Diagnostic tests
Resources for teaching LR etc
Even when studies are published, they often omit important elements that enable readers to assess and apply the findings.
One review of 519 randomized trials published in reputable journals during December 2000 found that 82% did not describe the process of allocation concealment and 52% did not provide details of measures to reduce [getit term=”bias, measurement”]observer biases[/getit] – both features that are crucial to good studies. [10]
This poor reporting of details extends even to the description of the treatments used. A trial showing that giving a specific booklet (compared with no booklet) helped patients with irritable bowel syndrome, omitted to describe the contents of the booklet or how to obtain it; the ‘treatment’ could therefore not be used by any other patients or doctors. This was just one example in an analysis of trials in major journals that found about a third omit such crucial details. [11]
Finally, most published trials do not set their results in the context of previous similar trials. Without this key step, it is impossible to know what the results actually mean.
Four-yearly checks of randomized trials reported in five major medical journals over a period of 12 years – 1997-2009 – illustrate the extent of the problem. Overall, only 25 of 94 (27%) reports made any reference at all to systematic reviews of similar trials. Only 3 of 94 reports actually contained updated reviews integrating the new results, and so showing what difference the new results had made to the totality of evidence. Sadly, there was no evidence of improvement in reporting practice with the passage of time. [12]
This failure can lead to clinicians using different treatments depending on which journals they happen to read.
Next: Blueprint for a better future
Discussion
0 Comments