Urine output predicts serum gentamicin concentration in preterm infants

There are no reliable predictors of gentamicin (G) levels in premature infants. As G is excreted by the kidneys, we hypothesized G levels can be predicted by urine output. We conducted a retrospective study of neonates (GA 30-36wks, n= 20) who received G therapy (3 mg/kg/day). Infants with a Dx of acute tubular necrosis, asphyxia, hypotension or received vasopressor support or indomethacin were excluded. We collected data on gestational age (GA), body weight, urine output (UOP, cc/m²/hr and cc/k/hr), fluid intake and output (I/O), G intervals/route, body surface area, G levels and day of life G levels drawn. Volume of distribution (V_D, liters) was calculated as dose given/[peak-trough] levels. Data were analyzed by student's test, chi square test, correlation coefficient as appropriate. Results showed urine output ≥2 and ≤ 4.0 cc/k/hr predicted therapeutic G levels (peak= 5-10 μg/mL) in premature infants (sensitivity 0.78, specificity 1 and positive predictive value 1). Urine output >4 cc/k/hr resulted in peak levels < 5 μg/mL. There were significant inverse correlations comparing peak G levels to UOP and V_D (Table). We conclude that urine output, 24hrs prior to G level measurement, is a non-invasive, highly predictive measure of adequacy of dose. We also conclude that the inverse correlation of peak G levels with urine output reflects the change in volume of distribution. Table: Correlations of Peak Gentamicin (G) Concentrations UOP (cc/kg/hr) UOP (cc/m²/hr) V_D r -0.572 -0.648 -0.49 r² 0.328 0.42 0.24 p value 0.008 0.002 0.03.