Page 31 - IJPS-5-2

Costa et al.

This will be significant, as almost half of these under-five deaths are newborns whose deaths can be prevented through
higher coverage of quality prenatal care, skilled care at birth, postnatal care for both the mother and infant and the care
of small, sick newborns (UN IGME, 2019), that is, conditions that could be prevented or treated with access to simple,
low-cost interventions. These avoidable deaths are one of the focuses of the UN goals for 2030 (WHO, 2020; ODS, 2020;
Golding et al., 2017).
Monitoring infant mortality and associated risk factors are essential to evaluating public policy and development. This
measure is a strong indicator of socioeconomic conditions, such as poverty, access to education and health services (Gaiva,
Fujimori and Sato, 2015; Carvalho et al., 2020). A study on the determinants of infant mortality can be of considerable
help to the better targeting of public policy funding, which is increasingly possible through online platforms that make data
available, such as DATASUS, which is a Department of Brazilian Health Ministry responsible for collecting, processing,
and disseminating public health data.
In Brazil, the IMR has undergone a continual decrease in the recent decades, mainly due to the reduction in post-
neonatal deaths (those occurring between 28 days and 1 year) and improvements in sanitary conditions. The share of
post-neonatal deaths among total infant deaths declined from 51% in 1990 to 38% in 2015 (IGME, 2020). Neonatal
mortality (deaths occuring less than 28 days), which is the main component of the IMR in Brazil, has also been following
a declining trend, mainly due to favorable changes in factors related to pregnancy and childbirth, despite the increase of
its shared proportion. Neonatal mortality is harder to address and extends to the perinatal period (Duarte, 2007; Carvalho
et al., 2020). Indeed, neonatal mortality has become the biggest challenge in fighting infant mortality, as it currently
corresponds to the majority of such deaths and involves various biological, socioeconomic, and health-care factors.
Neonatal mortality rates (NMR) have decreased but have also become the focus of public policies on infant mortality
due to its proportion of the global IMR. The greater availability and the quality of health data in Brazil have enabled
more precise analyses of this issue on a nationwide scale. This availability of data and the increasing importance of
neonatal deaths on infant mortality have led to a significant number of studies covering different factors, regions, and
methods concerning neonatal deaths.
The aim of the present study is to investigate the maternal factors associated with neonatal mortality by employing the
framework proposed by Mosley and Chen (2003), using a hierarchical model based on the hypothesis that socioeconomic
factors determine behaviors that exert an impact on biological factors. In their influential paper, Mosley and Chen argue
that mortality studies usually have a bias toward the social or biological approach, isolating the external determinants
according to the field of study. Therefore, even if biological factors are directly responsible for deaths, this information
may be insufficient regarding the establishment of adequate recommendations and effective public policies. Maternal
characteristics also have the advantage of being available earlier compared to childbirth or pregnancy-related factors.
Moreover, some studies have shown the predictive power of these factors in affecting the IMR (Fonseca, Flores, Camargo
Jr. et al., 2017; Bertoldi et al., 2019).
This study offers an analysis and discussion of neonatal mortality in Brazil considering a broader perspective available
through a nationwide sample, using data visualization techniques and classifications to summarize the results. The aim
is to help predict the risk of neonatal mortality in Brazil by evaluating important characteristics related to the mother and
considering the period between 2006 and 2016.

2. Materials and Methods
An observational, retrospective, and cohort study was conducted with secondary data on births and deaths of infants in
Brazil between 2006 and 2016. Data were obtained from two sources: Sistema de Informação sobre Mortalidade (SIM
[Mortality Information System]) and Sistema de Informação sobre Nascidos Vivos (SINASC [Live Birth Information
System]), both of which are available through DATASUS (Health Informatics Department of the Brazilian Ministry of
Health) and the Instituto Brasileiro de Geografia e Estatística (IBGE [Brazilian Institute of Geography and Statistics]).
Figure 1 illustrates the linkage process between the two datasets (SIM and SINASC) concerning the characteristics selected
for this study as well as the data selection process. The main problem is the availability of a standard variable between
the two datasets to enable successful merging. The standard variable Declaration of Live Birth Number (NUMERODN)
was only available in 36.5% of the cases (197,971 out of 543,437 deaths), despite the fact that filling in this information is
mandatory. From this percentage, it was possible to link 95% of the cases, resulting in a large dataset. Some entries were
also excluded due to inaccurate data (such as extreme outliers of age – above 90 years). The final sample was 28,362,359
children, representing 151,473 neonatal deaths records. Moreover, the datasets are unbalanced, i.e., the percentage of
death class samples outnumbers the percentage of living class samples.

International Journal of Population Studies | 2019, Volume 5, Issue 2 25