Efficacy and safety of quinolones for the treatment of uncomplicated urinary tract infections in women: a network meta-analysis

Uncomplicated urinary tract infection (uUTI) is defined as the presence of pathogenic organisms in the urinary tract without anatomical and functional abnormalities, is accompanied by inflammatory leukocytes and cytokines and may or may not develop clinical symptoms. The frequency of uncomplicated urinary tract infection is higher in young women. Several quinolone treatment regimens are available; however, since we do not know which is the best antibiotic regimen for the treatment of this urinary infection, we analyzed the published evidence and conducted a systematic review with network meta-analysis. The aim was to compare and hierarchize quinolones according to their efficacy and safety and to identify the best treatment for uncomplicated urinary tract infection in women through a systematic review with network meta-analysis. Medline, Embase, LILACS, Cochrane CENTRAL and other databases were searched for trials. Bias in the trials was assessed using the Cochrane Collaboration tool. To analyze efficacy and adverse events, for direct comparisons, we obtained risk ratios and 95% confidence intervals by applying a fixed-effects model using tau2 and Q2 tests to calculate the heterogeneity. For the network meta-analysis, we analyzed the indirect comparisons by Bucher’s method. We included 18 trials (8765 women). For premenopausal women, ofloxacin had a 57% probability of achieving remission but an 83% frequency of adverse events. For postmenopausal women, ofloxacin was 82% more effective for remission, with a 49% frequency of adverse events, compared with other types of quinolones. Compared with other quinolones, ofloxacin 200 mg once daily for a treatment duration < 3 days provides the highest clinical and bacteriological remission rates with the lowest relapse and resistance rates for the treatment of women with uUTIs. However, additional trials are needed to confirm our findings, especially when the treatment duration exceeds 3 days.

UTIs are regarded as infections with low morbidity and mortality in the community environment, but they have a high incidence, with 250 million cases reported annually worldwide, are more frequent in young women (approximately 0.5 episodes per person per year) and represent up to 5% of primary medical health care visits [1,[3][4][5].
This high incidence may be attributed in premenopausal women to sexual activity and a history of UTI in the last 12 months and in postmenopausal women (> 65 years) to urinary incontinence, the prolapse of pelvic organs and vaginal infections as a result of the change in the vaginal flora secondary to the decrease in estrogen levels [4][5][6].
Overall, the most frequent pathogens that cause uUTIs are gram-negative bacilli (80% to 90% Escherichia coli, Klebsiella pneumoniae and Proteus mirabilis), followed by group B Streptococcus, Enterococcus faecalis and Staphylococcus saprophyticus; however, the selection of antimicrobials for the treatment of uUTIs depends not only on the causative organism but also on the site of infection, dose and time of administration of the antibiotic regimen [4][5][6][7].
Due to the above, the guidelines of the Infectious Diseases Society of America recommend the use of TMP/SMX (160/ 800 mg twice daily for 3 days) or nitrofurantoin (100 mg twice daily for 5 days) as a first-line treatment for urinary tract infection [8]. However, some low-income countries have reported resistance rates > 20% for TMP/SMX; therefore, studies have been conducted to identify other effective antibiotics for the treatment of UTIs.
Several clinical trials have concluded that the administration of agents of the quinolone family is an appropriate therapeutic option [9,10]. However, due to their variety, it has not been possible to identify which quinolone is most effective and safe for the treatment of uUTIs because it would be necessary to perform multiple head-to-head trials. Therefore, this has resulted in clinicians prescribing drugs as per convenience, without considering their dose, adverse effects and time of administration necessary for the healing of the patients [11][12][13].
Based on the above, we conducted this systematic review with a network meta-analysis to compare and hierarchize quinolones according to their efficacy and safety and to identify the best treatment for uUTIs in women.

Search strategy
Two reviewers (RRL and AGG) conducted a search of the following databases: Medline, Embase, LILACS, Cochrane CENTRAL and the WHO International Clinical Trials Registry Platform. In addition, references cited in the identified studies and relevant abstracts presented at various conferences (e.g., Infectious Diseases Society of America, European Association of Urology and American Urological Association) were manually searched without language restriction. The search included the following MeSH terms:

Study and participant selection criteria
We included any randomized controlled trial that examined the use of any quinolone in healthy adult women who had a normal urinary tract, had not been catheterized and had been diagnosed with a uUTI according to any of the following criteria: (1) culture (> 10,000 colony-forming units/ml); (2) urinalysis (> 20 leucocytes per field, > 3 erythrocytes per field and the presence of nitrites); or (3) clinical symptoms (e.g., dysuria, urinary urgency, urinary frequency or suprapubic pain).
The quinolone regimens were adjusted according to age, dose and time of treatment to evaluate the following outcomes: (1) efficacy: clinical remission (the remission of the clinical symptoms such as dysuria, urgency and urinary frequency or suprapubic pain experienced by patients) and bacteriological remission (negative urine culture after treatment); (2) safety (according to the frequency of any adverse events); and (3) relapse rate (the reappearance of signs and symptoms of a UTI from which the participant was convalescing) and resistance rate (continued presence of signs and symptoms of a UTI despite treatment).
A study was considered ineligible if the following occurred: the participants presented with immunosuppression, renal failure, pyelonephritis, chemotherapy or pregnancy; prophylactic antibiotics or catheters were used; UTI was considered complicated (functional or structural abnormalities of the genitourinary tract); or the studies included drugs withdrawn by the FDA (sparfloxacin, lomefloxacin, gatifloxacin and temafloxacin). Crossover, quasi-experimental, noninferiority, observational, narrative, case report and consensus studies were also excluded from this review.

Assessment of the risk of bias
Two independent reviewers (AGG, LVH) analyzed the study titles and abstracts to determine their inclusion; disagreements were resolved by discussion and consultation with a third reviewer (EOH). We used the Cochrane "risk of bias" assessment tool to judge the risk of bias for the following individual items: (1) random sequence generation, (2) allocation concealment, (3) blinding of participants and personnel, (4) incomplete data and (5) selective reporting of information [12].
Discrepancies were resolved by discussion with the third reviewer (EOH); for cases involving unclear information, the authors were contacted via email [12].
We present an adapted Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram to show the process of trial selection (Fig. 1) and the assessment of the risk of bias (Fig. 2, Supplementary Fig. 1) [14].

Data extraction
This process was performed independently by two researchers, who used a standardized form that included the following information: author's name, the year of publication, country, the number of participants included, treatment and dosage, the duration of follow-up, outcomes analyzed and findings obtained (Table 1).

Statistical analysis
Direct comparisons For each pairwise comparison and each outcome, we obtained RRs with 95% CI as a measure of the association between the interventions. Conventional metaanalyses were conducted using a fixed-and random-effects model, with the inverse variance for each outcome and comparison. We used the standard chi 2 test with a significance level of 0.1. Heterogeneity was considered important when the I 2 value was > 50%.
For the assessment of publication bias, we formed funnel plots to evaluate asymmetry and confirmed the findings using Egger's test [15].
For all outcomes, we conducted an analysis according to the intention to treat; in cases where this information was not reported, we contacted the study authors.
Indirect comparisons Since a network meta-analysis is a method of synthesizing information from studies with the same outcomes but different interventions, it requires the information of direct and indirect comparisons between interventions to calculate a single effect size. An indirect comparison is the relative effect obtained from different treatments adjusted according to the results of their direct comparisons with a common comparator (transitivity).
To obtain indirect comparisons and to generate the network, it was necessary to use TMP/SMX (160/800 mg twice daily) as a common comparator because it was the most commonly used drug in the included clinical trials, it is not a quinolone and it is considered the conventional treatment for uUTIs according to the Infectious Diseases Society of America [8]. We calculated the indirect comparisons using Bucher's method, considering a cutoff value of 0.05.
Subsequently, we analyzed the inconsistency in each loop using the τ test; if all loops in the net had consistency, we performed a correlation matrix and obtained the SUCRA, which shows the cumulative probability of an intervention being among the best options, using STATA software v15.1 [13,16,17]. This review was registered and approved in the International Prospective Register of Systematic Reviews (PROSPERO: CRD42015025886).
We identified seven different quinolone-based treatment regimens (ciprofloxacin 100, 250 and 500 mg; levofloxacin 250 mg; norfloxacin 400 mg; ofloxacin 200 and 400 mg) with different administration times (1 to 14 days) administered to women of different ages (18 to 80 years). Therefore, to reduce the heterogeneity, we grouped the population into four groups of patients according to age (pre-and postmenopausal) and the duration of treatment (< 3 days and > 3 days), with which we could generate a network meta-analysis. For details, please see Table 1.
Fourteen percent of the trials were considered to have a high risk of bias due to the lack of detailed descriptions for random sequence generation, allocation concealment and blinding of participants and personnel. We noted that 61% of trials had dropout rates > 30%. However, 97% of the trials properly reported selective bias and other potential sources of bias ( Fig. 2 and Supplementary Fig. 1).

Clinical remission
The six analyzed trials [19,23,25,26,28,30] involved six different types of interventions (ciprofloxacin 100, 250 and 500 mg; norfloxacin 400 mg; ofloxacin 200 and 400 mg) (for regimens of administration, see Table 1). We calculated both direct and indirect comparisons and generated a network plot with the cumulative ranking curve plots (Supplementary Table 1). The IF was not significant (p = 0.84).
Overall, we did not observe a significant difference between regimens when comparing different types of quinolones or TMP/SMX. The overall ranking curve plots the antibiotics most likely to yield a clinical remission of UTIs, such as ciprofloxacin 250 mg and ofloxacin 200 mg, with an area under the ranking curve of 58.5% and 57.5%, respectively (Table 2).

Bacteriological remission
The meta-analysis was performed with six trials, and we generated a network plot with a nonsignificant IF (p = 0.95). The cumulative curve plots indicated that ciprofloxacin 100 mg and ofloxacin 200 mg were most likely to yield bacteriological remission with an area under the ranking curve of 65.5% and 63.2%, respectively ( Table 2, Supplementary Table 2).

Adverse events
The main adverse events reported in these trials were gastrointestinal issues (diarrhea, nausea and vomiting), dizziness, headache, rash and genital itching (for details, please see Table 1). The related network meta-analysis included six trials   Table 3) [19,23,25,26,28,29]. The ranking curve plots reported that the antibiotic associated with a lower risk of developing any adverse events was ciprofloxacin (100 and 250 mg), with an area under the curve of 26.4% to 29.5% and 35.1%, respectively (Table 2).

Relapse and resistance
For relapse, a meta-analysis was performed with the same trials with a nonsignificant IF (p = 0.74), in which we observed that ciprofloxacin (250 and 500 mg) was the antibiotic with the highest probability of relapse (77.7% to 80.4%, respectively) ( Table 2, Supplementary Table 4). For the resistance rate, we could not perform the analysis because of the heterogeneity among the studies.
Finally, we compared the cumulative probabilities for each outcome to identify the overall efficacy and safety of the quinolone regimens and observed that 200 mg ofloxacin once daily was the quinolone with better probability of clinical and bacteriological remission, with a low frequency of relapse rate but with the highest frequency of adverse events compared with the other types of quinolones (Fig. 5).

Clinical remission
Of the seven analyzed trials [21,22,28,[30][31][32]34] involving six different types of interventions (ciprofloxacin 100, 250 and 500 mg; levofloxacin 250 mg; norfloxacin 400 mg; ofloxacin 200 mg) (for details, please see Table 1), we Overall, we did not observe a significant difference between regimens when comparing either quinolones or TMP/ SMX, with the exception of ofloxacin 200 mg (RR 1.16; 95% CI 1.02 to 1.32; p = 0.023). In the ranking curve plots, the antibiotics most likely to yield a clinical remission of UTIs were ciprofloxacin 500 mg and ofloxacin 200 mg, with an area under the ranking curve of 82.6% and 75.3%, respectively (Table 3, Supplementary Table 5).

Bacteriological remission
We performed a network plot with a nonsignificant IF (p = 0.68). Ciprofloxacin 250 mg was the only quinolone that demonstrated a significant difference compared with TMP/SMX (RR 1.10; 95% CI 1.0 to 1.21; p = 0.04). The area under the curve plots indicated that ciprofloxacin 100 mg was most likely to yield bacteriological remission, with a cumulative probability of 79.6% (Table 3, Supplementary Table 6).

Adverse events
The adverse events in these trials were the same as those reported in studies with premenopausal women. The related network meta-analysis included seven trials [20-22, 28, 29, 32, 34] with an IF of p = 0.76 (for details, see Table 1). Treatments associated with a lower risk of any adverse event were ofloxacin 200 mg (RR 0.56; 95% CI 0.36 to 0.88; p = 0.013) and levofloxacin 250 mg (RR 0.52; 95% CI 0.31 to 0.87; p = 0.013) compared with TMP/SMX. The cumulative curve plots reported that 250 mg levofloxacin was the quinolone with the smallest area under the curve to develop adverse events (28.6%) ( Table 3, Supplementary Table 7).

Relapse and resistance
We could not perform an analysis of relapse and resistance because of the heterogeneity among the studies. However, for resistance, we generated the network with five trials [20,22,28,29,32] with an IF of p = 0.44. The treatment associated with a lower risk of resistance was ofloxacin 200 mg, with an   (Table 3, Supplementary  Table 8). Subsequently, we compared the area under the ranking curves for each outcome and observed that ciprofloxacin 500 mg was the quinolone with the best probabilities of clinical remission but with a high frequency of adverse events compared with the other types of quinolones (Fig. 6).
With respect to the analysis of the duration of treatment > 3 days, we could not generate a network in any outcome because of the heterogeneity among the studies.

Discussion
UTIs are considered infections with low morbidity and mortality and are commonly caused by gram-negative bacilli; their treatment depends mostly on the type of bacteria. Currently, the Infectious Diseases Society of America recommends the use of TMP/SMX as a first-line antibiotic; however, resistance rates > 20% have been reported in some countries worldwide, encouraging a search for alternative drugs active in secondline regimens.
Although quinolones share similar characteristics, some of them are associated with specific adverse events, making it unsafe to assume that they are interchangeable. Therefore, we conducted this network meta-analysis to hierarchize the quinolones and to identify the best treatment for patients with uUTIs.
This review included 18 trials with 7 different treatment regimens (ciprofloxacin 100, 250 and 500 mg; levofloxacin 250 mg; norfloxacin 400 mg; ofloxacin 200 and 400 mg) with administration times from 1 to 14 days administered to 8765 women aged 18 to 80 years. In this study, we observed that despite diversity among the studies, only 14% had a high risk of bias due the lack of detailed descriptions for random sequence generation and high dropout rates.
Overall, regarding the clinical and bacteriological remission rates, we did not observe significant differences for any type of quinolone compared with TMP/SMX. Nonetheless, when comparing the efficacy and safety among quinolones, we observed that there were apparently disagreements in our  findings since a higher dose of ciprofloxacin appeared to decrease the rate of bacteriological remission, while the use of higher doses of ofloxacin seemed to reduce the frequency of adverse events. This could be due to the high dropout rate of trial participants (> 30%) and some trials not being included in all quantitative network analyses, which limits comparisons between studies. However, despite the above limitations, we observed that ofloxacin 200 mg once daily not exceeding 3 days of therapy duration was the most effective quinolone regimen for clinical and bacteriological remission of uUTIs in premenopausal women, with a low probability of relapse compared with other types of quinolones; however, a high probability of developing adverse effects (mainly dizziness, nausea and vomiting) was reported for this regimen. Therefore, an alternative that could be used for the treatment of uUTIs is norfloxacin 400 mg twice daily, which has good effectiveness for clinical and bacteriological remission, with a higher relapse rate than ofloxacin, but with a low probability of adverse events.
Regarding postmenopausal women, ofloxacin 200 mg was found to yield the best probabilities of clinical remission and low resistance rates and adverse effects. Although ciprofloxacin 500 mg once daily showed the best probability of clinical remission, a high probability of resistance and of developing adverse effects was reported; therefore, it might not be the best choice for the treatment of uUTIs.
These findings coincide with the results obtained in the studies of Gupta, Sotomayor and Zalmanovici [8,10,11,42] in the sense that they recommend the use of these quinolones for short periods as an alternative for the treatment of UTIs.
It is worth mentioning that although the cost of TMP/SMX is lower (US $2.12 per 100 tablet) than that of ofloxacin (US $4.97 per 100 tablet) worldwide, there are more people allergic to TMP/SMX (3 to 5%) than to ofloxacin (0.4 to 2%) [43][44][45][46]. We hence consider that these findings could be taken into account when building an appropriate treatment strategy for uUTIs, without the need to use the latest-generation quinolones, which, although they are more effective, also have a high risk of causing adverse events, which favors the abandonment of treatment and the increase in bacterial strains resistant to antibiotic treatment.
Limitations of the review: Due to the great diversity of interventions in the trials included in this review, and as a consequence of the use of the age of the participants, different doses and administration times, we had to reduce the heterogeneity existing through the generation of subgroups. However, this limited the analysis of those studies that analyzed treatment durations > 3 days.
It is worth mentioning that despite these limitations, we were able to perform the network meta-analyses for all results without substantial inconsistency. However, we suggest additional studies in which there is a dropout rate < 30% to improve the findings. Implications for practice: The results of this study suggest that ofloxacin would be a good option in terms of efficacy and safety if a quinolone is to be selected for the treatment of uUTIs. However, it would be advisable to carry out new studies that incorporate evidence on treatment regimen durations > 3 days and cost analysis studies to assist in the selection of a particular quinolone.

Conclusions
In this study, we did not observe significant differences for any type of quinolone compared with TMP/SMX. Nonetheless, we observed that, compared with other quinolones, ofloxacin 200 mg once daily for treatment durations < 3 days provides the highest clinical and bacteriological remission rates, with the lowest relapse and resistance rates for women with uUTIs, albeit at a high probability of adverse events such as dizziness, nausea and vomiting. We consider norfloxacin 400 mg twice daily to be an alternative for the treatment of uUTIs, as it has a low probability compared with other quinolones of leading to adverse events in pre-and postmenopausal women.
Additional trials with treatment regimen durations > 3 days and cost analysis studies to assist in the selection of a particular quinolone are recommended.