Univariable regression analysis Multivariable regression analysis with time‐dependent trend Multivariable regression analysis without time‐ dependent trend Access site Subclavian Jugular Femoral Brachial Other 1.08 (0.84 – 1.38) 1 1.30 (0.92 – 1.82) 0.44 (0.16 – 1.24) 1.29 (0.52 – 3.18) 0.81 (0.63 – 1.06) 1 1.18 (0.81 – 1.72) 0.42 (0.14 – 1.22) 1.53 (0.57 – 4.08) CVC type CVC Swan‐Ganz 1 0.59 (0.25 – 1.39) Lumen Multi Single 1 1.00 (0.56 – 1.72) Indications for CVC use Antibioticsd Blood products Dialysis Total parenteral nutritione Other indicationsf 1.47 (1.11 – 1.95) 1.15 (1.12 – 1.88) 0.94 (0.67 – 1.32) 1.70 (1.31 – 2.20) 0.88 (0.65 – 1.19) 1.38 (1.01 – 1.87) 1.32 (1.01 – 1.71) 1.68 (1.25 – 2.26) BSI, bloodstream infection; CRBSI, central venous catheter‐related bloodstream infection; CVC, central venous catheter; HH, hand hygiene; ICU, intensive care unit a according to the SAPS II criteria: medical: no surgery within one week of admission to ICU; scheduled surgical: surgery was scheduled at least 24 hours in advance +/‐ seven days intensive care unit admission; unscheduled surgical: patients added to the operating room schedule within 24 hours of the operation. b in univariable analysis the interaction term of Intermediate Care with CVC duration was significant with a HR of 0.90 (0.82‐0.99, p‐value = 0.02). HR for IM was 10.3 (1.2‐86.4, p‐value = 0.03). This effect decreased with increasing CVC duration (10% per day). In the multivariable analysis this interaction was not significant anymore. c including subsequent ICU stays (maximum 3) until insertion; in univariable analysis the interaction term of an ICU stay of >20 days with CVC duration was significant with a HR of 1.04 (1.004‐1.08). HR for ICU stay > 20 days was 1.78 (0.99‐3.20, p‐value = 0.06). This effect increased with increasing CVC duration (4% per day). In the multivariable analysis, this interaction was not significant anymore. d data of first two months of one hospital excluded because of invalid or missing data. e one hospital excluded because of invalid or missing data; in univariable analysis, the interaction term with CVC duration was significant with a HR of 1.03 (1.005‐1.71, p‐value = 0.02). HR for TPN was 1.12 (0.73‐1.71). This effect increased with increasing CVC duration (3% per day). In multivariable analysis this interaction was not significant anymore. f data of first year of three hospitals and of first four months of another hospital excluded because of different interpretation.
RkJQdWJsaXNoZXIy MTk4NDMw