The estimated total expected non-health GDP loss ascribed to child deaths of Int$ 150.3 billion is about 6 % of the combined 2013 GDP of the 47 African countries . This estimate denotes the expected loss in potential GDP in the future from the 2 976 000 child deaths revalued relative to the base year 2013, i.e. present values. The use of forgone future earnings assumes that changes in child mortality rates are reflected in changes in future earnings and national income (as measured by GDP). This assumption may not always hold because such estimates are influenced by a number of transient factors such as distribution of income, education and employment opportunities. This means that a reduction in child mortality may not necessarily translate into increases in GDP. Thus, the expected loss of Int$ 150.3 should be viewed as an estimate of the economic value of lives lost due to premature mortality; and not an indicator of resources that would be saved if those lives were saved.
We applied a discount rate of 3 % because it was used also in the WHO health systems’ performance assessment , the global burden of disease studies , the Institute for Health Metrics and Evaluation’s global burden of disease studies  and the economic evaluation studies on health interventions in Africa . Nevertheless, to test the effect of the discount rate on the total expected non-health GDP loss estimate, a one-way sensitivity analysis was conducted at 5 % and 10 % discount rates. Using a 5 % discount rate reduced the total expected non-health GDP loss by Int$ 39.3 billion (26 %) and the average non-health cost per child death by Int$ 13 193, whilst application of the 10 % discount rate decreased the grand total non-health GDP loss by Int$ 87.2 billion (58 %) and the average non-health cost per child death by Int$ 29 316. This signifies that the magnitude of the total economic loss is partially dependent on the discount rate utilized.
We used 2.5 years (a simple average) as the average age at death. This value was used owing to the lack of data on age distribution of child deaths. Nonetheless, since the distribution of child deaths is unlikely to be uniform over the 0–5 year range, a sensitivity analysis was conducted to determine the effect of age on the total non-health GDP loss estimate. The model was first re-estimated assuming an average age at death of 0 years. The utilization of this value raised the total non-health GDP loss by Int$ 3.7 billion, a 2.5 % increase.
The model was re-estimated assuming an average age at death of 5 years. This average reduced the total non-health GDP loss by Int$ 5.98 billion, a 4 % decrease. This implies that the magnitude of the expected non-health GDP loss to a limited extent also depends on the average age used for the onset of child deaths. Therefore, there is need for more investments in research to come up with reliable data on age distribution of child deaths in Africa.
To a large extent child morbidity and deaths and the associated microeconomic and macroeconomic losses could be prevented if all children had unfettered access to the available and cost-effective newborn, infancy and childhood interventions [42, 43]. WHO provides details on the packages of interventions essential for children for the home or community level, and primary level and referral health facilities, and which, if implemented to scale, could end preventable child deaths . Over a decade and half ago, WHO and United Nations Children’s Fund (UNICEF) published a document presenting an integrated approach to improving management of childhood illnesses, which is still effective .
For childhood interventions to be effectively and efficiently delivered in an integrated manner to the needy population groups, the national and local health systems need to be strengthened to become resilient to shocks of whatever kind [46, 47]. That entails programmatic leadership and governance to plan, guide, support, monitor and evaluate health promotion and service delivery within the model of a continuum of care, where the health services are always available, accessible, safe and acceptable; the health workforce is of adequate numbers and mix and has the required range of competencies; life-saving supplies and commodities are available; technology is up to date; the health financing system covers health promotion and services for pregnant women, newborns, infants and children; and health management information systems are effective .
There is need for investments in other sectors to adequately address socioeconomic determinants of health, including building or strengthening relevant structures to ensure that civil and vital registration systems that facilitate tracking of child births, mortality and causes of death are functional  and strengthening national health research systems to promote the generation and use of epidemiological and clinical research, social-cultural and behavioural change research, implementation research, and health systems and economic research [50–53]. Similarly, human rights tools and frameworks will need to be strengthened to achieve better outcomes, to apportion accountability for women’s and children’s health, and to institutionalize maternal, newborn and child mortality censuses [13, 48].
Limitations of the study
Cost-of-illness studies like the one reported in this paper strictly are not meant to inform public health priority setting because they do not compare the costs and consequences of alternative interventions that could prevent child morbidity and mortality [54, 55]. Therefore, the purpose of our study was not to guide priority setting but rather to raise awareness of the public and policy-makers in the ministries of health and finance on the negative impact of child deaths on non-health GDP.
The study did not include direct health-care costs such as those related to vaccines, drugs, tests, supplies, hospital personnel, diagnostic equipment and physical facilities; direct non-health-care costs of treatment such as transport to and from the health service provider; patient time costs for treatment such as those relating to travel and waiting and treatment time; cost of the time informal caregivers, volunteers, family or friends spend accompanying or visiting the sick person; loss in productivity due to morbidity; or intangible costs such as pain and grief [56, 57].
The analysis reported in this paper is based on estimates of under-five mortality reported in the World Health Statistics 2015 . Those estimates are derived wherever possible from death registration data reported annually to WHO. Unfortunately, very few African Region countries have civil registration and vital statistics systems (CRVS) that permit adequate and regular tracking of mortality and causes of death . For instance, out of the 46 WHO African Region Member States in 2007, only Algeria, Mauritius, Seychelles and South Africa had a death registration coverage rate of 75 % or higher . For countries where such data are not available or are of poor quality, WHO uses household surveys (for births and child deaths) and censuses to prepare estimates of mortality rates and life expectancy. As AbouZahr et al.  eloquently state, the need for support to countries to develop functional CRVS and to institutionalize international classification of diseases cannot be overemphasized.