INT J TUBERC LUNG DIS 20(5):696–703

Q 2016 The Union

http://dx.doi.org/10.5588/ijtld.15.0796

Solar-powered oxygen delivery: proof of concept

H. Turnbull,* A. Conroy,† R. O. Opoka,‡ S. Namasopo,§ K. C. Kain,‡¶#**†† M. Hawkes*‡‡§§

*Department of Paediatrics, University of Alberta, Edmonton, Alberta, †Department of Medicine, University of
Toronto, Toronto, Ontario, Canada; ‡Department of Paediatrics and Child Health, Mulago Hospital and Makerere
University, Kampala, §Department of Paediatrics, Jinja Regional Referral Hospital, Jinja, Uganda; ¶Institute of
Medical Sciences, University of Toronto, Toronto, #Sandra A Rotman Laboratories, McLaughlin-Rotman Centre for
Global Health, Toronto, **McLaughlin Centre for Molecular Medicine, Toronto, ††Tropical Disease Unit, Toronto
General Hospital, Toronto, Ontario, ‡‡School of Public Health, University of Alberta, Edmonton, §§Department of
Medical Microbiology and Immunology, University of Alberta, Edmonton, Alberta, Canada

SUMMARY

SETTING: A resource-limited paediatric hospital in
Uganda.
O B J E C T I V E : Pneumonia is a leading cause of child
mortality worldwide. Access to life-saving oxygen
therapy is limited in many areas. We designed and
implemented a solar-powered oxygen delivery system
for the treatment of paediatric pneumonia.
D E S I G N : Proof-of-concept pilot study. A solar-powered
oxygen delivery system was designed and piloted in a
cohort of children with hypoxaemic illness.
R E S U LT S : The system consisted of 25 3 80 W photo-
voltaic solar panels (daily output 7.5 kWh [range 3.8–
9.7kWh]), 8 3 220 Ah batteries and a 300 W oxygen
concentrator (output up to 5 l/min oxygen at 88%
[62%] purity). A series of 28 patients with hypoxaemia
were treated with solar-powered oxygen. Immediate
improvement in peripheral blood oxygen saturation was
documented (median change þ12% [range 5–15%], P ,
0.0001). Tachypnoea, tachycardia and composite illness
severity score improved over the first 24 h of hospital-
isation (P , 0.01 for all comparisons). The case fatality
rate was 6/28 (21%). The median recovery times to sit,
eat, wean oxygen and hospital discharge were respec-
tively 7.5 h, 9.8 h, 44 h and 4 days.
C O N C L U S I O N : Solar energy can be used to concentrate
oxygen from ambient air and oxygenate children with
respiratory distress and hypoxaemia in a resource-
limited setting.
K E Y W O R D S : pneumonia; oxygen; solar energy; hyp-
oxaemia; child

PNEUMONIA IS A LEADING CAUSE of child
mortality worldwide, causing over 900 000 deaths
per year in children aged ,5 years.1 Countries in
Africa and Asia report 2–10 times more cases of
pneumonia than industrialised countries, and the
majority of global pneumonia deaths.2–4 In Uganda
alone, 21 000 children aged ,5 years die of pneumo-
nia every year.5 The World Health Organization’s
(WHO’s) 2013 integrated global action plan for the
prevention and control of pneumonia and diarrhoea
aims to end preventable deaths from pneumonia and
diarrhoea by 2025.6 This mandate includes providing
90% access to essential pneumonia treatment with
medications and oxygen therapy.
Hypoxaemia is prevalent among children presenting
with lower respiratory tract infection such as bacterial
pneumonia, viral respiratory tract infections and
tuberculosis.7,8 In a recent systematic review of more
than 16 000 children in developing countries with
acute pneumonia or other lower respiratory tract
infections, the median prevalence of hypoxaemia
(pulse oximeter oxygen saturation [SpO2] ,90%) of
children with severe pneumonia requiring hospitalisa-
tion was 13.3%.9 Hypoxaemia is a risk factor for
morbidity and mortality in paediatric pneumonia,
associated with a five-fold increased risk of death.10,11
In resource-constrained settings, oxygen delivery
systems can lead to measurable improvements in
survival from childhood pneumonia. A study in Papua
New Guinea demonstrated a reduction in mortality
from childhood pneumonia from 5.0% to 3.2% (35%
reduction in mortality) after implementation of an
enhanced oxygen delivery system.12 Improving and
expanding oxygen delivery systems could save up to
122 000 children’s lives per year from pneumonia.13
Oxygen is a lifesaving therapy for children with
pneumonia and hypoxaemia; however, challenges
remain in oxygen delivery globally. Two main systems
of oxygen delivery have been implemented and
evaluated in resource-constrained settings: oxygen
cylinders and oxygen concentrators. Oxygen cylinders
are ready to use, simple to operate and do not require

Correspondence to: Michael Hawkes, Division of Infectious Diseases, Department of Paediatrics, University of Alberta, 3-
588D Edmonton Clinic Health Academy, 11405 87 Avenue NW, Edmonton, AL, Canada T6G 1C9. Fax: (þ1) 888 790 1176.
e-mail: mthawkes@ualberta.ca
Article submitted 22 September 2015. Final version accepted 30 November 2015.

Figure 1 Diagram of SPO2 delivery system. Numbered system
components are described in detail in the text, and included 1)
solar panels installed on the roof of the hospital ward; 2) a
charge controller to optimise power output from the array of
solar panels; 3) a bank of batteries to store the energy; 4) a
DCAC inverter to operate the concentrator; 5) a commercially
available oxygen concentrator; and 6) oxygen stream delivered
to patients via nasal cannula or mask. DC ¼ direct current; AC ¼
alternating current; SPO2 ¼ solar-powered oxygen.
any electricity. However, cylinders are very costly, and
their distribution and use is challenging.14 Oxygen
concentrators have proven an effective means of
delivering oxygen and are significantly less expensive
than cylinders.15–21 Nevertheless, to operate, oxygen
concentrators require continuous and reliable electric-
ity, which is not readily available in many regions,
particularly in resource-limited settings, where the
majority of pneumonia deaths occur.22
To meet the demand for oxygen therapy in
resource-limited settings, in this study we investigate
a novel strategy: solar-powered oxygen delivery
(SPO2). This system uses free input (sun and air)
and can be implemented in remote locations with
minimal access to an electrical power supply. Few
studies have reported on the use of solar-powered
oxygen delivery. One study from the Gambia has
described the cost and practical experience of
extending an existing hospital-wide solar system used
for lighting and support for other electrical equip-
ment to run an oxygen concentrator.23
The objective of the present study was to design
and implement an SPO2 delivery system at a resource-
limited hospital in Uganda for the treatment of
paediatric pneumonia. Here we present the system
characteristics and the results of a pilot study of 28
children with hypoxaemia treated with SPO2.
STUDY POPULATION AND METHODS
Setting
The study was performed at Jinja Regional Referral
Hospital (JRRH), Jinja, Uganda. JRRH serves six
districts in mid-eastern Uganda under severe resource
limitations. The SPO2 system was installed in a four-
Solar-powered oxygen 697
bed paediatric intensive care unit (ICU) staffed by a
dedicated study nurse 24 h per day.
System design
The key performance goal of the system was to
produce oxygen reliably and continuously (purity
.85% at a flow rate of 5 l/min, adequate for the
treatment of up to two children simultaneously) using
solar power as the only electricity source. The
components selected to meet these goals included
photovoltaic panels to generate electrical energy from
sunlight using the photo-electric effect, a battery bank
to store the energy, a commercially available oxygen
concentrator to purify oxygen from the ambient air
and the electrical circuit accessories (charge control-
ler, current inverter) to optimise the power harnessed
and link the components compatibly.
The SPO2 system consisted of the following
commercially available products: 25 3 80 W solar
panels (Solarworld, Freiburg, Germany), a charge
controller (FLEXmax 80, OutBack Power, Arlington,
WA, USA), a bank of 8 3 220 Ah batteries (Ultra-
Power gel batteries, Victron Energy BV, Almere
Haven, The Netherlands), a DCAC current inverter
(Victron Energy BV, Almere Haven, The Nether-
lands) and a 300 W oxygen concentrator (model 525
KS, DeVilbiss Healthcare LLC, Somerset, PA, USA).
A diagram of the system is provided in Figure 1. The
system components were purchased from and in-
stalled by Ugandan electrical engineers and techni-
cians (Ultratec [U] Ltd, Kampala, Uganda).
For the purposes of our study, which involved
experimental methods for oxygen delivery, the ICU
was also equipped with cylinder oxygen as a failsafe
backup electrical supply with a fuel-powered gener-
ator.
System characteristics
The available hydroelectric grid power was moni-
tored continuously over a representative 2-month
period using a data logging multimeter (Fluke 287
Multimeter, Montreal, QC, Canada) connected to the
hospital circuit. The power generated by solar panels
was measured hourly over a representative 7-day
period (Fluke 287 Multimeter). The maximum
duration of the battery bank was tested by fully
charging the bank, disconnecting all input current
(solar panels and hydroelectric grid) and running the
oxygen concentrator continuously at maximum flow
rate (5 l/min) until failure. The power consumed by
the concentrator (load) was measured at the begin-
ning and end of oxygen therapy for all patients
enrolled in the pilot study described below (Fluke 287
Multimeter). As a quality control measure, the
oxygen content of the output stream from the
concentrator was measured at the beginning and
end of oxygen therapy for all patients enrolled in the
pilot study, using an oxygen analyser (Guangdong
698 The International Journal of Tuberculosis and Lung Disease
Figure 2 Electric power output from hydroelectric grid
measured continuously over a 64-day period (October–Decem-
ber 2014) at Jinja Regional Referral Hospital, Jinja, Uganda.
Frequent losses in power supply cause challenges to oxygen
delivery by concentrators. AC ¼ alternating current.
Pigeon Medical Apparatus Co, Canton, China), for
each patient in the pilot study.
Proof-of-concept pilot study
Paediatric patients presenting to JRRH with respira-
tory illness and hypoxaemia (SpO2 , 90%) were
recruited. Study inclusion criteria were age ,5 years;
pneumonia, severe pneumonia or very severe disease
as defined in the Integrated Management of Child-
hood Illness;24 and hypoxaemia (SpO2 , 90%),
measured using a portable pulse oximeter (Masimo
Rad-57e, Masimo Corporation, Irvine, CA, USA).
Pneumonia, severe pneumonia or very severe disease
was defined as cough or difficulty breathing, with
chest indrawing or any of the following danger signs:
history of convulsions, inability to feed, vomiting
everything, or lethargy or unconsciousness.24 Patients
were excluded if there was clinical suspicion of
pulmonary tuberculosis (infection control hazard),
lack of essential supplies in the hospital (e.g., blood
for transfusion) requiring urgent referral or refusal of
parental consent. All patients received standard care
for their underlying disease according to WHO
recommendations. This study also ensured standard-
ised care for patients with respect to oxygen therapy,
including quality-assured devices, well-maintained,
consistent power and quality single-use equipment
for oxygen delivery (masks, nasal prongs).
Measures of effectiveness included changes in vital
signs, composite disease severity score, recovery times
and outcome. Peripheral blood oxygen saturation
and vital signs were measured before and every 4 h
during the administration of supplemental SPO2.
Need for oxygen was assessed by a standardised
protocol at least once daily. Each patient’s peripheral
oxygen saturation was monitored for at least 15 min
from the removal of the oxygen-delivering device
(nasal prongs, mask). If SpO2 dropped below 90% on
room air, oxygen therapy was restarted. If SpO2
remained above 90% for at least 15 min, the oxygen
was discontinued, according to WHO guidelines.
Once off oxygen, the child’s SpO2 was then rechecked
every 4 h until discharge from the hospital. In
addition, a validated composite severity score, Signs
of Inflammation in Children that Kill (SICK),25 was
computed at each observation period (every 4 h
during hospitalisation).
Patient outcomes were categorised as death,
discharge without disability, discharge with disability
and transfer to another facility. The length of stay
among survivors was determined systematically. Prior
to discharge, the patient had to meet all of the
following standardised criteria: SpO2 over 90% for
24 h breathing room air, afebrile for at least 24 h,
eating or breastfeeding well (not requiring intrave-
nous fluids), stable vital signs and absence of
respiratory distress.
Analysis
Data were analysed using descriptive statistics,
including Kaplan-Meier survival curves for time-to-
event analysis. GraphPad/PRISM software, version
5.0 (GraphPad Software, La Jolla, CA, USA) was
used for data analysis. Longitudinal improvements in
vital signs and disease score were modelled using
linear mixed-effects models (R version 3.0.1, R
Computing, Vienna, Austria).
Ethics
The Makerere University School of Biomedical
Sciences Research and Ethics Committee (Kampala),
the Research Ethics Board of the University of
Toronto, Toronto, ON, Canada, and the Uganda
National Council for Science and Technology (Kam-
pala, Uganda) approved the study.
RESULTS
Adequacy of local power supply
An alternative power supply for oxygen concentra-
tors is necessary in situations where the grid supply is
unreliable. We recorded 120 episodes of power failure
(measured output ,200 V lasting for .5 min) during
continuous monitoring of the local power supply over
a 64-day period between October and December
2014 at the JRRH. There was an average of 13
episodes of power failure per week. The median
duration of the power failures was 30 min (range 5
min to 17.5 h); the cumulative proportion of time
without adequate power supply during the measure-
ment period was 10% (Figure 2).
Characteristics of the solar-powered oxygen delivery
system
The system was designed to run continuously
(including at night time and on cloudy days) using
solar energy alone, without any hydroelectric power
input (24/7 ‘off the grid’ operation). The cost of the
system was ~US$18 000. Measured output from the

Figure 3 Solar power and oxygen purity of SPO2 delivery
system. A) Output (power ¼ voltage x current) measured from
solar panel array vs. time of day over a continuous period of 7
days. As expected, array output was highest during sunny
daytime hours, and varied according to the angle of incident
sun. B) The oxygen generated was measured at the beginning
(‘start’) and end (‘stop’) of oxygen therapy for 28 hypoxaemic
children treated with SPO2 over a 5-month period in a pilot
study. SPO2 ¼ solar-powered oxygen.
array of 25 80 W photovoltaic panels varied over the
course of the day, as expected, according to the angle
of incident sun (Figure 3A). The daily energy
generated was median 7.5 kWh (range 3.8–9.7).
Compared to the theoretical maximum daily energy
output (80 W per panel, adjusted for changing the
angle of incident sun), the array efficiency was
median 50% (range 27–66%). Periods with visible
cloud and/or rain were associated with lower array
efficiency (median 17%, range 0–43%) compared to
sunny periods (median 55%, range 11–96%, P ,
0.0001). The battery bank could run the concentrator
at maximum output (5 l/min) for 69 h without an
external power source (solar panels and hydroelectric
grid disconnected).
Measured throughout the period of the pilot study,
mean power consumption by the oxygen concentra-
tor (load) was constant, at 320 W (standard deviation
[SD] 660 W). Power consumption did not vary
according to the flow rate of oxygen generated within
the manufacturer-specified range (0–5 l/min). The
concentrator delivered up to 5 l/min of oxygen, and
could be used for up to two patients simultaneously.
Measured throughout the period of the pilot study,
the oxygen purity was constant, at mean 88% (SD 6
2.6) (Figure 3B). The ambient temperature and
humidity in Jinja during the pilot study period were
Solar-powered oxygen 699
mean 228C (range 11–328C) and 78% (14–100%),
respectively.26
Participant information
We piloted the solar-powered oxygen delivery system
among 28 children presenting to JRRH with hypox-
aemic respiratory illness between 10 September 2013
and 18 February 2014. Patient characteristics at
admission are shown in the Table.27,28 Health care
providers and users of the SPO2 delivery system were
six Ugandan nurses and one Ugandan medical officer.
They had no prior experience with solar-powered
oxygen delivery or electrical systems generally, but
had prior experience with oxygen concentrators. The
electrical wiring and equipment were installed such
that users did not need to perform any steps other
than turn on the oxygen concentrator. Anecdotally,
they found the system easy to use, particularly as it
required no knowledge or steps to operate other than
the use of the concentrator.
Solar-powered oxygen treatment
Patients were treated with oxygen for a median of 2.5
days (range 7 h–7 days) until standardised weaning
criteria were met. The median power consumption of
the concentrator over the period of hospitalisation
was 15 kWh (range 2.6–48) per patient. In addition to
oxygen, patients received antibiotics (ceftriaxone in
25/28, 89%), antimalarials (6/28, 21%) and antipy-
retics.
Equipment failure occurred in one instance. The
system failure occurred during a period of heavy
rainfall in Jinja, when cloud cover was substantial.
Two patients were using the concentrator simulta-
neously (total flow rate 5 l/min). At the time of the
system failure, the concentrator had been running for
12 days (approximately 850 h) consecutively. The
cause of the system failure was depletion of the
battery bank due to low solar energy input (cloudy
skies) and high demand (12 days continuous opera-
tion of the concentrator). The action taken was to
switch the concentrator to grid power to continue to
provide oxygen to the patients and to recharge the
battery bank from the hydroelectric grid.
Course in hospital and outcome
We documented immediate improvement in periph-
eral blood oxygenation in patients receiving solar-
powered oxygen. Median oxygen saturation in-
creased by 12% (interquartile range 5–15%), com-
paring measurements taken before (median 30 min)
and after (median 45 min) initiating oxygen (Figure
4A). Administration of SPO2 was associated with
rapid normalisation of hypoxaemia and vital sign
abnormalities (tachypnoea and tachycardia). Exam-
ples of vital sign profiles of selected patients receiving
SPO2 are given in Figure 5A–C. The severity of illness
(SICK) score25 decreased over the course of hospital-
700 The International Journal of Tuberculosis and Lung Disease

Table Characteristics of study participants 4C and D). Overall, 20/28 (72%) patients were
Characteristics n (%)
discharged from hospital, 2/28 (7.1%) were trans-
ferred to another facility and 6/28 (21%) died. For
Female sex 13 (46)
Age, months, median (range) 4 (1 day–3 years)
those patients who survived to discharge and were
History
not transferred to another facility (n ¼ 20), the
Fever before enrolment (last 48 h) 21 (75) median length of hospitalisation was 4 days (range 0–
Cough 23 (81) 29).
Difficulty breathing 25 (89)
Altered level of consciousness 2 (7.4)
Convulsions 4 (15)
Vomiting 5 (18)
DISCUSSION
Diarrhoea 5 (19)
Unable to feed/drink 9 (32)
This proof-of-concept study demonstrates that solar
Clinical examination
energy can be used to concentrate oxygen from
Weight, kg, mean 6 SD 7.31 6 3.2 ambient air and oxygenate critically ill children with
Temperature, 8C, mean 6 SD 37.2 6 1.0 hypoxaemia in a resource-limited setting. Although
Fever (axillary temperature .37.58C) 12 (43)
the WHO lists oxygen as an essential medicine, it is
Blood pressure, mmHg, mean 6 SD*
Systolic 100 6 14
not available in many hospitals in developing
Diastolic 53 6 11 countries.16,29,30 A 2012 survey of 12 African
Altered consciousness† 5 (18) countries found that only 44% of 231 health centres,
HR, mean 6 SD 161 6 25 district hospitals and provincial/general hospitals had
Tachycardia (HR .99% quartile)‡ 13 (46)
RR, mean 6 SD 61 6 15
access to oxygen on a continuous basis.31 In Uganda,
Tachypnoea (RR .99% quartile)‡ 15 (54) only 36% of facilities offering chronic respiratory
Saturation, %, median [IQR] 83 [73–85] disease services had oxygen, and a verbal autopsy
Hypoxaemia (SaO2 6 90%) 28 (100)
Deep breathing 26 (96)
study found that only 4% of children who died from
Nasal flaring 24 (86) pneumonia received oxygen therapy.32 SPO2 could
Intercostal retractions 25 (89) help address the significant gaps in oxygen availabil-
Subcostal retractions 27 (96)
Wheeze 3 (11)
ity by bringing oxygen therapy to the periphery,
Stridor 2 (7.4) where most pneumonia deaths occur.
Crackles on auscultation 13 (46) Methods commonly used to deliver hospital
Laboratory assessments oxygen in resource-limited settings include com-
Hypoglycaemia (,3 mmol/l) 3 (11)
Lactate, mean 6 SD 6.16 6 6.7
pressed oxygen cylinders and oxygen concentrators.
Lactate .2.0, mmol/l 16 (64) Cylinders require a reliable supply chain linking the
Methaemoglobin, g/l, mean 6 SD 3.3 6 3.5 oxygen production plant to the hospital, which may
pH, mean 6 SD§ 7.3 6 0.16
PaCO2, mm Hg, mean 6 SD§ 40 6 16
be compromised by poor road conditions, especially
HCO3, mmol/l, mean 6 SD§ 19 6 7.3 during the rainy season, costs of transportation and
Plasma base excess,§ mmol/l, mean 6 weak stock management.30,31 Fuel for transportation
SD§ 9.3 6 8.2
of oxygen cylinders to health centres distant from an
Diagnoses
IMCI-defined severe pneumonia or very
oxygen plant represents an environmental cost of this
severe disease 28 (100) inefficient delivery method; in contrast, solar pow-
Malaria with respiratory distress 6 (21) ered oxygen uses ‘green energy’, with essentially a
HIV with pneumonia¶ 1 (10)
Other# 7 (25)
zero carbon footprint.
Oxygen concentrators purify oxygen from ambient
* Data available for 12/28 patients.
†
Blantyre Coma Scale , 5.27
air through selective adsorption of nitrogen using
‡
Normal range of vital signs from Fleming et al.28 aluminum silicate sieve beds.14,16,19 They can provide
§
Data available for 6/28 patients.
¶
Data available for 10/28 patients.
a reliable source of oxygen for many years, with
#
Down’s syndrome (n ¼ 2), diarrhoea and dehydration (n ¼ 2), congenital minimal service and maintenance,33 and are more
heart disease (n ¼ 1), birth asphyxia (n ¼ 1) and neonatal sepsis (n ¼ 1).
SD ¼ standard deviation; HR ¼ heart rate; RR ¼ respiratory rate; IQR ¼
cost-effective and user-friendly than cylinders.34
interquartile range; SaO2 ¼ arterial blood oxygen saturation; PaCO2 ¼ partial However, concentrators require a constant uninter-
carbon dioxide; HCO3 ¼ bicarbonate; IMCI ¼ Integrated Management of
Childhood Illness; HIV ¼ human immunodeficiency virus.
rupted electrical supply, which may not be available
in resource-poor settings.21,35 We observed that, at a
high-level referral hospital in Uganda serving a large
isation (0.11/day, P , 0.0001) following initiation
catchment area and treating often critically ill
of solar-powered oxygen therapy (Figure 4B). After
children, power failures averaged 13 episodes per
commencing oxygen therapy, the respiratory rate week and lasted a median of 30 min, with some
decreased by 10 breaths/min (95% confidence lasting as long as 17.5 h, which is unacceptable for
interval [CI] 3.2–17, P ¼ 0.004) and the heart rate concentrator-based management of patients in need
decreased by 23 beats/min (95%CI 16–31, P , of continuous oxygen therapy for survival. Lack of
0.0001) over the first day of admission, based on access to electrical energy is a problem in Uganda and
estimates from linear mixed-effects models (Figure many countries; a recent systematic review looking at
 Solar-powered oxygen 701

Figure 4 Vital signs of patients treated with SPO2. A–C) Longitudinal profiles of SpO2 (dotted line), HR (dashed line) and RR (solid
line) of three illustrative patients, together with the flow rate of SPO2 (grey area). SPO2 ¼ solar-powered oxygen; SpO2 ¼ peripheral
blood oxygen saturation; HR ¼ heart rate; RR ¼ respiratory rate.

13 health care facilities in sub-Saharan Africa found
that 26% of health facilities reported no access to
electricity, and that only 28% of those facilities with
electricity access reported reliable access (range 15–
49%).22 In Uganda, in a national survey in 2007, only
19% of government hospitals and 40% of private
facilities had electricity routinely available during
service hours or a backup generator with fuel.36
The baseline characteristics of patients in our pilot
study reflect a cohort with severe cardiopulmonary
compromise and guarded prognosis. Immediate
improvements in SpO2, resolution of vital sign
abnormalities over the first 24 h, recovery of function
(sitting and eating) over days, and ultimately dis-
charge without disability, occurred in the majority of
patients. This favourable clinical course in most
patients indicates the potential utility of SPO2, even
among patients with advanced disease. Nonetheless,
six patients in this study died despite the administra-
tion of quality-assured oxygen. This likely reflects the
severity of the underlying disease, and this study is
limited by the lack of a comparator (control) group
against which to gauge the efficacy of SPO2 delivery.
Ethical considerations of course prohibit a compar-
ison with placebo (no oxygen treatment). We are
currently conducting a randomised controlled trial

Figure 5 Improvement in vital signs associated with SPO2. A) Peripheral blood oxygenation
increased after initiating SPO2 therapy by a median 12% (range 5–15%, P , 0.0001). Dashed lines
link data from each individual patient. B) Composite SICK severity score decreased after initiating
SPO2 therapy (P , 0.0001). Dashed lines link data from each individual patient. C) RR decreased
over the first 24 h of admission following the initiation of SPO2 therapy by 10 breaths/min on
average (95%CI 3.2–17, P ¼ 0.004). Trend line corresponds to fitted LME model through individual
repeated measurements on the cohort. D) HR decreased over the first 24 h of admission following
initiation of SPO2 therapy by 23 beats/min on average (95%CI 16–31, P , 0.0001). Trend line
corresponds to fitted LME model through individual repeated measurements on the cohort. SPO2
¼ solar-powered oxygen; SpO2 ¼ peripheral blood oxygen saturation; SICK ¼ Signs of Infection in
Children that Kill; RR ¼ respiratory rate; HR ¼ heart rate; LME ¼ linear mixed-effect.
702 The International Journal of Tuberculosis and Lung Disease
(RCT) comparing SPO2 delivery with standard
delivery using compressed oxygen cylinders, under
the hypothesis of non-inferiority.37 This study will
provide context for the high mortality in this critically
ill patient population in a resource-limited hospital.
The SPO2 system was successful in harnessing solar
power to drive an oxygen concentrator. The system
was engineered and assembled using locally sourced,
commercially available components, which are likely
available worldwide. Although this technology has a
high initial capital assessment, the operating costs for
maintenance and use are low, due to the freely
available input: sun and air. Routine maintenance of
the oxygen concentrator (e.g., periodic replacement
of filters, humidifier bottle, compressor gaskets)
incurs costs, estimated at US$51 per month in a
previous study.23 The system components have an
estimated life of 20 years (solar panels) and 5 years
(batteries), and could therefore be used to treat
numerous patients prior to replacement. We designed
a system capable of running ‘off the grid’ without
recourse to grid electricity, for maximal general-
isability to the most remote settings. However,
significant cost savings on system components
(photovoltaic cells and batteries) could be achieved
in a hybrid system where solar energy supplements,
rather than replaces, grid electricity. In addition,
combining solar-powered oxygen generation with
low-pressure oxygen storage and/or oxygen sparing
strategies at the bedside could result in further
reductions in capital costs. Remote and resource-
limited settings are ideal for the implementation of
solar-powered oxygen delivery because of a context-
specific need for an oxygen delivery system that does
not rely on electrical power or cylinder distribution.
The SPO2 delivery system described was reliable,
simple to use by health care workers familiar with the
operation of an oxygen concentrator and may be a
sustainable solution to oxygen requirements in
remote hospitals. Relative to cylinders, which require
a reliable supply chain and need health workers to
change tank regulators with substantial mechanical
force and some safety risk, the SPO2 delivery system
was ‘hands-off’, apart from use of the concentrator.
Compared to operating a concentrator from a fuel-
powered generator, where cost, theft and procure-
ment may be challenges, the SPO2 delivery system
was always accessible and operational for health care
workers at no cost. Anecdotally, the generator at
JRRH was rarely started during power cuts, due to
lack of fuel or lack of access by ward staff to the
locked generator. Determining the sustainability of
SPO2 delivery will require longer-term studies.
However, the simplicity, reliability and availability
of the system are attractive features for users within
the hospital that may promote sustainable use of the
system. Maintenance of the concentrator and em-
ploying an electrician familiar with components will
be necessary to implement an SPO2 delivery system
sustainably.
CONCLUSION
Solar-powered oxygen delivery is feasible in a
resource-poor paediatric hospital. This exciting ad-
vance over existing oxygen delivery modes warrants
further study, and we are currently conducting an
RCT of SPO2.37 Development and scale-up of a
prototype that can supply oxygen to the numerous
hospitals across Africa and Asia with high pneumonia
burden and poor access to oxygen are needed.
Acknowledgements
The study was supported by a ‘Stars in Global Health’ Award from
Grand Challenges Canada, Toronto, ON, Canada.
Conflicts of interest: none declared.
References
1 Liu L, Oza S, Hogan D, et al. Global, regional, and national
causes of child mortality in 2000–2013, with projections to
inform post-2015 priorities: an updated systematic analysis.
Lancet 2015; 385: 430–440.
2 Nascimento-Carvalho C M. Etiology of childhood community
acquired pneumonia and its implications for vaccination. Braz J
Infect Dis 2001; 5: 87–97.
3 Mulholland K. Childhood pneumonia mortality—a permanent
global emergency. Lancet 2007; 370: 285–289.
4 Murray C J, Lopez A D. Mortality by cause for eight regions of
the world: Global Burden of Disease Study. Lancet 1997; 349:
1269–1276.
5 Walker C L, Rudan I, Liu L, et al. Global burden of childhood
pneumonia and diarrhoea. Lancet 2013; 381: 1405–1416.
6 World Health Organization/United Nations Children’s Fund.
Ending preventable deaths from pneumonia and diarrhoea by
2025: integrated global action plan for prevention and control
of pneumonia and diarrhoea. Geneva, Switzerland: WHO,
2013. http://www.who.int/maternal_child_adolescent/
documents/global_action_plan_pneumonia_diarrhoea/en/.
Accessed January 2016.
7 Onyango F E, Steinhoff M C, Wafula E M, Wariua S, Musia J,
Kitonyi J. Hypoxaemia in young Kenyan children with acute
lower respiratory infection. BMJ 1993; 306: 612–615.
8 Ibraheem R M, Johnson W B, Abdulkarim AA. Hypoxaemia in
hospitalised under-five Nigerian children with pneumonia.
West Afr J Med 2014; 33: 37–43.
9 Subhi R, Adamson M, Campbell H, et al. The prevalence of
hypoxaemia among ill children in developing countries: a
systematic review. Lancet Infect Dis 2009; 9: 219–227.
10 Salah E T, Algasim S H, Mhamoud A S, Husian N E. Prevalence
of hypoxemia in under-five children with pneumonia in an
emergency pediatrics hospital in Sudan. Indian J Crit Care Med
2015; 19: 203–207.
11 Lozano J M. Epidemiology of hypoxaemia in children with
acute lower respiratory infection. Int J Tuberc Lung Dis 2001;
5: 496–504.
12 Duke T, Wandi F, Jonathan M, et al. Improved oxygen systems
for childhood pneumonia: a multihospital effectiveness study in
Papua New Guinea. Lancet 2008; 372: 1328–1333.
13 Catto A G, Zgaga L, Theodoratou E, et al. An evaluation of
oxygen systems for treatment of childhood pneumonia. BMC
Public Health 2011; 11 (Suppl 3): S28.
14 Dobson M B. Oxygen concentrators and cylinders. Int J Tuberc
Lung Dis 2001; 5: 520–523.

15 Dobson M B. Oxygen concentrators offer cost savings for
developing countries. A study based on Papua New Guinea.
Anaesthesia 1991; 46: 217–219.
16 Duke T, Peel D, Graham S, Howie S, Enarson P M, Jacobson R.
Oxygen concentrators: a practical guide for clinicians and
technicians in developing countries. Ann Trop Paediatr 2010;
30: 87–101.
17 Dobson M, Peel D, Khallaf N. Field trial of oxygen
concentrators in upper Egypt. Lancet 1996; 347: 1597–1599.
18 Wood P B. Oxygen concentrator in a remote hospital in Zaire.
Trop Doct 1985; 15: 26–27.
19 Enarson P, La Vincente S, Gie R, Maganga E, Chokani C.
Implementation of an oxygen concentrator system in district
hospital paediatric wards throughout Malawi. Bull World
Health Organ 2008; 86: 344–348.
20 Shrestha B M, Singh B B, Gautam M P, Chand M B. The oxygen
concentrator is a suitable alternative to oxygen cylinders in
Nepal. Can J Anaesth 2002; 49: 8–12.
21 Mokuolu O A, Ajayi O A. Use of an oxygen concentrator in a
Nigerian neonatal unit: economic implications and reliability.
Ann Trop Paediatr 2002; 22: 209–212.
22 Adair-Rohani H, Zukor K, Bonjour S, et al. Limited electricity
access in health facilities of sub-Saharan Africa: a systematic
review of data on electricity access, sources, and reliability.
Glob Health Sci Pract 2013; 1: 249–261.
23 Schneider G. Oxygen supply in rural Africa: a personal
experience. Int J Tuberc Lung Dis 2001; 5: 524–526.
24 World Health Organization. Integrated management of
childhood illness. Geneva, Switzerland: WHO, 2014. http: //
www.who.int/maternal_child_adolescent/topics/child/imci/en/.
Accessed January 2016.
25 Kumar N, Thomas N, Singhal D, Puliyel J M, Sreenivas V.
Triage score for severity of illness. Indian Pediatr 2003; 40:
204–210.
26 The Weather Channel. Jinja climate. Weather Underground.
Online database of daily weather conditions. Atlanta, GA,
USA: 2016. www.wunderground.com.
Solar-powered oxygen 703
27 Molyneux M E, Taylor T E, Wirima J J, Borgstein A. Clinical
features and prognostic indicators in paediatric cerebral
malaria: a study of 131 comatose Malawian children. Q J
Med 1989; 71: 441–459.
28 Fleming S, Thompson M, Stevens R, et al. Normal ranges of
heart rate and respiratory rate in children from birth to 18 years
of age: a systematic review of observational studies. Lancet
2011; 377: 1011–1018.
29 Nolan T, Angos P, Cunha A J, et al. Quality of hospital care for
seriously ill children in less-developed countries. Lancet 2001;
357: 106–110.
30 Hill S E, Njie O, Sanneh M, et al. Oxygen for treatment of
severe pneumonia in The Gambia, West Africa: a situational
analysis. Int J Tuberc Lung Dis 2009; 13: 587–593.
31 Belle J, Cohen H, Shindo N, et al. Influenza preparedness in
low-resource settings: a look at oxygen delivery in 12 African
countries. J Infect Dev Ctries 2010; 4: 419–424.
32 Kallander K, Hildenwall H, Waiswa P, Galiwango E, Peterson
S, Pariyo G. Delayed care seeking for fatal pneumonia in
children aged under five years in Uganda: a case-series study.
Bull World Health Organ 2008; 86: 332–338.
33 La Vincente S F, Peel D, Carai S, et al. The functioning of
oxygen concentrators in resource-limited settings: a situation
assessment in two countries. Int J Tuberc Lung Dis 2011; 15:
693–699.
34 Perrelet A, Zellweger J P, Talla I, Ndiaye Y, Gautier E, Gehri M.
The oxygen concentrator: an appropriate technology for
treating hypoxaemic children in developing countries. Int J
Tuberc Lung Dis 2004; 8: 1138–1141.
35 Malkin R A. Design of health care technologies for the
developing world. Annu Rev Biomed Eng 2007; 9: 567–587.
36 Uganda Ministry of Health & Macro International Inc. Uganda
Service Provision Assessment Survey, 2007. Kampala, Uganda:
Ministry of Health & Macro International Inc, 2008.
37 Nyende S, Conroy A, Opoka R O, et al. Solar-powered oxygen
delivery: study protocol for a randomized controlled trial.
Trials 2015; 16: 297.

C A D R E : Un hôpital d’Ouganda aux ressources limitées
pour le traitement des pneumonies pédiatriques.
O B J E C T I F : La pneumonie est une cause majeure de
mortalité des enfants dans le monde. L’accès au
traitement salvateur par oxygène est limité dans de
nombreuses régions du monde. L’objectif de l’étude a été
de concevoir et de mettre en oeuvre un système de
distribution d’oxygène à énergie solaire.
S C H É M A : Etude pilote de démonstration de faisabilité.
Un système de distribution d’oxygène à énergie solaire a
été conçu et piloté dans une cohorte d’enfants atteints
d’une maladie hypoxémiante.
R É S U L T A T S : Le syst ème a consist é en panneaux
solaires photovoltaı̈ques de 25 3 80 W (fourniture
quotidienne de 7,5 kWh [fourchette 3,8–9,7]), en
batteries de 8 3 220 Ah et en un concentrateur
d’oxygène de 300 W (fourniture jusqu’à 5 l/min
M A R C O D E R E F E R E N C I A: Un hospital pediátrico de
referencia con recursos limitados, en la ciudad de Jinja
de Uganda.
O B J E T I V O: La neumonı́a es una de las principales
causas de mortalidad infantil en todo el mundo. El
acceso a una oxigenoterapia salvavidas es limitado en
muchas regiones del mundo. El presente estudio tuvo por
objetivo desarrollar y poner en práctica un sistema de
aporte de oxı́geno que funciona con energı́a solar, para el
tratamiento de la neumonı́a de los niños en un hospital
con recursos limitados en Uganda.
M É T O D O: Fue este un estudio preliminar de eficacia de
un nuevo sistema de oxigenoterapia que funciona con
energı́a solar, en una cohorte de niños aquejados de una
enfermedad que provoca hipoxemia.
R E S U L T A D O S: El sistema consistió en 25 tableros
fotovoltaicos de 80 W (con una potencia de salida
diaria de 7,5 kWh [entre 3,8–9,7]), ocho pilas de 220 Ah
y un concentrador de oxı́geno de 300 W (salida hasta de
Solar-powered oxygen i
RESUME
d’oxygène à 88% [6 2%] de pureté). Une série de 28
patients ayant une hypoxémie (âgés de 3 mois à 3 ans)
ont été enrôlés et traités avec l’oxygène à énergie solaire.
Une amélioration immédiate de la saturation en oxygène
du sang périphérique a été documentée (amélioration
médiane þ12% [fourchette 5–15] ; P , 0,0001). La
tachypnée, la tachycardie et le score de gravité composite
de la maladie se sont améliorés au cours des premières 24
h d’hospitalisation (P , 0,01 pour toutes les
comparaisons). Le taux de létalité a été de 6/28 (21%).
Le délai médian de récupération pour s’asseoir, manger,
sevrer l’oxygène et quitter l’hôpital a été de 7,5 h, 9,8 h,
44 h et 4 jours, respectivement.
C O N C L U S I O N : L’énergie solaire peut être utilisée pour
concentrer l’oxygène de l’air ambiant et oxygéner les
enfants atteints de détresse respiratoire et d’hypoxémie
dans un contexte de ressources limitées.
RESUMEN
5 l/min de oxı́geno con una pureza de 88% [6 2%]).
Una serie de 28 pacientes con hipoxemia participó en el
estudio (de 3 meses a 3 años de edad) y recibió
oxigenoterapia con el dispositivo de energı́a solar. Se
puso en evidencia una mejora inmediata de la saturación
de oxı́geno periférica (mediana del cambio þ 12% [entre
5% y 15%]; P , 0,0001). La taquipnea, la taquicardia y
la puntuación de la escala compuesta de gravedad de la
neumonı́a mejoraron a las 24 h de la hospitalización (P
, 0,01 para todas las comparaciones). El ı́ndice de
letalidad fue 6 de 28 (21%). La mediana del lapso de
recuperación hasta poderse sentar fue 7,5 h, hasta comer
fue 9,8 h, hasta la retirada del oxı́geno fue 44 h y 4 dı́as
hasta el alta hospitalaria.
C O N C L U S I Ó N: Es posible utilizar la energı́a solar con el
propósito de concentrar el oxı́geno del medio ambiente y
ofrecer oxigenoterapia a los niños con insuficiencia
respiratoria e hipoxemia en un entorno con recursos
limitados.