Does poorly reported research make treatments look good?

Nick Freemantle and Greta Rait publish an editorial in the BMJ on a meta-analysis which found that in the cutting edge field of stem cell research in heart disease the more discrepancies found in a paper the greater the improvement in outcome.  On average, trials with many errors show improved ejection fraction, while trials with no errors find no benefit. What should we make of this?

Find the paper here

The authors work with PRIMENT Clinical Trials Unit, a partnership between the UCL Research Department of Primary Care & Population Health, Division of Psychiatry and the Department of Statistical Science.