<<

. 24
( 45 .)



>>

sumed during the lifetime of the animals affected from the compounders of
these rations. The result of this was that m every case of BSE for which full
records were available, meat and bone meal had been mcorporated as a protein
supplement in at least one product This, therefore, provided circumstantial
evidence for a common exposure factor. It also highhghted the complexity,
and therefore diffculttes, of mvesttgations mvolving data on the exposure of
cattle to mdividual ingredients m cattle rations. First, cattle, even m their first
6 mo of life, can be fed a number of different products not necessarrly pro-
duced by the same company. Second, the details of mdtvidual products would
not always be available from the manufacturers, either because of the passage
of time, with records not being kept for a sufficiently long period, or because
companies had been taken over or had simply ceased trading. These factors
were certainly important m thinking about the potential analytical epidemto-
logtcal studies. This identtlication of a common exposure factor was clearly
important, but needed to be examined for tts plaustbthty m the first instance
with respect to the descriptive eptdemiological ptcture, especially because meat
and bone meal was a commonly used ingredtent m commerctally produced
cattle rations.
The first intriguing aspect m the occurrence of BSE was its appearance more
or less comcidentally throughout Great Britain (2.5) However, this needed to
be considered with the evidence of the mittal occurrence of BSE. This was
identified as April 1985, from clinical histortes and examination of archived
fixed brain, suggesting that BSE was a truly novel disease (I 6) Therefore, the
widespread occurrence of BSE was consistent with an exposure factor com-
mon throughout Great Britain. Another strikmg aspect of the descriptive epi-
demtology was the markedly greater incidence of drsease m dairy herds
compared with beef suckler herds (14). This difference was certainly exph-
cable by the meat and bone meal hypothesis because the feeding of concen-
trated rations is relatively uncommon in commercial beef herds, m which
nutrition during the winter period is largely based on conserved grass and
homegrown cereals.
The descriptive epidemiologtcal features that were most dtfficult to recon-
cile with the working hypothesis were, first, the geographical variation m risk,
with animals m the south of England having a notably greater risk of disease
than those m the north of England and Scotland. The second was that meat and
bone meal had been used as a protein supplement in commerctal foodstuffs for
decades; tts uttltzatton was not a recent event. In addmon, the question of why
BSE had only occurred, or been reported, m Great Britain pervaded, at least
subconsctously, epidemiological thinking.
BSE Analysis m the UK 161

Two observations were made tending to countermand these findings that were
potentially detrimental to the working hypothesis. The first was the occurrence of
BSE on the Channel Islands of the British Isles. In the early years, BSE was con-
firmed on the island of Guernsey, but not on Jersey. This stimulated an investiga-
tion of the suppliers of commercial cattle feedstuffs to these two mam Channel
Islands, and revealed that the principal manufacturer of cattle rations imported by
each island was different, although both were m mainland Great Britain and feed
was not imported from any other sources. The main supplier to Guernsey was
found to use meat and bone meal relatively frequently, whereas that for Jersey had
not considered meat and bone meal as a nutritionally/economically valued mgre-
dient in the formulation of the maJority of its cattle rations. The eventual occur-
rence of BSE on the island of Jersey,but at a much lower incidence, provided the
necessary natural experiment, the results of which supported the hypothesis.
The second observation was the occurrence of cases of TSE m a nyala m
1986 and a gemsbok m 1987 m a zoological collection m the south of England
(17) These had been identified in the course of the normal surveillance of
animal diseases. At the time, thoughts on the etiology centered on a solely
genetic ortgm However, further mvestigations followmg the establishment of
the working hypothesis for BSE proved helpful. The nutrition of the collec-
tions of these two species had included commercially produced feedstuffs, but
meat and bone meal had been specifically excluded from the ingredients at the
request of the keepers, for reasons that were not related to any concern about
the risk of mfection with a scrapie-like agent. Vegetable proteins, such as soya,
had been used, but when the price of this commodity increased markedly, meat
and bone meal was used for economy for a very short period of time (13). The
most probable mcubation period was shorter than that which had been observed
in cattle, but the coincidence could not be ignored.
In retrospect, these “observations” were only made possible by a good
degree of communication and enticing the curiosity of mdividuals to assist m
explammg unusual occurrences that would be a help m understanding the larger
problem. However, the basic surveillance was a key to providing supportmg
epidemiological evidence as advised by Hill (Z Z)
One problem with the meat and bone meal hypothesis was that meat and
bone had been used m the formulation of cattle rations for many years; it was
not a recent development. At this stage, discussions had obviously taken place
wtth nutritionists m the animal feedstuffs industry, but not with the rendering
industry, which produces meat and bone meal. Instead, efforts were concen-
trated on an attempt to determine the most likely time of the onset of effective
exposure of the British cattle population.
The occurrence of the AIDS epidemic has stimulated the development of
methods of estimating the time of exposure to infection (28) However, a rela-
162 Wilesmlth

ttvely sample determmtsttc stmulatton approach was used for BSE when the
number of cases was small (12,13) This involved mvestigatmg the probabilt-
ties of exposure as calves, yearlmgs, and adults, estimating the mcubation
period distribution, and exammmg the possibilities of exposure at various
times. The validity of the outputs from the simulattons was assessed by the fit
to the age-specific mctdences. The best tit indicated that the onset of effective
exposure occurred suddenly m 1981-1982 and that the majortty of clinical
cases observed had been infected m calfhood. More sophtsticated stattsttcal
techniques that have been used during the course of the epidemic have not
altered these baste conclusions, but have provided improved estimates of the
incubation period dtstrtbution (I 9)
This finding provided some confidence m addressing the question of why
cattle became exposed This was directed at examining the production pro-
cesses used to render waste animal tissues to produce meat and bone meal A
prerequisite for this was to discuss the eptdemiological evidence with repre-
sentatives of both the animal feedstuffs and rendering mdustries, to gam their
confidence and cooperatton. As a result, a survey of all rendering plants was
conducted m the second half of 1988 The mam objectrves of this were*

1 To obtain details of the changes in rendering processes during the 1970s and 198Os,
2 To determine current time-temperature treatments in the various processes, and
3 To obtain details of the species composition of the waste tissues rendered in each
plant over time
The survey was conducted by three veterinary officers with a knowledge of
the rendering mdustry who had been briefed specifically about the objectives.
Data were obtained by a questionnaire and time-temperature measurements were
recorded m a uniform manner, having evaluated a number of potential methods.
The results of this survey indicated that the estimated onset of effective exposure
was comcident with a marked reduction m the use of hydrocarbon solvents to maxi-
mize the yield of tallow (fat) during the rendering process (20) Only limited
laboratory-based studies had been carried out on the use of organic solvents
for the disinfection of surgical and laboratory mstruments and equipment. How-
ever, there was some indication that this change was a btologtcally plausible
explanatton for the onset of exposure. The survey also provided a further expla-
nation for the geographical variation in the risk of infection of cattle, notably the
continued and exclustve use of the solvent extraction process in Scotland, where
the majority of meat and bone meal produced had undergone this treatment.
This whole stage of the investigation was obviously important because of
the need to assess the validity and strength of the mttial hypotheses. If the
assessment had failed to provide substantiating and explanatory data and mfor-
mation, second thoughts on mvestigating the epidemic would have been
BSE Analysis m the UK 163

required. This, fortunately, proved not to be the case and the success,at least m
part, could be attributed to the collection of sound descriptive eptdemiologtcal
and other data based on documented sources. The results, therefore, were used
as a basis for formulatmg the first statutory control measure, mtroduced m July
1988, to prevent further exposure from the feed-borne source, and for destgn-
mg further, more analytical, eptdemrologtcal studies.
5. Analytical Epidemiological Studies
to Test the Working Hypothesis and to Determine
the Occurrence of Other Potential Means of Transmission
5.1. Analytical Epidemiological Studies
The most useful epidemiologtcal approach to examme both the valtdtty of
the meat and bone meal hypothesis in general and the solvent-extracted meat
and bone meal hypothesis m particular would have been a cohort study. This
was not possible for either because of the changes m feed supphet that can
occur even during calfhood, and because the contents of rattons change result-
ing from the use of the least-cost formulatton method and variation in the sup-
pliers of mdivtdual ingredients.
A case-control approach (21), therefore, needed to be contemplated m early
1988, following the identification of the most likely source. The usual constd-
erattons of selecting an appropriate control population and mmtmtzmg all
sources of bras were necessary. At this stage sample stze was not constdered
and efforts were concentrated on identifying a suitable control population.
Because the disease had not yet been made statutorily nottfiable, but became
so m June 1988, tt was considered important to prevent the misclassrficatton of
herds, and, therefore, possibly mdrvtduals Also, the early tdentificatton of
potential control herds was an attempt to select controls before the meat and
bone meal hypothesis had gained general currency. To these ends the use of
neighborhood controls was the chosen approach, but with the knowledge that
overmatching could ulttmately be a problem. Veterinary practices that had
reported cases voluntarily were contacted to Identify potentially suitable con-
trol herds. The criteria for these mcluded:
1 The herds were subject to regular routine visits by the attendant veterinary surgeon,
2 The herd owners or workers generally sought veterinary advtce when animals
became 111;
3 Record keeprngwas a routme; and
4. The majortty of replacement animals for the adult herd were homebred.
This was successful and the necessary list was compiled.
it was intended to conduct the next phase, obtaining detailed feeding histo-
ries, by visiting the farms and interviewing herd owners and workers to secure
164 WilesmIth

comparable mformation to that being obtained routmely for BSE cases, with
confirmatory documentation from receipts and other records, but this proved
impossible because msufficient field staff resources were available. As a result,
a postal questionnaire had to be employed that, although less satisfactory, drd
produce the mformation that was essential
The final decision in completing this study involved three aspects First,
which birth cohort would be used m the analyses? This was a relatively straight-
forward decision and was based on the modal age of onset of 4-5 yr. In order to
conduct the analyses at the earliest opportunity, sufficient cases must have
occurred m the cohort born nearest to the estimated time of the onset of effec-
tive exposure The second aspect, sample size, was necessarily of lesser tmpor-
tance, essentially because the number of neighborhood control herds could not
be increased prospectively However, the remammg control herds, which had
not experienced a case of BSE, were considered to be sufficiently numerous to
be used m the analysis. The third aspect, which was an overridmg one, was the
assessment of the completeness of the details available from the manufacturers
of the mgredients of the feedstuffs consumed by both the cases and controls.
Although it was clearly not possible to examine the potential dose-response
relattonship because of the lack of records on the mclusion rates of meat and
bone meal, the data on the presence or absence of meat and bone meal was
considered to be sufficient. The final results of this study provided supportmg
evidence for the hypothesis (22).
This was, therefore, a difficult study to design and rf the disease had been
less important the study may not have been pursued, notably because of the
possibility of the specific problem of overmatchmg and the necessity to limit
the study to calfhood exposure. These could have conspired to produce a false
negative result. However, it could be regarded as an example of having to cope
with such difficulties, and, providmg the basic design could be made sound,
persevering rather than abandoning all hope
5.2. The Question of Maternal Transmission
It ts, of course, good scientific practice to read the appropriate scienttfic
literature when “new” diseases occur. The literature on kuru in the Fore-speak-
ing people of Papua New Guinea indicates that maternal transmission does not
occur (23). However, a cursory reading of the hterature is persuasive that
maternal transmission is the means by which sheep scrapie is naturally mam-
tamed. A more critical review of the published studies suggests quite strongly
that this could have been a self-perpetuatmg myth, and that m high-incidence
flocks hortzontal transmission is of at least equal importance (24).
The initial approach to the question of maternal transmission of BSE was to
determine whether or not this means of spread, on its own and n-respective of
BSE Analysis in the UK

the mechanism, could mamtam the disease m the cattle population. The only
method available to examine this question was simulation modelmg, which
revealed that maternal transmission alone could not sustain the epidemic (5).
This result was of considerable importance, but, returnmg to the literature,
the only excreted material m which the sheep scrapie agent was known to be
present was the placenta. Therefore, exposure to this tissue could result in trans-
mission to either offspring (maternal) or to unrelated animals (horizontal). In the
quest to determine whether BSE was transmitted naturally, It seemed likely
that the occurrence of maternal transmission would be identifiable earlier than
the occurrence of horizontal transmission. If evidence of maternal transmis-
sion was found, then a worst case scenario assuming horizontal transmission
would have to be considered in deciding what animal disease control measures
needed to be taken.
As a result, the decision was made to expend a great deal of time and effort
on determining whether or not there is maternal transmission of the BSE agent,
and a cohort study was instigated m July 1989 (25). This is a long-term study
for which the result is not available at the time of wrmng this chapter, but m the
meantime other opportunities have been identified and acted on to obtain the
earliest possible valid answer to the question (see Section 7.).
It is unfortunate that the reasons for studying the question of maternal trans-
misston m such detail have often been mtsmterpreted, and that the myth of the
occurrence and importance of maternal transmission m animal TSEs, particu-
larly sheep scrapie, has apparently been perpetuated. With hindsight, this IS a
problem of education to counter preconceived ideas and myths, but even wtth
the benefit of hindsight it IS difficult to know how the problem, which in the
case of BSE has proved to be important m discussions on the trade m animals
and animal products, could have been prevented.
6. Monitoring the Epidemic
The monitormg of the BSE epidemic was facihtated by the diseasebeing statu-
torily notifiable, particularly becausethis maximized the ascertainment of cases
and therefore removed any bras.The mam aims of monitoring the epidemic were
to assessthe effects of the statutory action, to identify any changes m the disease,
to mvesttgate means of transmission other than the feed-borne source, and to
provide eptdemiological data to construct analytical studies. It required the con-
tinued allocatton of veterinary field staff resources to visit animals suspectedof
having BSE, examine them clmically, and complete the standard epidemio-
logical questionnaire, not an mconsiderable undertaking given the size and
duration of the epidemic and therefore the fortitude and endurance necessary.
An underlying need was the computmg facthty to handle the eptdemiologi-
cal data efficiently and conduct the analyses to fulfil, at least, the first aim. The
166 Wilesmith

mam descrlptlve epldemtologlcal features used to assess the effects of the statu-
tory control measures that have been taken were age-specific mcldences, the
weekly reportmg rate of suspect cases, the monthly animal and herd Incidence,
and the within herd incidence. All of these have indicated that the statutory ban
on the feeding of ruminant-derived protein to ruminants has had a posmve
effect (26-28). The fulfilment of the second arm was naturally mtermmgled with
investigating the first, but required addmonal analyses-notably to detect any
change m the incubation period distribution. This was achieved by cohort analy-
ses (29), which fortunately did not reveal any changes that would have made
the analysts and interpretation of the eprdemiologrcal findings more complex
Specific analyses have concentrated on two other potential changes: the
clmlcal mamfestatlon of the disease and any change m the dlstrlbutton of
lesions m the bram that could invalidate the routme hlstologtcal diagnosis. The
first was examined by a comparison of the frequency of clinical signs between
cohorts, but no change was identified (30). The second change could have
occurred because of a change m the BSE agent as a result of its passage m
cattle. This was examined by taking a systematic sample of whole brams from
suspect cases throughout the epidemic and studying the drstrtbutton of lesions.

<<

. 24
( 45 .)



>>