Table 3.
PCT Guidance as an Independent Protective Factor from Development of IAAEs by Day 180
| Parameters | IAAE (−) (n = 227) | IAAE (+) (n = 29) | Univariate Analysis |
Multivariate Analysis |
||
|---|---|---|---|---|---|---|
| HR (95% CI) | P Value | HR (95% CI) | P Value | |||
| PCT guidance | 116 (51.1) | 9 (31.0) | 0.45 (0.20–0.98) | 0.045 | 0.38 (0.17–0.85) | 0.01 |
| Dementia | 54 (23.8) | 16 (55.2) | 3.53 (1.70–7.34) | 0.001 | 4.30 (2.02–9.14) | <0.001 |
| Residency in healthcare facilities | 14 (6.2) | 6 (20.7) | 3.56 (1.45–8.74) | 0.006 | — | — |
| Hospitalization in last 3 mo | 39 (17.2) | 10 (34.5) | 2.47 (1.15–5.31) | 0.02 | — | — |
| SOFA score >4* | 74 (32.6) | 15 (51.7) | 2.09 (1.01–4.34) | 0.05 | — | — |
| Community-acquired pneumonia | 105 (46.3) | 7 (24.1) | 0.39 (0.17–0.90) | 0.03 | — | — |
| Primary or secondary bacteremia | 31 (13.7) | 11 (37.9) | 3.52 (1.66–7.45) | 0.001 | 2.93 (1.34–6.43) | 0.01 |
| Septic shock | 13 (5.7) | 5 (17.2) | 3.17 (1.21–8.32) | 0.02 | — | — |
| Empiric treatment according to the ESCMID guidelines | 86 (37.9) | 17 (58.6) | 2.25 (1.07–4.70) | 0.001 | — | — |
| Intake of cefepime | 3 (1.3) | 3 (10.3) | 5.58 (1.68–18.51) | 0.005 | — | — |
| Intake of piperacillin/tazobactam | 122 (53.7) | 24 (82.8) | 3.81 (1.45–9.98) | 0.007 | — | — |
| Intake of a carbapenem | 27 (11.9) | 10 (34.5) | 3.37 (1.57–7.26) | 0.002 | 2.91 (1.29–6.52) | 0.01 |
| Intake of a glycopeptide | 35 (15.4) | 10 (34.5) | 2.66 (1.24–5.72) | 0.01 | — | — |
| Intake of fluconazole | 8 (3.5) | 5 (17.2) | 4.46 (1.70–11.70) | 0.002 | — | — |
Definition of abbreviations: CI = confidence interval; ESCMID = European Society of Clinical Microbiology and Infectious Diseases; HR = hazard ratio; IAAEs = infection-associated adverse events; PCT = procalcitonin; SOFA = Sequential Organ Failure Assessment.
Univariate and multivariate (Cox Forward Conditional) models for the intention-to-treat population (N = 256) are presented. Only parameters reaching statistical significance are provided. All baseline parameters that may act as confounders for the primary outcome were tested in univariate models (shown in Table E3). All 14 parameters reaching statistical significance in univariate models were subsequently entered in a multivariate model. Statistical probability threshold for entering the model was set at 0.05 and for removal at 0.10. Data are shown as n (%).
Cutoff value was calculated by the Youden’s Index of the respective receiver operator characteristics curve of the score in relation to primary outcome.