Study Design

Descriptions of key aspects of study design were very poor. No paper gave a justification for the sample size for the study. It was very hard to identify which studies were prospective, in the sense that there was a clear a priori intent to collect data with the specific aim of generating a prognostic model.

It was rarely possible to derive the number of eligible patients, the number included in univariate analyses, and, in particular, the number contributing data to the generation of the prognostic model. Thus, the number of patients with complete data was often not stated, yet missing data can lead to a substantial reduction in overall sample size. The number of events per variable could be calculated for only about half of the studies, many of which had fewer than 10 events per candidate variable.

0 0

Post a comment