Caffeine in apnoeic Asian neonates: a sparse data analysis.

AIMS

To monitor plasma caffeine concentrations and adverse effects and to study the pharmacokinetics of caffeine in neonatal apnoea in the local Asian population after intravenous administration of caffeine.

METHODS

Eighteen neonates with apnoea were treated with caffeine citrate at a loading dose of 10 mg caffeine base kg-1 and a maintenance dose of 2.5 mg kg-1 day-1. Blood samples, three after loading and two after the maintenance dose on day 2, 3, 7, 14 and 21 were taken and analysed for caffeine and its main metabolites using solid phase extraction and h.p.l.c. Adverse effects were monitored. Sparse data pharmacokinetic analysis was performed using P-Pharm.

RESULTS

Mean caffeine concentrations varied from 10 to 20 mg l-1 throughout treatment (range 3.6-28.4 mg l-1). These concentrations were efficacious; less so in those with lower concentrations. Adverse effects included gastrointestinal disturbances, diuresis and hyperglycaemia. Pharmacokinetic parameter estimates [mean (coefficient of variation%)] were CL=0.00628 (17.5%) l h-1 and V=0.961 (20.3%) l. CL (l h-1)=0.004248 * wt(kg)+0.00154; r=0.8, P<0.01, explained 64% of the variation. V (l)=0.6299 * wt(kg)+0.259; r=0.67, P<0.01, explained 45% of variation. Model-predicted compared with observed plasma concentrations in a separate group of 10 neonates were unbiased and of good precision.

CONCLUSIONS

The dosing regimen studied was suitable for our local Asian neonates as it resulted in therapeutic caffeine concentrations for adequate treatment of apnoea. Adverse effects were tolerable. Therefore, to avoid a higher incidence of adverse effects, this regimen should be retained and not increased as proposed by other workers. CL and V were within values of those reported for neonatal apnoea. Sparse data analysis showed that weight alone was adequate as the influential variable for the accurate prediction of individual pharmacokinetic parameters, plasma concentrations and for dosage adjustment if required.