Key points

Pulse oximetry plays an essential role in early detection of hypoxaemia and in guiding oxygen therapy.

Although pulse oximetry has become a basic standard of care in the developed world, its availability and use is

severely limited in low-resource settings.

Barriers preventing the uptake of pulse oximetry need to be identified and overcome so that oximetry becomes

available to clinicians and patients throughout the world.

Pulse oximetry in low-resourcesettings

Educational aims[ To increase awareness about the issues of making pulse oximetry available in low-

resource settings[ To increase awareness of the World Health Organization Surgical Safety checklist

and how this has triggered the concept of Global Oximetry.

[ To discuss the relevance of increasing availability of pulse oximetry to areas otherthan operating theatres (such as medical wards) in hospitals in low-resource settings.

SummaryPulse oximetry is used to detect hypoxaemia. It is widely used in both the pre-hospital and hospital settings in developed countries and has become a basicstandard of care. There are substantial differences in healthcare betweendeveloped and developing countries and it has been recorded that surgery isassociated with a much higher number of complications and deaths occurring inresource-limited settings. To address this issue, the World Health Organization(WHO) introduced the Safe Surgery Saves Lives Programme in 2007. With thisprogramme, the WHO Surgical Safety Checklist was launched. Included in thischecklist is the step to ensure that a pulse oximeter is on the patient andfunctioning. WHO now has the Global Pulse Oximetry Project, an initiative topromote the use of pulse oximeters in every operating room in the world. TheLifebox project is a charity that aims to help the supply of pulse oximeters at lowcost to anaesthesia providers in low-resource settings.Increasing availability of pulse oximetry in low-resource settings is relevant to allphysicians because hypoxaemia is a common complication of many illnesses,particularly pneumonia. Pneumonia impacts developing countries disproportionately,and accounts for over 2 million deaths a year worldwide. Hypoxaemia is a recognisedrisk factor for death, correlates with disease severity and is difficult to detect clinicallyuntil cyanosis is present. Oximetry plays an essential role in early detection ofhypoxaemia and in guiding oxygen therapy, which is often a scarce resource.

Statement of InterestI.H. Wilson ispresident of theAssociation ofAnaesthetists of GreatBritain and Ireland andis a trustee of Lifebox.

HERMES syllabus link:Module D.1.4

Breathe | December 2012 | Volume 9 | No 2 91DOI: 10.1183/20734735.038612

Lara J. Herbert1,

Iain H. Wilson2

1Dept of Anaesthesia,Bristol Royal Infirmary,Bristol, and2Dept of Anaesthesia, RoyalDevon and Exeter Hospital,Exeter, UK

L.J. Herbert: Dept ofAnaesthesia, Level 7, BristolRoyal Infirmary,Marlborough Street, Bristol,BS28HW, UK

laraherbert@doctors.org.uk

Oximetry

Hypoxaemia can be detected using clinicalsigns, blood gas analysis or pulse oximetry.

Clinical signs are often unreliable in thediagnosis of the presence or absence of hypox-aemia. For example, cyanosis has poor sensitivity:the lack of cyanosis, despite severe significantcentral nervous system symptoms from hypox-aemia was recognised by J.S. Haldane in 1920[1]. Blood gas analysis is expensive, invasive andprovides a single measure in time only. Anaemiais a common condition in poorer parts of theworld and makes the detection of cyanosis moredifficult.

Oximetry is the determination of arterialoxygen saturation of haemoglobin by com-paring absorbance of light of different wave-lengths by blood. It relies on the principlethat absorbance of light energy by haemoglo-bin depends upon its level of oxygenation.Oxygenated and deoxygenated haemoglobin(HbO and Hb respectively) have differentabsorbance spectra. Comparison of absor-bances at different wavelengths allows esti-mation of the relative concentrations ofHbO and Hb (i.e. arterial oxygen saturation(SaO2)).

Pulse oximetry is the accepted standardfor detecting hypoxaemia. It was introducedin the 1980s and is now widely used in modernhealthcare systems as a basic monitoring tool.It is simple, non- invasive technique and whenused correctly, can provide reliable monitoringwithout distress to the patient. Using a probeplaced on the finger, toe or ear lobe, theabsorption of light (emitted by light emit-ting diodes (LEDs)) passing through tissue ismeasured and processed to produce a readingof pulse rate and oxygen saturation.

Pulse oximeters are reliable and robustdevices. The component parts include theprobe, which is delicate and will generally last6–12 months unless damaged, the powersupply (mains or battery), the processor andthe display.

Use of pulse oximeters

The importance of pulse oximeters wasquickly recognised in many different clinicalareas, particularly in intensive care, anaes-thesia, respiratory medicine, paediatric and

emergency care. Since the late 1980s theyhave been introduced into all clinical areas,including home therapy.

In addition to monitoring for hypoxaemia,pulse oximeters can be used to ensure themost efficient use of oxygen therapy, which isespecially important in resource-limited set-tings, where oxygen may be scarce. Further-more, oximetry is considered essential inneonatal practice to prevent the devastatingcomplication of blindness due to retrolentalfibroplasia, which is caused by excess use ofoxygen.

Pulse oximetry foranaesthesia and surgery

For more than 20 years, the use of pulseoximetry for anaesthesia monitoring duringsurgery has been a standard of care in thedeveloped world. Pulse oximeters are used innearly every procedure that involves anaesthe-sia or sedation. The safety of anaesthesiaimproved dramatically following the availabilityof routine oximetry, and the risk of mortalityfrom anaesthesia reduced from around 1:10,000to 1:100,000–1:200,000. Other developmentshappening around the same time, includingcapnography, training, drugs and equipment,also contributed to improving patient safetyduring anaesthesia. However, these safetypractices have not been routinely instigatedin the developing world. Estimates suggestthat more than half of operating rooms are notequipped with pulse oximeters.

It has been estimated that over 230 millionsurgical procedures are performed around theworld each year [2]. In the developed world, 3–16% of hospitalised surgical patients havemajor complications and nearly 1% experiencepermanent disability or death as a result oftheir operation [3, 4], If these numbers areextrapolated globally, at least 7 million peopledevelop serious complications and 1 milliondie; 50% of these complications are thought tobe preventable [5].

Due to substantial differences in thesafety of surgery between developed anddeveloping countries, a disproportionate num-ber of complications and deaths are likely tooccur in resource-limited settings. Anaesthe-sia death rates in these settings are reportedly100- to 1,000-times higher than in the devel-oped world [6]. For example, a mortality rate

Pulse oximetry in low-resource settings

Breathe | December 2012 | Volume 9 | No 292

of 1:133 has been recorded in Togo, mostlyhypoxia related [7].

The WHO Surgical SafetyChecklist

To improve the safety of surgery, WHOlaunched the Safe Surgery Saves Lives pro-gramme in 2007 [8]. One goal was to define aminimum set of surgical safety standards thatcould be applied in all countries and hospitalsettings. The result was the WHO surgicalsafety checklist, which was launched in June2008 [9]. The checklist is simple and can becompleted in less than two minutes.

Over 280 professional bodies, hospitalsand ministries of health have approved thechecklist, which includes a set of basic stepsto follow before, during and after surgery.Examples of the steps include: confirmingpatient identity, recording medication allergies,administering antibiotics on time, countinginstruments, sponges and needles and ensur-ing that a pulse oximeter is on the patient andfunctioning.

At the time of the checklist launch, WHOreleased preliminary results from over 1,000patients in eight pilot hospitals across theworld. The checklist nearly doubled the chancethat patients would receive proven standardsof surgical care and substantially reducedcomplications and deaths [10].

The WHO decided that pulse oximetryshould become a standard of care and featureas a requirement on the checklist. Followingthe publication of the checklist, the WHO andthe World Federation of Societies of Anesthe-siologists (WFSA) confirmed oximetry as astandard of care for all patients undergoinganaesthesia and surgery [11, 12].

Despite the establishment of pulse oxime-try as a monitoring standard, the technology isstill not currently available in many operatingrooms around the world. WHO started theGlobal Pulse Oximetry Project, an initiative toimprove the availability of pulse oximeters inevery operating room in the world.

Although it has been estimated that around77,000 operating rooms need pulse oximetersin resource-poor settings, the overall figure ismuch higher. If all clinical areas are taken intoaccount, the numbers of oximeters requiredhas been estimated at over 1,000,000 [13].

Barriers to achieving globalpulse oximetry

The relatively high initial cost of pulseoximetry technology has been a significantbarrier in many developing world settings.Markets in resource poor areas are small andmanufacturers do not find it profitable tomake the investment in developing a salesinfrastructure in many economies.

Operating theatres have to compete forfunds with other parts of the hospitals andcapital investment is often low, limiting theavailability of all equipment that might not beconsidered absolutely essential.

Where pulse oximeters have been pur-chased, the costs are generally higher than inwell-resourced settings due to the difficulty ofmarket creation and resultant small numbersof orders.

The fragility of the probe is another pro-blem; in many settings, donated oximeters lieunused in cupboards due to broken probes,which have not been, or cannot be replaced.Spare probes may be prohibitively expensive inthis setting, as a single import.

Affordability is not the only factor affectingaccessibility of pulse oximetry in low-resourcesettings. As with all equipment, challenges ofdistribution within country including importa-tion and taxation costs, coupled with inadequatesupply chains pose difficulties. An absence of areliable electricity supply is common, requiringbattery function.

Many countries have a serious lack oftrained healthcare workers. This difficulty isgreatest in sub-Saharan Africa, which has 11% ofthe world’s population and 24% of the world’sdisease burden, but only 3% of the world’s heal-thcare workers. Although pulse oximetry is asimple technique, there is still a requirement forinterpretation by someone trained in their use.75% might be considered an excellent examresult, but it is a dangerously low oxygen satura-tion. Training is required to understand theimplications!

The realities of workforce and traininghave been highlighted in anaesthesia. In 2007,Uganda had around 15 physician anaesthetistsfor a population of 27 million. The UK has12,000 physician anaesthetists for a populationof 60 million. Due to the severe shortage ofphysician anaesthetists in Uganda, most anaes-thetics are administered by 350 anaestheticofficers in the country who have received only

Pulse oximetry in low-resource settings

Breathe | December 2012 | Volume 9 | No 2 93

1–2 years of anaesthesia training. Electricalsupplies were constant for only 20% of pro-viders and 10% of anaesthetic providers workedwithout an oxygen supply. The item mostfrequently unavailable for 75% of providerswas a pulse oximeter [14].

Pulse oximetry outside theoperating theatre in low-resource settings

The launch of the WHO Surgical SafetyChecklist has meant that much of the focuson pulse oximetry availability in low-resourcesettings has been on operating theatres. Dataon availability of pulse oximetry on wards andother areas of hospitals in low-resource set-tings is lacking.

Hypoxaemia is a major cause of morbidityand mortality associated with chronic and acutelung disease in children and adults. Oxygen hasbeen available as therapy for hypoxaemia andused with significant clinical benefit for morethan a century. It is a life-saving therapy in manycritical serious illnesses that can and should bewidely available. Unfortunately, oxygen therapyis often scarce in low- and middle-incomesettings that have by far the greatest burden ofdeath due to lung disease.

Every year, nearly 9 million children diemostly from preventable or treatable diseases,and more than 95% of these deaths occur indeveloping countries. Pneumonia is the lead-ing cause of death in children under 5 years ofage, responsible for an estimated 18% of alldeaths in this age category. Hypoxaemia is amajor fatal complication of pneumonia [15]. Ifpneumonia is combined with hypoxaemia, ashappens in 13% of cases, children are 5-timesmore likely to die than are those with onlypneumonia [16].

Even conditions that are rarely complicatedby hypoxaemia contribute considerably to theglobal hypoxaemia burden because they are socommon. For example 3–5% of all hospitalisedcases of malaria have hypoxaemia. Of 13,183children aged o60 days admitted to a districthospital in rural coastal Kenya, 5.3% werehypoxaemic and the most frequent final diag-noses among hypoxaemic children were malaria(35%), pneumonia (32%), malnutrition (10%)and gastroenteritis (7%) [17].

Apnoea and hypoventilation may occur inotherwise healthy premature babies due to

immature respiratory drive, and in babies of anygestational age with sepsis, asphyxia, seizuresor hypoglycaemia. Cautious oxygen therapy, toavoid the preventable complication of fibrolen-tal hyperplasia, is often necessary for thesecommon conditions that cause the large pro-portion of the neonatal deaths occurring glo-bally each year [18].

Therefore, oxygen is a therapy that is a basicrequirement to save lives of seriously ill patientsand is included on the WHO ‘‘EssentialMedicine’’ list. Despite this, it is frequentlyunavailable in primary health facilities and smallhospitals.

Clearly where oxygen supplies are limited,the best use of a scarce resource has to bemade. Oximetry accurately determines theneed for oxygen, how much is required andfor how long.

Pulse oximetryspecifications

Having adopted the concept of GlobalOximetry, WHO developed specifications forpulse oximeters that would be suitable for theenvironment of resource poor hospital oper-ating theatres (table 2) [19].

Pulse oximetry probe designand prices

There is a wide range of oximetry models onthe market. They differ in cost, durability,accuracy and the variety of information theyare able to provide. Units available for salegenerally fall into three distinct groups:

1. Finger probe pulse oximeters intended forpersonal use or spot checks.

2. Hand-held units or stand-alone unitsintended to measure just oxygen saturationand pulse rate only.

3. Multimodal units, which incorporate otherparameters such as electrocardiography(ECG), capnography and blood pressuremonitoring.

The finger probe pulse oximetry units arecurrently available from US$20–50. They areable to measure SpO2 and pulse as a wave-form and a numeric digital display. They arelightweight and are even available from

Educationalquestions1. Light is absorbed byhaemoglobin at differentfrequencies. Which of thefollowing statements aretrue?

a. Absorption of light byoxyhaemoglobin is greater at660 nm than 940 nm

b. Fordeoxyhaemoglobin,absorption of light is muchgreater at 660nm than940nm

c. At 860nm, bothoxyhaemoglobin anddeoxyhaemoglobin have thesame rate of absorption

d. Carboxyhaemoglobinabsorbs light at 660nm

e. Methaemoglobinabsorbs light at the samefrequency asdeoxyhaemoglobin

2. Which of the followingstatements regardingoxygen transport in theblood are true?

a. pO2 depends on theamount of O2 bound to Hb

b. Each Hb molecule canbind to four O2 molecules

c. Hb binds to O2 moreavidly at lower pH

d. The Bohr effect refersto the effect of temperatureon the Hb–O2 dissociationcurve

e. In the lungs, theHb–O2 dissociation curve isshifted to the left

3. Which of the followingstatements about pulseoximetry are true?

a. Burns from LEDs mayoccur in poorly perfusedpatients

b. Pulse oximeteraccuracy is between ¡2%between 90–100% SpO2

c. Acute changes are notdetectable

d. Pulse oximeters detectsaturation and adequacy ofventilation

e. External fluorescentlighting can interfere withpulse oximetry readings

4. What is the leadingcause of death in childrenunder 5 years of age?

a. Gastroenteritis

b. Pneumonia

c. Malaria

d. Acute asthma

e. Meningitis

Pulse oximetry in low-resource settings

Breathe | December 2012 | Volume 9 | No 294

Amazon. Although these units are ideal forpersonal use and may be used in somelimited clinical settings, they are not ideal forthe operating theatre environment.

In the operating theatre, the specificationsrequire a more robust model with a screen thatis easy to read and an audible signal. Longlasting rechargeable batteries are essential tocope with long power cuts. (Replacementbatteries (e.g. AA cells) are often expensiveand not easily afforded. A cable running fromthe probe to the readout device means that thescreen can be positioned at a distance from thepatient and a variety of probes can be attacheddepending on need (e.g. infants or adults).Currently prices vary considerably depending onthe features of the machine.

Oximetry requirements for the operatingtheatre are likely to be different to thoserequired for the ward or outpatient setting. Inthe ward or clinic setting, where spot checks ofoxygen saturation are performed, less sophis-ticated and cheaper pulse oximetry modelsmay be more appropriate.

Multimodal monitors are increasingly com-mon in Europe. On the ward, a single unit willmonitor SpO2, pulse rate and non-invasive bloodpressure. In theatre, capnography and gasanalysis is included. These units are morecomplex and expensive, costing up to £10,000for theatre monitoring.

The mobile phone oximeter is being devel-oped using a specialised probe and customsoftware for a Smartphone. User-friendly soft-ware has signal-processing algorithms for

oxygen saturation, respiratory rate and heartrate from the plethysmographic waveform. Thisproduct may find use in developing countrieswhere mobile phones are widely available. Thedevice and software is still undergoing devel-opment and evaluation [20].

The choice of units will always be a balancebetween utility, reliability and price. Howeverthe barriers to purchase at a competitive pricewill always disadvantage the small ordertypically placed by healthcare providers inlow-resource settings.

Following the WHO initiative, the WFSAconducted a tender exercise to try to source asuitable low-cost pulse oximeter for wide-spread use by colleagues overseas, selectinga model from Acare, Taiwan (fig. 1).

Lifebox

To facilitate the availability of these oximeters,the Lifebox Project, a charity developed by theWFSA, the Association of Anaesthetists ofGreat Britain and Ireland (AAGBI) and theHarvard School of Public Health (HSPH),aims to help the supply of pulse oximeters atlow cost to anaesthesia providers in not-for-profit hospitals in low-income and low-middleincome countries. Details can be found atwww.lifebox.org where oximeters are availablefor $250 including delivery.

In addition, Lifebox provides education inthe use of oximetry. This is essential, as manyclinicians who have not used this technology

Table 1. Causes of hypoxaemia

Neonates Children Adults

Respiratory distress syndromeBirth asphyxiaSepsisApnoeaHypoventilation

PneumoniaMeningitisAcute asthmaAcute sepsisSevere malaria

Chronic obstructive pulmonarydiseaseAcute asthmaPneumoniaSepsisShockMajor traumaAnaphylaxisAcute heart failurePulmonary embolismPleural effusionPneumothoraxLung fibrosisCarbon monoxide poisoningObstetric and surgical emergenciesSickle cell crises

Pulse oximetry in low-resource settings

Breathe | December 2012 | Volume 9 | No 2 95

previously may be unaware of the normaloxygen saturation levels and how to managea falling saturation. Instruction and informa-tion in the use of the WHO Surgical SafetyChecklist is also provided.

Lifebox is an innovative approach toprocurement for hospitals and ministries. Byreceiving pooled purchases, a manufacturer isable to construct a more viable businessmodel. The role of specialist professionalgroups is key to creating demand and also tostimulating donations to the project. Delivery,customs and warranty are all potential issues,but oximetry is a particularly reliable and low-cost technology, suitable for this approach.

Importantly, as Lifebox does not work inwealthy established markets (part of the

Table 2. Pulse oximetry specifications for hospital operating theatres

Environmental issues and other specifications Oximeters should comply with relevant standards such as InternationalElectrotechnical Commission (IEC) 69691-1 and International StandardsOrganisation (ISO)9919:2005. Units should work reliably in a wide range ofoperating temperatures (e.g. 10uC to 40uC) and humidities (e.g. relativehumidity 15–95%).

Physical features Units should be robust enough to withstand falls of 1 m onto concrete.Portability and ease of handling is desirable.

Power supply Units must be able to be powered by rechargeable batteries, normalbatteries and mains supply electricity.

Ease of use Oximeters should be intuitive and simple to use. Interfaces should belanguage free to the extent possible, although it may be suitable to haveconfigurable language displays.

Alarms Alarms indicating breaches of safe limits of oxygen saturation should beaudible and supplemented with a visible change of display (such asflashing). Ideally an alarm should indicate sensor misplacement.

Oxygen saturation Arterial oxygen saturation should be measured between clinically relevantlimits (e.g. 70–100%) with reasonable accuracy (e.g. ¡2% of truesaturation).

Pulse display The device should have a plethysmograph display of the pulse (eitherwaveform or bar graph). The unit should measure pulse rate betweenclinically appropriate limits (e.g. 20–200 beats?min-1) to ¡3 beats?min-1.The pulse rate should also be displayed numerically and be represented byan audible tone, which changes as saturation levels deteriorate.

Display The readout should be clearly visible from 5 m.

Sensors/probes Probes should be as robust as possible. A range of sensors covering varioussizes and ages of patient should be available, and all probes should bereusable.

Warranties and maintenance The expected life of the oximeter and probes should be specified andappropriate warranties provided. User and service manuals should beprovided, electronically and by hard copy, in a variety of languages.

Figure 1The Lifebox Pulse Oximeter.

Pulse oximetry in low-resource settings

Breathe | December 2012 | Volume 9 | No 296

charitable status), the market for major man-ufacturers has not altered, but rather newmarkets are being developed.

Conclusion

There is no doubt that pulse oximetry shouldbecome as readily available to cliniciansthroughout the world as the mobile phone.The technology provides immediate detectionof falling levels of oxygen and in the UK and

other wealthy settings, are mandated as a partof theatre monitoring.

Patients’ lives are saved daily by theavailability of pulse oximetry and by healthcareworkers trained to react to oximetry readings. Itis time that oximeters were universally availableand affordable anywhere in the world, not onlyto patients undergoing surgery and anaesthe-sia, but also to medical patients. This is a vitalpatient safety initiative and will need thecooperation and leadership of many people tomake this happen.

References

1. Haldane JS. Discussions on the therapeutic uses ofoxygen. Proc Royal Soc Med 1920; 13: 76.

2. Weiser TG, Regenbogen SE, Thompson KD, et al. Anestimation of the global volume of surgery: amodeling strategy based on available data. Lancet2008; 372: 139–144.

3. Gawande AA, Thomas EJ, Zinner MJ, et al. Theincidence and nature of surgical adverse events inColorado and Utah in 1992. Surgery 1999; 126: 66–75.

4. Kable AK, Gibberd RW, Spigelman AD. Adverseevents in surgical patients in Australia. J Qual HealthCare 2002; 14: 269–276.

5. Haynes A, Weiser T, Berry W, et al. A surgical safetychecklist to reduce morbidity and mortality in aglobal population. New Eng J Med 2009; 360:491–499.

6. Walker IA, Wilson IH. Anaesthesia in developingcountries- a risk for patients. Lancet 2009; 371:968–969.

7. Ouro-Bang’na Maman AF, Tomta K, Ahouanbevi S,et al. Deaths associated with anaesthesia inTogo, West Africa. Tropical Doctor 2005; 35:220–222.

8. World Health Organization. Safe Surgery Saves Lives.www.who.int/patientsafety/safesurgery Date lastaccessed: August 31, 2012.

9. World Health Organization. WHO surgical safetychecklist and implementation manual. www.who.int/patientsafety/safesurgery/ss_checklist/en/index.htmlDate last accessed: August 31, 2012.

10. World Health Organization. New checklist to helpmake surgery safer. www.who.int/mediacentre/news/releases/2008/pr20/en/index.html Date last accessedon August 31, 2012.

11. Merry AF, Cooper JB, Soyannwo O, et al. Internationalstandards for a safe practice of anaesthesia 2010. CanJ Anaesth 2010; 57: 1027–1034.

12. World Health Organization. WHO guidelinesfor safe surgery 2009. Safe surgery saves lives.http://whqlibdoc.who.int/publications/2009/9789241598552_eng.pdf Date last accessed: August31, 2012.

13. Howitt P, Darzi A, Yang GZ, et al. Technologies forGlobal Health Lancet 2012; 380: 507–535.

14. Hodges SC, Mijumbi C, Okello, et al. Anaesthesiaservices in developing countries: defining the pro-blems. Anaesthesia 2007; 62: 4–11.

15. Duke T, Graham SM, Cherian MN, et al. UnionOxygen Systems Working Group. Oxygen is anessential medicine: a call for international action. IntJ Tuberc Lung Dis 2010; 14: 1362–1368.

16. Ashraf H, Jobayer M, Alam N. Treatment of child-hood pneumonia in developing countries. www.intechopen.com/books/health-management/treatment-of-childhood-pneumonia-in-developing-countries.Date last accessed: August 31, 2012.

17. Mwaniki MK, Nokes DJ, Ignas J, et al. Emergencytriage assessment for hypoxaemia in neonates andyoung children in a Kenyan hospital: an observationalstudy. Bull World Health Organ 2009; 87: 263–270.

18. Lawn JE, Cousens S, Zupan J, et al. 4 million neonataldeaths: When? Where? Why? Lancet 2005; 365:891–900.

19. World Health Organization. WHO Global PulseOximetry Project background document.www.who.int/entity/patientsafety/events/08/1st_pulse_oximetry_meeting_background_doc.pdf Datelast accessed: August 31, 2012.

20. World Health Organization. Compendium of new andemerging technologies that address global healthconcerns 2011. Mobile phone pulse oximeter. www.who.int/medical_devices/innovation/new_emerg_tech/en/index2.html Date last accessed: August 31, 2012.

Suggestedanswers

1. b, d, e.2. b, e.3. a, c, e.4. b.

Pulse oximetry in low-resource settings

Breathe | December 2012 | Volume 9 | No 2 97