Prescribing an equilibrated intermittent hemodialysis dose in intensive care unit acute renal failure

Prospective, formal, blood-side, urea kinetic modeling (UKM) has yet to be applied in intermittent hemodialysis for acute renal failure (ARF). Methods for prescribing a target, equilibrated Kt/V (eKt/V) are described for this setting.

Serial sessions (N= 108) were studied in 28 intensive care unit ARF patients. eKt/V was derived using delayed posthemodialyis urea samples and formal, double-pool UKM (eKt/Vref), and by applying the Daugirdas-Schneditz venous rate equation to pre- and posthemodialysis samples (eKt/Vrate). Individual components of prescribed and delivered dose were compared. Prescribed eKt/V values were determined using in vivo dialyzer clearance estimates and anthropometric (Watson and adjusted Chertow) and modeled urea volumes.

eKt/Vref (mean +/- SD = 0.91 +/- 0.26) was well-approximated by eKt/Vrate (0.92 +/- 0.25), R= 0.92. Modeled V exceeded Watson V by 25%+/- 29% (P < 0.001) and Adjusted Chertow V by 18%+/- 28% (P < 0.001), although the degree of overestimation diminished over time. This difference was influenced by access recirculation (AR) and use of saline flushes. The median % difference between Vdprate and Watson V was reduced to 1% after adjusting for AR for the 22 sessions with < or =1 saline flush. The median coefficients of variation for serial determinations of Adjusted Chertow V, modeled V, urea generation rate, and eKt/Vref were 2.7%, 12.2%, 30.1%, and 16.4%, respectively. Because of comparatively higher modeled urea Vs, delivered eKt/Vref was lower than prescribed eKt/V, based on Watson V or Adjusted Chertow V, by 0.13 and 0.08 Kt/V units. The median absolute errors of prescribed eKt/V vs. delivered therapy (eKt/Vref) were not large and were similar in prescriptions based on the Adjusted Chertow V (0.127) vs. those based on various double-pool modeled urea volumes (approximately 0.127).

Equilibrated Kt/V can be derived using formal, double-pool UKM in intensive care unit ARF patients, with the venous rate equation providing a practical alternative. A target eKt/V can be prescribed to within a median absolute error of less than 0.14 Kt/V units using practical prescription algorithms. The causes of the increased apparent volume of urea distribution appear to be multifactorial and deserve further investigation.