Acute pyelonephritis (aPN) is a severe form of urinary tract infection (UTI), which results from bacterial invasion of the renal parenchyma. The incidence of aPN is estimated at around 9-11 cases per 10,000 inhabitants and is four times more frequent in women than in men.1
Diagnosis is based on clinical criteria (fever, costovertebral pain, and dysuria symptoms) and microbiological criteria [pyuria and a positive urine culture with ≥104 colony-forming units (cfu) per milliliter of urine].2
Bacteria usually reach the kidneys by ascending from the lower urinary tract or, more rarely, through the bloodstream.5
Several factors may be involved in the etiology of aPN: anatomic or functional abnormalities of urinary tract drainage6 or metabolic disorders7,8 (intravenous drug abuse and endocarditis are implicated in gram-positive hematogenous infections).
Complicated aPN was defined as aPN occurring in any male patient, in patients with functional or anatomical abnormalities of the urinary tract, immunosuppressed persons, patients with a single kidney, permanent bladder catheter, nephrostomy or double-J catheter, or those patients who had experienced urinary tract manipulation in the previous two weeks.9
Pyelonephritis becomes a potentially fatal disease when secondary conditions develop, such as emphysematous pyelonephritis (20-80% mortality rate), perinephric abscess (20-50% mortality rate), or one of the sepsis syndromes (>25% overall mortality rate).4 Pregnant women are more likely to develop pyelonephritis and pregnancy is associated with major risk of complications.10
The extensive use of antibiotics has increased antibiotic resistance during recent years.13,14 For example Escherichia coli rates of resistance have reached 50%, 20%, and 10%-20%, respectively for ampicillin, fluoroquinolones, and trimethoprim/sulfamethoxazole (TMP-SMX).15 Given the high incidence of acute pyelonephritis in the community setting, measures should be taken to avoid further development of antimicrobial resistance.16
In the current International Clinical Practice Guidelines for the treatment of acute pyelonephritis, the recommended duration of treatment for pyelonephritis is 7 days for fluoroquinolones, 10-14 days for b-lactams and 14 days for TMP-SMX.17-19
No recommendation is provided for women or men hospitalized with acute pyelonephritis.20
The guidelines address only young, otherwise healthy women who are not pregnant. Best management for elderly individuals, men and patients with comorbidities remains unclear. Despite publication of the guidelines, studies demonstrate a wide variation in prescribing practices regarding the selection of antimicrobial agents and duration of therapy.21-23
A reduction in the duration of the administered regimes could minimize the selection pressure on potential pathogens, thereby reducing the risk for the emergence of new resistant strains.24
Under this perspective, we sought to compare short-course with long-course treatment with the same antibiotic regimes, administered by the same route and in the same total daily dosage for acute pyelonephritis, in terms of effectiveness and tolerability, by performing a systematic review and metaanalysis of randomized controlled trials (RCTs).
Materials and Methods
We included RCTs that compared a long-course versus a short-course antibiotic therapy of the same antibiotic agents administered by the same route and in the same total daily dosage. Long course was defined as a therapy, which lasts at least 2 days longer than the corresponding short-course treatment. The participants were eligible for inclusion if they were 18 years and older with acute PN, diagnosed based on clinical criteria (fever, costovertebral pain, and dysuria symptoms) and microbiologic criteria (pyuria and a positive culture with >=104 cfu per milliliter of urine. Pregnant women and patients, both hospitalized and outpatients, with anatomical or functional abnormalities of the urinary tract, permanent bladder catheter, immunosuppressed, oncological and diabetic were also considered. Trials with a mixed population were included in the systematic review if they provided separate data for aPN population.
Types of outcome measures and follow-up assessment
Our primary outcome was Clinical success, defined as a number of subjects with resolution of symptoms (fever, costovertebral pain and dysuria), signs (biohumoral tests as leukocyte count, inflammatory markers and renal function) and microbiologic criteria (pyuria and a positive culture with ≥104 cfu per milliliter of urine) at the end of treatment.
Secondary outcomes were: Microbiological success, at the end of treatment defined as yielding sterile urine cultures or positive cultures with <103 cfu/mL of urine at the end of therapy;24 Clinical relapse at 4-6 weeks after the end of treatment, defined as the reappearance of signs and symptoms.
Microbiological relapse, defined as the reappearance of the original strain in a urine culture at 4-6 weeks after the end of treatment, based on species identification and serotyping results; Microbiological recurrence or reinfection, defined as the appearance of another bacteriologic strain in a urine culture at 4-6 weeks after the end of treatment, based on species identification and serotyping results;24 Renal impairment, defined as a glomerular fraction rate <30 mL/min or creatinine increase >50% from baseline level; Intensive Care Unit requirement; Mortality for any cause occurring during the study; Readmission for the same cause until the end of the follow-up period; Patients with any Adverse event defined as any untoward occurrence of any unfavorable and unintended clinically relevant medical sign, symptom or any disease temporally associated with the study, which did not necessarily have a causal relationship with the study procedure and patient withdrawals due to serious adverse events during the study.
We searched the following electronic databases: MEDLINE (PubMed) (January 1966 to June 2016), EMBASE (Elsevier, EMBASE.com) (January 1974 to June 2016) and CENTRAL (up June 2016) using the search strategy outlined in the Appendix.
We checked the reference lists of all studies included and of any systematic reviews we have identified during the search process.
We also searched the following clinical trial registries to identify ongoing trials: ClinicalTrials.gov (http://clinicaltrials.gov/) and Current Controlled Trials (http://www.controlledtrials.com/); we included studies written in English, French, Spanish and Italian.
Selection of studies and data extraction
Three pairs of authors (BF, ATM, PS, TL, TD, TM) screened titles and abstracts obtained by the search strategy. Then they independently assessed the full text of potentially relevant studies for inclusion. Any disagreement was solved by discussion with a further author (SM).
We adopted a standardized data collection form to extract the following information: number and characteristics of participants, setting, type of experimental and control intervention (antibiotic agent, length of treatment, any further treatment), length of followup, types of outcomes, country of origin, funding and conflict of interest, clinical success, bacteriologic efficacy, relapses, recurrences, any adverse events and/or patient withdrawals due to adverse events. We contacted authors if the reported data were insufficient or unclear.
Risk of bias (quality) assessment
Two authors (TL, TD) independently assessed the risk of bias of the included studies. Any disagreement was resolved by discussion with a further author (SM). We assessed the risk bias using the Cochrane criteria.25 We considered the following specific domains: sequence generation and allocation concealment (selection bias), blinding of participants and providers (performance bias), blinding of outcome assessors (detection bias), incomplete outcome data (attrition bias) and selective outcome reporting (reporting bias). Each domain was judged as high, low or unclear risk of bias. To incorporate our assessment of risk of bias in the review process, we first plotted the intervention effects estimates, stratified by risk of bias for allocation concealment (selection bias), blinding of outcome assessors (detection bias) and attrition bias. If differences in the results were present among studies at different risks of bias, we performed sensitivity analysis by excluding studies with high risk of bias from the analysis.
We analyzed dichotomous outcomes by calculating the risk ratio (RR) for each trial with the uncertainty in each result being expressed with a 95% confidence interval (CI).
We combined the outcomes from the individual trials through meta-analysis where possible using a random- effects model26 because a certain degree of heterogeneity was expected among trials. We analyzed heterogeneity by means of the I2 statistic and the Chi2 test. The cut-off points were I2 value of more than 50% and a P value for the Chi2 test of less than 0.1.27 If a very high heterogeneity was found (i.e., greater than 90%) no meta-analyses were performed and results were described narratively.
We planned to investigate the publication bias using visual inspection for asymmetry of funnel plots if there were at least 10 studies included in the metaanalysis. We assessed the overall quality of the evidence for the primary outcome using the GRADE methodology.28,29 The GRADE approach uses five dimensions (risk of bias, consistency of effect, imprecision, indirectness and publication bias) to assess the quality of the body of evidence. The evidence is downgraded from high quality by one level if serious, or by two levels if very serious limitations are found for each of the five dimensions. We developed a summary of findings table presenting the quality of the evidence, reasons for limitation and main findings for the primary outcome in simple tabular format.
A total of 725 articles were retrieved from PubMed, 608 from EMBASE, and 185 from CENTRAL. After 175 duplicates were removed we were left 1343 articles and following a preliminary screening by reading the titles and abstracts we removed other 1331 articles. Out of 12 articles assessed as potentially eligible, 6 were excluded for the following reasons: two studies were not RCT,30,31 one study considered patients with recurrent urinary-tract infections and not acute pyelonephritis,32 two studies did not compare two regimens with the same antibiotic33,34 and one study used two different routes of administration; 35 two further studies were considered awaiting assessment: one study was published in Chinese language36 and one was a protocol in recruiting phase.37 We finally included 4 RCTs (Figure 1).
Characteristics of included studies
We included 4RCTs23,38-40 with 439 participants conducted between 1987 and 2012. Three studies were conducted in European countries23,38,40 and one in the USA.40 All studies included patients ≥15 years with aPN. Three trials38-40 included only patients with aPN, while the remaining trial23 included patients with complicated urinary tract infections; we extracted data only for the subgroups (n=34) with aPN. The majority of patients were women (87%, range 66.7 to 100%); mean age was 50.75 years, range 16-94.
One study23 enrolled only patients who were hospitalized, two studies38,39 included both inpatients and outpatients and the last one40 involved only outpatients. One study40 analyzed two subgroups of population treated with different antibiotic and, in this review, we classified it as Stamm A and Stamm B.
The duration of treatment for short antibiotic therapy ranged from 4 to 14 days. The duration for long therapy was ≥7 days (ranging from 7 to 42 days), but at least 2 days longer than the corresponding shortcourse treatment. The timing of outcomes assessment varied among studies and ranged from end of treatment up to 180 days.
Three trials were funded by pharmaceutics company: one from Hoffmann La Roche,23 one from Leo Pharmaceutical38 and one from Bayer.39 One trial was partially funded by National Institutes of Health and by a grant from Burroughs Wellcome Foundation40 (Table 1).
Risk of bias assessment
Three studies23,38,40 reported an appropriate method of randomization. One study39 did not report sufficient information to permit judgment of low or high risk. Only one study39 reported an adequate procedure for allocation concealment, the others23,38,40 were at unclear risk of bias not reporting enough information to make a judgment. The main characteristics of the participants in the two groups were similar in all selected studies.
Two studies were double-blind (blinding of providers and patients)38,39 while the other two23,40 were open label and judged at high risk of performance bias. Two studies reported that the outcome assessors were blinded,23,39 one was open label and judged at high risk of detection bias40 while the fourth38 did not report sufficient information. All the studies were judged at high risk of attrition bias and at low risk of selective outcome reporting. Risk of bias assessment is presented in Figure 2. The risk of publication bias has not been evaluated because less than ten studies were included.
Effects of interventions
Clinical success at the end of treatment
We found no significant differences in clinical success between short and long antibiotic therapy at the end of treatment (RR 1.01, 95% CI 0.96-1.07, 4 studies, 250 participants) with a moderate quality of evidence (Figure 3A and Table 2).
Microbiological success at the end of treatment
We found no significant differences in the microbiological success at the end of antibiotic treatment between short or long-term therapy (RR 0.99, 95% CI 0.92-1.07, 2 studies, 190 participants) with low quality of evidence (Figure 3B and Table 2).
Clinical relapse at 4-6 weeks
We found no significant differences in clinical relapse at 4-6 weeks after the end of treatment between short and long-term therapy (RR 1.20, 95% CI 0.43-3.30, 2 studies, 218 participants) with very low quality of evidence (Figure 4A and Table 2).
Microbiological relapse at 4-6 week follow-up
The long-term therapy seemed to prevent recurrences of the same biological germ up to 4-6 weeks after the end of treatment compared to short-term therapy (RR 2.39, 95% CI 1.19-4.83, 2 studies, 95 participants) with very low quality of evidence (Figure 4B and Table 2).
Microbiological reinfection at 4-6 week follow-up
We found no significant differences between longand short-term therapy in the prevention of re-infection by different germs from the one originally isolated (RR 2.40, 95% CI 0.68-8.49, 2 studies, 95 participants) with very low quality of evidence (Figure 4C and Table 2).
Number of patients with adverse effects
The number of patients with at least one adverse effect from antibiotic therapy seemed to be higher in the long-term therapy compared to that of short-term but the difference was not statistically significant (RR 0.63, 95% CI 0.39-1.02, 4 studies, 375 participants) (Figure 5 and Table 2). None of the included studies reported results for the outcomes Renal impairment, Intensive Care Unit requirement, Mortality for any cause, Hospital readmission for the same cause.
Summary of main results
This systematic review offers an update of results of all randomized trials that compare short- versus long-term antimicrobic therapy for acute pyelonephritis in adults. It is based on an extensive research, including incomplete data by attempting to contact all authors.
We found moderate quality evidence that shortand long-term treatments probably did not differ in terms of clinical success as well as in microbiological success at the end of treatment.
Conversely at 4-6 weeks after the end of treatment we found with a very low quality of evidence that microbiological relapse of the same biological germ may be more frequent with short-term therapy, but there may be no difference in the frequency of clinical relapse and re-infection by different germs.
Finally, we found that the incidence of any adverse effect seemed to be lower with the short-term therapy, though the results are not statistically significant and the quality of evidence is moderate. It was not possible to interpret the finding related to Mortality, Renal impairment, Intensive Care Unit requirement, Hospital readmission for the same cause since no data were reported in any of the included trials.
Strengths, limitations and uncertainties
The strengths of this review include the adherence to accepted standards for the conduct of systematic reviews and the use of extensive literature searches to identify relevant data.41-43Moreover we included studies written in English, French, Spanish and Italian, so reducing the risk of language bias.
The major limitation of this review is the small number of participants on which we could base our conclusions: we found only four randomized studies fulfilling our inclusion criteria and most of them had had small sample size. Since aPN could be considered almost a rare condition because presented few cases every 10,000 habitants44 multicenter RCTs should be encouraged in order to increase the precision and the strength of the recommendation for the evidence of findings. Further studies should address this limitation by providing a larger sample size and improve the methodological quality of the studies, for example, opting for double-blind studies.
Furthermore, the retrieved evidence comes to studies - except Sandberg39 - that were >10 years old and some of the antibiotics used in the included studies (i.e., fleroxacin) are not available in a number of countries worldwide. The included studies did not provide data on mortality and only one40 reported information about the degree of severity of aPN.
In addition, we detected some deficiencies in the methodological quality of some of the included studies in this review. Only two trials were double blind and all were at high risk of attrition bias. Overall, the quality of evidence was judged from moderate to very low due to serious risk of bias and imprecision.
We could not assess the risk of publication bias by visual inspection for asymmetry of funnel plots because only four studies were included in meta-analysis. So, the risk of publication bias cannot be excluded though we performed a very sensitive search looking also for unpublished studies and studies published in languages other than English.
Agreements and disagreements with other studies or reviews
We obtained the same findings of previous metaanalyses. 24,45 The review published in 2008 by Kyriakidou et al.24 included 4 studies with 283 patients (one double-blinded RCT and three open-label RCTs) comparing a short (defined as 7-14 days) versus a long arm (defined as 14-42 days) for the same antibiotic treatment. Authors found no significant difference between the short- and long-course regimens efficacy at the end of treatment or at follow-up. Another systematic review published in 2013 by Eliakim-Raz et al.45 investigated the duration of treatment for acute pyelonephritis including 2515 patients, 1239 treated for ≤7 days versus 1276 treated for >7 days. Five trials made comparisons using the same antibiotic, and three made comparisons using different antibiotics. They found out that seven days of treatment for acute pyelonephritis is equivalent to longer treatment in terms of clinical failure and microbiological failure, including bacteremic patients. In patients with urogenital abnormalities, the evidence, although weak, suggests that longer treatment is required.
Published Guidelines17 do not address the specific question of duration of treatment in acute pyelonephritis patients. However the following types of good clinical practice are performed: 7 days for fluoroquinolones, 10-14 days for β-lactams and 14 days for TMP-SMX.17-19 Since in our study we found no difference in terms of clinical and microbiological success or tolerability for short (from 4 to 14 days) and long course antibiotic therapy (ranging from 7 to 42 days), we can confirm the actual antibiotic recommendations even if the grade of recommendation is week for low quality of evidence due the paucity of patients and risk of bias of trials.
Finally, the reduction in patient exposure to antibiotics may also limit the increasing rates of antimicrobial drug resistance,46 decreasing costs,47 and may improve patient adherence and tolerability.
This review suggests that short-term treatment for acute pyelonephritis may be equivalent to longer-term treatment in terms of clinical success and microbiological success at the end of treatment adverse event may be slightly more frequent with the long therapy. The long-term therapy may be more effective for the prevention of microbiological relapse of the same biological germ up to 4-6 weeks after the end of treatment compared to short-term therapy.
Further high-quality research, through the launch of multicenter RCTs, is needed to confirm the clinical and microbiological equivalence of short and longerterm antibiotic treatment for aPN, including in patients with serious prognostic categories.