Attempts have been made to try to produce better models that would fit the oddities but it can always be said that models fitting does not infer that the models are correct and in fact further better models may appear later as information appears.
Please see the announcement by the Spongiform Encephalopathy Advisory Committee (SEAC) as a result of data given to them by Dealler, Wilesmith, and Anderson on the 19.7.96. It came into press on around 30.7.96 but information was given to the EC before the UK parliament.
It should be noted that a major editorial on vertical transmission is going to be published in the Internet Journal of Transmissible Spongiform Encephalopathy soon
Initially it was assumed that the disease was derived from scrapie in sheep and that this was presented to cattle as part of their diet. The spread of the disease was considered to be due to the cattle remains being fed to further cattle and the number of cattle that became infected in this way from one infected cattle being slaughtered could be estimated.
Specific problems arose in this epidemiological model and other models were chosen to try to explain this:
The first was that the infection was present in 'nuggets' that were present in the food. This model was required to explain why the age distribution of cattle infected in different years should be similar. If infection was not in 'nuggets' then the age distribution of cattle dying of disease would have been expected to grow younger. The nugget hypothesis would suggest that an infected animal passed infection to a limited number of bags of cattle feed and that throughout the epidemic there was not so many cattle slaughtered while infective as to cause the number of infective cattle per bag to be multiplied. This model also required that infective nuggets were relatively uncommon and that a bovine would not be eating multiple numbers, this also suggests that infectivity was passed almost entirely when the animal was young.
The next problem was that a 'nugget' model was simply not what one would expect from looking at the method of manufacture of the meat and bone meal. Also, the number of infected cattle involved in specific batches of feed might actually have multiplied during the epidemic according to the mathematics. To get around this it a different model was considered in that the infectivity was not so much in nuggets of the same size but rather of infectivity spread throughout a batch (e.g a bag of meal). Within that batch there would be some parts much more infective than others and as such the distribution of infection would be such as to produce the same effect as a 'nugget' and as such there would be a wide distribution of incubation periods but that this distribution would not be expected to change as the epidemic progressed. This would also depend on the number of infected cattle per batch not multiplying or at least the effect not being great enough to alter the age distribution.
Specific problems have arisen with these models and the possibility of both horizontal and vertical transmission must now be taken into account.
Here I put forward the epidemiological model of vertical transmission and the possibility that:
The calculated number of infected cattle becoming infected in specific years and dying by the age of 10. It was expected that the number born in these years would drop dramatically (line W) but in fact they did not (line C). When calculated, the number expected if the disease was vertically transmitted was really very similar (lines A or B) and these were calculated as the number of offspring of the proportion of adult dairy cattle at the time of the feed ban compared with the following years. As these cattle were slaughtered at the end of their milking lives, fewer and fewer offspring would be of cattle born before the feed ban. Assumptions used were those used in the statistical analysis of Dealler and Kent (i.e. that all cattle with symptoms prior to 1992 were reported, and that the age distribution of disease remained steady).
Various explanations have been given for the slow drop in cases. e.g. that pig meal still containing bovine material entered bovine food (with a nugget model this would have to be such a high proportion that it is impractical) e.g. that farmers retained feed from prior to the feed ban on the farm. Farmers have a monthly turnover of feed and in order to keep the level of infectivity at this rate this hypothesis would require that such an enormous percentage of meal was retained that the hypothesis could not be practical. e.g. that the meal manufacturers were not carrying out the regulations and that infectivity continued in the feed. My own experience is that the renderers were quite determined to keep infectivity out of the feed in order to keep their industry going. e.g. that other parts of the cow, not thought of as infective were entering the bovine feed (tallow?). Again the level of infectivity in the tallow would have had to be very high indeed to produce the effect. A 'nugget' model and these are all difficult to fit in with the number of BABs.
B.Cattle develop BSE in Portugal as offspring of UK dams.
The number of cases of BSE in Portugal has in fact gone up to around 35 but, unlike Germany, some of the cases seen are the offspring of the cattle exported. Further investigation into this by UK journalists came out with the indication that at least 12 were in this category, and that one was the offspring of a Dutch cow and one was the offspring of the offspring of an imported dam. One BSE case has been reported in the offspring of a dam imported to an 'organic farm' in the UK. Clearly more information is required.
C. The distribution of cases of BSE in the national herd.
What was expected with the 'nugget' hypothesis was that one infected batch of feed might go to one herd but the next infected batch was just as likely to go to that herd again as to any other another herd. i.e. the cases of BSE would appear in the herds like the winners of the lottery, and cases would appear randomly. What seems to have happened however, is that once a herd has had one case, it is more likely to get another than a herd that has not already had a case. i.e. it is as if winning the lottery once made you more likely to win it again. One explanation of this is that once a farmer had seen one cow with BSE , he was more likely to recognise it again. However, the information that I get is the opposite; once a farmer has recognised the first case he is knows what to look for in a second case and takes it to market i.e. if you've won the lottery once you would be expected to be less likely to win it again. Farmers talk about 'infected herds'. It is difficult to explain the appearance of cases in apparently previously unaffected herds (this is still happening). Model 2 will explain these cases but only in small herds would this be statistically valid.
Epidemiology seen fits precisely in what would be expected with vertical transmission. An infective batch would affect the mothers and hence the offspring in later years would be more likely to develop disease.
My own attempt at doing a sibling study was poor. Although there was an excess of siblings going down with BSE (p=0.05) I am unhappy with the size of the study (30 affected herds) and neither model would expect an excess (except the vertical transmission model would expect an excess born early in the epidemic).
D. Steady age distribution of BSE in cattle born in different years.
The surprising steadiness of the age distribution (until the feed ban, after which the cattle were apparently showing symptoms when younger). It was expected that the age distribution peak gradually drops as the amount of infectivity present in the feed increases. This would depend on the number of infective animals per batch reaching multiple levels, or that the disease came from scrapie originally.
The steady age distribution is exactly what would be expected with vertical transmission, however.
E. Steady in-herd rate of BSE
As more lottery tickets are sold, there should be more winners, and if they are distributed randomly then the numbers per herd should rise in parallel (statistically) the national rate. What actually happened was that the rate stayed relatively steady. After 'under-reporting' is taken into account the figure stays between 2 and 3 after 1089. In other words to have an epidemic what was required was that more herds would be infected. By 1988 around 90% of the dairy cattle in the UK were in an infected herd. i.e. the annual national rate could rise little further. The rate represents a much higher rate seen in cattle that are infected but slaughtered before symptoms appear (around 20-25% of the dairy cattle born in 1988). (This agrees with the data from the Statistical group at Reading University). It is as if only 20-25% of the cattle in a herd could be infected with BSE (no matter how much infection is given). This was agreed by Reading, who were involved in looking at the genetics of the cattle (they only looked at the PrP gene) for a reason why one cow did develop disease and another did not. They found that although there was genetic variation, it did not correlate with disease.
What the pattern fits well with is vertical transmission. It would be the mothers that were becoming infected because they were fed relatively large quantities of infective feed and a percentage of their offspring would develop disease. This would fit with the data concerning scrapie and vertical transmission of Dickinson. The reason that only a certain percentage 'can' be apparently infected is that all the mothers are infected to produce the annual steady in-herd rate.
F. Age distribution of BSE in cattle has apparently 2 peaks.
Initially it was difficult to demonstrate the graph accurately because the confidence intervals for cattle dying at ages over 7 were becoming unacceptably wide. However, with further statistical analysis it was possible to show that 2 peaks existed. This is under argument in that the number of cattle in each age group in the national herd is poorly known. My data is derived from the former Milk Marketing Board and includes information on beef suckler dams and bulls. When these last two groups are removed the second peak still appears to rise at the age of 10. It has been suggested that the effect is due to the younger group becoming infected when very young and open to infection more easily (the first few days of life? - this seems unlikely as much of the bovine meal used at that time does not contain MBM) and the second peak is that of the milking cattle being fed extra MBM to maintain milk yields.
A. Cattle born after the feed ban are more likely to be from a farm with pigs
The reason that this is thought to be important is that material meant for feeding pigs may have been used to feed the cattle. Unfortunately there was inadequate data in the study to make this certain. For instance large farms are more likely to have pigs and are also more likely to have BSE cases. Similarly wealthy farmers are more likely to have pigs and to have fed their cattle with expensive materials for milk production (the meat and bone meal is relatively expensive).
B. There appears to be no difference between the herd of cattle with BSE as without it.
This is put forward as a reason but in fact would produce the same result with models 1 and 2.
C. Cattle with BSE are similarly likely to have offspring with BSE as cattle from the same herd without BSE.
Again this could produce the same result for both the models.
D. Only 40 cases of BSE have been seen in the 600+ cattle of the cohort study.
These are the cattle that are either the offspring of mothers
with BSE or the offspring of other mothers from the
E. It is denied that there are two peaks to the age distribution of cattle dying of BSE
When looking at the age distribution of cattle dying of BSE in herds with homebred cases it is possible to know how many cattle are present in each age group and from that work out how many cases would have existed if there had been no slaughter. This suggests that the second peak is not really there. The data is difficult to interpret in that it disagrees with the overall study for the bovine population.
F. There was no excess of BSE in the siblings of cases This work was done by the statistics group at Reading University and so is reliable. It was done mainly on a single herd. When work was done by directly asking farmers in a major agricultural show, the reverse was found (p less than 0.05). Neither model would expect a sibling excess, however.
Also: remember that there are problems with some of the research done so far and its interpretation: The data on transmision of scrapie by embryos is quite equivocal because some of the experiments were not well designed. I understand that experiments have been ongoing for some time to determine the capacity of zona-intact embryos to resist infection by BSE. The term "vertical transmission" can be confusing because it does not differentiate between infection of the embryo prehatching from the zona pellucida and infection post hatching and dependency on a functional trophoblast and placenta. I cannot detect any effort in the information you present to elucidate the importance of ascertaining the susceptibility of the embryo-fetus at these definitive stages to infection with BSE agent. This is an important facit of designing control and prevention strategies. (sent by John Shelton, Australia)
It is important that we know who wants information about BSE and what you think about what is here. If you would like regular information mailed to you on the mailing list or the internet journal please tell us.