’ fear of oxygen therapy provided an additional barrier to oxygen delivery to children. Through the prioritisation of oxygen supplies by administrative staff, effective communication between the stores and wards, large back-up supplies of oxygen cylinders with functional flow-meters, a paediatric-specific oxygen concentrator, access to and use of pulse oximetry, healthcare professionals at Pallisa DGH reported use of oxygen therapy that more closely adhered to international recommendations [12].
The oxygen supply and delivery problems highlighted by our study reflect many issues raised in existing literature [3, 4, 6, 7]. There may be a number of reasons for this. In 2014, health spending per capita in Uganda, Papua New Guinea and the Gambia was 48, 107 and 23 USD respectively, contrasting with 4568 USD in the UK [15]. Although direct economy comparison between settings is difficult, prioritising care with these small budgets is challenging.
Oxygen concentrators are frequently advocated as oxygen sources throughout the developing world [16,17,18], for their relative low cost and ability to ‘make’ oxygen on site. They are the WHO preferred means of oxygen delivery in resource-limited health centres. At all three sites, concentrators were often discussed with frustration due to the frequency of breakdown, time taken for repair work to be undertaken and inadequate power supplies. Enarson and colleagues discuss that although a plausible oxygen supply strategy in their resource-limited setting [17], they advocated the use of flow-splitters with concentrators to address the issue of demand overburdening oxygen supplies [17]. Flow-splitters were not used in any of the hospitals in our study, limiting the number of patients able to receive oxygen from each concentrator. Enarson and colleagues highlight that equipment, training and maintenance must accompany oxygen concentrator systems in order to achieve effective oxygen delivery [17].
The reliance on a stable power supply also presented a large problem for concentrators in all hospitals. A study by Peel and colleagues demonstrated lower voltages and frequent power outages caused electrical faults in concentrators and break down, possibly explaining the problems reported at each hospital in our study [19]. Power supply issues in Uganda were reflected in a World Bank assessment in 2009 [20]. Uganda’s capacity factor- an indicator of power system’s proximity to overload - was calculated at 94%, the highest of all sub-Saharan African countries studied [20]. This problem demands a satisfactory back-up power solution. A study undertaken in a large Eastern Uganda referral hospital concluded solar energy was sufficient to power oxygen concentrator systems to treat children with hypoxaemia and respiratory distress [21] and perhaps could be considered at other hospitals with electricity supply problems.
The role of the nurses was comprehensive with regards to oxygen therapy, corresponding with the contents of the ‘Ugandan National Council for Higher Education Minimum Standards for Courses of Study for the Bachelor of Nursing Science’ [22]. However, this publication highlights use of oxygen funnels, tents and cylinders but does not discuss concentrators. This contrasts with the eastern Ugandan DGHs’ equipment availability and WHO recommendation [23]. National Ugandan Clinical Guidelines [24] also advocate the use of targeted oxygen saturations; however our study indicates this rarely occurs in practice. From an oxygen therapy perspective, national guidelines appear out-of-sync with the reality of Ugandan healthcare available in the DGH setting studied.
Patient/parent perceptions of oxygen in this setting are rarely reported on. Our study has highlighted the fear of oxygen amongst parents of paediatric patients and reflects a study by Stevenson and colleagues [25], who also noted the association between oxygen therapy and death by those interviewed. Education of local populations regarding the benefits of oxygen therapy may improve uptake and reduce fear of oxygen therapy.
Considering the reliance on signs and symptoms at Atutur and Bududa DGHs to establish hypoxaemia, it is concerning that signs and symptoms have been identified as poor hypoxaemia predictors. A study by Rao and colleagues demonstrated that, although poor, chest wall in-drawing, crepitation’s and nasal flaring were the three most sensitive predictors of hypoxaemia and cyanosis and nasal flaring had the best positive predictive value [26]. This partly corresponds with interviewee responses in our study, who cited ‘shortness of breath/ difficulty in breathing’, ‘cyanosis’, ‘chest in-drawing’ and ‘nasal flaring’ as signs indicative of oxygen need. A large prospective study in Kenya involving 15,289 paediatric admissions found 977 (6.4%) had hypoxaemia (SpO2 < 90%) measured by pulse oximetry [27] . In this study, the most predictive signs for hypoxaemia were compensated shock, impaired consciousness, bradycardia (a heart rate < 80 beats/minute) and alterations of normal respiratory pattern including irregular breathing and a respiratory rate > 60 breaths/minute.
Considering the variation in reliability of symptoms used to detect hypoxaemia, emphasis must be placed on routine oximetry use.
Study limitations
There were a number of limitations to this study. The selected sites were all government-funded district general hospitals in eastern Uganda and therefore external validity of results outside of this setting is questionable.
No formal cost analysis or detailed auditing process of oxygen supply and delivery systems was conducted due to difficulty accessing comprehensive hospital records containing this information.
No comprehensive assessment of power supply was undertaken in this study. However, triangulated insight from local healthcare professionals indicated that they could not rely on their current power supply. This highlights a specific area for quality improvement.
The authors’ involvement with oxygen systems in similar settings to the study prior to carrying out this study may have biased data collection and the interview approach. However, data collection adhered closely to qualitative data acquisition techniques, ensuring reliability and validity [28]. The Delphi technique and member checking would have a been useful techniques to establish consensus responses, strengthening conclusions and enhancing inter-disciplinary view-points [29], but this was not undertaken in our study.
There are greater numbers of coded passages from Pallisa DGH compared to Bududa and Atutur DGHs, biasing responses towards Pallisa DGH. The semi-structured interview design meant some aspects of the supply and delivery chain were not discussed with all interviewees, which may have been mitigated using a structured interview framework. However, this inductive approach allowed response freedom from participants, allowing them to focus on aspects important to them and therefore gave a truer reflection of local issues. These include teamwork, communication, parents’ oxygen fears and patient transfers to other facilities not previously highlighted by other large studies [3, 4, 6, 7].
The coding process in transcript analysis should ideally be undertaken by multiple coders to incorporate multiple opinions [14]. This study was coded by a single individual (JD), which may have increased bias towards expected responses.
Future studies assessing oxygen therapy should strive to understand local problems by interviewing providers of oxygen therapy. In addition, they should undertake a comprehensive site-specific cost analyses of running cylinders and concentrators over specific time periods coupled with reliability tests. This should be alongside assessment of back-up generators, fuel and cylinder transport costs. This will deduce the most suitable oxygen supply for a specific site, which may include cylinders, concentrators, a combination of the two or a local oxygen plant. Linking improvements in oxygen supply to changes in patient outcomes will help target quality improvement efforts, alongside other strategies that aim to reduce paediatric mortality in resource-limited settings [1].