Providing oxygen to children and newborns: a multi-faceted technical and clinical assessment of oxygen access and oxygen use in secondary-level hospitals in southwest Nigeria

Abstract Background Oxygen is an essential medical therapy that is poorly available globally. We evaluated the quality of oxygen therapy in 12 secondary-level Nigerian hospitals, including access to oxygen equipment, equipment functionality, healthcare worker knowledge and appropriateness of use. Methods We conducted a three-part evaluation of oxygen access and use involving: (1) facility assessment (including technical evaluation of oxygen equipment), (2) clinical audit (children and neonates admitted January 2014–December 2015) and (3) survey of healthcare worker training and experience on the clinical use of oxygen (November 2015). Results Oxygen access for children and newborns is compromised by faulty equipment, lack of pulse oximetry and inadequate care practices. One hospital used pulse oximetry for paediatric care. Eleven hospitals had some access to oxygen supplies. Testing of 57 oxygen concentrators revealed two (3.5%) that were ‘fit for use’. Overall, 14.4% (3708/25 677) of children and neonates received oxygen some time during their admission; 19.4% (1944/10 000) of hypoxaemic children received oxygen; 38.5% (1217/3161) of children who received oxygen therapy were not hypoxaemic. Conclusions Oxygen access for children in Nigerian hospitals is poor, and likely results in substantial excess mortality. To improve oxygen access for children globally we must focus on actual provision of oxygen to patients—not simply the presence of oxygen equipment at the facility level. This requires a systematic approach to improve both oxygen (access [including equipment, maintenance and affordability]) and oxygen use (including pulse oximetry, guidelines and continuing education).


Introduction
Oxygen is an essential medication that is important for treating hypoxaemia from pneumonia, other lung diseases and many other conditions (e.g. sepsis, meningitis, trauma and complications of prematurity) and situations (e.g. emergency, obstetric and perioperative care). 1 Global estimates suggest that at least 10-15% of children hospitalized with pneumonia, malaria or meningitis, and about 20% of sick neonates, have hypoxaemia. 2 Hypoxaemia is a major risk factor for death-increasing the odds of death fivefold among children with acute lower respiratory infection. 3 Improving the provision of oxygen to children in hospital can reduce inpatient childhood mortality by 10-20%, and pneumonia mortality by 20-40%. 4,5 However, despite high need, clear evidence of benefit, Int Health 2020; 12 : 60-68 doi: 10.1093/inthealth/ihz009 Advance Access publication 21 March 2019 and global recognition as an essential medicine, oxygen therapy is poorly available for many sick children and adults globally. 1,6 Providing oxygen to patients requires reliable access to medical oxygen and appropriate oxygen use. 5 Oxygen access can be limited by lack of equipment, poor equipment functionality or cost. 5 Good oxygen use requires prompt and accurate recognition of hypoxaemia, correct administration of oxygen therapy and regular monitoring, as well as treatment of the underlying condition. Pulse oximetry, a non-invasive method of measuring blood oxygen levels, is the preferred method to detect hypoxaemia (especially in resource-limited settings where blood gas analysis is rarely feasible), 7,8 as clinical signs of hypoxaemia are inaccurate. 9,10 Healthcare workers also need the skills to administer oxygen according to guidelines, and a work environment that ensures they give oxygen to the right patient, at the right time, every time. 5 Nigeria has a high under-5 y mortality rate (104.3 under-5 y deaths per 1000 live births), 11 and pneumonia is the number one killer of under-5 y children in Nigeria, responsible for about 140 520 (19%) under-5 y deaths in 2016. 12 The Nigerian Government has recognized child pneumonia as a priority, and has identified oxygen therapy as an important strategy to reduce deaths from childhood pneumonia. 13,14 This study aimed to evaluate the quality of oxygen therapy currently available to children and neonates in 12 secondarylevel Nigerian hospitals, including: (1) functionality of oxygen source(s); (2) appropriateness of oxygen use; and (3) healthcare workers' knowledge and experience using oxygen therapy.

Participants
We conducted our study in 12 secondary health facilities in southwest Nigeria, as part of a large field trial to improve oxygen provision to children (UTN U1111-1193-6364). 15 We selected sites that would be representative of medium-sized hospitals (government or mission) that regularly admit children (described in detail elsewhere). 15 Our inclusion criteria required hospitals to be secondary health facilities that admitted at least 150 children per year. Our initial focus was on government hospitals in Oyo state. When we could not achieve a sufficient number of hospitals, we decided to include hospitals from other states in southwest Nigeria that met the inclusion criteria. See Appendix 1 (supplementary data) for more detail on participating hospitals.

Procedures
Our study involved three parts: cross-sectional facility assessment (including technical evaluation of oxygen equipment); retrospective audit of oxygen-related care practices for children and newborns (2 y of data); and assessment of healthcare worker oxygen knowledge and experience.
In April 2015, a team of doctors, technicians and nurses conducted a cross-sectional assessment of each hospital's capacity to provide oxygen therapy to patients. We used a Facility Assessment Form (adapted from previously used WHO facility assessment tools) [16][17][18] to collect data on a broad range of structural, technical and clinical factors that may influence the provision of safe and effective oxygen therapy. The study team collected data during field visits from direct observation and interviews with key informants (including hospital directors, medical officers, nursing officers, technicians, medical records staff and other administrative staff). After completion of the field visit, the study coordinators collated the data and returned it to the medical director (or equivalent) for confirmation of accuracy, and performed follow-up visits and phone calls as necessary. Our technicians followed up the initial facility assessment with a more detailed technical assessment of the existing oxygen concentrators at each hospital, using a standardized data collection tool (Appendix 2) and calibrated oxygen analysers (Longfian Scitech, Baoding China; Cambridge-Sensotec, St Ives, UK). We also recorded informal feedback from technicians and hospital staff during all field visits.
We conducted a retrospective clinical audit of oxygen therapy for children using data extracted from patient case notes. Trained research nurses reviewed ward admission books and administrative records to identify all children (age <15 y) and neonates admitted during the study period (January 2014-December 2015). They extracted data from patients' case notes using a standardized case report form that had been successfully pilot-tested. We collected demographic and clinical data, including detailed data on oxygen use, diagnoses (using WHO case definitions) and outcomes.
In October and November 2015, we assessed healthcare workers' knowledge of and experience with the use of oxygen using a standardized written test (Appendix 3). This was administered before and after a basic training module on pulse oximetry, which was conducted at the time of distributing pulse oximeters for the first phase of the main field trial. 15 The test involved a series of Yes/No questions and a set of scenarios, designed to test basic knowledge and decision-making skills about oxygen therapy, and some questions on prior oxygen-related training.

Analysis
We present descriptive statistics for health facility characteristics, including the presence of oxygen equipment, associated guidelines and the functionality of oxygen concentrators. We defined a concentrator as being fit for use if: (1) it produces oxygen purity of at least 85%, meeting the WHO recommendation for medical oxygen 8 and (2) its electrical configuration is appropriate for use in Nigeria.
We report summary statistics to describe the study population, healthcare worker characteristics and the appropriateness of oxygen and pulse oximetry use. We used WHO definitions for hypoxaemia (SpO 2 <90%) and signs of hypoxaemia (any of fast breathing, head nodding, central cyanosis, respiratory rate ≥70 breaths per min, inability to drink due to respiratory distress and grunting with every breath). 8 We used data from the facility assessment to identify what proportion of oxygen was supplied from functional and non-functional concentrators, and combined this with clinical data to calculate the number of children who received oxygen from a concentrator producing substandard oxygen (<85% purity) (Appendix 4).
We report summary statistics of the healthcare workers who participated in knowledge tests and present descriptive statistics International Health Notes: neonate ≤28 d; child 29 d-15 y; i family medicine consultant; ii part-time; iii piped system connected to large oxygen cylinder; iv not available in paediatric areas; v present but not fit for use (see Table 2 for details); vi H6 was the only hospital routinely practising pulse oximetry; vii average daily cost per patient (individual hospitals variably charged per hour, per day, per patient, per cylinder), with conversion to US$ at 1 January 2015 exchange rate (184:1).
on healthcare worker knowledge, including mean scores with 95% confidence intervals. We report median and interquartile ranges for parameters that are not normally distributed. We used simple linear regression to look for associations between healthcare worker characteristics and experience, and score outcomes. We then used backward step-wise regression to build a multiple linear regression model that included the most relevant variables.

Oxygen access
Eleven of the 12 hospitals surveyed had some access to oxygen supplies at the time of assessment (Table 1). Four hospitals used oxygen cylinders alone. Eight hospitals used oxygen concentrators (three hospitals relied mostly on oxygen concentrators and five hospitals used both cylinders and concentrators). Our audit  International Health of oxygen practices revealed that staff at 10 of the hospitals had used oxygen for children or neonates during the 2014-2015 study period. Three hospitals had a pulse oximeter on site, but only a single hospital reported routine use of pulse oximeters for children. Our clinical audit confirmed that no other hospitals had been using pulse oximeters prior to our introduction of pulse oximeters in October-November 2015.
We tested a total of 57 oxygen concentrators across the eight facilities (Table 2). Hospitals had acquired the concentrators through donation from missionary and international agencies and/or through direct procurement from local suppliers. Procurement was usually handled by the hospital head/administrator or (less commonly) the State Ministry of Health. The majority of oxygen concentrators turned on and blew gas (50/ 57, 87.7%). Of these, most were blowing air (24/50, 48%) or concentrations of oxygen below medical oxygen standards (23/ 50, 46%). Overall, two (3.5%) concentrators were 'fit for use' (producing >85% oxygen and electrically compatible). No concentrators had a clean external filter (a marker of equipment maintenance).
We identified various technical problems, including faulty sieve beds (small cylindrical units inside oxygen concentrators that contain zeolite, which helps to remove nitrogen from the air), leaking humidifiers, faulty oxygen concentrator knob, no low-oxygen alarm, electrical faults and lack of routine preventive maintenance for oxygen equipment. Users reported that they did not know the concentrators were producing poor-quality oxygen, and expressed dissatisfaction and frustration with the current situation. Hospital staff (nurses, doctors, technicians and administrators) identified varied reasons for oxygen equipment failure (Box 1). Staff showed particular interest in how they could access better quality concentrators. Following our evaluation, hospital staff removed all the concentrators certified 'not fit for use' from clinical use.
Eleven of the 12 hospitals required patients to pay for oxygen (and other services and supplies). The median cost per day of oxygen was approximately 3800 Naira (range 0-19 200; US$ 20.65, 0-104.35).

Oxygen use
During 1 January 2014-31 December 2015, a total of 25 995 children (aged <15 y) were admitted to the 12 participating hospitals. The median age was 9 month, and approximately onethird were neonates (aged <28 d).
Overall, 14.4% (3708/25 677) of children received oxygen therapy at some point during their admission (Table 3a). Pulse oximetry was rarely used (with the exception of H6), with health workers relying on clinical signs to decide when to commence or stop oxygen therapy for patients. Overall, 19.4% (1944/10 000) of children who were hypoxaemic (SpO 2 <90% or signs of hypoxaemia if SpO 2 not recorded) on admission received oxygen (Table 3b). Hypoxaemic neonates and infants were more likely to be given oxygen than older children. Children with documented SpO 2 <90% were more likely to be given oxygen than those who had signs of hypoxaemia (134/192 69.8% vs 2011/10507 19.1%). Conversely, 61.5% (1944/3161) of children who received oxygen therapy had evidence of hypoxaemia (SpO 2 <90% or signs of hypoxaemia if SpO 2 not recorded), with negligible variation with age.
Hospital staff reported limited exposure to preservice training on pulse oximetry (approximately 32%) or oxygen use (52%), and pretraining assessment showed low knowledge scores (particularly relating to newborns) ( Table 4).
Multiple linear regression analysis showed that doctors scored higher than nurses (+4.1 points, 95% CI 2.7-5.7), and longer duration of service was positively associated with score (+0.08 patients per year, 0.02-0.35) (adjusted for age, sex, professional role and duration of service).
Based on clinical data, observation from research nurses and data from our technical assessment of oxygen concentrators (Appendix 4), we estimated that 43% (1595/3708) of children prescribed oxygen during 2014-2015 had been given substandard oxygen therapy (oxygen purity <85%). Overall, approximately 90% (9037/10 000) of children who had evidence of hypoxaemia did not receive appropriate oxygen therapy. We recorded 1105 deaths in the hypoxaemic cohort. Assuming that appropriate oxygen therapy could have reduced case fatality rates in this cohort by 20-40%, we estimated that lack of appropriate oxygen therapy may have contributed to 220-440 excess deaths over this 2 y period (of which 35-70 related to use of faulty equipment).

Discussion
This study provides insight into oxygen access and use in 12 secondary hospitals in southwest Nigeria and identifies key opportunities to improve oxygen systems in the country. Previous studies have identified deficiencies in oxygen equipment access, 6,19 equipment functionality 18,[20][21][22][23] or the clinical use of oxygen. 23 Surveys of healthcare workers in low-and middle-income countries have reported that oxygen and pulse oximeters were  available at approximately 75% and 60% of hospitals, respectively, with lower availability in smaller hospitals. 6,19 Other facility assessments reported lower availability of pulse oximeters (<10%) and identified major issues with non-functional concentrators. 18,[20][21][22] Few studies have looked at how oxygen is being used at the patient level. Multi-hospital studies in Laos and Malawi reported very low use of pulse oximetry for children with pneumonia (<5%) and the Malawi study found that healthcare workers provided oxygen to approximately 22% of eligible patients. 23,24 Data from Laos identified cost to patients as a major barrier to appropriate oxygen use. 24 Our study complements existing data by combining all these elements to provide a comprehensive evaluation of oxygen availability and oxygen use across multiple hospitals and a reasonably long time frame (2 y). Our findings are broadly consistent with previous studies, demonstrating that actual oxygen access is significantly poorer than might be suggested by the oxygen equipment that is available on-site. Despite most hospitals (11/12) having an oxygen source, only 20% of the children who warranted oxygen therapy received it.
Clearly, oxygen access must focus on actual provision of oxygen to patients-not simply the presence of oxygen equipment at the facility level. We identified four key barriers to the effective provision of oxygen to patients: low user knowledge and awareness about oxygen therapy (and pulse oximetry); reliance on clinical signs in the absence of pulse oximetry; lack of reliable oxygen source (e.g. concentrator, cylinder); and high cost to patients.
We were surprised to find how poorly many of the oxygen concentrators being used in hospitals were functioning-only 2 of the 57 concentrators tested were fit for use. These findings expand on previous reports on oxygen concentrator failures, 16,[20][21][22]25,26 and are particularly relevant given growing interest (and investment) in oxygen concentrators as cost-efficient sources of oxygen in Nigeria and elsewhere. 5,13,27 Our findings highlight the importance of selecting quality concentrators that can work in hot, humid, dusty conditions and building local maintenance capacity to monitor concentrator function and troubleshoot faults before they become irrepairable. 5,16,20,21,28 In particular, users need to be able to know whether their oxygen concentrators are providing oxygen at adequate purity using oxygen analysers. While many concentrators offer inbuilt oxygen concentration indicators (OCIs) as an option, none of the concentrators evaluated in our study had OCIs and none of the hospitals had access to oxygen analysers.
The WHO and PATH have recently released technical specifications for oxygen concentrators, 29,30 and independent assessments have found that most concentrators do not meet these specifications. 31,32 Given the new oxygen concentrators available on the market, 33 we need further independent testing to determine suitability for use.
When oxygen therapy is available, it is being poorly usedresulting in no oxygen for many who need it, and unnecessary oxygen use for some who do not need it. Our results complement findings from other studies showing that improved provision of oxygen to patients requires as much attention to oxygen use as it does to oxygen supply. 5 Healthcare workers need pulse oximetry to make rational decisions about oxygen therapy. Therefore, we must ensure that healthcare workers are In a user-pay environment, cost to patients may be a major barrier to appropriate oxygen use. Data from other studies suggest that addressing affordability is essential to improving the use of oxygen. 24 This study has some limitations. It represents a point evaluation of current oxygen systems at the hospitals and does not include any improvement or deterioration in practices postintervention. We assessed the appropriateness of oxygen use based on whether children with likely hypoxaemia on admission received oxygen, and without accounting for whether children were given appropriate flowrates for the full duration of time required (which may result in even lower estimates of appropriate oxygen use). We relied on clinical documentation of pulse oximetry and oxygen practices with a focus on the note written at the time of admission to enable correlation with clinical signs. This provided slightly lower proportions than including data from the whole admission (oxygen use 12.3% vs 14.4%, hypoxaemia prevalence 14.4% vs 17.4%). The retrospective nature of our chart review limited our ability to see what was not documented but also removed the risk of unintentionally biasing the healthcare worker. Our estimates of the number of children that received substandard medical oxygen and excess mortality was performed using facility-level data without knowledge of exactly what oxygen source was used for individual patients.

Conclusion
Oxygen access for children in the 12 secondary-level hospitals in southwest Nigeria is poor, and likely results in substantial excess mortality. To improve oxygen access for children globally we must focus on actual provision of medical oxygen to patients-not simply the presence of oxygen equipment at the facility level. This requires a systematic approach to both improve oxygen access (including equipment, maintenance and affordability) and oxygen use (including pulse oximetry, guidelines and continuing education). International Health