I’ve only been at this a few days, and already a repost! Here’re some comments on some articles in the Annals of Internal Medicine from earlier in the month. I think these are (or should be) of interest to the rheumatologist. (Login may be required to read these.)
The first looks at acupuncture for knee osteoarthritis (OA). It had three arms: one received 10 treatments of traditional Chinese acupuncture, one got 10 sham needling treatments, and the third received 10 physician visits. All these took place within 6 weeks, and all patients got nonsteroidals and physical therapy. The primary outcome was a 36% improvement in the WOMAC (a measure of pain). Results: No difference in response in the first two groups (about 52% responded in each group), but both were better than physician visits (29% response). What does it all mean? As far as I can see, it means that the placebo effect is a pretty powerful thing, but that acupuncture itself ain’t that useful. You wonder how they’d compare after 12 or 18 weeks rather than just 6. I bet the effect goes away.
The second article looks at C-reactive protein (CRP) as a predictor in coronary disease. Rheumatologists know that lots of cardiologists are all hot about this as the new thing for predicting ischemic heart disease, but lots of us rheumatologists aren’t so sure. The article seems to back up the rheumatology position. The authors basically looked at whether or not CRP added anything to a standard assessment using the Framingham risk score. They point out that most studies have looked at CRP as an independent risk factor, find a low to middling odds ratio for predicting coronary disease (OR about 2.0-3.0), and conclude that it’s useful. The authors point out that the real question is not whether CRP is independently associated with ischemic heart disease, but whether or not it actually adds anything to the current prediction model, and if so, is it enough to be clinically useful.
Basically, they say the hype is overblown. If you add CRP to traditional risk factors, you don’t get a statistical difference in the c-statistic (which, as I understand it, is a measure of how well the model differentiates an individual who will get the disease from one who will not). And even if you divide subjects into low and high risk groups by Framingham criteria, and then see whether CRP helps predict disease when high in the first group, or predict no disease if it’s low in the second group, it doesn’t seem to help more than the Framingham risk alone.
This has been my suspicion all along, and it’s nice to see an analysis that seems to be looking at the right parameters.
The last article is a review of noninferiority trials. I’m not going to try to summarize it, you can read it for yourself, but I think it’s important because I’m anticipating that rheumatologists are going to see a lot more of these in the future. Now that we have some effective treatments for rheumatoid arthritis and other autoimmune diseases, manufacturers are going to be more interested in showing that new drugs are as good as what’s already out there rather than just showing that they’re better than placebo. And besides, it’s going to be more and more difficult to use placebo comparisons when we clearly have effective treatments. My point is that we need to understand how these trials work, and how to interpret the results. I’d encourage you to read it.