A simulation study of the number of events per variable in logistic regression analysis.
Citations
7,895 citations
Additional excerpts
...When outcomes are either binary or time-to-event in nature, prior research has suggested that at least 10 events should be observed for every covariate that is entered into a regression model ( Peduzzi, Concato, Feinstein, & Holford, 1995 ; Peduzzi, Concato, Kemper, Holford, & Feinstein, 1996 )....
[...]
4,211 citations
2,982 citations
2,943 citations
Cites background from "A simulation study of the number of..."
...(2, 3) more closely, we also examined models with all binary predictors....
[...]
2,486 citations
Cites methods from "A simulation study of the number of..."
...Statistical issues for future research into meta-regression methods include how the number of covariates that can reliably be included depends on the number of trials (and their imprecisions) [38], the handling of multi-arm trials when individual patient data are not available, the appropriate use of regression diagnostics and sensitivity analyses [39; 40], and whether there are biases in using derived statistics measured after baseline as covariates....
[...]
References
10,964 citations
2,890 citations
1,650 citations
1,087 citations
643 citations