Table 2

The test characteristics of the different rules and algorithms for the detection of heart failure

Non-referralReferral
Number of non-referrals
n (%)
Heart failure absent
n
Sensitivity
(%), 95% CI
NPV
(%), 95% CI
LR− 95% CINumber of referrals
n (%)
Heart failure present
n
Specificity
(%), 95% CI
PPV
(%), 95% CI
LR+ 95% CI
ESC guideline, NT-proBNP≥125 pg/mL5106 (29)10092 (83 to 97)94 (88 to 98)0.21 (0.18 to 0.52)259 (71)6834 (29 to 40)26 (21 to 32)1.4 (1.3 to 1.6)
Kelder* diagnostic rule7181 (50)16781 (70 to 89)92 (87 to 96)0.33 (0.20 to 0.53)184 (50)6057 (52 to 63)33 (26 to 40)1.9 (1.6 to 2.3)
Oudejans† diagnostic rule8232 (64)21374 (63 to 84)92 (88 to 95)0.35 (0.24 to 0.52)133 (36)5573 (68 to 78)41 (33 to 50)2.8 (2.2 to 3.5)
Mant et al clinical decision rule6158 (43)14380 (69 to 88)91 (85 to 95)0.41 (0.26 to 0.66)207 (57)5949 (43 to 55)29 (23 to 35)1.6 (1.3 to 1.8)
  • *Dichotomous variable with a cut-off >54.

  • †Dichotomous variable with a cut-off >16.

  • ESC, European Society of Cardiology; LR−, negative likelihood ratio; LR+, positive likelihood ratio; NPV, negative predictive value; NT-proBNP, N-terminal probrain natriuretic peptide; PPV, positive predictive value.