Orientadores: Jose Guilherme Cecatti, Mary Ângela Parpinelli / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Ciências Médicas / Made available in DSpace on 2018-08-12T12:09:48Z (GMT). No. of bitstreams: 1 Souza_JoaoPauloDiasde_D.pdf: 4674272 bytes, checksum: c4d150bbef9de45ab0244529ccfa9117 (MD5) Previous issue date: 2008 / Resumo: Desde os anos 1990, um grupo especial de mulheres que sobrevivem a complicações graves da gestação tem despertado o interesse de pesquisadores e formuladores de políticas públicas. O melhor entendimento deste grupo, conhecido como "near miss", poderia contribuir para o desenvolvimento de políticas de saúde para a melhora da saúde materna. A presente tese de doutorado foi desenvolvida com a finalidade mais ampla de avaliar se o estudo do near miss materno seria factível através de inquéritos populacionais. Para este fim, foi desenvolvida uma série de projetos, implementados em três fases: preparação, desenvolvimento e validação de um instrumento para o inquérito populacional, e aplicação (através de uma pesquisa nacional de demografia e saúde). Na fase de preparação foram analisadas informações obtidas por inquéritos demográficos de saúde realizados na América Latina e Caribe sobre a morbidade materna. Durante a fase de desenvolvimento e validação do questionário, foi realizada uma revisão sistemática sobre o assunto, desenvolvida uma definição pragmática de near miss e avaliadas vivências da morbidade materna grave. A partir de achados das duas primeiras fases foi desenvolvido um instrumento para a realização de inquéritos populacionais, que foi submetido a um estudo de validação. A fase de aplicação se deu com aincorporação deste questionário na mais recente Pesquisa Nacional de Demografia e Saúde realizada no Brasil. Ao final deste processo, e utilizando-se uma definição pragmática, baseada na ocorrência de eclâmpsia, histerectomia, internação à UTI e transfusão de sangue, estimam-se a incidência anual de 54.000 casos de near miss materno no Brasil, com prevalência aproximada ao redor de 20 casos/1.000 nascidos vivos. Mulheres com mais de 40 anos ou de baixa escolaridade apresentam risco aumentado de desenvolver a condição. No conjunto, esta tese demonstra a versatilidade do conceito de near miss materno, cuja definição pode ser estrita ou pragmática, e utilizada para diversos fins, incluindo monitoramento, vigilância epidemiológica, controle de qualidade e para orientação do cuidado. A versatilidade do conceito, a maior freqüência dos casos e a possibilidade de entrevistar as mulheres suportam o propósito de se estudar o near miss materno para guiar os esforços locais na redução da mortalidade materna e da carga das complicações graves. / Abstract: Since the 1990s, a special group of women who survive severe complications of pregnancy has attracted the interest of researchers and public policymakers. A better understanding of this group, known as "near miss", could contribute to the development of health policies to improve maternal health. This doctoral thesis was developed with the wider aim of assessing whether the study of maternal near miss would be feasible through population surveys. To this end we developed a series of projects, implemented in three phases: preparation, development and validation of an instrument for the survey, and implementation (through a national survey of population and health). At the stage of preparation, information obtained by demographic health surveys conducted in Latin America and the Caribbean on maternal morbidity was analyzed. During the development and validation of the questionnaire, a systematic review was conducted and a pragmatic definition of near miss was developed. Womens experiences of severe maternal morbidity were evaluated. The findings from the first two phases supported the development of tool for population inquiries, which has undergone a validation study. The phase of implementation was possible due to incorporation of this questionnaire in the most recent national Demographic and Health Survey held in Brazil (2006/07). At the end of this process, and using a pragmatic definition, based on the occurrence of eclampsia, hysterectomy, admission to the ICU and blood transfusion, it is estimated the annual incidence of 54,000 cases of maternal near miss in Brazil, with prevalence around 20 cases/1.000 live births. Women aged > 40 years or with low education are at increased risk of developing the condition. Overall, this thesis demonstrates the versatility of the concept of near miss, whose definition can be strict or pragmatic, and used for various purposes, including monitoring, surveillance, quality control and guidance of care. The versatility of the concept, the greater frequency of cases and the possibility of interviewing women support the purpose of studying the maternal near miss to guide local efforts in reducing. / Doutorado / Doutor em Tocoginecologia

Bien que devenu un événement rare dans les pays riches, la mortalité maternelle est toujours considéré comme un marqueur du dysfonctionnement du système de soins d'un pays. Le premier travail de cette thèse avait pour objectif d'étudier l'évolution du profil de la mortalité maternelle en France entre 1998 et 2007 à partir des données de l'Enquête Nationale Confidentielle sur les Morts Maternelles (ENCMM). Le ratio de mortalité maternelle (RMM) est resté stable, avec 9 décès pour 100 000 naissances vivantes. Il se situe dans la moyenne basse des pays européens ayant un système renforcé de mesure, ce qui est relativement satisfaisant compte-tenu de l'évolution du profil des femmes et de l'augmentation observée ailleurs. La principale cause de décès était l'hémorragie obstétricale. La proportion des morts maternelles évitables n'a pas évolué et est restée au tour de 50%. Ces résultats soulignent qu'une réduction du RMM est envisageable. Pour cela, une meilleure compréhension de ses déterminants est essentielle. Le deuxième travail a abordé la question des disparités régionales. Cette analyse a montré que le risque de mort maternelle en postpartum, après prise en compte des certaines caractéristiques individuelles des femmes, était supérieur dans les DOM et en Ile-de-France, comparé au reste de la France métropolitaine, et que des facteurs liés au contenu et à l'organisation des soins pourraient être impliqués. La pertinence des leçons tirées de l’étude de la mortalité maternelle dépend de la validité de sa mesure. Le dernier travail avait pour objectif d’évaluer les écueils des statistiques de routine de mortalité pour étudier la fréquence et les causes de mortalité maternelle. La sous-estimation du RMM dans les statistiques de mortalité a diminué significativement au cours du temps. Cependant, les inexactitudes sont plus importantes que le RMM ne le montre ; en effet, pour 2007-2009 une mort maternelle sur trois de l’ENCMM n’est pas repérée dans les statistiques de mortalité et le profil de causes diffère de celui issu de l’ENCMM. / While maternal mortality has become a very rare event in developed countries, it remains a marker of the performance of a country’s health system. The first work of this thesis was to study the evolution of maternal mortality profile in France between 1998 and 2007 using data from the Confidential enquiry into maternal deaths (ENCMM). The maternal mortality ratio (MMR) remained stable, with 9 deaths per 100,000 live births, it is in the low average of other European countries that also have enhanced measurement systems, which is consider satisfactory in view of the evolution of profile of women and the rise of MMR observed elsewhere. The main cause of death was obstetric hemorrhage. The proportion of avoidable maternal deaths has not changed and remained around 50%. These results emphasize that reducing the MMR is possible, for this, a better understanding of its determinants is essential. The second work explored regional disparities of maternal mortality. The results showed that the risk of postpartum maternal death was higher, after taking women's individual characteristics into account, in the overseas districts (DOM) but also in Ile-de-France, compared with the rest of metropolitan France; we thus deduce that factors related to care may be involved.The relevance of lessons learned from the study of maternal mortality depends on the validity of the data used. The last work assessed the pitfalls of routine mortality statistics to study the frequency and profile of maternal mortality. The underestimation of MMR in mortality statistics has decreased significantly over time. However, inaccuracies are greater than suggested by the MMR; the routine statistics failed to identify one third of the maternal deaths identified by the ENCMM and the causes profile differs from that resulting from the ENCMM.

Representations of natural disasters in Black Southern literature identify social location as the greatest indicator of risk vulnerability. Moreover, they can expose the precarious subjectivity of the Black female reproductive body, as addressed through characters Lulu in Richard Wright’s “Down by the Riverside,” Janie in Zora Neale Hurston’s Their Eyes Were Watching God, and Jesmyn Ward’s Esch in Salvage the Bones. Together, these female characters share a legacy of social marginalization and Black female resistance that is (re)shaped through their experiences with ecological catastrophe. This thesis considers these three texts together as an ongoing testimony and as a means to bear witness to a socio-historical record of disaster oppression and Black female resistance.

Maternal deaths in Ethiopia are mainly due to complications of pregnancy and delivery. The socio-cultural contexts under which these pregnancies and deliveries occur that pave the way for these complications and mortality. In Ethiopia, the maternal mortality ratio had been 353/100,000 live births in 2015. Therefore, the purposes of this study were to examine, and describe the socio-cultural determinants and missed opportunities of maternal health care in Eastern Ethiopia. The study was conducted in selected districts of Grawa, Chelenko and Haramaya Woreda, East Hararghe, Oromia National Regional State, Ethiopia. A community-based survey involving pregnant women in their third trimester and women who gave birth in the last five years, husbands, mothers-in-law, sisters-in-law, health workers, religious and community leaders were conducted between September up to December 2017. A systematic sampling technique was used to get a total of 422 study participants for quantitative and 24 FGD participants to qualitative study were adopted using triangulation of data collection. Pre-tested and structured questionnaire was used to collect relevant data. The main instrument used for quantitative data collection was the structured questionnaire, specifically in-depth interview methods. Bivariate and Multivariate data analysis were performed using SPSS version 25.0 and focus group discussion (FGD) was used to collect qualitative information and the information was analysed using thematic analysis method based on Atlas.ti version 8.2 statistical software packages. The study revealed that among 359 (85%) pregnant women who planned for ANC visit, 16 (4.5%) received ANC four or more times during their last pregnancies, the respondents (81.3%) claimed that they were taken care of by skilled delivery attendant during delivery, 18.5% of them said that they delivered at home and 71.1% of them received medical care after delivery (missed opportunity). Women in the age group 15-24 years [AOR: 1.18, 95%CI: 1.18 (0.37, 3.74)], primary school [AOR: 4.09, 95%CI: 4.09(0.96, 15.50)], women intended their last pregnancy [AOR: 3.1, 95% CI: 0.32(0.11, 0.94)], and women living in urban residences [AOR: 1.2, 95%CI: 0.86(0.25, 2.95)] were significant predictors of unplanned home delivery. For optimal and effective interventions of maternal health services utilization, provisions should be made for better women‘s education, family planning, community-based health insurance, health facilities access, job opportunity and women empowerment; provisions should also be made for creating income generating activities to women. Strengthening village women‘s army wing, refreshing and enabling health extension workers and traditional birth attendants. What is more, optimal measures should be taken to discourage traditional practices such as female genital mutilation, polygamy, violence against women and teenage marriage. Finally, free maternal and child health services should be advocated for so that the gap in maternal healthcare services is bridged. / Health Studies / D. Litt. et Phil. (Health Studies)

Developing an Automatic Speech Recognition (ASR) system for the Somali language, though not novel, is not actively explored; hence there has been no success in a model for conversational speech. Neither are related works accessible as open-source. The unavailability of digital data is what labels Somali as a low resource language and poses the greatest impediment to the development of an ASR for Somali. The incentive to develop an ASR system for the Somali language is to contribute to reducing the Maternal Mortality Rate (MMR) in Somalia. Researchers acquire interview audio data regarding maternal health and behaviour in the Somali language; to be able to engage the relevant stakeholders to bring about the needed change, these audios must be transcribed into text, which is an important step towards translation into any language. This work investigates available ASR for Somali and attempts to develop a prototype ASR system to convert Somali audios into Somali text. To achieve this target, we first identified the available open-source systems for speech recognition and selected the DeepSpeech engine for the implementation of the prototype. With three hours of audio data, the accuracy of transcription is not as required and cannot be deployed for use. This we attribute to insufficient training data and estimate that the effort towards an ASR for Somali will be more significant by acquiring about 1200 hours of audio to train the DeepSpeech engine

Die hohe Müttersterblichkeit in Entwicklungsländern ist zum Teil auf indirekte Faktoren wie Analphabetismus und eingeschränkten Zugang zu Gesundheitsinformationen für Mütter zurückzuführen. Während gebildete Frauen auf Gesundheitsinformationen über Online-Plattformen und mHealth-Apps zugreifen können, müssen Analphabetinnen diese in Gesundheitseinrichtungen abrufen, was aufgrund der Transportkosten oft nicht möglich ist. Mobilfunktechnologie hat in der Gesundheitsversorgung Chancen für ressourcenarme Gemeinschaften eröffnet, die sonst nicht von den digitalen Technologien profitiert hätten. Obwohl Mobilfunktechnologie in der Müttergesundheit eingesetzt wird, können die meisten Maßnahmen nicht von Analphabeten genutzt werden, verwenden Sicherheitsmodelle die nicht auf den Kontext von Entwicklungsländern zugeschnitten sind, und wurden nicht auf ihre Auswirkungen auf die Müttergesundheit hin evaluiert. In dieser Arbeit wurden zwei (Web und Mobile) Apps entwickelt, die die Übermittlung von multimedialen Nachrichten zur Müttergesundheit, Terminerinnerungen und Anrufe/Chats erleichtern. Um die Anforderungen der Nutzer zu erfassen, wurde eine Feldstudie mit halbstrukturierten Interviews und Fokusgruppendiskussionen mit schwangeren Analphabetinnen, Gesundheitsexperten und Entwicklern durchgeführt. Es folgte die Entwicklung eines Sicherheitsmodells (T2RoL) zur Sicherung der Gesundheitsinformationen in den Apps, die dann nach einem nutzerzentrierten Designansatz entwickelt wurden. Eine zweite Feldstudie in Form von halbstrukturierten Interviews und Umfragen wurde durchgeführt, um die mobile App in einer randomisierten kontrollierten Studie mit 80 schwangeren Analphabetinnen über 9 Monate zu evaluieren. Die Auswertung zeigte, dass die App akzeptiert wurde sowie einfach zu erlernen und zu benutzen war. Das Wissen über Müttergesundheit in der Interventionsgruppe verbesserte sich, was sich positiv auf gesundheitsbezogene Entscheidungen und Gesundheitsmaßnahmen auswirkte. / Maternal mortality is high in developing countries partly due to indirect factors such as illiteracy and limited access to maternal health information. While literate women can access health information from online platforms, and mHealth apps, illiterate women must get it from health facilities which is often not possible due to lack of transport fees. Mobile technology has opened opportunities in maternal health care for low resource communities that would otherwise not have benefited from digital technologies. Although used in maternal health, most interventions are not usable by the illiterate, use security models that are not tailored to the developing countries’ context, and have not been evaluated to assess their impact on maternal health care. In this thesis, two (web and mobile) apps that facilitate delivery of multimedia-based maternal health messages, appointment reminders, and calls/ chats were developed. To gather user requirements, a field study in form of semi-structured interviews and focus group discussions was conducted with illiterate pregnant women, health practitioners and developers. Development of a security model (T2RoL) to secure the health information in the apps followed. The apps were then developed following a user-centered design approach. A second field study in form of semi-structured interviews and surveys was conducted to evaluate the mobile app through a randomized controlled trial with 80 illiterate pregnant women that were followed for 9 months. Overall, results show that the app was acceptable, easy to learn and use. There was improved maternal health knowledge among the intervention group which positively influenced health related decision making and health practices.

Abortion and severe pre-eclampsia/eclampsia remain the major causes of maternal mortality in Zimbabwe. Based on this problem, factors associated with maternal mortality due to abortion and severe pre-eclampsia/eclampsia were investigated to improve pregnancy outcomes. Cases and controls were selected from 4895 abortion and 318 severe preeclampsia/ eclampsia obstetric records to conduct a retrospective case-control study. Significant risk factors identified for reducing maternal mortality due to postabortion complications included the administration of oxytocic drugs and evacuations of the uterus whilst anaemia and sepsis apparently reduced these women's chances of survival. No significant factors could be identified which influenced maternal deaths among women suffering from severe pre-eclampsia/eclampsia. Magnesium sulphate was not routinely administered, as recommended internationally. In both groups, cases apparently received better reported quantitative care than controls. Recommendations based on this research report include improved midwifery education and in-service training, regular audits of patients' records and changed policies for managing these conditions more effectively in Zimbabwe. / Health Studies / D. Litt. et Phil. (Advanced Nursing Sciences)

The study attempted to identify factors influencing women’s preference for home births in the Mutare District, Zimbabwe. A quantitative, descriptive, exploratory, cross sectional survey, gathering data by conducting structured interviews with 150 women, was used. All 150 women attended antenatal clinics but did not deliver their babies at health care facilities. The research results indicated that home deliveries might decline if: • the hospital/clinic fees were reduced or removed • transport would be available for women in labour to reach hospitals/clinics • shelters were built for pregnant women at hospitals/clinics • clinics were well equipped and had sufficient numbers of midwives • women had received more effective health education on the advantages of institutional deliveries and on the danger signs of pregnancy/labour complications • nurses/midwives would treat patients respectfully. Unless these factors are addressed, the number of home deliveries might not decline, and the high maternal/infant mortality and morbidity rates in this district will persist. / Health Studies / M. Public Health

Résumé Objectif : Identifier les facteurs institutionnels qui influencent la mortalité maternelle (MM) hospitalière dans les maternités chirurgicales au Sénégal. Méthode : cette étude est une analyse secondaire des données de la troisième Enquête Nationale sur la Couverture Obstétrico-chirurgicale au Sénégal en 2001. Les données analysées, issues des fiches d'activité des maternités, comptaient pour 38,239 admissions en obstétrique dans 19 hôpitaux et 450 décès maternels. Les taux de mortalité maternelle hospitalière (TMMH) brut et ajusté ont été utilisés comme variables dépendantes. Le TMMH ajusté sur les caractéristiques de la clientèle ('cases-mix') a été estimé pour chaque établissement de santé par la méthode de standardisation directe. Les indicateurs de la qualité des structures, de la gestion des ressources, et un score de qualité ont été utilisés comme variables indépendantes pour prédire la MM hospitalière. Les tests de Mann-Whitney et de Kruskal-Wallis ont été utilisés pour analyser l’association entre les variables indépendantes, le score de qualité et la MM. Une analyse multivariée a été utilisée pour estimer l’impact du score de qualité sur la MM, en tenant compte de la situation géographique (Dakar versus autre région).Résultats: En analyse bivariée, la présence d'anesthésiste, la disponibilité de boîtes de césarienne complète et la supervision de tous les accouchements par du personnel qualifié sont les facteurs institutionnels associés significativement à une réduction du TMMH brut. Quant au TMMH ajusté ce sont la présence de scialytique, la disponibilité du sulfate de magnésium, l'utilisation des guides de pratiques cliniques (GPC) pour la prise en charge des complications obstétricales. Le score de qualité est associé significativement au TMMH brut, y compris en analyse multivariée, mais pas au TMMH ajusté. Conclusion : La disponibilité du Sulfate de magnésium, et du scialytique pourrait contribuer à la réduction de la MM. En complément, une réorganisation adéquate des ressources pour réduire la disparité géographique rurale/urbaine est essentielle ainsi qu’une sensibilisation du personnel à l’usage des GPC. De plus, l’assistance par un personnel qualifié de tous les accouchements est nécessaire pour améliorer la qualité des soins et la prise en charge des complications obstétricales. / Abstract Objective: To identify which institutional factors are associated to the hospital maternal mortality in surgical maternities in Senegal. Method: This study is a secondary data analysis of the third National survey of the Surgical-Obstetric Coverage in Senegal in 2001. The hospital statistics data analyzed represent 38,239 admissions in obstetrics in 19 hospitals and 450 maternal deaths. The hospital maternal death rates (TMMH) crude and adjusted were used as dependent variables. The TMMH adjusted on the characteristics of the customers (`cases-mix') was estimated for each health structure by the direct standardization method. The indicators of the structure, resources management’s complexity score quality were used as independent variables to predict hospital MM. The Mann-Whitney and Kruskal-Wallis tests were used to analyze the association between independent variables, the complexity score of quality and MM. A multivariate analysis was used to estimate the impact of quality score on the MM, taking into account the geographical location (Dakar versus other region). Results: In bivariate analysis, the presence of anaesthetist, the availability of Caesarean kit and supervision of all births by trained personnel are the institutional factors significantly associated with the crude TMMH decreased. For the adjusted TMMH, the presence of Scialytic, the availability of Magnesium Sulphate, the use of clinical guidelines of good practice (GPC) for obstetrical complications are the institutional factors associated to his decreased. The complexity score quality is significantly associated with the crude TMMH, including multivariate analysis, but not with adjusted TMMH. Conclusion: The availability of the Magnesium Sulphate, and the Scialytic could contribute to the reduction of maternal mortality. In addition, an adequate reorganization of the resources to reduce the geographical rural/urban disparity is essential, as well as a developing the personnel’s awareness for the use of GPC. Moreover, qualified assistance at birth for all women is necessary to improve the quality of care and the treatment obstetrical complications.

La mortalité maternelle et périnatale est un problème majeur de santé publique dans les pays en développement. Elle illustre l’écart important entre les pays développés et les pays en développement. Les interventions techniques pour améliorer la santé maternelle et périnatale sont connues dans les pays en développement, mais ce sont la faiblesse des systèmes de santé et les défis liés aux ressources qui freinent leur généralisation. L’objectif principal de ce travail était de mieux comprendre le rôle des ressources humaines en particulier ceux de la première ligne dans la performance d’un système de référence maternelle. Au Mali, la mise en place d’un système de référence maternelle, système de référence-évacuation « SRE », fait partie des mesures nationales de lutte contre la mortalité maternelle et périnatale. Les trois composantes du SRE, soit les caisses de solidarité, le transport et la communication et la mise à niveau des soins obstétricaux, permettent une action simultanée du côté de la demande et de l’offre de soins maternels et périnatals. Néanmoins, la pénurie de personnel qualifié a conduit à des compromis sur la qualification du personnel dans l’implantation de ce système. La région de Kayes, première région administrative du Mali, est une région de forte émigration. Elle dispose d’une offre de soins plus diversifiée qu’ailleurs au Mali, grâce à l’appui des Maliens de l’extérieur. Son SRE offre ainsi un terrain d’études adéquat pour l’analyse du rôle des professionnels de première ligne. De façon plus spécifique, ce travail avait pour objectifs 1) d’identifier les caractéristiques des équipes de soins de première ligne qui sont associées à une meilleure performance du SRE en termes de survie simultanée de la mère et du nouveau-né et 2) d’approfondir la compréhension des pratiques de gestion des ressources humaines, susceptibles d’expliquer les variations de la performance du SRE de Kayes. Pour atteindre ces objectifs, nous avons, à partir du cadre de référence de Michie et West modélisé les facteurs liés aux ressources humaines qui ont une influence potentielle sur la performance du SRE de Kayes. L’exploration des variations du processus motivationnel a été faite à partir de la théorie de l’attente de Vroom. Nous avons ensuite combiné une revue de la littérature et un devis de recherche mixte (quantitative et qualitative). Les données pour les analyses quantitatives proviennent d’un système d’enregistrement continu de toutes les urgences obstétricales (GESYRE : Gestion du Système de Référence Évacuation mis en place depuis 2004 dans le cadre du suivi et de l’évaluation du SRE de Kayes) et des enquêtes à passages répétés sur les données administratives et du personnel des centres de santé. Un modèle de régression biprobit a permis d’évaluer les effets du niveau d’entrée dans le SRE et des équipes de soins sur la survie jointe de la mère et du nouveau-né. A l’aide d’entrevues semi-structurées et d’observations, nous avons exploré les pratiques de gestion des personnes dans des centres de santé communautaires « CScom » sélectionnés par un échantillonnage raisonné. Les résultats de ce travail ont confirmé que la main d’œuvre humaine demeure cruciale pour la performance du SRE. Les professionnels de première ligne ont influencé la survie des femmes et des nouveau-nés, à morbidités égales, et lorsque la distance parcourue est prise en compte. La meilleure survie de la mère et du nouveau-né est retrouvée dans les cas d’accès direct à l’hôpital régional. Les femmes qui sont évacuées des centres de première ligne où il y a plus de professionnels ou un personnel plus qualifié avaient un meilleur pronostic materno-fœtal que celles qui ont consulté dans des centres qui disposent de personnel peu qualifié. Dans les centres de première ligne dirigés par un médecin, des variations favorables à la performance comme une implication directe des médecins dans les soins, un environnement de soins concurrentiel ont été retrouvés. Concernant les pratiques de gestion dans les centres de première ligne, les chefs de poste ont mis en place des incitatifs pour motiver le personnel à plus de performance. Le processus motivationnel demeure toutefois très complexe et variable. La désirabilité de bons résultats des soins (valence) est élevée pour tous les professionnels ; cependant les motifs étaient différents entre les catégories de personnel. Par ailleurs, le faible niveau d’équipements et la multiplicité des acteurs ont empêché l’établissement d’un lien entre l’effort fourni par les professionnels et les résultats de soins. Cette compréhension du rôle des professionnels de première ligne pourra aider le personnel administratif à mieux cibler le monitorage de la performance du SRE. Le personnel de soins pourra s’en servir pour reconnaitre et appliquer les pratiques associées à une bonne performance. Dans le domaine de la recherche, les défis de recherche ultérieurs sur les facteurs humains de la performance du SRE seront mieux identifiés. / Although proven effective interventions have been transferred from the developed world, developing countries still face high maternal and perinatal deaths. Weak health systems and human resources crisis hinder the scaling up of those interventions. This study is aimed at improving the understanding of the role of human resources, especially the first line staff in a maternal referral system, the evacuation and referral system (ERS) of Kayes (Mali). The implantation of the ERS as part of major strategies for fighting maternal and perinatal deaths has been generalised in Mali’s eight administrative regions. The ERS main components are: upgrading of emergency obstetric care (EmOC); funding by a solidarity fund; and improving transport and communication by ambulance. These components allows for joint action regarding the supply and the use of maternal and perinatal services. Yet, due to the shortages in skilled birth attendants, the extension of the coverage in emergency obstetric services has been made with less qualified staff. Kayes is Mali’s first administrative region and a cradle of emigration. It therefore offers, with the support of its immigrants, a more diversified primary care provision than elsewhere in Mali. This gives an opportunity of studying the role of primary care staff in the ERS performance. More precisely, the objectives of this research were to: 1 identify characteristics of primary care teams that influenced the ERS performance assessed by the joint mother-child survival and 2) improve the understanding of human resources practices associated with ERS performance. We combined a systematic literature review with a mixed quantitative and qualitative research design. Kayes ERS performance was modelised using Michie and West conceptual framework and the motivational process was explored with Vroom’s expectancy theory. Data for the quantitative analyses derived from an ongoing system of registering all obstetric emergencies implemented since 2004 (GESYRE: Gestion du Système de Référence Évacuation / Management of the Evacuation Referral System) and periodic surveys. A biprobit regression model has been fitted to estimate the effects of the point of entry in the ERS and the primary care team composition on the joint mother-newborn survival. We also conducted semi structured interviews and non participant observations in order to explore human resources practices in purposely selected community health centers. This research confirmed the importance of human resources in ERS performance. When women’ individual risk factors are controlled for, there is a combined effect of distance traveled and the point of entry in the ERS on one side and the community health centers staff on the other hand, on the mother-child joint survival. The best prognosis of care was found in women who directly accessed care at the regional hospital. Having been transferred from a community health center where there is a physician or more than three professionals increased the mother-child survival probability. Physicians in community health centers develop strategies for a better performance of the health centers they managed: direct implication in care and a competitive care environment. Primary care managers set up incentives to motivate staff. The motivation process remains however a variable and complex one. The desirability of good results was common for all staff though its motivations varied according to the type of staff. Besides, due to insufficiency of material and multiples intermediate factors, perception of a linkage between effort and result was low. This increased understanding of the role of primary care human resources in the ERS performance would focus the management of ERS performance on targeted but effective interventions. The workforce could benefit from the study conclusions in identifying and applying best practices. Further research challenges would also be more precisely identified.