Does the outcome change depending on the use of an administrative database versus a high resolution database? In order to answer the question: Is rrt an independent predictor for hospital mortality during ICU?

We identified rrt as:

Patients containing: rrt OR ultrafiltration OR cavhd OR dialysis OR cvvh OR sled and not containing chronic

Our Inclusion/Exclusion criteria are:

• Including only First ICU admission.
• Excluding age <16.
• Including only Apache admission diagnosis containing '%sepsis%'.
• Excluding patients who had a readmission.
• Excluding patients with history of ESRD.
• Including only patients who developed AKI during ICU stay.
• Including only patients with mechanical ventilation.
• Including patients with actualhospitalmortality NOT NULL.
• Including all years.