The mortality rate in children younger than 5 years decreased rapidly in the 10 CI districts from 219 child deaths per 1000 live births (95% CI 189–249) in the period 1991–1995 to 119 child deaths per 1000 live births in the period 2006–2010 (105–132) with an average annual change of –4.1%. The mortality decline was similar nationally and in the non–CI districts ( Figure 3 ).

Figure 3.  Under–5 mortality rates from 1991 to 2010 in Malawi. Data are from analysis of the 2010 national DHS survey in Malawi. Vertical lines show 95% CIs for survival probabilities for the Catalytic Initiative district estimates. Dates on the x–axis represent the 5–year periods preceding the 2010 Malawi DHS.

Improvements in the coverage of interventions relevant to child survival in the 10 CI districts across the time period 2000 to 2013 were found with regard to all indicators except vitamin A supplementation ( Table 1 ). For certain indicators the AAC was higher in the CI period compared to the pre–CI period (early breastfeeding (initiation within 1 hour), measles immunisation and oral rehydration salts (ORS) use) whilst for others (DPT3 immunisation and intermittent preventive treatment of malaria for pregnant women (IPTp) the coverage increases were maintained at the same rate as the pre–CI period. The AAC decreased during the CI period for tetanus toxoid vaccination for pregnant women, exclusive breastfeeding, vitamin A supplementation, care–seeking for pneumonia and insecticide–treated bednets (ITNs) ( Table 1 ). While AAC with regard to care seeking for pneumonia and ITNs was not as large during the CI period, both the pre–CI and CI periods reflected gains in coverage.

The AAC in coverage for the examined high impact child health interventions in the CI districts compared to the non CI districts between 2000 and 2010 showed no statistically significant difference for any of the indicators (Table s7 in Online Supplementary Document(Online Supplementary Document) ), meaning that although coverage levels differed by intervention and by district, the AAC was consistent across the country for each intervention examined.

The effect of the introduction of HSA delivered iCCM can be seen in the ‘place of treatment’ data. Figure 4 shows a steady and significant increase in care–seeking for fever, suspected pneumonia and diarrhoea at community level (HSAs at village clinics) from less than 1% in 2000 to 9% in 2013 with a corresponding significant decrease in children receiving no care from 56% in 2000 to 18% in 2013.

Figure 4.  Place of treatment for fever, suspected pneumonia and diarrhoea in children under–5, 10 Catalytic Initiative districts. DHS – Demographic and Health Survey, LQAS – Lot Quality Assurance Survey. *Other – Private pharmacy, drug vendor/ store, shop, traditional healer, relative/friend.

Given the mortality declines and increases in coverage of critical interventions for child survival, the lives saved analysis predicted an under–5 mortality rate in the 10 CI districts of 84 per 1000 live births in 2013 when starting at a baseline mortality rate of 119 per 1000 live births in 2008.

The proportion of child lives saved in 2013, by intervention, was calculated using the LiST estimation of 10 800 additional deaths averted in 2013 (relative to the 2008 baseline) as a denominator ( Figure 5 ). The pneumococcal vaccine introduction in 2011, with rapid scale up given the existing immunisation platform was estimated to have averted one in four deaths and was the single largest contributor to lives saved in 2013 (24%, 2600 lives saved), followed by ITNs for households (19%, 2100 lives saved) and malaria treatment with artemisinin–combination therapies (15%, 1700 lives saved). Care seeking for suspected pneumonia contributed to 7% of lives saved in 2013 (800 lives saved for pneumonia) and case management of diarrhoea (ORS and zinc) contributed to 5% (400 lives saved for oral rehydration salts (ORS) and 150 for zinc for treatment of diarrhoea). Changes in breastfeeding practices contributed to approximately 6% of deaths averted (600 lives saved). Facility deliveries in the CI districts increased 29 percentage points (from 55% to 84%) in the CI period resulting in improvements in care at birth. These interventions accounted for 11% of all deaths averted even though this was not a direct focus of the CI programme. Of the remaining interventions, no single intervention saved more than 3% of child lives in 2013 ( Figure 5 ). Stunting and wasting rates did not decline resulting in no measurable mortality reduction from interventions to address stunting and wasting. When comparing the pre–CI and CI periods, the proportion of deaths averted pre–CI was 15% in 2007 compared to baseline mortality in 2000, whereas during the CI period, the proportion was nearly double at 30% in 2013 compared to 2008 baseline mortality.

Figure 5.  Percentage of child lives saved in Malawi (10 Catalytic Initiative districts), by intervention, in 2013, relative to a 2008 baseline. Improvements to care at birth include: labour and delivery management, antenatal corticosteroids for preterm labour, neonatal resuscitation, and clean birth practices. WASH indicators include improved water and sanitation and access to water connection in the home. ITNs – Percent of children <5 who slept under an Insecticide Treated Net the previous night; ACT – Artemisinin–combination therapy; ORS – Percentage of children <5 with diarrhoea in the last 2 weeks who received oral rehydration salts; PMTCT – Prevention of mother to child transmission of HIV; WASH – water, sanitation and hygiene; Hib – Haemophilus influenzae type B vaccine.

Major policies and programmes related to child survival were initiated in Malawi between 2004 and 2011 ( Figure 2 ). The Essential Health Package in 2004 prioritised and strengthened community participation and delivery of free community health services through a set of evidence–based high impact interventions for children and adults. In 2005 the Reach Every District strategy was adopted to increase access to these interventions. In 2006, the Government of Malawi, together with UNICEF, WHO and the World Bank launched the Accelerated Child Survival and Development (ACSD) strategy which is an integrated approach based on the IMCI 5–year strategic plan 2006–2011 [25]. Implementation of the CI programme began in 2007 in support of the ACSD policy [8]. In 2008 guidelines for implementation of iCCM were developed and training of existing HSAs began across the country.

The desk review identified improvements in several macroeconomic indicators ( Table 2 ). Between 2000 and 2012 there were noticeable changes in the per capita GDP which increased by $ 110 per capita (in constant 2011 international dollars) from $ 639 in 2000 to $ 749 in 2012) [2]. Per capita total expenditure on health increased from $ 9 in 2000 to $ 24 in 2012. Although external funding for health had been declining since 2006, it remained substantial at 53% of total expenditure for health in 2012, and more than double what it was in 2000 (26%). The poverty headcount (<$ 1.90/ day) declined from 74% in 2004 to 71% in 2010 [2].

A major set–back occurred in 2011 due to a successive combination of the impact of the global economic crisis and the withdrawal of funding locally by bilateral and multilateral partners from the Sector Wide Approach (SWAp) [26]. These events led to the devaluation of the Malawi currency as well as severe shortages in foreign currency and fuel. With Malawi’s extreme dependence on foreign aid, the country’s health budget was substantially reduced in the second quarter of 2011. Development assistance to the SWAp has not resumed but direct funding of projects by partners continues.

Our results show that Malawi, one of the least developed countries in the world, ranking 170 out of 187 on the Human Development Index in 2012 [27], has managed to reduce under–5 mortality in the 10 CI districts by 100 deaths per 1000 live births (a 46% reduction) between 1991 and 2010. This has occurred in the context of changes, mainly at the policy and budgetary level, including increases in funding from both the government and donors [28], which enabled a conducive environment for implementation of the nation’s ambitious child survival policies.

The coverage of many priority child health interventions increased significantly since 2000, particularly preventive interventions including IPTp, immunisations and use of ITNs, as well as health behaviours such as early and exclusive breastfeeding and care–seeking for suspected pneumonia. The health interventions that account for the majority of the deaths averted are delivered at village clinic level through both outreach and stimulating increased demand (eg, immunisation, ITN distribution, care–seeking behaviours and treatment coverage). A number of high impact interventions increased in coverage at a higher rate than prior to the CI implementation. A similar AAC in coverage between 2000 and 2010 was seen in all districts across the country, evidence of the Ministry’s approach of scaling up high impact interventions across the entire country.

The contextual analysis provides insight into the lack of difference in mortality and coverage change between the CI and non–CI districts. The implementation of the Strategic Plan for Child Survival and subsequent scale up of iCCM occurred simultaneously across the whole country supported by a collective of partners and donors allocated to each of the 28 districts with a partnership agreement to guide implementation. Although the 10 CI districts were the focus of DFATD support, other donors implemented the same packages of support in the remaining 18 districts.

Our findings suggest that several factors worked synergistically to achieve the decreases in child mortality in Malawi. First, these improvements have taken place within a context of strong child health policies and clear leadership of the Ministry of Health, which, although donor dependent, channelled support in a co–ordinated manner to effectively implement national child survival policies [10,29,30]. This finding reinforces previous results from analyses of success factors related to declines in neonatal mortality in Malawi [13,31]. Second, the implementation of these policies and high impact interventions has been made possible through substantial external investments in the health system, particularly the contribution of donors. The official development assistance to maternal, child and newborn health in Malawi increased from US$ 51.7 million in 2003 to US$ 154.7 million in 2012 [28]. Third, investments in the capacity of the health work force has plausibly led to improvements in important health behaviours such as use of ITNs, facility deliveries, immunisation coverage, care–seeking and treatment for common childhood illnesses. These investments include strengthening, over decades, a dense network of over 10 000 HSAs nationally as well as the Emergency Human Resource Programme launched in 2004 leading to a doubling of professional health workers (including doctors, clinical officers and nurses) [32].

While HSAs play a very important role in the health system, the contribution that community level care (through HSAs) has made to child survival should not be overstated. While the 7 percentage point increase in care–seeking at the community level is substantial in a context where care–seeking at this level was 2% at the start of iCCM scale–up and almost non–existent prior to this, it does point to a need for greater efforts to generate demand for community care–seeking. However, it should be noted that there is a theoretical ceiling for treatment of childhood illness at community level since HSAs delivering iCCM services are situated in hard–to–reach areas and therefore not all communities sampled in household surveys will have been exposed to iCCM–trained HSAs. With just over a third of HSAs trained in iCCM, the Government of Malawi and its partners have yet to fully leverage the potential of this service delivery platform. However, this task–shifting needs to be aligned with investments to enable effective supervision and mentorship of this front–line cadre [33,34].

A strength of this evaluation is that the 10 CI districts are geographically–dispersed across all three regions of the country, thus limiting regional biases. Population–based household survey data on mortality and coverage from all 10 CI districts were available for analysis from the time of implementation of the CI (2006), during the CI period (sampling in 2010), and at the end of the programme (2013). These data were triangulated with an in–country document review to gain a more in depth understanding of the reasons for the quantitative changes in coverage and mortality.

There were weaknesses to this evaluation. First, the lack of oversampling to provide district level coverage estimates for all districts in the 2000 DHS. Although the data were aggregated across the two strata (CI and non–CI districts), there could be regional biases from the pre–CI estimates. A sensitivity analysis revealed no significant changes to the AAC comparisons (Table s8 in Online Supplementary Document(Online Supplementary Document) ). Second, due to the mix of methods used to collect contextual data, we are unable to quantitatively estimate the relative contributions of wider changes in the health system and beyond to the reduction of child mortality. However we have followed the approach used in two previous country case studies [35,36], one of which is part of the Countdown to 2015 multi–institutional, multi–agency collaboration to track progress towards MDG goals 4 and 5 [36]. This case study will therefore add to the body of literature describing unique national pathways towards improved child survival.

Third, while the LiST model predicted within a margin of the measured mortality change, all factors must be applied when directly linking measured mortality reduction with coverage change. For this reason the results of the LiST analysis should be treated with caution. For example, factors outside of the health sector could have contributed to mortality declines, and/or incorrect assumptions could have been used for coverage of high impact interventions without empirical data available (eg, Kangaroo Mother Care) to run the LiST model.

Malawi has achieved remarkable progress in reducing child mortality and a recently released household survey reports a further decline which will almost certainly enable the country to achieve MDG4 [37]. However, one also needs to consider whether this trend is sustainable given Malawi’s heavy reliance on donor funding for the health sector and the risk of shifting global priorities. Malawi provides a strong example for countries in sub–Saharan Africa of how high impact child health interventions implemented within a decentralised health system with an established community–based delivery platform, can lead to significant reductions in child mortality. This example also highlights the imperative for the international community to commit to longer–term investments in health system strengthening and development as countries move from the “quick wins” related to survival to the more complex goals related to overall child health and development.