From evidence- to science-based medicine

Kault, David (2014) From evidence- to science-based medicine. Annals of the Australasian College of Tropical Medicine, 15 (3). p. 53.

[img]
Preview
PDF (Abstract Only) - Published Version
Download (807kB) | Preview
View at Publisher Website: https://tropmed.org/publications/annals/...
 
53


Abstract

Background/Aims: Evidence-based medicine (EBM) tends to dismiss treatments which lead to non-statistically significant results and labels such treatments as having 'no effect'. If all such treatments really had no effect, there would be as many non-significantly negative results as non-significantly positive results. In practice there is a considerable excess of weakly positive results. This presentation suggests possible ways of using data on the distribution of summary statistical values to improve the EBM paradigm.

Methods: A sample of 100 summary statistical values was chosen randomly from the Cochrane collection for evidence based medicine. Several models were used to analyse the data.

Results: Weakly positive values outnumber weakly negative by about 3:1, suggesting that about two-thirds of weakly positive results are being produced by treatments which have at least a small positive effect rather than deserving the 'no effect' label. More complex models of this data agree with this estimate, albeit with wide confidence intervals.

Conclusion: With further data collection and research, it will become possible to give a probability that a treatment is effective rather than assign dichotomous labels 'effective' and of 'no effect' to treatments. Knowledge about the treatment aside from the statistical result, and knowledge of the cost of making a wrong assessment of treatments can be further incorporated into decision making, so that treatments are based on science and not just the vagaries of statistical assessment.

Item ID: 35981
Item Type: Article (Abstract)
Keywords: evidence Based Medicine, false negatives, Bayesian prior
Related URLs:
ISSN: 1448-4706
Date Deposited: 11 Nov 2014 01:39
FoR Codes: 01 MATHEMATICAL SCIENCES > 0104 Statistics > 010402 Biostatistics @ 100%
SEO Codes: 92 HEALTH > 9201 Clinical Health (Organs, Diseases and Abnormal Conditions) > 920102 Cancer and Related Disorders @ 100%
Downloads: Total: 53
Last 12 Months: 1
More Statistics

Actions (Repository Staff Only)

Item Control Page Item Control Page