Presentation on theme: "The case of Sierra Leone Nadine de Lamotte - MSF OCB London Scientific day, 7 June 07 Use of mortality data in humanitarian response."— Presentation transcript:
The case of Sierra Leone Nadine de Lamotte - MSF OCB London Scientific day, 7 June 07 Use of mortality data in humanitarian response
Introduction Two mortality surveys. Focus on 2 nd survey. Operational response to surveys with specific focus on malaria.
Map of Sierra Leone Liberia Guinea Atlantic ocean Freetown
Country background. War officially over in January Sierra Leone “famous”for its poor health indicators (OMS 2006): – MMR: 2,000/100,000 live births. – Under 5 mortality the highest in the world at 282/1000 live births. –Life expectancy at birth: 37 male / 40 women.
Local context: Bo. Second largest city in Sierra Leone. Population of the district: Hyper-endemic for malaria. National malaria protocol changed in 2004 to ASAQ after efficacy studies showed high failure to SP & CQ.
OCB operations in Bo. MSF in Bo since Actual target population: MSF hospital (530 admissions/month). 1 therapeutic feeding centre (150 admissions/month). 5 clinics ( consults/month). Malaria is key morbidity/mortality hence lobbying for country ACT implementation.
1 st mortality survey: April – June 2005 Part of 3-sample access to health care survey to document access barriers in different systems of payment: - Cost recovery in MOH area - Flat fee in MSF H area - Free care in OCB area
Results: death/10.000/day. Total deaths reported as being due to malaria /fever: 39%. In < 5 deaths: 62%.
Operational response to survey => Need to do sensitisation of local population on malaria, “show” Paracheck and ACT in the villages, distribute bed nets. Jan - June 2006: mapping of villages, population data, recruitment & training of outreach teams. Outreach & bed net distribution started in June 06. Monitoring bed net use: around 80% of the bed nets were seen hanging.
2 nd mortality survey Sept 2006: Reassess mortality following 2005 survey: -Retrospective mortality in catchment area of the clinics. -Causes of death (verbal autopsy). -Health seeking behaviour in those that died.
Methods Study population: ( ) 4 chiefdoms Sth Bo. Sampling method: 3 level cluster; each cluster= 30 children/ families. Family questionnaire: composition, mortality (recall period 97 days), health seeking behaviour. Child questionnaire: anthropometric data. Analysis: EpiInfo, deaths / / day.
Results (1) 907 families included. Total n = 5179 (<2yrs=8.4%; <5yrs=76%) 89 deaths ( 5yrs=44) Mortality rate / people / day 95% confidence interval CMR <2 MR – 9.5 <5 MR – 4.9
Results (2) Malaria related mortality. < 2 yrs = 71% (n=23), but recall period covering peak season (39% in June 05). < 5yrs = 53% (n=7) (62% in June 05). All malaria deaths = 42% (n=37). Died at home (all) = 74%. Died at hospital (all) = 25%.
Limitations Sampling error: sampling methodology, rainy season means remote villages inaccessible and more malaria (versus 2005 survey). Measurement error: definition of malaria as fever in survey (over-estimation?). Recall bias: long period of recollection, lack of maternal deaths (stigma?).
Operational response to 2 nd survey: 3 year pilot plan. => Bring ACT closer to population via PHUs: Identification / mapping of 5 PHUs per clinic, staff training on RDT & ACT use (March 07). Prospective mortality follow-up through weekly data collection at community level. Continue sensitisation of population on malaria and health seeking behaviour. Op research agenda: study ACT efficacy < 2, mortality surveys, baseline study…