Oxygen delivery systems for adults in Sub-Saharan Africa: A scoping review

Background Respiratory diseases are the leading cause of death and disability worldwide. Oxygen is an essential medicine used to treat hypoxemia from respiratory diseases. However, the availability and utilization of oxygen delivery systems for adults in sub-Saharan Africa is not well-described. We aim to identify and describe existing data around oxygen availability and provision for adults in sub-Saharan Africa, determine knowledge or research gaps, and make recommendations for future research and capacity building. Methods We systematically searched four databases for articles on April 22, 2020, for variations of keywords related to oxygen with a focus on countries in sub-Saharan Africa. Inclusion criteria were studies that included adults and addressed hypoxemia assessment or outcome, oxygen delivery mechanisms, oxygen availability, oxygen provision infrastructure, and oxygen therapy and outcomes. Results 35 studies representing 22 countries met inclusion criteria. Availability of oxygen delivery systems ranged from 42%-94% between facilities, with wide variability in the consistency of availability. There was also wide reported prevalence of hypoxemia, with most studies focusing on specific populations. In facilities where oxygen is available, health care workers are ill-equipped to identify adult patients with hypoxemia, provide oxygen to those who need it, and titrate or discontinue oxygen appropriately. Oxygen concentrators were shown to be the most cost-effective delivery system in areas where power is readily available. Conclusions There is a substantial need for building capacity for oxygen delivery throughout sub-Saharan Africa. Addressing this critical issue will require innovation and a multi-faceted approach of developing infrastructure, better equipping facilities, and health care worker training.

Respiratory diseases are the leading cause of death and disability worldwide. Oxygen is an essential, life-saving medical therapy that has been used to treat respiratory disease since the late 1800s [1,2]. It is used to treat both acute and chronic conditions which result in hypoxemia, which is an abnormally low concentration of oxygen in the blood. For adult patients with chronic hypoxemia from primary lung disease or heart failure, long term oxygen therapy is a well-established cornerstone of management, improving survival and quality of life [3,4].
Oxygen has been listed as one of the World Health Organization's (WHO) Essential Medicines since the first online edition in 2002, but only for anesthesia. Hypoxemia was added as an indication only recently in 2017 [5,6]. The simplest way to identify patients who are hypoxemic is by measuring oxygen saturation of the blood with a pulse oximeter, which uses infrared light refraction to non-invasively measure the percentage of oxygen in red blood cells. Supplemental oxygen can be provided via cylinders (gas or liquid), oxygen concentrators, or larger oxygen plants, each of which have unique advantages and disadvantages depending on environment and resource infrastructure. WHO guidelines on the clinical use of oxygen in children and technical specifications for oxygen equipment are available [7,8]. These highlight the need for pulse oximetry, appropriate clinical evaluation and management of underlying etiology, and basic administration standards. They notably do not specify guidelines for adults.
Despite widespread recognition of the importance of oxygen therapy for treatment of hypoxemia, its use and implementation remain inadequate in much of the world. Specifically, there is limited data about hypoxemia recognition and oxygen provision across sub-Saharan Africa (SSA), with most existing data focusing on acute oxygen needs among children and neonates. In order to design interventions and implementation efforts, a better understanding of the existing ecosystem is required, especially as it relates to adult populations.
In this scoping review, we aimed to identify and describe existing data around oxygen availability and provision for adults in SSA in order to determine areas of research or knowledge gaps on this topic and make recommendations for future research and capacity building. We hypothesized that the literature on adults would be limited as compared to pediatric populations, that what did exist would highlight the lack of availability, usage, and utilization of oxygen delivery systems for adults, and would identify key barriers of oxygen usage and opportunities for future research and capacity building.

METHODS
We used the Arksey and O'Malley methodological framework, along with Levac's recommendations for each stage of the framework, to perform a scoping review of what is known about oxygen delivery systems in SSA [9,10]. A review framework was prepared to develop the overall study protocol including identifying the research question, searching for relevant studies, selecting studies, charting the data, and collating, summarizing, and reporting the results.

Identifying the research question
The central research question of this scoping review is: what is known in existing literature about oxygen provision to adult patients with hypoxemia in SSA, what knowledge and research gaps exist, and what recommendations for future research and capacity building can be made?

Search strategy and selection criteria
We performed a systematic search of online databases (PubMed, EMBASE, African Index Medicus, and Web of Science) on April 22, 2020, with the assistance of a medical librarian. All articles from 1970-2020 were included. The review was conducted applying the search words "Africa," "oxygen inhalation therapy," "oxygen/ supply and distribution," "oxygen/therapeutic use," and "oxygen" with "domiciliary," "home," "therapy," "concentrator," "tank," "cylinder," "delivery," "distribution," or "supply," as the Medical Subject Heading (MeSH) headings ( Table 1).We included any study which included adults and addressed any of the following areas: hypoxemia assessment or outcome, oxygen delivery mechanisms, oxygen availability, oxygen provision infrastructure, and oxygen therapy and outcomes. Basic science studies, case reports, studies focusing only on neonates, children, or fetal outcomes among pregnant women, hyperbaric oxygen therapy, mechanical ventilation or oxygen in the setting of anesthesia, studies based wholly outside of SSA, and those for which no full article could be found were excluded. One investigator reviewed the titles and abstract of the studies for inclusion and an additional investigator assisted in full-text review for final inclusion. These two investigators also charted, collated, and summarized the data.

Charting the data, summarizing and reporting
We used a data collection form adapted from the Cochrane Effective Practice and Organization of Care group [11]. For each included study, we extracted data on study design, country or region of study, sample size, oxygen delivery device and patient interface, power source, population setting, facility, measured outcomes, and interventions where available.
The data were recorded in a series of tables, enabling repeated exploring of themes. After an iterative process of data extraction, recording, and thematic review, we aggregated studies into major themes and then collated the data to help describe the existing knowledge around oxygen availability and provision and analyze for key knowledge and research gaps.

Patient and public involvement
No patients nor members of the public were involved in the review.

Description of studies
Of 2424 papers screened, 173 full-text articles were assessed for eligibility (Figure 1). We ultimately included 35 articles describing 34 different studies.
A total of 22 SSA countries were represented in the studies (Figure 2), with 5 studies including data from multiple countries. Studies were published between 1995 and 2020 with data collection ranging from 1995 to 2017.
We identified three key themes among these articles, each with multiple sub-themes about the use of oxygen therapy in SSA ( Table 2 and Table 3). These themes are oxygen availability, infrastructure, and usage (n = 26), hypoxia assessment and clinical understanding and management (n = 24), and cost and cost-effectiveness (n = 4), and each are elaborated upon below. Eight articles contained two themes.   Oxygen availability, infrastructure, and usage Several of the reviewed studies used surveys to assess the available equipment in health care settings throughout SSA. These surveys showed a lack of basic equipment such as pulse oximeters and oxygen delivery systems, eg, cylinders or concentrators. Availability of oxygen delivery systems ranged from 42%-94% between facilities, with wide variability in the consistency of availability [13][14][15]18,20,22,24,25,28,30,33]. For example, the highest reported availability was in a study of private and public facilities in Uganda which showed that 15 of 16 facilities had access to oxygen at least half of the time. However, six of those hospitals lacked access to oxygen more than 25% of the time and one hospital never had access to oxygen [14]. The availability of pulse oximetry was significantly more limited, ranging from 0%-64% of facilities assessed [20,27,33]. This suggests that while facilities may have oxygen available, their ability to accurately identify patients who may need oxygen and titrate the amount delivered appropriately is limited. These limitations are especially true for public facilities and those providing lower tiers of care [13,14,28].
Three studies assessed infrastructure in low-and middle-income countries (LMICs) across the world; data specific to SSA was extracted from these studies and demonstrate similar findings [26,27,29]. These studies were focused on surgery and anesthesia capacity, but included assessments of the facilities at large and thus were included in this review. Kushner et al found 46% of facilities across eight LMICs (four of which were in SSA -Tanzania, Sierra Leone, Liberia, and The Gambia) never had oxygen available, 33% had it sometimes available, and 21% had it always available [26]. Likewise, a cross-sectional survey assessing cesarean-section delivery capacity across 26 LMICs (13 in SSA -Congo, Ethiopia, The Gambia, Ghana, Kenya, Liberia, Malawi, Niger, Nigeria, Sierra Leone, Somalia, Uganda and Tanzania) found that 21% of facilities performing cesareans reported not having a reliable supply of oxygen and that 26% of those referring out did not have any supply [29]. An additional study which compared in-hospital mortality among inpatients between an Ugandan hospital and Canadian hospital noted that there was only one oxygen concentrator available in a large regional hospital in Uganda [31].
Issues in oxygen availability extend beyond the availability of the oxygen delivery systems themselves. Delivering oxygen from a cylinder or concentrator to a patient requires basic equipment such as tubing to connect the system to a patient delivery device such as a face mask or nasal prongs. A notable multi-SSA country assessment demonstrated that not only do less than half of the facilities report access to an oxygen source, but that only 34.3% had at least one face mask and tube set available [15]. Furthermore, for oxygen concentrators to function, electricity must be readily available. This is a major issue for many facilities, with only 35%-68% having electricity fully available [15,27,32]. Backup power generators are often utilized in areas where consistent electricity may be lacking, but many facilities do not have access to functioning generators, relying instead on solar power [15,27,30,32]. Even in places where oxygen concentrators and power are available, many do not function properly or provide the indicated amount of oxygen [21,23].
There are promising strides in systematically determining the best approach to providing oxygen and developing innovative oxygen delivery and storage systems in under-resourced areas. Oxygen concentrators have been demonstrated to be much more cost-effective than cylinders for oxygen delivery without sacrificing medical benefit [24]. Replacing cylinders with oxygen concentrators and addressing the issue of reliable power by installing an uninterrupted power supply has been shown to be easy to maintain and cost-effective [17]. Oxygen storage systems or reservoirs, which store oxygen in low-pressure devices, have been effective in maintaining flow regardless of interruption in power supply for up to 30 days and reducing the number of oxygen outage events [19,34]. While there are limitations to large-scale implementation of these types of reservoirs including clinical trial data, space requirements, and significant need for personnel education, innovations such as these will be imperative in overcoming the "energy poverty" that challenges SSA [32]. At facilities that offer cesarean delivery, 78% had access to oxygen whereas only 21% of facilities that did not offer the procedure had available oxygen. The survey did not distinguish between cylinders and concentrators.

Otiangala, 2020 [30] Kenya 11 rural health facilities Key informant interviews
Of 11 facilities surveyed, 82% had at least one cylinder and at least one concentrator. A back-up generator was available at 64% of facilities. The study also found a high case fatality rate in hypoxemic patients with an upward trend in mortality in those who experienced interruptions in therapeutic oxygen supply.

Prospective data
This study compared prospective data from Uganda to retrospective data from Canada to compare in-hospital mortality in patients with similar admission characteristics. It noted that there was only 1 oxygen concentrator at the hospital in Uganda.

Retrospective clinical audit
This study assessed the compliance with traumatic brain injury (TBI) management criteria, one of which is oxygen therapy. Only 21% of patients 13yrs or older (mean age 29.8) with TBI were administered oxygen as indicated. The authors speculate that lack of availability is a significant factor.

Opio, 2014 [31] Uganda
Kitovu Hospital -220 bed private not-forprofit regional hospital Prospective data This study compared prospective data from Uganda to retrospective data from Canada to compare in-hospital mortality in patients with similar admission characteristics. It noted 3.9% of patients were on supplemental oxygen vs 31.2% of patients in Canada. There are also issues in the way in which concentrators are provided. In 2000, a group of North American philanthropists donated more than 20 oxygen concentrators to the Royal Victoria Teaching Hospital, The Gambia's tertiary referral hospital. However, within weeks, none of the concentrators remained in use. A technical and qualitative assessment found that the donation process was flawed, the receiving personnel were not adequately trained in the use or maintenance of the machines, and the electrical frequency of the devices was different than the hospital's electrical supply and would never have worked, even with a transformer [23].

Hypoxemia assessment and clinical understanding and management
An understanding of the prevalence of hypoxemia is critical in ensuring that the supply of oxygen in areas where it may be available is truly meeting the demand. Furthermore, medical staff and providers should have a clear understanding of the indications of oxygen, the appropriate route and titration of oxygen therapy, and when to discontinue it. Ten studies addressed the question of prevalence of hypoxemia and demonstrated variability in the assessment of hypoxemia across institutions and countries, as well as the lack of generalized data with most focusing on very specific populations. An additional five provided insight on the understanding of oxygen therapy and management of hypoxemic patients in SSA.
Overall prevalence of hypoxemia, as measured by handheld pulse oximetry among adults in a district hospital in Zambia was 9%, among adult inpatients in Malawi was 9%, and among adults in the emergency department at a teaching hospital in Rwanda was 12.1% [21,38,44]. The study comparing a Ugandan hospital and Canadian hospital assessed oxygen saturation but did not remove oxygen supplementation at the time of measurement, making their values difficult to interpret [31].
Five studies assessed hypoxemia among specific populations, ranging from those with cor pulmonale to sepsis to Ebola Virus Disease. The heterogeneity in patient populations makes it difficult to discern the overall need in various settings, but prevalence ranged from 11%-89% [35,37,40,45,46]. Several of these studies demonstrated high mortality among those with hypoxemia and an additional two studies focused specifically on hypoxemia as a risk factor for mortality [36,42]. For example, among hospitalized patients who presented with lower respiratory tract infection and cough at Mulago Hospital in Uganda, 10.6% had an oxygen saturation of <90%, and hypoxemia was associated with both in-hospital and two-month mortality [46]. Among patients with Ebola Virus Disease in West Africa, 18 of 41 (44%) patients required and received supplemental oxygen via concentrators, and 16 of 18 (89%) died [37]. The highest prevalence was seen among 34 patients with chronic cor pulmonale in Senegal where 88.8% were hypoxemic [45]. Unsurprisingly, oxygen saturation on admission was found to be a predictor of mortality among patients with heart failure across sub-Saharan Africa, [42] adults admitted for community-acquired pneumonia in Malawi, [36] and associated with mortality among hospitalized patients managed for chronic obstructive pulmonary disease (COPD) in Nigeria [47].
In order to ensure understanding and management of hypoxemia, there is a critical need for equipment such as pulse oximetry, more training in identifying and responding to hypoxemia with oxygen, and standardized protocols to guide initiation, titration, and discontinuation of oxygen therapy. Even when hypoxemia was identified, oxygen was not always provided or used appropriately with only 27%-29% of patients with hypoxemia receiving oxygen in two different studies [21,40]. Similarly, compliance with oxygen therapy in traumatic brain injury patients was found to be "dismal" [41]. An assessment of first year doctors in Ghana found that 28% prescribed oxygen for mild to moderate asthma and 74% for severe asthma, with many unaware that there was some degree of hypoxemia even with mild asthma [39]. A qualitative assessment in Ghana found nurses had a lack of knowledge about the appropriate amount of oxygen to provide and when to discontinue oxygen, and they strongly desired standardized protocols [12].
Efforts are being made to improve these metrics as highlighted by a quality improvement project in which daily team meetings and monthly feedback sessions improved supplemental oxygen administration for hypoxia from 75% to 92% among pre-hospital care providers in Rwanda [43].

Cost and cost-effectiveness
Four articles addressed the cost and cost-effectiveness of various oxygen delivery mechanisms, three of which were based on projects in The Gambia [16,17,24,34]. Oxygen concentrator systems with an uninterruptable power supply could save up to 51% of oxygen supply costs as compared to cylinders, amounting to a total cost of only US$45 000 over the course of eight years [17]. However, the cost-advantage only applies in areas where power is reliable, which as previously described is a major barrier for many facilities [24]. Fortunately, innovations such as the low-pressure oxygen storage system tested in Uganda cost as little as US$460 and generate little extra electricity costs [34]. Concentrators can also be properly maintained in resource-limited settings with most concentrator faults repairable for less than US$10 on average [16].

DISCUSSION
Our scoping review of available studies reveals that there is a widespread lack of access to and infrastructure for oxygen delivery systems across SSA, and that while this is a relatively understudied area with limited literature, there are several key opportunities to address this issue. It is noted that the majority of studies on oxygen therapy in SSA focus on neonates and children, which is understandable given the high mortality from pneumonia for children under five years. However, respiratory diseases remain the leading cause of death and disability in the world, with diseases such as COPD, asthma, pulmonary hypertension, lung cancer, and tuberculosis killing millions of adults annually. Although efforts are being made to address prevention of these diseases, huge strides are needed to reduce the burden of these diseases and resulting mortality among adults [1]. Availability of and access to oxygen therapy to treat patients with acute or chronic hypoxemia from these diseases is paramount.
Our review of the available studies involving adults demonstrates a dire lack of access to oxygen delivery systems across SSA and that most facilities are ill-equipped to identify adult patients with hypoxemia, provide oxygen to those who need it, and titrate or discontinue oxygen appropriately. Data are limited to mainly surveys, assessments, and observational studies, which cannot be validated, and the simple existence of an oxygen delivery system cannot be directly correlated with patient outcomes. Yet hypoxemia is still associated with significant mortality in many adult populations, and the ability to address it is limited in many care settings. Together, our findings highlight a number of important limitations and opportunities in addressing a critical health issue throughout SSA.
First, oxygen must be made more readily available and accessible at health care facilities providing care to adults, with emphasis being placed on public and lower-tier health centers. Stakeholder engagement is key to this process and needs assessments should be done to ensure facilities and communities are involved in decision-making. This will require a concerted effort by national, regional and local governments, ministries of health, policy experts, health care workers, and health facility leadership to identify the appropriate oxygen delivery system for each setting and to ensure adequate resources are available for the maintenance of these systems.
Our analysis shows that concentrators are more cost-effective than oxygen cylinders in areas where there is reliable access to power. Innovation around lowering the cost of devices and providing reservoirs during outages may be helpful, and there will need to be a significant focus on building infrastructure around reliable power. Projects in resource-limited settings outside of SSA may be helpful in this regard. For example, studies in Papua New Guinea describe the process and effectiveness of an oxygen program which included provision of pulse oximeters, training of staff and installation of oxygen concentrators, as well as the design and feasibility of a solar-powered oxygen system [48,49]. Similarly, trials focused on improving access to oxygen among pe-diatric populations in SSA, especially those exploring solar-power as an energy source, may provide the data and basic infrastructure needed to improve access for adults [50][51][52]. Innovations such as these are critical in facilitating oxygen availability in lower-tier health centers.
Second, health care workers need appropriate equipment, education, training, and feedback in order to use oxygen appropriately and effectively. Identifying patients who need oxygen therapy is limited by a lack of pulse oximeters and knowledge, and appropriately initiating, titrating, and discontinuing oxygen therapy is limited by a lack of knowledge and training among health care workers at all levels. Pulse oximeters should be in place at every health care facility. They are an easy-to-use and relatively affordable method of identifying patients at greatest risk of mortality. Oxygen tubing and patient delivery devices such as nasal prongs or face masks must be readily available, and a steady supply chain must be maintained. Education and training of health care workers around oxygen saturation, hypoxemia, and effective oxygen dosing can ensure that resources are being used efficiently [44].
A challenge in administering oxygen not yet addressed in this review is the oxygen saturation threshold below which oxygen should be provided in various settings and among various populations. Most included studies used an oxygen saturation below 90% as their threshold. This is reasonable for most patients with acute hypoxemia and is the threshold recommended in a WHO manual on oxygen therapy for children [53]. However, for patients with chronic hypoxemia from primary lung disease or for populations living at higher altitude where the partial pressure of oxygen is lower, a lower threshold may be more appropriate. These populations often have physiologic compensations allowing them to tolerate lower oxygen levels; providing excess oxygen may in fact cause harm. Therefore, education and training has to be contextualized for each setting and practitioners need to be capable of individualizing therapy for each patient.
Third, there were no studies assessing the prevalence of hypoxemia among outpatients and the availability of long-term oxygen therapy for home use. If we extrapolate from the severe limitations in hospital settings, we can infer this infrastructure is severely limited. In 2017, over 544 million people had a chronic respiratory disease, representing an increase of nearly 40% since 1990 [54]. Prevalence in SSA is likely underestimated due to a lack of diagnostic capabilities, but is expected to grow as life-expectancy increases in many countries [55]. Any efforts to reduce disability or mortality from chronic respiratory disease that do not include building infrastructure for long-term oxygen therapy will fall short given what is known about its mortality and quality of life benefits [3,4].
Finally, while this review was conceptualized and undertaken prior to the COVID-19 pandemic, it is impossible to ignore the disparities in resource allocation that the pandemic has underscored. Articles in the popular media have highlighted the enormous need for oxygen in countries across the world, many in SSA, as well as the incredible barriers in accessing it [56,57]. This enhanced awareness has facilitated increased work and innovation in this area [58]. As donations of oxygen concentrators pour in from international organizations and aid agencies, it will be important to ensure that aid extends past simple provision of these systems to maintenance, training, and local capacity building.
The major strength of our scoping review is its comprehensive scope and wide inclusion of studies addressing oxygen delivery systems in various ways. We aimed to evaluate the depth and breadth of knowledge and research on oxygen delivery systems in SSA and were able to summarize the evidence in this field. Limitations of our study include that this is not a systematic review, and therefore we cannot aim to assess the quality of articles or make definitive inferences; similarly, we cannot aim to assess the risk of bias given the descriptive nature of our objectives and the types of studies presented which were mainly surveys and observational data; by limiting our review to full-text articles, we may have missed relevant data available in abstracts; and included studies encompassed only 22 of 46 SSA countries so may not be representative of the general population or region.
In conclusion, our findings highlight the substantial need for further research and building capacity for oxygen therapy for adults across SSA and signals a call to action. We provide multiple potential action items for health care workers, researchers, policy makers, and organizations to consider as we move towards improving the care of and outcomes among patients with respiratory diseases.