Hepatitis A—New information on an old virus

✍ Scribed by Chris J. Dickinson

• Publisher: John Wiley and Sons
• Year: 1992
• Tongue: English
• Weight: 408 KB
• Volume: 16
• Category: Article
• ISSN: 0270-9139
• DOI: 10.1002/hep.1840160442

✦ Synopsis

Clinicians have been trained to use statistics to identify differences and to reject apparent differences that may have occurred by chance. The objective of studies such as that by Ramond et al. is not to prove that the groups of patients are different at randomization but that they are similar. Similarity and therefore comparability are not necessarily the same as statistical nondifference. However, although it is reasonable to insist that differences at randomization should not favor the treatment effect, it is obviously unreasonable to insist that the groups should be similar to the point of near identify (i.e., a p value approaching 1). This issue may need to be addressed in a different context.

It has been suggested that steroids should be withheld in the management of patients with alcoholic hepatitis "even in desperation" (3). Ramond et al. have made another dent in that recommendation even if it is not as deep as initial inspection of the data suggests. However, they have served us well by raising another question that interaction with our statistical colleagues may be able to address.