Presentation is loading. Please wait.

Presentation is loading. Please wait.

National Institute for Public Health and the Environment Volksgezondheids toegevoegde waarde van GIS/ruimtelijke analyse bij enkele infectieziekten. Wilfrid.

Similar presentations


Presentation on theme: "National Institute for Public Health and the Environment Volksgezondheids toegevoegde waarde van GIS/ruimtelijke analyse bij enkele infectieziekten. Wilfrid."— Presentation transcript:

1 National Institute for Public Health and the Environment Volksgezondheids toegevoegde waarde van GIS/ruimtelijke analyse bij enkele infectieziekten. Wilfrid van Pelt, Agnetha Hofhuis, Ingrid Friesema, Jan van de Kassteele en vele vele anderen 1.Een simpele vorm van clustering in ruimte en tijd geimplementeerd in 1999 met een internet interface, van 1000 salmonella types voor wekelijkse signalering en retrospectieve inspectie. 2.Eenvoudige ecologische analyse van de ziekte van Lyme bij NL huisartsen (1995, 2001 en 2006). 3.Eerste test van een Bayesiaanse analyse van regionale dichtheid van runderen, kippen en varkens en het voorkomen van patienten met STEC ( ).

2 National Institute for Public Health and the Environment Verloop van de vogelpest epidemie, 2003 pluimveehouderijen, besmet, geruimd Met dank aan Michiel van Boven

3 National Institute for Public Health and the Environment Rubella notifications by 4 digit postal code The Netherlands, –

4 National Institute for Public Health and the Environment MMR-1 coverage by municipality The Netherlands, 2004

5 National Institute for Public Health and the Environment Votes for SGP party by municipality National Elections, The Netherlands, 2003 Percentage of those allowed to vote Met dank aan Susan Hahne

6 National Institute for Public Health and the Environment Outbreak detectie door het vinden van clusters in regio & tijd bij Salmonella

7 National Institute for Public Health and the Environment Algorithm for detection of outbreaks Prospectively expected frequencies and tolerances Optimising sensitivity and specificity Not miss outbreaks but also not too much false alarms time-geography and time-age clusters and demographic aberrations

8 National Institute for Public Health and the Environment Intranet Catalogue / Atlas since Mai 1998 human/ animal/ food/ environment All sero and phagetypes Resistance Actual trends, Early Warning, GIS-clusters Each week backcalculated starting in 1984 > Tables and Figures

9 National Institute for Public Health and the Environment Explosions of S. Typhimurium ft 20 Early-Warning application

10 National Institute for Public Health and the Environment 1st step signal verification: region Silver wedding: Case-control study rPHA Coburgerham, salting process insufficient

11 National Institute for Public Health and the Environment 1st step signalverification: region-crossing Place, Age, Gender

12 National Institute for Public Health and the Environment 1st step signalverification: precedent No cause found: RIVM trawling questionnair, too late Animal Health Service, no clou in region

13 National Institute for Public Health and the Environment How to detect clusters in space

14 National Institute for Public Health and the Environment Methode is simpel, werkt goed en test duizenden potentiële clusters in enkele minuten Ja, maar, regio’s verschillen toch in bevolkingsdichtheid? Klopt!! Maar dat heeft v.n.l. invloed op de grootte en het aantal clusters in een regio.

15 National Institute for Public Health and the Environment Say CHEESE

16 National Institute for Public Health and the Environment Explosion of Salmonella Typhimurium DT7 cases In January 2006 up to April 2007 an explosion of S.T. DT7 infections occurred, resulting in an extra 297 lab- confirmed cases of salmonellosis. (tip of the Iceberg). Labconfirmed: 297 cases Hospitalized: cases Death (<2jr): Symptomatic Infected General population General Practices Laboratories Hospital † Doctor visits: +/- 650 General Population: +/ GE-cases COI: € 0.6 milj DALY: 54 Schattingen!!

17 National Institute for Public Health and the Environment 1 jan 2006 half mei 2006 april 2007 Automatische geografische outbreak detectie Kaas affaire Twente Salmonella Typhimurium Ft561

18 National Institute for Public Health and the Environment Automatische geografische outbreak detectie Kaas affaire Twente Salmonella Typhimurium Ft561 Januari tot half mei 2006 half mei tot eind 2006

19 National Institute for Public Health and the Environment Handmatig aangeven van geografie outbreak Kaas affaire Twente Salmonella Typhimurium Ft561/DT7 januari-december 2006

20 National Institute for Public Health and the Environment Explosion of Salmonella Typhimurium DT104 cases In the after summer of 2005 an explosion of DT104 infections occurred, resulting in an extra 261 lab- confirmed cases of salmonellosis (tip of the Iceberg). Labconfirmed: 261 cases Hospitalized: cases Death (<2jr): Symptomatic Infected General population General Practices Laboratories Hospital † Doctor visits: +/- 650 General Population: +/ GE-cases COI: € 0.5 milj DALY: 47 Schattingen!!

21 National Institute for Public Health and the Environment Outbreak strain different of endemic Importance molecular typing (MLVA), identical DK strain

22 National Institute for Public Health and the Environment Case-control study Case Control study Cases (109 eligible) Controls (411 eligible) Result -Filet-Americain OR 4.2 (1.5–12.0) -Mobile-Caterer OR 4.9 (1.1–22.1)

23 National Institute for Public Health and the Environment Eind december 2007 Begin April 2008 Automatische geografische outbreak detectie Onopgeloste zuivel affaire Salm. Typhimurium Ft651/DT14a

24 National Institute for Public Health and the Environment Automatische geografische outbreak detectie Onopgeloste zuivel affaire Salm. Typhimurium Ft651/DT14a Eind december 2007 tot begin april 2008

25 National Institute for Public Health and the Environment Lyme disease in the Netherlands Agnetha Hofhuis & Wilfrid van Pelt and many others

26 National Institute for Public Health and the Environment Lyme disease in Europe is caused by the Borrelia burgdorferi sensu lato group; B. burgdorferi sensu stricto, B. afzelii, B. garinii Transmission of Lyme disease Transmission by the sheep tick (Ixodes ricinus).

27 National Institute for Public Health and the Environment Lyme disease  Early local infection: erythema migrans (EM) % of B. burgdorferi infections  Early disseminated infection: manifestations in nervous system, skin, joints and heart  Chronic Lyme borreliosis…

28 National Institute for Public Health and the Environment  retrospective studies among general practitioners (GP’s) ?Incidence of tick bites and erythema migrans ?Geographical distribution in the Netherlands ?Ecological risk factors for tick bites and erythema migrans  retrospective analysis of hospital admissions for Lyme disease ?Occurrence of hospital admissions for Lyme disease ?Seasonal and annual trends in hospital admissions for Lyme disease  collecting ticks in 4 different biotopes ?Density of ticks & infection rate of ticks with Borrelia ?Seasonal & annual trends Studies on Lyme disease in the Netherlands

29 National Institute for Public Health and the Environment All (± 8.000) general practitioners (GP’s) in 1995, 2002 & 2006 received pre-coded questionnaire about previous year 1. How many patients with tick bites have you seen? 2. How many erythema migrans case-patients have you seen? 3. How many people are included in your practice population? Retrospective GP-study  postal questionnaire

30 National Institute for Public Health and the Environment  Response, coverage: 88% in % in % in 2005  Tick bite consultations: 1994    EM consultations: 1994   Retrospective GP-study  results Incidence of EM & tick bites per inhabitants

31 National Institute for Public Health and the Environment

32 National Institute for Public Health and the Environment Information on risk factors per municipality Roe deer Rabbits Horses Sheep & goats Cattle Woods Degree of urbanization Tourist nights per year Precipitation Parks & public gardens Sandy soil Uncultivated wet soil Uncultivated dry soil Dunes Retrospective GP-study  ecological risk factors Risk analysis for GP-studies of 1994, 2001 & 2005 together: Poisson regression: (offset: city population)  city repeated measure  year “confounder”  no random (un)structered effects, sofar  no Bayesian smoothing, sofar

33 National Institute for Public Health and the Environment Retrospective GP-study  ecological risk factors Risk factors for tick bites  area with sandy soil  mean precipitation  density of roe deer  rural areas (  houses/ km 2 )  tourist areas  cattle / km 2  rabbits / km 2 Risk factors for erythema migrans  area covered with woods  area with sandy soil  mean precipitation  roe deer / km 2  rural area (  houses/ km 2 )  tourist areas

34 National Institute for Public Health and the Environment  Risk of infection after a tick bite?  Infection rate of ticks: Borrelia, Ehrlichia, Babesia, Rickettsia.  Serology and clinical aspects after a tick bite or erythema migrans.  Case-control study: risk factors for tick bites and erythema migrans.  In 2007 and GPs selected in hotspot areas for tick bites and erythema migrans consultations. National Tick Bites study  GP-based prospective study

35 National Institute for Public Health and the Environment Aims  Accurate incidence figures across Europe for Lyme disease - regional comparison and analysis of regional risk factors  Feasibility of setting up a network of GPs across Europe - answer simple health care questions with a high response - with respect to a known denominator population.  Internet GIS-tool for questioning of physicians - immediately mapping and feeding back the results to GP EU GIS internet GP-study (MedVetNet)

36 National Institute for Public Health and the Environment Geographical distribution of STEC in the Netherlands Wilfrid van Pelt,Loes Bertens, Ingrid Friesema, Jan van de Kassteele (spatial statistician)

37 National Institute for Public Health and the Environment Bekende genoemde risicofactoren:  Consumptie rauwe melk/kaas (16%)  Contact landbouwhuisdier (21%)  Persoon-persoon overdracht (18%)  Consumptie kant en klare groente (28%) Belangrijkste reservoir:  Runderen en Kalveren Acute gastroenteritis met Complicaties  HUS (15%) en Ziekenhuisopname (41%), m.n. 0-4 jarigen Wat algemeenheden STEC-O157 infecties Doel was inventariserend: Spatial relation Cattle density and STEC incidence

38 National Institute for Public Health and the Environment STEC cases Incidence / [0,1.62] (1.62,4.7] (4.7,7.34] (7.34,11.1] (11.1,16.8] (16.8,22.4] (22.4,31.6] STEC-O157 in NL (2.4 / 10 6 inhabitants; / yr.; N=400) → Bayesian smoothing for low population areas

39 National Institute for Public Health and the Environment Seasonality STEC in Veals (1 st ), 1-2 wks later Humans and Dairy cattle week of onset of disease Human STEC O157 cases 0% 5% 10% 15% 20% 25% 30% 35% 40% Positive farms Dairy cattle (faeces) Veal calves (faeces)

40 National Institute for Public Health and the Environment STEC-O157 cases incidence and Cattle densities,

41 National Institute for Public Health and the Environment Average ( ) densities of farm animals / km 2

42 National Institute for Public Health and the Environment Neighbour matrix of 496 communities Rotterdam

43 National Institute for Public Health and the Environment # y[i] = observed counts # lambda[i] = RR.total[i]*E[i] = intensity # E[i] = expected number based on population properties # beta[1] = baseline log(RR) # beta[.] = association log(RR) for covariates # b.struc[i] = area specific spatially structured random effect for residual or unexplained log(RR) # b.unstr[i] = area specific unstructured (exchangeable) random effect for residual or unplained log(RR) # b.struc[i]+b.unstr[i] = effect of latent (unobserved) risk factors (convolution prior) # RR.total[i] = RR.fixed[i]*RR.resid[i] = total relative risk # RR.fixed[i] = RR of all known risk factors together # RR.resid[i] = RR of latent (unobserved) risk factors # # model for (i in 1:n) { y[i] ~ dpois(lambda[i]) log(lambda[i]) <- log(E[i]) + fixed[i] + resid[i] fixed[i] <- beta[1] + beta[2]*cattle[i] + beta[3]*pigs[i] + beta[4]*poultry[i] + resid[i] <- b.struc[i] + b.unstr[i] WinBUGS model called from R using R2WinBUGS library (i is community.age.sex unit) (E[i] is community.age.sex population offset) (classical poisson model) (age and sex fixed components) (standard random effect)(spatial dependence)

44 National Institute for Public Health and the Environment Model results RRP2.5P97.5 (Intercept)0,60,40,8 cattle: 1000/km26,40,845,2 pigs: 1000/km21,00,71,4 poultry: 1000/km21,00,91,0 Age 0-44,53,45,9 Age 5-91,61,22,1 Age ,50,40,7 Age 50+1,0 Males1,0 Females1,31,01,6

45 National Institute for Public Health and the Environment STEC cases Incidence / [0,1.62] (1.62,4.7] (4.7,7.34] (7.34,11.1] (11.1,16.8] (16.8,22.4] (22.4,31.6] STEC-O157 in NL (2.4 / 10 6 inhabitants; / yr.; N=400) Bayesian “smoothed” version (SMR)

46 National Institute for Public Health and the Environment STEC-O157 model: fixed part (explained by cattle/pigs/poultry)

47 National Institute for Public Health and the Environment STEC-O157 model: Structured and unstructured random effects (residuals)

48 National Institute for Public Health and the Environment  First WINBUGS-modelling attempt (1 week work) succeeded, but needs refinement  WinBUGS model called from R using R2WinBUGS library seems elegant  Seasonality, year are not included in the model as yet  Stratified analysis for age and season probably necessary to clearly show the effect of cattle density  Clearly other unknown regional effects are involved Spatial relation Cattle density and STEC incidence Conclusions


Download ppt "National Institute for Public Health and the Environment Volksgezondheids toegevoegde waarde van GIS/ruimtelijke analyse bij enkele infectieziekten. Wilfrid."

Similar presentations


Ads by Google