Skip to main content

Table 1 Numerical and graphical presentation of model (n = 47)

From: Reporting performance of prognostic models in cancer: a review

 

% (n) articles

Statistical model used

 

   Cox only

88 (41)

   Cox plus other (two RPA, one ANN)

6 (3)

   Other (one Weibull, one RPA, one unclear)

6 (3)

Assumption of proportional hazards tested†

21 (10)

Final prognostic model reported*

96 (45)

Regression coefficient reported**

72 (34)

Reproducibility of model development assessed††

11 (5)

   Model with same variables, not same coefficients

9 (4)

   Model generating both new variables and coefficients

4 (2)

  1. † In three articles assumption of proportional hazards was not applicable as models used were RPA (recursive partitioning analysis) and ANN (artificial neural network).
  2. * Two articles did not report the final model
  3. ** Not applicable in two articles using RPA or ANN model
  4. †† One article used both methods to examine model reproducibility.