• No results found

Governance timeline

7 Unpacking effects – outcomes

7.1

Has the FHCI contributed to saving lives in target groups?

7.1.1

Maternal mortality

Maternal mortality levels are extremely difficult to measure accurately in many developing countries because these countries tend to lack the vital registration systems that can accurately record births, deaths and the causes of death.

It is also difficult to measure maternal mortality through surveys. Even in countries where maternal mortality levels are high, the number of maternal deaths that will be recorded in a survey such as the DHS will be relatively small, even with large sample sizes. This means there is a great deal of uncertainty about the estimates produced.

In addition to trying to measure maternal mortality itself, there are often a number of attempts to model maternal mortality levels. These use available variables that are thought to be related to maternal mortality such as place and attendance at birth, fertility levels, GDP and other factors.

For Sierra Leone, Figure 67 shows three series of estimates for the MMR. The DHS aims to measure maternal mortality itself (indirectly using the sisterhood approach). The UN inter-agency group on maternal mortality produces regularly updated modelled estimates for all countries including Sierra Leone, while the Institute for Health Metrics and Evaluation (IHME) also produces modelled estimates at the country level.

Figure 66: Maternal mortality ratio

0 500 1000 1500 2000 2500 3000 3500 4000 4500 1990 2000 2010 Dea ths per 100,000 li ve bi rths UN centralestimate UN uncertainty interval DHS central estimate DHS confidence interval IHME central estimate IHME uncertainty interval

Two overall conclusions can be drawn from the information in the chart. First, Sierra Leone has had and still has extremely high levels of maternal mortality. For example, the UN estimate shows an MMR figure of 1,360 maternal deaths per 100,000 live births in 2015 – the highest ratio in the world.

Second, the exact level and trend of maternal mortality levels in Sierra Leone is very difficult to measure with any certainty. Each of the series comes with wide margins of uncertainty and the mortality levels and the direction and shape of the trends are different between the three sources. As the IHME report (2010) notes, ‘The differences between global modelling efforts, which are at times substantial, emphasise the influence of each of the analytical steps used to estimates maternal mortality.’

DHS estimates

The DHS asks women aged 15 to 49 about their siblings who have died, including when they died, and, for sisters, whether they were pregnant or had recently been pregnant at the time of death. Deaths to sisters who were pregnant or within two months of the birth or termination of a

pregnancy are included in the estimates of maternal mortality.

As the numbers of maternal deaths recorded in the surveys are small, the estimates relate to a period of seven years before each of the surveys. Thus, the chart shows the estimate centred at 2009 for the 2013 DHS as it relates to the period 2006 to 2013 – and likewise for the 2008 survey.

The estimates show MMRs of 857 from the 2008 survey and 1,165 from the 2013 DHS. However, because of the wide confidence intervals for both these estimates we are not able to say with certainty whether these estimates are different. They do not provide evidence that levels of maternal mortality have changed between the two periods.

UN inter-agency group estimates

The UN estimates are modelled and the explanatory variables include the following: GDP (rising since the end of the civil war); general fertility rate (declining slowly); skilled birth attendance levels (increasing slowly); and the measures of MMR from the DHSs.

The central estimate shows a falling trend for MMR from 1,990 in 2005 to 1,360 in 2015. However, the uncertainty intervals are so wide that it is not possible to draw conclusions about the true direction of the trend over the last 10 years with any certainty.

IHME estimates

The IHME estimates were produced after testing a wider range of covariates. In addition to those used by the UN, IHME tested the inclusion of the following in their model: HIV death rates for women aged 15 to 49 years; neonatal death rates; coverage of four or more ANC appointments; and malnutrition in children under five. Not all covariates were retained in the models for each country.

IHME produces the lowest figures for MMR of the three sources. However, the trend for the central estimate is gently rising from 521 in 1990 to 623 in 2013. As with the DHS and UN figures, the uncertainty intervals are so wide that we cannot draw conclusions about the size and direction of any change in MMR.

7.1.2

Child mortality

Mortality in relation to children under five years is easier to measure than maternal mortality. Estimates are shown in Table 34 from the 2013 DHS, while Figure 68 compares the UN inter- agency group estimates with those from the 2013 DHS.

Table 34 shows the trends in neonatal, infant and under-five mortality over the last 15 years from the 2013 DHS. Mortality rates for all three indicators are falling, although the declines in neonatal mortality (i.e. deaths in the first month of life) are relatively small. For under-five mortality, rates fell by one-third between 1999 to 2003 and 2009 to 2013.

Early neonatal deaths (within the first week) represented 81% of all neonatal deaths in the 2013 DHS and this was unchanged from the 2008 survey. The perinatal mortality rate (which includes both stillbirths and early neonatal deaths) was 39 per 1,000 pregnancies in 2013 compared with 34 in 2008.

Table 34: Neonatal, infant and under-five mortality rates

Years

1999 to 2003 2004 to 2008 2009 to 2013

Neonatal mortality* 48 46 39

Infant mortality** 152 127 92

Under-five mortality*** 227 194 156

* Deaths under the age of one month per 1,000 live births. ** Deaths under the age of 12 months per 1,000 live births. *** Deaths under the age of five years per 1,000 live births. Source: DHS 2013

As for the estimates for maternal mortality, the UN inter-agency group uses additional variables to model under-five mortality including, in particular, mortality estimates from the MICS – the latest of which was conducted in 2010. The UN has also incorporated their own annual disaggregations of the 2013 DHS under-five mortality data into their model. Figure 68 shows both the modelled figures going back to the 1950s and the disaggregated DHS data for the years between 1997 and 2012.

The modelled data show that under-five mortality has clearly been declining gradually, from very high rates, for many years. This decline appears to have accelerated slightly from around 2000 onwards.

The annual estimates of under-five mortality from the 2013 DHS show a gradual decline from 1997 to 2009. However, there is then a very sharp decline from 187 deaths per 1,000 live births in 2009 to 147 in 2010. This coincides closely with the launch of the FHCI. The following years also saw continued declines to reach 126 deaths per 1,000 live births in 2012.

Figure 67: Under-five mortality

7.1.3

Summary

The picture on changes in mortality following the introduction of the FHCI is mixed. This is partly due to the difficulty in measuring maternal mortality in the absence of a robust vital registration system.

The latest UN estimates of maternal mortality put the levels in Sierra Leone at the highest in the world. Their central estimates do show declining levels but these are accompanied by wide uncertainly intervals that make it difficult to draw firm conclusions on the trend. It is not possible to say from the figures how, if at all, maternal mortality has changed as a result of the FHCI.

The situation for child mortality is more positive. The UN-modelled estimates show a declining trend. The UN has also produced annual estimates of under-five mortality using the 2013 DHS. These show a sharp reduction in rates immediately after the start of the FHCI. The levels fell from 187 deaths per 1,000 live births in 2009 to 147 in 2010. The level continued to fall in the following two years reaching 126 per 1,000 live births in 2012.

After