Preventing stroke with ramiprilBMJ 2002; 325 doi: https://doi.org/10.1136/bmj.325.7361.439 (Published 24 August 2002) Cite this as: BMJ 2002;325:439
Results should have been presented in ways that help practising clinicians
- P Badrinath, specialist registrar in public health medicine (email@example.com)
- Suffolk Public Health Network, St Clements Hospital, Ipswich IP3 8LS
- Burntwood, Lichfield and Tamworth Primary Care Trust, Lichfield WS13 6JB
- Langton Medical Group, St Chad's Health Centre, Lichfield WS13 7HT
- Department of Medicine, Diabetes and Cardiovascular Disease Academic Unit, Royal Free and University College London Medical School, London N19 5LW
- Timmins and District Hospital, Timmins, Ontario, Canada P4N 8R1
EDITOR—I was surprised to see that the BMJ published a trial that presented the results in a way that exaggerates the findings.1 Stroke prevention, the topic under discussion in the paper by Bosch et al, is important for patients, doctors, and funders of care. Hence the results should have been presented in a way that would help practising clinicians—by giving numbers needed to treat (NNT) along with relative risk reductions (RRR). The authors report a relative risk reduction of 32% in all strokes and of 61% in fatal strokes. For all strokes this translates into a number needed to treat of 67 for four and a half years' treatment.
Evidence shows that the way results of clinical trials are presented influences both physicians and funders of health care. 2 3 In the randomised controlled trial by Bucher et al, doctors gave higher ratings for the effectiveness of the drug and were more inclined to prescribe lipid lowering drugs when the results were presented as relative risks.2 A study from a health authority in the United Kingdom reported that health authority members' willingness to purchase services was influenced by the methods used to present results.3 Interestingly, both these papers were published in the BMJ.
The problem of biased reporting of clinical trials is not a new phenomenon. Pocock et al in their survey of three medical journals in 1987 found that, overall, the reporting of clinical trials seems to be biased towards an exaggeration of treatment differences.4 What do the CONSORT guidelines say?5 The following quote may be relevant here: “For both binary and survival time data, expressing the results also as the number needed to treat for benefit (NNTB) or harm (NNTH) can be helpful.” The two citations supporting this …