Skip to main content

Table 2 Treatment ranking measures in the example of all-grade treatment-related adverse events

From: Predictive P-score for treatment ranking in Bayesian network meta-analysis

Treatment Mean (P-score) Median 95% credible interval
Frequentist P-score:
 1 0.365 NA NA
 2 0.821 NA NA
 3 0.677 NA NA
 4 0.174 NA NA
 5 0.096 NA NA
 6 0.944 NA NA
 7 0.432 NA NA
Scaled rank in the NMA:
 1 0.362 0.333 (0.167, 0.500)
 2 0.764 0.667 (0.667, 1.000)
 3 0.780 0.833 (0.500, 1.000)
 4 0.164 0.167 (0.000, 0.500)
 5 0.092 0.000 (0.000, 0.333)
 6 0.924 1.000 (0.667, 1.000)
 7 0.415 0.500 (0.000, 0.833)
Expected scaled rank in a new study:
 1 0.360 0.365 (0.190, 0.509)
 2 0.748 0.749 (0.588, 0.897)
 3 0.757 0.771 (0.488, 0.943)
 4 0.202 0.174 (0.008, 0.573)
 5 0.141 0.124 (0.009, 0.369)
 6 0.873 0.897 (0.613, 0.993)
 7 0.418 0.413 (0.155, 0.729)
  1. Note: NA, not applicable. The posterior means of the scaled ranks in the NMA are the Bayesian P-scores, and those of the expected scaled ranks in a new study are the predictive P-scores