Abstruse comparisons : The problems of numerical contrasts of two groups.
The most common quantitative comparison in medical literature is a contrast of two numbers, such as two means or two rates.
The two numbers, A and B, can be compared as a direct increment (A-B), ratio (A/B), relative change ([A-B]/B), or other index of contrast.
To appreciate the quantitative distinction, a reader must know the « setting » reflected by the basic values of A and B. For example, a ratio of 2.0 does not distinguish comparisons between rates of 60% versus 30% and 0.006% versus 0.003%. Despite the frequency of published comparisons, they can he expressed with two types of ahstrusity : quantitatives, if the basic values for A and B are not readily evident ; and qualitative, if the component underlying variables are unfamiliar and not suitably explained.
Among the published articles during the first six months of 1995 for JAMA and New England Journal of Medicine, 57 that satisfied inclusion criteria were reviewed for compliance with standards for avoiding the two types of abstrusity.
The standards for quantitative abstrusity were applied to the published abstract-summary, because it is often the only « sound bite » that is read and remembered by most readers.
The standards for qualitative ahstrusity, however, could be fulfilled in the text, not just in the abstract-summaries of each article.
Among the 57 abstract-summaries, 30% were abstruse quantitatively, and 11 (48%) of 23 pertinent papers were qualitatively abstruse. (...)
Mots-clés Pascal : Homme, Etats Unis, Amérique du Nord, Amérique, Médecine, Document publié, Résultat, Etude comparative, Essai clinique, Méthodologie, Analyse statistique, Interprétation information
Mots-clés Pascal anglais : Human, United States, North America, America, Medicine, Published document, Result, Comparative study, Clinical trial, Methodology, Statistical analysis, Information interpretation
Notice produite par :
Inist-CNRS - Institut de l'Information Scientifique et Technique
Cote : 99-0148867
Code Inist : 002B30A01A1. Création : 16/11/1999.