Trials of autologous bone marrow stem cells for heart disease

The report by Nowbar and colleagues (doi:10.1136/bmj.g2688) on the relation between the number of discrepancies observed in clinical trials of autologous bone marrow stem cells for ischaemic heart disease and ejection fraction includes shambolic and poorly conducted research that reflects poorly on every part of the research process.1 On average, trials with many errors show improved ejection fraction, while trials with no errors find no benefit. How should we react to this finding and what lessons can we learn? In An Essay on Criticism, Alexander Pope stated that “to err is human,” but surely there is no place for errors in development of evidence that will inform treatment decisions?2

Errors range from the unfortunate typo to poor design, conduct, analysis, and reporting. Many of the errors described by the authors could be explained and are perhaps less harmful than Nowbar and colleagues imply. There are many descriptions of “impossible” percentages in their paper.1 The authors are quite correct, for example, to say that 65% of 42 participants does not give an integer result, but 65% of 40 participants does and the trialists might simply be correctly accounting for missing data in their …