Discussion
Bacterial resistance to antimicrobial agents is an incessantly evolving phenomenon which threatens one of the greatest achievements of medical science, the effective treatment of infectious diseases. Overprescribing and suboptimal selection of antimicrobial agents are believed to have contributed to the acceleration of the selection of resistant strains. Thus antimicrobial stewardship has provoked the interest of the medical community as a multifaceted set of interventions which aim to optimise antimicrobial use and thus stem the onset of resistant bacterial strains.
Despite, however, the public health importance of this issue, there is a notable lack of standardised high-quality research on the field to provide definite answers as to which, if any, initiatives are effective or not. We have already examined antimicrobial restrictions and audit with feedback in two papers that were recently published[9][10.] The absence of randomised models and the great heterogeneity in study protocols limited the ability to draw any firm conclusions on the aspects researched. It highlights the need for future high-quality, reproducible research. Standardisation in study design would increase the utility of clinical research in this field, as meta-synthesis of studies would be possible, providing greater statistical power to detect and map the effects of intervening to try to reduce resistance, and guide clinicians.
Examination of the available literature on the potential efficacy of antimicrobial cycling gives an overall impression of rather limited success. Research papers could be roughly divided to those which evaluated cycling versus a control group and produced conflicting results and those that compared cycling with mixing with none of the strategies appearing superior to the other. Lack of success becomes more evident if one takes into account the most rigorous studies conducted by Toltzis et al[14] as well as Van Duijn et al[23 ]both of which failed to record any favourable results comparing cycling with a control group and a mixing group respectively.
Fair interpretation of the relevant data must take into account some core limitations which could influence results in either way. One such limitation is the lack of standardization of antibiotic protocols across intervention and control groups of different studies, though a general tendency to increase heterogeneity of antibiotic administration in the experimental arms was observable. It is rational to assume that the relevant baseline practices would influence whether significant changes in antibiotic resistance patterns would be recorded post-intervention. A pertinent paradigm is probably provided by Nijssen et al who compared antibiotic rotation with a control group receiving fluoroquinolones in a highly homogeneous manner. Fluoroquinolone resistance rates were decreased in the rotation arm, a trend not seen for cephalosporins. It is well-known that the main mechanism of fluoroquinolone resistance comprises point mutations in chromosomal DNA which are obviously particularly prone to selective pressures. Radical reduction in fluoroquinolone administration along with the main relevant mechanism of resistance could provide a likely explanation for the observed results further supported in the clinical literature after the application of restrictive fluoroquinolone strategies[9].
We cannot exclude the possibility that the potential of success could be pathogen-specific and depending on the monitoring protocol it could be potentially missed; a pathogen-specific effect has indeed been suggested by researchers in the past[8]. It is true that the majority of the available positive findings in our dataset relate to P. aeruginosa although we are not aware of any pathophysiological mechanism that would account for such a theory.
Failure of antibiotic cycling to produce clear benefits is consistent with the theoretical predictions generated by many mathematical models that challenge its intuitively presumed efficacy. On the basis of the aforementioned models, though, one would expect that antibiotic mixing would be more effective via maximising heterogeneous antimicrobial use. Neither assumption was confirmed in practice. Although there is high variability in research protocols and the overall quality of our data is far from satisfying to reach definite conclusions, we should bear in mind that the evolution of bacterial resistance is a complex process and the strategies tested may rely on an oversimplified model of how it may be manipulated. It is worth mentioning that antimicrobial agents of similar spectrum may possess totally different mechanisms of action, and thus may affect bacteria in different ways. In addition, infection control is a hard to standardise parameter which could influence relevant studies drastically.
At this point, it would be useful to discuss the third set of studies included in our review. The latter evaluated resistance dynamics of each of the on-cycle antibiotics during the application of antimicrobial cycling protocols. They provide little information as to the overall efficacy of cycling but could offer some ground for future research as to which agents are actually less prone to the selection of resistant strains. Ginn et al compared periods of predominant cefepime and piperacillin-tazobactam use and found that cefepime, a fourth-generation cephalosporin, was associated with higher overall resistance rates (including co- and cross-resistance). There is plenty of observational research which supports the notion that piperacillin-tazobactam is a less important driver of antibiotic resistance than broad-spectrum cephalosporins[9]. A rational explanation could lie on the fact that broad-spectrum cephalosporins are less effective than inhibitor-based beta-lactams in vitro against ESBLs, which are the among most widespread multidrug-resistant strains within nosocomial environments and could be theoretically preferentially selected under the pressure of inappropriate antibiotic treatment.
On the other hand, Van Loon et al concluded that the homogeneous use of cefpirome, another fourth-generation cephalosporin, was not associated with an increase in the incidence of cefpirome-resistant strains, while both piperacillin-tazobactam and levofloxacin use provoked resistance. The results of those studies are seemingly contradictory and could be confounded either by seasonality or breaks in infection control. Such discrepancies underline the importance of the use of contemporaneous controls as well as the need for bacterial typing data in future research to facilitate a more meaningful interpretation of the data. Bacterial typing becomes especially important in view of the fact that most studies to date have used the unit-wide incidence of resistant strains as the primary outcome indicator, but this is easily affected by changes in colonization pressure and/or breaks in infection control. An idea for future research would also be to differentiate colonization rates in patient groups within the same ward who have and have not participated in study protocols and use additional wards with similar baseline characteristics as comparison units.
Lack of standardization of research protocols was once again a crucial issue which limited our ability to evaluate with confidence the replicability of findings and reach safer conclusions. Research protocols differed in terms of the cycle length, the choice of empirical agents, the opportunity to de-escalate, the acquisition of typing data to assess cross-transmission dynamics, and the measurement of indicators of potential collateral damage induced by the established policies. Among the studies of our dataset it was only Van Duijn et al in 2018 who utilised a cluster-randomised cross-over design to compare cycling with mixing, which was a stronger study design than most. A more thorough evaluation would be possible only if the study included control groups and/or baseline data as well as bacterial typing to assess bacterial clonality. It is true that the conduct of research well-designed and rigorous to be of practical use to clinicians requires specialist expertise of multiple kinds, and is logistically difficult. Nevertheless, it is a worthwhile investment which should be co-ordinated by national or international public health agencies with the ultimate aim to safeguard the future value of antimicrobial agents.