|Nielsen, Jens. 2005. Mortality pattern and interventions during the war in Guinea-Bissau 1998-99. Statistical methods and inference in complex emergency situations. Bandim Health Project, Danish Epidemiology Science Centre, Statens Serum Institut. University of Copenhagen.
|During a complex humanitarian emergency, as the war in Guinea-Bissau 1998-99, the infrastructure, public services and food security will deteriorate or collapse and the risk of infectious diseases will increase; implying increased morbidity and mortality especially for vulnerable groups. The aim of humanitarian aid interventions is to prevent or mitigate the suffering of the affected population or vulnerable groups. Humanitarian interventions must be evaluated; both internally to optimise management of the actual intervention and externally by the effect on community to show which interventions are effective.
CESs as well as humanitarian interventions cannot be randomised and will be performed in a complex setting. Hence, interventions cannot be evaluated in one design/analysis, but should be analysed in different designs. If these analyses converge, a plausible impact may be concluded.
Mortality is the measure that most accurately represents the health status of a population. CESs are by international standards defined by a doubling in mortality, and interventions should also be evaluated by their effect on mortality. For comparison this demands a reference, a baseline/expected mortality, i.e. mortality as it would have been had the CES not occurred. We used Time Series methods to forecast an expected mortality adjusted for trend, seasonal variations and periods of excess mortality (epidemics) based on the previous 3.5 years before the war. Standard base-line levels of mortality exist, and the local level of mortality prior to the CES may be used.
However, they will not compensate for trends and fluctuations.
The war in Guinea-Bissau may be divided into a first half-year (June-November l998) during which most fighting took place, and a second half-year (December 1998 - May 1999) during which a peace treaty had been signed, peace-keeping troops had arrived and internally displaced persons (IDPs) had returned more permanently. In the first half-year, the over-all crude mortality (CMR) in the Bandim Health Project (BHP) area of Bissau was increased by 78% (95% CI: 61%-97%) and the crude mortality for children under five years of age (CMRU5) was 2.07 (95% CI: 1.79-2.38) increased. Hence, the war was a complex humanitarian emergency according to international standards. In the second half-year mortality returned to pre~war level, though CMRU5 was elevated in the first three months. The mortality pattern during the war paralleled the percentage of children who were internally displaced.
Most inequalities in mortality associated with socio-economic differentials persisted during the war even though the better-off households initially had a lower relative mortality compared to the expected inequality in mortality. Behaviour like breastfeeding and a motivated staff at the paediatric ward at the national hospital, and to some extent education in the household, contributed to reduced childhood mortality. The inequality associated with the cultural differential ethnicity showed, surprisingly, an over-all negative effect of being of the ethnic group Pepel. Though Pepeis normally have a relatively increased mortality, the exaggeration probably should be attributed to the fact that it was the land ethnically affiliated with Pepeis that mainly was exhausted by IDPs, deteriorating conditions for this ethnic group more than other groups.
Children 6 to 59 month of age were offered vitamin A from October 1998 at the regular three-monthly home visits. We evaluated this intervention in different designs. A step-wedged design indicated a 51% (MR 049; 95% CI: 0.09-2.70) reduction in mortality, and a comparison with mean pre-war mortality indicated a 12% (MR 0.88; 95% CI: 0.41-1.87) reduction. From September 1998, malnourished children, MUAC 130 mm, observed at the three-monthly home visits were referred to treatment at the local health centres, where they received supplementary feeding and treatment. An internal evaluation of the programme showed that among those receiving treatment 1% died, 67% recovered and 32% abandoned; compliance being 89%. Seventy-four percent of the referred children received treatment. Externally evaluated by impact on community, the level of malnutrition among undernourished children did not deteriorate. In addition, the prevalence of malnutrition did increase initially, but decreased during the war, and no increase in mortality of malnourished children was observed, though it would have been expected. Further, at the paediatric ward, where drugs, medical equipment and food were provided, case fatality halved during the war and the post-discharged mortality was reduced. In all three interventions equitable impact was
The tradition of the Guinean population for sharing few resources and its custom to cope with hardship may have been to the major advantage of the civil population in this conflict, though humanitarian aid may also have provided some assistance and it is plausible interventions have reduced the impact of the CBS and prevented further deterioration of the public health situation.
We have shown that it is possible to make a plausible estimation of the impact of a war and to evaluate intervention. The latter evaluation used at least two - non trivial - designs, to document a plausible impact of the intervention. Moreover, methods to project expected mortality during an CES have been explored and used.