Communicating Possible Harms and Benefits of Treatment and Lifestyle (MtG)

Communicating Possible Harms and Benefits of Treatment and Lifestyle (MtG)


Good morning thank you for joining us
for today’s Medicine Mind the Gap lecture featuring Professor David
Spiegelhalter. This Mind the Gap lecture series explores issues at the
intersection of research evidence and clinical practice areas in which
conventional wisdom may be contradicted by recent evidence from the role of
advocacy organizations in medical research and policy to off-label drug
use to the effectiveness of continuing medical education the seminar series
aims to engage the NIH community in thought-provoking discussions to
challenge what we think we know and to think critically about our role in
today’s research environment today’s lecture will focus on the increasing
attention being paid to presenting potential benefits and harms of
treatments and lifestyle choices in a balanced and transparent way Professor
Spiegelhalter will discuss some proposals for how this might be done
focusing on alternative ways in which numbers and graphics may be used and
emphasizing the role of interactive animations and videos recent research on
public preferences and understanding of different formats strongly suggests that
one size does not fit all and a range of alternative presentations may be
appropriate a professor Spiegelhalter is Winton professor of the Public
Understanding of Risk at the Cambridge at the University of Cambridge where he
is also a Senior Scientist in the Medical Research Council Biostatistics
Unit his background is in medical statistics particularly the use of
Bayesian methods in clinical trials health technology assessment and drug
safety he led the statistical team in the Bristol Royal Infirmary inquiry and
also gave evidence to the Shipman inquiry in his current post Professor
Spiegelhalter leads a small team that is attempting to improve the way in
which the quantitative aspects of risk and uncertainty are discussed in society
he was elected a fellow of the Royal Society in 2005 and awarded an Order of
the British Empire in 2006 for services to medical statistics
please join me in welcoming professor David Spiegelhalter thank you hello
thanks very much for for coming along and I see the schedule for two hours I’m
not going to drone on for two hours though that’s really expecting 200 I
will then yeah well if you want I can I can keep going until you drop but I I’ll
try not sick and there also I’m very like happy to take interruptions or
comments as I go along which I think helps the the flow of it so let’s see if
I can get this moving which are strange this usually works and yeah I got this
odd title Wynton professor for the public understanding of this Wynton is a
hedge fund Wynton Capital Management and they um they put up the money for my
post to endow my post which is quite nice to them oh this isn’t gonna work
excuse me Asha might machine has clearly completely seized up yeah that’s good
it’s uncie’s right so a Wynton Capital Management is is a hedge fund that we we
go to scoop understanding uncertainty and try to keep their own risk out of
what I do because it’s got such negative connotations particularly when you go
into schools they’re thinking to do with health and safety and stuff like that so
and and so we obviously talked a lot about uncertainty in sports and cardiac
kids interested in in risk and probability I’ve got a YouTube video
professor risk you can’t quite see that there’s two of me there one of them
taking my clothes off and and it’s all about 60,000 hits and it’s to where it’s
not quite up to lady gaga but it’s not bad he listened and and we go into
schools to do a lot of stuff on chance trying to improve the low probability is
taught in school so I feel very strongly that’s really not David though well in
the moment and try to engage with probability in the real world and I
working with the B receive interest on this big risk test which is the biggest
ever the psychological test of people who are
scattered seeds and risk appetites having more than hundred thousand
participants on that and I’m working with a for example working the Met
Office and alternative ways to communicate uncertainty about weather
forecasts and in the u.s. you get probabilistic precipitation measures
already but in the UK this is not not done routinely and I also try to take
some risks myself some of you may recognize those so you either if you an
eight year old child of the mind of an eight year old child you may recognize
this as did people recognize this who recognizes this as the wipeout course
and for the US got his own course this is a idiotic game showing which you get
beaten to a pulp by large plastic objects and fallen a lot of mud and
water and this is an Argentine because the rest of the world feels films films
their versions are just outside Buenos Aires so I flew to Argentina for this
just before Christmas I did quite well so there but I’m not going to show you
that and I could show you the video there I’m not going to know though this
is serious talk I’m not going to show video me being knocked into mud by large
objects but I will if it cost later if you ask what I can talk about is using
communicating harms and benefits and risks and uncertainties and therefore
out using words using pictures using numbers using graphics and I’ve gone
straight about some the alternative metaphors maybe some alternative ways to
think about this which might sort of grasp the attention of the public and
physicians as well and I’m particularly interested in communicating small
everyday risks the things we do the way we get around our sports or activities
what we eat or be drink et cetera what that means to our lives because this is
what is in the news all the time and like to finish they only want to make
sure I’ve got some timeless to talk about the deeper uncertainties about
risk what I’m you just don’t know or we won’t grow we might have some
statistical measure of uncertainty but maybe we’ve got some deeper
uncertainties what do we do then okay there’s a paper in science last year we
had about about many of these things which is downloadable okay
I’d like to refer first of all to for this document from really from the
Commission by the FDA is by the risk communication group in the FDA and
committee which is absolutely wonderful so it would fail your valuable downline
and they’re online and it’s um yeah it’s really good you know that it covers a
lot of things I’m talking about fully reference very well written very nicely
and really led by boat fish off as a team but two H’s which I think I’ve
spelt wrong and here this is a quote from BIOS there’s the first step in any
communication is define risk costs and benefits in ways that allow people to
construct stable informed preferences the emphasis in this document is very
much about the ethical basis the ethical need to communicate harms and benefits
in a transparent way there’s an absolutely need demand for whatever the
consequences whatever this might do people’s behavior there’s an ethical
need to do that if we believe in shared care and an informed decision making so
doing so requires knowing what is at stake
crafting suitable measures and testing the results see how well it serves users
need so this is empirical science as well as an ethical to drive behind it
that we can develop things and test how how well they work both fish-off spelt
wrong two H’s and so the issue is that either the traditional in a really
old-fashioned scientific view will just be a work at the numbers and tell people
the numbers on land then they’ll be clear and they’ll be
formed well and the point is that people’s responses to risks
uncertainties times of crisis and dominated by factors that are not to do
with the numbers I mean there’s a lot of research over the decades by people like
Paul slovic and others just shown that those dominates by our ideas of their
emotions personality their experiences themselves and their families their
feeling of control or what the risk has been imposed upon
them just just compare you know the radiation risk from CT scans compared
with their attitudes to Fukushima and you’ll see this being you know appearing
and use all the time cultural attitudes to do with what your
particular tribe and I mean Metin modern sense feel around you trust of authority
lack of it familiarity redness of the hazard innocence of victims social norms
etc properties are largely ignored I mean the study of this is fascinating
because it certainly is a very transient the very local and in the it depends
where you are and when you are they if you look at the original research done
here in the late 70s people are terrified of microwave ovens it was at
the top of the phase right up there with it with radiation leaks of nuclear power
microwave ovens but people actually mind too much by my I still treat them with
great suspicion but in mind about them anymore and different things are
nanotechnology and you came nobody expresses any concern whatsoever in
France are actually obsessed by it and they meetings are broken up we were
discussing nanotechnology attitudes of vaccines except from very enormous tea
from country to country so and strong cultural attitudes in these things so
but I’m gonna make some assumptions what I’m going to talk about and assume that
you know following Bo’s comments that there’s an obligation for transparent
communication is telling people reassuring because they rely them token
on trust of the of the young of the authority and we need to cover the
uncertainty okay and we need to acknowledge patients
concerns we need to listen to to the audience and take their concerns
now that whether we think they’re rational or not to be it does serious
things now using words for example they you
know the standard information leaflets you get talk about something might
happen there’s a small chance something might happen etc etc now these endless
researchers showing these were interpreted completely and consistently
from audience to audience there’s be attempts to try to impose some order on
it and I’m gonna make a lot of today I’m going to talk about side medicine quite
a lot and looking what other groups do example the IPCC the internet
Intergovernmental Panel for climate change and spent years trying to
regulate or or impose some consistency in the way that uncertainty is discussed
in this highly controversial area very political area and one of the things
they do say is that if we’re going to use any words like that likely unlikely
very virtually certain that something will happen
it’s got to be defined in terms of their probability so and so they when it says
you know after the morphic you know when global warming is very likely to be
caused by human activities what they say that we mean ninety did they reckon
there’s a 90% probability that has been caused by human activity so and so that
I think this is a step along the way there’s been various research and the
evenness doesn’t really help too much in people’s understanding of these words or
the consistency of the use of these words but as a step in the right
direction there still was enough and the problem is that you know is this is the
hope numerators the public this is Kings Cross station London which
now has a platform 0 and which is caused endless confusion they keep on having to
announce you know passengers are advised that platform 0 is situated next to
platform 1 the people can’t find it this is but on the site and if they extend
the station need more in that direction they’re gonna have to go into minus
numbers and then for the mathematicians among you the undergoing platforms that
have to be given complex numbers and these all vary
very prominent numbers and this is one of my favourite bit of numbers from
Tesco’s my local supermarket one pound each by 2 for 3 pounds for pomegranates
which is which Oh deep philosophical if what do you want to pomegranates you
know do you have to go through the checkout twice or so it is problems with
numeracy and a lot of psychological research has shown importance of
numeracy in people’s understanding and I talked about doctors they said we and I
see you got your patient information leaflets in 35 languages or something
only for one level of numeracy and yet that stratifies the population enormous
lis terms of their understanding their willingness to engage in factual matters
so people are now searching doing a lot of work on understanding of numbers is
the no recent paper in archives internal medicine by its group based in Berlin
who are collaborating with and this committee and there’s a simple one
either what you which of the following numbers represents the biggest risk of
getting a disease one in 100 100 thousand one in 10 now I’m not going to
test you it’s all right I’m not gonna ask the show of hands and you should
read work is in a population turnover in the population and your telephone survey
in the US how many people got that wrong you think in a random popped it and a
thousand people in the US randomly selected and token so how many people
didn’t got that wrong two-thirds got it wrong well that’s a bit so quite that
it’s in fact my quarter got it wrong in the years and slaying them in Germany
and we just asked that question of 80,000 people in the US the UK and then
the point is that a sizeable minority this sort of language is odds language
which is used quite a lot in in the UK for things like risk good for homeless
and tsys and things that they’re being found again and again to be a very
confusing very poor way to communicate risks than 1 in 9 so I think that one
number maybe even two but if you feel lots that with them they’re confusing
because the biggest number is associated with the smallest risk of these
inversion all the time so the jury considered an
extremely poor way to communicate risk and that I get so fed up in my turf I
live in a flood zone and my flood risks are all communicating this way because
it’s the technical way they do things in terms of u-turn periods of 200 years or
100 years and that the technical work is then dumped on the public and with some
assumption that they’ll understand it but they don’t the other the classical
thing is the importance of this faction is in this experiment has been repeated
again and again and I mean as mathematicians we sort of think oh well
a probability is 0.25 or 1 out of 4 or 25 out of 100 is they’re all the same
you know they’re not at all the same those are all totally different those
are completely different numbers to to many people
so this is a standard study done be repeated many times the odds of general
population your two-bit bowls where you go this has got ten beans
one of which is red and it’s got a hundred beans eight of which are red
you only get $1 if you if you pick a red beam which bold you choose your bean
from this is this do choosing this bowl or that bowl now again I’m not going to
test you on this because it’s still quite early in the morning for you
anyway not for me and so but they’re interesting how many people think choose
from in the pod general public in this experiment how many people do you think
chose from that side which one you think most people chose from the outside or
outside yeah then majorities chose from that side because there’s more chances
of winning makes this completely logical is more chances of winning on the right
hand side this is good denominator mingle act or ratio biases they look and
see ratio biases the fact that people tend
to look at the numerators the number times that and the actual denominator it
says to ignore so a risk of 20 as if a hundred looks a lot bigger than two out
of ten and not as big as 200 out of a thousand yet so
oh yeah yeah yeah yeah yeah no they ignore the pictures ignore the fear
exactly I’ll go with it he’s got it wrong you’ve
got the wrong number of beans oh yeah yeah no no no they’re really really a
mess yeah yeah you can’t even count them after it’s okay yeah I got no I’m
skipping over that now yeah those are a ratio but a very reproducible finding
that people are strongly influenced by the size of the numbers denominator
detectors when you’ve taken absolutely no notice the denominator I think or who
does that who does that the newspapers do that every single day because the
story is in the numerator not in the denominator the denominator is how many
times something didn’t happen no child knocked down on the way to school we
don’t you know which is a terrible thing I’m an incredibly rare now in the UK and
so you don’t hear how many millions of children went to school all year without
being knocked down don’t hear the denominator and Cass Sunstein is really
is written a lot about denominator neglect particularly in relation to
terrorism where people only focus on on the event and not on the number of times
it has it didn’t happen essentially so narratives only like numerators in use I
always remember those newspaper stories and narratives mainly look at numerators
they never tell you how many times something didn’t happen so this is you
know stand a bit of media reporting which we have to watch out for because
we can be easily tempted into that okay the other that standard problem is
showing familiar with is the frequent use of relative risks to communicate the
benefits or of interventions so this is bathers in this area at a greater than
average risk of shark attack and so you know well how you average average for
what you know average for England you know it’s really not very useful piece
of information to notice things greater than average but we get that all the
time I made my examples I’m sure I could have
spent five spend a couple of days I could have found examples
u.s. newspapers of a similar ilk but in UK oh obsessed with these what I tend to
think is often rather second-rate epidemiological studies which constantly
find the increased risks got something by doing something and cats causing
cancer as always my new one but the classic one which you get so this is
daily fry up fried breakfast increases the chance of pancreatic cancer while
mr. v so new some newspaper headlines there are your daily bacon sandwich risk
of paint a lifetime risk of pancreatic cancer increase by fifth no I will
believe the science of the myelin question the science it’s probably quite
questionable but I won’t question the science the moment let’s pretend that’s
really true what does it actually mean ah you can’t quite see there’s five
people up there the screen is slightly off it so that says of four hundred
people eating that for breakfast every day in their lifetime five will get
pancreatic cancer is about what an 80 lifetime risk of pancreatic cancer
terrible disease to get what an 80 so if but if all if you if forged other
people whoíve at every day that so many people get pancreatic cancer so you see
the difference that’s going from that to the yeah so that’s the effect there in
the absolute risk is one in four hundred absolute and absolutely increase in
absolute risk so the number needed to not number needed to treat the number
needed to eat the number needed to eat a bacon sandwich for every day that our
life is four hundred and order gets one extra and pancreatic cancer may be put
in those terms that certainly stops pursuing quite so impressive now you may
think oh that’s just ridiculous and now that this happens all the time
it’s an in particular has been identified this either mismatched
framing when benefits are reported in relative risks and harms reported
adverse drug you add vers reactions almost always reporting it in in
absolute risks there’s a less than one enough that has a chance of something
wrong as the standard reporting for absolute risk so this is from a cat an
information leaflet and for a sigmoidoscopy
in the okay reduce the risk of death 59% for
cancers and plug this is rate of perforations so that’s called mismatch
framing you’re defining the improvement as a relative risk which is known from a
lot of experiments in a Cochrane Collaboration review to give the biggest
impression of magnitude and use absolute risks for the hunt now that is what I
can see what this talk is about that this is absolutely disgraceful and
really should not be shouldn’t be allowed shouldn’t be allowed and
absolute deserves people who says that this is not allowed and the Association
for the British pharmaceutical industry code of practice in 2011 has banned this
that practice it says I’ve read all that it says you’ve got to the reader needs
to know the relative risks relative the absolute risk relative risk should never
be referred to without those who referring to the absolute risk doesn’t
say you can’t use relative risk says you must refer to absolute lists when
communicating about a drug now those in the UK we don’t have direct to consumer
advertising so when you do and adverts like that come up but is this Britain oh
okay this is known as the British ad so it doesn’t need to follow the code but
as supposing the wait does follow the code because the relative risks the 36
percent is up there in great big letters but a much smaller fault down here if
you look carefully it says what this means is that 3 percent of patients
taking a sugar pill or placebo had a heart attack head of 2 percent taking
lipitor so there’s 36 reduction I think is only over about 5 years as a
reduction from 3 percent to 2 percent so again this will absolutely adduction of
1 percent and means a number needed to treat is a
hundred people taking the drug taking the statin every day for 5 years to
prevent one heart attack now it’s not that that form of communication is never
used in marketing because it’s the one that makes things look least effective
so this in a way obeys the guidelines but still the relative risk as BAM is
that looking at you so what does it is to do is with this classic psychological
thing of framing you know Danny Kahneman with his recent book I’m thinking fast
and slow talk slow though the framing and how people
are manipulated by the framing of messages and and we see this again all
the time this is just one I took a picture up for not a very good picture
on the London Underground which said 99% of young Londoners do not commit serious
youth violence can that’s called a positive frame now that sounds great
anything hang on one percent of young Londoners do commit serious youth
violence about a million young Londoners oh my god there’s 10,000 maniacs running
in London this is this is terrifying but you’re not supposed to do the sums
you’re supposed to just accept that positive message and I want to show a
really the best example I’ve seen of this of framing now this was a paper
from I can’t what is the u.s. University in Nature Genetics 2009 one of those
tedious papers ever been published and it found 10% of people had a genetic
variants that reduce the risk of high blood pressure worthy but dull paper
until a very skilled press officer got hold of it and turned it into a story in
which 9 in 10 people carry gene which increases the chance of high blood
pressure same story didn’t change in one little
bit just changing it from positive to a negative frame changing the baseline
category now when we’re in the world you know here introduced by the coverage
great success for the press officer absolutely nonsensical reporting so it
just watch out for these it’s very clever and the press officers are the
ones who are very good at this that was a particularly skillful I thought I
think she should have been or she or he should have got an award for that and
for the most manipulative bit of scientific reporting ever been seen so
we’ve got to watch out for this kind of thing so how are you trying to avoid
that well the people who have been really clear and trying to avoid that to
think of a group of people like can steal ocean Lisa Schwartz they’re
developed ideas of the drug facts box which tries to get around this ideas of
frame you whether it’s called basically the little phrase is uniform
porting and benefits and harms no relative risks everything referred to
absolutely in an unbiased way about whether it does you good or whether it’s
done harm reporting there’s still psychological argument I can do it in
the quite vehement argument within the psychological community about whether
it’s better to say six percent or six and a hundred as again as mathematicians
we might think oh yeah it’s the poor they’re both the same well there’s a lot
of discussion about which is the most effective communication device and so
they overdose actually doing both at the moment saying that you know for the drug
if people give me the placebo people given the drug how many will have an
unpleasant taste in their mouth twenty six percent with the drug six percent
birth of deceiver so well we’ve got some you know as we’ll see later one of the
things is these some of these side effects might be quite minor some of the
major outcomes what MIT might buy major so generally all stratify about the
severity of the outcomes but the basic thing is that harms and benefits were
reported in creating balanced way now the interesting thing about that though
they managed that could manage to get the FDA risk communication Advisory
Committee that book I showed earlier is really plugging the drug facts box as a
ways to communicate benefits enhance and I think you know this is this is for
prescription drugs but of course it could be applied to it forms any
intervention that you make and they managed to get it in the act so a
mandating that the FDA should at least evaluate the drug Facts box it’s not yet
mandated that he will be used and be compulsory in them in prescriptions but
and it’s you know that it’s mandated that the updation least evaluate it and
it has been you know empirical evaluation showing the benefits of using
it so I think I said you know one of them a really great innovation okay
that’s the only miracle so you want to talk about suppose getting numbers right
and graphics I love graphics now part of the things things that out of that
framing literature and the relative risk has come this idea that we need to be
able to make part-to-whole comparisons you can’t just look at what’s bad you
need to look at what will happen you know the fate of the Casa bit of
framing is that in the u.s. they report you know mortality rates for cardiac
surgery two percent mortality rate in the UK we report survival rates 98
percent survival rates for cardiac so it sounds much better and submit they ask
what that’s is pure framing device so in really if you say there’s 2 percent some
mortality you should say 98% survivorship in both numbers so the best
way to do that one way to do is using a pie chart that’s the worst pie chart
I’ve ever seen from Fox News last year and any one spot the error in the pie
chart but I like I do this with schools classes who are learning about pie
charts and they they just can’t believe it yet no that’s one of that’s a great
pie chart that’s and so the pie charts do show part-to-whole comparisons not
not like that and but other ways part to hook me is to use a bar chart or
something that just represents say a hundred people this is the one used in
the adjuvant decision support system that’s available online in fact one of
my colleagues were a wife and got handed this output for her and in this format
and we’re looking at what would happen in the decision good about having
chemotherapy or not and so rather than just saying well you’re the number
people died within five years will go down from 12 to 7 in other words a
reduction of 512 which sounds rather good is a relative risk when you put it
in the totality of things you can say the benefit of the chemotherapy is is
that much there’s five people out of 100 are benefiting from it in terms of their
survival so new one in 20 the people taking it are benefiting from it and
there’s the side effects yeah yep yep now the color and design layout
font everything is very important in this in
color has got well-known associations of red happiness you know an association
with some with their using it here the red is the worst outcome excetera so we
tend to be the sort of colors that people use so the color is important and
that’s something that again is empirically testable you know in any
design I will be talking about that particular today but everything right
and use of sensor the language is important as well it isn’t just some
other ideas just you know how do ideas for communicating uncertainty people
used to tag tags about frequency of words in this one the size of the font
represents the chance of this person winning rim buildin and according to the
bookmaker in the weeks before we moved in the last year so this was you know
this is the favorite and you see the chance now the probably that so the eye
is drawn to the ones most likely to win over that is the people with long names
tend to feature if I was on there I’d look like I was gonna win Wimbledon
there’s got a stupid long name so an alternative ways to use this either this
you I’m not sure people see these tree maps which again used to show
decomposition in tumult of into small bits they often use for budget tree
things to show exploded clever government expenditure how it breaks up
and you can put any list of numbers in into on to website many eyes or anything
really produce one of these for you but this is one where the area represents
the chance this person winning Wimbledon so again quite people like what people
think of that’s quite an effective way of showing out so it’s when you’ve got
lots of different things that can happen the known probabilities people like AI
yeah like it about that is that like that or not yeah no people and it’s very
qualitative that you can t get difficult to make comparisons I think okay so the
way probably most common way now being recommended is icon arrays or Pictou
pictographs or something like that so this is I’ve done 23andme you know spit
into the tube and send the stuff off and they you know
northern her website and find out what your genetic risks are according to the
markers that they may use so my lifetime risk of type 2 diabetes according to
this out of a hundred people like me with my share my genotype that could
work or during these few markers that they measure and hundred people 26 on
average would would get diabetes and 31 or people like me will get diabetes
between 20 and 79 because these aren’t really my risks because I’m 58 and
haven’t got it yet and this only looks with my genetic stereo a few genetic
markers and doesn’t consider all my other lifestyle factors
so it’s not my risk but it gives an image of what what my genes were saying
so this deliberately there it’s a hundred people you should color we’re
getting a the part-to-whole comparison we’re showing the 400 not just the ones
we’re going to get it not quoting relative risks there if they do give
relative risks and blocking the icons and be in experiments to show the
effects of scattering the icons are showing example minute and which which
actually has some predictable consequences of impure increasing the
impression of unpredictability they’re scattered but decreasing the ability to
compare things because people can’t count them on the scale so for example
here’s doing me as an example my eyelid my GP got me through a risk calculator
framing and noting yours you tell me a twenty-foot 12 percent chance of heart
attack or stroke in the next 10 years and if I took the statins every day I
can reduce it by about 33 percent by right or third okay so that’s I’m told
what I want to do well okay one way to think about that is to think of these as
a hundred people like me so those I’m to be blind me and if none of them take
their statins twelve of them are going to have a heart attack or stroke over
the next ten years but if they all take their statins
like good boys every day for ten years and four of them won’t so these are the
ones are going to have their heart attack these are gonna have them anyway
these will not have it anyway and force them or have it prevented so that’s what
that information the DGP my doctor was telling me could be communicated
communicate way that it doesn’t look great yeah 7 1
in 25 enters number needed to treat again 25 people gonna have to be like me
would have to be treated 10 years taking tablets every day with possible side
effects for to prevent and one heart attack or stroke it doesn’t look great
in that format we’ll come back to that actually in a minute I’d like to I’m
going to come back to that representation and think of some other
ideas about it and just to quickly look over what in some other non-medical
areas how people are in other words the Bank of England for example is
communicates its uncertainty about what will happen in the economy with these
fan charts we’re getting into the naree idea of blurring to represent
uncertainty using shading of color so they produce an interval like that and
that’s a 90 percent interval in the middle that’s a 80% 70% 60% in the
middle like that so that’s what the Bank of England it does they don’t produce
the central forecasts either but the governor the Bank of England Mervyn King
refuses to put a center of forecasting because people just say oh that’s the
bank room there’s forecasts so just as I do in view of shading no central
forecasts because that’s what actually happened this is because I picked
November 2007 so and then down there somewhere so what you know just shows
the luck of the draw that’s actually they’re unapologetic about that because
this was only a 90 percent interval it’s not very clear from the graphic that
there’s actually five percent probability somewhere down here
somewhere and so I think that could be improved
and they for how good morning in the States the standard technique until
recently was what’s called a cone of uncertainty where they produce a white
air blur there where this is a two-to-one chance the horican
will go within the white area rather than outside so it’s the two-thirds
chance and there where the black line showing the most likely path now that’s
been quite as strongly which there’s been no criticism that there’s published
runs about the fact that the central path again
encourages people to almost fixate on that and so you know people live there
and they don’t evacuate and then house gets flattened and they complain well
actually they shouldn’t complain in recent when Hurricane Irene they’ve
dropped the central path I just using the cone responded to that you as
criticisms of having the sensual area in but the one thing I found fascinating
for Hurricane Irene is I think it was the first time that NBC used the
spaghetti plot spaghetti pot is a I don’t know there are like no but you
think this one it just puts different computer runs and traces on the plot
with the idea of possible path now this is changing the metaphor now is changing
the metaphor into other possible futures these are the things that might happen
we don’t know one which one will happen but one of these might happen different
computer runs now is amazing they didn’t do any great psychological testing they
must have just thought oh this is a cool idea we’ll just stick it on national
news and then in see this is in the notice and then see that yeah what did
you think yeah yeah exactly I thought this is completely wonderful they could
spaghetti up what I used within the trade of weather forecasting and things
of that severe event ports yeah one old one yes all right sorry yeah
it’s trouble with them and interacting with hoods yes so I guess well I’m just
asking what do people think of this if they noticed it and what do people think
of it can I ask you to say sir well I don’t know do you have to go up
there sorry yeah sure I was just gonna say it’s what’s really interesting about
some of these it’s just that some of the models will be very different than the
rest yeah yeah you guys raised all the outlaws but again I think that’s you
know again I think people can take that kind of thing oh yeah there’s not one
well that’s possible but I don’t even believe that we believe the bulk of them
so he’s a very powerful metaphor now we could be learn something from that
metaphor from this instead of this you know rather faceless hundred people like
me we could change the metaphor to possible futures for me some personalize
it more so the metaphor being a hundred ways things might turn out for me now
you could you can slightly joke this by you know putting in there’s a hundred
knees and the future and so there’s a hundred ways things might turn out for
me oh yeah I don’t know I’m grinning year this is 100 meters taking my
statins every day across I know I’m Gooding because I haven’t had a heart
attack or stroke in the next in the last ten years but I didn’t know it’s because
I was taking the statins it’s more like far more likely that I wouldn’t find it
anyway yeah if I don’t have a heart attack or stroke there’s only a four in
this many chance that it was because of the statins I didn’t have it most likely
I wouldn’t apply there anyways all the equity waste of time but never mind
and what this could all also shows that you know maybe that as a representative
to scatter the icons gives a much better impression of the unpredictability you
know one of these will happen to me if I take my statins
so throwing a dart where they’ve been a hit and don’t know which so these are
trying to images of the trying to promote the idea of unpredictability and
of the uncle stability is to do with we just don’t
know which of the possible futures might occur so these are quite powerful ideas
and in terms of iconography and we’ve been strongly influenced by this pre
will work of Otto Neurath and ger towns this idea of isotypes
which you can think of stacked icon arrays the identical is seen these are
sort of children’s be all children’s books used to use this quite a lot and
these were you know born out of the Bauhaus movement and is part of pre-war
and socialist Vienna to communicate the means of production everything that then
the ownership of means production as well to the to the masses and they use
this idea of colored or and blocked up little icons of people so this is
employment of women in different countries in 1930 and we’ve been quite
strongly influenced by this so just using little people and it’s used in
newspapers a lot as well so let’s look at how we might communicate the harms
and benefits all statins I’m interested in that I’m a moderate risk person now
I’m not necessarily necessarily saying as data is perfect this is just from one
study or the general practice database in the UK a couple years ago so I’m not
saying these numbers are fine I will talk later about how we do with
uncertainty about these numbers but that’s really says out of a thousand
people like me taking the statins over these are the people taking their
statins and these people not taking the statins what are the negative events
that are going to happen over those five years well this is many heart attacks or
strokes rather a lot more without statins and but it looks like more
cataracts now how can we make comparative the crucial thing we have to
do is take the small story from the biggest and when we do that we can see
what they know the risk balance is from taking the statin so a thousand people
taking the statins will have this many extra of these side effects but they
will save this many art attacks or strokes and soft your counsels on
average a thousand people okay so don’t take it well depends what
I though I feel about these I’ve colored these all the same as I’ve been I which
I’ve been criticized for it was made later on the music’s like pink for the
milder side effect though the big ones but that Chris Lee is the issue and how
do I weigh these possible outcomes can I sue me at the moment of these unknown
numbers summers and risk and risk balance okay so yeah and no J trig off
topic a bit by talking about and I’m going to come back to that because I
want to talk about when we don’t know the risks and we’re not sure about the
risks and let’s talk that now is the stuff I get asked about all the time
into a media performances and things which is about everyday risks they old
all new stuff that we do every day how do we communicate that and now I know
it’s use a few a couple of units for communicating this now this is so this
is much more um anyway yeah no it’s ten sitive experimental work but I’d like to
know what you feel about it there’s a there is a units a deadly risk that was
developed in Stanford by Ron Howard about thirty forty years ago called the
micro Mort so one-in-a-million chance of dying micromort now I feel that’s a
silly thing I find it really useful I know use it all the time right
everyday life and I’m every activity I’m looking at well how many micronauts is
that it’s quite a good measure because I’m using a UK comparison but it’s
probably similar in the u.s. in the UK in the England of those about fifty
million people about 50 people died non natural causes everyday accidents and
violence essentially so fifty people a day in the UK in the England Wales to
does one in a million so just living on average now it’s going
to be much higher for older people and men and things like that in different
groups will have much higher s and some people are very cautious and safe and
don’t have any died from accidents and violence very much at all but on average
of the whole population we’ve got about one micromort a day a deadly risk a
sudden death and that’s like flipping a coin 20 times in a row and it coming up
heads every time for that kind of chance we play out on average every day in our
lives so we sort of yeah well now are we prepared to take it we’re not we don’t
go mad about it well you know it’s pretty unlikely one-in-a-million chance
I’m sure you take one in a million chances quite a lot and so how can you
take one of the million chance in the UK at the moment again I don’t think it’s
too much different in years you get about 250 miles from micromort
on average on the road no chance how you drive them where you drive and be more
on motorways on freeways be less on otherwise and if you’re a young man
we’ve got a lot less but about 250 miles from micromort by cycling about 20 miles
and so you cycle etc and Cambridge is for the lot less and walking there 15
miles motorbike that’s six miles so you can see immediately how you can
translate everyday risk for example like miles and a motorbike so when I use
quite a lot because it’s quite a got a stronger bit quite a strong image on
people and then in terms of other things that we do scuba diving’s v quite quite
your bus leaf I’ve micromort so for going scuba diving a million chance why
don’t one or two hundred thousand dives end up with the diver dead and skiing is
about half Michael Moore today horse riding so it depends of venting or
competitive force ridings a lot more than that hanged items by they
parachuting us parachuting Association in three million jumps last year there
20 deaths so that’s seven micronauts of time it’s about seven Michael is running
a marathon in the US so these he’s fairly competitive sports that people
engaged and all round about the same back between five and ten micromort
seems to me what people are prepared to do as well the scene is quite an extreme
activity but one that was still fairly popular much above that he gets very
extreme BASE jumping leaping off buildings with parachutes
and things or off cliffs with the parachutes
one it’s released four hundred micro mort’s offer one of the cliffs in Norway
and and you have terrible lot of base jumpers have died so we can compare
these things you know going up Everest I mean one in every twenty five people who
summit along Everest someone’s died and mountains above going up above 8,000
meters until recently it was deputy filters and you know one in twenty five
chance of dynast about forty thousand micromoles this is a lot yeah forty
thousand micromoles which is about the same as two missions in Bomber Command
over Germany in the Second World War so it is a lot now our health ones um
giving birth from UK’s bed after giving birth in the on average in the u.s. is
about double that and that’s on average so a lot of people would be a lot less
so it’s a huge variability in the US Sweden’s about safest place in the
world’s about fifty UK no the cesarean is 170 micronauts general anesthetic is
about ten or five for an elective operation so we can make this go so the
general anesthetic in most general sense about sixty miles on a motorbike so you
having an anaesthetic and your doctor says oh it’s just is just like crossing
there you coulda sign the consent form this is just like crossing the road you
say no it’s not just like crossing the road it’s like walking about 150 miles
so it’s so we start making these comparisons okay so that’s one that this
is the but these are sort of risks that could kill you straight away actually
sort of drop you dead the ones that I think we’re probably interested here in
this institution at home are the chronic risks the ones are gonna niggle away in
you and shorten your life so for example here and this is madness of my sort of
restaurant this is the sort of food I I mean you probably don’t even know half
those things are that’s best not to know because these are all vile foods notice
no calories on the menu and they wouldn’t dare imagine if
they’re the perfect calories on and they’ll be fascinated by become a so
since I’ve been here being obsessive calorie counter it’s ridiculous
doesn’t make much difference nothing so until everything the warnings here I’d
let me Japanese melius I’m tucking into my you know raw tuna for my and my so
she me or something I look up and there’s this big notice saying eating
raw fish increases your risk of foodborne diseases oh no I don’t want to
be warned anyway how could we translate these into the risks so let’s take an
example this is there a recent study from Marchand red meat consumption out
of Harvard the this from the nurses stereo they put together the Nurses
Health Study in the health professionals follow-up study into one thing on red
meat now I don’t talk too much about the science of this and the I’d like to
Martin water-dwelling reporting of it how it can be reported this was reported
one night another portion was like in the u.s. in the UK it was absolutely
terrible and you know nobody could understand nobody’s it clue what they
were talking about so this sort of stuff Daily Mirror replacing mitla nuts
thought to reduce the risk of dying about 20% so read nuts one in five of us
need never die so that was good and Daily Express said this is good if
people cut dynamited red meat be ten percent of all deaths could be avoided amazing they could they no idea how to
translate the basic finding of the study the basic finding of the study was the
one point one to hear total mortality from one serving a day increases 1.13
for unprocessed red meat so a three eighths length of meat every day by that
much and iPhone worth of meat every day and increases your risk of death by
thirteen percent on average so that was one of the main findings of the study
now we could argue that and is buried though it is buried down there of that
number so we can see why people find it quite difficult to if but they should
have been able to interpret that a bit better and we could argue about the
study in about adjustment but I won’t do that man what does that mean mean how
can we commune take that in a way this is a relative
risk we know that we all have risks for the worst way to communicate risk how
can we do it better well and what does this mean for each of us
well we have to look at what the effect is on our lives that 13% extra risk and
we have to go back to life tables I’m working in when we go back to the life
tables we can work out a hazard ratio 1.13 applied throughout their adult life
because you’re eating this thing every day and eating the burger every day and
what’s it going to do to you and it turns out that for 30 year old for
example if he’s if he’s eating for lunch not eating is meat and not eating meat
if we believe this absolutely his life expectancy down another 40 years now his
mate who stuffs that learners goal every lunchtime if you can imagine it and if
we believe that he’s got a 30% extra risk every year of dying because of that
his life expectancy is actually reduced by one year it’s not a lot not reduced
by 30% that 13 percent increased risk very roughly fewer the hazard ratio of
about 1.1 about 10% it’s about one year of your life so you know that’s random
it says slight is for slightly more than one year it’s bit of rounding but it’s
about the rule of thumb I remember there never see a hazard ratio of 1.1 as a
year of your life as an advert as a chronic exposure over your lifetime and
now so life expectancy reduced by one year now that doesn’t sound too exciting
and there’s all the kinky Amos you know I’ve got this crate saying I’m not going
to give anything up for the sake of another year in a nursing home in
weston-super-mare now you probably don’t know where weston-super-mare is but you
don’t want to spend another year nobody wants to spend a year there it’s a real
dump but and so people say this isn’t very impressive is this going to
influence anybody this year in your life but you could also say that only a year
lost of your wife’s equivalent of losing 30 minutes of every day of your adult
life on average so that if you use 30 minutes off you’re better off your each
day you lose a year of your life so that means eating that 30 minutes off my life
for that burger every day every day so one way to think about that of course is
that by eating that burger I’m going towards mine death instead of going
towards my death for 24 today I’m going at 24 and a half pound
Rome Russia I’m aging fast and aging half an hour extra from eating this
burger so that’s all way of communicating it which is really is
quite a strong constructs quite a strong image knows lots of huge amounts of
assumptions it’s a metaphor because eating this we’ve no idea what effect
this will actually have on anybody and this is an averaging over people and
their lives and things like that now I’ve called this a micro life 30 minutes
off your life expectancy young people in their 20s or some here who have got a
life expectancy say between 55 and 60 years ahead of them by adult life ahead
of people have 55 60 years of our life roughly that’s a million half hours
million half houses 57 years so half an hour is a millionth of you out on life a
roughly so so you can have you especially uglier Microlife micro adult
life it could be so you could you know watch an episode of friends or something
around as a micro I’ve gotten to try that back never to be repeated so how
else can you lose micro lives well alcohol alcohols like a standard one
this was in the newspaper a couple of weeks ago in the UK optimal level
modeling the impact of changed an average consumption levels and this got
really badly reported and another one it’s started saying it’s about official
alcohol there’s nothing to live with official limits anyway so and we don’t
care about that but really badly reporting and nobody could understand
what it meant four and a half thousand lives again that’s an absolute risk
figure but it’s not put in any context what does it mean it said they could
save four-and-a-half thousand lives a year if everyone cut down their drink
and yeah it’s good it’s not least it’s not a relative risk but it it’s an
absolute risk this was the main part of the paper basically says at the moment
median alcohol consumption according to survey is 13 grams a day and totally
this is total deaths and this is deaths delayed or averted and so if people
increase their drinking the number of deaths would go up but if they decreased
median alcohol consumption 10 to 5 grams a day went up and then it started going
down again so it was a meta-analysis that took account of the of the
protective effect of low doses of alcohol on card
vascular diseases so as a peak so 5 gram the optimum amount of drinking and well
they didn’t cycles that if they could get exactly the same benefit of all the
twenty seventy percent of teetotalers started drinking and if they did the
decent thing for the community and started boozing they could get exactly
the same benefit and leave everybody else to carry on drinking exactly the
same but that didn’t get in the news I don’t know quite why not but never mind
so what does that mean and well they are they model the effect of reducing that
everyone dropping a unit of alcohols eight grams of the drink and I’m doing
25 mils of wine whatever of alcohol so as equipment or everyone cutting down
their drinking about sixty percent or so four and half thousand T that’s some
chronic diseases are avoided each year I swear I’m one percent of total deaths
in England this is ring fire half a million people died in England each year
so it didn’t the life accidents and violence that say it’s two percent so
what they’re really saying is that if this we you know the hell average hazard
have annual death would be reduced two percent with this if everyone just
stopped this amount of drinking when you turn that hazard ratio of 0.98 again
looking at overpopulation it actually only increases life expectancy web eight
eleven weeks so it again it doesn’t look terribly exciting less than three months
extra on average that the population would live if they cut down their booze
by sixty percent the don’t look great it corresponds to
five minutes a day over sixty years so that drink cutting down cutting down
that drink or that extra that drink is costing five minutes a day and we’ve
done other calculations previously to this which also showed the six units of
alcohol roughly and was about half an hour or was about a Microlife over
lifetime interested you it means that the first drink is puts five minutes on
your life essentially there’s medicine and the second one takes it off again
it’s poison now the third one takes stop and it
doesn’t go medicine poison medicine poison medicine poison
and it goes medicine poison poison poison which is poison with five minutes
knocking off every time yeah I see you know that’s not going to carry on to
extreme levels but that’s the the interpretation of the of the data so
what does this mean on average though it means that on average one make a note
this is some terrible photo and we can lose one Microlife from there’s me
eating a burger stuff in my face of the burger and six you decide is about three
cans of lager tucking in for my favorite games lager two cigarettes
I have it’s two cigarettes as half an hour for the life expectancy you know
that’s another people done these calculations then we don’t know what the
effect of to exactly two cigarettes is but a pro-rata multiplied up over a
lifetime to cigarettes often now this next picture I’m afraid those of you of
a nervous disposition may look one to look away now
this is an awful sight get rid of it now that is only is done by CGI rules on
Photoshop and five don’t take a picture there
the latest meta-analysis by Richard Peto group that appears shows that being 5
kilograms overweight each day is half an hour for your life expectancy this is
what among five kilos overweight so and yes a ghastly sight so I’m yeah I’m
notching up a few of those I won’t tell you which ones so not this against and
so that’s five five kilograms leg weighs half an hour for life so you can add
these up now the blue these are terrible things you can get those half hours back
by eating or by exercising statin or course material recent meta-analysis by
taking my stats in every day it’s the hazard ratio is about point
nine which is one year which is a micro life half an hour on your life each day
you take the statin now I think when I showed that previous enough of the
Staffie I don’t take that as a moment I showed those previous pictures with the
poor yellen’s I thought I probably only take statin
I almost convinced myself because of my own analysis I should be taking a statin
next I think watch the half an hour just for taking a pill the ones I can try to
control side effects doesn’t sound too bad
says half an hour for statin exercise an interesting one because
seem to get is highly nonlinear their latest meta-analysis again should highly
nominate that compare leading slob the first 20 minutes of moderate exercise
each day gets you about an hour about to microlights and the second 20 minutes
only gets you 20 minutes so you live longer but you’ve spent the whole time
pounding up and down the trail if you want to do it so and after that it’s
going so highly nonlinear on exercise and we can do the same with with fruit
and vegetables etc etc now it is all very rough is there any metaphors any
stories but they’re quite gripping ones and they my public health people say oh
you can’t do this people will do trade offs they’ll say oh I’ll take the statin
that means I can have some extra dreams like stop smoking things like that so
there is that danger so that’s can be gained Michael eyes but we shouldn’t
believe what we hear was this about a dried food just as healthy as fresh and
this is from the world nut and dried fruit Congress they say you know I’m
going to be careful people are trying to sell us stuff all the time okay and I’m
going to go your ID gun on all the wait a bit longer than I thought but and I
had Carol a bit more for the last bit which is talking about the trickier
stuff when we don’t know what’s going on and I never quit your drink so this is
facing up to ambiguity generally we don’t know what these numbers are what
these risks are and then only reasons we might not know this if we turn them to a
patient we might not know we only know a few bits of information about them we haven’t got great population data to
do the statistical modeling and finally you know actually England about what’s
really going on you know can we really make these projections you know is there
hazard ratio the benefit for statin is just going to carry on for decades and
when nobody’s been taking them that long so stuff we just don’t know about and
I’d like to refer to a particular work and Paul hammers in the audience here on
doing some really nice stuff I’m trying to deconstruct these sources of
ambiguities and difficulties and thinking ways to communicate them for
example this is what he’s been doing National Cancer Institute their
colorectal cancer risk assessment tool now this is just one source of
uncertainty statistical uncertainty confidence intervals on the risks how do
we say that this risk is not quite know well they’ve been using the blurring
idea Bank of England thing using the blurring to show distribution to show
the uncertainty rather than a fixed quantity to show they don’t quite know
where it is so slightly cone of uncertainty bit better with a blur on it
and the important things that have done at the doing empirical research and
she’s finding out how this changes people’s perceptions and their
understanding and now we’ve done to show something here that we’ve been doing is
a little video this is silent and this is about the Cochrane Collaboration some
of the findings table they pictures the Cochrane Collaboration when they do
review produce a summary of findings table which basically says out of a
thousand people what would happen if you took the treatment didn’t take the
treatment so it’s like a drug Facts box and they produce that at the moment and
they want to visualize it this is some ideas we’ve we had and I just want to
illustrate where and yeah this is the sort of summary findings table at the
moment this is adjuvant radiotherapy after surgery for cervical cancer what
might happen to a thousand people with and without the treatment if a bubble’s
speech but you can read the speech models or listen to me this is
visualization showing 400 people who don’t
do have the intervention compared with 100 women who don’t and this is adjuvant
ready for PR surgery of cervical cancer for adverse outcomes to ours and side
effects adverse events and others the main outcomes and that’s the expected
many women who suffer these adverse events from the meta-analysis these are
the grade scales who should do the quality of evidence theater for ease and
bad evidence so you can see we’ve shaded the side effects slightly paler now
assuming in but trying to keep an idea that this is actually the stories people
whom this doesn’t affect trying to keep the idea of the part to hold comparison
and now we take the difference as we did before to show you getting extra in the
people don’t have it of extra disease progression and deaths and but the
people do have it have got extra side effects because that’s the crucial risk
balance is that worth that the chance of that with the chance of that happening
and that depends how you feel and you know for things like HRT is could be
really interesting might and really be different for different people but
actually out of a hundred women a diversity of more than 10 the benefit
and only a couple will be harmed now this is the final pictures this is
the optional bit this is doing that low calden on these things we’re blurring
the images to give an idea that we’re actually quite sure how many there are
here the blurring the people now so this is the point is what this shows is that
the death difference was not statistically significant and those s it
was expected to be three there’s a lot of uncertainty about that you can’t
actually be sure there is more deaths than the whether it’s more deaths in
that group or that group so in fact by introducing the
uncertainty it suggests that she’s pretty well ignoring this not taking so
much notice of this and being much more focused on these these outcomes and
we’re very unsure about the death data okay so that’s the kind of ideas that
actually the group who designed these are are doing
this they’ve they use gonna use something very soon for this one of the
problems is that we always thought it seems much easier to compare sizes
vertically when rather than a cross way so they even stack them up vertically
okay so that’s the kind of thing that we’ve been doing now that only allows
though for confidence intervals it’s detecting the quantifiable uncertainties
but for years has been known that not all uncertainties are susceptible of
measurement and this has identified million hundred years ago Frank Knight
in the economics literature to say he used the word uncertainty to say if the
things were just competent number and some we just don’t know what these
chances I’ll be coming for their confidence interval we just we’re not
happy about the evidence we just don’t think Keynes in 1937 so they’ve got no
scientific basis to make a probability up
here’s them about predicting copper prices in forty years or something
suddenly don’t know so this is getting out there with a qualitative assessment
of how happy we are with our quantitative analysis now this is really
quite it’s not so safe to get taught and statistics course you’re not Megan
really taught is anywhere again the people are really trying to struggle
with this in the IPCC consistent treatment of uncertainties
and they’ve got this idea of confidence in their analysis they try to assess how
confident they are with the whole thing they’ve gone through their whole so if
you’ve got and they predominately their confidences got to sort of main
dimensions whether you get the experts agreeing and whether the evidence had
good the evidences so if you’re low agreement there’s no evidence and nobody
can agree you actually haven’t got a clue or unconfident they don’t even use
any probability so refused to use numbers in those circumstances
don’t use numbers but up at the top these are highly robust evidence then
they’re happy to use probabilities and then in between they might say things
like well you might say ordered magnitudes only very rough assessments
of probability not giving fine numbers at all refusing to give that precise
numbers I think it’s a very interesting attempt whether it be success a lot the
way remains to be seen but very interesting attempt
the grade scale which is used by the Cochrane Collaboration many other
organizations tries to do a similar thing it sets the quality of the
evidence high moderate load very low they used to use this as a definition
which was in terms of it was it was me for example yo is for the researchers
very likely to have an important impact on our confidence and estimate of
effects in other words how robust that their conclusions were to further things
there might be found out I like that there now each tit which i think is are
they very unfortunate they now say our confidence and the effect destler’s
limited the true effect may be substantially different from the
estimate of the effect which i think is a shame because only in terms now in
terms of expression some of the uncertainty absolutely division
distribution around the the effect and i prefer the older definition so I’d like
to just I think finish now with a quotes actually from Paul Han without his
summarizes up what this puts the situation in difficult situation it puts
us into is one where the communication uncertainty clinical evidence burdens
patients collisions are like those sort of nearly difficult tasks that mean
historically ill prepared to undertake to affirm the value of available
evidence while simultaneously recognizing its inevitable limitations
yeah this is great but there are these caveats to understate the sites of
action while acknowledging all the reasons for indecision say yep you know
we’re not sure that everything but actually we do feel this is the best
course of action and have faith by the line of someone’s actions and out what
the future holds while affirming the irreducibility attack to say well you
know we can’t know exactly we can’t pretend we know everything and they want
to do know everything but actually we do feel confident in the action now this is
a quite a political idea of risk communication we could get a whole lot
about doing this in all sorts of areas and but it seems to be incredibly
important just in this area so in my final conclusions on communication we
need uniform reporting of benefits and harms multiple formats and metaphors of
risk one size does not fit all acknowledged patients concerns and for
deeper uncertainties we need a qualitative language for expressing
confidence and analysis eg grade or something else and the finish off oh yes
yeah this is what and if you really want to do this is a
some orange juice I bought in Finland which advertises itself as promising to
make you live longer than your friends so maybe maybe that’s assuming maybe
that’s the sort of marketing we need to improve people’s lifestyle I’m not sure
and so with that thank you very to deke that was a great talk
I was interested in your statin example and when you presented the net
differences the it looked like stroke and heart disease deaths averted I guess
was sort of comparable in magnitude to some of these other things yeah I
thought of haddock and and but yet later on in the talk you started talking about
micro lives and yeah yeah it seemed much doubt in your mind yeah and I just
wonder if you’d comment on the weighting of the various yeah points and what and
also do you think people should subjectively assign weights or or could
there be some agreement on weights oh yeah okay well that that risk balance
thing you know naturally you know if in a way you could put a number on these
are the relative importance of these it would be decided you know you tip over
and you could in fact we were thinking doing a little model because it is like
a center of gravity thing if you got four people and it’s three times as
important you move it out and and you can actually model the balance whether
this you know this tip so how important is I I’m not fond of I mean there are
four policy decisions yes we can put numbers on those we’ve we do that within
cost-effectiveness analysis in the UK you know with our death panels we’ve got
more like and stuff I’m on one of them so where we cost up treatments and use
quality adjusted life years to do it for policy decisions the individual decision
I feel that that’s you know that’s maybe going too far I think that it should be
left up to an individual for me I mean that I suppose I’ve been slightly short
by people saying that for the myopathy for example if you could change the
statin and you know you can avoid some of these these hobbies these heart side
these side effects and also that while the evidence for all cause mortality on
statins was was disputed into you know a few years ago
recent meta-analysis come to our the Fermi on the all cause mortality there’s
being beneficial from sensors thanks Davis terrific talk and I have a
kind of philosophical question for you and you know a lot of what you talked
about is how we improve the public’s understanding of risk information but do
you think there’s a way to improve their tolerance of uncertainty and I think
there’s kind of a cultural psychological components of this that you alluded to
but yeah there’s our way of doing that apart from I guess or in addition to
sort of improving numeracy and understand the yeah it’s a it’s a it’s a
question of anthropology is life that they the fascinating worked I don’t know
not enough bad on different cultural attitudes to risk and different in Japan
and Germany compared certain there was a risk aversion compared with in other
areas where there’s a general tolerance for for uncertainty and risk and the
uncertainties is people’s feelings better hugely influenced by their trust
of the authority you know whether they actually can rely on somebody and so if
they and if it’s a fairly reliable or trusted source they’re willing to you
know accept their the ambiguities and the necessary uncertainties in
scientific knowledge when making a judgement about something if they don’t
trust you then the one any any chink in the armor and they’re they’ll go for it
and I will just decrease trust acknowledgement of uncertainty so is a
very as you as you know it’s very complex issue it’s another a simple
issue at all and it by will depend on for some of the personality of the
individual in terms of their and their optimism you know and their and their
attitudes to the authority it’s a it’s a comment that I I now try to watch all
the time when people are dealing with with vaccines or food scares or and
Fukushima or anything where and you know some something has to be communicated to
the public how they doing it we know what metaphors that they’re using are
they are they owning up to scientific uncertainty that you know the all the
advice says yes we’ve got to be an open and transparent about by darn certainty
there it comes to the cross and people politicians seem unhappy about doing
that but some are quite good at it some people quite good at it because they
they can say oh well you know we’re not sure that
thing that people say is that you must when even in a situation deeper
uncertainties you must be clear what you do know clear what you don’t know and
clear what people can do now and that’s whether they have something action to
take clear that you are doing what you can to reduce your uncertainties and
that you this is what you should do for the moment but we made just to be on the
safe side but we may change your advice when we know more so those those things
and then you can get away with it essentially and you kind of say oh we
have another clue what’s going on and that’s not a good risk communication but
it is it’s something that’s needs a lot more study and attention but they’re
only can study is through I think through case studies – something
something I there’s some empirical research going on with them then down
Karen and the cultural cognition project and how it is very exciting stuff with
actually get actors to play out different roles in different communities
and see how they how they can change their opinions about these aren’t we
know run HPV vaccines for example are there any other questions for
professor Spiegel halter please yeah I just want to know if you had some
comments on sort of I assigning reliability indices to risk projection
models or something like that and comments on validation size independent
validation studies and their role in all this
oh yeah yeah no absolutely the I mean this is I get this more for me though as
your weather forecasting and things like that
yeah these idea that if you’re gonna produce a probability for someone you
know people can quite reason the up order that mean you know does it is it a
a meaningful probability and I think you know we are being told what the chances
are all things in different circumstances whether it’s sports or
whether it’s politics of others whether all weathers are in health then then you
speak I think a greater attention of familiarity with the tools there are for
checking the reliability of those probabilities a whole I did so I’m going
to know that scoring rules but calibration discrimination two very
well-established and worked and in weather forecasting for checking these
probability of precipitation numbers and and so that when they say 60% 60% of the
time it does right you know that’s what you need to be able to make sure of that
and and of course that has to be assessed in a new sample of completely
new data that hasn’t have any part that serving the development of the tool so
people are it’s getting better I think your medicine but the you know for so
long have been people just inventing risk indices and then actually you know
didn’t win how reliable the numbers are at all so validation is really important
in this and but again it’s still I think you know there’s multiple things that’s
checking that the numbers are reasonable and then there’s how they’re presented
to people I think and how they’re the language used for those numbers is
important before I close this out oh please if you’ve got a question um I was
wondering if you could talk a little bit about how risk communication changes if
the patient is under earth becomes more distressed about this issue
I guess I’m particularly thinking about prenatal testing and attention all
parents yeah yeah yeah no no I mean there are jobs thing by all these things
is that then the numbers in any decision aids and graphics and things like that
can only be part of a of a process that’s dealt with sensitively and in a
in a humane way and and so that the language and the use is is needs to be
handled very carefully and I mean I’m not I’ve got a whole stuff they could
demonstrate though there’s anything I didn’t know but the crucial thing that
having the multiple levels is that you know what you get people information
that they don’t want that they really don’t that isn’t gonna help is that I
because it’s going to confuse or that you know it’s just too much they’re too
upset they don’t want to hear it so that’s why if you can’t just say Malloy
here it is and think you’ve got rid of your responsibility to inform that
patient it’s absolutely ridiculous so it has to be sensed it has to be sense
particularly I think how much actually that the person would like to pass the
decision back to you that and that’s you know absolutely fair enough but you know
this is part of the general idea of shared care and informed decision-making
that there has to be the sensitive process and these tools only one one
part of it but they are an important part of that they know it and they’re
not just as an ethical duty but because some people like me want to see this
stuff and get infuriated when I’m armed off with a whole lot of logos
patronizing childish drivel which is what I get when I leave the health
information it’s one of the numbers but not everybody does so you have to and
you know have those so we’re hoping and desiring patient information leaflets
them for cancer screening and cervix creams very conscious count there and
breast cancer screening is very controversial in UK at the moment we’ve
really advising on the leaflets and they’re the idea if you do it online
it’s quite easy to make multiple levels so you can have a know though broad
qualitative thing at the front and do can drill down and find out more detail
at they want no difficult in a leaflet but I think you can still have parts to
the leaflet that are clearly for those who want more information
got the numbers got the flowcharts through what you’d expect to happen to a
thousand people you don’t want to dump that right in at the beginning in fact I
think I’ll keep numbers out of the first bit you know just and and again the fact
that in many situations it’s when it especially when there’s some uncertainty
about science that it is a reasonable thing to say these authorities recommend
this you know they’re essentially if there are trusted authorities that’s it
and that might be enough for some people but not enough for everybody so I think
that’s is it as I said is a complex issue and that has to be handled
sensitively but I think the things that the NCI ones in the the Agilent ones you
know are using the extremely delicate situations and and have been found
useful it just made me think of a few examples
that have been in the news recently about changing changing recommendations
for prostate screenings and for mammograms for women under 40 and there
was this big uproar when they changed their recommendations because they say
look at all these lives that we are saving and it’s hard to also consider
the additional risks because of the screenings and I think those graphics
that you showed would have been a great way to help address the concerns of the
public exactly but it’s really difficult to I mean this business of overdiagnosis
and screening as being you know there’s some excellent books that have come out
in the u.s. recently about that and you know it’s a real problem because you
just show that you put in the screening and there’s vast mass of over treatment
for almost no benefit very differently came to that the person who gets up and
says this saved my life even though it’s almost certainly didn’t say not though
they said it very likely that actually you know this wouldn’t have harmed them
anyway very difficult to counter that personal input and so but I think it
just needs that have to have this evidence ready and in a way that’s clear
that shows show we’re working particularly on this idea on screen
you’re saying what would happen to a thousand people if they were screened a
thousand people in one screen now through this sort of frequency tree
showing it as it happens how many go down this how many goes down that and
how many ended up in these particular situations and seems to be you know as a
way to try to do it now the problem with that I think a lot of people would agree
that’s the thing to do they can’t agree though on how what the numbers are I
mean that the breast cancer screening of the thing of the moments one interesting
way is not just that statistical confidence intervals it’s is the
uncertainty there’s actually no scientifically you know is disagreement
about what the the extent of overdiagnosis is in breast cancer
screening and so that’s a delicate situation where people are those dispute
about what the numbers are and how you have the Latin a patient
information leaflet has got to be faced and I’m sort of looking forward to
finding out how people manage it I think you make a really interesting point
about change in historical context for prevention research so you know in the
days when children were dying from pertussis having a vaccine was
miraculous and now that most of us have never seen that in this country at least
you know people are more concerned about the potential side effects of the
vaccine and I’m wondering whether you know how you communicate that kind of
things we’ve seen that for example with antivirals
antiretrovirals yeah changing even if a relatively short course of time and and
it’s you know prevention as you say it is extremely uncertain if you
successfully how do you show that you know any way to you know use the
historical knowledge where we know that we’ve prevented to inform those and
there’s not cause to mention say I think the historical bit has to be part of the
story you know these are stories which others would mean what I’m saying is
that you have to turn these into gripping narratives that to a certain
extent arouse emotions but not too much I mean that’s the general general
recommendation so it has to be images and stories that do grab people and
their historic rather at moment people influence so much by individual stories
about about me or my family celebrities and whereas the story of
what’s happened to the population is an important story and I think that should
be part of that of the message that this is what things used to be like and now
they’re not like this but in West it could be like that again you know or the
you know that’s why this is this is effective and that’s why we have to do
it so I think that the historical bit is important this one because in the sense
is giving an image for what it would be we didn’t have this which otherwise you
haven’t gone yeah any other questions for David before I
close it out I thought I’d let you know you you refer to a few times to press
accounts of risk interpretation one of the other things that the office of
disease prevention here does is to put on an annual course called medicine in
the media Oh sniffing oh yeah and Steve will lotion and Lisa Schwartz do you
know the faculty who who organized this with us we can only teach 50 journalists
a year so it’ll be a long while before we penetrate but it it really is a
challenge yeah during during the course what what we’ve
done is to issue the students a calculator so they can do their own
calculations and a whistle when somebody really does something screamingly bad
they get to whistle this test case really you know fit the bill of bad bad
reporting you gave a couple of awfully good examples that yourself I would like
to close on behalf of everybody I’d like to thank you for a really stimulating
lecture it’s very much in the tradition of the medicine mind the gap series that
really is shaped by ideas that come from our colleagues at the NIH and we’re
grateful for the co-sponsorship of the National Institute on Alcohol Alcohol
Abuse and Alcoholism and the division of cancer prevention at
the NCI Office of Biostatistics Research at the NHLBI so on behalf of
all of us I’d like to thank you and wish you safe travels
thank you very much thank you

Leave a Reply

Your email address will not be published. Required fields are marked *