Table 2.
Univariable and multivariable analysis 30-day mortality predictors in COVID-19 patients with ICU-acquired Enterococcus spp. BSI.
| Variable | Non-survivorsa 18 (42) | Survivorsa 25 (58) | Univariable analysis |
Multivariable analysis** |
||
|---|---|---|---|---|---|---|
| OR (95% CI) | p-Value | OR (95% CI) | p-Value | |||
| Age in years, median (IQR) | 65 (60–70) | 61 (54–67) | 1.06 (0.98–1.14) | .150 | 1.07 (0.98–1.17) | .128 |
| Male gender | 14 (78) | 19 (76) | 1.11 (0.26–4.67) | .892 | ||
| Charlson score, median (IQR) | 3 (2–5) | 3 (1–4) | 1.23 (0.87–1.73) | .246 | ||
| Diabetes mellitus | 4 (22) | 5 (20) | 1.14 (0.26–5.03) | .860 | ||
| Chronic obstructive pulmonary disease | 2 (11) | 1 (4) | 3.00 (0.25–35.91) | .386 | ||
| Chronic kidney disease | 3 (17) | 5 (20) | 0.80 (0.16–3.88) | .782 | ||
| Previous myocardial infarction | 1 (6) | 3 (12) | 0.43 (0.04–4.52) | .483 | ||
| Presence of solid neoplasm | 1 (6) | 0 (0) | Model not converging | .419* | ||
| Presence of hematological malignancy | 1 (6) | 0 (0) | Model not converging | .419* | ||
| Solid organ transplant | 0 (0) | 0 (0) | – | 1.000* | ||
| Haematopoietic stem cell transplantation | 0 (0) | 0 (0) | – | 1.000* | ||
| Admission from LTCF | 3 (17) | 1 (4) | 4.80 (0.46–50.50) | .191 | ||
| Previous hospitalisation (within 6 months) | 6 (33) | 1 (4) | 12.00 (1.29–111.32) | .029 | 6.52 (0.62–68.44) | .118 |
| Previous VRE isolation (within 6 months) | 3 (17) | 2 (8) | 2.30 (0.34–15.44) | .391 | ||
| Previous therapy with glycopeptides (within 6 months) | 1 (6) | 4 (16) | 0.31 (0.03–3.03) | .313 | ||
| ARDS at hospital admission | 16 (89) | 24 (96) | 0.33 (0.03–3.99) | .386 | ||
| Need for invasive mechanical ventilation | 18 (100) | 25 (100) | – | 1.000* | ||
| Treatment with steroids | 15 (83) | 19 (76) | 1.58 (0.34–7.38) | .562 | ||
| Treatment with tocilizumab | 5 (28) | 8 (32) | 0.82 (0.22–3.09) | .766 | ||
| Duration of ICU stay before BSI onset in days, median (IQR) | 17 (12–32) | 21 (11–32) | 1.01 (0.98–1.05) | .476 | ||
| Recent treatment with cephalosporins | 15 (83) | 19 (76) | 1.58 (0.34–7.38) | .562 | ||
| Neutropenia (ANC < 500 cell/mm3) | 0 (0) | 0 (0) | – | 1.000* | ||
| Presence of CVC | 17 (94) | 25 (100) | Model not converging | .419* | ||
| Pitt bacteraemia score, median (IQR) | 8 (7–8) | 6 (4–8) | 1.24 (0.97–1.58) | .086 | ||
| SOFA score, median (IQR) | 11 (8–13) | 7 (6–10) | 1.31 (1.07–1.60) | .010 | 1.32 (1.04–1.66) | .021 |
| Septic shock | 10 (56) | 11 (44) | 1.59 (0.47–5.39) | .456 | ||
| VRE as aetiological agent | 4 (22) | 3 (12) | 2.10 (0.41–10.80) | .377 | ||
| Polymicrobial BSI | 7 (39) | 12 (48) | 0.69 (0.20–2.36) | .553 | ||
| DENOVA score, median (IQR) | 0 (1–2) | 1 (1–2) | 0.92 (0.45–1.86) | .808 | ||
| CRBSI | 12 (67) | 19 (76) | 0.63 (0.16–2.42) | .502 | ||
| Endocarditis | 0 (0) | 0 (0) | – | 1.000* | ||
| Source control performed/unnecessary | 15 (83) | 25 (100) | Model not converging | .066* | ||
| Empirical therapy | 17 (94) | 24 (96) | 0.71 (0.04–12.13) | .812 | ||
| In vitro active empirical therapy | 16 (89) | 22 (88) | 1.09 (0.16–7.31) | .929 | ||
ANC, absolute neutrophil count; ARDS, acute respiratory distress syndrome; BSI, bloodstream infection; COVID-19, coronavirus disease 2019; CRBSI, catheter-related bloodstream infection; CI, confidence intervals; CVC, central venous catheter; DENOVA, long Duration of symptoms/Embolization/Number of positive cultures/Origin of infection unknown/Valve disease/Auscultation of murmur; ICU, intensive care unit; IQR, interquartile range; LTCF, log-term care facility; SOFA, sequential organ failure assessment; VRE, vancomycin-resistant enterococci.
aResults are presented as No. of patients/total of patients unless otherwise indicated.
*In the case of zero events in both groups and non-converging univariable logistic regression models, p values are from Fisher exact test. Nonconvergence was also observed when including the variable source control performed/unnecessary (p < .20 in univariable analysis) in the multivariable logistic regression model, that was eventually built without this variable. An additional, penalised, multivariable logistic regression model with Firth’s correction, which included source control performed/unnecessary plus all the variables included in the final standard multivariable model, confirmed the results observed in the standard model (age: OR 1.07, 95% CI 0.99–1.18, p = .112; previous hospitalisation: OR 2.82, 95% CI 0.40–31.13, p = .302; SOFA score: OR 1.26, 95% CI 1.03–1.61, p = .025) and no independent association with treatment failure was observed for source control performed/unnecessary (OR 0.23, 95% CI 0.00–3.27, p = .296). The additional analysis with Firth’s correction was performed using the logistf package for R Statistical Software version 3.6.0 (R Foundation for Statistical Computing, Vienna, Austria).
**Only results for variables retained in the final multivariable model are presented. The discriminatory performance and the calibration of the model were evaluated using the C-statistic (area under the curve [AUC] 0.835, with 95% CI from 0.702 to 0.967) and the Hosmer–Lemeshow’s test (p = .149), respectively.