Name: Veirum, Jens Erik Home Country: Denmark Research Country: Guinea Bissau Project period: 2003-2003 Type: Ph.D. thesis Title Childhood Hospitalisations in a West African City. Community-based studies on hospitalisation, acute and post-hospitalisation mortality and long term survival among children admitted to the paediatric ward. Hospital Nacional Simão Mendes, Bissau, Guinea Bissau Abstract The present thesis was planned to describe a clinical intervention and quality assessment project based on analyses of hospital data collected by the Bandim Health Project at the only paediatric ward in Guinea Bissau from 1991 to 1996. Improved case-management was expected to have major impact on hospital morta1ity, and by closely monitoring the implementation process the effects of different interventions were to be examined. The intervention observed turned out to be an unexpected civil war that allowed us to describe the effect of an emergency situation on in-hospital mortality and on post-discharge survival. Linking information from a longitudinal community surveillance system to prospectively registered hospital information, we examined the use of hospital services, hospital mortality and post-discharge mortality prior to the intervention. The estimation of hospitalisation rates was based on an open cohort consisting of 33,727 children, whereas first-ever admission risks were examined within a birth cohort of 8,184 children, contributing with 85,678 and 15,542 person-years at risk respectively. In-hospital mortality was examined for 5,544 paediatric admissions from 1991-96. A total of 4,153 admissions could be identified as belonging to 3,373 persons in the population register and post- discharge mortality could be examined for the 3,467 admissions resulting in live discharges. During the civil war, the hospital surveillance system revealed a significant decline in in- hospital mortality. This may have been due to improved hospital cafe and we therefore examined possible bias and implications. Furthermore, we examined the effects of standard vaccinations on in-hospital mortality and post-discharge survival. Results Almost 45% of all children in the area were hospitalised at least once before the age of 5 years and every fourth death occurred at the hospital. Overall in-hospital mortality was 12.2 % with some differences between diagnostic groups; half of all deaths occurred during the first day after admission and younger children, especially neonates were subject to higher and increasing mortality during the period. Community mortality rates for infants and children under 3 years of age were 110 and 207 per 1000, respectively, during the period 1991 and 1996. The overall in-hospital and 12-month post-discharge mortality was 20%, and compared with community mortality, the mortality risk was 12 times higher during the first 2 weeks after discharge and remained significantly higher in the period 3-6 months after hospitalisation (RR = 2.5 (CI95% 1.6-3.8)). Risk factor analyses were performed for hospitalization, for community and hospital mortality as well as for post-discharge mortality. The risk factors for hospitalisation did not reflect the risk factors for the hospital case fatality rate, whereas good consistency was found between risk factors for community and hospital-related mortality. For instance, having a mother with some schooling was associated with a higher risk of hospitalisation but a lower risk of mortality in the immunity, at the hospital and after discharge. Leaving the hospital without medical discharge was found to be the all-dominating risk factor for early post-discharge mortality, while other significant risk factors were ethnic group, house quality and maternal education, being similar to risk factors for community mortality. The diagnoses associated with high in-hospital case fatality were also associated to high post-discharge mortality. During a 12-month war period the hospital CFR and post-hospital mortality were significantly reduced (0.68 and 0.57 respectively), which could not be explained by changes in recruitment, discharge policy or general changes in mortality. We found effects of immunisations on in-hospital mortality which cannot be explained by the expected disease specific effects of particular vaccines; a strong protective effect of measles vaccine on overall in-hospital mortality (MR = 0.51 (CI95% 0.27-0.98)), being strongest for girls, and a significantly higher CFR for DTP vaccinated girls than for DTP vaccinated males (MR=1.63 (CI95% 1.03-2.59)). Discussion The fact that almost half of all children are hospitalised before the age of 5 years, that one fourth of all deaths before the age of 5 years occurs at the hospital, and that high percentage of the children will pass through the health system during their last illness, indicates that the population, in spite of high in-hospital mortality, does seek the assistance of a weak public health system, and that an improved ability to deal with severe childhood illness could have a considerable effect on the overall child morta1ity in the capital. Considering the resources available at the paediatric service of the National Hospital and the effect of improved case management during an emergency situation, a prioritised management of the severely ill child at arrival and at risk children at discharge, as well as improvements of staff dedication and morale, would be expected to have a major impact overall childhood mortality. Furthermore, the changes in hospital case fatality rates associated with different vaccines were consistent with community findings and warrant further research to understand and control the immune stimulatory effects of routine vaccines. The present study emphasized the importance of primary care activities, as well as a need for a profound understanding of the divergent non-specific effects of common vaccines as these might contribute to better child survival in developing countries. Involved research institution(s) Projecto de Saúde, Bandim / MINSAP Republica da Guinea Bissau Department of Epidemiology Research, the State Serum Institute, Denmark Supervisor(s) Peter Aaby, Department of Epidemiology, the State Serum Institute, Denmark Correspondence