Do Levels of Infant Mortality Reach Millennium Goal? A classification from DHS Data
maria Silvana Salvini, University f Florence
Fausta Ongaro, University of Padova
As it regards methods used, firstly we show a descriptive analysis of the trends, divided according to the gap with Millennium Goals and secondly we will use techniques of cluster analysis on historical series of infant mortality rates and related variables.
From a descriptive point of view, for example Jordan, Liberia, Peru and Senegal have reduced infant mortality of more than two/thirds so reaching the target previewed by the Goals, while the large majority of countries of Africa, Asia and America Latina present at the end of 2010-2015 a value of infant mortality around the half of late ‘80s. These are the cases of Malawi, Madagascar, Mali, Niger and Rwanda, for sub-Saharan Africa. Other countries present a worst pattern, citing firstly Ghana and Ethiopia in Africa and Haiti in Central America.
The clusters obtained with the above cited variables show the evolution of the countries during the times of the surveys. The pattern of the countries passing from a cluster to another illustrates the passage from a situation to another, generally from a cluster characterized by worse condition to another with a better condition.
Introductionand goals - The decline and differences in infant mortality over time - muchhigher in poor countries than in rich countries - are phenomena worthy ofserious attention by demographers, economists and others (Cutler, Deaton andLleras-Muney, 2006).
InSeptember 2000 United Nations adopted the Millennium Declarationoutlining eight Millennium development goals, with a deadline of 2015, includedthe lowering of infant mortality by two-thirds.
It isrecognized that in developing countries the decline of infant mortality ratesis influenced by intermediate determinants like improved nutrition, vaccinationand medical treatment, as well as by background factors like public health,urbanization, education and female status, economic development (Preston,1980). However, the causal link between the (direct and indirect) factors andinfant mortality is far to be completely explored (Desai and Alva, 1998).
Amongthe background variables women education mostly affects infant mortality(Mosley and Chen, 1984), acting also on fertility levels. The relevance of thisfactor depends on the fact that, as main caregivers, women are most likely toimplement the behaviors that can improve their childrens health (Shapiro andTenikue, 2017). However, it is dubious whether the role of mothers education ismore or less strong depending on the context (i.e, level of health services or economicresources) in which she lives.
Thiscontribution has a two-fold aim:
a) describing the trends of infant mortality ofdeveloping countries looking at the gap, if any, between the current value andthe millennium goal;
b) outlining which are the correlated variables ofthe above-cited patterns
Data andmethods - Our data come from the Demographic and Heath Surveys.Several countries have conducted more than one survey so permitting to disposeof infant mortality data referring to more than one point in time.
Moreover,we dispose of time series of immunization and malnutrition that is theproximate variables of mortality and that represent the most important causesof death (Mosley and Chen, 1984). These factors acts as direct variables and areinfluenced by some background variables, such as mothers education andsocio-economic status of the family.
Finally,we could count on time series of macro-data about infant mortality rates andtheir correlated variables (see legenda table 1) from 183surveys conducted between 1990 and 2015 for 66 countries. We use clusteranalysis to outline the groups of countries according the time pattern ofmortality and correlated variables.
PreliminaryResults - Many countries, i.e. Jordan, have reducedinfant mortality of more than 2/3 so reaching the target of the Goals, whilethe majority of countries of Africa, Asia and America Latina present at the endof 2010-2015 a value of infant mortality around the half of late 80s. Othercountries present a worst pattern, citing Ghana and Ethiopia in Africa.
Theresults of the cluster analysis allowed us to identify 4 clusters. Table 1 showsthe characteristics of them.
Legend:TFR=Total Fertility Rate; IMR=Infant Mortality Rate; IMMU=Percentage immunizedchildren; ISTR_SEC=Percentage of Women with Secondary Level Education;MAFM=Mean age at first Marriage; Place_del=Place of delivery;Child_Stunt=Percentage of children with stunting, that is have interrupted thegrowth of their body; Child_UW= Percentage children born under- weight;HH_Elec=Percentage Household with electricity.
Fig.1 Clustersover time for some countries
Accordingto the level of mortality, the worse situation refers to cluster 4 (87.12),while the best is that of cluster 2 (28.32); cluster 1 and 3 are inintermediate positions (respectively 58.35 and 49.09). The levels of themortality rate determinants generally are consistent with this ranking list:for example, the lowest TFR (2.79) refers to cluster 2 and the highest (5.62)to cluster 4, while cluster 1 and cluster 3 present TFT which are in between.However, the clusters show even results that do not completely support thehypothesis of a strong correlation between infant mortality and somedeterminants (see for example immunization in cluster 1).
Thepattern of the countries passing from a cluster to another illustrates thepassage from a situation to another, generally from a cluster characterized byworse condition to another with a better condition. Fig.1 exemplifies thechanges of pattern over time of 4 countries: Bangladesh moved from C4 to C3 inthe first decade of the new century passing through C1; Burkina Faso moved onlyfrom C4 to C1 during the same decade; Kenia shows a temporary worsening of themortality rate at the beginning of the century, and finally, Niger remains inthe lowest position at least till 2012.
Presented in Session 1234: Mortality and Longevity