Glucosamine and Chondroitin

I have bad knees (too much running). I have used glucosamine in the past and indeed still have some in the cupboard. It seems I have wasted my money. But I am not alone and the global sales of glucosamine reached $2bn in 2008. A detailed meta-analysis has now been performed and has shown no significant benefit for either of these medication in preventing hip and knee pain (http://www.bmj.com/content/341/bmj.c4675). Let’s have a look in more detail.

Glucosamine and chondroitin are constituents of cartilage. They can be absorbed orally and the theory is that they make their way to the joints and help reconstitute worn-out cartilage. As is often the case with “alternative therapies” there are a number of studies which have suggested they help. Most of these studies have been in relatively small numbers of patients and, from a methodological approach, of poor quality. The better the study, it seems, the smaller has been the measured effect (if any has been detectable at all).

The researchers identified 10 suitably sized studies with adequate methodology, which included 3803 patients and combined the results in a  meta-analysis. Although meta-analysis has problems of its own as a technique, it is still generally regarded as the gold standard when determining whether a treatment has an effect or not.

The conclusions of the authors were fairly damning – there was no clinically relevant improvement in pain or joint space in patients taking glucosamine, chondroitin or a combination of the two. On the bright side there was no evidence of harm. The study in the BMJ has been backed up by conclusions from the Cochrane Collaboration, who have reviewed glucosamine (http://www2.cochrane.org/reviews/en/ab002946.html) and also by the European Food Safety Authority (EFSA) who have examined the use of glucosamine alone or in combination with chondroitin (http://www.efsa.europa.eu/fr/scdocs/scdoc/1264.htm).

This doesn’t mean that these treatments don’t work though – just that we have no robust evidence that they do. One positive is that the treatments seem unlikely to do any harm – except to your wallet.

CURE (http://www.nejm.org/doi/pdf/10.1056/NEJMoa010746)

So I have been thinking about relative and absolute risk recently. In cardiology we make decisions that are “evidence-based” and we are fond of quoting relative risk. But the reality is that the differences we make with each treatment change outcomes a few percent either way and that we have to treat tens or hundreds of patients to improve the outcome for just one. I admit that the following blog posts do take a slightly jaundiced view…

Let’s start with the CURE (the Clopidogrel in Unstable Angina to Prevent Recurrent Events) study. This arguably is the trial which has ensured that patients with an acute coronary syndrome (http://en.wikipedia.org/wiki/Acute_coronary_syndrome), not ST-elevation myocardial infarction , receive both aspirin and clopidogrel. Clopidogrel is a thienopyridine derivative that inhibits platelet aggregation induced by adenosine diphosphate.

12,562 patients were enrolled. Patients had to be within 24 hours of admission. They had to be over 60 years old. They had to present with a story of angina and either ECG changes (not ST-elevation) or elevation of cardiac biomarkers (or both). There were the usual, predictable exclusion criteria, the most notable of which were patients with severe heart failure and those who had undergone recent revascularisation. They were divided into two groups and the groups were well matched for basic epidemiological characteristics.

Patients were given a loading dose of clopidogrel (300mg) and then 75mg daily. The mean duration of treatment was 9 months. The study was performed between 1998 and 2000 when revascularisation was not routine (as it is today).

It is arguable that the only important endpoints are death and quality of life. But that is too simplistic, and a variety of measures are required, including those relevant to health economics. For this study the primary outcome measure was a composite endpoint – death from cardiovascular causes, nonfatal myocardial infarction, or stroke. A number of other endpoints were also recorded.

The primary endpoint was of course positive. The primary endpoint occurred in 582/6259 (9.3%) patients in the clopidogrel group vs. 717/6303 (11.4%) patients in the control group. This is quoted as a relative risk of 0.80 in the paper, with a p value of <0.001. Relative risk (http://en.wikipedia.org/wiki/Relative_risk) can be calculated as follows: (582/6259)/(717/6303) = 0.815…

This is an absolute difference of 2.1%. The number needed to treat (http://en.wikipedia.org/wiki/Number_needed_to_treat) to prevent one primary endpoint is therefore 47.

But let’s look at some other endpoints and also harm; aspirin and clopidogrel compared with aspirin causes more bleeding.

Death occurred in 359/6259 (5.7%) patients in the clopdiogrel group vs. 390/6303 (6.2%) patients in the control group. This difference was not significant. There is no evidence from this trial that you save lives with clopidogrel.

Major bleeding episodes were defined as substantially disabling bleeding, intraocular bleeding leading to the loss of vision, or bleeding necessitating the transfusion of at least 2 units of blood. 231/6259 (3.7%) patients in the clopdiogrel group vs. 169/6303 (2.7%) patients in the control group experienced a major bleed. The absolute difference was 1%. The number needed to treat to cause one additional major bleed is 99.

In summary, the data from this trial suggest that if you treat patients with an acute coronary syndrome with aspirin and clopidogrel rather than aspirin alone for 9 months, no deaths are prevented, 2 people have one less death from cardiovascular causes/nonfatal myocardial infarction/stroke and one additional person will have a significant bleed.

How often do you inform patients of this when prescribing clopidogrel in this situation?