- Findings run contrary to current guidelines
- The findings are correct, but the lack of information on downstream side-effects of fluoroquinolones impairs risk-benefit assessments
- Take home message
- Funding source
Findings run contrary to current guidelines
Daneman et al. found that fluoroquinolones were associated with better clinical outcomes than non-fluoroquinolone antibiotics for treatment of uncomplicated urinary tract infections (UTI). Thosewomen who received fluoroquinolones experienced decreased odds of subsequent UTI-related hospitalizations, emergency department visits and return outpatient visits. In addition, repeat antibiotic prescriptions within 30 days were less common among those who were prescribed fluoroquinolones.
These findings are unexpected given that randomized controlled trials and meta-analyses have shown that the first-line recommended agents (nitrofurantoin, trimethoprim-sulfamethoxazole, fosfomycin trometamol and pivmecillinam) have both clinical and microbiological efficacy in UTI that is comparable with fluoroquinolones. The results of this observational study will likely be unwelcome news to countless antibiotic stewardship programmes that are focused on reducing fluoroquinolone use for simple infections, in line with the US Food and Drug Administration (FDA) black box warnings, and to help avoid inducing Clostridioides difficile infections.
Specifically, the goals of the new antimicrobial stewardship requirements for Joint Commission accredited ambulatory health-care organizations effective January 2020 include decreasing overuse of fluoroquinolones. The Joint Commission evaluates and accredits more than 21 000 health-care organizations and programmes in the USA. The study findings also run counter to the 2011 Infectious Diseases Society of America guideline recommendations, and fluoroquinolone-related warnings from the FDA and the European Medicines Agency.
The findings are incorrect because of biases
One possibility is that the findings of this observational study are incorrect as the result of confounding factors or other sources of bias. We commend the authors on what appears to be a well thought-out and implemented determination of the study cohort, application of appropriate inclusion exclusion criteria, operationalization of study outcomes, base-case statistical analyses, and meaningful set of sensitivity analyses to explore important ‘what if’ questions that are relevant to the conclusions made. However, we have several concerns, some of which are difficult or impossible to avoid with observational research using administrative health data. C
ertainly, one strength of Daneman et al.’s study is their use of the increasingly popular high-dimensional propensity score model, which resulted in adjustment for 500 covariates across various clinical dimensions. However, whereas high-quality randomized, controlled trials can control for both measured and unmeasured/ unknown confounders, observational studies cannot make this claim. We also lack details about the main characteristics of the population, the selected covariates, balancing of any covariates after inverse probability of treatment weighting, the distribution of propensity scores, and whether there was sufficient overlap in the propensity score distribution between the fluoroquinolone and non-fluoroquinolone groups.
The findings are correct, but the lack of information on downstream side-effects of fluoroquinolones impairs risk-benefit assessments
The second possibility is that the results of the observational study are correct and that fluoroquinolones are actually more effective than other UTI-relevant antibiotics for treatment of uncomplicated UTI. However, important balancing measures addressing downstream side-effects of antibiotics were not included in the study by Daneman et al. These harmful events include subsequent emergence of resistant organisms, adverse effects of antibiotics, and C. difficile infections.
Fluoroquinolone use is one of the factors that has led to the spread of multidrug-resistant Escherichia coli strain sequence type 131. Several studies have indicated an association between fluoroquinolone use and tendonitis and tendon rupture, QT-prolongation, dysglycaemia, neuropathy and, potentially, aortic rupture. Compared with the first-line agents for treatment of UTI, fluoroquinolones confer a higher risk of C. difficile infections. Ultimately, this observational study is not able to provide important evidence on whether the collateral damage of fluoroquinolones outweighs the reported increase in treatment effectiveness.
The findings are correct, but changes in antibiotic resistance since the end of data collection make the results less relevant
The third possibility is that the results of the observational study are correct, but rising resistance of uropathogens to fluoroquinolones since end of data collection makes the findings less relevant. Although the authors found no difference when comparing the outcomes between 2005e2007, and 2012e2014, the resistance to trimethoprim-sulfamethoxazole and fluoroquinolones has climbed since then, making it challenging to apply these study findings to current clinical practice. For example, susceptibility of urinary E. coli to trimethoprim-sulfamethoxazole ranged from 68% to 78% in data compiled from 145 emergency departments across the USA from 2016 to 2017 .
Similarly, in the same study, susceptibility of urinary E. coli to fluoroquinolones ranged from 69% to 87%. As the result of increased prevalence of resistance to UTI-relevant antibiotics, the results of the Daneman et al. study may not be applicable in contemporary treatment of UTI, which makes it challenging to use their findings to inform current clinical practice.
The findings are correct, but the marginal benefits of using fluoroquinolones for acute uncomplicated cystitis are not worth the downside
The fourth possibility is that the results of this observational study are correct, however, the absolute risk reductions in repeat visits associated with fluoroquinolone use reflect small clinical benefit when weighed against potential consequences. For example, the absolute risk reductions for emergency department visits and hospitalizations were 0.28% and 0.05%, respectively, which means that the number of women with uncomplicated cystitis who would need to be treated with fluoroquinolones to prevent just one emergency department visit is 242. That number increases to over 1800 to prevent one cystitis-related hospitalization. Additionally, given the non-severe nature of many episodes of cystitis, treatment with a next-line agent that has lower toxicity than fluoroquinolones seems a reasonable strategy.
Take home message
In summary, a relatively modest magnitude of effect was found in a very large observational study using administrative data and showing superiority of fluoroquinolones compared with nonfluoroquinolone antibiotics for uncomplicated UTI. When deciding how to weigh these findings against the existing evidence, the devil is in the details. The potential impact of unmeasured confounders, concern for downstream effects of fluoroquinolones, and the rising rates of bacterial resistance to fluoroquinolones undermine the clinical applicability of these findings.
Strong consideration should be given to avoiding use of fluoroquinolones for cystitis to preserve the clinical activity of these important antibiotics for more invasive infections. Still, these provocative findings suggest that future randomized, controlled trials are necessary, especially given the changing picture of bacterial resistance among common uropathogens. For the time being, however, the FDA and the European Medicines Agency strongly caution against use of fluoroquinolones for uncomplicated UTI.
LG reports grants from the US National Institute of Health, Agency for Healthcare Research and Quality, Veterans Affairs Health Services Research and Development, and investigatorinitiated grant from Zambon Pharmaceuticals, Italy, outside the submitted work. BWT reports grants from the US National Institute of Health, Agency for Healthcare Research and Quality, Centers for Disease Control and Prevention, Veterans Affairs Health Services Research and Development, an investigator-initiated grant from Zambon Pharmaceuticals and personal fees from Paratek pharmaceuticals, outside the submitted work. JLS has no conflicts of interest to report.
No grant support was received for this commentary. BWT’s work is supported in part by the Houston Veterans Affairs Health Services Research & Development Center for Innovations in Quality, effectiveness and Safety (CIN 13-413).
Author : L. Grigoryan, J.L. Salemi, B.W. Trautner