Servicios Personalizados
Revista
Articulo
Español (pdf)
Articulo en XML
Referencias del artículo
Traducción automática
Enviar articulo por emailIndicadores
Citado por SciELO 
Accesos
Links relacionados
Citado por Google 
Similares en
SciELO 
Similares en Google
Compartir
Permalink
CES Medicina
versión impresa ISSN 0120-8705
Resumen
BENAVIDES-ZORA, David; SALDARRIAGA-BETANCUR, Santiago y DONADO, Jorge Hernando. The Fragility Index and its characteristics in randomized clinical trials of diabetes mellitus. CES Med. [online]. 2022, vol.36, n.2, pp.106-121. Epub 03-Oct-2022. ISSN 0120-8705. https://doi.org/10.21615/cesmedicina.6632.
Introduction:
to evaluate the robustness of the results, a tool called the Fragility Index (FI) has been proposed, which is defined as the minimum number of patients that have to be changed from "No events" to "Events" in the intervention group to change a statistically significant to nonsignificant result. Showing that among a lower Fragility Index, the results of the trial will be less robust or more fragile. Different authors have found that the significance of the results of many controlled clinical trials (RCTs) depend on very few events. The objective of the study is to evaluate the FI of controlled clinical trials in diabetes mellitus in five of the general medical journals with the greatest impact factor worldwide.
Methods:
an electronic search was conducted in PubMed to identify randomized clinical trials in The Annals of Internal Medicine, BMJ, The Lancet, The New England Journal of Medicine and JAMA. Clinical trials were reviewed with diabetic or prediabetic patients and the FI was then calculated for each outcome according to the method described by Walsh et al, using 2x2 contingency tables. A priori was planned to use the Spearman correlation coefficient to evaluate the direct correlation between the Fragility Index and sample size, number of events, p-value and follow-up time. The significance of all the results was evaluated with a value of p <0.05.
Results:
the median Fragility Index was 11, and in three studies (7.3%) the result were not statistically significant after recalculating the p value with Fisher's exact test. A slight direct relationship between the number of events and the Fragility Index (Rho = 0.343, p = 0.02) was found and a moderate inverse correlation was observed between the p value and the FI (Rho = -0.632, p = 0.000 ). No statistically significant correlation was found between sample size, follow-up time and losses with the FI.
Conclusions:
in controlled clinical trials on diabetes, we found that the statistically significant results depend on a few events, evidenced by a low value in the Fragility Index. The values of this measurement are related to the number of events and negatively to the p value.
Palabras clave : fragility index; diabetes mellitus; controlled clinical trials.
