Computer generated transcript

Warning!
The following transcript was generated automatically from the content and has not been checked or corrected manually.

Articles Anaesthesia-relatedmaternalmortalityinlow-income andmiddle-incomecountries:asystematicreviewand meta-analysis SohaSobhy*,JavierZamora*,KuhanDharmarajah,DavidArroyo-Manzano,MatthewWilson,RamesanNavaratnarajah,ArriCoomarasamy, KhalidSKhan,ShakilaThangaratinam Summary Background The risk factors contributing to maternal mortality from anaesthesia in low-income and middle-income LancetGlobHealth2016; countries and the burden of the problem have not been comprehensively studied up to now. We aimed to obtain precise 4:e320–27 estimates of anaesthesia-attributed deaths in pregnant women exposed to anaesthesia and to identify the factors linked SeeCommentpagee290 to adverse outcomes in pregnant women exposed to anaesthesia in low-income and middle-income countries. *Jointfirstauthors Women’sHealthResearchUnit (SSobhyMBBS, Methods In this systematic review and meta-analysis, we searched major electronic databases from inception until Oct 1, 2015, for studies reporting risks of maternal death from anaesthesia in low-income and middle-income ProfJZamoraPhD, ProfKSKhanMSc, countries. Studies were included if they assessed maternal and perinatal outcomes in pregnant women exposed to ProfSThangaratinamPhD), anaesthesia for an obstetric procedure in countries categorised as low-income or middle-income by the World Bank. MultidisciplinaryEvidence We excluded studies in high-income countries, those involving non-pregnant women, case reports, and studies SynthesisHub(mEsh) (ProfJZamora,ProfKSKhan, published before 1990 to ensure that the estimates reflect the current burden of the condition. Two independent ProfSThangaratinam),Barts reviewersundertookqualityassessmentanddataextraction.Wecomputedoddsratiosforriskfactorsandanaesthesia- andTheLondonSchoolof related complications, and pooled them using a random effects model. This study is registered with PROSPERO, MedicineandDentistry,Queen MaryUniversityofLondon, number CRD42015015805. London,UK;Clinical BiostatisticsUnit,Hospital Findings 44 studies (632556 pregnancies) reported risks of death from anaesthesia in women who had an obstetric RamonyCajal(IRYCIS, surgical procedure; 95 (32149636 pregnancies and 36144 deaths) provided rates of anaesthesia-attributed deaths as a CIBERESP),Madrid,Spain proportion of maternal deaths. The risk of death from anaesthesia in women undergoing obstetric procedures was (ProfJZamora, 1·2 per 1000 women undergoing obstetric procedures (95% CI 0·8–1·7, I²=83%). Anaesthesia accounted for 2·8% DArroyo-ManzanoMSc); NorthwickParkHospital, (2·4–3·4, I²=75%) of all maternal deaths, 3·5% (2·9–4·3, I²=79%) of direct maternal deaths (ie, those that resulted Harrow,Middlesex,UK from obstetric complications), and 13·8% (9·0–20·7, I²=84%) of deaths after caesarean section. Exposure to general (KDharmarajahMRCOG);School anaesthesia increased the odds of maternal (odds ratio [OR] 3·3, 95% CI 1·2–9·0, I²=58%), and perinatal deaths (2·3, ofHealthandRelatedResearch, UniversityofSheffield, 1·2–4·1, I²=73%) compared with neuraxial anaesthesia. The rate of any maternal death was 9·8 per 1000 anaesthetics Sheffield,UK(MWilsonMD); (5·2–15·7, I²=92%) when managed by non-physician anaesthetists compared with 5·2 per 1000 (0·9–12·6, I²=95%) BartsHealthNHSTrust, when managed by physician anaesthetists. Whitechapel,London,UK (RNavaratnarajahMRCOG);and SchoolofClinicaland Interpretation The current international priority on strengthening health systems should address the risk factors ExperimentalMedicine,College such as general anaesthesia and rural setting for improving anaesthetic care in pregnant women. ofMedicalandDentalSciences, UniversityofBirmingham, Funding Ammalife Charity and ELLY Appeal, Bart’s Charity. Birmingham,UK (ProfACoomarasamyMD) Correspondenceto: Copyright © Sobhy et al. Open Access article distributed under the terms of CC BY. ProfKhalidSKhan,Women’s HealthResearchUnit, MultidisciplinaryEvidence Introduction estimates are available of maternal deaths from obstetric SynthesisHub(mEsh),Bartsand A quarter of a million women die every year during or anaesthesia, or of overall maternal mortality attributable after pregnancy and childbirth, and 99% of these are from to anaesthesia, in low-income and middle-income TheLondonSchoolofMedicine 1 andDentistry,QueenMary low-income and middle-income countries. Anaesthetic countries. Factors that contribute to maternal and UniversityofLondon, interventions are an integral part of emergency obstetric perinatal mortality in women exposed to anaesthesia in LondonE12AD,UK care. However,thereisapaucityofphysiciananaesthetists low-income and middle-income countries need to be k.s.khan@qmul.ac.uk in many of the poorest countries, with an estimated ratio identified. of one physician anaesthetist per million women. There Individual studies have provided varied and imprecise is also a lack of infrastructure, drugs, and equipment. results, with up to a fifth of all direct maternal deaths 6 The need for safe, affordable surgery and anaesthesia attributed to anaesthesia-related procedures. Systematic in low-income and middle-income countries is reviews report estimates of complications in all recognised, with perioperative death as a global safety individuals exposed to anaesthesia, not specifically in 4 7 indicator. In high-income countries, very few maternal pregnant women. We undertook a systematic review to deaths are attributed to anaesthesia. 5However, no robust obtain precise estimates of anaesthesia-attributed deaths www.thelancet.com/lancetgh Vol 4 May 2016 e320 Articles Researchincontext Evidencebeforethisstudy anaesthesiatooverallmaternalmortality,particularlyindeaths Existingsystematicreviewsonglobalcausesofmaternaldeath relatedtocaesareansection(13·8%).Weidentifiedtherisk havenotpreviouslyassessedtheroleofobstetricanaesthesia. factorsformaternaldeathsrelatedtoanaestheticexposure Arecentsystematicreviewofanaesthesia-relatedmortalityin suchasgeneralanaesthesiaandruralsetting.Weprovided low-incomeandmiddle-incomecountriesfocusedonriskstothe estimatesofdeathrelatedtoanaesthesiaaccordingtothetype generalpopulation,andnotspecificallytopregnantwomen, ofanaestheticpractitionerinlow-incomeandmiddle-income whoareathighrisk.Individualobservationalstudiesvaryintheircountries.Abouttwo-thirdsofreporteddeathsfrom estimatesforanaesthesia-relatedmaternaldeathsinlow-income anaesthesiawereduetopreventablecomplicationsrelatedto andmiddle-incomecountriesandrelevantriskfactors. airwaymanagementandpulmonaryaspiration. Addedvalueofthisstudy Implicationsofalltheavailableevidence Wehaveprovidedrobustestimatesofriskofanaesthesia- Ongoingandfutureeffortstoimprovethesafetyofobstetric attributedmaternaldeathinpregnantwomenwhohave anaesthesiainlow-incomeandmiddle-incomecountries obstetricproceduresinlow-incomeandmiddle-income shouldtargettheriskfactorsidentifiedinourreviewtoimprove countriesoverall,andinvariousgeographicregionsgroupedby training,infrastructure,andprovisionofresources. incomestatus.Wehavehighlightedthecontributionof in pregnant women exposed to anaesthesia and to countries, those including non-pregnant women, case identify the factors linked to adverse outcomes in reports, and studies published before 1990 to ensure that pregnant women exposed to anaesthesia in low-income the estimates reflect the current burden of the condition. and middle-income countries. We defined anaesthesia-attributed complications as those that occurred directly as a result of anaesthesia (as Methods established by the primary study authors), and Searchstrategyandselectioncriteria anaesthesia-related outcomes as those that were directly In this systematic review and meta-analysis, we used a or indirectly associated with anaesthesia. Maternal prospective protocol (PROSPERO CRD42015015805) in 8 mortality was defined as the death of a woman during line with current recommendations, and reported as per pregnancy or at any time until 42 days after delivery, 9 the PRISMA guidelines. irrespective of the duration and site of the pregnancy, as We searched MEDLINE, Embase, Scopus, the defined by WHO. This definition included deaths from Cumulative Index to Nursing and Allied Health any cause related to or aggravated by pregnancy and its Literature (CINAHL), Web of Science, and the WHO management, but not from accidental or incidental 11 Library and Global Index Medicus from inception until causes. Direct maternal deaths were those that resulted Oct 1, 2015. We used MeSH headings, text words, and from obstetric complications; indirect maternal deaths word variants for “pregnancy” and combined them with from disorders aggravated by physiological effects of terms for low-resource countries such as “low-income” pregnancy, by pre-existing disease, or by diseases that 11 or “middle-income” or “developing country”. We developed during pregnancy. We grouped direct and combined these with terms related to anaesthesia and indirect maternal deaths together as overall maternal surgery such as “an(a)esthesia” or “an(a)esthetist” death. or “nurse an(a)esthetist” or “c(a)esarean section” Perinatal death included any fetal death that occurred SeeOnlineforappendixppendix p 1). There were no language restrictions. after 28 completed weeks of gestation, stillbirths, and Additionally, we searched the reference lists of the earlyneonataldeathsupto1weekafterbirth. Weclassed included studies and relevant reviews for eligible studies.Apgar scores as low if they were less than or equal to 7 at We selected studies in two stages. In the first stage, we 1 and 5 min. We accepted the primary study authors’ screened the titles and abstracts of all citations for definitions for maternal and fetal complications such as potentially relevant papers. In the second, we assessed post-partum haemorrhage, cardiac arrest, and admission the full texts of the retrieved papers. Two independent to the intensive care unit. reviewers (SS, KD) selected the papers against prespecified inclusion criteria. Any discrepancies were Studyqualityassessmentanddataextraction resolved after discussion with a third reviewer (ST). Two independent reviewers (SS and KD) undertook Studies were included if they assessed maternal and study quality assessment and data extraction, and any perinatal outcomes in pregnant women exposed to discrepancies were resolved with input from the third anaesthesia for an obstetric procedure in countries reviewer (ST). For studies of rates of anaesthesia- categorised as low-income and middle-income countries attributed maternal death, we assessed the following by the World Bank. 1We excluded studies in high-income criteria: representativeness of the population, sample e321 www.thelancet.com/lancetgh Vol 4 May 2016 Articles selection, outcome assessment, adequacy of sample size, middle, or upper middle), setting (urban or rural), and year and ascertainment of the cause of maternal death to of publication (before and after 2000). We assessed the anaesthesia. 13,We deemed a study to be adequate for effectsofstudyquality(loworhigh)anddesign(prospective representativenessifitincludedinstitutionsfromvarious or retrospective) on the maternal mortality rates. We used settings such as rural and urban hospitals in a region or multilevelrandomeffectslogisticmodels,andincludedthe country, and to be inadequate if it included only one above factors. The meta-regressions were run as separate hospital or unit. We classed sample selection as adequate univariate analyses. We did sensitivity analysis by limiting if all deliveries or maternal deaths were included, and as our findings to only direct maternal deaths. We also inadequate if a particular group of women were excluded. assessed the proportion of all maternal deaths during or We deemed outcome assessment to be adequate when a after caesarean section attributed to anaesthesia. confidential inquiry, verbal autopsy, or professional panel We assessed for publication bias and the effects of small established the cause of death and inadequate when studies using funnel plots, and Begg’s 18and Egger’s 19 20 there was no special effort or use of registry data from tests. All analyses were done with Stata (version 13). only one source. An adequate sample size included data for at least 10000 births. We regarded studies that Roleofthefundingsource accounted for the cause of death in at least 95% of The funder of the study had no role in the study design, maternal deaths to be adequate for ascertainment of data collection, data analysis, data interpretation, or cause of death. A study was classed as hig13quality if writing of the report. The corresponding author had full three of the above five criteria were met. access to all the data in the study and had final For comparative studies, we used the Newcastle-Ottawa responsibility for the decision to submit for publication. scaletoestablishtheriskofbiasinselection,comparability of cohorts, and outcome assessment. Studies that scored Results four stars for selection, two stars for comparability, and From 11782 citations, we included 140 studies. 44 studies three stars for ascertainment of the outcome were (632556 pregnancies) provided data for risk of death from regarded to have a low risk of bias. Studies with two or anaesthesia in women undergoing obstetric surgical three stars for selection, one for comparability, and two procedures, and 95 studies (32149636 pregnancies, for outcome ascertainment were considered to have a 36144 deaths) reported anaesthesia-attributed maternal medium risk of bias. We deemed any study with a score mortality as a proportion of maternal deaths. 25 studies of one for selection or outcome ascertainment, or zero for 15 any of the three domains, to have a high risk of bias. 11782citationsidentified To compute corresponding proportions for individual 11722citationsidentifiedfromelectronicdatabases studies, we extracted data for the number of women 60citationsidentifiedthroughreferencesand othersources exposed to anaesthesia, total and direct maternal deaths, anddeathsduringoraftercaesareansection.Weobtained information about the number of events (anaesthesia- 11312articlesexcluded 1450duplicates related maternal death and pregnancy complications) in 9862withexclusioncriteria women exposed and unexposed to risk factors such as type of anaesthesia (neuraxial or general), setting (urban 470citationsincludedforfulltextreview or rural), and practitioner (physician or non-physician). Dataanalysis 330studiesexcluded We computed odds ratios for various risk factors and 103notprimarystudies anaesthesia-related complications in individual studies, 10systematicreviews 16 93narrativereviews, and pooled them using a random effects model. We commentary used Peto odds ratios when the numbers of events were 74inappropriateexposuresns too few. We assessed heterogeneity with the I² statistic. 75inappropriateoutcomes 6articlesnotavailable When comparative data were not available, we reported the proportion of complications for each risk factor separately, and provided summary estimates. 140studiesincluded Summary rates of risk of death from anaesthesia in 25studies*(4149pregnancies)of pregnancy were reported as deaths per 1000 women 44studies*(632556pregnancies)ofcomplications undergoing obstetric procedures. We also reported anaesthesia-attributeddeathinpregnantwomen undergoingsurgicalprocedure anaesthesia-attributeddeathsasaproportionofallmaternal 95studies*(32149636pregnancies)ofratesof deaths (direct and indirect). We did subgroup analysis and anaesthesia-attributedmaternalmortality meta-regression for the following factors that were prespecified before the analysis: geographical location Figure1:Studyselection (World Bank classification), country income (low, lower *Somestudieshavebeenusedinmorethanonecategory. www.thelancet.com/lancetgh Vol 4 May 2016 e322 Articles The risk of maternal death directly attributed to A Representativenessofpopulation 49 46 anaesthesia after obstetric procedures was reported in 44 studies from 15 low-income and middle-income Sampleselection 85 7 3 countries, which were grouped into the following regions: sub-Saharan Africa (n=38 studies), south Asia Outcomesassessment 13 23 59 (n=4), and east Asia and the Pacific (n=2). Most studies Samplesize>1000 74 17 4 were facility based (42 of 44) and in nearly three-quarters of studies (31 of 44), women were managed in an urban <5%ofmaternaldeathsunaccounted 59 23 13 setting. Of the 95 studies (31 countries) that reported Overallriskofbias 65 30 anaesthesia-attributed mortality as a proportion of all 0 20 40 60 80 100 maternal deaths, 52 provided facility-based data, and 29 provided countrywide data. In 45 studies, women Proportionofstudies(%) Lowriskofbias Highriskofbias Unclearriskofbias weremanagedinanurbansetting(appendixp3andp20). Studies compared the odds of adverse maternal and B fetal outcomes for risk factors such as the type of Selection 11 12 2 anaesthesia (25 studies, 414069 pregnancies), setting (one Comparability 4 1 20 study, 8070 pregnancies), and anaesthesia provider (one study, 8070 pregnancies). Rates of any maternal death in Ascertainmentofoutcome 14 11 anaesthesia administered by a non-physician were assessed in eight studies (27714 pregnancies), and by a 0 20 40 60 80 100 Proportionofstudies(%) physician anaesthetist in six studies (20313 pregnancies). Both high-risk and low-risk women were studied, and Lowriskofbias Mediumriskofbias Highriskofbias caesarean section was the most common surgical Figure2:Riskofbiasassessment procedure. Studies ascertained the cause of maternal (A)Studiesofanaesthesia-attributedmaternalmortalityrates.(B)Studiesevaluatinganaesthesia-relatedrideaths and exposure to anaesthesia from theatre records, factorsandmaternalandfetalcomplications. patient notes, facility and countrywide maternal death reviews, and verbal autopsies. Number Number Numberof Maternal 95%CI I2 Meta- 65 (68%) of 95 included studies of anaesthesia- of of women deathsper regression attributed maternal mortality had low risk of bias. About studies deaths undergoing 1000women pvalue surgical undergoing half had high risk of bias for representativeness of the population and setting, and 90% had adequate sample procedures surgery Overall 44 264 632556 1·2 0·82–1·7 83% .. selection,andaquarterhadhighriskofbiasforoutcomes reporting (figure 2A). Three-quarters of all studies had WorldBank regions adequate sample size and about two-thirds adequately accounted for maternal deaths (figure 2A). SouthAsia 4 16 37132 0·34 0·13–0·90 71% 0·004 Sub-Saharan 38 237 567431 1·5 1·1–2·2 85% Four-fifths of studies of risk factors for complications Africa in women exposed to obstetric anaesthesia had high risk EastAsiaand 2 11 27993 0·40 0·22–0·71 .. of bias (figure 2B). Nearly half of the included studies Pacific had low risk of bias for study selection; a fifth of studies Year had low or medium risk of bias for comparability of the <2000 17 55 49232 1·2 0·76–1·7 50% 0·74 cohorts, and more than half of studies had low risk of ≥2000 27 209 583324 1·2 0·74–2·1 86% bias for ascertainment of the outcome (figure 2B). Setting In women undergoing an obstetric procedure, the risk Urban 31 85 72203 1·5 1·0–2·3 67% 0·02 ofdeathattributedtoanaesthesiawas1·2per1000women Rural 1 1 69 14·5 0·00–42·7 .. (95% CI 0·82–1·7, I²=83%), with the highest rates in Both 12 178 560284 0·67 0·37–1·2 89% sub-Saharan Africa (1·5 per 1000 women, 1·1–2·2, Countryincome I²=85%; table 1). The appendix provides estimates for Low 13 53 33431 1·5 0·84–2·8 63% 0·003 individual countries (appendix p 14). Caesarean section Lowermiddle 27 90 80295 1·4 0·92–2·2 72% wasthesurgicalproceduredonein37studies,comprising 97% (611291/632556) of the included women. Seven Uppermiddle 4 121 518830 0·23 0·20–0·28 24% studies either reported other obstetric procedures (eg, Table1:Mortalityratesfromanaesthesiainwomenundergoingobstetricsurgicalproceduresin cervical cerclage, laparotomy for ectopic pregnancy, low-incomeandmiddle-incomecountries dilatation and curettage, and removal of retained placenta), or failed to specify the type of obstetric surgery. (414069 pregnancies) assessed the association between Subgroup analysis and meta-regression showed a anaesthesia-related risk factors and complications in significant difference between regions (p=0·004). The women undergoing obstetric procedure (figure 1). risks of death from anaesthesia were higher in rural than e323 www.thelancet.com/lancetgh Vol 4 May 2016 Articles urban settings (p=0·02), and in low-income and lower- Number Numberof Total Anaesthesia-95%CI I Meta- middle-income than upper-middle-income countries ofstudies deathsfrom number attributed regression (p=0·003). There were no differences for year of anaesthesia of mortality pvalue publication (p=0·74; table 1). maternal Anaesthesia was reported as the main cause of death in deaths 2·8% (95% CI 2·4–3·4, I²=75%) of all maternal deaths Overall 95 987 36144 2·8% 2·4–3·4 75% (direct and indirect), with the highest rates in Middle WorldBank regions* East and north Africa (6·2%, 3·9–9·7, I²=86%), and the lowest in east Asia and Pacific (1·5%, 0·9–2·3, I²=63%; Sub-Saharan 50 675 24873 2·9% 2·3–3·6 61% 0·004 Africa table2).Theappendixprovides estimates from individual SouthAsia 18 89 4317 2·4% 1·5–3·8 79% countries (appendix p 14). Anaesthesia was reported as the cause of death in 3·5 % (95% CI 2·9–4·3, I²=79%) of MiddleEast 10 136 2555 6·2% 3·9–9·7 86% andnorth direct maternal deaths (76 studies, 20780 deaths, Africa 26750727 pregnancies), and 13·8% (95% CI 9·0–20·7, EastAsiaand 9 49 3276 1·5% 0·9–2·3 63% Pacific I²=84%) of all deaths that occurred during or after caesarean section (31 studies, 1028 deaths; Europeand 4 14 455 3·0% 1·3–6·7 64% centralAsia appendix pp 17–18). LatinAmerica 4 24 668 3·6% 2·4–5·3 0% Meta-regression showed a significant difference in the overall anaesthesia-attributed mortality rates by andthe Caribbean geographical region (p=0·004) and year of publication Year (p=0·002). We noted no significant differences by setting, <2000 28 250 6589 4·1% 3·0–5·5 77% 0·002 study design, income level, or study quality (table 2). >2000 67 737 29555 2·4% 1·9–2·9 72% Compared with neuraxial anaesthesia, administration Setting 0·29 of general anaesthesia tripled the odds of maternal death Rural 8 14 894 1·9% 0·9–3·8 38% (OR 3·3, 95% CI 1·2–9·0, I²=58%), with mortality rates Urban 45 245 7987 3·3% 2·5–4·2 72% of5·9per1000and1·2per1000forgeneralandneuraxial Both 42 728 27263 2·5% 1·9–3·4 80% anaesthesia, respectively. General anaesthesia also Studydesign 0·68 doubled the odds of perinatal death (odds ratio [OR] 2·3, Prospective 9 35 1819 2·5% 1·1–5·6 77% 95% CI 1·2–4·1, I²=73%) compared with neuraxial Retrospective86 952 34325 2·9% 2·4–3·4 75% anaesthesia.Therewasanincreaseinothercomplications Countryincome 0·57 such as post-partum haemorrhage (OR 9·4, 95% CI Low 17 89 3171 2·6% 1·9–3·6 35% 5·1– 17·2, I²=65%), low Apgar score at 1 min (6·3, 2·9–13·6, I²=64%) and at 5 min (3·2, 2·3–4·4, I²=0%) Lowermiddle 38 199 8130 2·6% 1·9–3·6 78% Uppermiddle 40 699 24843 3·1% 2·4–4·1 78% with general compared to neuraxial anaesthesia (figure 3). There were no differences in cardiac arrest Studyquality 0·12 High 65 859 32099 2·6% 2·1–3·2 79% (OR 3·6, 95% CI 0·6–20·6, I²=71%) between the two groups. Management in a rural setting was associated Low 30 128 4045 3·6% 2·6–4·8 59% with an increase in the odds of maternal death (2·1, *Basedon2015WorldBankdata. 1·2–3·7) compared with an urban setting. 21 No direct comparative data were available for physician Table2:Anaesthesia-attributedmaternalmortalityinlow-incomeandmiddle-incomecountries versus non-physician providers of anaesthetic care. The overall risk of any maternal death when non-physicians ventilation difficulties, and hypoxia; 38 (31%) from provided care was 9·8 per 1000 (95% CI 5·2–15·7, pulmonary aspiration; 34 (27%) from issues related to I²=92%), and the rates of anaesthesia-attributed maternal staff competency, poor pre-assessment, intraoperative deaths was 1·8 per 1000 (0·25–4·3, I²=85%). The monitoring,andequipmentfailure.Othercausesincluded corresponding estimates for physician anaesthetists were cardiac arrest at induction or during the procedure (seven 5·2 per 1000 (0·9–12·6, I²= 95%) for any maternal death, [6%]), high spinal anaesthesia (eight [6%]), and drug and 1·3 per 1000 (0·16–3·1, I²=79%) for anaesthesia- overdose or adverse reactions (seven [6%]). attributed maternal deaths, respectively (appendix p 13). With regard to publication bias and small studies One study reported an increase in the odds of maternal effects, a funnel plot of studies reporting risk of death deaths (OR 2·7, 95% CI 1·6–4·6) when maternal care was from anaesthesia in women undergoing obstetric managed by non-physician anaesthetists without formal procedure showed significant asymmetry (Egger’s test structured training compared to those with training. 21 p=0·001), although Begg’s test failed to confirm this The underlying causes were reported for 124 maternal effect. There was no asymmetry in a funnel plot of deaths (24 studies). 56 (45%) of all deaths resulted from studies that reported anaesthesia-attributed mortality as airway complications such as difficult or failed tracheal a proportion of all maternal deaths (Egger’s and Begg’s intubation, oesophageal intubation, bronchospasm, tests p>0·05; appendix p 19). www.thelancet.com/lancetgh Vol 4 May 2016 e324 Articles Number GeneralanaesthesiaRegionalanaesthesia OR(95%CI) I² ofstudies Number Number Number Number ofevents ofwomen ofevents ofwomen Maternaldeath 9 138 23256 25 21035 3·3(1·2–9) 58% Perinataldeath 5 860 4882 181 3459 2·3(1·2–4·1) 73% Post-partumhaemorrhage 4 183 2923 302 67694 9·4(5·1–17·2)65% Intensivecareunitadmissionofmother 319 14243 126 286682 34·4(8·9–133·2)5% Cardiacarrest 4 21 8234 10 15130 3·6(0·62–20·6)71% LowApgarscoreat5min 7 282 909 62 919 3·2(2·3–4·4) 0% LowApgarscoreat1min 6 254 303 57 395 6·3(2·9–13·6)64% 0·1 1 10 100 FavoursgeneralFavoursregional anaesthesia anaesthesia Figure3:Maternalandfetalcomplicationsinwomenexposedtogeneralversusneuraxialanaesthesiainlow-incomeandmiddle-incomecountries Discussion higher than current estimates because of scarce data from Anaesthesia contributes disproportionately to maternal these low-income and middle-income countries with high mortality in low-income and middle-income countries. maternal mortality and poor health-care resources. We About one in seven maternal deaths during or after used the 2015 World Bank atlas for classification of a caesarean section was due to anaesthesia, a very high country’s income status, and adjustment for year of 5 mortality rate compared with developed countries. publication could have resulted in some differences in Exposure to general anaesthesia, and administration of classification. We noted significant heterogeneity in our anaesthesia by non-physicians, especially those with no findings despite adjusting for various factors because of formal training, were major risk factors for maternal variations in the characteristics of population, setting, type deaths from anaesthesia. We have mapped the safety of of anaesthesia, availability of caesarean section, and obstetric anaesthesia across various economic regions and provider. Few studies provided detailed reports on the individual countries. Most studies were from the sub- underlying cause of death from anaesthesia. Saharan African region, which also had the highest risk of The asymmetry recorded in the funnel plot was the deaths from anaesthesia in women undergoing surgery. result of high rates of maternal deaths in small studies Ours is the first review, to our knowledge, to making them more likely to be published, similar to the comprehensively assess the risk factors for maternal and effect reported in many non-comparative reviews on perinatal deaths and complications from anaesthesia in proportions. 23 This asymmetry could have slightly low-income and middle-income countries, and the overestimated the recorded maternal death rates. overall risk of maternal death from anaesthesia. Our However, the magnitude of the bias is probably small in estimates of the risk in low-income and middle-income view of the low weight of these studies in the analysis. countries are significantly higher than those reported in Pregnant women requiring general anaesthesia need high-income countries such as the USA, where the case tracheal intubation to ensure the airway is secure from fatality rate from general and regional anaesthesia given aspiration. Compared with the general surgical for caesarean section were 6·5 and 3·8 per million population, pregnant women are at increased risk of 22 anaesthetics, respectively. We assessed the extent of the complications from general anaesthesia, with eight times problem in detail by assessing rates of death in women higher risk of failed intubation and its associated who had surgery, and as a proportion of any, direct, and hazards. 24,2Our findings support existing data for the caesarean-section-related maternal deaths. We reported role of airway complications, and pulmonary aspiration the effects of study quality on mortality estimates. We of gastric contents as major causes of death from 26 looked for variations in anaesthesia-attributed maternal anaesthesia, and the need for specific training. mortality rates according to economic regions, individual The increased mortality and morbidity that we countries, setting, year, and anaesthesia provider. identified with general anaesthesia could be due to the Our findings were limited by the differences in quality following reasons: inadequate training and resources, and reporting of outcomes in the studies. Studies focused poor general condition of the mother, or concomitant mainly on assessing the risks associated with type of complications such as post-partum haemorrhage. The anaesthesia, and less on other factors, which limited our low Apgar scores associated with general anaesthesia synthesis, and we could only provide rates for these risk exposure could be indicators of neonatal and anaesthetic factors separately. Fewer studies were published in low- facilities, including limited access to modern, volatile income countries that are outside sub-Saharan Africa. The anaesthetic agents that minimise fetal respiratory 27 actual rates of anaesthesia-attributed deaths are probably depression. The increased blood loss associated with e325 www.thelancet.com/lancetgh Vol 4 May 2016 Articles exposure to general anaesthesia is similar to the findings MW provided expert anaesthetic advice. SS and ST prepared the initial of the Cochrane review, which reported higher blood loss drafts of the manuscript, with additional input from KSK, AC, MW, and RN. with general than regional anaesthesia in pregnant All authors contributed to the drafts and final version of the manuscript. women undergoing caesarean section. 28 Declarationofinterests In high-income countries such as the USA, no We declare no competing interests. Acknowledgments measured differences were recorded in anaesthetic complications between physician and non-physician This study was funded by Ammalife Charity (Registered UK Charity 29–31 1120236) and ELLY Appeal, Bart’s Charity (Registered UK Charity anaesthetists. However, compared with the rigorous 212563). We thank Angela Enright, Clinical Professor of Anaesthesia at additional training provided to non-physician the University of British Columbia, for her input into the report. anaesthetists in the USA, their counterparts in low- References income and middle-income countries have very little 1 Say L, Chou D, Gemmill A, et al. Global causes of maternal death: training, which varies between countries. This a WHO systematic analysis. 2014; 2: e323–33. 2 Clyburn P, Morris S, Hall J. Anaesthesia and safe motherhood. discrepancy could have contributed to the recorded Anaesthesia 2007; 62: 21–25. increase in maternal mortality when anaesthesia was 3 Hoyler M, Finlayson SR, McClain CD, Meara JG, Hagander L. Shortage of doctors, shortage of data: a review of the global surgery, given by non-physician anaesthetists than their physician obstetrics, and anesthesia workforce literature. World J Surg 2014; counterparts; the risks were also high for non-physicians 38: 269–80. 21 without adequate training. Many of the reported causes 4 Meara JG, Leather AJM, Hagander L, et al. Global Surgery of anaesthesia-attributed deaths such as complications of 2030: evidence and solutions for achieving health, welfare, and airway management, pulmonary aspiration with general economic development. Lancet 2015; 386: 569–624. 5 McClure JH, Cooper GM, Clutton-Brock TH, Centre for Maternal anaesthesia, and hypotension and high spinal with and Child Enquiries. Saving mothers’ lives: reviewing maternal neuraxial anaesthesia are preventable with appropriate deaths to make motherhood safer: 2006–8: a review. Br J Anaesth 32 2011; 107: 127–32. training and resources. 6 Larsen JV, Janowski KA, Krolikowski A. Maternal mortality in The global definition and classification of anaesthesia- hospitals in Zululand, July 1993 June 1994. S Afr Med J 1996; 86: 424–30. attributed deaths need standardisation to identify the real burden.Anaesthetistsshouldbepartofthepanelanalysing 7 Bainbridge D, Martin J, Arango M, Cheng D, Evidence-based thecausesofmaternaldeaths,andthelevelofcontribution Peri-operative Clinical Outcomes Research Group. Perioperative and 33 anaesthetic-related mortality in developed and developing countries: of anaesthesia to the death should be reported clearly. a systematic review and meta-analysis. Lancet 2012; 380: 1075–81. Strategies to reduce maternal mortality should include 8 Sobhy S, Thangaratinam S, Zamora J, Damaraajah K. Anaesthesia related maternal and perinatal mortality and morbidity in increasing the number of anaesthetic practitioners developing countries: a systematic review and meta-analysis. managing pregnancy, enhancing the resources available to Prospero 2015; CRD42015015805. 9 Moher D, Leberati A, Tetzlaff J, Altman DG, Group PRISMA. them, and increasing their level of training in low-income Preferred reporting items for systematic reviews and meta-analyses: and middle-income countries. Implementation of simple the PRISMA statement. PLoS Med 2009; 6: e1000097. measures,suchastheWHOSaferSurgerychecklistbefore 10 World Bank. Country and lending groups. http://data.worldbank. and during surgery, and access to simple monitoring org/about/country-and-lending-groups (accessed Oct 31, 2015). technology such as pulse oximetry, could potentially 11 WHO. ICD-10, international statistical classification of diseases and 34 related health problems, 10th revision. Geneva: World Health reduce adverse outcomes. The introduction of these Organization, 2010. measures in anaesthesia providers in high-income 12 WHO. Maternal and perinatal health. http://www.who.int/ maternal_child_adolescent/topics/maternal/maternal_perinatal/ countries has reduced maternal death from anaesthetic en/ (accessed Aug 10, 2014). complications to very low levels. Recent global initiatives 13 Khan KS, Wojdyla D, Say L, Gülmezoglu AM, Van Look PFA. WHO such as Lifebox, and training courses done in partnership analysis of causes of maternal death: a systematic review. Lancet FormoreonLifeboxsee 36 with non-profit organisations such as Kybele, have 2006; 367: 1066–74. http://www.lifebox.org/ focusedeffortstoimprovethesafetyofsurgicalprocedures. 14 Glasziou P, Irwig L, Bain C, Colditz G. Frequency and rate. FormoreonKybeleseehttps:// Systematic reviews in health care: a practical guide (2nd edn). www.oaa-anaes.ac.uk/Kybele Governmental and non-governmental organisations Cambridge, UK: Cambridge University Press, 2001. 67–73. should prioritise investment in obstetric anaesthesia, to 15 Wells GA, Shea B, O’Connell D, et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in implement the World Health Assembly’s resolution to meta-analyses. http://www.ohri.ca/programs/clinical_ include emergency and essential surgical care and epidemiology/oxford.asp (accessed Aug 10, 2014). anaesthesia as a component of universal health coverage. 37 16 DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986; 7: 177–88. In conclusion, anaesthesia is a major contributor to maternal deaths in pregnant women undergoing surgery 17 Brockhaus AC, Bender R, Skipka G. The Peto odds ratio viewed as a new effect measure. Stat Med 2014; 33: 4861–74. in low-income and middle-income countries. Targeted 18 Begg CB, Mazumdar M. Operating characteristics of a rank efforts are needed to provide safe obstetric anaesthesia by correlation test for publication bias. Biometrics 1994; 50: 1088–101. 19 Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta- improving training, infrastructure, and resources. analysis detected by a simple, graphical test. BMJ 1997; 315: 629–34. Contributors 20 StataCorp. Stata Statistical Software: release 12. College Station, TX: ST, JZ, and SS came up with the research question and designed the StataCorp LP, 2015. protocol. SS and KD did the literature search, study selection, and data 21 Fenton PM, Whitty CJM, Reynolds F. Caesarean section in Malawi: extraction with the help of ST. JZ and DA-M did statistical analysis. prospective study of early maternal and perinatal mortali. BMJ SS designed the tables, figures, and appendices, with input from ST. 2003; 327: 587–90. www.thelancet.com/lancetgh Vol 4 May 2016 e326 Articles 22 Hawkins JL, Chang J, Palmer SK, Gibbs CP, Callaghan WM. 30 Needleman J, Minnick AF. Anesthesia provider model, hospital Anesthesia-related maternal mortality in the United States: resources, and maternal outcomes. Health Serv Res 2009; 1979–2002. Obstet Gynecol 2011; 117: 69–74. 44: 464–82. 23 Yousef F, Cardwell C, Cantwell MM, Galway K, Johnston BT, 31 Lewis SR, Nicholson A, Smith AF, Alderson P. Physician Murray L. The incidence of esophageal cancer and high-grade anaesthetists versus non-physician providers of anaesthesia for dysplasia in Barrett’s esophagus: a systematic review and meta- surgical patients. Cochrane Database Syst Rev 2014; 7: CD010357. analysis. Am J Epidemiol 2008; 168: 237–49. 32 Pinder A. Complications of obstetric anaesthesia. 24 Hawkins JL. Maternal mortality: anesthetic implications. Curr Anaesth Crit Care 2006; 17: 151–62. Int Anesthesiol Clin 2002; 40: 1–11. 33 Dalina AM, Inbasegaran K. Anaesthetic related maternal deaths in 25 Samsoon GL, Young JR. Difficult tracheal intubation: a Malaysia: a review. Med J Malaysia 1996; 51: 52–63. retrospective study. Anaesthesia 1987; 42: 487–90. 34 Walker IA, Reshamwalla S, Wilson IH. Surgical safety checklists: 26 Cook TM, Woodall N, Frerk C, Fourth National Audit Project. Major do they improve outcomes? Br J Anaesth 2012; 109: 47–54. complications of airway management in the UK: results of the 35 Sama HD, Maman AF, Walker IA. Incidence of hypoxia and related Fourth National Audit Project of the Royal College of Anaesthetists and the Difficult Airway Society. Part 1: anaesthesia. Br J Anaesth events detected by pulse oximeters provided by the Lifebox Foundation in the maternity unit at Sylvanus Olympio University 2011; 106: 617–31. Teaching Hospital, Togo. J Anesth 2015; 29: 971–73. 27 Clyburn P, Collins R, Harries S. General anaesthesia. Obstetric 36 SAFE. Obstetric anaesthesia course. http://www.aagbi.org/sites/ Anaesthesia for Developing Countries. Oxford, UK: Oxford default/files/Report of Uganda Safe Course June- July 2011_0.pdf. University Press, 2010. (accessed April 4, 2016). 28 Afolabi BB, Lesi FE. Regional versus general anaesthesia for 37 WHO. 68th World health assembly resolution: Strengthening caesarean section. Cochrane Database Syst Rev 2012; 10: CD004350. Emergency and Essential Surgical Care and Anaesthesia as a 29 Simonson DC, Ahern MM, Hendryx MS. Anesthesia staffing and Component of Universal Health Coverage. http://apps.who.int/gb/ anesthetic complications during caesarean delivery: a retrospective ebwha/pdf_files/WHA68/A68_R15-en.pdf?ua=1 (accessed analysis. Nurs Res 2007; 56: 9–17. April 4, 2016). e327 www.thelancet.com/lancetgh Vol 4 May 2016