Reports of new research should begin and end with systematic reviews
The report of a study  to assess the effects of giving steroids to people with acute traumatic brain injury shows how to address all of Bradford Hill’s four questions.
Instructions to authors to put research results in context
issued by the editors of the medical journal The Lancet.
They described the measures they had taken to minimize biases and to achieve adequate control of the play of chance by studying a sufficiently large number of patients. They reported that their study had shown that steroids given to patients with serious brain injury increased the likelihood that these patients would die.
Finally and importantly, they provided readers of their report with all the evidence needed for action to prevent thousands of deaths from this widely used treatment because they updated their original systematic review of previous studies by incorporating the new evidence generated by their study.
Review of the Key points
- A single study rarely provides enough evidence to guide treatment choices in healthcare
- Assessments of the relative merits of alternative treatments should be based on systematic reviews of all the relevant, reliable evidence
- As in individual studies testing treatments, steps must be taken to reduce the misleading influences of biases and the play of chance
- Failure to take account of the findings of systematic reviews has resulted in avoidable harm to patients, and wasted resources in healthcare and research.
More: References (Section 8)
Next Chapter: Regulating tests of treatments: help or hindrance?