Animal health economics and precision farming

Economics of animal health and precision farming

Henk Hogeveen

Who am I Born on a dairy farm (1966) Animal science at Wageningen University

●Epidemiology (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 management

Business economics group, Wageningen UniversityFaculty of Veterinary Medicine, Utrecht University

@henkhogeveen slideshare.net

animal-health-management.blogspot.com

Outline The Dutch dairy sector Economics of animal health Disease control: optimization Modeling economics of animal health Examples normative modelling Empirical modelling Modern management: Precision dairy farming Final remarks

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

1999 2002 2005 2008 2011 20143650000

3700000

3750000

3800000

3850000

3900000

3950000

4000000

4050000

4100000

Grass-based system

Summer●Fresh grass + corn silage + concentrates

Winter●Grass silage + corn silage + concentrates●Half mixed ratio

http://www.google.com/url?sa=i&rct=j&q=&esrc=s&frm=1&source=images&cd=&cad=rja&uact=8&docid=lf0smABMICCYCM&tbnid=2jRB1RzyQcjH2M:&ved=0CAUQjRw&url=http://www.holaras.nl/nl/product/product/6/voermachines-blokkendoseerwagen-rabbit.html&ei=5QWLU9zNHOb60gWj-ICIAw&bvm=bv.67720277,d.ZGU&psig=AFQjCNHbLaH3GWKmoKXEfs3CMHsCrBvT9g&ust=1401706319039993

http://www.google.co.za/url?sa=i&rct=j&q=&esrc=s&frm=1&source=images&cd=&cad=rja&uact=8&docid=p7uHcdYFwko1hM&tbnid=BOaLrKkN3qf54M:&ved=0CAUQjRw&url=http://www.archief.veeteelt.nl/nieuws/voeding/2012/voerrobot-van-lely-op-film&ei=_waLU4ijNILS0QWVhYCQCw&bvm=bv.67720277,d.ZGU&psig=AFQjCNE-xyzYDNHE0Tep8TTuC7vD3paSPQ&ust=1401706621450863

http://www.google.com/url?sa=i&rct=j&q=&esrc=s&frm=1&source=images&cd=&cad=rja&uact=8&docid=RGjFDtR5p9rzxM&tbnid=582A69gm0eVv5M:&ved=0CAUQjRw&url=http://www.youtube.com/watch?v%3DY300ns2ye_o&ei=NgWLU5O2OoKd0QWF-IHIBQ&bvm=bv.67720277,d.ZGU&psig=AFQjCNE7nlB6tDWDtURsG8v2i16y2yAerw&ust=1401706155918861

Dutch dairy sector

Increasing farm seizes Mostly own (family) labour Half grass-based system

●Grazing under pressure (farm management)●Stimulated (societal preference)

Volatile milk prices More challenges on management

http://www.google.com/url?sa=i&rct=j&q=&esrc=s&frm=1&source=images&cd=&cad=rja&uact=8&docid=ncIredPz0Jq00M&tbnid=cZ6IrlLrC7ehbM:&ved=0CAUQjRw&url=http://boerderijkrokus.nl/archives/1099&ei=7gSLU-KyDMrH0QXn04CgBw&bvm=bv.67720277,d.ZGU&psig=AFQjCNFrixq2ZLwc0-PwMvLtfqp4466TAA&ust=1401706073982985

and …..

Animal health is becoming more and more important Stimulated by dairy industry

●Improvement of udder health●Improvement of claw health●Improvement of longevity


Economic effects of animal disease

Output

MilkMeatEggsDraft power…….

After: McInerney, 1996

Human benefit (utility)

Input

LandLabourCapital

The field: Economic effects of animal disease

Output

MilkMeatEggsDraft power…….

Disease

1. Lower efficiency

2. Lower suitability for consumption

3. Lower human well-being

Human benefit (utility)

Input

LandLabourCapital

1.

2. 3.

Most economic work

J. Agricultural Economics, 1996

Types of animal diseases Production diseases

●On-farm optimization●Externalities ●E.g., mastitis, lameness, APP

Endemic contagious diseases●On-farm control decision● Interaction between farms●E.g., BVD, Aujeszky’s disease

Notifiable contagious diseases●Regional control decisions (eradication)●Surveillance●E.g., FMD, AI, rabies, BSE

The management problem

Consequences animal health

Epidemiological consequences

Veterinary knowledge of diseases


Consequences animal welfare

Consequences human health



Knowledge about externalities




Costs of intervention



Decisons become increasingly complex

Decision makerObjectives

Available resources



Costs of intervention



Levels of decision making Individual animals

● Treatment● Culling● Interaction

Groups of animals (herd/farm)● Prevention● Eradication

Sector● Control● Eradication

Region● Control● Eradication






Production diseases& Endemic contagious diseases

Type of disease

Contagious notifiable diseases






Farmer, supported by advisor

Farmer’s organisationProcessors

Government

Decision maker


Maximization

Veterinarians want to maximize animal health

●If there is a vaccine, use it●If there is a (better) treatment, use it●In case of doubt: treat (better safe than sorry)

Medical doctors also want to maximize health

●And minimize risk of infectionMicrobiologists want to maximize detection

●If there are more precise tests, use it

But ……. Is it optimal?

There is more than only the health of animals:●Money●Time●Availability of drugs or vaccin

So measures need to be balanced

Economic effects of diseases

Expenditures (additional resources) ●Drugs●Veterinarian●Labour●Expenditures to control disease

Losses (decrease in production)●Decreased production level●Discarded milk●Changes in milk price (milk quality)●Culling

Total costs

Expenditures + losses Often overlooked 90 % of studies only look at losses Farmers tend to look at expenditures We need to optimize

Control vs failure

Control expenditures (€)

Output losses (€)

J. Agricultural Economics, 1996

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

High losses, low control expenditures

Control vs failure


Output losses (€)


Low losses, high control expenditures

Control vs failure


Output losses (€)


Optimal

Control vs failure


Output losses (€)


Basic approach

Normative modelling●Relate costs of intervention

with animal health andepidemiological consequences

●Cost-benefit analysis (alternative: cost effective or cost utility analysis)

●Assuming profit maximising behaviour of farmers●Basis for on-farm decision support tools

Empirical modelling●Use data to compare farms/animals/groups of

animals with and without intervention●Experiments or existing datasets (accountancy data)

Normative models

Simulation model Input data based on data, literature, expertise Relatively cheap Pragmatic approach Bio-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 dynamic

●behaviour over timeDeterministic vs stochastic

●definite predictions or averages (deterministic)●output is probability distributions (stochastic)●variability of the system uncertainty of knowledge

Spatial●Space effects play a role

Optimization vs simulation●optimum solution, given an objective●outcome given a pre-defined set of input

Economics is about money. Right?

What about●Human disease (zoönoses)●Welfare●The environment●…….

Express these in money ……

Different approaches

Cost-minization analysis Cost-effectiveness analysis Cost-utility analysis Cost-benefit analysis

Differ in: measurement of effect

utility benefit


First step in all modelling

Costs of disease●Costs of rabies control on Flores island●Costs of Foot and Mouth Disease in Ethiopia●Costs of Avian Influenza in Central Java

Costs of control of rabies

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 20110

100

200

300

400

500

600

700

800 PET human

Trace back investigationDiagnostic costs in animals

Dog-bite investigationDog cullingDog vaccination

Year

Cost

s of

con

trol

mea

sure

s (x

1000

$U

S)

SPIN project:

Data collection to calculate costs of poultry diseases in Western Java

4 types of poultry farms Overall costs Variation between farms and farm types

Dikky Indrawan, Much. Gumilang & Eko Rudi Cahjadi

Dry cow therapy

Treatment of cows that dry off (before calving) Antibiotics Individual cow decision Two modes of action:

●Cure of existing (chronic) intramammary infections●Prevention of new infections during dry period

Often herd decision (blanket dry cow therapy)

Debate on selective vs blanket dry cow therapy

Stochastic model

Cow as basic unit Dynamic around dry period Results summarized for whole herd Accounting for differences between pathogens Dutch circumstances

Selective dry cow treatment cheapest

Blanket Selective NoIMIdo (%) 15 (7.7, 23.1) 15 (7.7, 23.1) 15 (7.7, 23.1)

Treatment (%) 100 35 (23, 46) 0IMI at calving 7.5 (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) 0

Production 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 resistance

Resistance of mastitis pathogens●Self-interest●No increase seen (Hogan, IDF-factsheet)

Antibiotic resistance in humans●Externality●Dairy cattle has very minor contribution (Oliver et al., 2011)

Decision of government In the Netherlands (self) regulation

●Maximum amount of antibiotics to be used (< 50 %)

Optimizing: linear programming (Maas, 2014, MSc thesis)

Farm level Cows with high SCC are treated

●Primiparous > 150.000 cells/ml●Multiparous > 250.000 cells/ml

Other cows selective Categorized at SCC level Optimization to minimize total costs of treatment and

mastitis around dry periodBased on: Maas, 2014, MSc thesis

We’re also interested in amount of AB

Constraining antibiotic use has economic effects

100%

95%90%85%80%75%70%65%60%55%50%45%40%35%30%25%20%15%10% €39

€41

€43

€45

€47

€49

€51

€53

Average farmLow BTSCC farmHigh BTSCC farm

Percentage allowed antibiotics (%)

Cost

s pe

r lo

w S

CC c

ow

Outline

Decision making on animal health●The decision problem●The levels of decision making

Some examples of analyses●Dry cow therapy●Slaughterhouse measures to reduce the BSE

load●Blue tongue control●Veterinary herd health and management programs

Final words

Bovine Somato Encyphaleomytis

Mad cow disease

BSE

1986 first described 1996 -> link with Creutzveldt Jacobs Disease (vCJD) Since August 1989 measures against BSE in the

Netherlands●Since 1990 feed ban (no animal protein)●Since 2000 dead cattle older than 30 m tested●Since 2001 slaughtered cattle older than 30 m

tested●Disposal of BSE risk materials●Culling of cohort of detected animal

Incidence of BSE is virtually 0

Are preventive measures cost-effective?

Simulation modelling●Static●Stochastic●Simulation

Monte carlo model●1 iteration = 1 year●Baseline: no intervention●Alternative: one or more interventions

Model

3 types of BSE●Clinically affected●Test detectable●Non detectable (3 for every detectable)

Per BSE type of BSE load (from different organs) of the food supply was calculated

Based on Infectious doses, risk of vCJD Prevented case of vCJD -> life years saved (most likely

51) Comparison: do nothing vs intervention

Costs

Removal of specific risk material (~60 kg): €/kg slaughtered weight

Transport of specific risk material Post mortem testing: € 90 per head Costs of cohort culling

http://www.google.nl/url?sa=i&rct=j&q=&esrc=s&frm=1&source=images&cd=&cad=rja&uact=8&docid=TbV2UaSOEumMgM&tbnid=vz1aUVBPXJDosM:&ved=0CAUQjRw&url=http://en.wikipedia.org/wiki/Specified_risk_material&ei=51cIVIjnNoj_OZCngIAG&bvm=bv.74649129,d.ZWU&psig=AFQjCNFj64Roaoh3Ekm7plCqtIjVpJR3Eg&ust=1409919327147872

Results - retrospective

Year 2002 2005

Number of BSE cases (total, at slaughter) 24, 12 3, 2

BSE load of the food supply Mean 5th – 95th Mean 5th – 95th.

Baseline scenario 34,857 30,213-39,602 5,502 3,592-7,620

SRM removal 2,330 2,020-2,648 368 240-509

Post-mortem testing (PMT) 7,455 4,846-10,306 939 198-2,091

PMT and cohort culling 7,059 4,505-9,865 939 197-2088

SRM removal and PMT 498 324-689 63 13-140

SRM removal and PMT and cohort culling 472 301-659 63 13-139

Food risk (life years lost) Mean 5th – 95tb Mean 5th – 95th pct.

Baseline scenario 16.98 8.66-26.70 2.69 1.25-4.61

SRM removal 1.14 0.58-1.79 0.18 0.08-0.31

Post-mortem testing (PMT) 3.63 1.67-6.27 0.46 0.08-1.11

PMT and cohort culling 3.44 1.56-5.94 0.46 0.08-1.11

SRM removal and PMT 0.24 0.11-0.42 0.03 0.005-0.07

SRM removal and PMT and cohort culling 0.23 0.10-0.40 0.03 0.005-0.07

Costs (mln €)

Year2002 2003 2004 2005

SRM removal19.22 18.27 19.29 19.82

Post-mortem testing38.16 29.56 26.57 21.12

Cohort culling6.97 4.80 3.41 2.43

Total costs64.34 52.64 49.27 43.37

Cost-effectiveness

Cost-effectiveness 2002-2005

Outline

Decision making on animal health●The decision problem●The levels of decision making

Some examples of analyses●Dry cow therapy●Slaughterhouse measures to reduce the BSE load●Blue tongue control●Veterinary herd health and management programs

Final words

Blue tongue disease

Viral disease Generic disease effects Production losses

Several subtypes BTV 8 problem

Spread by midges

First step: decision analysis

6 7

2

1

3 5

4

Vaccinate yes or no

Herd

exposure to BTV-8

Disease effects

Export effects

Income effects

Consideration 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)

DE 1

HE 1

V 1

EXR 1

NV1

NV2

HE 2

V 2

DE 2

Export restriction

Disease effects

Herd exposureVaccinate?Vaccinate?

Herd exposure

Disease effects

Export restriction

EXR 2

Probability of disease effects in year 1

Economic consequences of DE1 calculated with model Velthuis (2010)

Vaccinate againt BTV8 or not

Vaccination

No vaccination

Income losses (Euro)

Prob

abili

ty d

ensi

ty

BTV8 outbreakYear 1: high net expected utility of vaccinationYear 2: risk attitude important

After theory of reasoned action and theory of planned behaviour

Fishbein and Azjen reunite Reasoned action approach Addition of actual control Injunctive norms and descriptive norms Addition 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

Constructs Items Preceding phrase Item description

If Bluetongue would turn up in the environment this year, is preventive vaccination of my herd …

unsatisfying – satisfying1

disadvantageous – advantageous2

necessary – unnecessary2

unimportant – important2

acceptable – unacceptable1

If Bluetongue would turn up in the environment this year …

then people who have something to do with my farm expect me to vaccinate my herd preventively.

then people in the industry whose opinions I value would approve of me vaccinating my herd preventively.

then people who are important to me think that I should vaccinate my herd preventively.are farmers like me going to vaccinate their herd preventively.

If Bluetongue would turn up in the environment this year, and a voluntary vaccination programme would be announced ...

do I have the possibility to vaccine my herd preventively.3

could I vaccinate my herd preventively, if I wanted to.3

is it up to me whether I vaccinate my herd preventively.4

If Bluetongue would turn up in the environment this year, and a voluntary vaccination programme would be announced …

am I going to vaccinate my herd preventively.

do I want to vaccinate my herd preventively.

am I willing to vaccinate my herd preventively.1 = experiential dimension, 2 = instrumental dimension, 3 = capacity dimension and 4 = autonomy dimension.

QUESTIONNAIRE

If I vaccinate my herd preventively when Bluetongue would be in my vicinity next year,

Not very likely > > > > Very likely

1 2 3 4 5

get 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 -1 0 +1 +2

Negative side effect and/or stress with my cows ⃝ ⃝ ⃝ ⃝ ⃝

Approach: Vaccination for BTV8

• A 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

1.00 0.62* 1.00 0.33* 0.50* 1.00 0.36* 0.25* 0.06 1.00 0.78*

0.62* 0.38* 0.36* 1.00

* = < 0.001

0.61

( < 0.001)0.18

( < 0.002)0.09

( < 0.05)0.08

( < 0.10) 35%


Empirical modelling

Using economic performance of farms Difficult (lack of data) Link performance to disease situation Econometrics

Veterinary herd health management

We want farmers to uptake VHHM●Better health●Better welfare●Better

Prevention vs cure Farmers have to pay

We did not do any fully normative modelling

Data collection 1

5,000 farms (207 veterinary practises) Questionnaire about VHHM 695 replies (69 % VHHM) Milk production and fertility data available Analyses on MPR data in relation to VHHM

●ANOVA/T-test●Linear regression, herd size●Linear mixed model

Effects VHHM

Participation●More 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 studies

Semi-empirical Empirical Question:

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

Semi-empirical study

All farms from previous study Look at VHHM and estimate costs (normative)

●Farm seize●Reproductive performance●Intensity of VHHM

Calculate net returns milk production Calculate costs of replacement

Used normative factors

Variable Abbreviation Value

Call-out costs of veterinary visit (€/visit) Cv 30

Costs of time of veterinarian (€/hour) Ct 120

Time necessary for a pregnancy check (min.) Pt 2

Time necessary to discuss the first topic (min.) Tt1 10

Time necessary for each additional topic (min.) Tt2 5

Costs of replacement heifer (€) Ch 888

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

Results

VHHM* NVHHM

Min Max Mean Min Max Mean

NRmilk 1452 3160 2403 1173 3066 2293

Cvhhm

Startup cost vet visit 1.06 26.00 4.71 - - -

Cost of pregnancy check 0 10.00 8.09 - - -

Cost of time for discussion 0.56 40.48 6.79 - - -

Total 1.62 67 19.62 - - -

NRvhhm 1429 3138 2388 1173 3066 2293

Costs of replacement heifer 51 415 224 76 464 212

NRtot 1198 2887 2164 1018 2851 2081

Empirical study

Bookkeeping firm (Alfa Accountants) Questionnaire send to 572 farms 187 replied (85 participants, 102 non part) All farm economic data available Stochastic frontier analysis

●Looking at efficiency of farms●4 models: financial data per cow/kg milk and

including/excluding farm structure variables

Farm efficiency (DEA)

Variable Participant Mean (sd) se P-value

Total revenue/100 kg cmilk yes 35.52 (3.28) 0.36 0.16

no 34.71 (4.32) 0.43

Feed costs/100 kg cmilk yes 7.44 (1.40) 0.15 0.80

no 7.38 (1.75) 0.17Health and med costs/100 kg cmilk yes 1.31 (0.59) 0.07 <0.01

no 1.01 (0.43) 0.04Cattle costs/100 kg cmilk yes 2.84 (0.71) 0.08 <0.01

no 2.35 (0.63) 0.06Land costs/100 kg cmilk yes 1.58 (0.41) 0.04 0.26

no 1.50 (0.53) 0.05Non-operational costs/100 kg cmilk yes 39.58 (7.67) 0.87 0.99

no 39.57 (7.05) 0.72

Total revenue/cow yes 3185.39 (528.69) 57.34 <0.01

no 2949.44 (501.22) 49.63

Feed costs/cow yes 671.96 (174.43) 18.92 0.11

no 631.18 (173.88) 17.22

Health and med costs/cow yes 111.05 (48.09) 5.38 <0.01

no 11.98 (36.23) 3.72

Cattle costs/cow yes 256.56 (80.43) 8.72 <0.01

no 200.36 (62.51) 6.19

Land costs/cow yes 142.17 (42.84) 4.65 0.02

no 126.90 (46.48) 4.60

Non-operational costs /cow yes 3517.20 (830.11) 94.60 0.07

no 3314.43 (556.35) 57.08

No differences between groups

Problem with empirical studies

Ceteris paribus assumption Co-variance

Farmer

Disease managemen

t

Othermanagemen

t

Disease

Other diseases

Productivity

Financial results


Precision dairy farming

1970’s: Development of individual cow ID 1980’s: Sensors for disease detection 1990’s: Automatic milking 2000’s: Revival of sensors 2010’s: New generation of sensors

http://www.google.nl/url?sa=i&rct=j&q=cow+identification+nedap&source=images&cd=&cad=rja&docid=6g7lFGlyEOJ3iM&tbnid=sV8jrpK-VpRX2M:&ved=0CAUQjRw&url=http://www.nedap-lactivator.com/specifications/the-neck-version/&ei=sinIUY-3B4a40QXhjICIAQ&bvm=bv.48293060,d.d2k&psig=AFQjCNFuiJ-0zlu7PDmFiCiMTi-aTC6Yow&ust=1372158749520492

Past developments

0102030405060708090

100

0 10 20 30 40 50 60 70

year since introduction

Where are we now

700

102030405060708090

100

0 10 20 30 40 50

year since introduction

Milking throughout time

Milking with milking machines

Modern large milking machine

Automatic milking is not an issue anymore

They function Service and maintenance (infrastructure) is good Reasons pro

●Labour savings, quality of labour●More milkings per cow

Reasons against●Price (investment)●Flexibility of expansion

http://www.google.nl/url?sa=i&rct=j&q=robot&source=images&cd=&cad=rja&docid=v0QNhasfM90rAM&tbnid=9hzx_RMV8Rbk4M:&ved=0CAUQjRw&url=http://www.athomaz.com/?m=201006&ei=PxXIUbu4Buq30QXY64DIAQ&bvm=bv.48293060,d.d2k&psig=AFQjCNHNOFvVvnSRhqyw470_LeI7-1Nueg&ust=1372153525269617





Economic effects (€/100 kg milk) No robot robot Before After

Capital costs 10.38 9.72 13.97

Labour costs

12.38 11.69 11.30

Variable costs

19.45 18.66 19.80

Revenues

46.28 43.93 46.38

Profit 4.07 3.86 1.31

Current demands to dairy industry

Animal well-being Consumer demands Environment Labor Economics

We have to reduce the use of scarce resourcesSo: explore the full potential of each individual

dairy cow

Is individual cow management possible?

Easy

(too) difficult

Don’t even think about it

Sensor system

SensorData

AlertAlgorithm

SOPDecision support

Decision

Otherdata

Review of sensor systems until now

Success factors

System specifications Cost efficiency Non-economic factors


Take home message

Animal health management decisions are taken daily Economics are useful/necessary to support decisions A first step are “cost of disease” studies

●General interest●Supporting stakeholders (negotiations)

Start for “economics of intervention” studiesCost-effectivity, cost-utility and cost-benefit

Empirical studies useful, but difficult

Remember, there is more than money to motivate farmers

Thank you for your attention