Modeling serum creatinine in septic ICU patients

Serum creatinine is a metabolite assumed to be constantly produced by the normally functioning muscle mass and is a good measure for monitoring daily renal function in the intensive care unit (ICU). High serum creatinine levels or an abnormal departure from normal pre-disease basal levels are indices of acute renal failure (ARF). Septic ICU patients develop multi-organ failure and ARF is often an important complication. In order to attempt to substitute artificially for the failing renal function, septic ARF patients often undergo hemodialytic procedures until kidney damage resolves. The present work details the structure of a model describing observed creatinine serum concentration (CSC) variations, depending on the time-varying septic insult to renal function in ICU patients, as well as the estimation of its parameters. CSC determinations were routinely obtained from 12 patients, of whom six underwent continuous dialysis procedures and six did not. The model incorporates a delay term ?, to be estimated, which expresses the number of days between the beginning of sepsis and ICU admission, allowing, in this way, a general representation of the time course of renal function depending on the history of sepsis. In order to take into account the effect of dialysis, a linear CSC elimination rate was added in those days when dialysis therapy was actually administered. Estimates of the model's structural parameters were computed for each patient by means ordinary least squares and were then compared across dialysis groups. © 2004 Plenum Publishing Corporation.