A metadata driven module for managing and interpreting HDSS verbal autopsy datasets using interVA-4 model

Abstract

Standardised by theWorld Health Organisation (WHO), Verbal Autopsy (VA) is a research survey-based tool widely used to interview relatives, caregivers, friends or witnesses of the deceased to collect data related to the dead in order to determine causes of death in areas where there is no medical record or formal medical attention given and deaths are not recorded routinely. In the past, the findings of the probable cause of death is usually done using the Physician Certified VA (PCVA) method where a team of physicians were used to interpret VA data in order to assign causes of death. However, this method is very time consuming and expensive in term of resources consumption. This has necessitated the need for practitioners to seek other alternate methods of determining Causes of Death (CoD). Among the several methods of collecting and analysing VA datasets, the Tariff and InterVA-4 are most widely used because they are recognised by WHO and International Network for the Demographic Evaluation of Population and Their Health (INDEPTH) Network for their and Demographic Surveillance System (HDSS) sites. The InterVA-4 is a standardised WHO verbal autopsy software used to interpret death related datasets. In this work, we have addressed some data management challenges associated with VA and InterVA-4. Among such challenges is the iterative and continual change of the WHO VA questionnaire (2007, 2012, and 2014). These set of changes come in two folds; the first changes made to the original verbal autopsy instrument by WHO to get a new version. The second usually results from each INDEPTH site adapting instrument for their HDSS area realities. Although these datasets contain spatial information such as global positioning system coordinates, the visualisation on maps of the distribution of causes of death from the verbal autopsy datasets is still lacking in the literature. In this project, we seek to fill these gaps by developing a data model(an abstract model that organizes elements of data and standardizes how they relate to one another and to properties of the real world entities) and platform for VA based on model-driven(used mostly in software design) and meta-data architectures(data about data structure and organisation). We also implemented a geographic information system (GIS) layer that allows display on maps the causes of death from verbal autopsy datasets in the demographic surveillance area (DSA). The tool will enable research scientist to better understand the patterns of causes of death in HDSS sites and aid in accurate analysis of VA datasets.