Swansea-2010 Meeting

Title: Estimating the burden of injuries in sub-Saharan Africa

Location: Swansea, Wales, United Kingdom
Date: 17th September 2010

This meeting was one of a sequence of meetings related with injury measurement that was held at Swansea from 17th to 20th September 2010. In particular, the GBD-Injury Expert group meeting on 18th September discusses closely related issues and methods but in a global perspective.  To find out more about the other meetings, please visit the main meeting page (click here).

The purpose of this one-day meeting was to:
  • Discuss progress on developing regional and country estimates of the burden of injuries.
  • Discuss the technical basis for a cross-country collaboration on injury metrics. Thus, we examined the similarities and differences in information architecture, and identified the needs for common methods.
  • Discuss the value of such research collaboration to all participants with the aim of identifying how the partnership may be structured differently in the future.  
Meeting Program: View as web page or download pdf 

Meeting Summary


Kavi Bhalla started by providing an overview of the project. The primary purpose of this study is to improve estimates of the burden of injuries in Sub-Saharan Africa produced as part of the current revision of the GBD-2010 study. The project is structured around work in a selection of focus countries to develop methods that allow estimating national and regional burden of injuries using existing data sources.

Session 1 Overview _ Architecture of data sources

Session 1: Country Data Inventories

The first session examined the data sources inventory available from six countries and led to discussions on the similarities and differences in information architecture. Poster presentations were made describing country data inventories and early results from the key data sources from: Ghana (Adofo Koranteng), Sudan (Safa Abdalla), Ethiopia (Kunuz Abdella), Nigeria (Uwom Eze), Uganda (Olive Kobusingye), and Mozambique (Jerry Abraham). The following types of common data types were identified from the national data inventories:
  • Census data for estimating injury mortality envelopes (see, for instance, the Sudan poster)
  • Mortuary data for estimating injury mortality patterns (all country posters)
  • Health and Demographic Surveillance Sites (HDSS) for estimating rural injury mortality patterns (see, for instance, the Uganda and Mozambique posters)
  • Injury surveys for estimating non-fatal injury incidence (all country posters)
  • Hospital surveillance data (see, for instance, the Ethiopia poster)
 Sudan Poster
 Ghana Poster
 Nigeria Poster
 Uganda Poster
 Mozambique Poster
 Ethiopia Poster

Common Data Types

Analytical Methods

Session 2: Analytical Methods

Estimating national burden of injuries

The existence of a common architecture of country data sources provides an important technical basis for an international collaboration on analytical methods. The data sources have common technical issues (e.g. handling recall biases, estimating completeness) that require the development of a suite of analytical methods. Similarly, methods are needed to combine these data sources into coherent national burden of injury estimates..This was the focus of the second session. 
This session did not discuss some key data sources that were covered in later sessions and during the GBD-IEG meeting on 18th Sept. 

The second session was introduced by Kavi Bhalla with  an overview of methods, followed by a presentation on using census data, which is a previously untapped source for estimating injury mortality. Many population censuses in Africa include a household mortality module, which occasionally includes a question about whether the death was from injury. Typically such an injury question is included as a weed-out question for estimating maternal mortality. However, analysis of responses to this question can provide accurate estimates of national injury mortality. A comparison of the national death registration data and the 2001 S. African census was used to establish face validity of census-based measurements. Census results from five other countries were also presented.  

Session 3 Special Focus Mortuary Data Overview

This was followed by a presentation by Safa Abdalla that provided an early demonstration of how multiple data sources from Sudan can be combined to construct coherent estimates of the burden of injuries. Safa combined estimates of total injury mortality from the 2008 Census data with Omdurman mortuary data, to estimate the number (and road-user breakdown) of road injury deaths in Khartoum State and Urban Northern Sudan. She showed that her estimates were substantially higher than previously reported statistics. 


Session 3: Special Focus - Mortuary Data

The third session focussed on the use of mortuary data, which exist in most major urban centres in Africa. Several members of the project group have previously conducted research on injury mortality patterns using mortuary data collected retrospectively or prospectively. 
This session focused first on technical issues related with using mortuary data to estimate population- ‐based injury mortality rates in the current project. 

Kavi Bhalla led with an overview of methods to estimate national urban injury mortality patterns from mortuary data, followed by crude estimates of completeness of mortuary data. The results suggest that completeness of city mortuary data (relative to injury deaths in the city) are higher than 60% in many cases. 

Next, Adofo Koranteng, William Ackaah, and James Damsere Derry presented results from an investigation of police under-reporting by record linkage of police and mortuary data in Kumasi, Ghana. The results showed that approximately 32% of the mortuary records could not be linked with police data. 
Under-reporting levels were substantially higher for deaths that occurred outside Kumasi city, but did not vary much by site of death (hospital, brought-in-dead), gender, age, and road-user type. 

Cross National Comparison

Finally, Uwom Eze concluded with a presentation that compared results from all seven mortuary datasets. The presentation showed that, in the current form, the datasets are difficult to compare directly because of several broad issues with mortuary data that include quality of cause coding, especially in retrospectively collected data (e.g. Uganda, Zambia, Sudan-Khartoum). The mortuary data show a common age pattern (dominated by young adult males) and some general similarities in causes, such as the high proportion of road injury deaths in all datasets.

WHO Mortuary Guidelines

Following the discussion of the existing mortuary data, the longer-term perspective on mortuary data was considered. The WHO recognizes the potential value of conducting injury surveillance at mortuaries. Joan Ozanne-Smith and Kidist Bartolomeos presented on the development of WHO/Monash University guidelines which will help to systematize processes for prospective mortuary surveillance. The panel reviewed the proposed mortuary data form. It was agreed that the quality of cause-coding of mortuary-based data collection will improve substantially once the standardized WHO instrument are used. However, the panel also felt that there are several other considerations that are important for the use of mortuary data for estimating injury mortality (such as estimating data completeness, documentation of medico-legal laws and practices) that were not discussed as part of the WHO guidelines. 

Session 4 Panel Discussion

Session 4: Future Collaboration

The final (fourth) session involved a panel discussion on improving future international collaboration. Because of time constraints, part of this session was conducted in parallel with one of the ICE break-out sessions on 20th Sept. Collaboration is successful when all participants make progress towards their research and policy agendas. Thus, the session re-assessed the likely outputs of the current project in relation with the primary research and advocacy goals of the collaborators. Based on this, it was decided that two types of reports will be produced in this study. The first is directed at country policy makers and will be crafted either as national burden of injury reports (e.g. for Ghana), or on a topic of national importance, e.g. road injuries and/or homicide in Nigeria. The second are cross-country reports of the burden of injuries and thematic reports on methods (for mortuary, census, survey data). 

This project on estimating the burden of injuries in Africa is supported by a grant from the World Bank Global Road Safety Facility to Harvard University. 

The meeting was made possible by core support from the World Health Organization and help from a number of local organizations, including, St David’s Medical Foundation, the School of Medicine, Swansea University, Public Health Wales NHS Trust, Children in Wales and the Thematic Research network for emergency UnScheduled and Trauma Care (TRUST). Support from the Violence and Injury Prevention program of the World Health Organization and the World Bank Global Road Safety Facility enabled the participation of a considerable number of researchers and officials from low and middle income countries.

For further information: Please contact Kavi Bhalla (kavi_bhalla@harvard.edu)
Last Update: 6 October 2010
Kavi Bhalla,
Sep 26, 2010, 11:47 AM