Skip to main content

Table 2 Discrimination power of the tool

From: The development and appraisal of a tool designed to find patients harmed by falsely labelled, falsified (counterfeit) medicines

Health care setting/Medicines True positives a (n) False positives b (n) Cut-off level (scores) of 100: Sensitivity/Specificity (%) (Confidence intervalsc: Lower 95% CL; Upper 95% CL); Positive/Negatived Likelihood ratios (LR+/LR-)
Ambulatory Yerevan (Armenia)/Antibiotics 1 10 ≥40:
Sensitivity 100 (2.5 100)
Specificity 79.95 (65.66 89.76)
LR+ 4.9
LR- 0
  0 2 ≥60
Sensitivity NaN (0 100)
Specificity 96 (86.29 99.51)
LR+ N/A
LR- NaN
Ambulatory Zagreb & Pula (Croatia)/Anti-obesity Preparations, Anabolic Agents for Systemic use, Diuretics, Sex Hormones and Modulators of the Genital System, Urologicals, Psycho-analeptics 1 20 ≥40
Sensitivity 100 (2.5 100)
Specificity 71.01 (58.84 81.31)
LR+ 3.5
LR- 0
  1 13 ≥60
Sensitivity 100 (2.5 100)
Specificity 81.16 (69.94 89.57)
LR+ 5.3
LR- 0
Ambulatory Rome (Italy)/
Anabolic Agents for Systemic use, Diuretics, Sex Hormones and Modulators of the Genital System, Urologicals
3 0 ≥40; ≥60
Sensitivity 100 (29.24 100)
Specificity 100 (93.73 100)
LR+ -
LR- 0
All ambulatory sub-studies
Yerevan (Armenia), Zagreb & Pula (Croatia), Rome (Italy)
   
  5 30 ≥40
Sensitivity 100 (47.82 100)
Specificity 82.86 (76.44 88.12)
LR+ 5.8
LR- 0
Ambulatory sub-studies
Yerevan (Armenia), Zagreb & Pula (Croatia), Rome (Italy)
   
    ≥40
Sensitivity 100 (39.76 100)
Specificity 84.13 (76.56 90.03)
LR+ 6.3
LR- 0
    ≥60
Sensitivity 100 (39.76 100)
Specificity 89.68 (83 94.39)
LR+ 9.7
LR- 0
In-patient Aalst (Belgium)/Anti-obesity Preparations, Anabolic Agents for Systemic Use, Diuretics, Sex Hormones and Modulators of the Genital System; Urologicals, Psychoanaleptics 0 33 ≥40
Sensitivity   0 100
Specificity 21.43 (10.3 36.81)
LR+ N/A
LR-  
  0 11 ≥60
Sensitivity 0 100  
Specificity 73.81 (57.96 86.14)
LR+ N/A  
LR  
In-patient Reykjavík (Iceland)/
Anti-obesity Preparations
0 13 ≥40
Sensitivity 0 100  
Specificity 92.52 (87.01 96.21)
LR+ N/A  
LR-
  0 5 ≥60
Sensitivity 0 100  
Specificity 96.64 (92.34 98.9)
LR+ N/A  
LR-  
  1. Legend
  2. The cut-off level which classified correctly the cases of harm at a minimum of false classifications is highlighted in bold letters
  3. a“True positive”: The initial classification of the patient by the tool as “probably being at risk of health damage caused by FFC medicines” was maintained after the retrospective assessment of the medical record (“reference method”)
  4. b“False positive”: The initial classification of the patient by the tool as probably being at risk of health damage caused by FFC medicines” was rejected after the retrospective assessment of the medical records (“reference method”)
  5. cThe sensitivity and specificity of the tool were expressed as “exact two-sided 95% confidence intervals” according to the method of Clopper and Pearson for the presence of 2 and 3 of 5 risk indicators
  6. dPositive/negative likelihood ratios: the ability of a test to discriminate between people likely to have a disorder and those less likely to have a disorder is determined by the test’s likelihood ratio (LR)
  7. The ratio of true positives to false positives (LR+) is the likelihood ratio for a positive test result being correct
  8. The ratio of false negatives to true negatives (LR-) is the likelihood ratio for a negative test result being correct