Healthy Skepticism International News

Can MaLAM make constructive contributions to improving guidelines?

June 2000

Two problems with the WHO-ISH guidelines 1. Are all antihypertensives equal? 2. Should we aim for low BP targets? Recommendations for the Heart Foundation of Australia lipid guidelines

Contents

Introduction

Two problems with the WHO-ISH guidelines

1. Are all antihypertensives equal?

2. Should we aim for low BP targets?

Recommendations for the Heart Foundation of Australia lipid guidelines

Introduction

Last year the World Health Organisation and the International Society for Hypertension published Guidelines for Management of Hypertension and a glossy summary called Practice Guidelines for Primary Care Physicians. The these guidelines have been heavily criticised especially by members of the “E-Drug” e-mail network: www.healthnet.org/progams/edrug.html
(E-Drug is one of MaLAM’s favourite sources of information.)

The two main problems with the WHO-ISH guidelines are discussed in the first section of this edition.

Unfortunately, the WHO-ISH recommendations have influenced the development of guidelines in Australia and perhaps in other countries as well. In November 1999 MaLAM wrote to the Heart Foundation of Australia to express concern but also offered to assist with writing future guidelines. This led to MaLAM being invited to participate in writing the Heart Foundation of Australia’s next set of lipid management guidelines. The second section of this edition is based on some of the recommendations that Peter Mansfield made to the Heart Foundation of Australia.

MaLAM is a member of WHO-NGO Roundtable. In March 2000 Peter Mansfield represented MaLAM at a meeting of the Roundtable in Geneva. At that meeting, Margaret Ewen, representing the International Society of Independent Drug Bulletins, expressed concerns about the WHO-ISH guidelines. Peter assisted Margaret with explaining the scientific issues to WHO staff. Unfortunately it is still not clear how the WHO is going to respond.

Two problems with the WHO-ISH guidelines

Peter Mansfield

Email: .(JavaScript must be enabled to view this email address)

1. Are all antihypertensives equal?

The WHO-ISH guidelines claim that “The randomised trials to date have not provided any clear evidence of differential effects on outcome of different agents producing the same blood pressure reduction.” The WHO-ISH Practice Guidelines for Primary Care Physicians claims that “There are six main drug classes used worldwide. All are suitable for the initiation and maintenance of BP lowering therapy”

Those claims are not in accord with the evidence from all the published trials comparing dihydropyridine calcium channel blockers with alternative antihypertensives summarised in table 1.

Table 1 Key RCTs of dihydropyridine CCBs for hypertension

2. Should we aim for low BP targets?

The WHO-ISH Practice guidelines for primary care physicians claims that “The goal of antihypertensive treatment should be to achieve “optimal” or “normal” BP in young, middle-aged, or diabetic subjects (below 130/85) and at least “high normal” BP in elderly patients (below 140/90).”

The lower targets for diabetics are supported by evidence from two trials. (HOT and UKPDS 38)

For non-diabetics the evidence from HOT suggests not only that there is no benefit from very low BP levels but also that very low levels may be harmful. See table 2.

Table 2 Outcomes in the HOT trial

The evidence from controlled trials supports use of thiazides, beta-blockers and ACE Inhibitors rather than placebo. That evidence does not provide answers about what target to aim for because those studies were not designed to address target setting.

Recommendations for the Heart Foundation of Australia lipid guidelines

Peter Mansfield

Email : .(JavaScript must be enabled to view this email address)

The following recommendations are my personal opinions addressed to the Heart Foundation of Australia. My recommendations were not put through MaLAM’s usual quality control debate because of time pressures. I hope that readers will let me know if you disagree.

Recommendation 1

Please consider the context of the Australian society that these guidelines should serve. Problems to be considered include the following. Poor diet and physical activity are important established risk factors that are getting worse especially amongst the younger age groups. Smoking rates amongst less wealthy young women are not going in the right direction. There is observational evidence suggesting social status and job insecurity are major risk factors. Both the gap between the rich and poor and job insecurity are currently worsening. Currently most of the spending on reducing cardiovascular risk occurs amongst wealthy people. This is sometimes described as the inverse care law. Interventions which are intended to improve health care often result in increased inequality.

Recommendation 2

Please consider the context of the Australian GPs who may be influenced by these guidelines. Many of us feel overwhelmed by the constant flood of conflicting messages about how we should be doing this or that better. There is evidence that despite the fact that many of us feel bad about not being able to keep up, the dissemination of guidelines alone is not reliably effective. However guidelines may be very influential for recent graduates who are just establishing their practice habits. The Heart Foundation is very highly regarded by Australian GPs. Other organisations feel very reluctant to promote recommendations that differ from yours. In fact, National Prescribing Service funded staff may feel obliged to promote your guidelines using methods that are reliably effective. For all those reasons it is important that your guidelines make recommendations that are better than current practices.

The context into which guidelines go also includes the many misunderstandings and biased emphasises induced by drug company promotion. There is often a need to help people to unlearn unjustified claims before they can accept justified claims.

Recommendation 3

Instead of producing separate hypertension and lipid guidelines please consider producing just one set of guidelines for reducing the risk of premature cardiovascular disease. Amongst other advantages, this would enable you to put all of the risk factors and interventions into their proper context relative to each other. As indicated above I do not expect you to implement this recommendation immediately but I would be delighted to assist in any way that you feel is appropriate towards implementing this recommendation sometime in the future.

Recommendation 4

A good process for writing guidelines is to employ someone who has expertise at evaluating evidence in general rather than someone who has expertise in the subject area to write the initial literature review. Bias is normal. It is easier for someone who is relatively ignorant to produce a less biased, fresh evaluation of the evidence. That approach can produce a more solid foundation to present to the experts for debate about what spin to put on the data. That approach can reduce the requirement for people to expend a lot of energy and override feelings of modesty before sticking their necks out to disagree with something that has already been written down.

Recommendation 5

Please do not make recommendations that conflict with the Australian Government Pharmaceutical Benefit Scheme criteria unless you have compelling evidence to justify doing so. Otherwise you are inciting GPs to break the law. Many of us will do so because of our primary loyalty to the patient we are sitting with and because many of us trust the Heart Foundation more than we trust the government. We do not trust the government because we perceive the bureaucrats as having more interest in cost-cutting. (GPs like to believe that the Heart Foundation has purer motives.) The reality is that neither the Heart Foundation nor the government are Norman Lindsay’s Magic Pudding (which always renewed itself no matter how much was eaten). Total health expenditure is relatively constant but expenditure on pharmaceuticals has been soaring so that expenditure on other areas of health care has had to be cut back.

Recommendation 6

Evidence from animal studies, in vitro studies and observational studies and speculations about mechanisms can be very misleading. Consequently, it is better to leave them all out unless you have a very good reason to do otherwise. GPs have limited time so it is best to focus on clinical trials with clinically important end-points.

If you do want to discuss observational evidence then please explain the “cum hoc ergo propter hoc” fallacy for the benefit of GPs who have not been taught, or have forgotten, how to interpret observational studies. In this case all you have to do is point out that just because there is an association between A) “lower cholesterol levels” and B) “lower rates of cardiovascular disease” does not mean that A causes B. It is possible that people who have C) “healthier diets and more physical activity” have lower cholesterol levels because of C and lower rates of cardiovascular disease because of C. The only reliable evidence for a causal link between lower cholesterol and less cardiovascular disease comes from the controlled intervention studies. Even that evidence does not exclude the possibility that other mechanisms may be important nor the possibility that healthier diets and more physical activity have additional benefits other than just by lowering cholesterol levels.

Recommendation 7

There are many ways to communicate about the efficacy of drugs:

relative risk reduction with treatment after a specified time
number of patients you need to treat to delay a clinical outcome so it does not happen during a specified time
absolute risk reduction after a specified time
absolute risks in the treatment and control groups after a specified time
absolute survival rates in the treatment and control groups after a specified time
Kaplan-Meier survival analysis graphs for treatment and control groups during a specified time
Of the list above the relative risk has the highest potential to mislead.

This is a difficult area because clinical trials are only good at showing that there is a statistically significant difference between a treatment and a comparator. They are not really suitable for determining the magnitude of the treatment effect unless they continue until all subjects have died. However we all want to know the magnitude of treatment effects and have not got any better way to estimate them.

Ideally we would use Kaplan-Meier graphs with 95% confidence intervals to communicate about efficacy but unfortunately few GPs have been taught how to read them. Also, whilst Kaplan-Meier graphs are common I have yet to see one that includes 95% confidence intervals published in a medical journal. (The advantages of such graphs include that you can get a feel for the variation over time during the trial in how long the events have been delayed by. This is important because so-called “preventative treatments” usually do not “prevent” events but merely delay them. I concede that sometimes events are delayed for long enough to enable the patient to die from some other cause so that they never experience the event.)

I would like to be able to tell patients that for every year they take a therapy they will be able to delay their death by, on average, X weeks. Unfortunately, clinical trials do not provide enough information to enable us to communicate in that way.

In the real world, I agree with the National Prescribing Service that the best compromise is to state the absolute risk of events in the treatment and control groups. I would add that without disclosing the duration of the study those numbers alone are unhelpful. The reason for that is that the absolute risks and everything else vary throughout the duration of the study. Those numbers would continue to vary over time if the study was continued until all subjects died. In the long-run even with an excellent treatment the absolute risk of death with and without treatment is always 100% and the relative risk reduction is always 0%.

Recommendation 8

The number of subjects is not a measure of the quality of a clinical trial unless an inadequately powered trial is being used for a claim that there is no difference between a treatment and a comparator. Large trials are warning signs for a) a small treatment effect and b) a drug company risking a large amount of money so that there will be strong pressures to put a positive spin on the results. If smaller trials have found statistically significant results then they should be included.

It is important to include the HERS trial which found an increased rate of non-fatal myocardial infarct plus CHD deaths in the first year despite a reduction in cholesterol levels. This trial is a good illustration of the point that just because a drug reduces cholesterol does not guarantee that it will reduce the risk of cardiovascular disease because there may be other mechanisms involved. In the case of HRT it may be blood clotting.

It is even more important to include the Lyon Diet Heart Study to make the key point that dietary change / a referral to a dietician may be more effective than a prescription for a cholesterol lowering drug. I underlined the word “may” in the previous sentence because this trial has imperfections and lacks confirmation from other trials (because of funding bias). However I do not believe that there is any compelling reason for disbelieving the findings of this trial. Participants in the intervention group were asked to change to an individualised Mediterranean type diet by a dietician following endorsement of the diet by a cardiologist. The rate of cardiac deaths plus non-fatal myocardial infarction during just under 4 years was reduced from 21.5% in the control group down to 6.4%. By comparison, the best results with a lipid lowering drug were seen in the 4S study of subjects at a higher initial level of risk. The higher initial level of risk makes it easier to achieve greater benefit. However, 5.4 years of simvastatin only reduced the rate of major cardiovascular events from 28% in the control group to 19%. The majority of those who would have had a major cardiovascular event during the 5.4 years still had one.

I do accept that use of cholesterol lowering drugs is worthwhile for people at high initial levels of risk. However, it is important for GPs and our patients to know that diet may be more effective. Currently few GPs are aware of that. The drug companies have put a lot of effort into repeating the message that cholesterol lowering drugs are “powerful”. It is important to correct that misunderstanding. Many GPs believe that it is not possible to get people to improve their diets. This may result from GPs giving up trying to influence patients because of repeated failures due to lack of adequate training. By contrast, the model of a doctor and a dietician working separately but as a team as in the Lyon heart study is achievable in Australia if funding is provided to enable less wealthy people to get access to public dieticians. Otherwise we will have another example of the inverse care law as mentioned in recommendation 1 above. Problem solving often needs to start with opening the problem up for discussion.

Recommendation 9

The terms “primary prevention” and “secondary prevention” can mislead via oversimplification and via meaning different things to different people. The concept of a continuous spectrum of initial level of risk has many advantages. One of the advantages is that it enables us to make the point that a low status job-insecure hypertensive diabetic smoker with a family history of cardiovascular disease deserves as much attention as someone who has already had a myocardial infarction.

Recommendation 10

Please support the New Zealand cardiovascular risk chart promoted by the National Prescribing Service. That calculator has been promoted to GPs during one-to-one educational outreach visits Australia and is extremely popular. Competing against it is likely to be counter-productive.

However, I disagree with Rod Jackson’s emphasis on absolute risk alone because it can oversimplify things. A 10 percent absolute risk and of a myocardial infarction at 40 years of age is very different from a 10 percent risk of a myocardial infarction at 90. However if the concept is explained to GPs we can correct for it by showing patients how their risk level compares with others of the same age on the chart.

Recommendation 11

It would be good to endorse either the new Australian guide to healthy eating or the 12345+ diet or both, rather than appear to be in competition with both. The dieticians that I have spoken to prefer the 12345+ diet because it is easier to teach patients. It is also better for GPs because it enables us to recommend “doses” eg 3 serves of fruit and 4 serves of veg etc, so it fits in with our way of thinking about therapies. Many of us are a bit vague on how much is a “serve” but we can learn that quickly and then become confident about giving advice or reinforcing advise from a dietician.

Recommendation 12

Because there is no evidence to support lipid targets it is logical to not recommend any targets. However it does seem reasonable to aim for lower cholesterol levels for patients who are at higher initial levels of risk. If we accept low targets for the whole population without evidence then why should drug companies take the risk of funding target finding studies?

Recommendation 13

Statins are used more than the alternatives in large part because they are very heavily promoted. At low doses resins and nicotinic acid may be well tolerated and much more cost-effective and therefore deserve strong promotion in these guidelines. The cost savings would be more than enough to pay for the dieticians and other interventions that we need to really win the battle against cardiovascular disease.

HS Int News index

Comments

Our members can see and make comments on this page.