07-047076 ccare

7/27/2019 07-047076 CCare

1/9

30 Bull World Health Organ2009;87:3038 | doi:10.2471/BLT.07.047076

Objective To evaluate the eect o a national reerral system that aims to reduce maternal mortality rates through improving accessto and the quality o emergency obstetric care in rural Mali (sub-Saharan Arica).Methods A maternity reerral system that included basic and comprehensive emergency obstetric care, transportation to obstetrichealth services and community cost-sharing schemes was implemented in six rural health districts in Kayes region between December2002 and November 2005. In an uncontrolled beore and ater study, we recorded all obstetric emergencies, major obstetricinterventions and maternal deaths during a 4-year observation period (1 January 2003 to 30 November 2006): the year prior to theintervention (P1); the year o the intervention (P0), and 1 and 2 years ater the intervention (P1 and P2, respectively). The primaryoutcome was the risk o death among obstetric emergency patients, calculated with crude case atality rates and crude odds ratios.Analyses were adjusted or conounding variables using logistic regression.Findings The number o women receiving emergency obstetric care doubled between P1 and P2, and the rate o major obstetricinterventions (mainly Caesarean sections) perormed or absolute maternal indications increased rom 0.13% in P1 to 0.46% in P2.

In women treated or an obstetric emergency, the risk o death 2 years ater implementing the intervention was hal the risk recordedbeore the intervention (odds ratio, OR: 0.48; 95% confdence interval, CI: 0.300.76). Maternal mortality rates decreased more amongwomen reerred or emergency obstetric care than among those who presented to the district health centre without reerral. Nearlyhal (47.5%) o the reduction in deaths was attributable to ewer deaths rom haemorrhage.Conclusion The intervention showed rapid eects due to the availability o major obstetric interventions in district health centres,reduced transport time to such centres or treatment, and reduced fnancial barriers to care. Our results show that nationalprogrammes can be implemented in low-income countries without major external unding and that they can rapidly improve thecoverage o obstetric services and signifcantly reduce the risk o death associated with obstetric complications.

Une traduction en ranais de ce rsum fgure la fn de larticle. Al fnal del artculo se acilita una traduccin al espaol.

Improved access to comprehensive emergency obstetric care

and its effect on institutional maternal mortality in rural MaliPierre Fournier,a Alexandre Dumont,b Caroline Tourigny,a Georey Dunkley c & Skou Dram c

.

a Unit de Sant Internationale, Centre de Recherche du Centre Hospitalier de lUniversit de Montral, 3875 Saint-Urbain, Montreal, Quebec, H2W 1V1, Canada.b Centre Hospitalier Universitaire Ste-Justine, Universit de Montral, Montreal, Quebec, Canada.c Direction Rgionale de la Sant, Kayes, Mali.Correspondence to Pierre Fournier (e-mail: [email protected]).(Submitted: 24 August 2007 Revised version received: 22 April 2008 Accepted: 24 April 2008 Published online: 18 November 2008)

IntroductionMaternal mortality is a major public health problem, particu-larly in sub-Saharan Arica, where hal (50.4%) o all maternaldeaths worldwide occur.1 One objective o the MillenniumDevelopment Goals is to reduce maternal mortality by 75%between 1990 and 2015.2 In 2005, the maternal mortalityratio in sub-Saharan Arica, estimated at 900 maternal deathsper 100 000 live births, was by ar the highest in the world. 1Unlike other regions, sub-Saharan Arica has not seen improve-ments in indicators linked to maternal mortality, leading toears that the Millennium Development targets will not bemet.3 In response to this disquieting situation, many Aricancountries have adopted measures towards reducing maternal

mortality.Te context and causes o maternal mortality and mor-

bidity are well known,4 and strategies to ameliorate them wererecently reported.5 One proven eective strategy is to pro-vide access to basic emergency obstetric services (parenteraloxytocics, antibiotics and anticonvulsants; assisted deliver-ies; manual extraction o the placenta; removal o retainedproducts) and, i necessary, to comprehensive emergencyobstetric services (basic services plus Caesarean sections andblood transusions).6 Access to these services is a key element

o the WHO Making Pregnancy Saer programme.7 Ensur-ing timely Caesarean delivery when needed is a priority insub-Saharan Arica.

In western Arica, maternal mortality is highest in ruralareas where access to emergency obstetric care is limited bylarge geographic distances to health acilities and scarce re-sources.8 While progress has been made in reducing maternalmortality rates in urban areas, the situation in rural areas isnot improving. In Mali, or example, the population-basedrate o Caesarean delivery in urban areas rose rom 1.6% to3.5% between 1991 and 1998, while in rural areas it remainedunchanged (1.6% and 1.5%, respectively).9 Implementingemergency obstetric care programmes, and maternity reer-

ral systems in particular, is complicated in settings whereresources are scarce.10 In western Arica, where the BamakoInitiative has made cost recovery in health care the standard,11the costs o comprehensive emergency obstetric care representa major outlay or households12,13 and several strategies havebeen attempted to reduce this nancial burden.14,15

In 2002, the Government o Mali launched a nationwidematernity reerral system16 aimed at improving the quality andaccessibility o comprehensive emergency obstetric care servicesand at reducing the danger o death associated with obstetric

7/27/2019 07-047076 CCare

2/9

31Bull World Health Organ2009;87:3038 | doi:10.2471/BLT.07.047076

ResearchMaternal mortality in rural MaliPierre Fournier et al.

complications. In this study, we aim toevaluate the eects o the system in arural population o more than one mil-lion inhabitants.

Methods

Setting

O 177 countries on the Human De-velopment Index or 2005, Mali wasranked 173rd17 and had the 17th high-est maternal mortality ratio1 Te healthregion o Kayes, situated in the west oMali, has nearly 1.7 million inhabitantsunevenly distributed over 120 760 km.Rugged terrain and periods o intenserain with ooding make communica-tions difcult. Te study area consistso six o the regions seven districts.

Te provincial district was excludedbecause it is more urban. Te studyarea contains just over 1.25 million in-habitants, and the population densityo the districts varies between 9.7 and26.2 inhabitants per km (mean o 14.7inhabitants per km).

Despite eorts in recent years, thegeographic accessibility o health ser-vices remains poor: the distance romhome to a primary health care centreis more than 5 km or 56% o thepopulation and more than 15 km or

30%. Te public health system, whichis almost the only provider o modernhealth-care services, has ew resources(one doctor per 28 000 and one mid-

wie per 96 000 inhabitants). Te studyarea has 101 community health centres(15 to 20 in each district) and six districthealth centres. Whereas the latter oercomprehensive emergency obstetriccare, community health centres providebasic obstetric services, including as-sisted deliveries. When an emergencycomplication arises at the communityhealth centre, the patient is reerredto a district health centre. Te overallutilization rate o reproductive healthservices in the study area resemblesthe national average or rural areas inMali.18

Intervention

Te maternity reerral system is a na-tional programme launched in 2002to reduce the risk o maternal deathassociated with obstetric complica-

tions.16

Te system relies on three maincomponents (Fig. 1). First, it seeks toimprove communication and transportopportunities to eliminate delays in the

Fig. 1. Components and function of referral system for comprehensive emergencyobstetric care, Mali, 20032006

Detection o risk pregnanciesand obstetrical complications

at the community health centresGovernmental

decree orCaesarean ree

o charge

Reduction o the 2nd delay(transport time)

1. Transportand communication

Improved radiocommunication betweencommunity anddistrict health servicesImproved ambulanceservice betweenlevels o care

Elimination o inancialbarriers or women

2. Funding throughcommunity cost-sharingschemes

Community healthassociationLocal governmentDistrict councilDistrict health centreWomen (patients)

Reduction o risk o deathIncrease in coverage

Component o the intervention Causality link

Improved clinicalmanagement

3. Emergencyobstetric care

Physicians withsurgical competenceOperating roomsEssential obstetric caretraining or sta atthe district health centresEquipment

Programme objective

delivery o emergency obstetric services.Funds rom overseas donors are usedto improve radio communications be-tween community health acilities anddistrict health services, as well as ambu-lance transport between them. Second,alternative unding options, includingcommunity cost-sharing schemes, areaccessed to eliminate nancial barriersto obstetric care. Community-undedschemes receive unds rom the localgovernment, local health services andcommunity health associations andthen reimburse health providers or allservices they give to women, who con-tribute only a small co-payment. Tird,training and equipment are providedto improve the clinical management oobstetric emergencies.

Te programme was designed inaccordance with national guidelines butis implemented with adaptations or

regional contexts in cooperation withlocal health partners. In the study area,the start-up period was between 2002and 2005, depending on the district.

wo categories o women use theprogrammes obstetric services: (i) those

with obstetric complications who arereerred by community health centresand have beneted rom all compo-nents o the system, and (ii) those whoare sel-reerred to the district healthcentre.

Tere are six categories o obstet-

ric emergency, dened on the basis othe medical diagnosis or the reasonor reerral: haemorrhage, uterinerupture, pre-eclampsia/eclampsia, dys-

tocic labour, inection and other (orother obstetric emergencies that causematernal death directly, such as abor-tion, or indirectly, mainly malaria andanaemia).19

Study design and statisticalanalysis

In 2004, a system or ongoing registra-

tion o obstetric emergencies was set upin all districts o the Kayes region. Dur-ing an initial pilot phase, data collection

was supported and supervised by theRegional Health Authority o Kayes andthe research team. Te system allowedor the documentation o each patientdeemed to be an obstetric emergency,and the data collected included thesociodemographic characteristics othe woman, the obstetric diagnosis andoutcome, pregnancy ollow-up, etc.Data were collected retrospectively or

the period rom 1 January 2003 to 30June 2004, ater which the data werecollected prospectively until 30 No-vember 2006. All cases with direct andindirect obstetric complications wererecorded and classied in accordance

with accepted standards.19

Because the maternity reerral sys-tem is a national programme that allregions must implement, it was neitherethical nor practical to include a controlgroup that would be denied access tothis programme in our study design.

Tereore, we used a quasi-experimentaluncontrolled beore-and-ater studydesign. Te main criterion to assess theeects o the intervention is the risk o

7/27/2019 07-047076 CCare

3/9


ResearchMaternal mortality in rural Mali Pierre Fournier et al.

Fig. 2. Data availability, by period and district, in referral system for comprehensiveemergency obstetric care, Mali, 20032006

District

P1 P0 P1 P2

5

0

6

2003

Year

2004 2005 2006 2007

4

3

2

1

Start o intervention

P1, year beore the intervention; P0, year o the intervention; P1, 1 year ater the intervention; P2, 2 years aterthe intervention.

death among obstetric emergency cases.o evaluate the efcacy o the mater-nity reerral system, we considered ourperiods: beore the intervention (P1),the year during which the intervention

was implemented (P0) and two 12-

month post-intervention periods (P1and P2). Data availability and start dateo intervention by district is shown inFig. 2.

Te maternity reerral system wasimplemented on a dierent date in eacho the six districts o the study area. Asa result, the P1 period lasted rom3 to 34 months, with a mean o 12.2months; the P0, P1 and P2 periodsall lasted 1 year, except in district 4,

where P0 lasted 11 months, and ordistrict 1, where P2 lasted 7 months.

District 3 had no P1 or P2 while thestudy lasted since implementation othe programme was delayed because othe time it took to reach a consensusregarding community cost-sharingschemes.

We checked or group comparabil-ity or the main known risk actors ormaternal death (age, cause o obstetriccomplications). Patients came romdistricts where access to health servicesvaries greatly. Districts were classiedaccording to the percentage o the

population living within 15 km roma primary health care centre (good ac-cessibility: > 85%; average accessibility:rom 60 to 85%; poor accessibility:< 60%). We compared the risk o death

Table 1. Characteristics of the study area and of the obstetric care delivered before, during and after the adoption of a nationalmaternity referral programme, Mali, 20032006

Variable P1 P0 P1 P2No. (%) No. (%) No. (%) No. (%)

Population covered (both sexes) 1 040 917 (NA) 1 169 061 (NA) 1 037 367 (NA) 976 927 (NA)Expected deliveries a 52 046 (NA) 58 453 (NA) 51 868 (NA) 48 846 (NA)Institutional deliveriesb 9 871 (19.0) 15 576 (26.6) 16 573 (32.0) 19 235 (39.4)Obstetric emergencies treated 475 (0.9) 658 (1.1) 571 (1.1) 913 (1.9)Women reerred c 143 (0.27) 273 (0.47) 246 (0.47) 452 (0.93)Women not reerred 332 (0.64) 385 (0.66) 325 (0.63) 461 (0.94)Major obstetric interventionsd 120 (0.23) 273 (0.46) 292 (0.56) 396 (0.81)Absolute maternal indicationse 66 (0.13) 124 (0.21) 156 (0.30) 225 (0.46)Non-absolute maternal indications 33 (0.06) 107 (0.18) 92 (0.18) 144 (0.29)Unknown 21 (0.04) 42 (0.07) 44 (0.08) 27 (0.06)

NA, not applicable; P1, year beore the intervention; P0, year o the intervention; P1, 1 year ater the intervention; P2, 2 years ater the intervention.a Calculated as 5% o the population; this number has been used as a denominator to calculate percentages.b Includes deliveries at community and district health centres.c Women with obstetric complications who were reerred by a community health centre to a district health centre.d Includes Caesarean section, hysterectomy, laparotomy, internal version and craniotomy/embryotomy.e Includes severe antepartum and incoercible postpartum haemorrhage, uterine prerupture/rupture, major cephalopelvic disproportions, abnormal presentations

(transverse and brow).

among obstetric emergency cases acrossthe dierent periods o the study bycalculating odds ratios (ORs) and their95% condence intervals (CIs). oevaluate the potential eects o the in-tervention at dierent periods, adjustedORs were calculated using various

logistic regression models supported byStata sotware, version 9.1 (Stata Cor-poration, College Station, X, UnitedStates o America). Conounding vari-ables included in the model were age,

previous Caesarean section, diagnosis,district accessibility, Caesarean deliveryand transusion.

Results

able 1 presents data on obstetric ac-

tivities in the study area. During thestudy, the rate o institutional deliveriesand the number o obstetric emergen-cies treated in district health centresincreased. Te proportion o deliveries

7/27/2019 07-047076 CCare

4/9



beneting rom major obstetric inter-ventions or absolute maternal indica-tions increased markedly: rom 0.13%(95% CI: 0.100.16) at P1 to 0.46%(95% CI: 0.260.66) at P2.

o take into account the sharpincrease in the number o institutionaldeliveries during the study period, weperormed calculations with institu-tional deliveries as a denominator. Tese

calculations showed little change in therate o obstetric emergencies treated(P1 versus P2; 4.8% and 4.7%, respec-tively) while the rate o major obstetricinterventions has increased (25% to43%, respectively). Also o note is thatthere is little change in the percentageo major obstetric interventions per-ormed or absolute maternal indica-tions (55% in P1 and 57% in P2).

During the study, 2617 obstetricemergency patients received care in dis-trict health centres and beneted rom

all or some components o the mater-nity reerral system. Across the dierentperiods, they diered signicantly interms o age, diagnosis, previous obstet-

ric history and case management (Cae-sarean section), as well as in geographicalaccess to primary care services in theirdistrict o residence (able 2).

able 3 shows the causes o mater-nal death in the dierent study periods.Te crude case atality rate (denedas the ratio o total deaths observed,regardless o cause among cases, to thetotal number o cases observed) also de-creased rom 10.1% to 5.13% betweenP1 and P2. Nearly hal the reductionin mortality could be attributed to ewerdeaths rom haemorrhage. Overall, therisk o death decreased, and among

women with haemorrhage, the reduc-tion in risk was nearly three-old and

was statistically signicant (OR: 0.37;95% CI: 0.170.79). Te decrease inthe risk o death rom haemorrhage wasgreater among reerred women and wasalso statistically signicant (OR: 0.17;

95% CI: 0.040.68].able 4 presents the adjusted odds

ratios between the pre-intervention, im-plementation and two post-intervention

periods. Te risk o death was reducedby about hal or all women, and thereduction was statistically signicantbetween P1 and P1 (P= 0.027) andbetween P1 and P2 (P= 0.002). Tereduction was even more marked amongreerred women, or whom the risk odeath was three times lower in P2 thanin P1 (P= 0.002). Te statistical testsor trends in case atality rates did not

yield signicant results.

Discussion

Tis study had the advantage o beingset within a national programme, sothat the measurement o its eects andthe analysis o processes allowed us todraw lessons that are directly applicablenot only to Mali but also to otherresource-poor countries in sub-Saharan

Arica. Existing studies on the imple-mentation o transport systems, emer-

gency loans, community nancing,communications, or various combina-tions o these elements 2022 oer ewor no empirical data on whether they

Table 2. Characteristics of women with obstetric emergencies before, during and after the adoption of a national maternity referralprogramme, Mali, 20032006

Characteristic P1 P0 P1 P2 P-value (DF)

n= 475 n= 658 n= 571 n= 913

No. (%) No. (%) No. (%) No. (%)

Age in years

16 64 (13.5) 95 (14.4) 83 (14.5) 125 (13.7)

< 0.01 (9)1734* 300 (63.2) 435 (66.1) 370 (64.8) 664 (72.7) 35* 93 (19.6) 105 (16.0) 107 (18.7) 114 (12.5)Unknown* 18 (3.8) 23 (3.5) 11 (1.9) 10 (1)1Previous Caesarean section 15 (3.2) 17 (2.6) 24 (4.2) 46 (5.0) 0.07 (3)

Diagnosis

Haemorrhage* 115 (24.2) 163 (24.8) 135 (23.6) 140 (15.3)

< 0.01 (15)

Dystocic labour* 107 (22.5) 232 (35.3) 189 (33.1) 351 (38.4)Uterine rupture 12 (2.5) 18 (2.7) 10 (1.8) 16 (1.8)Pre-eclampsia/eclampsia** 59 (12.4) 47 (7.1) 49 (8.6) 91 (10.0)Inection** 11 (2.3) 5 (0.8) 3 (0.5) 12 (1.3)

Other 171 (36.0) 193 (29.3) 185 (32.4) 303 (33.2)Referred womena 143 (30.0) 273 (41.5) 246 (43.1) 452 (49.5) < 0.01 (3)

Accessibility of primary healthcare servicesb

Good* 40 (8.4) 176 (26.7) 198 (34.7) 300 (32.9)< 0.01 (6)Average* 354 (74.5) 225 (34.2) 166 (29.0) 284 (31.1)

Poor* 81 (17.1) 257 (39.1) 207 (36.3) 329 (36.0)Caesarean delivery 112 (23.6) 258 (39.2) 286 (50.1) 383 (41.9) < 0.01 (3)Transfusion 11 (2.3) 17 (2.6) 19 (3.3) 22 (2.4) 0.70 (3)

* c level o signifcance < 0.05; **c level o signifcance < 0.01.DF, degrees o reedom; P1, year beore the intervention; P0, year o the intervention; P1, 1 year ater the intervention; P2, 2 years ater the intervention.a Women with obstetric complications who were reerred by a community health centre to a district health centre.b Accessibility was categorized according to the percentage o the population in a womans district o residence that lived within 15 km rom a primary health care

centre: good accessibility, over 85%; average accessibility, rom 60% to 85%; poor accessibility, less than 60%.

7/27/2019 07-047076 CCare

5/9



reduce institutional maternal mortal-ity. Furthermore, the useulness o datarom these studies is limited by studydesigns that are less than robust.

Our data show that Malis nationalmaternity reerral system increases thecoverage o obstetric emergencies andreduces the risk o death among womendelivering with obstetric complica-tions. Tis risk reduction is achievedprimarily in those with haemorrhage,

whose prognosis is directly related towhether appropriate care is receivedwithin 2 hours or not.23 Furthermore,the reduction in risk among women

with haemorrhage is distinctly moremarked in those who are reerred bycommunity health centres than in those

who come to the district health centre

on their own. Tis point is particularlyimportant because the reerred womenhave beneted rom all componentso the intervention, particularly aster

modes o communication and transportthat reduce treatment waiting times.Tus, improved access to comprehen-sive emergency obstetric care and toCaesarean sections has contributed tothe programmes success.

Te eects o the intervention werequick to be noted because there wereconsiderable unmet needs in emergencyobstetric care and baseline case atalityrates were very high. Over the studyperiod, the number o women attend-ing community health centres or nor-mal deliveries increased progressively.Furthermore, obstetric complicationsbecame more likely to be diagnosed atan earlier stage in the labour or post-partum stages, allowing more womento benet rom the intervention sooner.

Indeed, the number o obstetric emer-gencies and o major obstetric inter-ventions perormed in reerence healthcentres increased dramatically during

the study period. However, 2 years aterimplementation o the reerral system,the rates o major obstetric interven-tions perormed or absolute maternalindications remained low (0.46%) bycomparison to rates reported in 2003rom a study in an urban setting inneighbouring Burkina Faso. In thatstudy, the number o major obstetricinterventions or absolute maternal in-dications increased signicantly (rom0.75% to 1.42%) the year ater imple-mentation o a cost-sharing mechanismor emergency obstetric care.15

An analysis o the conditions inwhich the intervention described hereinwas implemented made it possible toidentiy its strengths and weaknesses.10,24One o its strengths was the sustained

political support it received at both theregional and national levels. Reproduc-tive health is a major component oMalis national health plan, and the

Table 3. Case fatality rates (CFRs) and crude odds ratios, by cause of death,a before, during and after the adoption of a nationalmaternity referral programme, Mali, 20032006

Cause ofdeath

P1 P0 P1 P2 P1 versus P2

Deaths CFR Deaths CFR Deaths CFR Deaths CFR Absolutechangeb

Attributa-blec

Crude oddsratio (95% CI)

Haemorrhage 18 3.79 16 2.43 13 2.28 13 1.42 2.37 47.5 0.37 (0.170.79)Reerred 7 4.89 8 2.93 5 2.03 7 1.55 3.34 35.4 0.17 (0.040.68)Not reerred 11 3.31 8 2.08 8 2.46 6 1.30 2.01 57.4 0.38 (0.131.13)

Dystocic labour 8 1.68 11 1.67 13 1.23 8 0.88 1.81 16.2 0.52 (0.181.52)Reerred 6 4.19 7 2.56 2 0.81 5 1.11 3.08 32.7 0.26 (0.070.96)Not reerred 2 0.60 4 1.04 5 1.54 3 0.65 0.05 1.42 1.08 (0.159.27)

Uterine rupture 4 0.84 2 0.30 1 0.18 4 0.44 0.40 8.1 0.52 (0.112.47)Reerred 2 1.40 1 0.37 1 0.41 3 0.66 0.74 7.9 0.47 (0.064.06)Not reerred 2 0.60 1 0.26 0 0.00 1 0.22 0.38 10.9 0.36 (0.015.04)

Pre-eclampsia/eclampsia

10 2.11 10 1.52 8 1.40 9 0.99 1.12 22.5 0.46 (0.171.24)

Reerred 5 3.50 3 1.10 5 2.03 8 1.77 1.73 18.3 0.50 (0.141.78)Not reerred 5 1.51 7 1.82 3 0.92 1 0.22 1.29 36.9 0.14 (0.011.25)

Infection 2 0.42 2 0.30 0 0.0 3 0.33 0.09 1.9 0.78 (0.116.67)Reerred 0 0.00 1 0.37 0 0.0 2 0.44 0.44 4.67 NANot reerred 2 0.60 1 0.26 0 0.0 1 0.22 0.38 10.9 0.36 (0.015.04)

Other 6 1.26 9 1.37 6 1.05 10 1.07 0.19 3.8 0.87 (0.292.69)Reerred 2 1.40 5 1.83 2 0.81 2 0.44 0.96 10.2 0.31 (0.033.14)Not reerred 4 1.20 4 1.04 4 1.23 8 1.74 0.54 15.4 1.45 (0.395.76)

Total 48 10.1 50 7.59 35 6.14 47 5.13 4.98 100.0 0.48 (0.310.75)Reerred 22 15.4 25 9.16 15 6.09 27 5.97 9.43 100.0 0.35 (0.180.66)Not reerred 26 7.83 25 6.50 20 6.15 20 4.33 3.50 100.0 0.53 (0.281.01)

CFR, case atality rate; CI, confdence interval; P1, year beore the intervention; P0, year o the intervention; P1, 1 year ater the intervention; P2, 2 years ater the

intervention.a Total number o obstetric emergency cases or P1 = 475; P0 = 658; P1 = 571; P2 = 913.b The dierence between CFR P2 and CFR P1 or a specifc cause (e.g. 1.42 3.79 = 2.37 or haemorrhage).c Absolute change in maternal mortality or a specifc cause (e.g. 2.37 or haemorrhage) expressed as a percentage o the total absolute change in maternal

mortality rates (2.37 4.98 = 47.5).

7/27/2019 07-047076 CCare

6/9



Table 4. Adjusted odds ratios for comparisons of risk of death between differentperiods before, during and after the adoption of a national maternity referralprogramme, Mali, 20032006

Comparison All women Referred women Non-referred womenORa (95% CI) ORa (95% CI) ORa (95% CI)

P0 versus P1 0.72 (0.451.13) 0.62 (0.311.22) 0.76 (0.411.43)P1 versus P1 0.57 (0.350.94) 0.40 (0.190.84) 0.69 (0.351.36)P2 versus P1 0.48 (0.300.76) 0.34 (0.180.70) 0.60 (0.311.18)

CI, confdence interval; OR, odds ratio; P1, year beore the intervention; P0, year o the intervention; P1, 1year ater the intervention; P2, 2 years ater the intervention.a Adjusted or age, previous Caesarean, diagnosis, health-centre accessibility, Caesarean delivery and

transusion.

regional authorities have made the re-quired investments. A urther strengtho the programme was its reliance onthe establishment o community cost-sharing schemes, which not only reducenancial barriers to care, but also help

to ensure programme sustainabilityby eliminating the need or nancialcontributions rom outside the com-munity. Te last strength was the orga-nization o the reerral system and thecollaboration among various partners.Te programme was unique in that itsgeneral ramework was dened at thenational level, while regions supportedthe districts, which in turn designedtheir own local system. Tis methodhas the occasional drawback o slowingdown implementation in areas where

local dynamics are unavourable; how-ever, it generally has the advantage oproducing a system that is supportedby the community and local actors.In the Kayes region, implementationo these systems at the district leveltook 3 years (rom December 2002to November 2005). In Mali, theavailability o Caesarean sections atthe district level is made possible bya programme o surgical qualicationor general practitioners that allowsthem to perorm Caesarean deliveries.

Neighbouring countries with greaterresources have had difculty achievingsurgical training or general practi-tioners, resulting in reduced or non-existent access to surgical services inrural areas. Te implementation o thesystem or ongoing registration o ob-stetric emergencies will yield the datanecessary or continuous monitoringo the eectiveness o the maternityreerral system. Data collection alsoserves as a stimulus and a perormanceincentive or those locally responsible

or reproductive health services.Areas or additional improvement

include the availability and quality obasic emergency obstetric treatmentat the rst level o care. Most deliver-ies in community health centres areperormed by poorly qualied birthattendants who ail to detect obstetriccomplications early enough. Further-more, at district health centres morepatients could be saved i the currentlyinadequate supplies o blood or trans-usion were increased.

Tis study has some limitations. Ithad no control group or randomizationand was constrained by the act that thematernity reerral system is a national

programme whose eects should beevaluated under normal conditions oimplementation. Te absence o a con-trol group made it impossible to control

or other interventions or eects thatcould have modied the outcome oobstetric emergencies over time. How-ever, the study area did not undergo anyimportant social or economic changesduring the study period. Changes tothe health-care system in connection

with obstetric emergencies were madewithin the ramework o the maternityreerral system being evaluated. Teonly notable modication to healthservices outside o the programme wasthe governments decision to provideCaesarean sections ree o charge. Tisdecision completely eliminated the

womans nancial contribution, whichhad already been considerably reducedby community cost-sharing schemesassociated with the intervention.

Te second limitation o the studywas the variability o data quality,which showed improvements over timein tandem with improvements in themonitoring system (or example, miss-ing data or age decreased rom 3.8%

in P1 to 1.1% in P2). Decreases incase atality rates could be linked toan increase in the registration o lesssevere cases resulting rom better caseregistration overall and the detection ocomplications at an earlier stage.

Te limitations o the study weremitigated by the methods o analysis,

which allowed us to control or the prin-cipal coactors pertaining to patientspersonal characteristics and obstetric his-tory (age, previous Caesarean section),case mix (diagnosis), case management

(Caesarean section and transusion) andgeographical accessibility. Given theconstraints linked to the evaluation o anon-pilot reerral system, the design and

methods o analysis chosen providedthe most robust results possible.25

Our results show that in poor coun-tries, programmes to reduce barriers

to comprehensive emergency obstetricmedical care can substantially decreasedeaths associated with obstetric emer-gencies. Furthermore, they show thatsuch programmes can be implementedon a large scale without major externalunding. Our results were observed ina very poor country and in a region

where the geographic accessibil ity ohealth-care services is among the poor-est. Tereore, the system should be ap-plicable to most rural Arican contexts,

where most maternal deaths occur.Furthermore, our results allow us

to draw lessons not only about theintervention, but also about the re-search methods applied. More in-depthresearch should be carried out on theeects o this intervention on the popu-lation, especially related to geographi-cal accessibility. Te eects observedso quickly ater this intervention werethe result o the programmes responseto important unmet needs. However,eorts to satisy such needs will be

limited by geographic accessibility; anyuture eects o the intervention couldbe diminished unless patient recruit-ment is extended to those living arrom community health centres. Stud-ies are currently under way to modelaccessibility, not in terms o distance,but rather o time, taking into accountthe seasons and the modes o transportbeing used. A better understanding othe causes o the rst delay in seekingservices26 will also help improve accessto the system.

In contexts where the rate o in-stitutional deliveries is low, improve-ments in emergency obstetric care willenhance the eects o the maternity

7/27/2019 07-047076 CCare

7/9



reerral system on maternal mortality.However, to achieve a more compre-hensive approach or monitoring deliv-eries27 that will ensure better maternaloutcomes, the rate o institutional de-liveries must be increased. In turn, this

necessarily calls or improvements inservice quality at all levels o the healthsystem, some o which can derive romthe successes o the maternity reerralsystem. Our results suggest that parto the success o the intervention waslinked to increased rates o institutionaldeliveries during the study period.

Acknowledgements

We thank Karim Sangar, Aguissa Ma-ga, Diarrah Coulibaly, Odette Laplanteand Sylvie Charron or their work inthe implementation and ollow-up othe intervention and their input into

the research activities. We also thankMaria-Victoria Zunzunegui and AnnaKone or their important statisticaladvice, and the district teams, without

whose help this work could not havebeen accomplished. We also thank thereviewers or their useul commentstowards improving the paper.

Funding: Te research was supportedby the Governance, Equity and HealthInitiative o the International Devel-opment Research Centre in Ottawa,Ontario, Canada. Te Regional Health

Authority o Kayes receives support

rom the Canadian International Devel-opment Agency.

Competing interests: None declared.

Rsum

Amlioration de laccs des soins obsttricaux durgence complets et effets sur la mortalit maternelle en

milieu hospitalier, dans une rgion rurale du MaliObjectif Evaluer les eets dun systme national daiguillage visant rduire les taux de mortalit maternelle travers une amliorationde laccessibilit et de la qualit des soins obsttricaux durgencedans une rgion rurale du Mali (Arique sub-saharienne).Mthodes Un systme daiguillage maternel, comprenant laprestation de soins obsttricaux durgence de base et complets, letransport dans un centre de sant dispensant des soins obsttricauxet des systmes de partage des cots dans la collectivit, a tmis en uvre dans six districts sanitaires ruraux de la rgion deKayes, entre dcembre 2002 et novembre 2005. Dans le cadredune tude avant et aprs non contrle, nous avons enregistrtoutes les urgences obsttricales, les interventions obsttricales

majeures et les dcs maternels sur une priode dobservationde 4 ans (du 1er janvier 2003 au 30 novembre 2006), couvrantlanne avant lintervention (P-1), lanne de lintervention (P0) etles annes dbutant 1 an et 2 ans aprs lintervention (P1 et P2respectivement). La principale mesure de rsultat tait le risque dedcs chez les patientes prsentant une urgence obsttricale, calcul partir des taux de ltalit et des odds ratios bruts. Les rsultats desanalyses ont t ajusts par rgression logistique pour tenir comptedes acteurs de conusion.Rsultats Le nombre de emmes recevant des soins obsttricauxdurgence a doubl entre P-1 et P2 et le taux dinterventions

obsttricales majeures (principalement des csariennes),pratiques pour des indications maternelles absolues, est passde 0,13 % en P-1 0,46 % en P2. Pour les emmes prises encharge pour une urgence obsttricale, le risque de dcs 2 ansaprs lintervention tait inrieur de moiti celui enregistr avantlintervention (odds ratio, OR : 0,48 ; intervalle de confance 95 %,IC : 0,30-0,76). Les taux de mortalit maternelle ont diminu plusortement parmi les emmes aigui lles vers des soins obsttricauxdurgence que parmi celles stant prsentes dans des centresde sant de district, sans aiguillage. Prs de la moiti (47,5 %) decette baisse de mortalit tait attribuable la diminution des dcspar hmorragie.

Conclusion On a observ pour cette intervention des eets rapides,imputables la disponibilit des interventions obsttricales majeuresdans des centres de sant de district, la rduction du temps detransport dans ces centres pour y recevoir un traitement et larduction des obstacles fnanciers la dispensation des soins. Nosrsultats montrent que les programmes nationaux sont applicablesdans les pays aible revenu sans apport fnancier externe majeuret quils peuvent amliorer rapidement la couverture des servicesobsttricaux et diminuer notablement le risque de dcs associ auxcomplications obsttricales.

Resumen

Mejora del acceso a atencin obsttrica de urgencia integral y efecto sobre la mortalidad materna institucionalen zonas rurales de Mal

Objetivo Evaluar el eecto de un sistema nacional de derivacinconcebido para reducir las tasas de mortalidad materna mediantela mejora del acceso a la atencin obsttrica de urgencia y de lacalidad de la misma en el Mal rural (rica subsahariana).Mtodos Entre diciembre de 2002 y noviembre de 2005 seimplant en seis distritos de salud rurales de la regin de Kayesun sistema de derivacin para atencin de maternidad que incluaatencin obsttrica de urgencia bsica e integral, transporte a

servicios de obstetricia y planes comunitarios de participacin en lafnanciacin de los gastos. Mediante un estudio antes y despusno controlado, registramos todas las urgencias obsttricas, lasintervenciones obsttricas mayores y las deunciones maternas

a lo largo de un periodo de observacin de 4 aos (1 de enero de2003 a 30 de noviembre de 2006): el ao previo a la intervencin(P-1); el ao de intervencin (P0), y al cabo de 1 y 2 aos de laintervencin (P1 y P2, respectivamente). La variable principalde valoracin ue el riesgo de deuncin entre las pacientes conurgencias obsttricas, calculado mediante las tasas brutas deletalidad y las razones de posibilidades brutas. Los anlisis seajustaron por variables de conusin mediante regresin logstica.

Resultados El nmero de mujeres que recibieron atencinobsttrica de urgencia se duplic entre P-1 y P2, y la tasa deintervenciones obsttricas mayores (principalmente cesreas)para indicaciones maternas absolutas aument de 0,13% en

7/27/2019 07-047076 CCare

8/9



References

Maternal mortality in 2005: estimates developed by WHO, UNICEF, UNFPA1.and the World Bank. Geneva: World Health Organization; 2007.Health and the Millenium Development Goals2. . Geneva: World HealthOrganization; 2005.AbouZahr C, Wardlaw T. Maternal mortality at the end o a decade: signs o3.

progress? Bull World Health Organ2001;79:561-8. PMID:11436479Ronsmans4. C, Graham WJ. Maternal mortali ty: who, when, where andwhen. Lancet2006;368:1189-200. PMID:17011946doi:10.1016/S0140-6736(06)69380-XCampbell5. OMR, Graham WJ. Strategies or reducing maternal mortality:getting on with what works. Lancet2006;368:1284-99. PMID:17027735doi:10.1016/S0140-6736(06)69381-1Paxton6. A, Maine D, Freddman L, Fry D, Lobis S. The evidence or emergencyobstetric care. Int J Gynaecol Obstet2005;88:181-93. PMID:15694106doi:10.1016/j.ijgo.2004.11.026Weil7. O, Fernandez H. Is sae motherhood an orphan initiative? Lancet1999;354:940-3. PMID:10489970doi:10.1016/S0140-6736(99)02369-7Ronsmans8. C, Etard J-F, Walraven G, Hj L, Dumont A, de Bernis L, et al.Maternal mortality and access to obstetric services in West Arica. TropMed Int Health2003;8:940-8. PMID:14516306doi:10.1046/j.1365-

3156.2003.01111.x

Stanton9. CK, Holtz SA. Levels and trends in cesarean birth in the developingworld. Stud Fam Plann2006;37:41-8. PMID:16570729doi:10.1111/j.1728-4465.2006.00082.xMurray10. SF, Pearson SC. Maternity reerral systems in developing countries:current knowledge and uture research needs. Soc Sci Med2006;

62:2205-15. PMID:16330139doi:10.1016/j.socscimed.2005.10.025Palmer11. N, Mueller DH, Gilson L, Mills A, Haines A. Health fnancing topromote access in low income settings - how much do we know? Lancet2004;364:1365-70. PMID:15474141doi:10.1016/S0140-6736(04)17195-XEnsor12. T, Ronih J. Eective fnancing o maternal health services: a review othe literature. Health Policy2005;75:49-58. PMID:16298228doi:10.1016/j.healthpol.2005.02.002Borghi13. J, Hanson K, Adjei Acquah C, Ekanmian G, Filippi V, Ronsmans C, et al.Costs o near-miss obstetric complications or women and their amilies inBenin and Ghana. Health Policy Plan2003;18:383-92. PMID:14654514doi:10.1093/heapol/czg046Borghi14. J, Ensor T, Somanathan A, Lissner C, Mills A. Lancet MaternalSurvival Series steering group. Mobilising fnancial resources or maternalhealth. Lancet2006;368:1457-65. PMID:17055948doi:10.1016/S0140-6736(06)69383-5

P-1 a 0,46% en P2. En las mujeres tratadas por una urgenciaobsttrica, el riesgo de muerte a los 2 aos de la intervencinse redujo a la mitad respecto a antes de la intervencin (raznde posibilidades, OR: 0,48; intervalo de confanza, IC, del 95%:0,300,76). Las tasas de mortalidad materna disminuyeron msentre las mujeres derivadas para atencin obsttrica urgente que

entre las que acudieron al centro de salud de distrito sin derivacinprevia. Casi la mitad (47,5%) de la reduccin de la mortalidad esatribuible al menor nmero de deunciones por hemorragia.Conclusin La intervencin tuvo eectos rpidos debido a la

disponibilidad de intervenciones obsttricas mayores en loscentros de salud de distrito, la disminucin del tiempo de transportehasta esos centros para recibir tratamiento y la disminucin de lasbarreras fnancieras a la atencin. Nuestros resultados muestranque es posible aplicar programas nacionales en los pases deingresos bajos sin necesidad de ondos externos importantes, y

que dichos programas pueden mejorar rpidamente la coberturade servicios obsttricos y reducir de orma considerable el riesgode deuncin asociada a complicaciones obsttricas.

:

.) ( :

/ 2002 /.2005

2003 / 1 ()1 ( )2006 / 30

)1 ( )0 (.)2 (

.

.:

) (

%0.46 %0.13 .

0.30 0.48 (.)0.76

.

. )%47.5(:

.

.
http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=11436479&dopt=Abstracthttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=17011946&dopt=Abstracthttp://dx.doi.org/10.1016/S0140-6736(06)69380-Xhttp://dx.doi.org/10.1016/S0140-6736(06)69380-Xhttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=17027735&dopt=Abstracthttp://dx.doi.org/10.1016/S0140-6736(06)69381-1http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=15694106&dopt=Abstracthttp://dx.doi.org/10.1016/j.ijgo.2004.11.026http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=10489970&dopt=Abstracthttp://dx.doi.org/10.1016/S0140-6736(99)02369-7http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=14516306&dopt=Abstracthttp://dx.doi.org/10.1046/j.1365-3156.2003.01111.xhttp://dx.doi.org/10.1046/j.1365-3156.2003.01111.xhttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=16570729&dopt=Abstracthttp://dx.doi.org/10.1111/j.1728-4465.2006.00082.xhttp://dx.doi.org/10.1111/j.1728-4465.2006.00082.xhttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=16330139&dopt=Abstracthttp://dx.doi.org/10.1016/j.socscimed.2005.10.025http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=15474141&dopt=Abstracthttp://dx.doi.org/10.1016/S0140-6736(04)17195-Xhttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=16298228&dopt=Abstracthttp://dx.doi.org/10.1016/j.healthpol.2005.02.002http://dx.doi.org/10.1016/j.healthpol.2005.02.002http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=14654514&dopt=Abstracthttp://dx.doi.org/10.1093/heapol/czg046http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=17055948&dopt=Abstracthttp://dx.doi.org/10.1016/S0140-6736(06)69383-5http://dx.doi.org/10.1016/S0140-6736(06)69383-5http://dx.doi.org/10.1016/S0140-6736(06)69383-5http://dx.doi.org/10.1016/S0140-6736(06)69383-5http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=17055948&dopt=Abstracthttp://dx.doi.org/10.1093/heapol/czg046http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=14654514&dopt=Abstracthttp://dx.doi.org/10.1016/j.healthpol.2005.02.002http://dx.doi.org/10.1016/j.healthpol.2005.02.002http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=16298228&dopt=Abstracthttp://dx.doi.org/10.1016/S0140-6736(04)17195-Xhttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=15474141&dopt=Abstracthttp://dx.doi.org/10.1016/j.socscimed.2005.10.025http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=16330139&dopt=Abstracthttp://dx.doi.org/10.1111/j.1728-4465.2006.00082.xhttp://dx.doi.org/10.1111/j.1728-4465.2006.00082.xhttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=16570729&dopt=Abstracthttp://dx.doi.org/10.1046/j.1365-3156.2003.01111.xhttp://dx.doi.org/10.1046/j.1365-3156.2003.01111.xhttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=14516306&dopt=Abstracthttp://dx.doi.org/10.1016/S0140-6736(99)02369-7http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=10489970&dopt=Abstracthttp://dx.doi.org/10.1016/j.ijgo.2004.11.026http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=15694106&dopt=Abstracthttp://dx.doi.org/10.1016/S0140-6736(06)69381-1http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=17027735&dopt=Abstracthttp://dx.doi.org/10.1016/S0140-6736(06)69380-Xhttp://dx.doi.org/10.1016/S0140-6736(06)69380-Xhttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=17011946&dopt=Abstracthttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=11436479&dopt=Abstract

7/27/2019 07-047076 CCare

9/9



Richard15. F, Oudraogo C, Compaor J, Dubourg D, De Brouwere V. Reducingfnancial barriers to emergency obstetric care: experience o cost-sharingmechanism in a district hospital in Burkina Faso. Trop Med Int Health2003;12:972-81.The16. health o the people: Arican regional health report. Geneva: World HealthOrganization, Regional Ofce or Arica; 2006.Human17. development report 2007/2008: fghting climate change: humansolidarity in a divided world. New York: Palgrave Macmillan; 2007.Mali18. demographic and health survey 2006. Calverton, MD: CPN/DNSI andMacro International, Inc.; 2007.Maine19. D, Akalin MZ, Ward VM, Kamara A. The design and evaluation omaternal mortality programs. New York: Center or Population and FamilyHealth, School o Public Health, Columbia University; 1997.Krasovec20. K. Auxiliary technologies related to transport and communicationor obstetric emergencies. Int J Gynaecol Obstet2004;85 Suppl 1;S14-23.PMID:15147850doi:10.1016/j.ijgo.2004.02.007Samai21. O, Sengeh P. Facilitating emergency obstetric care throughtransportation, Bo, Sierra Leone. Int J Gynaecol Obstet1997;59 Suppl2;S157-64. PMID:9389627doi:10.1016/S0020-7292(97)00161-6Foana22. P, Samai O, Kebbie A. Sengeh (the BO PMM Team). Promoting the useo obstetric services through community loan unds, Bo, Sierra Leone. Int JGynaecol Obstet1997;59 Suppl 2;S225-30. PMID:9389635 doi:10.1016/

S0020-7292(97)00169-0

World23. health report 2005: making every mother and child count. Geneva:World Health Organization; 2005.Tourigny24. C, Fournier P, Sangar K, Laplante O, Coulibaly D, LaBossire F,et al. valuation de limpact dune intervention visant rduire la mortalitmaternelle en milieu rural en Arique (Kayes, Mali) [in French]. CanadianPublic Health Association, 97th Annual Conerence, Vancouver, Canada, 30

May 2005.Ross25. L, Simkhada P, Smith WCS. Evaluat ing eectiveness o complexinterventions aimed at reducing maternal mortality in developing countries. JPublic Health (Ox)2005;27:331-7. PMID:16234263doi:10.1093/pubmed/di058Thaddeus26. S, Maine D. Too ar to walk: maternal mortali ty in context.Soc Sci Med1994;38:1091-110. PMID:8042057doi:10.1016/0277-9536(94)90226-7Koblinsky M, Matthews Z, Hussein J, Mavalankar D, Mridha MK, Anwar I,27.et al; Lancet Maternal Survival Series steering group. Going to scale withproessional skilled care. Lancet2006;368:1377-86. PMID:17046470doi:10.1016/S0140-6736(06)69382-3
http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=15147850&dopt=Abstracthttp://dx.doi.org/10.1016/j.ijgo.2004.02.007http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=9389627&dopt=Abstracthttp://dx.doi.org/10.1016/S0020-7292(97)00161-6http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=9389635&dopt=Abstracthttp://dx.doi.org/10.1016/S0020-7292(97)00169-0http://dx.doi.org/10.1016/S0020-7292(97)00169-0http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=16234263&dopt=Abstracthttp://dx.doi.org/10.1093/pubmed/fdi058http://dx.doi.org/10.1093/pubmed/fdi058http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=8042057&dopt=Abstracthttp://dx.doi.org/10.1016/0277-9536(94)90226-7http://dx.doi.org/10.1016/0277-9536(94)90226-7http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=17046470&dopt=Abstracthttp://dx.doi.org/10.1016/S0140-6736(06)69382-3http://dx.doi.org/10.1016/S0140-6736(06)69382-3http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=17046470&dopt=Abstracthttp://dx.doi.org/10.1016/0277-9536(94)90226-7http://dx.doi.org/10.1016/0277-9536(94)90226-7http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=8042057&dopt=Abstracthttp://dx.doi.org/10.1093/pubmed/fdi058http://dx.doi.org/10.1093/pubmed/fdi058http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=16234263&dopt=Abstracthttp://dx.doi.org/10.1016/S0020-7292(97)00169-0http://dx.doi.org/10.1016/S0020-7292(97)00169-0http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=9389635&dopt=Abstracthttp://dx.doi.org/10.1016/S0020-7292(97)00161-6http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=9389627&dopt=Abstracthttp://dx.doi.org/10.1016/j.ijgo.2004.02.007http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=15147850&dopt=Abstract

07-047076 ccare

Documents