A pharmacometric model for factors in the development of bacterial resistance in acutely ill patients

143 acutely ill patients (mainly with lower respiratory tract infections), from 4 clinical trials, had PKs of antimicrobial drugs characterized & C&S, from the site of infection, performed ≈ daily. A pharmacometric model was developed (maximum likelihood, ADAPT II; model discrimination by Akaike's Information Criterion) to identify factors which were associated with percent probability (%P) of an apparent change in MIC, from sensitive to resistant (by NCCLS standards). Overall, 35 of 134 (26%) initially-sensitive, evaluable cases, subsequently 'developed' resistance. In an initial univariate screen, significant factors included: AUC₂₄/MIC (AUIC), treatment (including recent prior drugs), organism & MIC. The final pharmacometric model defined 4 groups. Group A was 42 cases with Pseudomonas aeruginosa (33 monotherapy & 9 combination; median AUIC: 182, range: 19→2005) in which, modelled %P = 100-[1-AUIC/(193+AUIC)]. In this Hill-type PD equation, 100 is the extrapolated %P at an AUIC = 0 & 193 is the AUIC at which %P = 50. For these organisms, the observed %P, in the 18 cases at AUIC < 150, was 72% &, in the 17 cases at AUIC > 250, the %P was 24%. Group B was 66 other gram negative rods, on monotherapy, for which %P was 21.2% & no AUIC- (median: 922, range: 8→23500) effect was apparent; Group C was 15 similar cases (median AUIC: 2142, range: 75→11430), on combination regimens, in which %P was 0; & Group D was 11 'other' organisms (median AUIC: 785, range: 174→24830) in which %P was also 0. Small sample size, in some of the implied cells, may have limited the sensitivity of the analysis, to reliably detect additional, significant factors. For example, AUIC was highly related to %P for Pseudomonas aeruginosa but the very high AUIC values, for the other organisms, may have prevented detection of similar relationships. Also, it appears probable (but not certain) that combination therapy decreases %P, beyond the effect predicted for the resulting increase in AUIC.