Table 2.

Oxygen availability, infrastructure, and usage and cost and cost-effectiveness

Author, year	Country	Facility type and setting	Data source	Key findings
Adipa, 2015 [12]	Ghana	Korle Bu Teaching Hospital – Emergency Department and ICU	Face to face interviews with health care workers (HCWs)	Oxygen was readily available via cylinders, but cumbersome tanks often had to be moved across wards for patient access.
Albutt, 2018 [13]	Uganda	17 public hospitals	Surgical Assessment Tool (SAT) developed by the Program in Global Surgery and Social Change and the WHO	Oxygen was available more than half the time at 68.8% of facilities while continuous pulse oximetry was always available in the operating room in only 62.5% of hospitals.
Albutt, 2019 [14]	Uganda	16 private and private not-for-profit hospitals	WHO Tool for Situational Analysis to Assess Emergency and Essential Surgical Care (TSAAEESC)	93.8% of hospitals had oxygen but only 37.5% of had it for more than 25% of the time.
Belle, 2010 [15]	Ethiopia, The Gambia, Ghana, Kenya, Liberia, Malawi, Mali, Nigeria, Sierra Leone, Sao Tome & Principe, Tanzania, Uganda	231 health centers and hospitals; 23% private, 38% primary care, 31% district or regional and 8% general hospitals	WHO TSAAEESC	43.8% of hospitals had a consistently available oxygen source; 31.4% had intermittent availability. Oxygen cylinders, concentrators and face masks and tubing were readily available in less than 35% of hospitals. Electricity and generator supply were similarly sporadic.
Bradley, 2015* [16]	The Gambia	Biomedical Engineering Department at the Medical Research Council Unit	Retrospective analysis of electronic maintenance records for 27 oxygen concentrators	The majority of concentrator faults were repairable for less than US$10, with the average costs of the most common repairs - filter and check valve replacements - US$4.53 and US$6.80 respectively. Median cost of repairs was US$9.44 with a maximum of US$573. The authors predict a seven year lifespan for concentrators in low resource settings with a US$15 per machine-year of service repair cost.
Bradley, 2016* [17]	The Gambia	42 bed hospital in the Medical Research Council Unit	Assessment of oxygen concentrator function and user feedback	The hospital system saved 51% of oxygen costs by using concentrators which amounted to US$45 000 over the course of 8 y.
Burssa, 2017 [18]	Ethiopia	29 facilities throughout Ethiopia	Ministry of Health Assessment	Continuous oxygen supply was not available in 33% of facilities and 59% had interrupted electricity. As part of their Safe Surgery initiative, 2 oxygen plants were built at referral hospitals.
Calderon, 2019 [19]	Uganda	Jinja Regional Referral Hospital	Prototype Assessment	A low-pressure reservoir storage system was able to deliver oxygen in 56% of power outage minutes and cover over 99% of power outage events.
Desalu, 2011 [20]	Nigeria	68 tertiary care hospitals	Cross-sectional survey	In the studied tertiary care centers, 52.9% reported having a standard oxygen delivery system. Under 40% had pulse oximeters.
Evans, 2012 [21]	Malawi	Queen Elizabeth Central Hospital (QECH) – 1200 bed public, teaching hospital in Southern Malawi	Cross-sectional study of adult medical inpatients and oxygen provision over 24 h	8concentrators were present, but only 4 functioned appropriately. Three had oxygen flow rates <60% of indicated; 1 did not function at all.
Hill, 2009 [22]	The Gambia	12 health facilities – 5 government referral hospitals, 7 health centers	Standardized WHO questionnaire	6 of the 12 facilities surveyed had available oxygen supplies (cylinders and concentrators). The government central referral hospitals generally had good reliability of supply whereas large health centers did not.
Howie, 2008 [23]	The Gambia	Royal Victoria Teaching Hospital (tertiary referral center)	Interviews with HCWs	Oxygen concentrators were donated from North America but required a different electrical frequency so were difficult to use. Very little training was given to providers and eventually the supply of concentrators went unused.
Howie, 2009* [24]	The Gambia	11 public hospitals	Health needs assessment framework	Concentrators have significant advantage compared to cylinders where power is reliable. In other settings cylinders preferred if transport is feasible. Cylinder costs are influenced by leakage whereas concentrator costs are affected by cost of power. Only 2 of 12 facilities in Gambia were suitable for concentrators over cylinders.
Kouo-Ngamby, 2015 [25]	Cameroon	12 public hospitals	WHO TSAAEESC	8 of the 12 hospitals had oxygen cylinder supply, but only 4 of the 12 had reliable oxygen concentrators. The greatest equipment needs were demonstrated in facilities providing lower tiers of care.
Kushner, 2010 [26]	Tanzania, Sierra Leone, Liberia, The Gambia	132 district hospitals in 30 LMICs	WHO TSAAEESC	No country reported 100% of facilities having continuous water, electricity and oxygen supplies. Oxygen was never available in 46% of facilities.
LeBrun, 2014 [27]	Ethiopia, Liberia, Rwanda, Uganda	78 district hospitals	WHO TSAAEESC	Approximately 80% of hospitals had a reliable oxygen source in the operating theater (OT) while 59% had pulse oximeters in each OT and 33% had them in surgical recovery rooms. Many of the hospitals surveyed reported power outages or interruptions.
Nyarko, 2016 [28]	Ghana	23 health facilities of various care levels; all received some if not all funding from the Ghanaian government	Facility-based survey to assess WHO- Package of Essential Noncommunicable Disease Interventions	None of the community-based health care facilities had functional oxygen cylinders or pulse oximeters while district and regional hospitals all did. Access to medications and medical equipment improved with increased levels of care.
Ologunde, 2014 [29]	Congo, Ethiopia, The Gambia, Ghana, Kenya, Liberia, Malawi, Niger, Nigeria, Sierra Leone, Somalia, Uganda, Tanzania	719 health facilities	WHO TSAAEESC	Oxygen was available in 73.3% of the facilities surveyed. 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.
Opio, 2014 [31]	Uganda	Kitovu Hospital – 220 bed private not-for-profit 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 that there was only 1 oxygen concentrator at the hospital in Uganda.
Ouedraogo, 2018 [32]	Senegal	All formal sector health facilities in the country	Senegal Service Provision Assessments	52% of health care facilities do not have access to regular uninterrupted electricity. Most of these facilities are connected to the central grid with only 18% of facilities using a generator or solar supply. 11% of facilities with oxygen concentrators do not have the electricity to power them.
Penoyar, 2012 [33]	Tanzania	48 public hospitals	WHO TSAAEESC	Of the facilities surveyed, which together serve 46% of the Tanzanian population, 42% had consistent access to oxygen delivery, most of which used concentrators. Only 37.5% had reliable running water and electricity. A total of 6 functional pulse oximeters were located across all 48 facilities.
Rassool, 2017* [34]	Uganda	Mbarara Regional Hospital (referral hospital)	Field testing of a low-pressure oxygen storage system	A low-pressure oxygen storage system designed by a team in Australia was able to provide continuous oxygen supply to a simulated patient without interruption for 30 d. The estimated cost of the system was US$460. The main drawback noted was the large amount of space the system occupies.
Rudd, 2017 [35]	Uganda	Bwindi Community Hospital – a private, 112-bed, rural hospital	Prospective observational single cohort study	Of 199 patients admitted to the hospital during the study period, 62 met SIRS criteria and were enrolled in the study. In the adult population, 44% of patients hypoxic to <94% O2 were treated with oxygen therapy; 100% of those hypoxic to <90% were. Only 1 of the patients studied died in the hospital while 92% of patients went home in improved condition.

*Article also addresses cost and/or cost-effectiveness.