As I'm not on the list I can't post this, could you do it for me?
I'm not on this list, but a colleague forwarded this to me.
> Dear all,
>
> there are different scoring systems to estimate the cardiovascular risk
> in primary prevention, e.g. Framingham-based scores, the SCORE project,
> or the PROCAM-score with different estimates. This has already been
> discussed on this list.
>
> But is it too naive to add two further questions?
>
> The first question is a methodological issue: Often or mostly such
> scores are produced by a multivariate analysis of different risk factors
> run over a sample of the population.
> Such an analysis requires that the factors are ideally independent from
> each other. But the concept or the theory of the Œmetabolic syndrome¹
> shows that some of these risk factors are closely interrelated, e.g.
> hypertension, hyperlipidæmia (different fractions), overweight, diabetes.
> Apparently as a consequence some factors like diabetes figure in one
> score and are eliminated in another solution or score. Of course, such
> equations, on which the scores are based, can be rated by their
> statistical quality or be tested within independent samples. But how to
> deal systematically with the interdependence?
Multivariate analysis doesn't require that predictors are independent of
each other. The process is easiest what they are not, but works with
predictors that are correlated as well. All that happens is that the
prediction becomes less certain. So it is that in many societies risk
factors tend to cluster in persons. This does not invalidate the maths, just
make it harder. Don't forget that risk estimation models are just that:
models. In order to combine risk factor information they do not have to
combine the factors in a physiological way, but in an informational way. The
'independence' of terms in a model refers to their information content
rather than their physiological independence.
>
> The second question is a little bit comprehensive, too: The risk
> estimates of these scores apply to primary prevention, i.e. in persons
> without (or not selected by) a diagnosis of CVD.
> People with angina or a history of myocardial infarction have this risk
> as a baseline risk, too. But they have already CVD and they have
> additional disease specific risk markers (cardiologic findings, degree
> of heart failure etc.), or risk reductions due to appropriate therapy
> (ASS, beta-blocker etc.). There is a lot of work on different
> constellations (risk of re-infarction after MI or different
> interventions, systematic reviews about risk reduction by treatments
> like ASS etc.), but I don¹t know any comprehensive risk score in
> secondary prevention / therapy.
This is still an issue that concerns me too. But we were unable to identify
any such thing as a low risk group of people with established CVD. In public
health terms, the real question is not, I think, what the person's risk is,
but how we can do useful things to lower it. People living with CVD are
overwhelmingly likely to die of it too.
>
> The second question is particularly relevant for the evaluation of
> comprehensive programs (or frameworks) of care for CVD. Disease
> management programs are complex interventions and the primary outcome is
> not a single intervention/ a treatment effect or simple adherence but an
> improvement of the risk situation of the patients involved. In
> controlled trials you therefore need an estimate of the baseline risk /
> risk reduction by appropriate care, and you cannot wait for the Œhard
> clinical endpoints¹, cardiovascular events which will occur ten years later.
I would beware of proxy endpoints. The entry criteria into a clinical trial
mean that population-based risk estimates are probably wrong. You are better
off calculating risk by using a control arm.
Ronan M Conroy ([log in to unmask])
Principal Investigator
The SCORE project
Royal College of Surgeons
Dublin 2, Ireland
+353 1 402 2431 (fax 2764)
--------------------
Just say no to drug reps
http://www.nofreelunch.org/
----------------------------------------------------------------------------
----------------------------------------
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom
they are addressed.
If you have received this email in error please notify the
originator of the message. This footer also confirms that this
email message has been scanned for the presence of computer viruses.
Any views expressed in this message are those of the individual
sender, except where the sender specifies and with authority,
states them to be the views of The Royal College Of Surgeons in Ireland.
----------------------------------------------------------------------------
----------------------------------------
From: Ronán Conroy <[log in to unmask]>
Date: Wed, 31 Mar 2004 13:47:40 +0100
To: <[log in to unmask]>
Subject: Cardiovascular risk scores
I'm not on this list, but a colleague forwarded this to me.
> Dear all,
>
> there are different scoring systems to estimate the cardiovascular risk
> in primary prevention, e.g. Framingham-based scores, the SCORE project,
> or the PROCAM-score with different estimates. This has already been
> discussed on this list.
>
> But is it too naive to add two further questions?
>
> The first question is a methodological issue: Often or mostly such
> scores are produced by a multivariate analysis of different risk factors
> run over a sample of the population.
> Such an analysis requires that the factors are ideally independent from
> each other. But the concept or the theory of the Œmetabolic syndrome¹
> shows that some of these risk factors are closely interrelated, e.g.
> hypertension, hyperlipidæmia (different fractions), overweight, diabetes.
> Apparently as a consequence some factors like diabetes figure in one
> score and are eliminated in another solution or score. Of course, such
> equations, on which the scores are based, can be rated by their
> statistical quality or be tested within independent samples. But how to
> deal systematically with the interdependence?
Multivariate analysis doesn't require that predictors are independent of
each other. The process is easiest what they are not, but works with
predictors that are correlated as well. All that happens is that the
prediction becomes less certain. So it is that in many societies risk
factors tend to cluster in persons. This does not invalidate the maths, just
make it harder. Don't forget that risk estimation models are just that:
models. In order to combine risk factor information they do not have to
combine the factors in a physiological way, but in an informational way. The
'independence' of terms in a model refers to their information content
rather than their physiological independence.
>
> The second question is a little bit comprehensive, too: The risk
> estimates of these scores apply to primary prevention, i.e. in persons
> without (or not selected by) a diagnosis of CVD.
> People with angina or a history of myocardial infarction have this risk
> as a baseline risk, too. But they have already CVD and they have
> additional disease specific risk markers (cardiologic findings, degree
> of heart failure etc.), or risk reductions due to appropriate therapy
> (ASS, beta-blocker etc.). There is a lot of work on different
> constellations (risk of re-infarction after MI or different
> interventions, systematic reviews about risk reduction by treatments
> like ASS etc.), but I don¹t know any comprehensive risk score in
> secondary prevention / therapy.
This is still an issue that concerns me too. But we were unable to identify
any such thing as a low risk group of people with established CVD. In public
health terms, the real question is not, I think, what the person's risk is,
but how we can do useful things to lower it. People living with CVD are
overwhelmingly likely to die of it too.
>
> The second question is particularly relevant for the evaluation of
> comprehensive programs (or frameworks) of care for CVD. Disease
> management programs are complex interventions and the primary outcome is
> not a single intervention/ a treatment effect or simple adherence but an
> improvement of the risk situation of the patients involved. In
> controlled trials you therefore need an estimate of the baseline risk /
> risk reduction by appropriate care, and you cannot wait for the Œhard
> clinical endpoints¹, cardiovascular events which will occur ten years later.
I would beware of proxy endpoints. The entry criteria into a clinical trial
mean that population-based risk estimates are probably wrong. You are better
off calculating risk by using a control arm.
Ronan M Conroy ([log in to unmask])
Principal Investigator
The SCORE project
Royal College of Surgeons
Dublin 2, Ireland
+353 1 402 2431 (fax 2764)
--------------------
Just say no to drug reps
http://www.nofreelunch.org/
----------------------------------------------------------------------------
----------------------------------------
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom
they are addressed.
If you have received this email in error please notify the
originator of the message. This footer also confirms that this
email message has been scanned for the presence of computer viruses.
Any views expressed in this message are those of the individual
sender, except where the sender specifies and with authority,
states them to be the views of The Royal College Of Surgeons in Ireland.
----------------------------------------------------------------------------
----------------------------------------
> From: Beyer <[log in to unmask]>
> Reply-To: Beyer <[log in to unmask]>
> Date: Tue, 30 Mar 2004 19:10:02 +0200
> To: [log in to unmask]
> Subject: cardiovascular risk scores
>
> Dear all,
>
> there are different scoring systems to estimate the cardiovascular risk
> in primary prevention, e.g. Framingham-based scores, the SCORE project,
> or the PROCAM-score with different estimates. This has already been
> discussed on this list.
>
> But is it too naive to add two further questions?
>
> The first question is a methodological issue: Often or mostly such
> scores are produced by a multivariate analysis of different risk factors
> run over a sample of the population.
> Such an analysis requires that the factors are ideally independent from
> each other. But the concept or the theory of the Œmetabolic syndrome¹
> shows that some of these risk factors are closely interrelated, e.g.
> hypertension, hyperlipidaemia (different fractions), overweight, diabetes.
> Apparently as a consequence some factors like diabetes figure in one
> score and are eliminated in another solution or score. Of course, such
> equations, on which the scores are based, can be rated by their
> statistical quality or be tested within independent samples. But how to
> deal systematically with the interdependence?
>
> The second question is a little bit comprehensive, too: The risk
> estimates of these scores apply to primary prevention, i.e. in persons
> without (or not selected by) a diagnosis of CVD.
> People with angina or a history of myocardial infarction have this risk
> as a baseline risk, too. But they have already CVD and they have
> additional disease specific risk markers (cardiologic findings, degree
> of heart failure etc.), or risk reductions due to appropriate therapy
> (ASS, beta-blocker etc.). There is a lot of work on different
> constellations (risk of re-infarction after MI or different
> interventions, systematic reviews about risk reduction by treatments
> like ASS etc.), but I don¹t know any comprehensive risk score in
> secondary prevention / therapy.
>
> The second question is particularly relevant for the evaluation of
> comprehensive programs (or frameworks) of care for CVD. Disease
> management programs are complex interventions and the primary outcome is
> not a single intervention/ a treatment effect or simple adherence but an
> improvement of the risk situation of the patients involved. In
> controlled trials you therefore need an estimate of the baseline risk /
> risk reduction by appropriate care, and you cannot wait for the Œhard
> clinical endpoints¹, cardiovascular events which will occur ten years later.
>
> Can anyone help?
>
> --
> ------------------------------------------------------------------------------
> --
> Dipl.Soz. Martin Beyer, Dr.med. Jochen Gensichen
>
> Institut für Allgemeinmedizin / Institute for General Practice
> Universitätsklinikum Schleswig-Holstein Campus Kiel / Kiel University
> Hospital
> Christian-Albrechts-Universität zu Kiel / Christian-Albrechts-University
> of Kiel
> Arnold-Heller-Straße 8
> D-24105 Kiel / Germany
> Tel.: ++49-(0)431-597-4121 / 2226
> Fax.: ++49-(0)431-597-1183
>
> eMail: [log in to unmask]
> od [log in to unmask]
>
> homepage: http://www.allgemeinmedizin.uni-kiel.de
> ------------------------------------------------------------------------------
> --
>
>
--------------------------------------------------------------------------------------------------------------------
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom
they are addressed.
If you have received this email in error please notify the
originator of the message. This footer also confirms that this
email message has been scanned for the presence of computer viruses.
Any views expressed in this message are those of the individual
sender, except where the sender specifies and with authority,
states them to be the views of The Royal College Of Surgeons in Ireland.
--------------------------------------------------------------------------------------------------------------------
|