Skip to main content

Table 2 Analyses of the predictive accuracy of the various models

From: A predictive model for respiratory syncytial virus (RSV) hospitalisation of premature infants born at 33–35 weeks of gestational age, based on data from the Spanish FLIP study

  True Positive False Positive False Negative True Negative Sensitivity Specificity PPV
%
NPV
%
LR Diagnostic Accuracy %
FLIP 15 variable model § 130 102 53 265 0.71 0.72 56 83 2.56 72
FLIP Final 7 variable model ¤ 139 113 45 254 0.76 0.69 55 85 2.45 71
Munich 6 variable model 14 106 4 247 0.78 0.70 12 98 2.59 70
  1. § Records for 550 infants were included within the analysis. Seven records were dropped from the analysis due to missing data for one or more of the predictor variables
  2. ¤ Records for 549 infants were included within the analysis. 8 records were dropped from the analysis due to missing data for one or more of the predictor variables
  3. Records for 370 infants were included within the analysis. Three records were dropped from the analysis due to missing data for one or more of the predictor variables. Two records for hospitalised cases were removed from the analysis, as they each had one negative RSV test
  4. PPV = positive predictive value
  5. NPV = negative predictive value
  6. LR = likelihood ratio of a positive test; for information about likelihood ratios see reference 25
  7. Standardised canonical discriminant function coefficients for the FLIP final 7 variable model: birth ± 10 weeks of start of season = 0.678, birth weight, kg = 0.184, breast fed ≤ 2 months or not = 0.511, number of siblings ≥ 2 years = 0.489, number of family with atopy = 0.151, female sex = -0.113, number of family with wheeze = 0.125