As rheumatologists, we specialize in immunosuppression. Its a knack really. Unfortunately this preoccupation occasionally leads to some rather serious infections. Fortunately its rather uncommon, but if we could reduce the incidence further, well all the better. Vaccination is of course the ideal solution and is in fact recommended by the ACR for all patients taking biologics.That said, there has been debate about the efficacy of the pneumococcus vaccine and this controversy was rekindled by a recent article in the CMAJ.
There are several controversies raised by the present CMAJ study.
The first should be: Just how many meta-analyses are too many? The study itself suggests that there had been seventeen meta-analysis or systematic reviews prior to this one. The editorial in the same issue calculates only twelve. I'm sure there must be several more stewing out there in the statosphere. Isn't there a law or at least a code of conduct amongst epidemiologists against this kind of thing. Maybe a mercy rule. "No answer after ten studies and we call the whole thing off" Did the granting agencies really think that number eighteen would be just the thing to settle the matter.
This study argues that previous studies probably shouldn't have used trial done in 1947. It also suggests that a study from the Papua New Guinea highlands is questionable. Personally, I think that if we are discussing sixty year old studies we are in serious trouble, and if we are including studies from the highlands of New Guinea, applying the results to my city dwelling, Starbucks sipping immunosuppressed RA patients might be a bit of a stretch.
By the way, the CMAJ article concludes that the vaccine doesn't work for adults, high risk or not. The editorial disagrees though, as does the World Health Organization, although neither with any gusto.
If this study has made one thing clearer, it's that another meta-analysis is not going to give us the answer without more data. Preferably data from the last half-century.
Couldn't agree more. The immense proliferation of metaanalyses is disturbing. Is it easier to sit in the office collating and reanalysing old data with ever-increasing degrees of apparent thoroughness than to do the hard yards of going out and conducting well-designed new RCTs that have sufficient statistical power to actually answer important outstanding questions?
Posted by: Hugh Campbell | July 03, 2009 at 06:34 PM