Fraud, fraud, fraud
Both the BMJ and Lancet last week had multiple articles about suspected fraud. They involve a series of alleged trials run by by Dr. Ram Singh from India that are suspect on a number of levels: his research output was prolific, particularly from someone in an impoverished area with minimal facilities, no association with a university or major medical center and no outside funding. (article from India times with response by Dr. Singh here)
There is now overwhelming evidence that his papers are, at best, deeply flawed and probably made up. To list some of the problems identified many data are missing, there are statistical anomalies and many other Indian researchers have stated that the type of research reported (e.g. detailed diet diaries) are improbable in a poor, largely illiterate population.
Remarkably, the BMJ study was published in 1992 and despite years of concerns and ongoing investigations, the journal has just now come forward with its concerns. Unfortunately, the Lancet would almost certainly not of published its article in 2002 had it known of the concerns.
It seems to me there is a inexplicable reluctance to question suspect reasearch and to label it as fraudulent (or wrong in the case where fraud isn't clear). In every field I've been in there are reports in the literature that those "in the know" are aware are simply wrong. Sometimes it is fraud, others people just f*^&ed up experiments, but no one ever demands a retraction or investigation.
While this may not matter for an obscure basic science study, it can wreak havoc in clinical work, particularly with increasing reliance on meta-analysis.
Another problem is who investigates once concerns are raised. Journals say they don't have the resources or authority, but other potential investigators have their own problems: specifically, universities and institutes have conflicts of interest and limited resources and government bodies aren't usually equipped for such work. Given the incredible demands journals put on authors during review it would seem they could take a bit more responsibility for ensuring that studies withstand questions that arise and air any doubts more quickly.
I think this is a major issue for science and one that no one really wants to deal with. I know I don't have great confidence in what is published, partly because of fraud and partly because of sloppiness. But someone needs to take responsibility for determining the validity of published research once questions are raised.
UPDATE: more on fraud and a less than vigorous response to it here. To be fair, the acts described seem a bit short of meeting what I'd call criteria for fraud. Researchers seem to have padded their grant applications and violated an obscure NIH regulation, not fabricated data.