Oral vs Intravenous Ciprofloxacin in the Initial Empirical Management of Severe Pyelonephritis or Complicated Urinary Tract Infections: A Prospective Randomized Clinical Trial

Background  There are few data on the efficacy of oral antibiotics in the initial empirical management of severe forms of urinary tract infection (UTI).

Methods  In a multicenter, prospective, randomized trial we compared oral (500 mg twice daily) vs intravenous ciprofloxacin (200 mg twice daily) in the initial empirical management of hospitalized patients with serious forms of UTI. Exclusion criteria were severe sepsis, inability to take oral medication, or the presence of obstruction or renal foci of suppuration. The study population included 66 women with pyelonephritis, 43 patients with community-acquired UTIs, and 32 patients with hospital-acquired UTIs. The frequency of bacteremia was 28 (42%) of 66 in the patients with pyelonephritis and 25 (33%) of 75 in those with complicated UTIs. Seventy-two patients were randomized to treatment with oral and 69 to intravenous ciprofloxacin.

Results  There were no infection-related deaths and no patients required an early change of antibiotics because of worsening clinical status during the initial empirical phase of treatment. The mean duration of fever was 1.7 days in patients treated by the oral vs 1.9 days in patients treated by the intravenous route (P=.15). The rates of microbiological failure (3% in the oral vs 2% in the intravenous treatment group) and of unsatisfactory clinical response (4% oral vs 3% intravenous) were low. A treatment change was eventually required in 14% of the patients assigned to the oral and 7% of the patients assigned to the intravenous regimen, mainly because of the isolation of enterococci or ciprofloxacin-resistant organisms in pretherapy urine specimens.

Conclusions  In the hospital setting, oral ciprofloxacin is as effective as the intravenous regimen in the initial empirical management of serious UTIs, including bacteremic forms, in patients without severe sepsis, obstruction, or renal foci of suppuration. The efficacy of the oral regimen indicates a potential use for ciprofloxacin in outpatient treatment of a subset of patients currently hospitalized on account of disease severity.

SEVERE FORMS of urinary tract infection (UTI) require immediate empirical treatment. Many authorities advise hospitalization and parenteral antibiotics, especially if bacteremia is suspected.^1-4 Several regimens have been recommended, including combinations (such as ampicillin or co-trimoxazole plus gentamicin) and single-drug regimens (such as third-generation cephalosporins, broad-spectrum β-lactam agents, or a fluoroquinolone).^1-8 However, these recommendations have not been based on the results of trials specifically conceived to delineate optimal initial treatment, which is generally empirical. In fact, studies on uncomplicated pyelonephritis have often excluded from efficacy analysis patients whose pretherapy isolates proved to be resistant to the study drug. With regard to complicated UTIs, the large majority of studies have focused on mildly symptomatic patients with known susceptibility of the causative agent to the assigned regimen. Finally, the recommendation for parenteral treatment, if bacteremia is suspected, is based on the unproven assumption that oral regimens are inadequate to treat bacteremic UTIs.

Since increasing attention is being given to cost-containment strategies, the issue of whether a subset of patients currently hospitalized for treatment of severe UTI can be safely managed on an outpatient basis is of particular importance. Parenteral regimens can be used to treat acutely ill patients at home, but they are impractical. Feasibility of outpatient management of serious UTIs requires the demonstration that oral regimens are effective in this setting. The present prospective, randomized study was specifically designed to compare oral with intravenous ciprofloxacin in the initial empirical management of serious pyelonephritis or complicated UTIs, including cases with suspected bacteremia. Feasibility of oral treatment was assessed in the subset of patients who were able to take oral antibiotics and did not have signs of severe sepsis, undrained obstruction, or renal foci of suppuration. Ciprofloxacin was selected because of the high bioavailability of its oral form, its broad spectrum of activity against uropathogens, and the favorable results reported in previous trials^9-16 when it was used in the treatment of various forms of UTI.

The study was a multicenter, prospective, randomized clinical trial. Recruitment took place in 4 community care centers in the south of Switzerland. During the study period (August 1992 through February 1996) all patients 18 years or older admitted for suspected UTI or who developed symptoms suggestive of UTI during hospitalization were considered for enrollment in the study. Patients who satisfied the entry criteria and provided informed consent were randomly allocated to oral or intravenous ciprofloxacin. Randomization was performed in each center according to a list of trial codes in sealed envelopes. The study protocol was approved by the local ethics committee.

Symptoms of UTI included lower urinary tract symptoms (dysuria, frequency, and urgency) and upper urinary tract symptoms (flank pain and costovertebral tenderness). Fever was defined as an axillary temperature of 37.5°C or higher, leukocytosis as a white blood cell count greater than 10x10⁹/L. Pyuria was defined as more than 10 leukocytes per high-power field in the sediment of centrifuged urine specimens. Urine for culture was obtained as a clean-voided midstream specimen in women, as a midstream specimen in men, and as a catheter-collected specimen in patients with an indwelling catheter. In women, cleansing was done with nonmedicated soap (detailed instructions were provided to ambulatory women). Quantitative cultures were performed using MacConkey and cystine-lactose-electrolyte-deficient media. Significant bacteriuria was defined as 10⁴ or more colony-forming units (CFUs) per milliliter on culture of midstream–voided urine specimens or at least 10⁵ CFUs/mL on culture of specimens collected from an indwelling catheter. Complicated UTIs were those associated with a structural or functional abnormality of the urinary tract (eg, an indwelling catheter, recent instrumentation, calculi, urinary tract obstruction, neurogenic bladder, or benign prostatic hyperplasia); those nosocomially acquired; or those occurring in men.¹⁷ Pathogens were identified according to standard methods and susceptibility to antimicrobials tested by disk diffusion.¹⁸

To be eligible for enrollment in this study, patients were required to have microscopic evidence of pyuria, symptoms of UTI or complicating urologic conditions, and fever or leukocytosis. Definitive inclusion in the study was confirmed by finding a significant bacteriuria in the pretreatment urine specimens. Cases where the uropathogen was not identified because of contamination of the urine during collection or because of recent antibiotic treatment were also included in the study.

Exclusion criteria were a history of allergic reaction or intolerance to ciprofloxacin; inability to take oral medication; severe sepsis (ie, at least 1 of the following: systolic blood pressure <90 mm Hg, oliguria, lactic acidosis, or an acute alteration in mental status)¹⁹; serum creatinine levels higher than 200 µmol/L or rapidly deteriorating renal function; obstruction not accessible to immediate drainage or renal foci of suppuration detected on ultrasonographic examination; high-risk conditions (eg, agranulocytosis or renal transplantation); or pregnancy. The administration of antibiotics other than ciprofloxacin within the 24 hours before admission was not considered a criterion for exclusion.

Patients allocated to intravenous treatment were given 200 mg of ciprofloxacin every 12 hours, infused over 30 minutes. Patients allocated to oral treatment received 500 mg of ciprofloxacin every 12 hours. Initial intravenous treatment was administered for a suggested minimum of 72 hours or at least until the patient was afebrile for 24 hours; treatment was then continued on the oral regimen (500 mg every 12 hours).

Initial evaluation included a complete medical history and physical examination, laboratory assessment of standard hematology parameters and of findings from plasma glucose, urea, creatinine, electrolyte, and liver function tests, urinalysis and culture, and 2 sets of blood cultures. In addition, renal ultrasonographic scans were performed in all patients within 12 hours of presentation. Axillary temperature was monitored twice daily, and patients were examined and questioned daily for signs and symptoms of UTI and for adverse effects of the treatment. At day 3 to 5 of treatment (before changing treatment in those patients who required a change), a second urine specimen (midstream-voided urine or catheter-collected urine) was collected for microscopic analysis to detect pyuria, for culture, and, in specimens yielding 10³ CFUs/mL or more, for identification of the organisms. In patients with documented bacteremia, blood was collected for culture if fever persisted for longer than 3 days. No attempt was made to blind investigators, but laboratory personnel were unaware of patients' treatment allocation.

Response to initial treatment was evaluated in terms of suppression of growth of the causative agent(s) at day 3 to 5 of treatment, and in terms of control of fever and UTI symptoms. In addition, adequacy of the initial empirical treatment was also evaluated in terms of the number of patients requiring treatment change. Microbiological failure of initial treatment was defined as the persistence of the causative agent by day 3 to 5 of treatment. Major clinical treatment failure was defined as the necessity to change initial empirical treatment before susceptibility testing results were available because of persistent fever (body temperatures of 39°C or higher) and/or persistent chills or because of the appearance of signs of severe sepsis. Minor clinical treatment failure was defined by 1 of the following: persistence of fever or UTI symptoms at the time when the infecting organism was reported to be resistant in vitro or to be an enterococcus; persistence of fever and/or UTI symptoms for 4 days or longer if the infecting organism was susceptible in vitro; or inability to retain oral medication in patients randomized to oral ciprofloxacin. Necessity for a change in treatment was defined by 1 or more of the following: major or minor clinical failure; microbiological failure; isolation under treatment of 10³ CFUs/mL or more of a new pathogen considered to require a treatment change (the decision to change treatment in these cases was taken individually, depending on the number of CFUs, the presence of an indwelling catheter, the response of pyuria, and the clinical response); and/or in vitro resistance of the causative agent(s) (the isolation of enterococci was considered to mandate a treatment change regardless of the results of the disk test).

Descriptive analysis included means or percentages with 95% confidence intervals (CIs) or medians and ranges, as appropriate. Comparison between therapy groups was made using Student t test or Wilcoxon rank sum test, as appropriate, and the χ² statistic (statistical significance P=.05).

A total of 163 patients were enrolled in the study. Of these patients, 83 were randomized to the oral and 80 to the intravenous regimen. Twenty-two patients (11 in each treatment group) were excluded secondarily for the following reasons: 5 patients did not fulfill inclusion criteria (they had asymptomatic catheter-associated bacteriuria), and 1 patient had an exclusion criterion (a rapidly deteriorating renal function); 11 patients did not have UTIs retrospectively, but harbored an infection of another kind or site (pulmonary infection, n=6; staphylococcal endocarditis, n=1; disseminated gonococcal disease, n=1; viral illnesses, n=3). In 2 patients symptoms were, in retrospect, unrelated to infection (renal infarction, n=1; polymyalgia rheumatica, n=1). In 2 patients collection of data was incomplete and 1 patient withdrew consent. No patients were excluded after study entry on the basis of obstructive uropathy or foci of renal suppuration revealed by ultrasonographic scan.

Seventy-two of the 141 evaluable patients were randomized to oral and 69 to intravenous ciprofloxacin. Demographic and clinical characteristics of the patients are summarized in Table 1. Women with uncomplicated pyelonephritis made up 66 (47%) of the patients. Half of these women were premenopausal, with a peak age in the 20s. Fifty-eight (77%) of the patients with complicated UTIs were men, 45 (60%) were patients older than 60, and the infection was nosocomially acquired in 32 cases (43%). Diabetes was present in 14 (21%) of the women with pyelonephritis and in 12 (16%) of the patients with complicated UTIs; other underlying diseases were present in 15 patients with pyelonephritis (23%) and in 46 patients with complicated UTIs (61%). Apart from diabetes, the most common underlying conditions were cardiovascular disease (25 patients, 18%), cerebrovascular disease and/or dementia (17 patients, 12%), and cancer (13 patients, 9%), including 4 cases of prostate and 3 cases of urothelial cancer. Among patients with complicated UTIs, the most common associated factors were the presence of an indwelling catheter in 23 patients (31%), recent instrumentation (mainly short-term catheterization) in 17 patients (23%), benign prostatic hyperplasia in 12 patients (16%), and obstructive uropathy and/or stones in 6 patients (8%). Four patients (2 allocated to oral and 2 to intravenous regimen) had received 1 to 2 doses of norfloxacin (n=2), co-trimoxazole (n=1), or amoxicillin-clavulanic acid (n=1) within 24 hours before entering the study.

At presentation, fever was present in 86% of the patients, the median temperature in febrile patients being 38.7°C (range, 37.5°C-40.7°C) in the oral treatment group and 38.7°C (range, 37.6°C.-40.0°C) in the intravenous treatment group. Symptoms of UTI were present in 84% of the patients with pyelonephritis and 46% of the patients with complicated UTIs.

The causative agent could not be identified in 10 patients, all admitted with fever, UTI symptoms, and pyuria. Four of these patients had negative urine cultures after recent exposure to antibiotics, whereas in 6 patients the urine yielded a mixed flora, which was considered contamination superimposed on infection, and the organisms were not identified further. All women with pyelonephritis and documented infection had 100,000 CFUs/mL or more of a single uropathogen. Polymicrobial infections were observed in 8 patients with complicated UTIs. Five of these patients had an indwelling catheter in place, and 3 had had recent short-term catheterization. The frequency of blood cultures positive for organisms was 53 of 141 (38%) overall, 28 of 66 (42%) in uncomplicated pyelonephritis, and 25 of 75 (33%) in complicated UTIs.

The spectrum of uropathogens isolated in the 131 documented infections and their susceptibility to ciprofloxacin are summarized in Table 2 and Table 3. In vitro resistance to ciprofloxacin was associated with enterococcal infections in 5 cases (3 of which were mixed infections), with infections caused by gram-negative bacilli in 3 cases (Pseudomonas aeruginosa, n=1; Escherichia coli, n=1; Enterobacter species, n=1), staphylococcal infections in 2 cases (Staphylococcus aureus, n=1; coagulase-negative staphylococci, n=1), and 1 fungal infection. On disk test, enterococci were intermediately resistant in 5 and sensitive in 2 cases. With regard to other antibiotics commonly used in the treatment of UTI, resistance rate to ampicillin was 33% in pyelonephritis and 56% in complicated UTI; resistance to co-trimoxazole was 12% and 29%, and resistance to gentamicin was 3% and 12%, respectively. As indicated in Table 1, Table 2, and Table 3, there were no significant differences between the 2 treatment groups with respect to demographic characteristics, type, signs and symptoms of infection, susceptibility to antibiotics, and frequency of associated bacteremia.

The results of response to initial treatment are summarized in Table 4. Fever and UTI symptoms abated rapidly in most patients without significant differences between the 2 treatment regimens. Moreover, there were no differences between the 2 regimens within patient subgroups (pyelonephritis in premenopausal or postmenopausal women, community-acquired complicated UTIs, hospital-acquired UTIs) (data not shown). In uncomplicated pyelonephritis (but not in complicated UTIs), the duration of fever was longer in the subset of patients with associated bacteremia (mean duration, 2.4 days [95% CI, 2.0-2.7 days]) than in patients without bacteremia (1.4 days [95% CI, 1.2-1.7 days]) (P < .002). In this same subset of patients, however, there was no difference between the 2 treatment groups (mean duration of fever in patients with bacteremia treated with oral ciprofloxacin, 2.2 days [95% CI, 1.7-2.6 days], vs 2.6 days [95% CI, 2.0-3.2 days] in patients with bacteremia treated intravenously) (P=.18).

Assessment at day 3 to 5 of treatment documented 3 cases of microbiological failure. All of these patients had complicated UTI and harbored in vitro–resistant microorganisms (P aeruginosa, n=1; S aureus, n=1; Candida albicans, n=1). In 8 patients (6 with an indwelling catheter), culture under treatment yielded a new microorganism (mainly 10³-10⁴ CFUs/mL of coagulase-negative staphylococci, resistant in vitro to ciprofloxacin). Because of the decrease in pyuria and the favorable clinical response (in particular, 4 of the patients with indwelling catheters had bacteremic infections that were responding well to treatment), none of these patients required a change in the initial treatment.

There were no infection-related deaths, nor did any patient require a change of antibiotics because of clinical deterioration during the initial empirical phase of treatment. In the case of 1 woman with pyelonephritis randomized to oral ciprofloxacin treatment, vomiting precluded further oral therapy and she was continued on the regimen with intravenous ciprofloxacin. The 2 other cases of minor clinical failure in the oral treatment group were of 1 woman with pyelonephritis caused by a resistant strain of E coli (she was changed to a gentamicin regimen) and 1 man with complicated UTI caused by enterococci plus Klebsiella species (he was changed to therapy with amoxicillin-clavulanic acid). The 2 minor clinical failures in the intravenous treatment group were 1 woman with pyelonephritis, whose causative agent was not identified (pretreatment specimen contaminated), who had persistent UTI symptoms after 4 days of treatment (she received gentamicin plus ampicillin); and 1 man with complicated UTI caused by a resistant strain of P aeruginosa (he was changed to treatment with imipenem-cilastin).

In addition to the 7 patients with microbiological and/or clinical failure, initial treatment had to be changed in 5 patients receiving oral and 3 patients receiving intravenous treatment because of the isolation of enterococci (n=6) or other in vitro–resistant strains (n=2) in pretreatment urine specimens. All of these patients (including 2 with bacteremic enterococcal infection) were responding favorably to ciprofloxacin and had sterile urine cultures when the treatment was changed.

The treatment was generally well tolerated. Adverse effects probably related to ciprofloxacin were observed in 2 patients. In 1 patient (randomized to initial oral treatment), ciprofloxacin treatment had to be discontinued because of the development of mental confusion. A second patient (randomized to initial intravenous treatment) developed pruritus, but the treatment could be completed.

There were no deaths during treatment. Three patients with nosocomially acquired UTIs (1 randomized to the oral and 2 to the intravenous treatment group) died of unrelated causes (lung cancer in 2, bladder cancer in 1) during hospitalization. In the other 138 patients there were no relapses or reinfection requiring treatment during the same hospitalization, and they were discharged home (n=131) or transferred to a nursing home (n=7). The duration of hospitalization was no different in the 2 treatment groups and depended largely on the presence of associated diseases and/or on the patient's age (Table 4). In the subset of patients with pyelonephritis in particular, the median duration of hospitalization was 4 days for premenopausal and 11 days for postmenopausal women.

Among the 22 patients excluded from efficacy analysis, 2 patients (1 allocated to oral and 1 to intravenous treatment) died (myocardial infarction in one, refractory pulmonary edema in the other). The other 20 patients were discharged after a median hospitalization of 14 days in patients allocated to regimens with oral and 10 days in patients allocated to regimens with intravenous ciprofloxacin. In particular, the 8 patients with bacterial infections other than UTI (3 allocated to oral and 5 to intravenous regimens) received further treatment according to their specific diagnoses, had a favorable clinical outcome, and were discharged 2 to 29 days after admission.

Previous studies on the treatment of UTI with oral ciprofloxacin have not been comparative,^11-13 have included only mild forms of complicated UTI¹⁴ or UTIs in which the susceptibility of the causative agent was known,^13,15 or have excluded patients with associated bacteremia.^11,13 Studies specifically addressing the issue of the feasibility of outpatient or of oral treatment of severe UTIs have been retrospective²⁰ or have excluded patients with associated bacteremia.^21,22 The present study is prospective and focuses on the initial empirical phase of treatment. Patients with severe sepsis, defined as the presence of infection-related organ dysfunction, were excluded¹⁹; however, patients with serious infections, many of whom proved to have bacteremia, were enrolled. In spite of the differences in spectrum of uropathogens, antimicrobial sensitivity, and propensity for relapse between uncomplicated pyelonephritis and complicated UTIs, both types of infection were included if obstruction or foci of renal suppuration were absent. In fact, the present study assessed the efficacy of the initial treatment in controlling symptoms and suppressing bacteriuria. The achievement of these goals, which is the first consideration in pyelonephritis as well as in complicated UTIs, depends primarily on the efficacy of the initial antibiotics, if complications requiring a specific urologic approach are excluded. Once the initial symptoms of infection are controlled, therapy can be modified according to susceptibility test results; several oral regimens have proved to be safe and effective for the completion of treatment.^1-8

The present results document the efficacy of ciprofloxacin in the initial empirical management of severe UTIs, and show that the oral regimen is as effective as the intravenous one. Given the high frequency of bacteremia, the first important result of the study is that there were no infection-related deaths and that no patient required rescue treatment because of clinical deterioration during the initial empirical phase of therapy. The duration of fever was longer in the subset of patients with pyelonephritis with bacteremia, but in these patients too, oral ciprofloxacin was as effective as the intravenous regimen. Minor clinical failures were observed in 3 patients assigned to oral and 2 patients assigned to intravenous treatment; microbiological failures (all associated with in vitro–resistant microorganisms) occurred in 2 patients receiving oral and 1 patient receiving intravenous treatment. Monitoring for relapse by microbiological methods was not done; this is the main limitation of this study. It is reassuring, however, that the clinical course was uncomplicated in all cases (including those patients who, in retrospect, had bacterial infections other than UTIs), with no significant difference in the length of hospital stay between the 2 treatment groups. Finally, the treatment was well tolerated; there was only 1 adverse reaction requiring discontinuation of treatment with ciprofloxacin.

Concerning the efficacy of ciprofloxacin in the subset of patients with uncomplicated pyelonephritis, our data can be summarized as follows: 98% of the causative agents were sensitive to ciprofloxacin in vitro; during therapy, the causative agents were suppressed in the urine in all cases; and there were 3 minor clinical failures, including 1 failure of the orally administered drug, which, however, was managed successfully with intravenous ciprofloxacin. In accordance with previous studies,^20,23E coli was the most frequent uropathogen (93% of the isolates) and Staphylococcus saprophyticus the most common gram-positive agent (3%). Enterococci were not encountered and are in fact rare in uncomplicated pyelonephritis in nonpregnant women.^20,23,24 The rate of resistance to ampicillin was 33%, a finding consistent with previous studies,^20,23,24 whereas the rate of resistance to co-trimoxazole (12%) was higher than previously reported.^20,23,24

Resistant infections were more frequent in complicated UTIs (10 of 70 patients, 14%) and were often caused by gram-positive cocci, particularly enterococci. The isolation of enterococci (in monomicrobial or mixed infections) was also the most frequent reason for modification of treatment when culture and susceptibility testing results became available. However, while the protocol advised a change of treatment regardless of the disk test results, enterococci were suppressed in the urine in all patients during treatment, and the clinical response to ciprofloxacin was favorable in 6 of 7 patients (including 2 cases of bacteremic infection). Therefore, while suspected or documented enterococcal infections must still be treated with other regimens, our observations indicate that empirical ciprofloxacin is partially effective against this pathogen until treatment can be adapted to the culture results.

In conclusion, our data demonstrate that oral ciprofloxacin is as effective as the intravenous regimen in the initial empirical treatment of severe UTIs, including bacteremic forms, provided that severe sepsis, obstruction, or foci of renal suppuration are not present. Since oral treatment is feasible at home, the present findings indicate a potential use for ciprofloxacin in the outpatient treatment of a subset of patients currently hospitalized because of infection severity. Our results, however, were obtained in the hospital setting and might not apply to the outpatient setting for several reasons: first, hospitalization allowed more accurate diagnostic evaluation; second, patients had standard supportive hospital care, including intravenous fluids in most cases, which is not routinely feasible at home; finally, hospitalization provided the opportunity for close monitoring of clinical response and ability to retain oral treatment. Therefore, before oral ciprofloxacin can be recommended for outpatient treatment of severe UTIs, our results need to be confirmed by studies directly comparing inpatient vs outpatient treatment. Such studies are needed urgently since UTIs are a frequent cause of hospitalization²⁵ and outpatient treatment would permit considerable financial savings.²⁰

Accepted for publication May 18, 1998.

We thank E. Zucca, MD, for his help in statistical analysis and P. Gafner for typing the manuscript.

Reprints: G. Mombelli, MD, Reparto di Medicina Interna, Ospedale Regionale di Locarno, CH-6600, Locarno, Switzerland (e-mail: odlin1@tinet.ch).