Table 3 Performance of the developed algorithm to detect relevant bacteriuria (≥ 105 colony-forming units/ml) for all samples, samples of male and female patients and for samples of different divisions, respectively
All samples (47,572) | Female (24,923) | Male (22,649) | Emergency& Outpatient (21,496) | Medicine (7968) | Surgery (7580) | Intensive care (2984) | Gynaecology & Obs. (3154) | Paediatrics (3305) | External (1048) | |
|---|---|---|---|---|---|---|---|---|---|---|
Rule-out (< 50 bacteria/μl) | ||||||||||
Sensitivity, % | 96 | 97 | 93 | 96 | 94 | 95 | 96 | 97 | 96 | 96 |
Negative predictive value, % | 98 | 97 | 99 | 98 | 98 | 98 | 99 | 97 | 98 | 97 |
Rule-in (≥ 250 bacteria/μl) | ||||||||||
Specificity, % | 86 | 76 | 94 | 83 | 89 | 91 | 90 | 69 | 87 | 85 |
Positive predictive value, % | 67 | 65 | 73 | 68 | 69 | 72 | 59 | 49 | 69 | 75 |
Number of undecided samples, i.e. ≥ 50 and < 250 bacteria/μl (% of total within that group) | 7365 (15) | 4694 (19) | 2671 (12) | 3245 (15) | 1168 (15) | 1040 (13) | 471 (16) | 673 (21) | 600 (18) | 155 (15) |
Relevant bacteriuria samples within undecided samples, % | 14 | 14 | 14 | 15 | 15 | 15 | 9 | 12 | 13 | 14 |