Research Plan

We intend to derive regional estimates of the burden of injury by analysis of data sources that already exist in the region. This implies that we will not do any prospective primary data collection. Instead, we will focus on developing analytical methods to correct for bias and extrapolate from existing data sources. The following illustrate a few key components of the process of estimating the burden of injuries in sub-Saharan Africa:

- Estimation of rural injury mortality patterns using verbal autopsy data collected by Demographic Surveillance Sites (DSS).
- Estimation of urban injury mortality patterns using data from various urban mortuaries and civil registration systems.
- Estimation of regional non-fatal injury incidence by extrapolation from data collected by community surveys and national health surveys.
- Estimation of burden using data collected by hospital based injury surveillance systems.

We anticipate that we will need analytical solutions to a cluster of methodological issues in such estimation. These relate with:

- Quality of cause-coding: These include methods to minimize biases in reallocation of cases coded to partially-specified causes and estimating external causes from data sets that only code nature of injuries.
- Methods for estimation underlying population: These are particularly important for estimating incidence with data sources that do not have a well-defined catchment population (e.g. hospitals, mortuaries).
- Methodological issues with household surveys: Particularly relating to recall biases, differential item functioning, and measurement of injury severity.
- Extrapolation and estimating uncertainty: We will develop extrapolation methods that rely on injury pyramids and mappings from external causes to nature of injuries.

Beyond the implications of this work for sub-Saharan Africa, this study seeks to establish methods for estimating injuries in information-poor settings.