Prior studies of cholesterol-lowering therapy in patients without overt coronary disease (primary prevention studies) have shown a reduction in coronary events but not in coronary mortality or all cause mortality and have raised questions about increases in non-cardiovascular morbidity and mortality. This study was designed to examine the effect of an HMG-CoA reductase inhibitor, pravastatin, in the primary prevention therapy of hypercholesterolemia in middle-aged men.
Screening clinics established at primary care facilities in West of Scotland.
The authors note that , in contrast to other primary prevention trials which used less powerful cholesterol-lowering agents, the reduction in events was noted within 6 months of initiation of therapy. They estimate that, by treating 1000 middle-aged men, there would be 14 fewer angiograms, 8 fewer revascularizations, 20 fewer non-fatal MI's, 7 fewer cardiovascular deaths and 2 fewer deaths from other causes.
1) Treatment of hypercholesterolemia in middle-aged men without a history of MI (primary prevention) is effective in reducing coronary event rates (CHD death plus non-fatal MI). This is not a new finding.
2) In this study, treatment also reduced the "probable" CHD death rate, although it did not achieve statistical significance for the definite CHD death rate. This is a new and significant finding.
3) Total death rate also was reduced, but it missed statistical significance by a hair. In a previous summary (summary #4 ) I was rightly taken to task for commenting on a trend that did not reach statistical significance; the data is so close here, however, that I will commit this error again, knowingly this time, at the risk of further approbation.
4) There was no evidence for an increase in non-cardiovascular deaths or in cancer incidence after five years of treatment with the study drug. This does not eliminate the possibility of an increase in cancer after a longer period of time, but is reassuring, nonetheless.
What population is this applicable to?
The study included middle-aged men, 45-64 years old, from an area of Scotland with a particularly high incidence of coronary disease. The study revealed a similar degree of reduction in CAD event rates in various subgroups with different a priori likelihoods of coronary events, indicating that these results might be extrapolatable to lower risk populations (younger men; women). Nevertheless, the absolute benefit, in terms of number of events prevented per 1000 treated, is highly dependent on the population's initial CAD risk. Preventing coronary disease in a lower risk population will require treating more people for longer periods of time. This needs to be taken into account, particularly when cost-benefit analyses are undertaken.
Would all of these men have been treated with drugs if NCEP guidelines were followed? Most, but not all. The guidelines recommend drug therapy (after diet) for LDL cholesterols > 190 or if LDL is >160 and there are two or more risk factors. About half of the men here had LDL's > 190. The other half were > 160. All had one risk factor (male, age over 45), probably about 2/3 had another risk factor (smoking or hypertension or diabetes) or had evidence of CAD. Thus, roughly 1/2 + (2/3 x 1/2) = approximately 5/6 would have been candidates for drug therapy by NCEP guidelines.
What drugs does this apply to?
One of the biggest problems is extrapolating the results of a study such as this to a particular drug or class of drugs. Is the protective effect due only to the actual, numerical reduction in cholesterol? Is there something else that we're not measuring, an effect on a cholesterol sub-fraction? Do the results apply only to pravastatin, only to the statins, or to some groups of drugs but not others?
In my opinion, the beneficial effect of the drug is most likely to be the result of the cholesterol lowering numbers, since other studies including diet and other drugs have shown reasonably similar effects. The safety information from this study, in particular the lack of a significant increase in cancers, is applicable mainly to the statins. One exception to this is the concern that has been raised about cholesterol-lowering and cancer, in general. This study seems to refute this hypothesis, at least after five years of follow-up.
Is there not a concern with this type of study that the results could be caused by the fact that the patient knows that the study is looking at CAD etc and cholesterol is a risk factor?
The Hawthorn effect?
In the case of this study, since the interventions were the same for placebo and treatment groups, the difference between these two groups should be real (and not due to any interventions common to both groups). There is one aspect of this study that is relevant to the Hawthorne effect, however: the intervention and regular follow-up makes medication compliance much more likely. The actual compliance with treatment is likely to be higher than it would in "real life". As a result, the quantitative "number of events prevented" per 1000 treated could well be higher than it would be in practice. -- mj
Date: Tue, 23 Jan 96 20:47:37
Dr. Brian O'Mahony, Family Practitioner
Health Centre, Lismore, County Waterford, Ireland.
Comments on "Prevention of coronary heart disease with pravastatin in men with hypercholesterolemia."
Thanks for including this study in the Journal Club. A couple of things about this study have been worrying me.
I have been trying to use the McMaster University method of calculating "number needed to treat (NNT)" to get an idea of the implications of this study for primary care. Using the formula NNT=1/(X-Y), where X is the risk of non-fatal MI or death in the placebo group and Y is the risk in the treatment group.
Taking figures from Table 2. End Points of the Study:
X = 248/3293 = 0.075
Y = 174/3302 = 0.053
(X-Y) = 0.022
NNT = 1/(X-Y) = 45
If my calculations are correct then I would need to treat 45 patients with pravastatin 40mg daily for 5 years to prevent one event (non-fatal MI or death).
Please correct me if I am wrong. Can I use this NNT calculation in a study going on over 5 years with a 30% withdrawal rate over this period?
This brings me to the cost implications of this study. The cost of pravastatin 40mg daily in Ireland is approximately 90 US $ a month. Yearly cost per patient is approximately 1,000 dollars. So 45 patients for 5 years would cost US $225,000, in drug costs alone, to prevent one event. What are the opportunity costs of this amount of money. What effects could a 'stop smoking programme' or 'exercise programme' have in preventing one event and how much would it cost?
Thanks for your time.
These numbers already take into account the 30% dropout rate at the end of 5 years. If, in reality, a greater proportion of patients dropped out, you would need to treat a higher number of patients to achieve the same benefit; if, on the contrary, patients were more compliant (unlikely), you would achieve the same benefit by treating a smaller number of patients.
As for the cost per event treated, if 30% drop out over 5 years, the number of events prevented is as calculated here, but the cost would be lower due to less drug actually taken. Assuming a relatively linear dropout rate, the average cost for the drug would be about 15% lower than if there were no dropouts ($195,000 instead of $225,000).
When looking at costs, it is also important to note that in addition to reducing CAD mortality and non-fatal MI, there was a similar reduction in coronary angiography plus revascularization procedures. Thus, very roughly, assuming $90 / month for the medication, it might cost about $100,000 to prevent one CAD death or nonfatal MI or angiogram or revascularization.
I'm not quite sure what the cost-benefit numbers are for hypertension therapy or smoking cessation; I'll try to get them. Perhaps someone else might be able to dig them up. -- mj
Date: Mon, 29 Jan 1996
From: fyacoub (u34750@UICVM.CC.UIC.EDU)
I have to admit that I am biased against primary prevention using cholesterol lowering therapy . The cost is great, and the benefit is marginal at best, I truly believe it is not cost effective. I think it is better to direct our efforts into stopping smoking, treating hypertension and diabetes, and improving diet and exercise among general population.
Date: Sun, 03 Mar 96
From: "Neil J. Stone, M.D." <firstname.lastname@example.org>
What makes the West of Scotland study significant is that it doesn't stand alone, but continues a trend of showing important decreases in coronary death rates in patients who have their cholesterol and more precisely their LDL-cholesterol lowered by diet and drug therapy.
The study is important because with its more than 6000 men it had the statistical power to look at total mortality that smaller trials did not. The lack of an increase in total mortality (if anything, a strong trend toward a decrease) is reassuring that this therapy does not have adverse consequences in the short term.
When one looks at the marked improvement in fatal and non fatal MIs, one realizes that the decline in age adjusted MI in this country has the effect of pushing MIs back into an older decade. This is an important effect which in itself has economic consequences that are not taken into account by the "models".
Date: Sun, 03 Mar 1996
From: David Winsemius <email@example.com>
Subject: Statins effect on all cause mortality
It seems to me that the time is at hand for a meta-analysis with all-cause mortality using cholesterol lowering manuevers. CHD "events" are interesting, but why is a CHD death any different than a non-CHD death. For pete's sake, these studies have several thousands of subjects.
If a life is valued at US$100,000, which is typical choice for these kind of analysis, and a typical MI costs US$20-40,000, the cost figures mentioned here suggest that the US$225,000 being spent on those 45 persons who neither died nor had MI's may not be a health-care bargain.
David Winsemius, MD,MPH
More importantly, since previous meta-analyses have found different
magnitudes of treatment effect, in the same issue of the AJC (pp 122C-126C)
two "prospective", clearly defined, large-scale meta-analyses are described
which will examine trials whose results are not yet known. -- mj
Date: Wed, 27 Mar 96
From: "Larry Hill, MD" <firstname.lastname@example.org>
I am not able to get readily a copy of this edition here in Manila.
Question - are there data comparing the dietary habits of the two groups,
i.e., did the drug treated group follow dietary instructions any more closely
than the placebo group. I suspect not. That then brings up the corollary
question that I have asked without getting an answer based in science and
which patients often ask me: "Now that my cholesterol is (say) 175 on medicines,
having been 250 before, can I go back to my fried eggs and bacon?" i.e.,
can the statins be used as "cholesterol sponges?" If we assume that the
benefits are indeed related to the lowered cholesterol, the answer should
be a qualified yes, the qualification being that levels should be followed
with the dietary liberalization. I'm anxious to hear the opinions of others
on this most practical question.
From a standpoint of CAD prevention, a given cholesterol level obtained by drugs is probably equivalent to one obtained by diet. How strict a diet a patient should follow when on an antilipidemic agent depends on a number of factors, in my opinion. Maintaining a fairly strict diet should reduce the drug dosage required to achieve a given result; since all medications have potential for side-effects that may be dose-related and since medication cost is certainly dose-related, the less medication a patient needs to take, the better. Not all patients' lipid levels respond to diet the same way, even when on medication. This is another factor that needs to be taken into account. Finally, high-fat diets carry other potential risks (weight gain and possible carcinogenicity), while some patients find dietary restriction more troublesome than others.
Bottom line, in my opinion: diet first, diet plus drugs next, then liberalize diet if the patient feels strongly about it, while watching what happens to lipid levels.
I would also be interested in other peoples' approaches. -- mj
Date: Mon, 08 Apr 1996
From: Colin Rose <email@example.com>
Rarely mentioned amid all the hoopla about cholesterol lowering drugs is a study from France (Delorgeril M et al. Mediterranean alpha-linolenic acid-rich diet in secondary prevention of coronary heart disease. Lancet 1994; 343:1454-1459) which showed an 80% reduction in cardiac events and a 60% reduction in total mortality with NO change in total cholesterol using the Mediterranean diet.
Also ignored is the followup of the original 7 countries study (Verschuren WMM et al. Serum total cholesterol and long-term coronary heart disease mortality in different cultures - twenty-five-year follow-up of the seven countries study. Jama 1995; 274:131-136) which showed the very weak association of cardiac mortality with total cholesterol and the very strong association with country of origin. These two papers are essential reading for anyone advising patients with atherosclerosis or who are at risk for same (virtually everyone in North American society).
Clearly the biggest risk factor for atherosclerosis is not your cholesterol, not your blood pressure, not your family history, not your smoking habit but your address. There is some very large negative factor which is absent or some very large positive factor which is present in southern Mediteranean and far eastern countries which has still not been identified. On the practical level I tell patients to adopt the Mediterranean diet and largely ignore the cholesterol. If they insist on taking a pill also I tell them that it is unlikely to prolong their lives. Usually their angina improves substantially or disappears just on the diet. NO pills, NO angioplasty, NO bypass. And they can acutally SAVE money!
Subject: pravastatin study
Date: Fri, 19 Apr 1996
This paper used relative risk reduction to express the results of the study. The absolute risk reduction, which, in my opinion, is less misleading, was a little more than 2%. My take on the findings was that if one treats 44 men with no previous MI, no abnormal ECG, and elevated cholesterol, and treats them for five years, one study- defined cv event will be prevented. Period. The drug acquisition cost alone for five years of pravastatin 40mg/day for 44 men for five years is almost $300,000 (Canuck; less in the States).
The statins are a physiologically elegant class of drugs that have an
impressive impact on lab results. What their role in therapy is, is yet
to be proven.
Letters to the Editor about this article from the NEJM
Many of the points raised in the discussion here were addressed in these letters to the editor.
Date: Wed, 05 Jun 1996
From: "R. Hal Baker" <firstname.lastname@example.org>
Organization: ACP Online
Why the West of Scotland? Why require several followup visits before enrolling the patients in the study? Because this study was really designed to answer the question "Can we find a group of patients at such risk for CAD and MI that it is possible to demonstrate a benefit from primary prevention?"
This is an efficacy study, not an effectiveness study. It affirmatively answers the question "Is it possible to show a benefit from primary prevention?" but that does not mean that the same benefit can be realized in your patient population in the US or elsewhere, where previous trials have failed to show the difference.
R. Hal Baker, MD
Date: Sat, 07 Sep 1996
From: "Dr. C. J. Martin" < email@example.com>
One of the principal flaws in this study was the inclusion of subjects with pre-existing heart disease. At the very least this would account for a reduction of 5 MI deaths in the treated group out of ?40.
It is correct that in this trial, the strict criteria for primary prevention were not met since 5% of patients had angina. However, one cannot directly extrapolate the patients from the 4S study to patients with angina included here, because 4S included patients with prior MI's whereas they were excluded in the WOSCOP trial.
Furthermore, when the subgroup without prior vascular disease (including patients with angina) was looked at, the authors found an equally significant reduction in the primary endpoint (definite MI and CHD death).
Thus, although the entry criteria do not strictly fit the definition of primary prevention, this departure does not seem to have significantly altered the results as far as applicability to primary prevention are concerned. -- mj
Subject: WOSCOPS, Absolute risk calculations
Date: Fri, 20 Dec 1996
From: Troels Thomsen <firstname.lastname@example.org>
Organization: Center for Sygdomsforebyggelse
A simple question regarding the absolute risk calculations in WOSCOPS:
How is the "absolute % risk at 5 years" calculated? From Table 2 in the article, if you divide the number of events by the number N of patients, you will get a result approximately 5% lower than that shown in the table. (e.g. non-fatal MI or death from CHD: 248 x 100 / 3293 = 7.5 and not 7.9). If you do the same in the 4S study there is no such problem.
Can anyone explain?
Troels Thomsen, MD, Glostrup, Denmark
Journal Club on the Web home page
Submit a comment about this