Back

There is no need for a computer to predict that "too little too late" does not work!......since policy makers tend to accept modelling results at face value, particularly if the results fit their policy, modelers have a serious socioeconomic responsibility to make sure that their models and the results are transparent and well qualified.

Dr Paul Sutmoller, writing here about the modelling work much hailed by the NFU last week. (This email was first posted on the proMed site. Dr Sutmoller has given his blessing for it to be reproduced here...and wishes us "Success!")

The [Morris] model tested 2 different FMD vaccination strategies: 1. FMD vaccination of cattle in a series of rather narrow, mostly East-West "bands" across the country, positioned to create a barrier between FMD-affected and FMD-unaffected areas, with 8000 head being vaccinated in each specified area per day.
2. FMD vaccination of high density zones in combination with preemptive slaughter of infected farms in 3 areas of more than 50 infected farms per 100 km2. Within each area 8000 cattle were vaccinated per day.

Any veterinarian with some practical experience in the planning and execution of FMD vaccinations would have predicted that such strategies would not work. Vaccines will work if applied on a sufficiently large scale in geographically well defined areas that adequately cover the spread of the disease. Few if any outbreaks will occur, once a sufficient level of population immunity is obtained, usually within a week after completion of vaccination in the area. Of course, if the disease also affects other species such as sheep, logic dictates that these species should be included in the vaccination campaign. There is no need for a computer to predict that "too little too late" does not work!

One can wonder what the results of the model would be with inputs like "FMD is now endemic in the sheep population" or "FMD spilled over into the deer population". Would the application of vaccine then perhaps show an advantage over contiguous cull? Are the modelers conscious that the modern vaccines protect the vaccinated population close to 100 percent? A model will give conventional answers when input parameters reflect conventional dogmas such as "vaccinated animals become carriers" and "carriers spread the disease". What happens if these dogmas are shown to be false?

The results of the model are based on depopulation of holdings within 24 hours of FMD confirmation and depopulation within 72 hours of farms within the preemptive slaughter list. Is this not a very different situation from the present epidemic, which seems to be smouldering among small ruminants (and possibly deer?) and where the main diagnosis is based on a rather slow program of bleeding and serological testing?

Computer simulation is entertaining and a good learning experience. But the old dictum still is "Garbage in, garbage out". However, since policy makers tend to accept modelling results at face value, particularly if the results fit their policy, modelers have a serious socioeconomic responsibility to make sure that their models and the results are transparent and well qualified.

--- Dr Paul Sutmoller, PhD DVM, Animal Health Consultant Presently: c/o Duinschooten 12/bungalow 127, 2211ZC Noordwijkerhout, The Netherlands 1502 Largo Road, Richmond, VA 23233, USA

[Paul Sutmoller spent a number of years at Plum Island before he moved to the PAHO Pan American Foot & Mouth Centre outside Rio de Janeiro where he was the senior virologist for many years and a key member of their team in developing their successful oil-adjuvant FMD vaccine. I first met him in 1975 at PANAFTOSA when I was on detached duty from MAFF and the centre was running extensive field trials of the vaccine. - Mod.MHJ]

Back