http://www.fao.org/ag/againfo/programmes/en/rinderpest/home.html
http://www-data.fao.org/ag/againfo/programmes/en/empres/event_200116.html
Presentation made during the international meeting titled “Maintaining global freedom from rinderpest” held in FAO-HQ from 20 to 22 January 2016.
Modelling to support rinderpest outbreaks preparedness
1. Name Benjamin McMahonName Benjamin McMahon
Title Scientist at Los Alamos National LaboratoryTitle Scientist at Los Alamos National Laboratory
Country USACountry USA
2. Modeling tool to support rinderpest
outbreak preparedness
Benjamin McMahon, Paul Fenimore, Judy
Mourant, Nick Hengartner, Carrie Manore, Mira
Dimitrijevic, Paul Rossiter, Samia Metwally
Los Alamos National Laboratory, FAO
3. Past rinderpest outbreaks
• Ethiopia, 1890, 90% cattle mortality
• Zimbabwe 1896, 90% mortality
• South Africa 1895-1896, 66.6% of 1.6 million cattle died or
slaughtered
• Nigeria and Chad basin 1982-84, 2 million deaths?
• Tanzania and Kenya, 1964-1968. Wildebeest population
increases from 250,000 to >1million after eradication of
rinderpest from cattle
• Pakistan, 1992, 40,000-50,000 cattle
4. Our motivation for modeling tool
- Quantify potential impact and motivate virus destruction & sequestration
- Significant stocks of rinderpest virus are maintained under laboratory
conditions for responding to contingencies (outbreaks) and research.
- These stocks create a potential risk for re-initiation of rinderpest infections.
- Historical outbreaks in cattle need to be considered carefully when
predicting the course of potential future outbreaks because of changes in:
- The level of pre-existing immunity
- The density and type (dairy, range, transported) of cattle
- The availability of surveillance and mitigations, such as:
- Vaccination
- Short range movement controls and hygiene
- Long range movement controls
- Culling
5. A rinderpest outbreak could be devastating:
Simulated spread of rinderpest in 101 days after point introduction to USA
Manore, McMahon, Fair, Hyman, Brown, & Labute, “Disease properties, geography, and mitigation
strategies in a simulation spread of rinderpest across the United States” Vet. Res., 42:1 (2011).
6. What modeling can and cannot do
• Modeling can:
– Translate historical events to contemporary situations, using best
available understanding of how diseases progress
– Provide specific numbers and their dependencies, to guide planning
• Modeling cannot:
– Predict the future. The actual course of the epidemic depends on
preparedness measures, responses, the particular viral strain, and
random events
7. Determinants of disease spread Frequency of long
and short range
cattle movement
I
S
S
S
S
S
FAO modeled
cattle density
5 km spatial resolution, from
http://www.fao.org/3/a-a1259e.pdf
8. Comparing types of mitigation
NumberofCountiesInfected
General impact of three types of mitigation
Time (days)
No Control
Movement Control
Vaccination
Culling
9. Disease progression rates determine required
timescale of intervention (a couple of weeks).
S = Susceptible
E = Exposed (incubating)
I = Infectious
H = Seriously diseased
D = Dead
R = Recovered (Immune)
VS= Vaccinated,
still susceptible
V = Vaccinated, immune
S E I H D
RVVS
Disease progression scheme
10. The planning tool http://bsvepi.lanl.gov/rinderpest
(password protected)
11. Phases of epidemic
1. Exponential growth (R0=5)
2. Short-range movement
controls and hygiene
measures
3. Plus vaccination
4. Long-range spread
5. Control
6. Eradication
Eradication
Control
12. Mitigations
Infectivity
(β * 5 days = R0 ~ 5)
Number of sick cattle when epidemic is identified
(25 cows)
Further delay and effectiveness of short-range movement restrictions & hygiene
(7 days, β cut in half)
Further delay and rate of vaccination
(4 weeks, 10,000 cattle / day)
Extent of long-range cattle trade occurring, and screening by illness
(0.15% cattle not seriously ill moved per day)
Total dead cattle = 885 cows
Vaccine doses given = 315,000 doses
Area affected by epidemic = 17,000 km2
Time to epidemic peak = 10 weeks
Time to eradication = 17 weeks
Consequence metrics
An example epidemic:
16. Further observations
• Long-range transport of animals has been added to model, with user-selected
destination and number of healthy animals moved. Infectious animals can be
selected from Incubating, Ill, and seriously ill fractions.
• Mixing lengths in model can account for differences between, for example,
dairy and range cattle.
• Virulent strains of rinderpest have a mortality rate of 80% and R0 ~ 5.
• Less virulent strains of rinderpest have lower mortality rate and R0.
• It is possible that naïve populations of cattle select for virulent strains as an
epidemic progresses.
• Studies of the 2001 FMD epidemic in Britain suggest short range movement
controls and hygiene measures can decrease R0 by a factor two.
• The calculations here were for cattle densities of ~150 cattle / km2
.
17. How to use this tool for planning
• Account for local variations in transmissibility, vaccination
rate, distance range of spread, long-range transmission,
escape from laboratory, movement controls.
• Convert estimates of dead and vaccinated cattle, duration
and geographic extent of epidemic and nature of control
measures into costs.
• Estimate required attributes of surveillance system to rapidly
identify epidemic, and the corresponding needed size of
vaccine stockpile.
• Balance cost of preparedness against acceptance of risk.
18. Conclusions & Acknowledgement
• We have provided a planning tool which provides
considerable flexibility in simulating a rinderpest outbreak
and mitigations
• It needs to be developed and validated in a country-specific
manner, in a collaboration between each country and FAO
• The earlier an outbreak is halted, the better.