Substandard emergency obstetric care - a confidential enquiry into maternal deaths at a regional hospital in Tanzania

Abstract

Objective (i) To identify clinical causes of maternal deaths at a regional hospital in Tanzania and through confidential enquiry (CE) assess major substandard care and make a comparison to the findings of the internal maternal deaths audits (MDAs); (ii) to describe hospital staff reflections on causes of substandard care. Methods A CE into maternal deaths was conducted based on information available from written sources supplemented with participatory observations and interviews with staff. The compiled information was summarized and presented anonymously for external expert review to assess for major substandard care. Hospital based maternal deaths between 2006 and 2008 (35 months) were included. Of 68 registered maternal deaths sufficient information for reviewing was retrieved for 62 cases (91%). As a supplement, in-depth interviews with staff about the underlying causes of substandard care were performed. Results The causes of death were infection (40%), abortion (25%), eclampsia (13%), post-partum haemorrhage (12%), obstructed labour (6%) and others (4%). The median time available for hospital staff to manage the fatal complication was 47 h. The CE identified major substandard care in 46 (74%) of the 62 cases reviewed. During the same time period MDA identified substandard care in 18 cases. Staff perceived poor organization of work and lack of training as important causes for substandard care. Local MDA was considered useful although time-consuming and sometimes threatening, and staff dedication to the process was questioned. Conclusion Quality assurance of emergency obstetric care might be strengthened by supplementing internal MDA with external CE. Soins obstétriques d’urgence de qualité inférieure - enquête confidentielle sur les décès maternels dans un hôpital régional en Tanzanie Objectif: (1) Identifier les causes cliniques de la mortalité maternelle dans un hôpital régional en Tanzanie et au moyen d’une enquête confidentielle (EC), évaluer les soins majeurs de qualité inférieure et comparer ces résultats avec ceux des audits internes des décès maternels (ADM), (2) décrire les réflexions du personnel hospitalier sur les causes de soins de qualité inférieure. Méthodes: Une EC sur les décès maternels a été réalisée sur base des informations disponibles à partir de sources écrites, complétée par des observations de participation et des entretiens avec le personnel. Les informations recueillies ont été résumées et présentées de façon anonyme pour examen par des experts externes afin d’évaluer les soins de qualité inférieure aux normes. Les décès maternels basés à l’hôpital entre 2006 et 2008 (35 mois) ont été inclus. Sur 68 décès maternels enregistrés des informations suffisantes pour l’examen ont été récupérées pour 62 cas (91%). Comme complément, des entretiens approfondis avec le personnel sur les causes sous-jacentes des soins de qualité inférieure ont été réalisées. Résultats: Les causes de décès étaient: infection (40%), avortement (25%), éclampsie (13%), hémorragie post-partum (12%), dystocies (6%) et autres (4%). Le délai médian disponible au personnel hospitalier pour gérer la complication fatale était de 47 heures. L’EC a identifié des soins de qualité inférieure sur 46 (74%) des 62 cas examinés. Pendant la même période les ADM ont identifié des soins de qualité inférieure dans 18 cas. Le personnel a perçu une mauvaise organisation du travail et un manque de formation comme causes importantes des soins de qualité inférieure. Une MDA locale a été jugé utile quoique exigeant en terme de temps et assez intimidant. Le dévouement du personnel au processus a aussi été questionné. Conclusion: L’assurance pour la qualité des soins obstétriques d’urgence pourrait être renforcée en la complétant avec des ADM internes et des EC externes. Cuidados obstétricos de emergencia por debajo del estándar – un análisis confidencial sobre las muertes maternas en un hospital regional de Tanzania Objetivo: (1) Identificar las causas clínicas de las muertes maternas ocurridas en un hospital regional en Tanzania, y mediante entrevistas confidenciales (EC), evaluar cuales son los mayores cuidados subestándar y compararlo con los hallazgos de las auditorías internas de muertes maternas (AMM); (2) describir las reflexiones del personal hospitalario sobre las causas de los cuidados subestándar. Métodos: Las EC de muertes maternas se realizaron basándose en información disponible de fuentes escritas, suplementadas con observaciones y entrevistas al personal sanitario. La información recopilada se resumió y presentó de forma anónima a expertos externos para su revisión con el fin de evaluar los principales cuidados subestándar. Se incluyeron todas las muertes maternas hospitalarias acaecidas entre 2006 y 2008 (35 meses). De 68 muertes maternas registradas había suficiente información para recuperar 62 casos (91%). Adicionalmente se realizaron entrevistas a fondo con el personal sanitario sobre las causas subyacentes de cuidados subestándar. Resultados: Las causas de muerte fueron infección (40%), aborto (25%), eclampsia (13%), hemorragia post parto (12%), parto obstruido (6%) y otros (4%). El tiempo medio disponible para que el equipo sanitario manejase una complicación fatal era de 47 horas. La EC identificó cuidados subestándar mayores en 18 casos. El equipo sanitario percibía la mala organización del trabajo y la falta de entrenamiento como causas importantes de cuidados subestándar. La AMM local se consideraba como útil aunque se percibía que consumía mucho tiempo y a veces era descrita como “amenazadora”, y se cuestionaba la dedicación del equipo médico al proceso. Conclusión: El control de calidad de EmOC podría fortalecerse mediante la suplementación de los AMM internos con EC externos. The 5th of the eight Millennium Development Goals (MDG 5) declared by the United Nations is a 75% reduction in maternal deaths from 1990 to 2015 (United Nations 2000). The global progress towards MDG 5 is minimal, especially in sub-Saharan Africa, where the problem is most severe (Hill et al. 2007; Bryce et al. 2008). The main strategy to reach MDG 5 has been to have a ‘skilled birth attendant’ (SBA) at 90% of all births (World Health Organization 2005). However, a review of ecological and observational studies in developing countries concludes that there is ‘little evidence that giving birth with a health professional reduces the risk of dying, and in some settings it appears to be associated with an increased risk of dying’ (Scott & Ronsmans 2009). An explanation for these findings could be that many SBAs are unable to provide emergency obstetric care (EmOC) when complications occur because they work with insufficient supplies, colleagues, training, supervision and quality assurance (Graham et al. 2001). Delays in receiving EmOC can be divided into three categories: (i) delayed decisions within the family to seek care; (ii) delays in transport; (iii) delays within the health facilities. If health facility delays are a major contributing factor to maternal deaths they must be reduced to an acceptable level before the other delays are addressed (Thaddeus & Maine 1994). Substandard EmOC can be assessed by both internal and external review of maternal deaths. An internal review takes place at health facility level performed by local staff. This process is an opportunity for reflection and direct feed-back (Wagaarachchi et al. 2001). Internal reviews are by their nature local and not anonymous. In contrast ‘confidential enquiries’ (CEs) are carried out at regional or national level and maintain the anonymity of staff and health institutions as cases are reviewed by external experts. The CE conclusions can be communicated to health professionals and decision makers and the civil society. In the United Kingdom (UK), CEs have been carried out for more than 50 years. At present CEs have been introduced to only a handful of developing countries despite being recommended by World Health Organization & Dept. of Reproductive Health and Research (2004). Little research has investigated the impact of maternal deaths audits (MDAs) (Pattinson et al. 2005) and even less the impact of CEs (Hussein 2007) in developing countries. In Tanzania, where 46% of births take place at health facilities, the estimated maternal mortality ratio (MMR: the number of maternal deaths per 100.000 live births) is around 950, one of the highest in East Africa (Bryce et al. 2008). This number seems to have increased between 1996 and 2004. Internal review of all maternal deaths called maternal death audit (MDA) has been imposed as a national policy to improve the quality of EmOC (Tanzanian Ministry of Health 2006). The purpose of this study was to identify clinical causes of maternal deaths at a regional hospital in Tanzania and through a CE to assess major substandard care and make a comparison to the findings of the internal MDA. The study further aimed to describe hospital staff reflections on causes for substandard care. A CE model was applied. To get as detailed a description as possible of circumstances surrounding maternal deaths, different data collection methods were used; written information was retrieved from hospital statistics, case records, hospital registration books and local MDA forms. Participatory observations and interviews with staff contributed additional information. The compiled information was summarized and presented as cases in an anonymous form for external expert review to assess for substandard care. As a supplement to the case-specific CE, in-depth interviews with staff about the causes of delays in care in general were conducted. The study took place in Tanzania at a regional hospital providing comprehensive EmOC. MDA was initiated in January 2006 and a consultant in obstetrics and gynaecology was appointed as medical officer in charge of the hospital in December 2007. One hundred and eight maternal deaths were registered at the regional hospital between 2004 and 2008. The CE was performed for all registered maternal deaths at the hospital from January 2006 to November 2008 (35 months), the same period during which MDAs had been carried out. In all, 68 maternal deaths had been registered and sufficient written information for reviewing was retrieved from 62 (91%) cases (47 from case files and 15 from the ‘hospital registration book’). These 62 cases were included in the CE. According to the Regional Medical Officer’s statistics, only two maternal deaths at peripheral health facilities referring to the study site were registered during the study period and no community based maternal deaths. Data were collected during the periods July–November 2007 and July–November 2008. An exploratory approach with the intention to exhaust all available data sources related to each case was applied. Data sources were: (i) written documentation from registers and case files supplemented by (ii) participatory observations, and (iii) interviews with staff. The case files for registered maternal deaths and the ‘hospital registration book’ were reviewed and detailed notes were made. For supplementary information, MDA reports and registration books from the operating theatre and labour ward were reviewed. Not included were six maternal deaths with insufficient or no written documentation about management (four cases of maternal death from infection and two from eclampsia). Normal deliveries at the labour ward for 4 weeks in August 2007 were observed. During this period and on two other occasions, six cases of maternal deaths were observed. Interviews with staff about clinical management were carried out and added minor information to 27 (44%) cases. As a supplement to the case-specific data collection, semi-structured in-depth interviews about causes in general for substandard care were performed with nine staff representing different professions: three health administrators (HA), three nurse-midwives (NM) and three assistant medical officers (AMO). AMOs have 4 years of pre-graduate training and are responsible for clinical management and surgery in the absence of medical doctors (MD). The interviews lasted from 20 to 45 min, six were recorded and transcribed and notes were written while being conducted. From the available information, the cases of maternal deaths were reviewed. Based on the material, the main clinical causes of death could be categorized under the following: ‘infections’, ‘complications from abortions’, ‘post-partum haemorrhage’ (PPH), ‘pregnancy-induced hypertension’ (PIH), ‘obstructed labour’ and ‘others’ and were listed by frequency (Table 2). The time available to provide EmOC at the hospital was estimated. If the patient arrived at the hospital with the complication, the ‘time to provide EmOC’ was the time from admission to death. Each maternal death was described as thoroughly as possible from the retrieved information and assessed by three external Tanzanian reviewers, all consultants in obstetrics and gynaecology. For each case the reviewers assessed if there was major substandard care involved, in other words, whether the maternal death could have been avoided without the delay. Answer options were ‘yes’, ‘no’ or ‘in doubt’. Each reviewer assessed the cases individually with the possibility of discussing cases or asking questions if clarification was required. If two or three reviewers had answered ‘yes’ the case was judged as involving a ‘significant’ delay. The in-depth interviews with staff about their reflections on causes in general for substandard care were analyzed separately. By reading through the interview transcripts the following main themes were identified: ‘organization of work’, ‘training’ and ‘maternal death audit’. It was a major concern to maintain strict anonymity of the staff and others in the maternal deaths that were reviewed. were presented to reviewers in a where the health and the could not be The study had from the of Medical Research in was from the of Medical Research and from of Health and Tanzania as as from the Medical One hundred and eight maternal deaths were registered between 2004 and 2008. The number of in the years the of MDA in the of and only in the a in obstetrics and gynaecology medical officer in charge of the hospital (Table of the 68 maternal deaths between 2006 and 2008 had been reviewed by MDA and 18 of these were identified as related to major substandard hospital care. The of the during the period 2006 and 2008 from to the median being years. were and had birth more than The clinical causes of maternal deaths between 2006 and 2008 were infection (40%), abortion (25%), (13%), (12%), obstructed labour (6%) and others (4%). substandard care was identified in 46 (74%) cases (Table 2). to the which were conducted during the same time the CE identified major substandard care more with an of The time available to provide EmOC at the hospital between 15 and with a median of 47 (Table 2). of the because of and had no at all, not have of post-partum infection and of these had of performed. of the were by two the while an was Of deaths to had no of performed. The three had an 18 and of the from not in cases the was not reduced to the recommended level of between and A was for a was registered on admission the and severe was by the on and with and was not at hospital for had a with at The of the was and no of was registered the when was A live was by had to post the patient was were not during and and was not The patient from in a of the eight from had been to the hospital from peripheral health facilities with a One arrived in a the other two were in on hospital management was In the had for of of the had performed and only two had been with 2). A for the time at in the severe post-partum haemorrhage and was and the was At that time had no and was not The on and had been had not The was by of the and emergency with and were from an was not available as there was a and the emergency not take was when the patient at The for was of no as it was and the was and a for was One of and the on arrived when the further had been from because of obstructed arrived at the hospital in the of the arrived at the hospital in The had a a and 5 was and in by a staff were about the causes of substandard care they organization of work and poor training as contributing further considered MDA as a useful in delays that had in work However, MDA was considered time-consuming and sometimes The organization of work was as a for the problem is that the staff are not they are not in the the on The can is not are not It takes an to the doctors which take are only a and with no the can One to the and the to in the they If they the in the they to in the to they are at the hospital to this a health are no important delays within the an which to some was by the the problem is that some deaths are out of some in if in in labour are staff EmOC to training of labour ward that they have more to these of staff is the training was in and only a were they are or doctors were have for more training was for training in in the and deaths audits was perceived as in delays in EmOC and the organization of before MDA was the for the to the can they can and an of for MDAs were time and not possible to in to the maternal death. of was as a to discussing quality of care have MDA as a that was not is they information. to make not to be the other not be If are on and to be It is not It is a the of the and supervision on a to on work is Staff to improve and one can that this was to be that has The and of staff were by some time when not when MDA and As it is MDAs are about the for the of health and not to in are to are out with and not The to maternal deaths are not The are during the staff not be for many deaths could be avoided at the is a for a to a In the CE major substandard care was identified in 46 (74%) of the 62 cases of maternal deaths by the external In the internal MDA had only identified major substandard care in 18 cases. EmOC of the and of and were delayed or not performed. The median time available for hospital staff to manage fatal complications was 47 h. In most developing the number of maternal deaths is and the of the study be as the number of maternal deaths at the hospital is However, a study from Tanzania that health facility had registered of all maternal deaths and was more than data from information from and et al. study that in two of three Tanzanian settings the of maternal deaths at et al. In this it be that the study a of the causes of maternal deaths in the only with maternal deaths this CE not the general quality of EmOC at the study the number of by care is not maternal deaths on and by nature a of substandard care. In the United into and Health between and major substandard care was identified in of maternal deaths from direct obstetric causes data collection methods were to get as detailed a description of each case as possible and as a to the of the The observations added information not by other data The and review of maternal death cases to some be although the of an external review can – not – The study not conclusions about the of CE to MDA to improve quality of EmOC. The that CE was more to substandard care than MDA. this is the by in-depth interviews with the staff and related to MDA. health facility delays in EmOC have been described in a number of studies et al. et al. et al. and poor quality care has been from other sub-Saharan settings et al. et al. et al. 2005) not described in as in the present In MDA has been to facility delays in EmOC (Wagaarachchi et al. et al. 2005) and et al. et al. in developing countries. In the in Tanzania was reduced by years when MDA and 20 other to improve EmOC were introduced at the regional hospital 2005). have that evidence for the of MDA is and that it is to assess and that the process can be to staff (World Health Organization & Dept. of Reproductive Health and Research In the present the MDA in a not to have reduced the in the years by the in the United Kingdom a developing countries have introduced Africa in and more and (Hussein For the external CE, the data quality to be than for an internal MDA where staff can information. One to CE is the quality of documentation of management in to maternal deaths. CEs were in and (Hussein et al. 2009). In a of 20 of registered health maternal deaths were reviewed based on case documentation et al. In maternal deaths in the community were reviewed based on interviews with et al. 2009). In both the CE identified substandard care as a main to maternal deaths. These studies that different data collection methods can sufficient information for an external expert In the present study from Tanzania, interviews with staff about the cases not information to the written to maternal mortality to be As in this and other health facility to maternal mortality in developing countries. 90% deliveries is not health facilities births by EmOC. be from community delays to health facility delays in EmOC. delays might be identified to some by registration and management in case records, and an to be important might be a to take the quality assurance process a level An external CE has a of and at the same time clinical in the The CE seems to be in substandard care and of for The CE could health and other important decision makers to for quality A national CE of maternal deaths from different of the health and the to In The CE identified substandard care in three of maternal deaths and identified more cases than MDA the same time It is that quality assurance of EmOC might be strengthened by supplementing internal MDA with external CE. The study was by the of and The no other in the and the are all of the