This multicentre SHARF4 (Stuivenberg Hospital Acute Renal Failure) study aims to investigate outcome in patients with acute renal failure (ARF), stratified according to severity of disease (SHARF score) and randomised to different treatment options.

All adult patients with a serum creatinine >2 mg/dl were included. Patients were stratified according to disease severity and those in need for RRT were randomised to intermittent (IRRT) or continuous renal replacement therapy (CRRT)

Subjects. Hospitals qualified for participation if they had at least 600 beds, a centre for the treatment of End Stage Kidney Disease (ESKD) patients, a multipurpose ICU with at least 12 beds, at least 30 patients with ARF treated by RRT during the last year and regular use of both intermittent and continuous techniques. A centre questionnaire was sent to candidate centres in order to check qualifying criteria.

Qualified Belgian ICU units included all adult (age ³ 18 years old) patients with a serum creatinine >2 mg/dl that were consecutively admitted. All patients with known pre-existing chronic renal disease, defined as a serum creatinine above 1.5 mg or with clearly reduced kidney size on ultrasound were excluded.

Severity was defined by calculating the SHARF score in all subjects that fulfilled the inclusion criteria. Patients were stratified in 3 classes (SHARF<30, 30-60, >60). Within each stratum, those in need for renal replacement therapy were randomised to daily IRRT (intermittent haemodialysis during 4-6 hours daily) or CRRT (continuous veno-venous hemofiltration), if none of the predefined contraindications for randomisation were present.

Data collection. The following basic data were collected: demographic (age, sex, weight and height), course of hospitalisation (date of admission to hospital and to ICU, date of discharge from ICU and from hospital or date of death), date of ARF diagnosis, type of ARF (prerenal, renal, postrenal, acute on chronic disease), cause of ARF (acute tubular necrosis, acute glomerulonephritis, acute interstitial nephritis, systemic disease), primary disease type (medical - toxic, medical - other, surgical, obstetric) and serum creatinine at different time points during hospitalisation. Parameters of the SHARF score were collected at the first day that the criteria of ARF were met. For patients referred to the ICU later in the course of their ARF, the day of admission to the ICU was the starting day. Overall severity was evaluated with the APACHE II score at baseline and with the SOFA score at diagnosis of ARF and after 24 and 48 hours.

The following short-term outcome parameters were measured: hospital mortality, length of stay at ICU and hospital and GFR at hospital discharge according to the Cockroft and Gault formula.

Allocation of treatment. From the previous SHARF studies, we learned that about 50 % of patients would need RRT treatment. The decision to treat conservatively or to start RRT was at the discretion of the attending physician, taking into account the rules of good clinical practice in this field. The choice of RRT treatment modality was randomised within each stratum of the SHARF score. Randomisation was performed electronically per centre. In order to avoid selection bias, all consecutive patients had to be included. If a patient could not be randomised for some reason, the reason had to be documented in the electronic Case Report Form (CRF).

Renal Replacement Therapy. The techniques used to perform RRT were in agreement with the standard procedures of the participating hospitals. Therefore, all candidate centres completed a questionnaire about their current renal replacement strategy in ARF. The strategy chosen in the protocol was the result of this questionnaire. For IRRT a central venous access, a synthetic membrane (polysulfone or AN 69), and a bicarbonate dialysate were used. Daily dialysis was performed during 4-6 hours per session with a blood flow of 100-300 ml/min and a dialysate flow of 100-500 ml/min. For CRRT a central venous access and a synthetic membrane (polysulfone or AN 69) were used. Continuous veno-venous hemofiltration (CVVH) was the preferred method. It was continued during 24 hours per day with a blood flow rate of 100-200 ml/min, a substitution rate of 1-2 L per hour and either lactate or bicarbonate solutions were used. For both modalities, anticoagulation was performed according to the centre practice with either unfractionated heparin, low molecular weight heparin or citrate. The randomly assigned treatment modality of IRRT or CRRT had to be continued daily during at least 3 consecutive days. Thereafter treatment could be continued according to the needs of the patients and at the discretion of the investigator. The motivation for any change in the randomised or planned treatment was recorded in the electronic CRF. Data of all RRT treatments were recorded including date, type of treatment, effective duration, ultrafiltration rate and artificial kidney.

Sample size calculation The sample calculation was based on the assumption that the overall mortality would be 50 % as in the former SHARF studies and that a difference of 10 % in mortality between IRRT and CRRT had to be detected to be clinically relevant. With a first order error of 5 % and a power of 80 % a sample size of 407 patients was needed in each renal replacement group. Since about 50 % of included patients would be considered for renal replacement therapy, a total of 1628 patients had to be included.

Statistical analysis. The data analysis was performed according to the intention-to-treat principle. Univariate analysis was performed on all parameters in order to find significant differences between groups using Student's t-test and Chi square test. Life table analysis was used to compare hospital survival in both treatment groups with Cox proportional hazards regression to control for covariates. Multivariate analysis was performed using logistic regression with mortality as dependant outcome variable. Confounding factors were selected if they show significant difference in the comparison between both treatment options and contribute effectively and independently to the observed outcome. For subgroup analysis, selection was based on reported evidence that these subgroups included the most complicated patients showing the highest co-morbidity and mortality. Statistical significance was set at the .05 level (two-sided). All analyses were performed using SPSS, version 12.0.

Institutional Review Board The protocol has been approved initially by the Ethics Committee of the Public Sector of Antwerp and by the Ethics Committee of each participating centre. A written informed consent has been asked from each patient or his representative in case the patient was unconscient or intubated.

Inclusion Criteria:

• adults (>17 years)
• having acute renal failure: serum creatinine > 2 mg/dl
• being in need for chronic renal replacement therapy (the decision to treat conservatively or to start RRT was at the discretion of the attending physician, taking into account the rules of good clinical practice in this field)

Exclusion Criteria:

• patients with known pre-existing chronic renal disease, defined as a serum creatinine above 1.5 mg/dl or with clearly reduced kidney size on ultrasound were excluded.