Skip to main content

Table 5 Classification matrix for the four mortality prediction systems

From: Assessment of performance of four mortality prediction systems in a Saudi Arabian intensive care unit

  Died Survived  
System PD PS PD PS Sensitivity Specificity PPV NPV OCCR
Cutoff 10%          
   MPM II0 286 24 334 324 92 (87–95) 49 (45–53) 46 (42–50) 93 (90–96) 63 (60–66)
   MPM II24 201 14 270 196 93 (89–96) 42 (38–47) 43 (38–47) 93 (89–96) 58 (55–62)
   APACHE II 206 9 278 188 96 (92–98) 40 (36–45) 43 (38–47) 95 (92–98) 58 (54–62)
   SAPS II 189 26 234 232 88 (83–92) 50 (45–54) 45 (40–50) 90 (86–93) 62 (58–65)
Cutoff 30%          
   MPM II0 235 75 143 515 76 (71–80) 78 (75–81) 62 (57–67) 87 (84–89) 77 (75–80)
   MPM II24 169 46 134 332 79 (73–84) 71 (67–75) 56 (50–61) 88 (84–91) 74 (70–77)
   APACHE II 161 54 107 359 75 (69–81) 77 (73–81) 60 (54–66) 87 (83–90) 76 (73–80)
   SAPS II 140 75 99 367 65 (58–71) 79 (74–82) 59 (52–65) 83 (79–86) 74 (71–78)
Cutoff 50%          
   MPM II0 189 121 58 600 61 (55–66) 91 (89–93) 77 (71–82) 83 (80–86) 82 (79–84)
   MPM II24 142 73 61 405 66 (59–72) 87 (84–90) 70 (63–76) 85 (81–87) 80 (77–83)
   APACHE II 110 105 53 413 51 (44–58) 89 (85–91) 67 (60–75) 80 (76–83) 77 (73–80)
   SAPS II 109 106 52 414 51 (44–58) 89 (86–92) 68 (60–75) 80 (76–83) 77 (73–80)
  1. Figures in parentheses are 95% confidence intervals. PD, predicted to die; PS, predicted to survive; PPV, positive predictive value; NPV, negative predictive value; OCCR, overall correct classification rate.