New approaches to the collection and analysis of data on mortality and cause-of-death

Mortality Information System (MIS) with responsibility for compiling cause of death data in Brazil was created by Ministry of Heath in 1975. Despite MIS continuous improvements, Brazil was categorized in 2000 as producing medium-quality death registration data. The aim of this proposal is to evaluate the evolution of quality of causes of death data, to describe the profile of deceased with ill-defined or unspecified cause of death and to evaluate the impact of the improvement of quality of data on mortality levels from some specific causes of death among adults and elders (15-59 and 60+) in Brazil from 1980 to 2010. For evaluating cause of death data quality we considered the criteria of completeness and validity. To analyze the profile of deceased with ill-defined cause of death, socio-demographic characteristics and variables on circumstances of death and its certification are considered. Although the important improvement in levels of completeness and validity of causes of death data in Brazil, especially after 2005, differences in quality of data among regions, age and other characteristics persist. With population rapidly aging, to assess quality of data on adult and elderly mortality in order to better evaluate health policy for these age groups is extremely important.

Levels and patterns of mortality by age, sex and race/color are among the most important indicators of health. Research shows that blacks and whites differ widely in their profiles of morbidity and mortality, but little is known about the sensitivity of interracial differences in mortality to different methods of estimation. This paper presents, compares and evaluates life tables by sex and race/color for all of Brazil, constructed on the basis of indirect demographic methods of estimation based on information regarding surviving children and mothers’ mortality (variants of the Brass indirect method). The data was taken from the Brazilian demographic censuses and nationally representative household surveys (PNADS). Information was also obtained by using direct methods, which combine statistics on reported deaths and intercensal growth rates. We discuss the advantages and disadvantages of each method and present confidence intervals for estimates of mortality by sex, age and race/color. Finally, we discuss the impact that racial reclassification may have on recent estimates.

Cause of death data for the period 1997-2009 have been used to estimate the number of mis-attributed AIDS deaths based on the distinct trend in the age pattern of cause specific rates. After adjusting for under-registration of deaths, the relationship between the increase in mortality and the lagged HIV prevalence was used to estimate the mortality prior to the impact of HIV. The excess mortality was considered to be mis-attributed AIDS deaths and accounted for 92% of the estimated AIDS deaths. The analysis indicates a steady increase in the estimated number of AIDS deaths with a peak of 246 000 in 2006, followed by a sustained decrease that accompanied the extensive roll-out of ARTs in South Africa and prevention of mother-to-child transmission programmes.

Des études récentes indiquent une persistance du niveau élevé de la mortalité maternelle dans un contexte de pauvreté au Cameroun. Ces deux informations cruciales justifient l’intérêt scientifique de cette étude sur l’impact du statut socio-économique du ménage sur le risque de mortalité maternelle au Cameroun. Basée sur trois EDS Cameroun entre 1991 et 2004, l’étude permet, après identification des facteurs de risque de mortalité maternelle, de les associer à la pauvreté : Une analyse contextuelle développe, grâce à la méthode de décomposition, le rôle joué par le statut socio-économique du ménage dans l'évolution des niveaux des risques spécifiques liés à la mortalité maternelle. Le changement social observé est analysé pour déterminer s’il découle d’un changement des comportements ou, au contraire, s’il découle d’une dynamique des structures en termes de taille et composition, dans certaines catégories à déterminer. La dernière phase utilise essentiellement des régressions multinomiales pour dresser les profils des femmes à haut risque et expliquer la survenance des risques de mortalité maternelle. Les résultats obtenus sont assez évocateurs et permettent non seulement de valider la littérature récente, mais aussi de plaider pour des actions ciblées susceptibles de réduire le niveau de la mortalité maternelle au Cameroun.

This study attempts to evaluate principal components concomitant to infant mortality. IMR is affected by several factors belonging to maternal characteristics, child characteristics, utilization of health services and socioeconomic status of the people. Each of these factors includes a large number of variables. Categorical Principal Component Analysis (CATPCA) method is applied as a data reduction tool to reduce dimensionality in data. Data for this work have been taken from NFHS III. The analysis produced a three-dimensional solution, which explains 50.47% of the variance with reliability of 89.1% (cronbach’s α 0.891). Total 26.17% variance is explained in dimension one, 12.49% variance in dimension two and 11.81% in third dimension. Maternal factors explain total 26.83% variance and child factors explain total 18.58% variance. Out of ten samples of size n/2 from the same data set, nine samples give similar results.
 

Monitoring maternal mortality is extremely challenging due to issues with data availability and quality. The maternal mortality estimates published by the WHO in 2012 include data adjustment parameters to account for data quality issues, but there is a discrepancy between the WHO assumption about, and observed variability in, misclassification errors in VR observations. Bayesian modeling approaches can be used to provide more data-driven insights into maternal mortality estimates and data adjustment parameters. We propose a Bayesian time series model for the VR adjustment parameters to assess the extent of VR misclassification errors and to provide a plausible assessment of the uncertainty associated with VR observations for which no external quantification of misreporting is available. We find that the proposed model gives a distribution for VR adjustments that is more comparable to the observed biases than the WHO expert distribution. However, given the lack of, and issues with maternal mortality data, validation of modeling assumptions and findings is challenging; more research on measuring maternal mortality and assessing data quality is urgently needed.

This study analyses the experience in Latin America regarding the use of the census question on last live-born child survival. Such question has been included in recent population census with a view of estimating infant mortality. However, it is not always possible to compute all the elements needed to obtain the probability of dying within the first year of life, thus the estimates suffer from some methodological vices that do not offset the costs of introducing a census question.

In this paper, practices across countries that included the question on last live-born child survival in their 2000 and 2010 census are compared, highlighting differences in the question formulation and the scope of the data. The authors estimate infant mortality for those countries having included this question. These estimations are compared to those presented in the publication Child mortality: a database of Latin America since 1960 drafted by ECLAC and UNICEF.

Findings show that figures derived from the census question on last live-born child differ significantly from those published in the ECLAC-UNICEF study for all scenarios and countries analysed. These results show that the use of the question on last live-born child survival in censuses is not performing robustly.

Population censuses are particularly important in countries lacking timely and reliable system of vital statistics. It provides the only opportunity for estimating vital rates that would not otherwise be available. Most countries therefore rely heavily on censuses to obtain information on fertility and mortality rates especially at sub national levels. Most censuses in Africa, reliable estimates of the basic mortality data have always been suspect. The use of indirect methods has been employed to estimate key indicators based on Brass Type methods (UN 1983). However, in recent times many researchers have argued that demographic trends observed do not match the strong mathematical and other assumptions regarding the use of these procedures. In the Kenyan 2009 census, the indirect techniques (both adult and childhood) produced implausible results and therefore mortality indicators were obtained from deaths in the last 12 months. Data on deaths in the last 12 months have been criticized to underestimate mortality levels because of omissions in reporting. The main objective of this study is to evaluate the utility of using this approach for mortality estimations.

We provide a joint overview of trends in mortality under age 5 and between ages 15 and 60 in Sub-Saharan Africa using data on the survival of children and siblings collected in Demographic and Health Surveys. Disregarding conspicuous stalls in the 1990s, child mortality levels mostly declined and converged over the last 30–40 years. By contrast, many Eastern and Southern African countries have witnessed an enormous surge in adult mortality that echoed earlier increases in HIV incidence. In Eastern Africa, adult mortality levels have begun to decline again, in some instances before the large-scale expansion of antiretroviral therapy programs. More surprising is the lack of sustained improvements in adult survival in some countries without severe HIV epidemics. Because trends in child and adult mortality do not always evolve in concert, model-based estimates that are inferred from matching indices of child survival onto standard mortality schedules can be very misleading.

One of the problems in monitoring maternal mortality is lack of reliable estimates. Several studies have provided estimates of maternal mortality from pregnancy related deaths on South Africa utilizing Census and survey data but these estimates have have methodological weaknesses. This study re-examines maternal mortality from pregnancy related deaths at national and provincial levels in South Africa and provides insight into the causes of the maternal mortality deaths.

Data from the 2001 census, 2007 Community survey and death registrations were utilised. Information on household deaths including pregnancy related deaths were collected. Direct and indirect methods were utilised to estimate maternal mortality ratio.
The results indicate that maternal mortality ratio increased from around 150 per 100,000 live births in 1992-1998 to about 764 per 100,000 live births in 2007 and ranged from 102 per 100,000 live births in the Western Cape Province to 1,639 in the Eastern Cape Province in 2007. Maternal infections and parasitic diseases as well as other maternal diseases complicating pregnancy, childbirth and the peurperium are the major causes.
Despite strategies to improve maternal and child health, maternal mortality is increasing and remains a public health challenge in South Africa.