Re National Audit Office -  The 2001 outbreak of foot and mouth disease
 
Nicola Morris wrote to Stewart Lingard about the National Audit Office Report on June 26th 2002. She pointed out with tact and detailed evidence that the NAO statement

'78% of infected premises confirmed on clinical grounds tested positive on laboratory test' is simply not true.

Her statistical evidence came from JCC at Defra itself - and her reminders about the accuracy of Pirbright testing were that, while clinical diagnosis of FMD is notoriously difficult, laboratory testing is straightforward and accurate - and she quoted Alex Donaldson's words in front of the EFRA committee in April 2001.
 
38 (Page 63)is wrong, I will quote 2 sources of figures because they are slightly different    Source 1 EFRA select committee report p 43 of minutes for 31/10/2001
 and source 2  figures I was given by the JCC data analysis department DEFRA
 

Test results Summary

                                               J Sucdamore 31/10/01                JCC data analysis             

No. Infected Premises                         2026                                        2026

No. laboratory positive                        1326                                        1324

No. laboratory negative                         402                                         401

No. untested                                         298                                          301

No. clinically negative, lab positive       175                                          171 

 

The clinically negative laboratory positive are those premises where - by the time DEFRA traced these premises - all clinical signs of disease had gone and the animals were confirmed on antibodies.   These farms require a separate analysis of their own because we need to identify whether or not the disease spread from these farms in the 3 week upto 6 month period in which it took to find them. In a nutshell there are over 171 farms in this epidemic where the disease remained undetected for up to 6 months (you have identified 1 in Settle ) where are the others? What impact did these premises have on the spread of the epidemic given that the key control intervention is slaughter within 24 hrs.

The statement ' 78% of infected premises confirmed on clinical grounds tested positive on laboratory test' is simply not true.   Using JCC stats which I believe are the most accurate

Clinical positive IPS   2026 - 171 =  1855

Laboratory positive    1324 - 171 =    1153  

so there were 1855 clinical positive cases of these 401 tested negative and 301 were not tested, that leaves 1153 laboratory positive.

Thus 62% of the clinically confirmed cases were laboratory positive. If note 1 still applies then this falls to 58%.

One of the problems in this epidemic was misdiagnosis. This problem is graphically illustrated if you look at the distribution of negative IPS across the country (data attached) . On a county basis the percentage of negative laboratory test results ranges from 7.5% to 60%.

The majority of these are not false negatives. I end with Dr Donaldson's views on the subject.

I have not done justice to your report and will read it thoroughly before I write to you again.   But, I would urge you to follow up the figures above, if I've got them wrong please tell me.

 
Stewart Lingard replied on August 6th.  Nicola Morris was alarmed at the way her serious concerns were being explained away.  She wrote again immediately.

Dear Mr Lingard

In reply to your Letter 6th August

I am sorry but I do not agree with your explanations:

(Extract from Stewart Lingard's letter)  'You question our statistics in Figure 38 and paragraph 3.75, in particular that 78% of infected premises confirmed on clinical grounds tested positive. The 78% figure relates only to those premises (cases) where laboratory samples were taken. It is consistent with the written evidence provided by the Chief Veterinary Officer on 31 October 2001 to the DEFRA Select Committee, and is based on improved data, available to the National Audit Office in April 2002.'

I am sorry but this is still not correct.  For it to be correct you have to state the full figures or alter the percentage  For this epidemic 78% of the clinically confirmed cases were not positive at laboratory test

Using the data given to EFRA committee :

In this epidemic of the 2026 infected premises 1851 premises were confirmed on clinical grounds, 175 were confirmed on laboratory grounds (not clinical grounds - for example the whole of the Brecon Beacons were antibody positive only, no clinical signs of disease- I am sorry you are not familiar with these figures- you should be )

Of those 1851 clinically confirmed cases 1156 were laboratory positive that is 62%.

If you say 78% of infected premises confirmed on clinical grounds tested positive then the reader will assume either 78% of 2026 premises tested positive i.e. 1580 premises

or that 78% of 1851 tested positive i.e. 1443

How can you square that with the actual figure of 1156 - clearly stated in EFRA memo? ( the figure 1554 given at the bottom is an obvious manipulation of the data which has occurred by deducting the untested clinically confirmed cases from the total number of clinically confirmed cases i.e 1851- 297= 1554 )

I have not assumed anything, because I have stated the facts. Which are 62% of the clinically confirmed cases were laboratory positive - I haven't said the others were negative they we either negative 398 cases or they weren't tested 298 cases .

(Extract from Stewart Lingard's letter)'  The description of cases as ‘clinically negative, lab positive’ is not one which we recognise. The supplementary memorandum to the DEFRA Select Committee report refers to them as ‘confirmed on laboratory results.’ This does not mean that clinical signs of disease were not found. '

May I suggest you check this. That is :clinically negative laboratory positive'

In August 2001 I received a letter from Marie Chapman stating some premises were confirmed on antibodies only, I also received data from ILU identifying various premises which were identified as being seropositive only, I have data from a court case relating to the Brecon Beacons stating that all infected hefts were identified on antibodies only. I now believe that there were infact 171 such premises - I think you will find that none of these premises had clinical signs of disease.

 (Extract from Stewart Lingard's letter)  'You say that most laboratory negative test results were not ‘false negatives’ (citing evidence from Dr Donaldson of the Institute for Animal Health to the DEFRA Select Committee). It is not possible to state with any accuracy how many laboratory negative results may have been ‘false negatives’. However, as we state in footnote 19 on page 63, it is possible that some were. I would interpret Dr Donaldson’s comments to the Select Committee to be referring to the accuracy of the testing process itself. By this, I mean that if an animal had the virus then the laboratory test would show it. However, as footnote 19 states, a premises may have been infected but this did not show up in laboratory results because a sample was not taken from the most appropriate animal (most likely in the cases of sheep). Damage to the sample in transit and a sample being taken from an animal at the very early incubation stage are other possible ways of a ‘false negative’ result might have been arrived at.'

I am sorry but I am shocked by your explanation it is not consistent with the scientific facts regarding diagnosis of FMD in sheep.

Pirbright state that accurate clinical diagnosis is very difficult. Accurate laboratory diagnosis on the other hand is relatively easy.

If you read Dr Donaldson's full explanation as given in his letter to Vet Record ( I will fax it to you) perhaps you will revise your opinion of the negative tests. I think we can safely say that at the most 40 of the negatives could be false negatives.

I am sorry but if a premises was diagnosed clinically, surely it is safe to assume that the animal with clinical symptoms was the one which was tested- ie the most appropriate one.

Viraemia ( ie blood mixed with virus) can be isolated pre clinical signs ( 17- 96 hours after exposure) - in this epidemic DEFRA on average found premises when lesions were 2 days old ( source History of the epidemic ILU). I think you will find that not one single premises was identified when animals were in the very early stages of incubation. Animals in the early stages of incubation do not have clinical signs of disease.

I am sorry if I sound irrated by your reply to my letter: there are many people like me who have watched DEFRA manipulate data and refuse to answer reasonable questions over the last 12 months; we had high hopes that NAO would not be fooled by many of DEFRA's explanations. On the face of it, it appears that you have been fooled.

The fact is that from the published data that I have we can say that the scale of unnecessary slaughter in this epidemic was unprecedented:

10509 farms were slaughtered, but on less than 13% of these farms was laboratory evidence of foot and mouth disease found. Laboratory testing is accepted to be atleast 90% accurate (IAH Pirbright).

in this epidemic of the truly infected farms visible signs of disease were found 4-7 days after exposure to the virus (Jim Sucdamore EFRA select committee 21/03/01).

8226 premises thought to have been exposed to virus were culled as a precautionary measure, many of these farms were not laboratory tested, but delays in slaughter (due to lack of resources) were such that up to 95% of these farms were slaughtered 7 or more days after possible exposure to FMD virus.

We can therefore say with some certainty that by the time slaughter occurred on nearly 8000 farms if animals on these farms had been exposed to foot and mouth virus visible signs of disease would have been apparent. We can conclude that in this epidemic up to 5 million adult animals were slaughtered unnecessarily.

I look forward to your reply to the issues raised again in this letter and for an opportunity to present to you the evidence to substantiate the facts presented in the paragraph above.

Yours sincerely

Nicola Morris

Copy to: Peter Luff

 


Letter from

Stewart Lingard, Audit Manager

6 August 2002

FOOT AND MOUTH DISEASE: LABORATORY TEST RESULTS

Thank you for your e-mail of 26th June to Mrs Thomas concerning our report on foot and mouth disease. I worked on aspects of the examination you refer to and have been asked to reply.

You question our statistics in Figure 38 and paragraph 3.75, in particular that 78% of infected premises confirmed on clinical grounds tested positive. The 78% figure relates only to those premises (cases) where laboratory samples were taken. It is consistent with the written evidence provided by the Chief Veterinary Officer on 31 October 2001 to the DEFRA Select Committee, and is based on improved data, available to the National Audit Office in April 2002. It includes adjustments (explained in Note 1 of Figure 38). We think that you have calculated a lower figure through basing the percentage calculation on all cases (premises). This means you have assumed that those premises (cases) from which samples were not taken would have produced negative test results for FMD virus.

You say that there were 171 infected premises where the Department arrived at so late that clinical signs of disease were not present, but laboratory tests subsequently confirmed disease. The description of cases as ‘clinically negative, lab positive’ is not one which we recognise. The supplementary memorandum to the DEFRA Select Committee report refers to them as ‘confirmed on laboratory results.’ This does not mean that clinical signs of disease were not found. On the contrary many were cases from the early weeks of the epidemic, where although a vet had found clinical suspicions of the disease, confirmation as an infected premises sometimes awaited receipt of a laboratory result. Others were cases which originated as slaughter on suspicion or dangerous contact cases in which animals were slaughtered but samples taken. When these results were received they were reclassified as infected premises.

You refer to the problems of misdiagnosis. Our report also makes clear, in paragraphs 3.75-3.76, that there were difficulties in diagnosis, particularly in the cases of sheep. Figures for results by Disease Control Centres are set out in paragraph 3.76 and Appendix 7. One of the factors that persuaded the Department to adopt the contiguous cull, was the difficulty of diagnosis with sheep.

You say that most laboratory negative test results were not ‘false negatives’ (citing evidence from Dr Donaldson of the Institute for Animal Health to the DEFRA Select Committee). It is not possible to state with any accuracy how many laboratory negative results may have been ‘false negatives’. However, as we state in footnote 19 on page 63, it is possible that some were. I would interpret Dr Donaldson’s comments to the Select Committee to be referring to the accuracy of the testing process itself. By this, I mean that if an animal had the virus then the laboratory test would show it. However, as footnote 19 states, a premises may have been infected but this did not show up in laboratory results because a sample was not taken from the most appropriate animal (most likely in the cases of sheep). Damage to the sample in transit and a sample being taken from an animal at the very early incubation stage are other possible ways of a ‘false negative’ result might have been arrived at.

I hope this response is helpful but if I can assist further please contact me.

 

Stewart Lingard, Audit Manager