Atkinson 1998.

Methods	Single‐centre RCT conducted in Los Angeles, USA.
Participants	63 children with cancer and tunnelled CVCs with a CVC‐related infection (for the exact definition of CVC‐related infections in this study see Table 1) and without a gross thrombus of fibrin within or attached to the CVC. Thrombus and fibrin deposits were investigated with a dye study injection of the CVC. Not all children received chemotherapy; the exact number is not reported. Type of CVCs; 11 single‐lumen Broviac (7 intervention, 4 control), 41 double‐lumen Broviac (18 intervention, 23 control) and 11 porth‐á‐cath (8 intervention, 3 control). Underlying diseases: 5 neuroblastoma (3 intervention, 2 control), 9 haemophilia (4 intervention, 5 control), 11 brain tumours (8 intervention, 3 control), 4 lymphomas (2 in each group), 11 ALL (4 intervention, 7 control), 8 ANLL (2 intervention, 6 control), 7 sarcomas (5 intervention, 2 control), 8 other (not further specified; 5 intervention, 3 control).
Interventions	Urokinase locks and concomitant systemic antibiotics (n = 33, intervention group) versus systemic antibiotics only (n = 30, control group). Lock treatment: intraluminal boluses of urokinase (Abbokinase Open‐Cath, Abbott laboratories, Chicago IL) were instilled after at least 24 hours of systemic antibiotics. Children received 2 x 5000 IU boluses of urokinase administered 12 hours apart via each lumen of the catheter to dwell for one hour. The choice of antibiotics was adjusted by the treating physician according to the sensitivities of the organisms that had been cultured.
Outcomes	(1) Number of children cured of their CVC‐related infection. (2) Number of days until first negative blood culture (defined as time from initiation of antibiotic therapy to negative catheter culture). (3) Premature removal of the CVC (defined as the number of catheters requiring removal because of persistent clinical or culture‐documented (i.e. culture positive for more than 72 hours beyond study entry) catheter sepsis). Children in both groups were monitored from the day of study entry with daily blood cultures drawn from all lumens of the catheters. They were also monitored for clinical signs of clearance or persistence of the infection including resolution of fever and leukocytosis.
Notes	A maximum of 17 participants with non‐malignant disease could have been included in this study (i.e. 9 with haemophilia; for 8 other participants their underlying disease was not mentioned in the article). Not all participants received chemotherapy. The urokinase lock group contains fewer children (n = 8) with haematologic malignancies than the systemic antibiotics alone group (n = 15). Quote: "Study follow‐up was completed when negative culture results were obtained and clinical signs resolved or when a failure was declared and the catheter removed". No further information regarding follow‐up was provided.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Randomisation was performed by the pharmacist who also supplied the urokinase for injection" Comment: the sequence generation was probably done randomly, as this is the purpose of randomisation; however a description of the method of randomisation is missing.
Allocation concealment (selection bias)	Unclear risk	The method of randomisation was not specified.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Complete outcome assessment of all included participants for all outcome measures investigated.
Selective reporting (reporting bias)	Low risk	Complete outcome reporting of all included participants for all outcome measures investigated.
Blinding of participants and personnel (performance bias) All outcomes	High risk	The study was open‐label: neither participants nor personnel were blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	The study was open‐label: outcome assessors were not blinded for any outcome evaluated.