Animal health economics and precision farming

PowerPoint-presentatie

Economics of animal health and precision farming

Henk Hogeveen

Who am IBorn on a dairy farm (1966)Animal science at Wageningen UniversityEpidemiology (simulation model of management regarding cystic ovaries)Economics (long term effects of herd health management programs)PhD at Fac. Veterinary Medicine (AI to diagnose mastitis)Professor in Animal health managementBusiness economics group, Wageningen UniversityFaculty of Veterinary Medicine, Utrecht University

@henkhogeveenslideshare.net

animal-health-management.blogspot.com

OutlineThe Dutch dairy sectorEconomics of animal healthDisease control: optimizationModeling economics of animal healthExamples normative modellingEmpirical modellingModern management: Precision dairy farmingFinal remarks

Cattle through the last 14 years (* 1,000)Now: ~1.3 bln kg milk

De totale rundveestapelDaling van 2000 -> 2005 daarna weer gestegen tot bijna 4.000.000

4

Grass-based systemSummerFresh grass + corn silage + concentratesWinterGrass silage + corn silage + concentratesHalf mixed ratio

Dutch dairy sector

Increasing farm seizesMostly own (family) labourHalf grass-based systemGrazing under pressure (farm management)Stimulated (societal preference)Volatile milk pricesMore challenges on management

and ..Animal health is becoming more and more importantStimulated by dairy industryImprovement of udder healthImprovement of claw healthImprovement of longevity

OutlineThe Dutch dairy sectorEconomics of animal healthDisease control: optimizationModeling economics of animal healthExamples normative modellingEmpirical modellingModern management: Precision dairy farmingFinal remarks

Economic effects of animal disease

Output

MilkMeatEggsDraft power.After: McInerney, 1996Human benefit (utility)Input

LandLabourCapital

The field: Economic effects of animal disease

Output

MilkMeatEggsDraft power.Disease

1. Lower efficiency2. Lower suitability for consumption3. Lower human well-beingHuman benefit (utility)Input

LandLabourCapital1. 2. 3. Most economic work

J. Agricultural Economics, 1996

Types of animal diseasesProduction diseasesOn-farm optimizationExternalities E.g., mastitis, lameness, APPEndemic contagious diseasesOn-farm control decisionInteraction between farmsE.g., BVD, Aujeszkys disease Notifiable contagious diseasesRegional control decisions (eradication)SurveillanceE.g., FMD, AI, rabies, BSE

The management problemConsequences animal healthEpidemiological consequencesVeterinary knowledge of diseases

The management problemConsequences animal welfareConsequences human healthConsequences animal healthEpidemiological consequencesKnowledge about externalities

The management problemConsequences animal welfareConsequences human healthCosts of interventionConsequences animal healthEpidemiological consequences

Decisons become increasingly complexDecision maker

ObjectivesAvailable resourcesConsequences animal welfareConsequences human healthCosts of interventionConsequences animal healthEpidemiological consequences

Levels of decision makingIndividual animalsTreatmentCullingInteractionGroups of animals (herd/farm)PreventionEradicationSectorControlEradicationRegionControlEradication

Levels of decision makingIndividual animalsTreatmentCullingInteractionGroups of animals (herd/farm)PreventionEradicationSectorControlEradicationRegionControlEradication

Production diseases& Endemic contagious diseasesType of disease

Contagious notifiable diseases

Levels of decision makingIndividual animalsTreatmentCullingInteractionGroups of animals (herd/farm)PreventionEradicationSectorControlEradicationRegionControlEradication

Farmer, supported by advisor

Farmers organisationProcessors

GovernmentDecision maker

OutlineThe Dutch dairy sectorEconomics of animal healthDisease control: optimizationModeling economics of animal healthExamples normative modellingEmpirical modellingModern management: Precision dairy farmingFinal remarks

MaximizationVeterinarians want to maximize animal healthIf there is a vaccine, use itIf there is a (better) treatment, use itIn case of doubt: treat (better safe than sorry)Medical doctors also want to maximize healthAnd minimize risk of infectionMicrobiologists want to maximize detectionIf there are more precise tests, use it

But . Is it optimal?There is more than only the health of animals:MoneyTimeAvailability of drugs or vaccinSo measures need to be balanced

Economic effects of diseasesExpenditures (additional resources) DrugsVeterinarianLabourExpenditures to control disease Losses (decrease in production)Decreased production levelDiscarded milkChanges in milk price (milk quality)Culling

Total costsExpenditures + lossesOften overlooked90 % of studies only look at lossesFarmers tend to look at expendituresWe need to optimize

Control vs failureControl expenditures ()Output losses ()

J. Agricultural Economics, 1996

Source: McInerney et al., Prev. Vet. Med, 1992

High losses, low control expendituresControl vs failureControl expenditures ()Output losses ()

Source: McInerney et al., Prev. Vet. Med, 1992

Low losses, high control expendituresControl vs failureControl expenditures ()Output losses ()

Source: McInerney et al., Prev. Vet. Med, 1992

OptimalControl vs failureControl expenditures ()Output losses ()

OutlineThe Dutch dairy sectorEconomics of animal healthDisease control: optimizationModeling economics of animal healthExamples normative modellingEmpirical modellingModern management: Precision dairy farmingFinal remarks

Basic approachNormative modellingRelate costs of interventionwith animal health andepidemiological consequencesCost-benefit analysis (alternative: cost effective or cost utility analysis)Assuming profit maximising behaviour of farmersBasis for on-farm decision support toolsEmpirical modellingUse data to compare farms/animals/groups of animals with and without interventionExperiments or existing datasets (accountancy data)

Normative models

Simulation modelInput data based on data, literature, expertiseRelatively cheapPragmatic approachBio-economic modelling: economics combined with detailed physiological basis

Models . do not capture the complexity of the real situation

Models. are sometimes nicer than reality (too good to be true)

Some terminologyStatic vs dynamicbehaviour over timeDeterministic vs stochasticdefinite predictions or averages (deterministic)output is probability distributions (stochastic)variability of the system uncertainty of knowledgeSpatialSpace effects play a roleOptimization vs simulationoptimum solution, given an objectiveoutcome given a pre-defined set of input

Economics is about money. Right?What aboutHuman disease (zonoses)WelfareThe environment.

Express these in money

Different approachesCost-minization analysisCost-effectiveness analysisCost-utility analysisCost-benefit analysis

Differ in: measurement of effect

utilitybenefit

OutlineThe Dutch dairy sectorEconomics of animal healthDisease control: optimizationModeling economics of animal healthExamples normative modellingEmpirical modellingModern management: Precision dairy farmingFinal remarks

First step in all modellingCosts of diseaseCosts of rabies control on Flores islandCosts of Foot and Mouth Disease in EthiopiaCosts of Avian Influenza in Central Java

Costs of control of rabies

SPIN project:Data collection to calculate costs of poultry diseases in Western Java4 types of poultry farmsOverall costsVariation between farms and farm types

Dikky Indrawan, Much. Gumilang & Eko Rudi Cahjadi

Dry cow therapyTreatment of cows that dry off (before calving)AntibioticsIndividual cow decisionTwo modes of action:Cure of existing (chronic) intramammary infectionsPrevention of new infections during dry periodOften herd decision (blanket dry cow therapy)

Debate on selective vs blanket dry cow therapy

Stochastic modelCow as basic unitDynamic around dry periodResults summarized for whole herdAccounting for differences between pathogensDutch circumstances

Selective dry cow treatment cheapest

BlanketSelectiveNoIMIdo (%)15 (7.7, 23.1)15 (7.7, 23.1)15 (7.7, 23.1)Treatment (%)10035 (23, 46)0IMI at calving7.5 (3.1, 12.3)12.3 (6.2, 20)19.3 (12.3, 27.7)Clinical mastitis (%)1.8 (0, 4.6)3.2 (0, 7.7)5.1 (1.5, 10.8)Treatment costs (/cow)10.1 (10.1, 10.1)3.5 (2.3, 4.7)0Production losses (/cow)1.3 (0.5, 2.2)2.1 (1.0, 3.4)3.3 (2.0, 4.7)Clinical mastitis (/cow)4.2 (0, 14.6)8.1 (0, 22.9)14.7 (2.0, 38.5)Total costs (/cow)15.6 (10.6, 26.6)13.7 (4.9, 29.4)18.0 (4.1, 42.6)

Little different results with extended model4 scenarios

Total net costs of scenarios:1 (BDCT): 8,8002 (BDCT + TS): 9,1783 (SDCT): 9.2434 (SDCT + TS): 9.435

New discussion onantibiotic resistanceResistance of mastitis pathogensSelf-interestNo increase seen (Hogan, IDF-factsheet)Antibiotic resistance in humansExternalityDairy cattle has very minor contribution (Oliver et al., 2011)Decision of governmentIn the Netherlands (self) regulationMaximum amount of antibiotics to be used (< 50 %)

Optimizing: linear programming (Maas, 2014, MSc thesis)Farm levelCows with high SCC are treatedPrimiparous > 150.000 cells/mlMultiparous > 250.000 cells/mlOther cows selectiveCategorized at SCC levelOptimization to minimize total costs of treatment and mastitis around dry period

Based on: Maas, 2014, MSc thesis

Were also interested in amount of AB

Constraining antibiotic use has economic effects

OutlineDecision making on animal healthThe decision problemThe levels of decision making Some examples of analysesDry cow therapySlaughterhouse measures to reduce the BSE loadBlue tongue controlVeterinary herd health and management programsFinal words

Bovine Somato EncyphaleomytisMad cow disease

BSE1986 first described1996 -> link with Creutzveldt Jacobs Disease (vCJD)Since August 1989 measures against BSE in the NetherlandsSince 1990 feed ban (no animal protein)Since 2000 dead cattle older than 30 m testedSince 2001 slaughtered cattle older than 30 m testedDisposal of BSE risk materialsCulling of cohort of detected animalIncidence of BSE is virtually 0

Are preventive measures cost-effective?Simulation modellingStaticStochasticSimulationMonte carlo model1 iteration = 1 yearBaseline: no interventionAlternative: one or more interventions

Model3 types of BSEClinically affectedTest detectableNon detectable (3 for every detectable)Per BSE type of BSE load (from different organs) of the food supply was calculatedBased on Infectious doses, risk of vCJDPrevented case of vCJD -> life years saved (most likely 51)Comparison: do nothing vs intervention

CostsRemoval of specific risk material (~60 kg): /kg slaughtered weightTransport of specific risk materialPost mortem testing: 90 per headCosts of cohort culling

Results - retrospectiveYear2002 2005Number of BSE cases (total, at slaughter)24, 123, 2BSE load of the food supplyMean5th 95thMean5th 95th.Baseline scenario34,85730,213-39,6025,5023,592-7,620SRM removal2,3302,020-2,648368240-509Post-mortem testing (PMT)7,4554,846-10,306939198-2,091PMT and cohort culling7,0594,505-9,865939197-2088SRM removal and PMT498324-6896313-140SRM removal and PMT and cohort culling472301-6596313-139Food risk (life years lost)Mean5th 95tbMean5th 95th pct.Baseline scenario16.988.66-26.702.691.25-4.61SRM removal1.140.58-1.790.180.08-0.31Post-mortem testing (PMT)3.631.67-6.270.460.08-1.11PMT and cohort culling3.441.56-5.940.460.08-1.11SRM removal and PMT0.240.11-0.420.030.005-0.07SRM removal and PMT and cohort culling0.230.10-0.400.030.005-0.07

Costs (mln )Year2002200320042005SRM removal19.2218.2719.2919.82Post-mortem testing38.1629.5626.5721.12Cohort culling6.974.803.412.43Total costs64.3452.6449.2743.37

Cost-effectiveness

Cost-effectiveness 2002-2005

OutlineDecision making on animal healthThe decision problemThe levels of decision making Some examples of analysesDry cow therapySlaughterhouse measures to reduce the BSE loadBlue tongue controlVeterinary herd health and management programsFinal words

Blue tongue diseaseViral diseaseGeneric disease effectsProduction losses

Several subtypes BTV 8 problem

Spread by midges

First step: decision analysis6721354

Vaccinate yes or noHerd exposure to BTV-8Disease effects

Export effectsIncome effectsConsideration 1: Reduces the impact of BTV-8 in the risk period of infection (1, 2, 3)

Consideration 2: Almost guarantees that heifer meant for export can be exported during the epidemic (4, 5)

Consideration 3: Helps to control the transmission (vaccination behaviour over time) (6, 7)

DE1HE1

V1EXR1NV1NV2HE2V2DE2Export restrictionDisease effectsHerd exposureVaccinate?Vaccinate?Herd exposureDisease effectsExport restrictionEXR2Probability of disease effects in year 1Economic consequences of DE1 calculated with model Velthuis (2010)

Vaccinate againt BTV8 or not

VaccinationNo vaccination

Income losses (Euro)Probability densityBTV8 outbreakYear 1: high net expected utility of vaccinationYear 2: risk attitude important

After theory of reasoned action and theory of planned behaviourFishbein and Azjen reuniteReasoned action approachAddition of actual controlInjunctive norms and descriptive normsAddition of background factors and feedback loops

I = A(b+e)+Ni(s+m)+Nd(s+m)+PBC(s+p)

Ni: perception of what the reference is thinking about what he/she should do

I = A(b+e)+Ni(s+m)+Nd(s+m)+PBC(s+p)

Nd: perceived behavior of others (farmers)

RRA and questionnaire

QUESTIONNAIREIf I vaccinate my herd preventively when Bluetongue would be in my vicinity next year, Not very likely> > > >Very likely12345get my cows to do with negative side effects and/or stress.Will the following motive be of importance when deciding to vaccinate your herd preventively, when Bluetongue would be in your vicinity next year?Of no importance< < > >Of importance-2-10+1+2Negative side effect and/or stress with my cows

Approach: Vaccination for BTV8A two-step modelling approach:Step 1: Estimate a measurement model (MM) using Confirmatory Factor Analysis (CFA)Step 2: Estimate a structural model (SM)

Final results

OutlineThe Dutch dairy sectorEconomics of animal healthDisease control: optimizationModeling economics of animal healthExamples normative modellingEmpirical modellingModern management: Precision dairy farmingFinal remarks

Empirical modellingUsing economic performance of farmsDifficult (lack of data)Link performance to disease situationEconometrics

Veterinary herd health managementWe want farmers to uptake VHHMBetter healthBetter welfareBetter Prevention vs cureFarmers have to pay

We did not do any fully normative modelling

Data collection 15,000 farms (207 veterinary practises)Questionnaire about VHHM695 replies (69 % VHHM)Milk production and fertility data availableAnalyses on MPR data in relation to VHHMANOVA/T-testLinear regression, herd sizeLinear mixed model

Effects VHHMParticipationMore milk (+336 kg/cow/year)Lower SCC (-8,340 cells/mL)Lower first calving age ALVA (-12 d) Lower % non return 56-d (3.34%)More inseminations per cow (+0.09). More culled cows (+1.05%)Lower age at culling (70 d).

Two economic studiesSemi-empiricalEmpiricalQuestion:If you compare the costs of VHHM with the economic value of improvements, is VHHM cost-effective?

Semi-empirical studyAll farms from previous studyLook at VHHM and estimate costs (normative)Farm seizeReproductive performanceIntensity of VHHMCalculate net returns milk productionCalculate costs of replacement

Used normative factorsVariableAbbreviationValueCall-out costs of veterinary visit (/visit) Cv30Costs of time of veterinarian (/hour)Ct120Time necessary for a pregnancy check (min.)Pt2Time necessary to discuss the first topic (min.)Tt110Time necessary for each additional topic (min.)Tt25Costs of replacement heifer ()Ch888

Table 2. Assumptions to calculate the costs of VHHM based on farm data.

ResultsVHHM*NVHHMMinMaxMeanMinMaxMeanNRmilk 145231602403117330662293Cvhhm Startup cost vet visit 1.0626.004.71---Cost of pregnancy check010.008.09---Cost of time for discussion0.5640.486.79---Total1.626719.62---NRvhhm142931382388117330662293Costs of replacement heifer5141522476464212NRtot119828872164101828512081

Empirical studyBookkeeping firm (Alfa Accountants)Questionnaire send to 572 farms187 replied (85 participants, 102 non part)All farm economic data availableStochastic frontier analysisLooking at efficiency of farms4 models: financial data per cow/kg milk and including/excluding farm structure variables

Farm efficiency (DEA)

VariableParticipantMean (sd)seP-valueTotal revenue/100 kg cmilkyes35.52 (3.28)0.360.16no34.71 (4.32)0.43Feed costs/100 kg cmilkyes7.44 (1.40)0.150.80no7.38 (1.75)0.17Health and med costs/100 kg cmilkyes1.31 (0.59)0.07