- Top of page
- Sources of inconsistency
- Moving forward
Monitoring progress in reducing maternal and perinatal mortality requires suitable indicators. The density of emergency obstetric care (EmOC) facilities has been proposed as a potentially useful indicator, but different UN documents make inconsistent recommendations, and its current formulation is not associated with maternal mortality. We compiled recently published indicator benchmarks and distinguished three sources of inconsistency: (i) use of different denominator metrics (per birth and per population), (ii) different assumptions on need for EmOC and for EmOC facilities and (iii) failure to specify facility capacity (birth load). The UN guidelines and handbook require fewer EmOC facilities than the World Health Report 2005 and do not specify capacity for deliveries or staffing levels. We recommend (i) always using births as the denominator for EmOC facility density, (ii) clearly stating assumptions on the proportion of deliveries needing basic and comprehensive emergency obstetric care and the desired proportion of deliveries in EmOC facilities and (iii) specifying facility capacity and staffing and adapting benchmarks for settings with different population density to ensure geographical accessibility.
La surveillance des progrès dans la réduction de la mortalité maternelle et périnatale nécessite des indicateurs appropriés. La densité des services de soins obstétricaux d’urgence (SOU) a été proposée comme un indicateur potentiellement utile, mais différents documents de l’ONU établissent des recommandations incohérentes et la formulation actuelle n’est pas associée à la mortalité maternelle. Nous avons compilé des repères indicateurs récemment publiés et avons distingué trois sources d’incohérence: (1) l’utilisation de différents dénominateurs métriques (par naissance et par population), (2) des hypothèses différentes sur les besoins en SOU et sur les services de soins obstétricaux d’urgence et (3) le manque de spécifier la capacité des services (charge des naissances). Les directives de l’ONU et le manuel exigent moins de SOU que le rapport 2005 de l’OMS et ne précisent pas la capacité des accouchements ou les niveaux du personnel. Nous recommandons (1) de toujours utiliser les naissances comme dénominateur pour la densité de SOU, (2) de clairement indiquer les hypothèses sur la proportion des accouchements nécessitant des soins obstétricaux d’urgence de base et complets, et la proportion souhaitée des accouchements dans les établissements offrant des SOU et (3) de préciser la capacité des services et du personnel et l’adaptation des repères pour les régions avec des densités différentes de population pour assurer une accessibilité géographique.
Monitorizar los progresos en la reducción de la mortalidad materna y perinatal requiere de indicadores adecuados. La densidad de centros para Cuidados Obstétricos de Emergencia (COEm) se ha propuesto como un posible indicador útil, aunque diferentes documentos de las Naciones Unidas hacen recomendaciones inconsistentes al respecto y su actual formulación no está asociada con la mortalidad materna. Hemos reunido indicadores de referencia recientemente publicados y encontrado tres fuentes de inconsistencia: (1) el uso de diferentes métricas como denominadores (por nacimiento y por población), (2) diferentes asunciones sobre la necesidad de COEm y de centros para COEm y (3) fallos para especificar la capacidad de los centros (carga de nacimientos). Las guías y el manual de las Naciones Unidas recomiendan menos centros para COEm que el Informe Mundial de la Salud 2005, y no especifican la capacidad para atender nacimientos o los niveles de dotación de personal. Recomendamos (1) siempre utilizar los nacimientos como el denominador de densidad de centros para COEm (2) dejar claras las suposiciones sobre la proporción de nacimientos que requieren cuidados básicos y cuidados obstétricos integrales de emergencia, y la proporción deseada de nacimientos en centros COEm, (3) especificando la capacidad de los centros y la dotación de personal, y adaptando las referencias para emplazamientos con diferentes densidades de población para asegurar la accesibilidad geográfica
- Top of page
- Sources of inconsistency
- Moving forward
Reducing maternal and perinatal mortality is a human rights imperative and is central to the Millennium Development Goals (MDGs). Suitable indicators are needed to guide and monitor the progress of interventions to reduce this mortality. The two main indicators used to track MDG 5a, namely the maternal mortality ratio and the proportion of births delivered by a skilled attendant, have limitations for monitoring. In the absence of civil registration, obtaining maternal mortality estimates requires large sample sizes and thus estimates either ‘tend to be imprecise’ or ‘costly to produce’ (Bailey & Paxton 2002). The proportion of births delivered by a skilled attendant as gathered from household surveys is only a proxy for skilled attendance, as it does not capture whether the attendant was truly skilled and worked in an enabling environment. Furthermore, it cannot distinguish whether low coverage stems from a lack of supply or a lack of demand for services.
In their ‘Guidelines for monitoring the availability and use of obstetric services’, UNICEF, WHO and UNFPA (1997) promoted a set of process indicators, the first of which is emergency obstetric care (EmOC) facility density [although they refer to it as essential obstetric care (Hussein & Clapham 2005)]. EmOC comprises basic emergency obstetric care (BEmOC), which can be provided by skilled birth attendants, such as midwives or doctors, at a well-equipped health centre, and comprehensive EmOC (CEmOC), which requires surgical skills and thus is usually provided by doctors in a hospital setting. EmOC functioning is generally assessed by checking whether certain signal functions are being provided at a facility (Box 1) (UNICEF, WHO, UNFPA 1997; WHO, UNFPA, UNICEF, AMDD 2009). EmOC facility density is a potentially useful monitoring tool, because it captures a supply-side intervention that is thought to be essential for saving lives, and having sufficient availability (density) of EmOC services is a prerequisite to their use (UNICEF, WHO, UNFPA 1997). This indicator was subsequently adopted and tracked for 27 countries by the ‘Countdown to 2015’, which seeks to hold governments to account for MDGs 4 and 5.
Table Box 1 . Signal functions used to identify basic and comprehensive emergency obstetric care services (from WHO, UNFPA, UNICEF, AMDD 2009‘Monitoring emergency obstetric care, a handbook’)
|Basic services||Comprehensive services|
|(1) Administer parenteral antibiotics*||Perform signal functions 1–7, plus|
|(2) Administer uterotonic drugs (i.e. parenteral oxytocin)*||(8) Perform surgery (e.g. caesarean section)|
|(3) Administer parenteral anticonvulsants for pre-eclampsia and eclampsia (i.e. magnesium sulphate)*||(9) Perform blood transfusion|
|(4) Manual removal of the placenta|
|(5) Remove retained products (e.g. manual vacuum extraction, dilation and curettage)|
|(6) Perform assisted vaginal delivery (e.g. vacuum extraction, forceps delivery)|
|(7) Perform basic neonatal resuscitation (e.g. with bag and mask)|
Definition and benchmarks
The indicator of EmOC availability is spelled out both as number of EmOC facilities per population and per births, and benchmarks (i.e. minimum recommended levels) are given for both, but the population denominator is used most commonly. Number of EmOC facilities per 500,000 population is the most frequent indicator in the ‘Needs Assessments’ of the Averting Maternal Death and Disability Program (AMDD) and partners (Paxton et al. 2006), and the ‘Countdown to 2015’ reports this indicator as percentage of the recommended minimum of five EmOC facilities per 500 000 population (UNICEF 2008). Combining these two sources, we identified 37 countries reporting this indicator, based on national or near-national surveys. However, we found that the number of EmOC facilities per 500 000 population correlated poorly with maternal mortality (Figure 1a) and that using births as the denominator instead yields a much better correlation (Figure 1b).
Figure 1. Association between emergency obstetric care facility density and maternal mortality. Maternal mortality ratio (MMR) and emergency obstetric care (EmOC) facility density are shown for 37 countries. While there is no correlation between MMR and number of EmOC facilities per population (Panel a), higher number of EmOC facilities per births is associated with lower maternal mortality (Panel b). EmOC, emergency obstetric care; MMR, maternal mortality ratio, from UN Maternal Mortality 2000–2008 estimates. (WHO, UNICEF, UNFPA, The World Bank 2010). aAs reported in Countdown 2015 (UNICEF 2008) and Paxton et al. (2006), bCalculated from a using national CBRs from UNdata (http://data.un.org). UN handbook benchmark of five EmOC per 500 000 population or 20 000 births (WHO, UNFPA, UNICEF, AMDD 2009) represented by the vertical line.
Download figure to PowerPoint
The World Health Report (WHR) 2005 (WHO 2005b) with its accompanying policy brief (WHO 2005c) and Technical Working Paper (WHO 2005a) revised the UN guidelines’ recommendations for EmOC facility density and provided additional EmOC staffing indicators. However, this attempt to improve this indicator and benchmark and to introduce other supply-side indicators has not been picked up in the 2009 update of the UN guidelines, named ‘Monitoring emergency obstetric care, a handbook’ (WHO, UNFPA, UNICEF, AMDD 2009). Researchers and policy makers wanting to employ meaningful EmOC indicators are thus presented with inconsistent recommendations from different UN documents.
Table 1 shows the main indicators recommended for monitoring EmOC availability, compares their benchmarks and lists their underlying assumptions.
Table 1. Comparison of the main EmOC indicator benchmarks and underlying assumptions. (implied numbers in italics)
|Indicator source:||UN guidelines 1997||Technical working paper for the World Health Report 2005||World Health Report 2005||UN handbook 2009|
|Reference population||500 000||120 000||120 000||500 000|
|Reference births (per year)||20 000||3600||3600||20 000|
|Crude birth rate (births per 1000 population)||40||30||30||40|
| EmOC facilities per reference population or per reference births|| At least 5 ||2|| 2–3 (or more) ||At least 5|
| CEmOC facilities per reference population or per reference births||At least 1||At least 1||1||At least 1|
| BEmOC facilities per reference population or per reference births||At least 4||At least 1||1–2 (or more)|| At least 5 minus CEmOC facilities |
| Midwives per reference population or per reference births|| ||At least 20||20|| |
| Doctors per reference population or per reference births|| ||4||3 part-time|| |
| Midwives per EmOC facility|| || 10 in 1 CEmOC and 10 in 1 BEmOC, or 10 in each of 2 BEmOC ||E.g. 10 in CEmOC, 5 in each of 2 BEmOC|| |
| Doctors per CEmOC facility|| ||4||3 part-time|| |
| Proportion of births needing EmOC facility||At least 15%||100%||100%||Up to government|
| Proportion of births needing CEmOC||At least 5%*||7%†|| 17–18%‡|| |
| Births per midwife|| ||175||175|| |
| Complicated births per doctor|| || || 200 || |
This article aims to reduce confusion by distinguishing different sources of inconsistency between proposed indicators and indicator benchmarks and provides suggestions for the modification and harmonization of future recommendations.
Sources of inconsistency
- Top of page
- Sources of inconsistency
- Moving forward
The inconsistencies between indicator benchmarks as evident in Table 2 stem from the following three main sources.
Table 2. Comparison of different EmOC benchmarks for a population of 500 000, and implications of crude birth rate on their consistency
|Source:||UN guidelines benchmarks||WHR 2005 benchmarks|
|Metric:||Per population||Per births||Per population||Per births|
|CBR||Births||Minimum EmOC (CEmOC) facilities||Minimum EmOC (CEmOC) facilities||Minimum EmOC (CEmOC) facilities||Minimum EmOC (CEmOC) facilities|
|50||25 000||5 (1)||6.25 (1.25)||8.3 (4.2)||13.9 (6.9)|
|40||20 000||5 (1)||5 (1)||8.3 (4.2)||11.1 (5.6)|
|30||15 000||5 (1)||3.75 (0.75)||8.3 (4.2)||8.3 (4.2)|
|20||10 000||5 (1)||2.5 (0.5)||8.3 (4.2)||5.6 (2.8)|
|10||5000||5 (1)||1.25 (0.25)||8.3 (4.2)||2.8 (1.4)|
(i) Different denominator metrics: Benchmarks for the number of EmOC facilities required are given in relation to the two different denominator metrics defined for the indicator, EmOC facilities per births and EmOC facilities per population, with different assumptions made regarding the crude birth rate (CBR). (The UN guidelines assume a CBR of 40 and the WHR 2005 of 30, Table 1). Worldwide, CBRs range from eight births per 1000 population in Japan to 54 births in Niger (United Nations Statistics Division). This means that even benchmarks within the UN guidelines or WHR 2005 will be inconsistent if a country’s CBR differs from the assumed one. Table 2 demonstrates that this can lead to big differences in the recommended number of facilities needed. For example, a Peruvian Departamento of 500 000 people with a CBR of 20 (leading to 10 000 annual births) would (according to UN guidelines) either need a minimum of five or of 2.5 EmOC facilities, depending on whether population or births were used.
(ii) Different assumptions on need for EmOC and for EmOC facilities: Assumptions regarding the proportion of births in need of emergency obstetric care (because they have a complication), and regarding the need for emergency obstetric facilities, are not clearly stated and appear to differ between guidelines. These assumptions are crucial for interpreting the benchmarks’ implications for facility size and staffing. Table 3 shows the birth loads implied by the UN guidelines and the WHR 2005 under different assumptions of need. For instance, if the UN guidelines assume that around 15% of births require EmOC and that only these need to be delivered in EmOC facilities, then the proposed benchmark of five EmOC facilities per 20 000 births implies that 3000 annual births need to be catered for, corresponding to an average of 600 births per EmOC facility. However, if they assume that provision should be made for all women to deliver in a facility where EmOC can be provided if needed, irrespective of whether women actually require emergency care, the same benchmark implies 4000 births per facility. Similarly, if 5% of births are assumed to require CEmOC, then a benchmark of one CEmOC facility per 20 000 births implies that this facility needs to care for 1000 complicated cases yearly.
Table 3. Implications of different assumptions regarding need on number of births in EmOC and CEmOC facilities
|Assumptions||UN guidelines benchmark||WHR 2005 benchmark|
|Percentage of births in EmOC facilities (%)||Percentage of births in CEmOC facilities (%)||Births per EmOC facility||Complicated births per CEmOC facility||Births per EmOC facility||Complicated births per CEmOC facility|
(iii) Lack of specification of facility capacity: The UN guidelines and handbook do not specify either the facility capacity (birth load that can be managed) or the staffing of EmOC facilities. This introduces ambiguity when comparing actual numbers of EmOC facilities with the benchmarks. For instance, countries may have apparently sufficient numbers of facilities, but these may be too small to serve all women in need. Conversely, in urban areas, a few large facilities may readily meet needs.
- Top of page
- Sources of inconsistency
- Moving forward
To move forward, we make the following suggestions:
(i) Providing benchmarks in both birth and population metric is confusing and can lead to fivefold differences in requirements. As the indicator is meant to be measuring the supply side of delivery care for pregnant women, relating the benchmarks to births is more useful and will give consistent measures across populations with different CBRs. Moreover, EmOC density per births appears to be a much better predictor of maternal mortality than EmOC density per population (Figure 1b vs. a). The population metric disadvantages countries where the CBR is relatively low; for example, Morocco with 21 births per 1000 population does not meet the UN guideline benchmark using the population metric (3.5 EmOC facilities per 500 000 population) but exceeds it using the birth metric (6.6 EmOC facilities per 20 000 births). If benchmarks per population are required for planning because ‘most health planning is done in relation to population size’ (WHO, UNFPA, UNICEF, AMDD 2009), they can easily be converted using local CBRs that tend to be readily available.
(ii) Assumptions on the proportion of deliveries needing basic and comprehensive emergency obstetric care and the desired proportion of deliveries in EmOC facilities need to be clearly stated and distinguished. While it is indisputable that all women with obstetric complications need EmOC, opinions differ on whether only these women, or all women, should be able to deliver in facilities capable of providing EmOC (Stanton 2008). The ‘skilled attendance for all’ approach promotes the latter, arguing that most complications are unpredictable and many need quick attention (Campbell & Graham 2006). Part of ensuring that women have ‘skilled attendance’ is to ensure that their skilled attendant is in an enabling environment that allows the attendant to manage at least the most basic complications, in other words, in a BEmOC facility at least. Others consider a clean setting sufficient for most births, provided timely detection, and referral to EmOC facilities can be ensured if complications arise (JHPIEGO 2004).
The UN guidelines appear to adopt the latter approach, stating that women with normal deliveries ‘may well be better off at home’ and proposing a ‘minimum acceptable level’ of 15% of births in EmOC facilities (UNICEF, WHO, UNFPA 1997). Their 2009 handbook update declares that the ‘optimal long-term objective is that all births take place in (or very near to) health facilities in which obstetric complications can be treated when they arise’ but refrains from proposing a percentage (WHO, UNFPA, UNICEF, AMDD 2009). However, the unchanged benchmark for EmOC facility density seems to imply only complications (15% of births) are catered for, as the implicit birth load per facility (4000 births) would otherwise be unrealistically high for most settings (Table 3).
By contrast, the WHR 2005 explicitly asserts that ‘all mothers and newborns, not just those considered to be at particular risk of developing complications, need skilled maternal and neonatal care’, and the report consequently based its benchmarks on meeting this need (WHO 2005b). The term used in the report to describe that type of skilled care is ‘first-level care’ rather than basic EmOC, thus emphasizing the complexity and continuity of preventive measures and attention to situations that may turn into complications as well as timely management and referral of complications. It would be desirable if other authors could be as explicit. Ideally, the discrepancies should be resolved, terminology clarified and all UN recommendations based on the same assumptions.
Assumptions on need for CEmOC, also referred to as ‘back-up care’ (WHO 2005b), lack a strong empirical basis (Ronsmans et al. 2002; WHO 2005b) and range from over 20% to 1–2% of deliveries. Need for CEmOC may partly depend on how well first-level care functions (WHO 2005b), and it would be desirable to have better estimates for different situations. It also seems judicious to include the need for neonatal emergency care as done in the WHR 2005.
Facilities offering CEmOC not only provide care for referred or self-referred complicated cases, but also attract uncomplicated cases not in need of CEmOC. To conceptually separate this double function, it may be useful to imagine a midwife-led BEmOC (first-level care) facility at the same site. For planning, birth loads and staffing required for both levels of care can then be calculated separately (as in Table 3).
(iii) Benchmarks on EmOC facility density should incorporate facility capacity (birth load) and/or staffing as otherwise figures can be grossly misleading. For example, five small facilities may be insufficient in some settings, while two large ones could cater for all births. The WHR 2005 attempted to estimate the minimum required number of midwives and doctors in delivery care, as 20 and three per 3600 births, respectively (WHO 2005b), and considered their distribution in facilities. Unfortunately, this was not taken up in the UN handbook 2009, which – like its predecessor, the UN guidelines – failed to consider facility size. It may make sense to develop the ideas set out in the WHR 2005 further and define a minimum first-level/EmOC unit, e.g. a team of 3–4 midwives and then use the number of units instead of the number of facilities for the EmOC density indicator. Following the WHR 2005 estimates (which require verification and country adaption), one such unit could cover around 600 births and thus a minimum of 6 units per 3600 births would be needed.
Furthermore, health professional density can be used as an indicator in itself. Data from Zambia and Sri Lanka (Gabrysch et al. 2011) suggest that it has better discriminatory power and thus, may be a more useful indicator than health facility density (particularly if the latter ignores facility capacity). While the WHR 2005 based its recommendations on district birth loads and an average work load of 175 births per midwife (WHO 2005b), the WHR 2006 also suggested a benchmark for health professional density and based this on ecological analyses (WHO 2006). It estimated that to achieve a minimum coverage of 80% births by skilled attendants, 2.28 total health professionals (counting all doctors, nurses and midwives, not just those working in delivery care) are needed per 1000 population (WHO 2006). It would probably be preferable to use births rather than population as the denominator metric, as discussed previously for facility density. While it may make sense to count all health professionals, given that numbers of health workers in delivery care may be difficult to obtain, detailed planning would require countries to start getting a handle on numbers of professionals actually providing delivery care. To establish an indicator of health professional density for delivery care and EmOC, further work would also be required to establish what a realistic workload per midwife and doctor is in different situations.
The WHR 2005 also briefly addressed the issue of geographical accessibility of facilities, proposing a larger number of smaller facilities for more dispersed populations and discussing the trade-off between efficiency, quality and accessibility. There are no clear guidelines, however, to apply to different contexts. It seems expedient for future guidelines to specify benchmarks in terms of facility number and capacity for settings with different population density. We understand that some of the ambiguities stem from a desire not to be too prescriptive as countries differ and ‘one size does not fit all’. But clarity on assumptions helps planners understand where and how they are planning differently. For example, a small urbanized country may have fewer but larger facilities. Using minimum EmOC units, as discussed earlier, and considering how to best distribute these for different population density scenarios, as sketched out in figure 6 in our previous article (Gabrysch et al. 2011), may be helpful in this context. Furthermore, calculation of the proportion of births within a certain distance of EmOC can indicate whether geographical coverage is adequate (Gabrysch et al. 2011).