DATA EXTRACTION AND QUALITY ASSESSMENT
Data were independently extracted from the included studies by two authors (J.S., S.A-M.) using a standardised database. Data extracted included: study design, patient demographics, reason for admission, indications for levetiracetam and CRRT, residual renal function, dosing regimen, CRRT characteristics and settings, and pharmacokinetic data. Any discrepancies on data extraction were arbitrated by a third investigator.
Our quality assessment was performed using a hybrid model that combined the Quality of Evidence (QoE) and Acute Dialysis Quality Initiative (ADQI) scoring systems [26-28] . Each study was assessed independently by two authors (J.S., S.A-M.) who allocated separate QoE and ADQI scores as per the protocol detailed in Appendix 2 . To eliminate discrepancies in ADQI scores between authors, a mean ADQI score was calculated for each study. Overall quality assessment scores were stratified into low, medium and high-quality equivalents based on the average quality of evidence across the QoE and ADQI assessments. For example, a study was ranked medium quality if it achieved a strong QoE score but a weak ADQI score. Any disputes were arbitrated by a third investigator.