Early warning Systems for Vector Borne Climate Sensitive Diseases to Improve Human Health
PART I: Early warning Systems for Vector Borne Climate Sensi<ve Diseases to Improve Human Health (Malaria and Ri* Valley Fever) Nanyingi M O, Kariuki N, Thumbi S,Kiama SG,Bett B, Estambale BPresented at KEMRI-WellcomeTrust, Kilifi Scientific Meeting, Monday 24th June 2013
One Health and Zoonoses in Kenya One Health IHAPPBASSVECTOR SURVEILLANCE
1.0 Study Background and Ra<onale : q The largest health impacts from climate change occurs from vector bornediseases, with mosquito transmitted infections leading in Africaq Climate change alters disease transmission by shifting vectors geographicrange and density , increasing reproductive and biting rates and vector- hostcontact. (Ro)q Climate change to alters land use patterns potentially influencing themosquito species composition and population size, resulting in changes inmalaria and RVF transmission.q Mathematical models for vector density and climate forecasts can predictdisease outbreaks by providing early lead times.q RVF Mortality and Morbidity in Kenya (1998,2006 cycles) (discussed)q In 2011,3.3 billion persons were at risk of acquiring malaria. 216 millionpeople developed clinical malaria in 2010 (81% in Africa), and 655,000 died(91% in Africa, most being children).
1.1 RVF Mortality and Outbreak Model: q Reduction of population vulnerability can be addressed through integratedassessment models which link climatic and non-climatic factors.q Basic dynamic infectious disease models to obtain the epidemic potential(EP) which can be used as an index to develop early warning tools
2.0 Study Goals and Objec<ves : q 2.1 Goal: To develop a framework for integrated early warningsystem for improved human health and resilience to climate–sensitive vector borne diseases in Kenya.2.1 Objectives:q To develop tools for detection of the likely occurrence ofclimate sensitive vector borne diseasesq To assess and compare the temporal and spatialcharacteristics of climatic, hydrological, ecosystems, andvector bionomics variability in Baringo and Garissa counties3.0 Output Indicator Geo-spatial maps of RVF-Garissa and Malaria- Baringooverlaid with climatic and hydrological ecosystems; andvector bionomics.
Study approach and design: q A multi site longitudinal study with quarterly visits.q Determination of point prevalence of P. falciparum infections and RVF inthe study population. testing will be carried out three times annually.q A stratified random sample of 1,220 primary school children aged 5 – 15 yr,RDT for Malaria and indirect IgG+M+A+D ELISA for RVF. Monthly caserecords will be aggregated into divisions and season (rainfall) and calibratedby total population(-ve autoregressive models)q Monthly values (rainfall, temperature, NDVI) will be plotted against logit-transformed diseases prevalence (spatial and inter-annual correlations).q Vector surveillance and risk profiling by site randomization: Habitat census,Adult and larval sampling(weighted probability index for malaria endemicity)q Molecular characterization (PCR) and Phylogenetic tree linkage to risk andvector density-distribution maps.q Arboviral Pathogen discovery (AVID-Google)- ILRI/ICIPE/CDC
Disease Early Warning Systems (DEWS) MEWS (MODIS_NDVI) Are these tools available universally and utilized adequately?HEALTH MAPPER
PART II: Perspec<ves of Predic<ve Epidemiology for RiR Valley Fever in Garissa, Kenya Nanyingi M O, Thumbi SM, Kiama SG,Muchemi GM,Njenga KN, Bett BProject code: C-‐9650-‐15 Presented at KEMRI-WellcomeTrust, Kilifi Scientific Meeting, Monday 24th June 2013
Etiology, Epidemiolgy and Economics of RVFMontgomery , 1912, Daubney 1931, Davies 1975, Jost et al., 2010q RVF viral zoonosis of cyclic occurrence(5-10yrs), described In Kenya in1912 isolated in 1931 in sheep with hepatic necrosis and fatal abortions.q Caused by a Phlebovirus virus in Bunyaviridae(Family) and transmitted bymosquitoes: Aedes, culicine spp.q RVFV is an OIE transboundary high impact pathogen and CDC category Aselect agent.q The RVFV genome contains tripartite RNA segments designated large (L),medium (M), and small (S) contained in a spherical (80–120 nm in diameter)lipid bilayerq Major epidemics have occurred throughout Africa and recently ArabianPeninsula; in Egypt (1977), Kenya (1997–1998, 2006-2007), Saudi Arabia(2000–2001) and Yemen (2000–2001), Sudan (2007) and Mauritania (2010).q Economic losses in 2007 outbreak due to livestock mortality was $10Million , in 3.4 DALYs per 1000 people and household costs of $10 for humancases. 158 human deaths.
10Risk Factors (Ecological and Climatic) q Precipitation: ENSO/Elnino above averagerainfall leading hydrographical modifications/flooding (“dambos”,dams, irrigationchannels).q Hydrological Vector emergency: 35/38spp. (interepidemic transovarialmaintenance by aedes 1º and culicine 2º,( vectorial capacity/ competency)q Dense vegetation cover =Persistent NDVI.(0.1 units > 3 months)q Soil types: Solonetz, Solanchaks,planosols (drainage/moisture)q Elevation : altitude <1,100m aslLinthicum et al., 1999; Anyamba et al., 2009; Hightower et al., 2012
11Study site: Garissa RVF Hotspots CRITERIAq Historical outbreaks in (2006-2007)q Shantabaq, Yumbis,Sankuri ,Ijara, Bura, jarajilla,Denyereq Large ruminant populationsq Transboundary livestock tradeq Transhumance corridorsq Animal clustering at water bodiesq Riverine and savannahecosysytems (vector host contactrates)q Sentinel herd surveillance
Research : RVF Spatiotemporal Epidemiology q Participatory Epidemiology: Ruralappraisal and Community EWS.q Sero-monitoring of sentinel herds andGeographical risk mapping of RVFhotspots?q Trans-boundary Surveillance forsecondary foci.q Disease burden analysis andpredictive modeling???q Decision support tools (Risk maps,brochures, radio…)q Subunit /Clone 13 Vaccine developmentShanta abaqDaadabShimbirye
Process based RVF Outbreak Predic<ve Modelling EPIDEMIOLOGICAL DATA GEOGRAPHIC/SPATIAL DATA Remote Sensing/GIS NDVI, Soil, ElevaIon TEMPORAL DATA Time Series Rainfall, Temperature, NDVI OUTCOMES: SEROLOGICAL DATA (case deﬁni<on) PCR/ELISA(IgM, IgG) Morbidity, Mortality, SOCIOECONOMIC DATA ParIcipatory IntervenIonal costs, Demographics, Income, Assets, CORRELATIONAL ANALYSIS Spatial auto correlationPREDICTIVE MODELLING LOGISTIC REGRESSION, GLM PRVFdiv = Prainfall + Ptemp+ PNDVI+ Psoil + PelevAnalysis of Spatial autocorrelation of serological incidence dataVECTOR PROFILE
Predictive modeling: Logistic Regression/GLM q Historical RVF data (1999-2010)*q Outcome: RVF cases were represent with 0 or 1(-ve/+ve): Cases in 8 of 15 divisions (Dec 2006 –Jan 2007 outbreak)q Predictors: Rainfall, NDVI, Elevationq Data used: 1999 – 2010: 2160 observationsq Univariable analysis done in R statistical computing environmentq model <- glm(case ~ predictor, data, family=“binomial”)- 6 modelsVariable(( Odds(Ratio(OR)( Lower(95%(CI( Upper(95%(CI( p:value(NDVI( 1.9$ 1.40$ 2.9$ <$0.001$Rainfall( 1.08$ 1.05$ 1.11$ <$0.001$Elevation( 1.01$ 0.99$ 1.01$ 0.695$$Univariate Model Mul<variate Model Variable(( Odds(Ratio(OR)( Lower(95%(CI( Upper(95%(CI( p:value(NDVI( 1.47% 1.05% 2.2% 0.03%Rainfall( 1.06% 1.03% 1.09% <%0.001%%??? Beta-binomial logistic regression model, with serologic incidence aggregated at thecompound level as the response, and the climatic metrics as the explanatory variables
Correlation Analysis: NDVI vs Rainfall Pearsons correlation coefficient (r) = 0.458NDVI= 0.411+ 0.764 × rainfall, p< 0.001Linear relationship between rainfall and NDVI: it is thus possible to utilizethese factors to examine and predict spatially and temporally RVFepidemics.
Garissa: Rainfall Es<mate Diﬀerences and MODIS(NDVI) 2013 Dekadalprecipitationona0.1x0.1deg.lat/longdCPC/FEWS RFE2.0*The short-term average may provide insight into changes in RVF risk in areaswhere precipitation anomalies are the principal cause of RVF epidemics byincrease vector competence.
Garissa: Mul< year NDVI Comparison(2006/2007/2012) 16-dayNDVIataresolutionof250m0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 MA A AM M MJ J JJ J JA A AS S SO O ON ND D D2 J JF F FM M 06 07 12 RVF OUTBREAKS USGS LandDAAC MODIS)q Persistence in positive NDVI anomalies (average greater than 0.1 NDVIunits) for 3 months would create the ecological conditions necessary forlarge scale mosquito vector breeding and subsequent transmission of RVFvirus to domestic animals and humans.q Climatic seasonal calendar concurrence with KMD (OND) short rainsand RVF alerts issued by DVS.interannual rainfall var, NDVI of 0.43-0.45/ SST by 0.5 ° ) epidemic indicative** Linthicum et al ., 1999
Where are the Vectors?: Outbreaks and Risk correlation Murithii et al 2010, Be` et al.,2012 q 5 fold probability of outbreak in endemic vs non endemic (62% to 11%)q Response can be geographically targeted (Disease Information Systems).q Vaccine allocation and distribution should be site specific(cost saving mechanism)q Vector surveillance for secondary foci and peri-urban locations (Vectorialcompetence and capacity) for hydrological vector emergence modelling
RVF Monitoring and Surveillance -Community Model q e-‐surveillance and data gathering by (Mobile phones, Digital pen, PDA) q Community sensiIzaIon/awareness by syndromic surveillance. q DisseminaIon of InformaIon through community vernacular radio,SMS Aanansen et al., 2009, Madder et al., 2012 e-‐surveillance
RVF: Decision making Collaborative tools Veterinary ,Public Health, Agriculture,Met UniversiIes,Research InsItuIons Government Vulnerable CommuniIes CAPACITY BUILDING § Risk Assessment § Lab Diagnosis § Informa<on MS § Simula<on Exercise COMMUNICATION § System Appraisal strategy § Par<cipatory message devt (FGD) § Media Engagement(Radio, TV) ONE HEALTH COORDINATION DISEASE CONTROL § Community Sen<nel Surveillance § Vaccina<ons and Vector Control
ACKNOWLEDGEMENTS Data and Financial Support Field work facilita<on q Rashid I M , Garissa q Kinyua J, Garissa q Asaava LL , Faﬁ q Obonyo M, Daadab Study Par<cipants q Bulla Medina CIG, Garissa q CommuniIes: Shanta abaq,Sankuri,Daadab,Ijara,Shimbirye