Abstract
To assess the difficulties in conducting pediatric research, we reviewed the 351 protocols dealing with research in infants and children in our institute between July 1982 and August 1988. Of the 16 rejected protocols (4.5%), 12 were drug studies, 3 dealt with the nature of course of disease states and 1 was in the area of behavioral sciences. Drug studies were significantly more likely to be rejected than all other studies. The most common reason for rejection (n = 10) were major scientific flaws which, according to the committee, would result in inability of the study to answer the questions posed by the researchers. In 9 cases, the committee judged a study to be physically invasive without a direct benefit to the involved infant/child. In 3 cases, the committee rejected a study because patients with serious medical conditions might be randomized to receive placebo and not a drugh which, based on current knowledge from adults, would possibly improve their condition. In 3 protocols current antimicrobial therapy covered all pathogens causing the infection and the proposed new therapy could not improve the prognosis further but only be equal or inferior. Researchers who had more than one protocol rejected had submitted significantly more protocols (7.17 ± 1.35) than those who had only one rejection (1.86 ± 0.36, p < 0.0005) or than the 10 researchers with the highest number of studies without a single rejection (4.2 0.4, p < 0.05). In trying to solve the problem of invasiveness in drug studies in neonates, we have conducted a pharmacokinetic analysis and have documented that 3 samples for drug concentration are all that is needed for pharmacokinetic analysis, as values achieved with these data are not different from those calculated from 8 concentration-time points. In a prospective study in neonates, the validity of these assumptions was proven for the use of the antibiotic vancomycin. This model may be applicable to other areas of pediatric research where careful analysis of existing data may reveal that accurate information can be derived from much fewer samples than previously believed.