 |
 |
Hi Mohammed: Not really sure what you want as far as evidence goes - I can send you the references if you would like.
1) There are lots of studies showing that health care providers struggle with risk assessment - in that they aren’t very good at estimating risks - typically overestimate risk - and benefits - typically overestimate benefits. Along with that is 30 years of anecdotal evidence as to how all over the board heath care professionals are when it comes to risk estimates. I would ask you if you have any studies showing that health care providers are good at risk estimates.
2) Philosophically, I am in favour of shared decision making because knowing a person’s values and preferences is very difficult to appreciate unless you specifically ask and have a reasonable idea of benefits and harms - for me I don’t require any studies of SDM for me to want to do SDM - hope that makes sense
3) Regardless of my philosophic stand on this there are studies showing that SDM decreases decisional conflict and also reduces costs and tests etc in general. - so that is encouraging
4) But even if studies showed that in some circumstances SDM leads to worse outcomes (on average) I would be OK with that because the people who would like to reduce their risk would be on treatment and those that don’t wouldn’t be on treatment - I can’t see how anyone can argue that this is a bad thing - would love to hear anyone’s argument against this sort of SDM
My only question for you is why would you not want to use risk assessment tools to help you do SDM and if you don’t use risk assessment tools how do you communicate risks and benefits to patients?
James
On Nov 3, 2018, at 4:02 PM, Mohammed T. Ansari <[log in to unmask]> wrote:
Hi James,
If you and Rod say that until such time that we have direct comparative assessment of the effectiveness of risk prediction-based versus alternative approach, we should based our decision on the best available evidence...and the best available indirectly supports that risk prediction-based will improve health outcomes on these grounds:.............and then you link your individual pieces of empiric evidence (including that "healthcare professionals are all over the board when it comes to estimating CVD risks") into an argument for your position....
When you do that, you'll find me on your side,.
Cheers,
Hi Mohammed - as you likely already now I agree with Rod and the concept of using ballpark risk assessment tools.
My question to you is how can a physician use "discretionary patient management" if they are primarily informed by guidelines that simply give thresholds for treatment. What I mean by that is, the evidence is overwhelming that heath care professionals are all over the board when it comes to estimating CVD risks and the benefits of treating risk factors. There is also no way that even years of clinical experience could allow a physician to gain knowledge of the increased risks associated with the commonly known (because of epidemiology) risk factors. Given that, what should inform a physicians discretionary management? Risk scores are miles from perfect but at present IMHO they are an improvement over the "high is bad low is good" concept.
James
On Nov 3, 2018, at 2:55 PM, Mohammed T. Ansari <[log in to unmask]> wrote:
Thanks Rod.
Yes that is also my understanding....that those with higher baseline risk will benefit more all else remaining the same.
The question still remains....whether formally triaging patients by their CVD risk and treating higher risk patients differently than lower risk patients based on a risk based protocol Vs. no formal CVD risk score based decision-making but physician's informal assessment and discretionary patient management will show meaningful differences. Some how you are convinced that there will certainly be incremental benefits of using a formal risk prediction tool over physician's discretionary patient management. I am not sure what makes you so convinced.....its not that the physicians is otherwise working with no information.....most of what goes into your model is otherwise also know to the physician (sex, age, smoking status, lipid levels, etc)....no?
Things will be even more different if the RR of therapy are lower for higher risk groups...
Hi Mohammed. Not sure if your question has been addressed sufficiently in subsequent emails.
I am saying that the RCT evidence I referenced from the Lancet (2012 and 2014) demonstrates that if you treat 200 patients with the same BP (or LDL) and lower their BP (or LDL) by the same amount, then the 100 patients with the highest predicted CVD risk, using a standard CVD risk prediction equation, will benefit more in absolute terms from treatment (i.e events per 100 on placebo minus events per 100 treated) than the remaining 100 at lower predicted risk will benefit from treatment. This evidence is as definitive as it gets.
I agree this is not the same as saying that a ‘management programme’ directed by absolute risk will prevent more events than one that is not. However the only reason why it wouldn’t be true would be if the management programme wasn’t implemented effectively and it didn’t lead to higher risk patients being treated. As I mentioned in a subsequent email, you cannot usefully assess implementation of such a programme in primary care in any meaningful and generalisable way using standard approaches like rcts because it would be extremely hard to do well and would be so context specific. Instead one needs to establish an implementation strategy in each setting using standard quality improvement methodologies.
In other words we have the evidence that it works if it is implemented. What we need to do is to implement it.
So I believe that statements like ‘we don’t have evidence that a predicted risk management approach is effective’ are not useful or relevant, because not only would such evidence be almost impossible to generate (e.g. by randomising patients to different strategies and following up for say 5 years with CVD events as the outcome), but even a good trial would not be generalisable in either space or time.
Too often we set impossible standards for evidence of effectiveness using approaches better suited for assessing efficacy. Implementation science is a quite different approach to the standard ebm approach.
Cheers Rod
* * * * * * * *
sent from my phone
Hi Rod,
I am being lazy and restricting to your response below rather than going into the papers. Relying exclusively on your response, I am not quite sure that the meta-analyses establish the effectiveness of risk prediction. Ideally I would want to see the prediction not of my CVD risk but prediction of response to therapy prescribed because of my CVD risk.....but thats another discussion for another day.
You say the "absolute benefits of BP-lowering drugs and statins are directly proportional to the pre-treatment predicted CVD risk"....which is not surprising because absolute benefits are dependent on the control event rates, which is what the predicted pre-treatment risk is predicting I guess. To me this does not establish that management guided by risk prediction versus unguided by it will lead to improved health outcomes, unless I am misreading you. Another way to look at that would be to see if the observed risk on therapy was lower than the pre-treatment predicted risk -- which could be because the therapy works (and should be offered when predicted risk is higher) or because risk-prediction was inaccurate if therapy does not work.
I thought what Juan was saying was that development of a risk prediction model does not tell us much about the effectiveness of using the model in practice.....I would agree I think.
m
Dear Juan – I challenge your statement that ‘using any cardiovascular risk calculator in the consultation room to take decisions is like using a magic crystal ball to take decisions has no scientific base ……,’
I recommend that you read the following two individual participant meta-analyses (Mihaylova B, Emberson J, Blackwell L, et al. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet 2012; 380: 581–90. AND Sundström J, Jackson R, Neal B, for the BPLTTC. Blood pressure-lowering treatment based on cardiovascular risk:
a meta-analysis of individual patient data. Lancet 2014; 384: 591–98). The key findings are nicely summarised in Figure 5 of Mihaylova et al and Figure 4 of Sundström et al.These two M-As, which retrospectively gave all randomised participants (from both intervention and control groups) a predicted pre-treatment CVD risk, show very clearly that the absolute benefits of BP-lowering drugs and statins are directly proportional to the pre-treatment predicted CVD risk.
We already know from a wealth of trial evidence that BP-lowering drugs and statins significantly reduce CVD risk (in relative terms), as James has eloquently described in this correspondence. However these two MAs now unequivocally demonstrate that the absolute benefits of these treatments are proportional to pre-treatment predicted risk.
Regards Rod Jackson
ps. the alternative old fashion approach – making treatment decisions primarily on BP levels or lipid levels - is simply a very crude alternative because single CVD risk factors are poor predictors of CVD risk.
From: "Evidence based health (EBH)" <[log in to unmask]> on behalf of Juan Gérvas <[log in to unmask]>
Reply-To: Juan Gérvas <[log in to unmask]>
Date: Wednesday, 31 October 2018 at 10:56 PM
To: "Evidence based health (EBH)" <[log in to unmask]>
Subject: Re: LDL and mortality (and use of statins)
-to finish this debate
1/ using any cardiovascular risk calculator in the consultation room to take decisions is like using a magic crystal ball to take decisions (no scientific base, but could be shared decisions)
2/ of course i never even think that you "cherry pick the meta-analysis", sorry, i never ever think that; i wrote " if your prefer cherry picking by the authors of meta-analysis" what is a critic to the way meta-analysis are done, by cherry picking
and 3/ i’m not pro or con statins, i'am only trying to debate about its rational use
-thanks in any case for your intelligent ideas and sorry for my horrible English
-un saludo juan gérvas @JuanGrvas
El mié., 31 oct. 2018 a las 4:17, McCormack, James (<[log in to unmask]>) escribió:
Thanks Juan for your response.
1) really glad you liked the calculator - and thanks to other people who said they also found it useful
2) risk calculators are all about how to help clinicians and patients make decisions - but agree they are not appropriate to use to decide on a specific threshold for treatment
3) not sure why a risk calculator that is used to give ballpark estimates of CVD needs to undergo a formal impact analysis - a formal impact analysis is used to evaluate if a prediction rule will be beneficial or harmful. When used for education to help make ballpark estimates the only outcome of interest is whether or not the person was properly informed - whether CVD events are impacted is irrelevant - a shared decision is the important endpoint
4) I didn’t cherry pick the meta-analysis - if you look at the actual numbers all MAs on this topic show pretty similar results - one should never just read the conclusion of a meta-analysis because as we have shown, many people who publish MAs don’t know how to interpret confidence intervals https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/1471-2288-13-134
The MA I sent and the one you suggested show pretty much the same thing - which is to be expected as they looked at mainly the same trials - although one (JACC) included a few more trials
See below for the result of each MA - the one I used was the JACC and your's was the AIM MA. The overall CVD and mortality point estimates were almost identical, one showed a larger point estimate for stroke and the opposite for MI - but there is such an overlap with the confidence intervals you can’t really say they are any different. Importantly, every point estimate leans in favour of statins.
Just for the record I’m not pro or con statins - I really don’t care if people use them or not - just that people are given the opportunity to make decisions based on the best available evidence.
A) Any cardiovascular event
JACC MA - 0.81 (0.75-0.89) – primary outcome endpoint so not any CVD event but similar
Arch Int Med MA - 0.81 (0.74-0.89)B) Mortality
JACC MA - 0.90 (0.82-0.99)
Arch Int Med MA - 0.92 (0.76-1.13)C) Stroke
JACC MA – 0.74 (0.55-0.99)
Arch Int Med MA - 0.92 (0.76-1.10)
D) MI
JACC MA - 0.78 (0.67-0.94) - CHD
Arch Int Med MA - 0.69 (0.52-0.91)
Hope all the above makes sense
James
On Oct 30, 2018, at 1:59 PM, Juan Gérvas <[log in to unmask]> wrote:
-excellent tool your risk calculator cvdcalculator.com
-but it is just a risk calculator, a table of risk not a table of decision
-to my knowledge you have not undergone formal impact analysisso you cannot use in a clinical context
-you cannot transform predicition rules in decision rules https://www.ncbi.nlm.nih.gov/pubmed/16461965
-about women and statins i prefer cherry picking by myself but if your prefer cherry picking by the authors of meta-analysis:
"Statin therapy is an effective intervention in the secondary prevention of cardiovascular events in both sexes, but there is no benefit on stroke and all-cause mortality in women" .https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/1195535
-un saludo juan gérvas @JuanGrvas
El mar., 30 oct. 2018 a las 20:57, McCormack, James (<[log in to unmask]>) escribió:
Thanks Juan - just a few comments - thanks for the info about the sugar industry - very interesting
A) CVD risk assessment
1) not sure why using CVD risk assessments needs to reduce CVD morbidity or mortality - in my opinion their only purpose should be to inform patients about a ballpark estimate of their CVD risk
2) agree that guidelines use them to inappropriately to establish treatment thresholds
3) what do you think of the risk calculator we developed cvdcalculator.com - always looking to make it better so would love feedback on how to improve it or if we got something wrong we would be happy to fix it
B) Women and statins
I think cherry picking 2 studies that didn’t show a statistical benefit is not the way one should discuss evidence. The meta-analysis (link below) seems to suggest that when you look at all the evidence there is a benefit in women - or am I missing something - would love to hear why this MA is flawed or what the concern is - you said you had lots of concerns
James
On Oct 30, 2018, at 12:25 PM, Juan Gérvas <[log in to unmask]> wrote:
-sorry for the delate in answering, James, i thought the question was over
-about my two questions
1/ do you agree with "it is uslees the global cardiovascular risk assessment in the primary prevention of cardiovascular disease in adults”?
There is currently no evidence reported that the prospective use of global cardiovascular risk assessment translates to reductions in CVD morbidity or mortality. THIS A KEY POINT IN PRIMARY PREVENTION AS IT IS USELESS THE USE OF ANY PREDICITION RULE https://bmjopen.bmj.com/content/7/3/e013650?rss=1
Prediction rules are not decision rules but may clinical guidelines transform predicition rules in decision rules https://www.ncbi.nlm.nih.gov/pubmed/16461965
2/ do you agree with "sugar lobby paid scientists to blur sugar's role in heart disease”?
"Sugar Industry and Coronary Heart Disease Research. A Historical Analysis of Internal Industry Documents" Together with other recent analyses of sugar industry documents, our findings suggest the industry sponsored a research program in the 1960s and 1970s that successfully cast doubt about the hazards of sucrose while promoting fat as the dietary culprit in CHD. https://jamanetwork.com/journals/jamainternalmedicine/article-abstract/2548255
-about your questions only one merit more debate, in my opinion
1) reduce cardiovascular events in secondary prevention? - I think they do Yes, almost no doubt in males. Lot of concern in females
For example: Scandinavian Simvastatin Survival Study (4S). Both CHD and total mortality were significantly reduced in men; however, there was no significant reduction in CHD mortality and no benefit on total mortality in women (RR, 1.16; 95% CI, 0.68-1.99)
Cholesterol and Recurrent Events (CARE). CHD mortality was significantly reduced in men but not in women-un saludo juan gérvas @JuanGrvas
El mar., 23 oct. 2018 a las 22:34, McCormack, James (<[log in to unmask]>) escribió:
Thanks Juan for agreeing to participate and appreciate your responses - see below for my response - capitals not because I am yelling but so you can easily see my response
On Oct 23, 2018, at 12:32 PM, Juan Gérvas <[log in to unmask]> wrote:
-thanks, James, everything is clear
-my answers in bold by your questions
-but two questions for you:
1/ do you agree with "it is uslees the global cardiovascular risk assessment in the primary prevention of cardiovascular disease in adults”?
NOT SURE WHAT YOU MEAN - DO YOU MEAN THAT RISK ASSESSMENTS (FRAMINGHAM ETC) ARE OF NO VALUE OR THAT LOWERING RISK BY STOPPING SMOKING, DECREASING BLOOD PRESSURE, BEING PHYSICALLY ACTIVE OR EATING HEALTHY FOOD IS USELESS
2/ do you agree with "sugar lobby paid scientists to blur sugar's role in heart disease”?
I HAVE NOT SEEN EVIDENCE THAT THEY HAVE BUT I SUPPOSE IT IS POSSIBLE. CAN’T INFER MOTIVE HOWEVER. PERSONALLY I THINK WE ARE DEMONIZING SUGAR IN A SIMILAR WAY IN WHICH WE DEMONIZED FAT 40 YEARS AGO. THE PROBLEM IS THAT TO ANSWER NUTRITION QUESTIONS LIKE LOW FAT VS LOW CARB REQUIRE HUGE TRIALS OVER YEARS AND I DON’T BELIEVE THESE WILL EVER BE DONE. I HOPE I AM WRONG
Statins based on the best available evidence - given that as with all evidence, it is never perfect -
1) reduce cardiovascular events in secondary prevention? - I think they do Yes, almost no doubt in males. Lot of concern in females
INTERESTING - I THINK EVERY META-ANALYSIS I HAVE SEEN SHOWS THAT THE RELATIVE BENEFIT IS SIMLAR IN MALES AND FEMALES - WHAT IS YOUR TAKE ON THIS OR YOUR CONCERN
2) reduce mortality in secondary prevention? - I think they do Yes, almost no doubt in males. Lot of concern in females
3) reduce overall SAE in secondary prevention? - I think they do Yes, almost no doubt in males. Lot of concern in females
4) reduce CVD events in primary prevention? - I think they do No. And we know it from more than two decades
I THINK EVERY META-ANALYSIS I HAVE SEEN SHOWS THERE IS A RELATIVE BENEFIT WHEN IT COMES TO CVD IN PRIMARY PREVENTION - WHAT SYSTEMATIC REVIEW OR MA DO YOU HAVE SHOWING NO BENEFIT - REMEMBER THE REASON WE THINK IT WORKS IN SECONDARY PREVENTION IS BECAUSE OF SYSTEMATIC REVIEWS AND MA FROM A COMBINATION OF INDUSTRY AND NON-INDUSTRY TRIALS
5) reduce mortality primary prevention? I think likely - all the point estimates are in favour but meta-analyses differ in whether or not the p-value is <0.05 https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/1471-2288-13-134 - they very likely don’t increase mortality No. And in any case, if yes, very little effect so harms are more important than benefits
AGREE
6) reduce SAE in primary prevention? they seem not to but I also believe that there is much inconsistency in definition/reporting of serious adverse events that while interesting we must remember that this is a secondary analysis and is really only hypotheses generating No.
With regards to tolerability
1) RCTs in general don’t show that the adverse effects are higher in the statin group compared to placebo - however the collection and reporting of adverse effects in all trials is at best weak Don't forget cerivastatin. RCT usually are not powered for adverse effects
DISAGREE - IF YOU THINK THAT ADVERSE EFFECTS (ABOVE PLACEBO) ARE SO PREVALENT, THESE STUDIES EVEN INDIVIDUALLY ARE MORE THAN POWERED TO PICK UP ADVERSE EFFECTS LIKE MUSCLE ACHES ETC - YO ARE CORRECT IN THAT THEY WON’T PICK UP ADVERSE EFFECTS THAT ARE SAY 1/1,000
2) enough people complain of muscle aches while on statins that it likely is a real issue but it is really hard to put a number to the incidence We can measure adverse effects with a proxy: rate of adherence to statin therapy was only 50% at six months, and further declined at one year.
DISAGREE - ADHERENCE IS A TERRIBLE PROXY MEASURE OF ADVERSE EFFECTS - SOME OF THE ADHERENCE MAY BE DUE TO PERCEIVED INTOLERABILITY BUT THE VAST MAJORITY IS THAT PEOPLE DON’T PERCEIVE A BENEFIT, DON’T LIKE TAKING PILLS, DON’T LIKE THE INCONVENIENCE OR THE COST
-un saludo juan gérvas @JuanGrvas
El mar., 23 oct. 2018 a las 0:14, McCormack, James (<[log in to unmask]>) escribió:
Hi Juan - agree this is an old debate but in debate it is always good to find common ground. Would you agree/or disagree with the following
Statins based on the best available evidence - given that as with all evidence, it is never perfect -
1) reduce cardiovascular events in secondary prevention? - I think they do
2) reduce mortality in secondary prevention? - I think they do
3) reduce overall SAE in secondary prevention? - I think they do
4) reduce CVD events in primary prevention? - I think they do
5) reduce mortality primary prevention? I think likely - all the point estimates are in favour but meta-analyses differ in whether or not the p-value is <0.05 https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/1471-2288-13-134 - they very likely don’t increase mortality
6) reduce SAE in primary prevention? they seem not to but I also believe that there is much inconsistency in definition/reporting of serious adverse events that while interesting we must remember that this is a secondary analysis and is really only hypotheses generating
With regards to tolerability
1) RCTs in general don’t show that the adverse effects are higher in the statin group compared to placebo - however the collection and reporting of adverse effects in all trials is at best weak
2) enough people complain of muscle aches while on statins that it likely is a real issue but it is really hard to put a number to the incidence
Hope the above is clear enough. Just wondering what you think specifically about these points.
James
On Oct 22, 2018, at 2:53 PM, Juan Gérvas <[log in to unmask]> wrote:
-this is an old debate
-the key is the frequency and severity of the adverse effects
Statins for primary prevention and serious adverse events
http://www.cmaj.ca/content/184/7/791.1
-but of course, it is easy to find studies pushing for the classical position:
Statins for the primary prevention of cardiovascular disease
https://www.bmj.com/content/348/bmj.g280
Statins for Prevention of Cardiovascular Disease in AdultsEvidence Report and Systematic Review for the US Preventive Services Task Force
-i have sent three links, this morning, and the third one is a critc of the previous ones
Behind the headlines. "Controversial report claims there's no link between 'bad cholesterol' and heart disease," "Bad cholesterol 'helps you live longer',"
Crítica a "el colesterol malo ayuda a vivir más. Lo que hay detrás de un titular.
https://www.nhs.uk/news/heart-and-lungs/study-says-theres-no-link-between-cholesterol-and-heart-disease/-un saludo juan gérvas @JuanGrvas
El lun., 22 oct. 2018 a las 19:24, Kev Hopayian (<[log in to unmask]>) escribió:
What I take away from this is that in the elderly, cholesterol levels have poorer predictive value for primary events and less or no benefit in primary prevention. Hardly surprising, because as you get older, age becomes a stronger predictor.
On 22 Oct 2018, at 15:41, Juan Gérvas <[log in to unmask]> wrote:
Cholesterol LDL-C does not cause and it is not associated with cardiovascular disease. Do not use statins.
El colesterol (ni "el malo", LDL) ni causa ni está asociado a enfermedad coronaria. No emplee estatinas.
[if you have problems, ask me directly the PDF, for personal use]
Lack of an association or an inverse association between low-density-lipoprotein cholesterol and mortality in the elderly: a systematic review.
En mayores de 60 años, menor mortalidad con colesterol "malo" (LDL) más alto.
https://bmjopen.bmj.com/content/6/6/e010401.full
Behind the headlines. "Controversial report claims there's no link between 'bad cholesterol' and heart disease," "Bad cholesterol 'helps you live longer',"
Crítica a "el colesterol malo ayuda a vivir más. Lo que hay detrás de un titular.
https://www.nhs.uk/news/heart-and-lungs/study-says-theres-no-link-between-cholesterol-and-heart-disease/-un saludo juan gérvas @JuanGrvas
To unsubscribe from the EVIDENCE-BASED-HEALTH list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=EVIDENCE-BASED-HEALTH&A=1
To unsubscribe from the EVIDENCE-BASED-HEALTH list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=EVIDENCE-BASED-HEALTH&A=1
To unsubscribe from the EVIDENCE-BASED-HEALTH list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=EVIDENCE-BASED-HEALTH&A=1
To unsubscribe from the EVIDENCE-BASED-HEALTH list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=EVIDENCE-BASED-HEALTH&A=1
To unsubscribe from the EVIDENCE-BASED-HEALTH list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=EVIDENCE-BASED-HEALTH&A=1
To unsubscribe from the EVIDENCE-BASED-HEALTH list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=EVIDENCE-BASED-HEALTH&A=1
To unsubscribe from the EVIDENCE-BASED-HEALTH list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=EVIDENCE-BASED-HEALTH&A=1
Dr. Kumara Mendis
MBBS, DFM, MD (Family Medicine), MSc (Medical Informatics), FCGP, FACHI
Board Certified Specialist in Family Medicine
There’s no limit to what you can accomplish if you don’t care who gets the credit
Bernard Baruch
To unsubscribe from the EVIDENCE-BASED-HEALTH list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=EVIDENCE-BASED-HEALTH&A=1