Investigators will warrant a correct protocol application. Study data will be reviewed by an external monitoring committee from the clinical assay research central unit (UCICEC - IDIBELL). Monitors will contrast registered data from the collection data form (CDF) with data from patient´s medical record. All patient´s medical records will be indefinitely saved in electronical format to be reviewed if necessary. Participants who meet inclusion criteria will be randomized for one of both arms with aleatory assignation using a randomisation sequential (RndSeq) program for Statistical Package for the Social Sciences (SPSS). Investigators will report adverse events (in less than 24 hours if severe) to the sponsor center to be properly evaluated. If the severe adverse event (SAE) is finally evaluated by the study board as related to the intervention arm, urgent notification to health authorities must proceed and study should be interrupted until further decision. Data registry has been created to include all variables with written individual data collection forms (DCF). Data will be bedside registered by the investigators but final software database registration will be done by the statistics outside investigator who has no contact with the participant situation. Cytokines levels will be introduced in DCF when measured (every six months). Statistical analysis will be done by the statistics investigator who wont have any role in patient´s selection, randomization, or follow up. SPSS v. 18.0 for statistical analysis will be used. Variable distribution will be studied and logarithmic transformation will be used on those variables that don't present normal distribution, presumably cytokine levels. Univariate analysis comparing clinical, demographic, biochemical, metabolic, hemodynamic and respiratory baseline variables between both arms (CVVHD and CVVH), will be done with two-tailed t test for continuous variables and chi-square test for categorical variables. Variables determined several times (T0, T24, T48, T72) will be analysed using a one-way repeated measures ANOVA test in order to demonstrate differences between both arms. Multivariate analysis will be completed to control those clinically relevant confounding variables as well as to discover baseline differences. According to hypothesis and to the dependent variable on study, investigators will use a survival analysis (to study mortality) with a cox regression model, or a hierarchic multiple linear regression model when the dependent variable is continuous. Arm intervention (CVVHD, CVVH) will be considered as the main independent variable adding other control independent variables. As the study is measuring cytokine levels in five different moments (T0, T24, T48, T72), in order to maximize statistical power and reduce control variables number, the area under the curve (AUC) we´ll be determined for every cytokine during the first 72 hours. Due to this statistical maneuver, investigators will obtain a continuous variable that represents each cytokine level during the biochemical study period (72 hours). To evaluate if the arm intervention improvement in terms of efficacy and safety could be related to cytokine levels during the first 72 hours, a mediation complementary analysis will be done considering cytokine (represented by AUC) as a mediator between the independent variable (intervention arm) and the evaluated effect.
