Sepsis is a major cause of acute renal failure in hospital patients, but its incidence and the associated prognostic factors have rarely been assessed prospectively by multivariate analysis.

We conducted a prospective 6-month study in 20 multidisciplinary intensive care units to assess the prognosis of patients hospitalized with acute renal failure due to sepsis. Sepsis syndrome and septic shock were defined according to the criteria of the Society of Critical Care Medicine Consensus Conference. Severity scoring indexes (SAPS, APACHE II, and organ system failure (OSF)) were measured on ICU admission and on inclusion. The end-point was hospital mortality.

Acute renal failure had a septic origin in 157 patients (Group 1), comprising 68 with septic shock and 89 with sepsis syndrome, and did not result from infection in 188 patients (Group 2). Patients with septic acute renal failure were older (mean age: 62.2 versus 57.9 years, P<0.02) and had on inclusion a higher SAPS (19.3 versus 16.1, P<0.001), APACHE II (29.6 versus 24.3, P<0.001), and OSF (2.07 versus 1.52, P<0.001) than patients with non-septic acute renal failure. They had a higher need for mechanical ventilation (69.1% versus 47.3%, P<0.001), and acute renal failure was more often delayed during the ICU stay than was present on admission (47.7% versus 32.4% respectively, P<0.005). Hospital mortality was higher in patients with septic acute renal failure (74.5%) than in those whose renal failure did not result from sepsis (45.2%, P<0.001). Mortality was influenced by the presence of a septic shock (79.4%) or of a sepsis syndrome on inclusion (70.8%). Using a stepwise logistic regression model, sepsis was an independent predictor of hospital mortality (OR, 2.51; 95% CI, 1.44–4.39) as well as a delayed occurrence of acute renal failure, oliguria, an altered previous health status, hospitalization prior to ICU, need for mechanical ventilation, age and severity scoring indexes on inclusion. In total patients, mortality was higher in dialyzed than in non-dialyzed patients (P<0.001), and in those treated by continuous compared to intermittent techniques (P<0.01). Patients dialysed with biocompatible membranes had a lower mortality than those treated with cellulose membranes (P<0.005).

Patients with acute renal failure due to sepsis have a worse prognosis than those with non-septic acute renal failure. Sepsis and the above-defined predictive factors are to be considered in studies on prognosis of ARF patients. Our results suggest that the use of biocompatible membranes may reduce significantly mortality in these patients.