Professor Carl Heneghan: How Positive COVID Tests are Forever

00:00
hi and welcome this is lockdown TB from
00:03
unheard and so earlier this week we
00:05
recorded a wine raging discussion with
00:08
Professor Karl Hennigan of Oxford
00:10
University’s Center for evidence-based
00:12
medicine and his colleague Tom Jefferson
00:15
covering all sorts of aspects of the
00:17
covered nineteen pandemic and we will be
00:19
publishing that shortly but yesterday
00:22
evening news broke of a discovery that
00:24
Professor Hanigan had made in these
00:26
statistics so we thought we’d quickly
00:28
catch up with him and hear about that
00:30
beforehand
00:31
professor Hanigan thanks for doing this
00:32
you’re welcome
00:34
so this concerns deaths and the numbers
00:38
of deaths that are attributed to Cobb in
00:40
nineteen and these matters because that
00:44
daily total that is is announced for
00:47
deaths in England affects the whole
00:50
atmosphere it affects whether people
00:52
think that the pandemic is coming to an
00:54
end or not it even affects politics for
00:57
example it’s referred to by Nicola
00:59
Sturgeon and neighboring countries as to
01:00
where what border arrangements should be
01:02
and you discovered that that daily
01:05
deaths total may not actually correlate
01:08
to people dying from carbon 19 tell us
01:11
about that yeah one of the things about
01:13
England and particularly UK is it’s
01:15
quite confusing when you look at the
01:17
different ways deaths that are reported
01:19
you’ve got office for national
01:20
statistics you’ve got public health
01:22
England and then you’ve got NHS England
01:24
then you’ve got the four devolved
01:26
nations who all do slightly different
01:28
things so it’s it’s very confusing but
01:30
one of the things that we noticed is
01:32
public health England grace lots of
01:35
confusion because one day for instance
01:37
on the six of July they will report 16
01:39
deaths and then the next day there’ll be
01:41
hundred and fifty-two deaths and at that
01:43
point the media takes it but notice
01:45
starts to get concerned and you see
01:47
these reports of hundreds of deaths per
01:49
day so what we did is go back and look
01:52
at the Office for National Statistics
01:53
who report the death who are registered
01:57
so when you go to the register your
02:00
death that death certificate is then
02:02
centrally registered so they’re very
02:04
accurate about the day they occurred and
02:06
what we noticed if you go back for
02:08
instance in the 30th of June the PAP
02:10
figures put health England are about
02:12
double
02:13
the ONS figures so we ask questions of
02:16
what’s going on here and what we noticed
02:18
is if you look at the way phe does their
02:21
reporting they are reporting anybody who
02:25
has had a test positive for Kovan in the
02:28
past therefore a death reported today
02:31
could have occurred in could’ve the
02:34
Cobra could have occurred in early March
02:36
you may not even be related but the way
02:39
they’re recording it it assumes that
02:41
people are thinking oh there are over a
02:43
hundred deaths today things are getting
02:45
worse
02:46
no it’s interested in the devolved
02:48
nations like Northern Ireland and
02:49
Scotland so just just so I’m really
02:51
understand it so if the individual
02:54
patient had covered 19 in March Gregg
02:57
recovered completely was subsequently
03:00
tested negative and then died in a
03:03
nursing home three months two months
03:05
later that would then be recorded as
03:07
occurred in nineteen death that is what
03:09
we assume is happening from a public
03:11
health England and that’s where the
03:12
disparity is coming from the wide
03:14
variation interestingly if you look at
03:17
Scotland and Northern Ireland they have
03:19
a sort of 28-day cutoff period and
03:22
that’s exactly what we’re asking for
03:25
what you really want to know is what are
03:27
the deaths occurring in the context of
03:29
an actual test that has occurred
03:31
positive tests in the last 28 days and
03:34
then you can then you can understand the
03:36
trends what’s happening and okay so k2
03:39
report the historical deaths but
03:41
actually they don’t help us understand
03:43
what’s happening now now this will get
03:46
increasingly confusing as we go into the
03:48
next winter because you could have a new
03:50
outbreak new death but you’d also report
03:53
in historical deaths we may not see that
03:56
trend in the way that they’re currently
03:58
reporting this you know if someone had
04:01
come in 19 now recovers and then dies of
04:06
something completely unrelated next yeah
04:10
in the current system yeah and that’s
04:13
why we’re saying it looks like you can
04:15
never recover under the current system
04:17
because he could die of influenza for
04:19
instance in January next year but
04:21
because he had a positive Coby test
04:22
you’d still appear in the figures now
04:25
that creates a real
04:26
problem for us as epidemiologists
04:27
because then we get confused but not
04:30
least the media’s getting confused so
04:33
one of the key aspects were asking here
04:35
is for some clarity in terms of how we
04:38
actually do this and some joined up
04:40
thinking between ons NHS England and
04:43
public health England so you get one
04:45
emerging datasets that’s consistent so
04:48
if they fix this which kind of we hope
04:52
they will understand they in Sweden
04:54
there’s a similar method for counting
04:56
carbon 19 deaths but there’s a 31-day
04:59
kind of elapsing period and also I think
05:03
many states in America count them in the
05:06
same way as well and are you are you
05:09
even happy then though I mean if there
05:12
is if there’s a sort of period where the
05:14
the fact of the carbon 19 infection is
05:18
counted on the death certificate and
05:20
that elapses odd to say a month do you
05:23
think we’ve then got accurate numbers
05:25
yes so there are two distinct issues you
05:28
get into is the immediate cause and the
05:30
immediate cause means you’ve had Kovac
05:32
and we suggest that actually that’s
05:34
within 21 days because that gives you a
05:37
time where you probably still not
05:39
recovered and you’re in hospital and you
05:41
died outside of that it tends to be an
05:44
underlying cause it contributed your
05:46
death but it wasn’t the direct cause of
05:48
your death and actually a 21 day cutoff
05:51
would be helpful because it would give
05:53
us a clearer understanding of that
05:55
distinction of immediate versus
05:57
underlying cause of death but there’s I
05:59
mean I spoke to a GP personally who said
06:04
that you know he’s he was going around
06:06
certifying deaths in old people’s homes
06:09
and you know he was encouraged to put
06:13
over 19 on the desk if he wasn’t sure so
06:16
you have a you know an elderly person
06:17
with dementia who has some sniffles or
06:20
whatever and then they say well maybe
06:23
that was Toby 19 should we be worried
06:26
that even beyond this technical problem
06:28
that could be fixed there might be
06:30
inflation of carbon 19 numbers more
06:33
generally
06:34
yeah and that’s important to remember
06:36
that when we have influenza outbreaks we
06:38
never test people and we put in
06:40
on the death certificate because of what
06:42
circulating at the time and there tends
06:45
to be that conflation and an inflation
06:47
in effect because you the sort of biases
06:50
in you’d is that at that moment in time
06:51
the most likely cause if you have a
06:54
fever is COBIT that’s one of the issues
06:56
that you see the deaths in ons are
06:59
higher than both sets of data in NHS
07:02
England in public health England and
07:04
it’s contributing to some of more of the
07:07
confusion if you like and I think this
07:09
is incredibly important because what we
07:12
follow then is the excess death and
07:15
that’s the the most accurate piece of
07:17
information that can tell you what’s
07:19
going on at that moment in time he can’t
07:22
tell you though what them excess deaths
07:24
are actually caused by and we’re
07:27
starting to see reports that some of
07:29
them in the home setting for instance
07:31
are a consequence of people not coming
07:33
forward with heart attacks and that’s
07:35
really important to understand that bit
07:37
of information because what we know now
07:40
is in the last three weeks the excess
07:42
deaths are in are actually come down
07:44
below the average for three weeks in a
07:46
row and now that the hospitals are more
07:48
open as well so people who go into
07:51
hospital again yeah but not in the same
07:54
numbers we’re still not back to normal
07:55
we’ve still got this reticence and I
07:58
think it’s across the board as we look
08:00
at like going back to the office people
08:02
are still concerned still worried and
08:04
anxious about going to hospital because
08:06
they perceive they’re gonna get COBIT
08:09
when they go in to a hospital
08:10
that’s why it’s incredibly important we
08:13
have accurate figures that we can
08:15
portray to the public to say this is
08:17
exactly what’s going on today as opposed
08:19
to what we see is sometimes in the media
08:22
it’s going up there are hundreds of
08:23
deaths things are getting worse when we
08:25
look at the data and say actually that’s
08:27
inaccurate and we don’t use the public
08:30
health England data to understand the
08:32
trends it’s in what’s difficult about
08:34
this and I think why people will
08:36
probably be arguing about this for years
08:38
to come is that so many of these
08:40
individuals are very elderly and many of
08:44
them are in care homes and you know a
08:46
huge push to test everybody in care
08:48
homes but people also die all the time
08:51
in care homes for obvious reasons and
08:54
and when you’re dealing with such old
08:56
people you know it’s gonna always be
08:59
hard as knit to sort of disaggregate if
09:01
someone tests positive for Kobe 19 but
09:03
then sadly at the end of their life what
09:06
actually is the cause of death so
09:09
there’s an important distinction the
09:10
difference between life lost and life
09:13
years lost and one of the issues we’ll
09:15
be watching very closely over the next
09:17
six months is to watch how many people
09:21
would have actually died in the next six
09:23
months and so their deaths were brought
09:25
forward by kovat because they were
09:28
vulnerable but actually their deaths was
09:30
their lives were shortened by a few
09:31
months and that’s where the excess death
09:34
calculations really matter and if we see
09:37
significantly trendy number now for the
09:39
next three to four months
09:40
we’ll start to come forward with
09:42
information that suggests yes exactly
09:44
there was a group of vulnerable people
09:47
who were actually any respiratory
09:50
infection could have been influenza
09:51
could have been covered would actually
09:55
have shortened their life by a few
09:56
months what’s interesting and important
09:59
about this is that actually it’s looking
10:01
like about 50% of the deaths are related
10:04
to care home ties are directly in care
10:06
homes are those people are admitted to
10:07
hospital in places like Spain it’s as
10:10
high as 70 percent so the number one
10:12
mitigation strategy should be to shield
10:16
Cairns and they should be real thoughts
10:18
about how to do that really well because
10:21
yes you’ve just nailed it on that
10:22
they’re the most vulnerable to this
10:25
infection and it seems to be once you’re
10:27
over 85 it really is has a big impact on
10:32
mortality in that age group when I read
10:35
your analysis yesterday the reason my
10:38
heart sank because I thought if we can’t
10:40
even trust the the death numbers that
10:43
are coming out from public health
10:44
England every day you know it’s gonna
10:47
make conspiracy theorists of us all in
10:49
the end because you know it seems like
10:52
the tilt on all of these things if there
10:54
is room for error and of course it’s
10:56
difficult it always seems to tilt in the
10:59
direction of making the numbers look
11:02
worse rather than look better we haven’t
11:04
had these kind of ambiguities in the
11:06
other direction
11:07
so much and it’s it gives the impression
11:10
that the kind of the structures are all
11:12
pointing in one way I mean do you feel
11:14
that yeah so that’s an important issue
11:16
that’s going on in the media the sort of
11:18
you’re always there about the
11:19
catastrophe and the consequences of that
11:22
one of the things we notice is when you
11:24
don’t hear about something there’s
11:25
probably good news happening so when
11:28
Sweden looks worse you hear about it but
11:30
when it’s not so bad like now you never
11:33
see it in the media but I think there
11:35
are two issues that are really important
11:37
one is out of this pandemic I think we
11:39
need a joined up thinking about how
11:41
statistics are produced it was in this
11:44
country and across the UK covered the
11:47
devolved nations and that will help us
11:48
understand what’s going on particularly
11:51
this issue about immediate causes of
11:53
death we’re having the same problem with
11:56
testing as well we’re never quite sure
11:58
if somebody’s been admitted where’s
12:00
Kovac or got Kovac while they were in
12:03
hospital and we can’t understand that
12:05
big piece of data so that’s one the
12:07
second is it’s incredibly difficult to
12:09
compare across countries as well because
12:12
deaths are all collected and collated in
12:14
different ways simple things like the
12:16
age reporting is different for instance
12:19
we wanted to compare England for
12:21
instance death rates to Island and
12:23
Island has a much younger population
12:26
than England so you can you can look at
12:28
that and think how can we compare what
12:30
would have happened in Ireland if we’d
12:31
have had an age structure like that we
12:33
couldn’t do it because they report their
12:35
age bands differently to England so we
12:37
need joined up thinking internationally
12:39
as well so we can make clearer
12:41
comparisons so at the moment you see
12:43
this problem of we’re doing worse or
12:45
better than this country but what
12:47
actually we look at and go it’s really
12:49
difficult to tell that given the way the
12:51
current reporting happens and and you
12:54
conclude this then that central number
12:57
which is the basis of all international
12:59
comparisons it’s the basis of a whole
13:01
our whole understanding of this which is
13:02
how many people have died we can’t
13:05
necessarily have confidence in because
13:07
the public health in the number we now
13:09
know seems to be including people who
13:11
have recovered the ons number which is
13:14
comes out of death certificates seems to
13:16
be including some people who may not
13:18
have had it
13:19
but certainly people who had might have
13:21
died of other things and just had it at
13:23
the time and then if we look at the
13:25
excess death number that includes people
13:27
who might have died for other reasons
13:28
like not going to hospital so we don’t
13:32
you know is it fair to say that the
13:33
actual common 19 death number is likely
13:36
to be lower than any of those official
13:39
numbers yeah so so there are two things
13:42
one thing we follow which has been
13:44
incredibly helpful is what NHS England
13:46
have done and in doing it they made a
13:48
change to report the date of the death
13:51
and when we look at that in hospitals
13:53
now we’re looking at about an average of
13:55
about 19 deaths per that day and that’s
13:58
slightly coming down because I’m
13:59
referring back to about five days ago
14:01
when it’s stabilized so each day they
14:03
report but they could go back five days
14:06
in time so NHS England is the best
14:08
dataset to understand the trend they
14:11
account for 60% of the death ons is
14:13
useful but the problem with OH&S is we
14:15
have to go back about 10 days and wait
14:17
till Tuesday for it to report so we
14:20
can’t accurately tell you what’s going
14:21
on today but we can tell you what was
14:23
going on 10 days ago so those are the
14:25
two data sets wheels and I tend to not
14:29
pay attention to phe because of this
14:31
problem with the way the data reported
14:34
and the variation from day to day and
14:36
and the ones that you do use you feel
14:39
confident in at least are a reasonable
14:42
reflection of Toby 19 deaths yeah and
14:46
what about yeah one of my jobs is and
14:48
our team’s job is to tell you whether
14:49
it’s coming down or very sensitively
14:52
when do we think it’s going up as soon
14:55
as it’s going up we would inform and say
14:57
look we think the trends going in the
14:59
opposite direction it is still coming
15:01
down it is slowed slightly but he’s
15:04
still going in the right direction and
15:05
that’s an important distinction because
15:07
we also want to be able to say oh maybe
15:10
there’s a second phase of an outbreak
15:11
it’s changing in the opposite direction
15:13
and that’s really important to have that
15:16
distinction within our ability to
15:18
understand the trends Carl Hannigan
15:21
thank you so much for paying close
15:23
attention to these things and there’s
15:25
more of you coming up later all right
15:26
thank you very much that was Professor
15:29
Hannigan just clearing up
15:31
what he’s
15:32
about the ambiguities around public
15:35
health England deaths so thanks that
15:37
you’ll be hearing more from him and Tom
15:39
Jefferson later today
15:48
you