- Good evening. This evening I'm going to talk about one of the most powerful
tools in public health but one which is often misunderstood, and that is screening. And I ask the question,
when is screening useful and when is it not? Now, historically, broadly
medicine can be divide into two types of activity, curative medicine and
preventative medicine. And this goes right back in history, both of these have existed
all the way through the history of medicine. Curative medicine has depended on people coming forward to their
doctor with symptoms and say, "These are the symptoms." And the doctor makes a diagnoses and then prescribes treatment. Alongside that, there
has been a long tradition of prevention through
advice on greater exercise, good diet, moderation in
alcohol, and in fact, many things which would be very recognizable
to public health today. And I've illustrated this with the famous medical school in Salerno. Which was the place which really brought together, for Europe, the previous learning of
Greek and Roman medicine, with Jewish and Islamic
medicine, and integrated it, but it had these strands
of preventative medicine and curative medicine. And these have been all the way through the history of medicine since that time and indeed before that time. Screening is neither of these,
it lies between these two. And with screening what
you're trying to do is identify people who are either at very high risk of
disease or alternatively people who've got very early
disease but have no symptoms. So it's not trying to give broad advice to the general public, nor is it trying to
diagnose based on symptoms. It's trying to pick people up early who do not have symptoms. And the reason for doing
this is fairly obvious. Many diseases, and in the
case of cancers for example, virtually all of the diseases
have a much better outlook if they are diagnosed earlier. These would include breast
cancer, prostate cancer, lung cancer, and bowel cancer, the four leading causes of
cancer mortality here in the UK, but many other cancers as well. They also include some genetic diseases. And then there are some things like high cholesterol or having
high blood pressure, which are risk factors for later disease, which if you treat them early, they lead to a much better outcome. And I've illustrated this
in this particular case by looking at breast cancer. And what you can see
is that if you identify the breast cancer very early, the 10 year survival is extremely good. And even with relatively advanced cancer in its earlier phases, the outlook over time is very good. But if cancer is diagnosed late, then there are serious treatments needed, many of which are extremely unpleasant. And the outlook in terms of mortality is significantly worse. So early diagnosis makes
a very big difference. And I've done a whole series of talks in the last year on different cancers. Identifying why it is that early treatment can make a very big difference. And if we look at just
four important cancers, these four I've talked about. These are the five year
survivals by stage at diagnosis. And early stage of
diagnosis means the cancer has been picked up very
early in its course. It's still definitely the cancer, but it's picked up very
early in its course. And as you go through the
stages, two, three, and four, we're talking about
later and later disease. And with all of these, if you
get stage one disease early on the five-year survival is over 50%. A great, the majority
of cases will survive. And most people with breast
cancer, prostate cancer, or bowel cancer picked
up in stage one or two, and in many cases in stage three, will have a good, medium
and long-term survival. But once you get to stage four disease, the outlook is a lot less good. So that's the reason
why identifying cancers and other diseases early is so critical. Because if you pick them up very early, then you can intervene
and prevent disease. And the same is true if we're
talking about things like a heart disease or a stroke, if you can pick up the fact that someone has significant
risk factors early and treat those risk factors, like lowering their cholesterol, you significantly delay the time before they actually get problems and hopefully delay
them almost indefinitely within their lifetimes. Now, another thing to say
with these diseases however, is that not all of them, and this is what the rest of the talk is really to talk about why this is. Not all of them are
well suited to screening with current levels of science. And if I choose the four
cancers I've identified here, all of which benefit
from earlier diagnosis. For two of them, breast
cancer and bowel cancer, we have a screening program and that screening program
undoubtedly saves lives. And in two of them, prostate
cancer and lung cancer, we do not have a screening program for technical reasons I
will go on on to talk about, particularly in the
case of prostate cancer. So early diagnosis really does matter. Picking up risk factors
early really does matter. But not all the diseases
where that is true will screening be useful. And in reality, at this point in time with our current diagnostic methods, for most serious diseases, screening is in fact
currently not a good idea. So it's a minority where
screening is helpful. To be an exception, to
be one of the diseases where screening is useful, it has first of all to
be a serious disease. Screening for very minor
diseases will only cause problems and worry and is very
unlikely to benefit anyone. You must be able to diagnose
it reliably and safely. And many diagnostic tests are fine if you're talking about someone
who's got obvious symptoms. But much less good, and
we'll talk about why, much less good if we're
trying to screen people who do not have symptoms. So you need to have a very
reliable diagnostic test. And some diagnostic tests
are also unpleasant, or in some cases even dangerous. And you wouldn't want to use
those in general for screening. Thirdly, you must be able
to prevent or treat it, in early disease very
effectively and safely relative to the risk of the disease. If you can't treat it,
then why diagnose it early? Because you're not going
to be able to intervene and actually improve the outcome. The next thing is it must
be reasonably common. Let's say you had a disease
of one in 30 million, doing a screening test in the UK would clearly not make
sense because you to end up only diagnosing two cases in
an absolutely ideal situation. So it has to be common
enough that the pickup rate is reasonably high. And there must finally be
a sufficiently long time between the first signs of the disease, which the test can pick up, and actually running into
problems that you can intervene. So if there's a disease where actually between the first point
you acquire disease and the point you actually
becomes a severe disease is only a few days. Clearly a screening test
is not likely to be useful. So the longer the time between the initial signs you've got disease and the time it you
have to have intervened the more amenable to having
a screening test it is. And many diseases fail
on one or more than one of those reasons. Let's start off with the diagnostic step. The practical question
for a diagnostic test in a screening system, it's
also true for diagnosis, but less important actually for diagnosis because of the fact
people have got symptoms. Is the risk of either a false
positive or a false negative. And there are false positive
and false negative tests for virtually every diagnostic test that could potentially
be used in screening. The question is, how many? So, you have group people, some of whom have got the disease and some of whom have not got the disease. The perfect test picks up all the people who have the disease and flags positive. And does not pick up anybody
who has not got the disease. So everybody else is picked
up by the test as negative. That's the perfect test. In reality, most tests have some rate where they will pick up some people who do not have the disease
but they flag positive. That's a false positive. And some people who
actually have the disease, but they flag negative, and
that is a false negative test. And if you have too many false positives off too many false negatives, that can be a serious problem. Because a false positive means
you will get a treatment, which you don't need because
you don't have a disease. And a false negative test means that you will be falsely reassured. So that is the concern
with the diagnostic test. And there are two bits to this. The first of it is the
property of the test itself. The sensitivity and the
specificity of the test or the screening tool. The sensitivity is the ability to actually pick up positive tests. The percentage of true
positives the test picks up. And many tests that may
people who are listening will be well used to, like ECGs, have a false positive rate. So for example the ECG, Electrocardiogram, is around 50% sensitive for picking up a ischemic heart disease, narrowing of the blood
vessels of the heart with chest pain. So, if you just do that
test, you will pick, you will miss some cases,
quite a significant proportion. The specificity of the
test is the percentage of true negatives that the
test flags as negative. Because you want the test to
say positive if it's positive and negative if it's negative. So for example, MRI scans of the brain are about 80% specific for
diagnosing multiple sclerosis. But what that means is 20% of people, where the MRI looks as if
it's multiple sclerosis, actually it is not. And that is a concern if you're the person who's the one of those 20% then your worry is that you might be falsely positively diagnosed. So all tests have a
sensitivity and a specificity. Very few of them have
virtually 100% for both. There are a few which are
extraordinarily reliable, provided you use them properly and provided you wait the
appropriate amount of time. Pregnancy tests for example,
are extraordinary reliable now. And provided someone
has waited long enough, a positive will mean they are pregnant and a negative will mean
they are not pregnant. It's a very important
question clearly for a woman who thinks she might
potentially be pregnant. And the same is true for modern HIV tests. A positive HIV test done in a lab is likely to be extremely reliable. If it's positive it will be HIV although that's almost always
repeated just to be sure. And a negative test, again, rules it out provide it's been done at the right time. So these are ones where the
sensitivity and the specificity is incredibly close to a
hundred percent used properly. But for most tests, it's
not quite that good. And there is generally
for most tests a trade-off between sensitivity and specificity. If you make the test more sensitive, so you pick up more cases, you also make it less specific so you also pick up more false positives. And therefore when you're
optimizing a test for screening, you need to know which is more important. Is it not missing a true case, maybe because it's too early in disease, or is it diagnosing a false positive? And depending on which disease it is one or other of those may be important, it depends on the clinical situation. There are some diseases
where all you just want to be is just reassured it's not there. There are some others where
you really do not want to miss a single case. And an example of this
is a very standard test for screening for diabetes. So there's a test which many people who are listening to this would have had, something called HbA1c, standard screening tool for diet diabetes. If you set the cutoff at 63%, 6.3%, the sensitivity of this test is about 80% and the specificity is 82%,
so both are around 80%. And therefore you will
miss 20% of diabetes cases and over-diagnosed roughly 20%. If you set the cutoff a bit higher, the sensitivity goes down and the specificity goes up. And therefore you'll miss 37% of people, but over-diagnose 6%. So, you can adjust the test but usually if you
increase the sensitivity, you will decrease the specificity. This is a property of the test. But the other thing which affects this, and this is rather less intuitive, is the importance of
the prior probability, the likelihood that the
person that you're screening or diagnosing has got the disease. And this is based on the
mathematical concepts of this gentleman, the
Reverend Thomas Bayes, a clergyman from Tunbridge Wells. And the Bayes mathematics,
the Bayesian theory, is essentially the mathematical proof that if you ask a silly question, you will get a silly answer. If I do a test for ischemic heart disease, for narrowing of the blood vessels due to clogging of the arteries, in large numbers of 20 year old runners. If I get a positive,
it'll almost certainly be a false positive. I've asked a stupid question and therefore have a reasonable chance I may get a stupid answer. So it's very important when thinking about tests for screening, what is the probability that
someone who's being screened actually has the disease? And I'm going to do one slide of maths. And I promise it is only one slide, but it is very important for understanding why so many diseases
screening is not appropriate. I'm going to take a disease
named as what diseases you wish and have a test that's
got a sensitivity of 90% and a specificity of 90%. That means it picks up 90% of
the cases as true positives and picks up 90% of the
negatives as true negatives. It's a pretty good test. By medical standards,
that's not a bad test. The key thing you've got
to think about though, it's not just the
sensitivity and specificity, which for this test is pretty good, but what is, what's called
the positive predictive value and the negative predictive value. The positive predictive
value is the probability this individual person, if
they have a positive test, is it a true positive? And the negative predictive
value is this individual person if they have a negative
test, is it a real negative? And what I've got here
is two probability trees. And as I said, this is the
single slide with maths, there's nothing else after this. But what I've got on the left, this is exactly, using
exactly the same test is a thousand people on the far left. And the prevalence, that probably the number of people in the community who've got this disease
in this case is 1%. So, true positives are 10
people out of this thousand and the true negatives are 990 people. That's the true number,
that not the test question, that's genuinely how many
people have got the disease. So you've got a test
that is 90% defective, so of those 10 people, it picks up nine. By a not very complicated
bit of mathematics. But therefore it misses one, so nine true positives and one false negative. But going down the other arm of the tree, 990 people do not have
disease in this situation. And this tests has got a 90% specificity. So you multiply 0.9 by 919 you get 891 where the
disease has truly picked up, that they are truly negative. That's correct. And a 10% the other way, which means of those 990, 99 people are flagged by
this test as positive. And therefore what you have is 99 false positives because the pre-test
probability is very low, but to only nine true positives. Most of the people that you've picked up with this actually pretty good test are in fact false positives. And this means the
positive predictive value is what's called-- It was was 8%, that's nine
out of 108 positive tests, and the negative predictive
value is very good. Now I take exactly the same
disease, exactly the same test, but this time, the prevalence
in the general population is 10%. So of that thousand people,
100 truly have the disease, 900 do not have the disease. Therefore, 90% sensitive tests, 90 of those are true positives. 10 are false negatives. And of the 900 who genuinely
don't have the disease, it correctly picks up
810 as true negatives and 90 of those are picked
up as false positives. So in this situation with
exactly the same test used for exactly the same disease, the ratio of true positives
to false positives is 50-50. 90 to 90. So positive predictive value of 50% negative predictive
value is still very good. So this means in this situation, if on the left-hand side, if I did a screening test, only 8% of the people who I screened would actually genuinely have the disease. Or the people who were
screened as positive in the right-hand side for the same test, because there's a higher probability before you go into the test. There's a 50% chance that
if I get a positive test, it's a true positive. So I've gone over that in some detail, but the key thing for those who have not wanted to go through the maths is to understand the fact
that if even with a good test, if you do this in a population where there's a very, very small chance of the disease in the first place, you will throw up a very large number of false positive tests. So both the test property and
the probability of disease in the population are very important. It's not just the test itself. Now this is important
because many treatments, in fact, probably most treatments and indeed some diagnostic
tests can do harm. And all of medicine, whether
it's curative medicine, or here in screening, is about a balance of risk of treatment, risk of intervention versus
risk of no treatment, risk of no intervention. And in many situations
actually doing something does more harm than doing nothing. Because medicine comes with some risk. At the extreme end, that might
be a very major operation, whether the person will not
survive without the operation but there might be a 50%
mortality from the operation. But it's still better to do the operation. The screening it's right at the other end of the risk spectrum. The probability of doing
harm is usually low, but the chance that someone
actually has disease is also relatively low. It's always about risk of
benefit against risk of harm. And if you treat someone with
no benefit from treatment, because you're treating
people with a false positive, they only get the risk of the treatment, they don't get the risk, or they don't get the
benefit of the treatment because they didn't have the
disease in the first place. So picking up false positives
is potentially very damaging. And you could use an example for example, of something which is a
well-known and very real risk factor for people having a stroke. This is something called carotid stenosis. This is a narrowing of the
blood vessel in the neck. And this is illustrated on the right. These are, this is the
blood vessel going up, forks and on the right, you can see a narrowing. That is potentially a
risk for someone to go on to have a stroke. Now, there's an operation
you can do to remove that. There's also a way you can
do it by putting stents in to open these areas up. These are perfectly
reasonable things to do if the risk is high but these procedures
themselves have risks. So what you don't want to
do is do a screening test, pick up a bunch of people who will not benefit from treatment, and then do the treatment,
the surgery, or stenting, because that itself can cause a stroke, which might've been preventable. So what we're always trying to do is only do screening tests, where it is much more likely than not that we will do benefit rather
than harm from interventions. And that is the key to screening. In addition to the direct harms, you also have the psychological harms that you can turn someone who was previously fit and
well feeding, perfectly happy. And you do a test which turns out to be a false positive
test, and as a minimum, it turns him into a medical
case to be followed up. Medicalizes people,
previously well people, can cause them significant worry, and sometimes can lead
to medical over-treatment because once people have
identified something, they tend to need to
chase it down to the end. And in some cases, particularly
infectious diseases can also cause stigma. So you do not want to embark
on a path potentially leading to significant numbers of false positives unless there is some benefit
at the end of that process. This reliance on the fact
that tests are not perfect, of course therefore means
that whether screening is a good idea for disease is not static. Because tests are
improving the whole time. The sensitivity of tests, the
specificity and the safety in some cases of the tests
are improving the whole time. We're also getting changes to the effectiveness
and safety of treatment. If you have a treatment
that is extremely benign, it really has very few side effects. Having a few false
positives who are treated is actually not a particularly big deal if you're then, if you by doing that are going to lead to a
very large number of people benefiting from the treatment. If and then they have the treatment has got significant risks, you're going to need
to be much more careful about false positives in a screening system leading to unnecessary treatment. And then finally, the
epidemiology of diseases changes. So, some diseases are getting more common. So with the same test, it's getting a better
predictive value over time. Some diseases are getting less common. So you get changes to the test,
changes to the effectiveness and safety of treatment, and
changes to the epidemiology of the disease. And then there are some
philosophical points too whether the screening is a good idea. Should we, for example, screen for pre-symptomatic Alzheimer's? That's a genuinely difficult
philosophical question. If we had a perfect screening
test for Alzheimer's, would someone wish to know that they are likely to
go on to get Alzheimer's since we do not currently
have treatment for that? And we don't currently have
very good tests for it. I put on the right, some
of the ways in which you can try and pick it up. None of them are perfect, but if we did have a perfect
test, you still might say, actually, let's wait to
do this until we know that we've got a good treatment. So if you tell someone you
could go on to get Alzheimer's, we can say, but we can intervene and that will go on to
reduce your probability. So there are sensitivity
and specificity questions, effectiveness and safety
of treatment questions, epidemiology questions, and
philosophical societal questions which have to be answered
for a screening test. So that's the kind of
principle preamble to this. Central to the screening
tests that we do use is the idea of risk stratification. So, because it's very important
you don't screen people who've a very, very low
chance of the disease unless you've got a perfect
test, which is quite rare. The key thing is to identify the people who are at greatest risk of disease. And the main way we do this through most screening
things at the moment, and we're going to get
on to a much better way of doing this over time, is to look at age above all, gender, and for some
diseases consider ethnicity, but you might also want
to consider other issues. If you've got some for example,
a woman in her sixties, then she is clearly at much
higher risk of breast cancer, which you might want to screen for, than a woman in her thirties, and absolutely a man in his sixties, although breast cancer in men
very occasionally does occur. So you stratify on relatively
crude criteria at the moment, largely age and gender related. And when you start thinking
about all of these reasons why you want to be very careful of not over-diagnosing someone
and over treating them, you realize it's actually
a relatively small number of diseases where
screening is a good idea. And on the right here what I've got is the current national
population screening programs. And I'm going to run
through in particular, the screening programs for adults. Some of these screening
programs are antenatal, this is to pick up major
fetal abnormalities. Things like down syndrome for
those parents who wish to. Some of them are immediately postnatal, these are to pick up genetic abnormalities where early treatment
would lead to a reduction or completely stopping
lifelong ill health. But then there's a group of things where there's a national screening program in adults in the UK. And these are the ones
I want to talk about in greatest detail. But there are also things where we screen high-risk individuals who
come from particular groups. This is not the general population, but someone comes from a
particular high risk group. And finally the risk is like health checks and other GP testing, which
is often called screening but often has a slightly
different purpose. So there's a number of screening programs, but the ones I want to concentrate on to illustrate how screening works are the adult diseases in the UK system. Now in the adults diseases, it's often said that
screening prevents disease. Actually for most of the
adult screening programs, it's to identify disease very early and prevent it progressing further often by in the case of
cancers, for example, actually just simply
cutting them out in surgery. And there are four
major national screening programs in adults. Cervical screening, in women, younger women, breast screening in slightly older women, bowel screening in men and
women looking for bowel cancer. And then aortic aneurysm screening which is looking for a
swelling in the blood vessel, major blood vessel in the abdomen. Let's start off with cervical screening, it's a extremely effective
screening program. It's the commonest cancer in young women. If people get this, they will get this early in
their lives in medical terms. So women aged 25 to 49 are offered screening every three years. And those 50 to 64 are offered
screening every five years. These numbers may change over time, but this is broadly where
it is at the moment. And this is therefore the screening is particularly concentrated
on earlier years. And that is because most
people acquire their risk for cervical cancer very
early in their life. It's a virus, HPV virus. And there'll be a period of time. And then that virus can cause
cells to become a pre-cancer a number of years later. And if that's not treated that can go on to cause
a cancer later in life. Screening of women for
cervical cancer began in 1988. And it's led to a substantial
reduction in cervical cancer by about probably 30 to 40%, according to the Office
for National Statistics. We're moving over to a better test, and have done in the last year, which improves the accuracy. And that may in fact lead
to an even better outcome in terms of this screening program. But on the right here, what you see is the
rate of cervical cancer up to the point of screen
when screening began, where the arrow is. And then you can see
that significant decline and that's because the
cancer was was prevented by picking up precancerous cells early on, and then intervening to make sure that those cells were killed. In a sense, you've very
localized our procedure. And that means the cancer
never goes on to happen. Unlike most cancers,
HPV decreases with age. And there's a clear and fairly easy diagnosable precancerous state. So this is a perfect
situation for screening. And you have these
precancerous cell changes. And then-- What you do is in the older methods, slightly, to say slightly changing, look at the cells, see
how abnormal they are. And if there are abnormal cells then the woman is invited
back for a procedure to try and make sure
that does not progress. So what are the advantages
of this screening program? What it means is that
women are picked up early in their disease. Very early in their disease before the cancer has gone
on to actually become, before the cells have gone
on to become cancer cells that precancerous. And then a very minor procedure is done, it may be extremely minor, really lasting a few minutes. It may be slightly more extensive but that means that a
small procedure early on means they don't go on to get cancer, which would otherwise have led
to very extensive treatment. and in some cases two
people or two women dying. So, the advantage of going
for the most sensitive tests by looking at the cells
with less advanced disease is you intervene even earlier and the procedures are even more minor. The disadvantages, some of those will be, would have actually gone
back probably to normal over time or would not,
it's a false positive. So there's a little bit of over-treatment, but the advantage of a
little bit of over treatment is it's a very minor treatment. If you wait too long till
you're certain it's a cancer you will have fewer people
have this very minor procedure, but more people will go
on to have a true cancer, and that can lead to people
having severe disease and in some cases dying. So, you're trying to intervene early with a little bit of over-treatment, but a very minor procedure to stop much more severe disease
and mortality later on. Cervical cancer is an
example where both the test and the epidemiology are changing. HPV has two particular types
that are responsible in the UK for somewhere between 50
and 70% of cervical cancer at the moment. But we also have a vaccine
which is widely used in the UK, which has over 95% effective against these if given to girls before
they become sexually active. And in the UK, coverage
among girls is about 89%. And this is likely to
reduce cervical cancer possibly by around 50% or more over time. And newer vaccines cover
a wider range of viruses, which can cause cancer and
therefore extend the protection. So what we expect is over time, the amount of cervical cancer will go down if we did nothing at all,
other than vaccinate, because it will become much rarer because we're getting rid of the virus. And therefore it may be
possible to do less screening and still have a significant effect in terms of reducing the
probability of disease. And then the epidemiology is changing, but also the test is changing
to make things more effective. And we're moving over to an HPV DNA test, and this will lead to a
reduction in over-treatment. So we've got a better
test and less disease. So things with cervical
cancer are heading very much in the right way. We've got an effective screening program and we've got an effective vaccine and we're getting better tests. This is a very clear good
news story in medicine. Despite the fact that
cervical cancer screening is highly effective, there is a good coverage rate in the UK for cervical cancer screening. But it's variable across the country and it's actually particularly
a problem here in London. And the rates have been gradually
drifting down over time. So if you compare rates in 2011, so 10 years ago, with rates now in every
part of the country, the rates of screening have gone down and I would strongly urge any
women in the appropriate age to get their screening cause
it really will lead to, over time in multiple people, reductions in their chances of
going on to get this cancer. The second major program to consider is breast cancer deaths. Breast cancer screening. There are about 11,000 breast
cancer deaths a year in the UK and around 55,000 cases. The ten-year survival over time has improved really substantially. So if you'd gone back to the seventies, breast cancer survival was around 40%. It's now close to 80%. And the peak time for
getting breast cancer is later than for cervical cancer. So breast cancer screening
is done by mammography, where people have a low
dose x-ray every three years between the ages of 50 and 70. And it's big benefit is
it catches early cancers. Obviously the risks are that
you will get some degree of over diagnosis and
over treatment and worry. But because breast cancer
is such a common cancer and treated early it is so easy to treat relative to how easy it is to
treat late on in the disease. Picking it up early through screening overall has a significant benefit. And the rate of breast
cancer detected in England is around eight per
thousand women screened. This is an example where actually the type of tests you use has different implications for whether it's useful for screening. So mammography, the type
of testing that's used, where women have low dose
X, ordinary x-ray done. Although it's a fact it's
a relatively low tech test, it is actually the best
screening test we currently have. MRI scanning, which is a much more sophisticated form of
testing, is more sensitive. But it would lead, but it
also picks up large numbers, more false positive tests if
used in a screening program. So it would lead to more
unnecessary procedures. It is better in particular people, people with very dense breasts, younger patients in some situations. But in general, this lower tech test is the most useful for screening. And then it can also be
benefited also from ultrasound where suspicious areas can be identified and help to identify where biopsy is. So it's the minimum amount
of procedures are done. These are some numbers, looking
at a mesh analysis of trials of breast screening in those 40 to 70. And this is from a large meta-analysis of very large numbers of people. Relative risk of reducing
mortality from screening about 0.8 for UK data. And about 0.8 for Canadian data. And so relatively consistently, a 20% reduction in
breast cancer mortality, and remember breast cancer mortality, breast cancer is a common cancer. So 20% reduction is really
quite a significant change. So in practical terms for every
10,000 women aged 50 years invited to screen for the next 20 years, you would end up preventing
43 deaths from breast cancer, but you would get some
degree of over-diagnosis with minor procedures
in general performed. But overall the benefits
of breast cancers, breast screening for breast
cancer are very significant. We now are trying to look at the question, should we extend the age of
people having breast screening lower in age and higher in age? And in both cases, what
we're trying to do, those who are doing the
studies are trying to do is to find out whether actually we will lead to a reduction in mortality without over-diagnosis. What you're trying to
avoid is over-diagnosis without a significant
reduction in mortality. And then there are also studies looking at the question about, should we be doing more
screening in high-risk women who for example, have a
first degree relative. That means a sister for example,
or a mother or a daughter who has a cancer at under 40 years, or has cancer in both
breasts under 50 years. And these people are at higher risk. So should we be screening
these people more intensively? And observational studies
would imply that we should. That if someone actually has
a very strong family history of young onset breast cancer, then you might want to screen
more intensively and earlier than in the general population. So this is an example of
stratifying based on risk. Breast cancer screening is
quite variable by country. These are a number of different countries. The UK is in red. Unlike cervical cancer screening, the rate of screening has
stayed relatively constant at about 75% over many
years now in the UK. This is higher than some countries but certainly lower than some others. For example, Finland has the highest rates and the Netherlands. And in fact, the USA also have very high rates of
screening for breast cancer. So, there's a certain
amount of variability among the countries. But it has remained fairly
stable in this country. The third cancer which is one which is, and it's even both men and
women, is bowel cancer. And for bowel cancer,
early diagnosis is the key to a good outcome. On the left here, you can see
the survival rates over time. Five-Year survival rates for
stage one through four disease. And if you pick this up
in stages one or two, and even actually in stage three, the outlook is really
very good for most people. Stage four disease much less so. So picking this up early
is absolutely critical. There are a variety of ways you
can screen for bowel cancer. One is what's called a bowel scope. This is actually looking up into the bowel for people who have otherwise
got no particular symptoms generally at age 55 in the UK. Or the other way this has been done over the last several years is something called
FOB, fecal occult blood. Three tests done every two
years from the age of 60 and can be done later in
some people done at home. And these have been posted in to test. And there's been a move over
to a rather more sensitive test or rather better test called FIT. And it's only requires one
specimen rather than three. And these are posted out to people and only if they are
positive do they then go on to the next stage, which is a colonoscopy, which is where someone looks
up their bowel with a telescope to try and see what's going on. So if there's a positive test,
that's the screening test. And then if there is a positive,
they have the colonoscopy, the second test which actually determines whether there is a cancer or not. And having this colonoscopy,
if they screen positive, allows both detection and
treatment of early disease. In some cases, certain
detection of early disease where a relatively minor
procedure can be undertaken. And someone who would have
gone on to get a bowel cancer is back to their normal life. So, in picking these things
up early is very worthwhile. And for both men and women, again, I strongly encourage people
who are potentially at risk of cancer to have these screening tests. If you get sent one of these
tests, please do do it. And we have in bowel cancer
several significant trials, large trials done and
all of them demonstrate that even using the rather less good tests we previously used, the FOB test, that there is a significant reduction in the case of a UK trial for example, a 13% reduction in colon
and rectal cancer deaths. So, this is not just
theory that this can work. This is been proven to work in trials where people were randomized
to a screening or not. And those who had screening had a significantly better outcome. And the same has been done
with the bowel scope test, where people had a single
test at the age of 55. Again, a very large study
17 years of follow-up, long period of time. And over time what we have
found was a 27% reduction in the amount of cancer
and a 30% reduction in cancer mortality. So again, clear evidence
that this form of screening reduces people's chances
of dying from bowel cancer over time. It's a long time lag, but over time, this makes a significant difference. So those are examples
where the trial evidence or the other studies are strongly supportive, that a screening program saves people going on to have disease and saves lives. But there are many diseases
where that is not the case, even where you might think
that it would make sense. And now, an example of this. And I'm only going to use the one, but there are others, is prostate cancer. Prostate cancer is the
commonest cancer in men and certainly in the UK. And there is a screening test, potentially there is a
blood test called PSA, prostate specific antigen. And... the screening test is pretty sensitive and fairly specific. But, if you screen
people to have this test, what you find is that actually, if you randomize people to have the test or not have the test, there's no increase in
improvement in mortality in the people who have the screening test compared to those who do not. So what you get is quite
a large number of men have a positive screening
test of the blood test. And then they have to go on to
have a relatively unpleasant, not massively unpleasant,
but relatively unpleasant would anyone choose to have it, procedure where they have a biopsy of the prostate. And only if that is
positive will they then need to go on to have further treatment. But these men are otherwise
living their blameless lives. They wouldn't otherwise
have had this biopsy and therefore it's only worth doing if there is evidence that
leads to a better outcome. And the randomized trials, large studies, this systematic review
looked at over 300,000 men randomized and found the
relative risk is one. There is no advantage to having this. Not a disadvantage from that point, from the point of view of mortality, but there's no advantage. And if you visualize this
with this prostate screening, this is a very nice way of visualizing it. On the left is a thousand
men without screening. The dark red arrows, what you
have is a small number of men on the top left who
without screening would die from prostate cancer. And a much larger number
of people who might die from other causes. And on the right, you
have the same situation. The same number of men
would have died without-- With screening. So it doesn't make any difference to them, but in the green, is what you
have, is the number of men without prostate cancer
who have false alarms and unnecessary tissue
samples, the biopsies. And in blue, you'll have the number of men who would actually have treatment. Might be surgery, might be radio therapy might be chemotherapy in
some cases for a disease, which they do not in fact need treating. They would not have come
to any harm from it. So here's a situation where if we screen large numbers of men with this blood test, we will make no benefit to them in terms of reducing mortality, but we will lead to a
lot more over-treatment. So this is important that
this trial has happened, but with the existing
technology it is very clear we should not be doing
prostate cancer screening using PSA testing. Finally, in the adults group, before I come on some
more general principles about where things are going, there are a number of diseases where you will do screening, but only in a very high risk group. So, for example, people who
have long-standing diabetes will have regular screening of their eyes to try and check if they're
a diabetic eye changes, because in those people
you can then intervene and prevent them from going
blind because of their diabetes. So that's screening, but only
in people who've got diabetes. And there are some screening programs that are done only in people who have a family history
of young onset cancers. Relatively small group of people, but if there's someone who very, very high rates of cancers at a very young age, those people might be screened, but only in those families. So there are some things where
screening is appropriate, but only for high risk individuals. So where are we going to go with screening in adults in the future? Well, there are several
diseases where it is clear that screening would
significantly improve outcome if we had better tests. These include lung cancer,
where there's a debate about whether we should be doing CT scanning of people who are lifelong smokers, because they're almost
all of the people at risk. Pancreatic cancer, ovarian cancer in women, cancer of the ovary,
and oesophageal cancer. These are all cancers which are unfortunately at this point
in time have a poor outlook. And I've talked about all of
them in the previous year. Over time, all of them have a poor outlook compared to breast cancer, prostate cancer and bowel cancer, for example. And that is in all of those cases because they are generally picked up late. Were they picked up much earlier treatment would be much more minor and out look could be much better. And here what we have on the right are the rates of mortality. Top of the list lung cancer, currently no screening need to debate it. Then breasts, bowel to ones
where we do have a screening. And that's one of the reasons why they have much lower
mortality than they used to. And then pancreas, ovary,
esophagus, the next ones down. These ones, really, we do need to think about how to improve it. high mortality, late disease. Scientifically, we are getting better at this stratification, identifying people who are
particularly high risk. And genomics and genetic
studies for example, are going to help us with this. But also a number of other tests, which allow us to say this man or woman isn't much the higher than average risk. They need more screening. And this man or woman
isn't much lower risk, and they can have less
screening than average. So something that we
only screen on the basis of age and gender may well
become a lot more sophisticated. And an example of a gene which I think a lot of
people will have heard of which if you, someone has it, you would want to do a much
more intensive screening is the BRACA, BRCA1 and 2 genes when there are mutations on those. Another technology which
is going to help us a lot is artificial intelligence
kind of methodologies used for diagnosis. This might be radiology,
X-rays, CTs, MRIs, or it might be histology where we look at cells
under the microscope. A lot is said about AI
that is exaggerated, but what the technique is very good at, and I think really will revolutionize, is pattern recognition. So just the routine churning
through of breast mammograms, cervical screening,
smear tests, and so on. So that's another technology, which I think is going
to significantly improve our ability to do screening. There's a lot of talk
about liquid biopsies, where by blood tests, we can pick up particularly
cancers at an early stage, often through genetic methods. And so that's all of that
is on the diagnostic side, and those are improving the whole time. And we're also getting safer treatments. And if the treatment gets safer, a screening program that might
previously have been risky because you're treating people
with a dangerous treatment and you definitely don't want
to pick up false positives may become a much more
attractive looking risk. And finally, we're getting
better tests all round over time. The whole time our ability to improve diagnosis is advancing. Antenatal screening is also
routinely offered to women. And that is looking for
things like down syndrome where you have three trisomy,
what's called trisomy, three of a particular
one of the chromosomes and similar things called
Edward's and Patau's syndrome. Women at about 20 have
about a one in 1,500 chance of a baby with Down's. And about 40, at the age
40 about one in a hundred. So it's still a minority,
but it increases over time. And the modern tests
combine an ultrasound test, looking at the neck, and a blood test, usually done at weeks 10 to 14. Now it's not perfect. And only if, if that's positive, that implies there's a problem. Women are then offered
the opportunity to go on to have what's called an amniocentesis where taking cells from the
the area around the baby or sampling of the villus, which is part of the way
the baby is attached. These are much more accurate tests, but they do have a very
small but non trivial risk of miscarriage. So women have to make a choice
as to whether they think this is something they would wish to do. But the first screening test means that only a very small
proportion of people need to go through to the second test if they wish it. The first test is very safe and does not pose any
threat to the pregnancy. And then when babies are born they'll have a heel prick
test generally at five days. And this is trying to test for a number of genetic conditions, which if they're picked up
early can significantly improve the probability that the
child can be treated early for things like cystic fibrosis,
congenital hypothyroidism. These are rare diseases
but if they're picked up it means that we can actually
prevent them from progressing. Finally, infectious diseases. I'm not going to talk at great length about screening for infectious diseases. And this is different from a mass testing, which has done under
operational conditions for a number of diseases
slightly different from screening as is ordinarily understood. The practicalities of this are
often quite tricky, actually. It's generally easier to
do classical screening for infectious diseases, where people get infected
and remain infected and infectious for long periods of time. So this is because that gives you time to actually, between the
time that you pick them up and the time they either have disease or pass it on to other people. And examples of this are
long standing diseases, diseases that once people got them they have for long periods
include tuberculosis, historically syphilis,
trachoma, which is a disease, a blinding disease of the
eye or sleeping sickness, which is a disease that
used to be very common in Central Africa, still it exists. This on the right is an example
of people being screened for trachoma as immigrants into the USA earlier in the last century. So you've got a longstanding,
long chronic disease, one that goes on for a long time. Some screening has sometimes been used for infectious diseases. And infectious diseases
also has the concept of what's called active
versus passive screening. Active screening is where you
go out and you find cases. Passive screening, you
wait until they come to you often with symptoms. And again, you can argue about whether that's screening at all. But in active screening, you
go out to try and find them. And there's a slight increased
risk of false positives cause you're at a lower
pre-test probability, but it means you'll miss fewer cases. In the UK, we used to
have a very extensive active TB screening program, for example, on the top what are some of
the trucks that were used in the active TB screening program in the earlier part of the last century. And on the bottom we've actually got our active screening
program here in the UK now. Much more smaller numbers, largely aiming to pick up cases in people who are homeless at the moment. But it is very important
we do that for their health and it also obviously reduces the risk they will pass things on. So this is screening
for infectious diseases. So screening is a very
useful potential tool under a restricted range of circumstances, but for many diseases it's not
the appropriate thing to do. It has to be a serious disease. You have to be able to diagnose
it reliably and safely. And ideally cheaply. There's got to be a highly effective treatment or prevention that's safe relative to the risk of the disease. It's got to be reasonably common or you won't be able
to pick up enough cases to make it justifiable. And also you'll get a very
large number of false positives. And there's got to be a
sufficiently long time from the point that you
pick them up in the disease to the point that you can intervene that actually it still benefits people. But if things fulfill all of those, screening is a very good way to mean that people have less disease and it's either stopped before they have any significant disease
at all, or early enough, they can have much more minor treatment which leads to longer survival and in most cases, a full and happy life. So we are constantly improving on this, but at this point in time, screening is only appropriate for a relatively restricted
range of diseases. Over the next decade, I
expect that to change. Thank you very much.
