Can we achieve Millennium Development Goal 4? New analysis of country trends and forecasts of under-5 mortality to 2015

Summary

Background

Global efforts have increased the accuracy and timeliness of estimates of under-5 mortality; however, these estimates fail to use all data available, do not use transparent and reproducible methods, do not distinguish predictions from measurements, and provide no indication of uncertainty around point estimates. We aimed to develop new reproducible methods and reanalyse existing data to elucidate detailed time trends.

Methods

We merged available databases, added to them when possible, and then applied Loess regression to estimate past trends and forecast to 2015 for 172 countries. We developed uncertainty estimates based on different model specifications and estimated levels and trends in neonatal, post-neonatal, and childhood mortality.

Findings

Global under-5 mortality has fallen from 110 (109–110) per 1000 in 1980 to 72 (70–74) per 1000 in 2005. Child deaths worldwide have decreased from 13·5 (13·4–13·6) million in 1980 to an estimated 9·7 (9·5–10·0) million in 2005. Global under-5 mortality is expected to decline by 27% from 1990 to 2015, substantially less than the target of Millennium Development Goal 4 (MDG4) of a 67% decrease. Several regions in Latin America, north Africa, the Middle East, Europe, and southeast Asia have had consistent annual rates of decline in excess of 4% over 35 years. Global progress on MDG4 is dominated by slow reductions in sub-Saharan Africa, which also has the slowest rates of decline in fertility.

Interpretation

Globally, we are not doing a better job of reducing child mortality now than we were three decades ago. Further improvements in the quality and timeliness of child-mortality measurements should be possible by more fully using existing datasets and applying standard analytical strategies.

Introduction

Child mortality is an important measure of health and development.¹ Sound measurement is needed so that we can learn from national success and identify countries where extra efforts are needed to accelerate the rate of decline in under-5 mortality (formally defined as the probability of death between birth and age 5 years per 1000 births). As performance-related disbursement by global-health initiatives gains momentum, robust measurement of under-5 mortality will take on even greater operational importance for development assistance.

In view of the importance of under-5 mortality, annual or periodic assessments are produced by UNICEF,² WHO,³ UN Population Division,⁴ and the World Bank.⁵ These assessments are based on a mixture of data sources: complete vital registration systems for some countries, partial vital registration systems, complete birth histories implemented in the Demographic and Health Surveys (DHS) or similar surveys, survey or census questions on the number of children ever born and the number surviving, and sample registration systems.6, 7 Over the past decade, WHO, UNICEF, UN Population Division, and the World Bank have not always agreed on past estimates, predictions for a recent period, or forecasts. Advances have been made to put all data sources used for tracking child mortality by these agencies in the public domain and to harmonise the work of identifying past trends and generating current estimates across the agencies.⁸

Despite these welcome advances, five main issues continue to limit the quality and usefulness of evidence on child mortality. First, for many countries we strongly suspect that data sources for child mortality are missing from international databases. For example, as of May 1, 2007, the most recent measurement in the UNICEF public domain database for Mexico was 1990 and for India it was 1997. In fact, since 1990, Mexico has had four national surveys containing data for child mortality, two censuses with questions on children ever born or children surviving, and nearly complete vital registration data available up until 2005. In the case of India, Indian sample registration system data are available up to the end of 2005, as are the 2000 census results and data from two rounds of district-level surveys that include child-mortality measurements.

Second, figures produced and published do not distinguish between statistics based on actual measurements corrected for a range of known biases, and predictions based on a model.9, 10 Distinguishing between observed, corrected, and predicted figures is essential for monitoring and evaluation work, for which predictions have no real role.⁹

Third, in principle, UNICEF fits historical data and predicts beyond the most recent measurement using a type of interval regression.⁷ However, the exact application of these methods and how and when forecasts are modified from the basic model are not reported for every country. For example, the estimate for Cote d'Ivoire of a 2005 under-5 mortality of 195 per 1000 is difficult to reproduce from the data and methods published by UNICEF. Fourth, despite calls from the scientific community,11, 12 estimates of levels and trends in child mortality do not routinely come with uncertainty bounds. In the absence of such limits, users of data naturally assume that all figures are known with equal precision.

Finally, there seems to have been a tendency to overestimate levels of child mortality in several sub-Saharan African countries. This situation might have arisen from a scarcity of recent data, judgments about the expected effects of the HIV epidemic influencing the interpretation of incomplete data, exclusion of some survey datapoints as outliers based on expectations of mortality increases, or a combination of these factors.

Here, we attempt to address some of these limitations by proposing new reproducible methods and, in doing so, contribute to the movement towards enhanced transparency of measurement by using all datasets available in the public domain and reanalysing existing surveys to elucidate more detailed time trends.