Caffeine is a potentially useful alternative to theophylline for the treatment and prevention of apnea of prematurity because of its lower toxicity and longer terminal half-life. Monitoring of salivary caffeine concentrations is less invasive than blood sampling, especially in very sick premature neonates. Caffeine citrate-3 mg/kg, 15 mg/kg, or 30 mg/kg-was administered once daily for 7 days in a randomized, parallel design to 59 newborn, premature infants with an initial loading dose of twice the maintenance dose. Serum and saliva samples (131 pairs) were collected and assayed by high-performance liquid chromatography (HPLC) for caffeine content. Measurable caffeine concentrations in serum ranged from 0.28 to 93.3 mg/L and in saliva from 0.35 to 91.5 mg/L. The mean ratio of the saliva-to-serum concentrations was 0.924. There was no significant difference in precision between the serum and salivary data. The mean serum caffeine concentration was 29.9 mg/L, and the mean salivary concentration was 27.7 mg/L, indicating a small negative bias for saliva versus serum monitoring. Salivary caffeine concentration monitoring is a satisfactory alternative to blood sampling across a wide range of caffeine doses used to treat apnea.

No abstract

The population pharmacokinetics of amoxicillin were determined in 40 very premature infants (< or = 32 week gestational age, < 1500 g birth weight) who were receiving intravenous amoxicillin (50 mg/ kg, every 12 h) during the first days after birth. Serum amoxicillin concentrations were measured by HPLC. Clearance (CL) and volume of distribution (Vd) were modeled alone and under the influence of demographic and clinical covariates with a 1-compartment model with first-order elimination. The final population models with influential covariates were: CL(L/h) = 0.0000610 x body weight (g) and CL (L/h) = 0.0000805 x body weight (g), for infants also receiving gentamicin and not receiving gentamicin, respectively; Vd(L) = 0.678. The interpatient standard deviation (SD) for CL was 0.0351 L/h, and for Vd was 0.365 L. The intrapatient variability (SD) among observed and model-predicted serum concentrations was 13.7 mg/L. Evaluation of the predictive performance of this model in another group of infants (n = 16) indicated statistically insignificant bias (p > 0.05) of 3 mg/L among pairs of observed and Bayesian-predicted amoxicillin concentrations. The average population CL was smaller, but the average Vd and terminal half-life (t1/2) were larger than previously reported for healthy adults.