 |
From: Terry S. Singeltary Sr. (216-119-144-33.ipset24.wt.net)
Subject: A case-control study of sporadic CJD in the U.K. Analysis of clustering [FULL TEXT]
Date: December 14, 2004 at 8:01 am PST
-------- Original Message --------
Subject: A case-control study of sporadic Creutzfeldt–Jakob disease in the United Kingdom Analysis of clustering
Date: Tue, 14 Dec 2004 10:02:48 -0600
From: "Terry S. Singeltary Sr."
To: Bovine Spongiform Encephalopathy
CC: cjdvoice@yahoogroups.com
CME A case-control study of sporadic
Creutzfeldt–Jakob disease in
the United Kingdom
Analysis of clustering
L. Linsell, MSc; S.N. Cousens, MA; P.G. Smith, DSc; R.S.G. Knight,
FRCPE; M. Zeidler, MRCP;
G. Stewart, MRCP; R. de Silva, MRCP; T.F.G. Esmonde, MD, FRCP; H.J.T.
Ward, MFPHM;
and R.G. Will, FRCP
Abstract—Background: The authors investigated whether cases of sporadic
Creutzfeldt–Jakob disease (CJD) had lived
closer to one another at some time in life than individuals without
sporadic CJD. Such a phenomenon would be compatible
with some cases resulting from transmission. Methods: UK sporadic CJD
cases occurring from 1990 to 1998 were
identified. Age-, sex- and hospital-matched controls were recruited.
Lifetime residential histories were obtained by
interview, usually with a proxy respondent. With use of Monte Carlo
simulation, the residential proximity of cases during
various time periods was compared with that expected in the absence of
any clustering, using the information collected on
the controls. Results: Two hundred twenty sporadic CJD disease cases and
220 controls were included. Cases lived closer
together than might be expected in the absence of any disease-clustering
mechanism. This evidence became stronger as
the critical period during which residential proximity was required to
have occurred extended further into the past.
Conclusions: These findings are consistent with some sporadic
Creutzfeldt–Jakob disease (CJD) cases resulting from
exposure to a common external factor. The rarity of sporadic CJD
suggests that repeated point-source outbreaks of
infection are more likely to explain our observations than direct
case-to-case transmission. Identifying sources of such
outbreaks many years after the event will be extremely difficult.
NEUROLOGY 2004;63:2077–2083
The transmissibility of Creutzfeldt–Jakob disease
(CJD) was first demonstrated experimentally in
1968.1 Later, evidence emerged that a small proportion
of CJD cases were the result of iatrogenic
transmission, most notably through the use of
human-derived pituitary hormone and dura mater
grafts.2 A further small proportion of cases are familial,
linked to mutations in the prion protein (PrP)
gene,3 and it has been suggested that the presence of
one of these mutations is, alone, sufficient to cause
disease.4 Prior to the identification of a new variant
of CJD in 1996,5 all other CJD cases were referred to
as “sporadic.”
Epidemiologic studies have attempted to elucidate
the role of environmental factors in the etiology of
sporadic CJD. Hypotheses investigated have included
that cases were infected through diet, occupation,
surgery, contact with animals, and contact with
other cases.6-13 Despite these efforts, the mode of
transmission, if any, of sporadic CJD remains unknown.
It remains a plausible hypothesis that sporadic
cases of CJD arise through “spontaneous”
endogenous events rather than by infection from another
individual or an environmental source.
Clusters or areas of high incidence of “sporadic”
CJD cases have been reported in the past. Those in
Slovakia,14 in Chile,6 and among Libyan Jews in Israel15
have since been shown to have occurred in
individuals with a mutation at codon 200 of the PrP
gene.16-18 Thus, these cases are familial; if this genetic
mutation is alone sufficient to cause disease,
the “clustering” is readily explained. Small “clusters”
of sporadic cases have also been reported in Australia,
19 England,20,21 France,22 Japan,23 and the USA.24
Such reports of clusters of rare diseases are extremely
difficult to interpret,25 especially when the
space and time dimensions of the cluster are defined
around the cases involved rather than on the basis of
an a priori hypothesis.26 However, if clusters of cases
in space or time are identified, and it can be shown
that they are unlikely to be due to chance, they may
provide important clues to the etiology of the disease.
From the Department of Infectious and Tropical Diseases (Dr. Smith, L.
Linsell and S.N. Cousens), London School of Hygiene and Tropical
Medicine, London,
and National CJD Surveillance Unit (Drs. Knight, Zeidler, Stewart, de
Silva, Esmonde, Ward, and Will), Western General Hospital, Edinburgh, UK.
Funded by a grant from the Department of Health. The CJD Surveillance
Unit is funded by the Department of Health and the Scottish Office
Department of
Health.
The views expressed in this publication are those of the authors and not
necessarily those of the Department of Health.
Received March 17, 2004. Accepted in final form August 12, 2004.
Address correspondence and reprint requests to S.N. Cousens, London
School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, UK;
e-mail: simon.cousens@lshtm.ac.uk
Copyright © 2004 by AAN Enterprises, Inc. 2077
This article describes the statistical analysis of
the lifetime residential histories of cases of sporadic
CJD occurring in the United Kingdom between May
1990 and December 1998 together with similar data
from age- and sex-matched controls.
Methods. Prospective surveillance for CJD in the United Kingdom
was instituted in May 1990. Surveillance methods have been
described in detail elsewhere.27 In brief, the two major mechanisms
of case identification have been through direct referral from
targeted professional groups (neurologists, neuropathologists, and
neurophysiologists) and from death certificates coded under rubrics
046.1 and 331.9 in the International Classification of Diseases
(9th rev.). A diagnosis of “definite” sporadic CJD requires
neuropathologic confirmation. A diagnosis of “probable” sporadic
CJD was assigned if there was a characteristic EEG or a positive
14-3-3 test in a patient with progressive dementia and at least two
of the following four symptoms or signs: myoclonus, visual or
cerebellar disturbance, pyramidal or extrapyramidal dysfunction,
akinetic mutism. Whenever possible, cases were examined clinically
in life by a neurologist from the National CJD Surveillance
Unit, and an interview was conducted with a relative using a
standard questionnaire. Among the information obtained was a
lifetime residential history for the case, up to the date of the
interview. From May 1, 1990, to December 31, 1998, an effort was
made to recruit as a control an inpatient at the same hospital as
the case, matched for sex and age 4 years. Individuals with
diseases that might be confused clinically with CJD were excluded
from acting as controls. Whenever possible, a relative of similar
degree as for the case was interviewed using the same standard
questionnaire. When this was not possible, the control was interviewed
directly.
As far as possible, precise addresses and dates of moving were
obtained for each study subject. The grid reference of each address
was identified through its postcode using AFD Postcode Plus (AFD
Software Ltd., Isle of Man, UK). Because generally only the year of
an address change was given, the address at which the individual
resided on January 1 of each year was taken as the location for that
year.
Three hundred fifty-five definite and probable cases of sporadic
CJD were identified between May 1, 1990, and December 31,
1998. Residential history data, together with control data, were
available for 220 (62%) of these (188 definite and 32 probable
cases).
For 120 controls, residential histories obtained from a relative
were available. For 99 of these, residential histories obtained directly
from the control were also available. For 100 controls, residential
histories were obtained only directly from the controls
themselves. In the case-control analyses presented below, residential
history data obtained from a relative have been used, when
available, to maximize comparability with the data obtained on
cases.
To look for evidence of past residential proximity that might
indicate transmission of CJD, we specified periods during which
an individual was assumed to have been “susceptible” and acquired
the infection and during which he or she was “infectious,”
that is, periods when the individual might have acquired the
infectious agent from another case and periods when that individual
might have infected others. All possible pairs of cases were
examined to determine whether they lived within a critical distance
of each other when one member of the pair was “infectious”
and the other “susceptible,” that is, when, under the postulated
model, transmission from one to the other could have occurred.28
Given the potentially long incubation period of CJD, this approach
offers advantages over methods based on place of residence at
diagnosis or death.
Two measures of “contact” were used. The first measure simply
counted the number of pairs of cases who had lived close to one
another when one was presumed infectious and the other susceptible.
The second measure counted the number of years during
which such a pair of cases lived close to one another. The distribution
of either measure of contact under the null hypothesis of no
clustering was computed by Monte Carlo simulation. For each
case-control pair, one individual, either the case or the control,
was selected at random. The amount of contact between these
randomly selected individuals was then computed (based on their
residential histories and the presumed periods of infectivity and
susceptibility). Repeating this procedure 500 times produced a
distribution for the measure of contact under the null hypothesis.
The observed amount of contact between cases was then compared
with this distribution to determine what proportion of the simulated
measures of contact exceed that actually observed between
the cases. This gave a measure of the probability of obtaining the
observed measure of contact if there was no underlying mechanism
(other than chance) that could lead to clustering.
Results. Among cases and controls, there were similar
proportions of addresses for which only the name of a
county or of a large conurbation was available (4.1% for
cases and 3.5% controls), similar proportions of addresses
for which dates had to be estimated (5.9% for cases and
7.3% controls), and similar proportions of addresses that
were missing or unknown (1.6% for cases and 1.9% for
controls). On average, cases had lived at 6.5 different addresses
during life, up to the time of the interview, and
controls had lived at 5.6 different addresses (p 0.004).
To assess the impact on data quality of using proxy
respondents, the residential histories obtained for 99 controls
from two sources, the control and a relative, were
compared. The distance between the map coordinates identified
from the relative’s history and that from the control’s
history were computed for each year. A total of 6,514 addresses
were compared. Of these, 68% were within 1 km of
each other, whereas for a further 2.6%, both respondents
reported a “missing” address (usually an address outside
the United Kingdom). A further 14.8% of addresses were
within 1 to 5 km of each other. The remaining addresses
(14.6%) were 5 km apart, or one of the addresses was
missing. Controls themselves reported an average of 5.9
different addresses throughout life, while their relatives
reported an average of 5.5 (p 0.25).
Various susceptible and infectious periods were investigated,
beginning up to 25 years before death, with critical
distances of 5, 10, and 20 km. Tables 1 and 2 present the
results for a critical distance of 20 km. Table 1 shows the
observed and expected numbers of pairs of cases meeting
the residential proximity criteria. Thus, if Cases A and B
lived within the 20 km of each other at a time when Case
A could have been infectious and Case B was susceptible,
the measure of “contact” for Pair A and B is 1. If, in
addition, Case B could have been infectious when Case A
was susceptible, then the measure for Pair A and B is 2.
Table 2 presents the observed and expected numbers of
years of such residential proximity “contacts.” So, if Pair A
and B were resident within 20 km of each other for 3 years
when A could have been infectious and B susceptible and
for 5 years when B could have been infectious and A susceptible,
then the measure for the pair is 8. For example,
for a critical distance of 20 km and susceptible and infectious
periods beginning 10 years prior to death, table 1
indicates that there were 1,314 pairs of cases meeting the
proximity criteria, whereas table 2 indicates that these
pairs met the proximity criteria for a total of 8,150 years
between them. The corresponding expected values of these
measures are 1,118 pairs and 7,316 years. These “expected”
values and the significance of the observed values
were obtained from the simulated distribution under the
null hypothesis.
There was no evidence that cases were more likely to
have lived within 5 km of each other during the specified
critical periods than might be expected by chance (data not
2078 NEUROLOGY 63 December (1 of 2) 2004
shown). At a critical distance of 10 km, the observed measure
was always greater than that expected under the null
hypothesis (data not shown). For a number of combinations
of susceptible and infectious periods, the excess was
close to the commonly used cut-off point of p 0.05.
The strongest evidence of residential clustering among
cases was observed when a critical distance of 20 km was
used (see tables 1 and 2), with a number of cells in the
tables showing significant excesses (p 0.05). The excesses
are most evident as the critical period during which
residential proximity must have occurred is allowed to extend
further into the past. Thus, relatively few cells suggest
an excess when residential proximity is required to
have occurred within 5 years of death. A high proportion of
cells show some evidence of an excess when infective and
susceptible periods are allowed to extend up to 10 years
prior to death.
Table 3 presents the results of analyses when it is assumed
that individuals may be susceptible or infectious
throughout life. There is no evidence of an excess when
individuals are assumed to be susceptible only during the
first year of life. When individuals are assumed to be susceptible
at any time during life, there is evidence of an
excess number of pairs of cases. The evidence of an excess
is greater as the period of infectivity extends further back
into the past and is greatest for a critical distance of 20 km.
Cells in which there was evidence of an excess were
examined more closely. Cases were grouped into mutually
exclusive sets or “chains” of potential transmission. Each
chain contains all cases linked to each other either directly
or indirectly through another case, based on the residential
proximity criteria. Thus, each case can be in only one
chain. For example, if Case A lived close to Case B at a
relevant time and, in turn, Case B lived close to Case C at
a relevant time, then Cases A, B, and C would be grouped
together in the same chain. Cases D and E, who were
linked to each other but not with any of the cases associated
with Cases A, B, and C, would lie in another chain.
For the narrower time periods of infectivity and susceptibility,
there were numerous small chains confined to different
areas of the country. For example, there were 172
cases that had lived within a distance of 20 km of another
case when infectivity and susceptibility are assumed to be
5 years before death. These cases form 24 chains ranging
in size from 2 to 45 cases. As the time periods for potential
transmission are allowed to lengthen, these relatively
small chains coalesce into a smaller number of much
larger chains, with cases in each chain scattered all over
the country. (As a case in one chain comes into contact
with a case in another chain, the two chains merge to
become a single chain.) The figure shows 2,086 pairs,
formed by 197 cases that were within a distance of 20 km
of each other when infectivity and susceptibility are assumed
to be 25 years before death. These cases form eight
chains ranging in size from 2 to 179 cases.
Unsurprisingly, the majority of residential links identified
occur around the large urban centers (London, Birmingham,
Liverpool, Manchester, Leeds, etc.). The cases
involved in some of the “more remote” links (Aberdeen,
Anglesey, Argyll, Great Yarmouth; see the figure) were
identified and their records examined in detail to see
whether any evidence of direct contact between them could
be found.
Four cases were linked by residence in or around Aberdeen,
northeast Scotland, during the periods 1968 to 1972
and 1989 to 1990. The distances between the places of
residence of the cases ranged from about 3 to 6 km. Ages at
death ranged from 49 to 66 years. One had worked as a
nurse and had been a vegetarian for 20 years. There was
nothing notable in the occupational or dietary histories of
the other three cases. Three of the cases had undergone
operations in the same hospital, but many years apart
(1964, 1974, and 1989).
Two cases, ages 77 and 80 at death, were linked by
residence on Anglesey, Wales, during the period 1970 to
1991, at a distance of just over 5 km. Both had lived close
to farms. There was nothing remarkable in the occupational
or dietary histories of either case. One of the cases
had undergone several operations in different hospitals,
whereas the other had no history of surgery.
Three cases, ages 45 to 64 at death, were linked by
residence in Argyll, west Scotland, from the 1960s to the
1990s. The distances between the cases (as the crow flies)
ranged from 6 to 13 km. However, all three cases were
separated by the Firth of Clyde or Gare Loch. One case
had worked on the father’s farm up to 1960. Another was
married to a butcher who worked in a slaughterhouse, and
the third had sold rabbit skins. None had unusual dietary
histories. Two of the cases had histories of surgery, but not
in the same hospital or in the same year.
Two cases, ages 63 and 73 at death, lived about 2.5 km
apart in Great Yarmouth, East Anglia, during the period
1990 to 1993. One had worked as a long-distance lorry
driver, delivering fertilizer and animal carcasses. Neither
had remarkable dietary histories. Both had histories of
surgery: one in London, the other in Norwich.
The statistical analyses described above were repeated
on the data excluding those addresses that were classified
as possibly having imprecise grid references or estimated
dates (as described in Methods). Very similar results were
obtained, with slightly stronger evidence of excess contact
between cases at critical distances of 10 and 20 km.
Discussion. In a previous study of cases of CJD
occurring in England and Wales between 1980 and
1984, using the same methods as the current study,
we observed an excess of cases born close to where
another (future) case was already living and an excess
of pairs of cases living within 5 km of each other at
some time during life.10 This time, we found no excess
of cases born close to where another (future) case was
already living (see table 3), but we again observed an
excess in the number of pairs of cases that had lived
close to each other at some time during life (see table
3). We also observed excess contact between cases during
periods 3 years before death (see tables 1 and 2).
In terms of distance, the excesses were most apparent
at a critical distance of 20 km and largely absent at a
critical distance of 5 km. In terms of time, the excesses
became increasingly apparent as the analyses went
further back into the past.
There are a number of possible explanations for
the observed excesses. Other than chance, these explanations
fall into two groups: those that attribute
the excess to an underlying disease-clustering mech-
December (1 of 2) 2004 NEUROLOGY 63 2079
anism and those that attribute the excess to the data
collection process (bias). Given the design of the
study, it is not surprising that the excesses are more
evident the further back into the past one goes. Controls
were recruited in the same hospitals as the
cases and thus at the time of recruitment will have
had a similar geographic distribution to the cases.
This “overmatching” on area of residence at the time
the case was ill makes it almost impossible to observe
excess residential proximity between cases in
the recent past using this method. The further back
into the past one goes, the more likely it is that cases
and controls were living in different locations and
thus the greater the likelihood of observing clustering,
whether it arises because of the etiology of the
disease or through some mechanism of bias.
One etiologic interpretation of our findings is that
these cases were involved in the transmission of the
agent among themselves many years prior to disease
onset. However, sporadic CJD is a very rare disease.
It would be remarkable if the infectious agent were
able to maintain itself in the human population in
Table 1 Observed and expected numbers of pairs of cases of sporadic CJD
living within 20 km of each other when one was assumed to
be “susceptible” and the other “infectious”*
Assumed period of infectivity
Assumed period of susceptibility
Death: 1 y Death: 2 y Death: 3 y Death: 5 y
Obs (Exp) p Value Obs (Exp) p Value Obs (Exp) p Value Obs (Exp) p Value
Death: 1 y 132 (130.6) 0.49 254 (251.2) 0.43 364 (347.7) 0.23 512
(476.4) 0.12
Death: 2 y 380 (371.3) 0.35 495 (471.1) 0.21 649 (604.2) 0.11
Death: 3 y 612 (570.3) 0.13 782 (711.4) 0.05
Death: 5 y 972 (869.6) 0.03
Death: 10 y
Death: 15 y
Death: 20 y
Death: 25 y
Analysis was based on 220 cases diagnosed from May 1990 to December 1998.
* A susceptible period of, e.g., Death: 10 y indicates that the analysis
assumed that individuals became susceptible 10 y prior to their
death and remained susceptible thereafter. Likewise, an infectious
period of Death: 5 y indicates that the analysis assumed that individuals
become infectious 5 y before they die of CJD and that they remain
infectious until death.
CJD Creutzfeldt–Jakob disease; Obs observed; Exp expected.
Table 2 Observed and expected numbers of years when cases of sporadic
CJD were living within 20 km of each other when one was
assumed to be “susceptible” and the other “infectious”*
Assumed period of infectivity
Assumed period of susceptibility
Death: 1 y Death: 2 y Death: 3 y Death: 5 y
Obs (Exp) p Value Obs (Exp) p Value Obs (Exp) p Value Obs (Exp) p Value
Death: 1 y 132 (130.6) 0.49 254 (251.2) 0.43 364 (347.7) 0.23 512
(476.4) 0.12
Death: 2 y 510 (499.4) 0.37 743 (716.2) 0.28 1,082 (1,010) 0.12
Death: 3 y 1,112 (1,060) 0.23 1,694 (1,575) 0.11
Death: 5 y 2,810 (2,615) 0.12
Death: 10 y
Death: 15 y
Death: 20 y
Death: 25 y
Analysis was based on 220 cases diagnosed from May 1990 to December 1998.
* A susceptible period of, e.g., Death: 10 y indicates that the analysis
assumed that individuals became susceptible 10 y prior to their
death and remained susceptible thereafter. Likewise, an infectious
period of Death: 5 y indicates that the analysis assumed that individuals
become infectious 5 y before they die of CJD and that they remain
infectious until death.
CJD Creutzfeldt–Jakob disease; Obs observed; Exp expected.
2080 NEUROLOGY 63 December (1 of 2) 2004
the absence of either a much larger number of
asymptomatically infected individuals able to transmit
the infection or the repeated reintroduction of
the agent into the human population, either as the
result of an endogenous event or as the result of
transmission from an animal or other source. (Studies
in mice have shown that prion infections that
remain asymptomatic over the mouse’s normal lifespan
can be induced experimentally.)29 In either circumstance,
one might expect direct case-to-case
transmission to account for only a small proportion
of cases, and one would expect detection of such
transmission to be very difficult. Furthermore, evidence
of geographic clustering of cases was stronger
the larger the critical distance considered. If direct
case-to-case transmission through ordinary daily
contact were responsible for the excess contact, one
might expect the excess to be greatest at the shortest
critical distance investigated (5 km).
The method of analysis used was originally developed
to seek evidence of direct case-to-case transmission
of the type discussed above. To this end, it
requires the specification of both “susceptible” and
“infectious” periods. For many of the analyses we
performed (the cells on the diagonals of tables 1 and
2 and the second and third rows of table 3), we set
the susceptible and infectious periods to be equal. In
these analyses, a pair of cases in which one individual
was susceptible when the other was infectious is
also a pair of cases in which both cases were susceptible
at the same time. The numbers of pairs of cases
who were living close together when both were susceptible
can therefore be obtained by dividing the
numbers in the appropriate cells by 2. Thus, an al-
Table 1 Continued
Assumed period of susceptibility
Death: 10 y Death: 15 y Death: 20 y Death: 25 y
Obs (Exp) p Value Obs (Exp) p Value Obs (Exp) p Value Obs (Exp) p Value
557 (520.8) 0.11
698 (649.3) 0.09
840 (760.3) 0.05
1,068 (937.0) 0.02 1,073 (948.5) 0.006
1,314 (1,118) 0.01 1,365 (1,157) 0.008
1,504 (1,273) 0.002 1,563 (1,307) 0.002
1,848 (1,539) 0.014 1,919 (1,590) 0.008
2,086 (1,719) 0.006
Table 2 Continued
Assumed period of susceptibility
Death: 10 y Death: 15 y Death: 20 y Death: 25 y
Obs (Exp) p Value Obs (Exp) p Value Obs (Exp) p Value Obs (Exp) p Value
557 (520.8) 0.11
1,234 (1,152) 0.11
2,038 (1,886) 0.11
3,915 (3,572) 0.05 3,941 (3,626) 0.08
8,150 (7,316) 0.04 9,440 (8,381) 0.02
14,210 (12,430) 0.01 15,521 (13,496) 0.002
20,918 (18,041) 0.01 22,551 (19,358) 0.01
28,332 (24,223) 0.01
December (1 of 2) 2004 NEUROLOGY 63 2081
ternative etiologic interpretation of the observed excesses
of contact between cases is that it reflects
shared exposure of susceptible individuals to unidentified,
localized environmental sources of either the
infectious agent or some other “trigger” capable of
inducing conformational changes in exposed individuals’
PrP, that is, that the observed excess is due to a
number of point-source “outbreaks.” We emphasize
that for sporadic CJD, there is no direct evidence
that such transmission or induction of the agent has
ever occurred. Identifying this type of source many
years after exposure occurred will be extremely difficult,
if not impossible.
Investigation of a small number of isolated “clusters”
did not identify any evidence of direct contact
between any of the small number of cases involved.
Nor did it identify any unusual environmental
factors.
A further etiologic hypothesis that must be considered
is that the mechanism underlying the apparent
clustering is genetic. If exposure to the agent or environmental
trigger were widespread and unknown
genetic factors located outside the PrP coding region
influenced individual susceptibility to infection or
disease, our observations might then reflect clustering
of the unknown genetic factor(s). However, if this
were the case, then we would expect the strongest
evidence of clustering to be observed around the time
of birth. We detected no evidence of clustering when
the period of susceptibility was restricted to the first
year of life (see table 3).
The retrospective nature of the data collection process,
with much of the data collected from proxy
respondents, presents considerable opportunity for
measurement error and thus bias. All data on cases,
but only some data on controls, were collected from
proxy respondents. If the data on cases were less
accurate than those on controls, with the result that
cases tended to be recorded as living at central locations
in towns and cities, this could result in an
Figure. Chains of contact between sporadic Creutzfeldt–
Jakob disease cases when using a critical distance of 20
km and assuming susceptible and infectious periods begin
25 years prior to death. Each symbol represents a pair of
cases having contact. Circles indicate the locations referred
to in the text. Numbers refer to chain numbers.
Table 3 Observed and expected links between pairs of cases of sporadic
CJD for specified distances during assumed periods of infectivity
and susceptibility
Assumed period
of susceptibility
Assumed period
of infectivity
Distance
Contact
measure
5 km 10 km 20 km
Obs (Exp) p Value Obs (Exp) p Value Obs (Exp) p Value
First year of life Life 0 or 1 133 (135.8) 0.57 336 (330.9) 0.42 799
(772.9) 0.38
Life Life 0 or 1 1,116 (931.3) 0.002 2,536 (2,044) 0.004 4,916 (3,997) 0.004
1/y of proximity 11,060 (11,073) 0.50 33,942 (30,478) 0.07 84,978
(75,417) 0.04
Life Last 40 y of life 0 or 1 650 (563.7) 0.08 1,613 (1,324.4) 0.02
3,400 (2,784.0) 0.02
1/y of proximity 5,806 (6,172.6) 0.72 19,569 (17,659) 0.08 51,874
(44,645) 0.01
Life Last 30 y of life 0 or 1 414 (370.1) 0.15 1,059 (922.7) 0.08 2,460
(2,058.3) 0.02
1/y of proximity 3,968 (4,218.4) 0.71 13,781 (12,429) 0.09 37,322
(31,888) 0.004
Life Last 20 y of life 0 or 1 267 (254.1) 0.34 773 (672.3) 0.07 1,922
(1,596.5) 0.01
1/y of proximity 2,309 (2,561.5) 0.81 8,531 (7,623.5) 0.07 22,590
(19,479) 0.01
Life Last 10 y of life 0 or 1 148 (163.4) 0.82 537 (466.3) 0.04 1,365
(1,151.9) 0.004
1/y of proximity 1,004 (1,126.3) 0.84 3,748 (3,336.2) 0.05 9,472 (835.2)
0.02
Analysis was based on 220 cases diagnosed from May 1990 to December 1998.
CJD Creutzfeldt–Jakob disease.
2082 NEUROLOGY 63 December (1 of 2) 2004
apparent excess of links between cases. There are
several reasons for thinking that this is unlikely to
be the explanation for our findings. First, a comparison
of data obtained directly from controls with that
obtained from proxy respondents suggests that proxy
respondents were able to provide remarkably accurate
data, with 86% agreement to within 5 km. Second,
similar proportions of data for cases and
controls relied on the use of the location of a “central”
point to cover a large area (4.1% of case addresses
and 3.5% of control addresses). Third, the
use of central locations would tend to produce “clustering”
at short critical distances. We found very little
evidence of an excess of cases living within 5 km
of each other. The most significant excesses occurred
with a critical distance of 20 km, the largest critical
distance that we considered. Finally, repeating the
analyses including only addresses considered reasonably
accurate did not alter our findings in any important
way. If anything, the strength of evidence of
clustering increased.
Alternatively, data on cases may be more accurate
or complete than data on controls, perhaps because
proxy respondents for cases may have been more
motivated to provide complete data than those of
controls, because the study was of direct relevance to
their family. The average number of addresses recorded
for cases was 6.5 compared with an average of
5.6 for controls. The greater the number of addresses
individuals live at, the greater is the probability that
they will live close to certain other individuals at
some time. Consider two individuals, A and B. B
lives in the same place all his life. If individual A
lives in the same place throughout her life, then she
has one chance of being in contact with B at some
time during her life. If A lives at two different addresses,
then A’s chances of being in contact with B
at sometime during her life are increased. If controls
were reported to live at fewer addresses than cases
because control respondents’ recall was poorer than
that of cases, then one might observe more than
expected pairs of cases with contact between them.
However, if cases are recorded as moving more often,
the average length (in years) of each contact between
cases will be reduced. Thus, whereas such a phenomenon
might explain excesses in the number of pairs
of contact (first contact measure), it would not explain
excesses in the number of years of contact (second
contact measure).
Acknowledgment
The authors thank the families of CJD cases and controls for their
participation, Jan Mackenzie for data management, and the neurologists
and neuropathologists in the United Kingdom for their
cooperation in surveillance activities. P.G. Smith, S.N. Cousens,
R.G. Will, and R.S.G. Knight initiated the study. M. Zeidler, G.
Stewart, R. de Silva, and T.G.G. Esmonde collected data. L. Linsell
prepared the data for analysis and performed the statistical
analyses. S.N. Cousens prepared the first draft of the article. All
investigators contributed to subsequent drafts of the article.
References
1. Gibbs CJ, Gajdusek DC, Asher DM, et al. Creutzfeldt–Jakob disease
(spongiform encephalopathy): transmission to the chimpanzee. Science
1968;161:388–389.
2. Brown P. Environmental causes of human spongiform encephalopathy.
In: Baker H, Ridley R, eds. Methods in molecular medicine: prion diseases.
Totowa: Human Press, 1996:139–154.
3. Owen F, Poulter M, Lofthouse R, et al. Insertion in prion protein gene
in familial Creutzfeldt–Jakob disease. Lancet 1989;1:51–52.
4. Hsiao KK, Scott M, Foster D, Groth DF, DeArmond SJ, Prusiner SB.
Spontaneous neurodegeneration in transgenic mice with mutant prion
protein. Science 1990;250:1587–1590.
5. Will RG, Ironside JW, Zeidler M, et al. A new variant of Creutzfeldt–
Jakob disease in the UK. Lancet 1996;347:921–925.
6. Galvez S, Masters C, Gajdusek DC. Descriptive epidemiology of
Creutzfeldt–Jakob disease in Chile. Arch Neurol 1980;37:11–14.
7. Brown P, Cathala F, Gajdusek DC. Creutzfeldt–Jakob disease in
France: III. Epidemiological study of 170 patients dying during the
decade 1968–77. Ann Neurol 1979;6:438–446.
8. Davanipour Z, Alter M, Sobel E, Asher DM, Gajdusek DC. A casecontrol
study of Creutzfeldt–Jakob disease—dietary risk factors. Am J
Epidemiol 1985;122:443–451.
9. Harries-Jones R, Knight R, Will RG, Cousens S, Smith PG, Matthews
WB. Creutzfeldt–Jakob disease in England and Wales, 1980–84: a
case-control study of potential risk factors. J Neurol Neurosurg Psychiatry
1988;51:1113–1119.
10. Cousens SN, Harries-Jones R, Knight R, Will RG, Smith PG, Mathews
WB. Geographical distribution of cases of Creutzfeldt–Jakob disease in
England and Wales 1970–84. J Neurol Neurosurg Psychiatry 1990;53:
459–465.
11. van Duijn CM, Delasnerie-Lauprêtre N, Masullo C, et al. Case-control
study of risk factors of Creutzfeldt–Jakob disease in Europe during
1993–1995. Lancet 1998;351:1081–1085.
12. Collins S, Law MG, Fletcher A, Boyd A, Kaldor J, Masters CL. Surgical
treatment and risk of sporadic Creutzfeldt–Jakob disease: a casecontrol
study. Lancet 1999;353:693–697.
13. Ward HJT, Everington D, Croes EA, et al. Sporadic Creutzfeldt–Jakob
disease and surgery: a case control study using community controls.
Neurology 2002;59:543–548.
14. Mayer V, Orolin D, Mitrova E. Cluster of Creutzfeldt–Jakob disease
and presenile dementia. Lancet 1977;2:256.
15. Kahana E, Alter M, Braham J, Sofer D. Creutzfeldt–Jakob disease:
focus among Libyan Jews in Israel. Science 1974;183:90–91.
16. Goldfarb LG, Mitrova E, Brown P, Toh BH, Gajdusek DC. Mutation in
codon 200 of scrapie amyloid protein gene in two clusters of
Creutzfeldt–Jakob disease in Slovakia. Lancet 1990;336:514–515.
17. Goldfarb LG, Korczyn AD, Brown P, Chapman J, Gajdusek DC. Mutation
in codon 200 of scrapie amyloid precursor gene linked to
Creutzfeldt–Jakob disease in Sephardic Jews of Libyan and non-Libyan
origin. Lancet 1990;336:637–638.
18. Brown P, Galvez S, Goldfarb LG, et al. Familial Creutzfeldt–Jakob
disease
in Chile is associated with the codon 200 mutation of the PRNP
amyloid precursor gene on chromosome 20. J Neurol Sci 1992;112:65–67.
19. Collins S, Boyd A, Fletcher A, et al. Creutzfeldt–Jakob disease cluster
in an Australian rural city. Ann Neurol 2002;52:115–118.
20. Matthews WB. Epidemiology of Creutzfeldt–Jakob disease in England
and Wales. J Neurol Neurosurg Psychiatry 1975;38:210–213.
21. Will RG, Matthews WB. Evidence for case-to-case transmission of
Creutzfeldt–Jakob disease. J Neurol Neurosurg Psychiatry 1982;45:
235–238.
22. Huillard d’Aignaux J, Cousens SN, Delasnerie-Lauprêtre N, et al.
Analysis
of the geographical distribution of sporadic Creutzfeldt–Jakob disease
in France between 1992 and 1998. Int J Epidemiol 2002;31:490–
495.
23. Arakawa K, Nagara H, Itoyama Y, et al. Clustering of 3 cases of
Creutzfeldt–Jakob disease near Fukuoka City, Japan. Acta Neurol
Scand 1991;84:445–447.
24. Farmer PM, Kane WC, Hollenberg-Sher J. Incidence of Creutzfeldt–
Jakob disease in Brooklyn and Staten Island. N Engl J Med 1978;298:
283–284.
25. Raubertas RF, Brown P, Cathala F, Brown I. The question of clustering
of Creutzfeldt–Jakob disease. Am J Epidemiol 1989;129:146–154.
26. Rothman KJ. A sobering start to the cluster busters’ conference. Am. J.
Epidemiol. 1990;132(suppl 1):S6–S13.
27. Will RG. Surveillance of prion diseases in humans. In: Baker H, Ridley
R, eds. Methods in molecular medicine: prion diseases. Totowa: Humana
Press, 1996:119–137.
28. Pike MC, Smith PG. A case-control approach to examine diseases for
evidence of contagion, including disease with long latent periods.
Biometrics
1974;30:263–279.
29. Hill AF, Joiner S, Linehan J, Desbruslais M, Lantos PL, Collinge J.
Species-barrier-independent prion replication in apparently resistant
species. Proc Natl Acad Sci USA 2000;97:10248–10253.
December (1 of 2) 2004 NEUROLOGY 63 2083tss
TSS
Follow Ups:
Post a Followup
|
