Antibiotics in the ICU: Cycling, Mixing, or Personalization?
Abstract & Commentary
By Stan Deresinski, MD, FACP, Clinical Professor of Medicine, Stanford, Associate Chief of Infectious Diseases, Santa Clara Valley Medical Center, is Editor for Infectious Disease Alert.
Synopsis: Maintenance of heterogeneity of antibiotic use in an ICU was superior to months-long cycling with regard to antibiotic resistance.
Source: Sandiumenge A, et al. Impact of diversity of antibiotic use on the development of antimicrobial resistance. J Antimicrob Chemother. 2006; 57:1197-1204.
Sandiumenge and colleagues evaluated the effects of 3 strategies of antibiotic prescribing in a 14-bed ICU. The strategies were applied serially, beginning with an initial 10-month period during which patients with suspected ventilator-associated pneumonia received "patient-specific therapy" in which multiple antibiotic regimens, chosen on the basis of length-of-stay and recent antibiotic exposure, were used. This was followed by a 24-month period consisting of 4-month cycles of preferential use, in order of anti-pseudomonal carbapenems, anti-pseudomonal cephalosporins, and piperacillin-tazobactam. This was followed by three 4 month cycles of these same agents, but in reverse order. In each instance, aminoglycosides or ciprofloxacin could be added to the regimen of the presence of Pseudomonas aeruginosa was suspected. Finally, during the final 10 months, a strategy of heterogeneity of antibiotic use (mixing), which actually consisted of extreme cycling, was implemented, with the preferred regimen changing with each consecutive patient. The order of patient-to-patient cycling was: antipseudomonal carbapenem in Patient 1, ciprofloxacin in Patient 2, clindamycin plus antipseudomonal cephalosporin in Patient 3, piperacillin-tazobactam in Patient 4, then restarting the cycle in Patient 4 and so on. An "antibiotic homogeneity index" was calculated, taking into account theoretically perfect heterogeneity of use, the proportion of individual antibiotic use, and the total number of antibiotics which could have been used.
Heterogeneity of use was maximal during the initial period of patient-specific therapy, followed close behind by the final period of "mixing." The least heterogeneity was achieved during cycling at 4-month intervals. High homogeneity (i.e., limited heterogeneity) was associated with significant increases in isolation of carbapenem-resistant Acinetobacter baumanii, ESBL-producing Enterobacteriaceae, and Enterococcus faecalis. Lesser homogeneity, in general, only partially reversed these increases and even when relevant antibiotics were restricted ("cycled out"), the incidence of isolation of carbapenem resistant A. baumanii remained high, although that of ESBL-producing Enterobacteriaceae and of E. faecalis was reduced.
The problem of antibiotic resistance, particularly in the ICU, is a formidable one. This has led to a variety of proposals for manipulating the way in which antibiotics are chosen with the aim of slowing or reversing antibiotic resistance patterns. One of the approaches taken, antibiotic cycling, in which specified antibiotics are used as empiric therapy for several months at a time before cycling to another antibacterial, became almost a fad for reasons I could never understand. However, it seems intuitive that overuse of any one antibiotic, or class of antibiotics, would drive resistance to that agent and other members of its class. In fact, this could also select for resistance across classes. An example of the latter can occur with ESBL-producing Enterobacteriaceae, since the plasmids on which the genes encoding the beta lactamase reside commonly carry genes encoding resistance to other antibiotics, such as aminoglycosides and their chromosomes often contain topoisomerase mutations causing fluoroquinolone resistance. In such cases, exposure to a single antibiotic may select out strains resistant to multiple antibiotics. Cycling might still make sense if cycling off an antibiotic rapidly led to replacement of the resistant strains to ones that were susceptible. Unfortunately, the evidence indicates that, while this can occur, it is often a very slow process, taking much longer than any practical cycling interval.
In fact, mathematical modeling suggests that when compared either to strategies of combination antibiotic use and heterogeneity of use, cycling is the least likely to prevent antibiotic resistance.1, 2 In addition, a number of studies have failed to demonstrate benefit from this strategy. On the other hand, one possibly unexpected observation in this study was an 18% increase in antibiotic use during the period of mixing. This suggests some of the complexities and unexpected consequences of protocolized interventions in antibiotic use.
It seems to me that the optimal strategy is "intelligent mixing" designed to optimize therapy for the individual patient while assuring that overall use of antibiotics in the unit is heterogeneous. Empiric therapy should be deescalated as appropriate when relevant clinical and laboratory information becomes available, as was done in this study. This approach, however, requires a higher level of functioning than cycling. Nonetheless, it's what I would expect from a good clinician.
- Bergstrom CT, et al. Ecological theory suggests that antimicrobial cycling will not reduce antimicrobial resistance in hospitals. Proc Natl Acad Sci USA. 2004;101:13285–13290.
- Bonhoeffer S, et al. Evaluating treatment protocols to prevent antibiotic resistance. Proc Natl Acad Sci USA. 1997;94:12106–12111.