Symptomatic therapy of uncomplicated lower urinary tract blind … · 2017-11-10 · Confidential: For Review Only page 1 / 26 1 Symptomatic therapy of uncomplicated lower urinary

Confidential: For Review O

nly

Symptomatic therapy of uncomplicated lower urinary tract

infections in the ambulatory setting. A randomized, double blind Trial.

Journal: BMJ

Manuscript ID BMJ.2017.039317

Article Type: Research

BMJ Journal: BMJ

Date Submitted by the Author: 11-May-2017

Complete List of Authors: Kronenberg, Andreas; University of Bern, Institute for Infectious Diseases; University of Bern, Department of Infectious Diseases Bütikofer, Lukas; University of Bern, CTU Bern; Universitat Bern Institut fur Sozial- und Praventivmedizin Odutayo, Ayodele; University of Toronto, Faculty of Medicine; University of Oxford, Centre for Statistics in Medicine Mühlemann, Kathrin; University of Bern, Department of Infectious Diseases Da Costa, Bruno; Universität Bern, Berner Institut für Hausarztmedizin (BIHAM) Battaglia, Markus; Medix General Practice Netzwork Meli, Damian; Medix General Practice Network Frey, Peter; University of Bern, Institute of Primary Health Care (BIHAM)

Limacher, Andreas; University of Bern Reichenbach, Stephan; Universitat Bern Institut fur Sozial- und Praventivmedizin; University of Bern, Department of Rheumatology, Immunology and Allergology Juni, Peter; University of Toronto, St. Michael's Hospital, Applied Health Research Centre (AHRC)

Keywords: cystitis, urinary tract infection, antibiotic therapy, symptomatic, ambulatory, randomized, outpatient, anti-inflammatory drugs

https://mc.manuscriptcentral.com/bmj

BMJ


nly

page 1 / 26

Symptomatic therapy of uncomplicated lower urinary tract 1

infections in the ambulatory setting. A randomized, double blind 2

trial 3

Andreas Kronenberg,1,2,3 Lukas Bütikofer,4,5 Ayodele Odutayo,6 Kathrin 4

Mühlemann,1,2 � Bruno R. da Costa,7 Markus Battaglia,3 Damian N. Meli,3,7 Peter 5

Frey,7 Andreas Limacher, 4,5 Stephan Reichenbach,5,8 Peter Jüni 6,7 6

1 Institute for Infectious Diseases, University of Bern, Bern, Switzerland 7

2 Department of Infectious Diseases, Bern University Hospital, University of Bern, Bern, Switzerland 8

3 Medix General Practice Network, Bern, Switzerland 9

4 CTU Bern, University of Bern, Bern, Switzerland 10

5 Institute of Social and Preventive Medicine (ISPM), University of Bern, Bern, Switzerland 11

6Applied Health Research Centre (AHRC), Li Ka Shing Knowledge Institute of St. Michael’s Hospital, 12

and Department of Medicine, University of Toronto, Canada 13

7 Institute of Primary Health Care (BIHAM), University of Bern, Bern, Switzerland 14

8 Department of Rheumatology, Immunology and Allergology, Inselspital, Bern University Hospital, 15

University of Bern, Bern, Switzerland 16

17

Corresponding address: Andreas Kronenberg, MD 18

Institute for Infectious Diseases 19

University of Bern 20

Friedbühlstrasse 51 21

3010 Bern 22

Tel: +41 (0)31 632 32 65 / Fax: +41 (0)31 632 49 66 23

e-mail: [email protected] 24

25

26

Running title Symptomatic therapy of cystitis 27

Keywords cystitis, urinary tract infections, antibiotic therapy, anti-inflammatory drugs, 28

symptomatic, ambulatory, outpatient, randomized, double-blind 29

Word count abstract 378 30

Word count text 4278 31

Registration: www.clinicaltrials.gov (NCT01039545) 32

33

Page 1 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 2 / 26

ABSTRACT 1

Background: Antibiotic resistance is increasing globally and largely stems from increasing 2

antibiotic use. It has been suggested, that symptomatic therapy with non-steroidal anti-3

inflammatory drugs (NSAIDs) might be non-inferior to antibiotics in treatment of 4

uncomplicated lower urinary tract infection (UTI) in women, thus offering an opportunity to 5

reduce antibiotic use in ambulatory care. 6

Methods: In this randomized, double-blind, non-inferiority trial among 17 general practices in 7

Switzerland, women with uncomplicated lower UTI were randomly assigned 1:1 to 8

symptomatic therapy with diclofenac or antibiotic treatment with norfloxacin. The 9

randomization sequence was computer-generated, stratified by practice, blocked and 10

concealed using sealed, sequentially numbered drug containers. Primary and secondary 11

outcomes were symptom resolution up to day 3 and antibiotic use up to day 30, respectively. 12

Analysis was by intention-to-treat. We also performed a meta-analysis combining the results 13

of the current trial with previously published randomized controlled trials comparing NSAIDs 14

and antibiotic treatment in women with UTIs. 15

Findings: We randomly assigned 253 women to NSAIDs (133) and antibiotics (120). 16

Thereof 72 (54%) and 96 women (80%) experienced symptom resolution at day 3 in the 17

NSAID and antibiotic groups, respectively (risk difference 27%, 95%-CI 15% to 38%, p for 18

non-inferiority 0.98, p for superiority <0.0001). The median time until resolution of symptoms 19

was 4 days in the diclofenac group and 2 days in the norfloxacin group. A total of 82 (62%) 20

and 118 women (98%) used antibiotics up to day 30, respectively (risk difference 37%, 95%-21

CI 28 to 46%, p for superiority <0.0001). Six women in the NSAID group (5%) but none in the 22

antibiotic group received clinical diagnosis of a pyelonephritis (p=0.031). CRP levels > 10 23

mg/L at baseline were observed in 21 women in the NSAID group without pyelonephritis 24

(17%) and in 3 women with pyelonephritis (50%), which resulted in a positive likelihood ratio 25

of 3.02 (CI 1.07 to 5.98, p=0.037). Finally, our results were consistent with two previously 26

published clinical trials when combined in a meta-analysis. 27

Page 2 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 3 / 26

Interpretation: Diclofenac is inferior to norfloxacin in terms of symptom relief, but superior in 1

terms of reducing antibiotic use. Deferring antibiotic therapy in women without CRP elevation 2

and short symptom duration could be justifiable, but should be tested in a randomized trial. 3

Funding: Swiss National Foundation, Swiss Academy of Medical Sciences, SwissLife and 4

Else Kroener-Fresenius Foundation. 5

6

Page 3 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 4 / 26

INTRODUCTION 1

Antimicrobial stewardship aims at reducing antibiotic resistance by optimising or 2

decreasing antibiotic use,1 which includes the prevention of antibiotic treatment in cases of 3

viral infections, the tailored prescription of narrow-spectrum antibiotics, shortening the course 4

of therapy, and deferring therapy for low risk bacterial infections. Urinary tract infection (UTI) 5

is one of the most common bacterial infections in adults, affecting considerably more women 6

than men.2 Approximately half of women have at least one UTI in their lifetime, and 20% to 7

30% have two or more.3 Antibiotic prescriptions for UTI accounts for 10-20% of all antibiotic 8

prescriptions in ambulatory care and is second in number only to respiratory tract infections,4, 9

5. Reducing antibiotic prescriptions for UTI could potentially decrease the risk of antimicrobial 10

resistance. Therefore, the benefit of antibiotic treatment needs to be weigthed against the 11

potential for adverse effects, both at the level of the individual (adverse drug reactions) and 12

population (as a driver of antibiotic resistance). 13

Symptoms of UTI may arise from local increases in pro-inflammatory factors like 14

prostaglandins and non-steroidal anti-inflammatory drugs (NSAIDs) may be useful in 15

alleviating symptoms.6-8 A small randomized pilot trial, which compared the NSAID ibuprofen 16

with the antibiotic ciprofloxacin in 80 women with uncomplicated lower UTI, concluded that 17

symptomatic therapy with NSAIDs may be non-inferior to antibiotics, but suggested 18

confirmation in a larger trial.9 Two adequately powered randomized double-blind trials were 19

therefore initiated simultaneously in February 2012 in Germany and Switzerland. Results of 20

the German trial, which compared ibuprofen with fosfomycin in 494 women, were recently 21

published.10 Here, we report results of the Swiss trial, which compared diclofenac with 22

norfloxacin in 253 women with uncomplicated lower UTI. To put our results into context, we 23

also performed a meta-analysis of all available trials comparing NSAIDs with antibiotics in 24

women with lower UTI. 25

26

Page 4 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 5 / 26

METHODS 1

Study design 2

This randomized controlled patient and assessor blind trial was conducted in 17 general 3

practices in Switzerland. The funding bodies were not involved in the design or the conduct 4

of the trial, nor in the writing of the manuscript or the decision to submit the manuscript for 5

publication. The first and senior author wrote the manuscript, had final responsibility for the 6

decision to submit the manuscript for publication and vouch for the accuracy of the data and 7

analyses and for the fidelity of the trial to the protocol. The funding bodies only had access to 8

the data after finalization of the statistical analysis plan and completion of all analyses. The 9

study was conducted in accordance with the Declaration of Helsinki and was approved by 10

the local research ethics committee and the Swiss Agency for Therapeutic Products 11

Swissmedic. The trial was registered before start of patient recruitment at clinicaltrials.gov 12

(NCT01039545). All women provided written informed consent. 13

14 15

Patients, Randomization and Treatment 16

Women aged 18 to 70, who visited their general practitioners (GPs) because of one 17

or more symptoms or signs typical of acute lower UTI (dysuria, frequency, macrohaematuria, 18

cloudy or smelly urine) or self-diagnosed cystitis were eligible if their urine dipstick was 19

positive for nitrite and/or leucocytes. We excluded pregnant women and women with clinical 20

signs of upper UTI such as fever (axillary body temperature >38oC), costovertebral pain or 21

tenderness, rigors, nausea or vomiting. We also excluded women with known or suspected 22

anatomical or functional abnormality of the urinary tract and comorbidities such as diabetes 23

mellitus, active gastric or duodenal ulcer disease or gastrointestinal bleeding, inflammatory 24

bowel disease, severe liver dysfunction (liver cirrhosis and ascites), coagulopathy (including 25

therapy with coumarine derivates), renal insufficiency grade III or higher (calculated GFR <60 26

ml/min), known congestive heart failure (NYHA III or higher), psychiatric illness or dementia, 27

inability to to communicate in German or French language and any other serious comorbidity 28

Page 5 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 6 / 26

as judged by the treating physician. In addition we excluded women with documented 1

immunosuppression (e.g. prednisone equivalent >10mg per day for >14 days, chemotherapy, 2

radiotherapy, immunmodulators, HIV infection, neutropenia) or hypersensitivity to one of the 3

study medications or history of asthma, urticaria or hypersensitivity-like reactions after 4

consumption of salicylic acid or other non-steroidal anti-inflammatory drugs, as well as 5

women with vaginal symptoms (discharge, irritation), bladder catheter in situ or during the 6

past 30 days, recurrent urinary tract infection (more than 3 infections during the last 12 7

months), antibiotic treatment during the last 4 weeks or duration of UTI symptoms for more 8

than 7 days before physician’s visit. 9

Women were randomly allocated in a 1:1 ratio to NSAID or antibiotic therapy. The 10

randomization sequence was computer-generated, stratified by practice and blocked with 11

randomly varying block sizes of 4 and 6. Allocation was concealed with sealed, sequentially 12

numbered opaque drug containers of identical appearance that contained opaque hard 13

gelatine capsules of identical size and colour. Women allocated to NSAIDs received 14

capsules containing 75 mg diclofenac retard for three days (Olfen-75 duo release®, Mepha 15

Pharma, Basel, Switzerland) and women allocated to antibiotics received capsules 16

containing 400 mg norfloxacin for three days (Norfloxacin-Teva® 400mg, Teva Pharma, Tel 17

Aviv, Israel; Norflocin-Mepha® 400mg Lactab, Mepha Pharma, Basel, Switzerland from 18

October 2013 onwards due to delivery restrictions). We chose norfloxacin because of the 19

high susceptibility rates in Switzerland and diclofenac because of its identical frequency of 20

administration which facilitated patient blinding. Women started therapy immediately after 21

randomization on day 0 and were advised to take 2 capsules per day: one in the morning 22

and one in the evening. All women were given a single open-label package of fosfomycin 23

(Monuril® 3g, Zambon, Cadempino, Switzerland) to be taken as a rescue antibiotic after 24

completion of the study medication on day 3 at their discretion, if symptoms persisted. 25

26

Patient involvement 27

Page 6 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 7 / 26

No patients were involved in setting the research question or the outcome measures, nor 1

were they involved in developing plans for recruitment, design, or implementation of the 2

study. No patients were asked to advise on interpretation or writing up of results. We plan to 3

disseminate the results of the research to all the scientific community, including study 4

participants. 5

6

Procedures and outcomes 7

The pre-specified primary outcome was symptom resolution up to day 3 (12 hours 8

after intake of the last study medication). In the absence of antimicrobial resistance, the 9

mean duration of symptoms in women with uncomplicated UTIs treated with antibiotics is 3 10

days11. Women rated the severity of five UTI symptoms (dysuria; frequency; urgency; 11

abdominal pain when passing urine; pain or tenderness in the lower back or loin) daily from 12

days 0 (randomization) to 10 by self-report diary and on day 30 by telephone interview on a 13

Likert scale from 0 to 6, with their composite score ranging from 0 to 30; symptom resolution 14

was defined as ≤ 2 points ("slight problems”) on all 5 components.12-14 Complete absence of 15

symptoms was defined as 0 points on all components. The pre-specified principal secondary 16

outcome was the use of any antibiotic (including norfloxacin and fosfomycin as trial 17

medications) up to day 30. The remaining pre-specified outcomes are reported in the 18

Supplementary Appendix. Side effects, quality of life, working days lost and use of the rescue 19

antibiotic were assessed on day 3, general satisfaction on day 10. Serious adverse events, 20

adverse events ≥grade 3 severity, any additional unplanned medical visits, including 21

telephone contacts, and co-medications were assessed by telephone interviews of the 22

treating physician on day 10 and of an independent, blinded interviewer on day 30. The 23

clinical diagnosis of pyelonephritis required the occurrence of loin pain and fever, leading to 24

an unplanned ambulatory visit. Mid-stream urinary samples obtained on days 0 and 10 were 25

processed according to standard laboratory procedures, using a cut-off of ≥103 colony 26

forming units per ml for a urinary culture to be considered positive;15 mixed flora with no 27

predominant microorganism, Lactobacilli or Streptococcus viridans group were considered 28

Page 7 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 8 / 26

negative. Details of baseline assessments, diary and scores used are described in the 1

Supplemental Appendix. 2

3

Systematic Review 4

We conducted a systematic search of MEDLINE (inception to April 9th 2017) using a 5

combination MeSH and keyword searching for the following terms: “Anti-bacterial Agents”, 6

“Antibiotics”, “Non-Steroidal Anti-inflammatory Agents”, “Urinary Tract Infections”, and 7

“Cystitis”. We included randomized clinical trials which compared NSAIDs and antibiotics for 8

the treatment of UTIs. No language restrictions were applied. We extracted information on 9

event rates in the NSAID and antibiotic treatment groups for the following outcomes: 10

symptom resolution at day 3 or 4, symptom resolution at 7 days, recurrent UTI, 11

pyelonephritis and use of antibiotics. 12

13

14

Statistical methods 15

We originally planned to recruit 400 women, but recruitment was slow and financial 16

constraints led us to decide in June 2014 to stop patient recruitment by December 2014, 17

when an expected 260 women would be included. The decision was made without inspecting 18

the data and after repeating the power analysis based on a closed-form normal 19

approximation test of proportions, 16 which was less conservative than the simulation-based 20

approach originally used. With the original assumption of 70% of women reaching symptom 21

resolution up to day 3 in both groups and the original, pre-specified non-inferiority margin of 22

15% on a risk difference scale, the projected sample size of 260 women would yield a power 23

of 84% to detect non-inferiority at a one-sided type I error of 5%. 24

The primary outcome was evaluated using a risk difference with a corresponding two-sided 25

95% confidence interval, a one-sided normal approximation test for non-inferiority and a two-26

sided Chi-squared test for superiority. Secondary outcomes were compared using 27

Page 8 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 9 / 26

conventional two-sided p-values for superiority and corresponding two-sided 95% confidence 1

intervals. We used risk differences with Chi-squared tests for binary data, Poisson regression 2

with robust standard errors for counts, and linear regression with robust standard errors for 3

continuous data. Kaplan-Meier curves accompanied by hazard ratios from Cox models were 4

used to analyse time to symptom resolution and time to antibiotic use. All women were 5

included in the analysis in the groups they were originally allocated to (intention-to-treat 6

analysis), missing values were accounted for by multiple imputation (see Supplementary 7

Appendix and Supplementary Table 1). We performed pre-specified subgroup analyses of 8

the primary outcome accompanied by Mantel-Haenszel tests for interaction by age (< 45 vs ≥ 9

45 years), symptom severity at baseline (≤ 20 vs > 20), symptom duration (≤ 3 vs > 3 days), 10

and presence of a positive urinary culture at baseline; post-hoc subgroup analyses were 11

performed by urine leucocytes (≤ 2+ vs >2+) and the presence of norfloxacin-resistant 12

Enterbacteriaceae. In per-protocol analyses, we excluded women with protocol deviations 13

who were defined as women with no documented intake of at least one dose of study 14

medication, cross-overs, or women who used rescue antibiotics before day 3. All analyses 15

were pre-specified in a statistical analysis plan before recruitment end and inspection of the 16

data. In post-hoc analyses of women allocated to NSAIDs, we compared baseline 17

characteristics and outcomes between those who used antibiotics until day 30 and those 18

never on antibiotics, determined the time between symptom onset and diagnosis in those 19

clinically diagnosed with pyelonephritis, and explored whether blood or urine findings at 20

baseline were associated with pyelonephritis with clinically relevant positive or negative 21

likelihood ratios above 5 or below 0.2, respectively, which could be used to rule in or rule out 22

future clinical diagnosis of pyelonephritis. 23

In view of the expected low number of trials and the considerable difference in sample size 24

between the initially available pilot trial and subsequent trials, we refrained from performing 25

conventional frequentist random-effects meta-analysis because these methods may yield 26

either overly optimistic or overly conservative results.17-19 Therefore, trial specific risk 27

differences were combined with a Mantel-Haenszel fixed-effect model and a Bayesian 28

Page 9 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 10 / 26

random-effects model. We chose the risk difference as treatment effect estimate as odds-1

ratios were not directly interpretable because some control group event rates were above 2

20%, and risk ratios were heterogeneous because of the variation in control group event 3

rates between trials. The specifications of the Bayesian random-effects model are described 4

in the web-appendix. We conducted two analyses. The first combined all trials and the 5

second was restricted to large trials with more than 100 patients per group; trials which would 6

have more than 50% power to detect a 15% difference in the primary outcome of incomplete 7

or no symptom resolution at day 3 or 4. All analyses were done in 8

Stata Release 14 (StataCorp. 2014. Stata Statistical Software: Release 14. College Station, 9

TX: StataCorp LP), RStudio version 1.0.143 (RStudio: Integrated Development for R. 10

RStudio, Inc., Boston, MA URL http://www.rstudio.com/), and WinBUGS version 1.4.3. 11

12

13

Page 10 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 11 / 26

RESULTS 1

Between Feb 7, 2012 and Dec 3, 2014, 253 women were included in the trial. Thirty-2

six patients were recruited in seven single practices, 108 in nine group practices and 109 in 3

one big medical centre with 17 physicians in the centre of Bern, Switzerland. One-hundred 4

and thirty-three patients were randomly allocated to NSAIDs and 120 to antibiotics. A total of 5

125 women (94%) received treatment as allocated in the NSAID group, and 118 women 6

(98%) in the antibiotic group. 119 (89%) and 112 women (93%) had complete follow-up until 7

day 30 (Figure 1). Groups were similar (Table 1), with a mean age of 36.8 years (SD 14.1), a 8

mean duration of symptoms of 3.4 days (SD 2.6) and a mean composite symptom score of 9

13.7 points out of 30 (SD 3.8), and urgency, frequency and dysuria as the most prevalent 10

symptoms. Thirty-five urine samples were nitrite-positive (14%) and 191 had >2+ leucocytes 11

(75%). For 185 urine samples, we found positive urinary cultures (73%), which resulted in a 12

total of 193 isolates, of which 187 isolates (97%) had documented norfloxacin susceptibility. 13

173 isolates contained Enterobacteriaceae, 160 of which were tested for fosfomycin 14

susceptibility and 158 were found susceptible (99%). 15

Table 1. Baseline characteristics 16

NSAIDs (N=133) Antibiotics (N=120)

Age - yr 37.8 (14.2) 35.6 (14.0)

Age <45 yr 94 (71%) 89 (74%)

Symptom duration - days since UTI onset 3.6 (3.1) 3.2 (2.0)

Symptom duration ≤3 days 80 (60%) 83 (69%)

Number of UTIs in the last 12 months 0.6 (1.1) 0.6 (0.9)

Baseline UTI symptoms - score 0 to 6

Dysuria 3.3 (1.3) 3.3 (1.2)

Urgency 3.7 (1.0) 3.6 (0.9)

Night frequency 2.6 (1.5) 2.6 (1.4)

Day frequency 3.5 (0.9) 3.5 (0.9)

Lower abdominal pain while urinating 2.6 (1.5) 2.6 (1.6)

Back or loin pain 1.1 (1.4) 1.2 (1.5)

Total symptom score 13.7 (3.9) 13.8 (3.8)

Symptom score ≤20 127 (95%) 115 (96%)

Blood tests

CRP - mg/l 6.7 (10.5) 8.5 (13.7)

CRP >10 mg/ml 24 (18%) 29 (24%)

Leucocytes - 1000/µl 8.5 (2.4) 8.7 (2.2)

Page 11 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 12 / 26

Urinary dipstick

Nitrites - positive 17 (13%) 18 (15%)

Erythrocytes - 1 to 4+ (median, IQR) 3+ (2+ to 4+) 3+ (2+ to 4+)

Leucocytes - 1 to 4+ (median, IQR) 3+ (3+ to 4+) 3+ (3+ to 4+)

Leucocytes >2+ 101 (76%) 90 (75%)

Urinary culture†

Negative 36 (27%) 31 (26%)

Positive‡ 96 (72%) 89 (74%)

Escherichia coli 82 (62%) 75 (63%)

Other Enterobacteriaceae 10 (8%) 6 (5%)

Staphylococcus saprophyticus 4 (3%) 7 (6%)

Enterococcus faecalis* 1 (1%) 6 (5%)

Beta-hemolytic Streptococcus group B 1 (1%) 1 (1%)

Susceptibility to norfloxacin 92 (69%) 87 (73%)

Enterobacteriaceae 88 (66%) 79 (66%)

Susceptibility to fosfomycin§ 84 (63%) 71 (59%)

Continuous variables are presented as mean (sd), binary variables as no. of patients (%). 1 † Urinary culture missing in 1 patients in NSAID group. 2 ‡ Do not sum up because of 2 double mixed infections in the NSAID group and 4 double and 1 triple mixed infection in the 3

antibiotic group. 4 $ Enterobacteriaceae only. Two isolates from the NSAID group (1 Proteus mirabilis, 1 Enterobacter cloacae) were not 5

susceptible to fosfomycin 6 7

Clinical outcomes are presented in Table 2. The primary outcome, resolution of 8

symptoms at day 3, was observed in 72 (54%) and 96 (80%) women in the NSAID and 9

antibiotic groups, respectively (risk difference 27%, 95%-CI 15 to 38%, one-sided p for non-10

inferiority=0.98, two-sided p for superiority in favour of antibiotic group <0.0001). The 11

principal secondary outcome, use of any antibiotic up to day 30, was observed in 82 (62%) 12

and 118 (98%) women, respectively (risk difference -37%, 95%-CI -46 to -28%, p for 13

superiority in favour of NSAID group <0.0001). Among the 82 women in the NSAID group 14

who used antibiotics, 58 (71%) decided to take antibiotics during the first three days and 55 15

of the 58 (95%) took the rescue antibiotic fosfomycin. 16

17

Table 2. Primary and secondary outcomes. Positive differences indicate an advantage of 18 antibiotics, negative differences an advantage of NSAIDs throughout. 19

Outcome NSAIDs (N=133) Antibiotics (N=120) Risk or mean

difference (95% CI)

P value

no. of patients (%) or mean (sd)

Resolution of symptoms

Day 3 (primary outcome) 72 (54%) 96 (80%) 27% (15 to 38%) <0.0001

Day 7 111 (83%) 115 (96%) 12% (4 to 19%) 0.003

Day 10 126 (95%) 116 (97%) 2% (-3 to 7%) 0.45

Day 30 127 (95%) 111 (93%) -3% (-9 to 3%) 0.32

Page 12 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 13 / 26

Complete absence of symptoms

Day 3 10 (8%) 20 (17%) 9% (0 to 17%) 0.038

Day 7 44 (33%) 65 (54%) 21% (9 to 34%) 0.001

Day 10 70 (53%) 77 (64%) 12% (-1 to 24%) 0.07

Day 30 101 (76%) 99 (83%) 6% (-4 to 17%) 0.22

Change of symptom score

Day 3 -7.3 (4.7) -10.3 (4.1) 3.0 (1.9 to 4.1) <0.0001

Day 7 -11.0 (4.8) -12.6 (4.2) 1.6 (0.5 to 2.7) 0.005

Day 10 -12.2 (4.3) -12.9 (4.1) 0.7 (-0.4 to 1.7) 0.20

Day 30 -13.0 (4.4) -13.1 (4.3) 0.1 (-1.0 to 1.1) 0.88

Use of any antibiotic

Up to day 3 58 (44%) 116 (97%) -54% (-63 to -44%) <0.0001

Up to day 30 (principal secondary outcome) 82 (62%) 118 (98%) -37% (-46 to -28%) <0.0001

Use of rescue antibiotic

Up to day 3 55 (41%) 9 (8%) 34% (24 to 43%) <0.0001

Up to day 30 73 (55%) 18 (15%) 40% (29 to 51%) <0.0001

Negative urinary culture at day 10 96 (72%) 112 (93%) 21% (11 to 30%) <0.0001

Re-consultations because of UTI 27 (20%) 10 (8%) 12% (3 to 20%) 0.010

Quality of life - range 0 to 10

EuroQol health state 8.8 (2.2) 9.4 (1.5) 0.6 (0.2 to 1.0) 0.005

EuroQol visual analogue scale 7.4 (1.9) 8.3 (1.5) 1.0 (0.5 to 1.4) <0.0001

Patient satisfaction with UTI-management 5.7 (3.0) 8.2 (2.1) 2.5 (1.9 to 3.2) <0.0001

Number of working days lost due to UTI 0.6 (0.8) 0.5 (0.8) 0.2 (-0.1 to 0.5)* 0.18

* Relative rate increase calculated from Poisson regression 1 2

3

Figure 2 presents time to event curves for resolution of symptoms (Panel A) and use 4

of antibiotics (Panel B) until day 10 (Figure 2 and Supplementary Table 2). The course of 5

mean symptom scores is shown in Supplementary Figure 1. The median time until resolution 6

was 4 days in the NSAID versus 2 days in the antibiotic group (HR 1.64, CI 1.26-2.14, 7

p<0.0001,). The median times until antibiotic use were 5 and 0 days, respectively (HR 10.06, 8

CI 6.67 to 15.17, p<0.0001). Figure 3 shows subgroup analyses for the primary and the 9

principal secondary outcome. Results appeared consistent across all subgroups. Sensitivity 10

analyses revealed consistent results for the primary and main secondary outcome 11

(Supplementary Table 3). 12

Remaining pre-specified outcomes favoured antibiotic therapy except for change of 13

symptom score on day 30 (p=0.88) and working days lost (p=0.18, Table 2). Post-hoc 14

analyses of additional time-points revealed little evidence of a difference between groups for 15

Page 13 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 14 / 26

resolution or complete absence of symptoms on day 10 and 30 or change of symptom score 1

on day 10 (p≥0.07) and strong evidence for a difference in favour of the NSAID group for 2

antibiotic use up to day 3 (p<0.0001, Table 2). 3

Supplementary Table 4 shows a post-hoc comparison of baseline characteristics of 4

women who had never been on antibiotics (n=51) with women who had used any antibiotics 5

until day 30 (n=82) among those randomly allocated to NSAIDs. We found little evidence for 6

a difference between groups, except for one component of the symptom composite score. 7

Supplementary Table 5 presents a comparison of outcomes. Resolution and complete 8

absence of symptoms at day 3 were more frequent among women never on antibiotics 9

(p≤0.006), changes of symptom scores were more pronounced at day 3 (p=0.0001), 10

reconsultations were less frequent (p<0.0001), and scores on quality of life and satisfaction 11

with care were higher (p≤0.015). The median time until symptom resolution in women never 12

on antibiotics was 3 days and the median time in women who had used antibiotics was 5 13

days (HR 0.61, CI 0.42 to 0.88, p=0.003, Supplementary Figure 2 and Supplementary Table 14

6). Thirty-four urinary cultures had been positive at baseline among women never on 15

antibiotics, of these 16 had become spontaneously negative on day 10 (47%). 16

Table 3 presents adverse events that resulted in reconsultations: 43 events in 41 17

women in the NSAID group (31%), and 22 events 21 women in the antibiotic group (18%). 18

Adverse events related to UTI were more frequent in the NSAID-group (p=0.012), with 6 19

cases of clinically diagnosed pyelonephritis in the NSAID group (5%) and none in the 20

antibiotic group (p=0.031) and one woman with clinically diagnosed pyelonephritis in the 21

NSAID group classified to have experienced a serious adverse event since she was 22

hospitalized to receive intravenous antibiotic therapy. The median time from symptom onset 23

to clinical diagnosis of pyelonephritis was 5.5 days (range 5 to 8 days). CRP levels > 10 mg/L 24

at baseline were observed in 21 women in the NSAID group without pyelonephritis (17%) 25

and in 3 women with pyelonephritis (50%, Supplementary Table 7). This resulted in a 26

positive likelihood ratio of 3.02 (CI 1.07 to 5.98, p=0.037) for CRP levels > 10 mg/L. None of 27

the blood or urine findings at baseline were associated with clinically relevant positive or 28

Page 14 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 15 / 26

negative likelihood ratios suitable to rule in or out future diagnosis of a pyelonephritis. 1

Although the symptom score for women who developed pyelonephritis was higher at 2

baseline compared to women who did not develop pyelonephritis, there were no differences 3

in symptom score from day 3 to 30 (Supplementary Table 8). 4

Table 3: Adverse events resulting in reconsultations up to 30 days 5

Adverse event NSAIDs (N=133) Antibiotics (N=120) P value

no. of patients (%)

Related to UTI 26 (20%) 10 (8%) 0.012

Persistence of symptoms 16 (12%) 4 (3%) 0.011

Additional symptoms 6 (5%) 2 (2%) 0.29

Recurrent UTI* 5 (4%) 4 (3%) 1.00

Pyelonephritis† 6 (5%) 0 (0%) 0.031

Other adverse event 17 (13%) 12 (10%) 0.56

Exanthema 1 (1%) 2 (2%) 0.61

Vaginitis 3 (2%) 0 (0%) 0.25

Gastrointestinal complaints‡ 3 (2%) 3 (3%) 1.00

Low back pain§ 5 (4%) 2 (2%) 0.45

Viral infection 1 (1%) 3 (3%) 0.35

Trauma 3 (2%) 1 (1%) 0.62

Miscellaneous# 3 (2%) 1 (1%) 0.62

Reported are numbers of patients experiencing at least one adverse event for each category, therefore, numbers do not add up. 6 P values are from 2-sided Fisher’s exact test 7

* Recurrent UTI was defined as additional visits after day 14 because of recurrent UTI symptoms after symptoms had resolved 8 by day 10, and the physician decided to treat with antibiotics. 9

† One patient in NSAID group hospitalized on day 4 because of pyelonephritis. 10 ‡ Includes one case of diverticulitis in antibiotic group. 11 § Considered to be of musculoskeletal origin by treating physician. 12 # Includes one case of external otitis and two cases of tonsillitis in NSAID group and 1 case of hair loss in Antibiotic group. 13 14

15

16

Systematic Review and Meta-analysis 17

We reviewed 81 studies and excluded 78 in the abstract screen. Of the three articles 18

undergoing full-text review, one article was a published study protocol for an ongoing clinical 19

trial that may be indeed by eligible for inclusion upon completion. Therefore, three trials, 20

including our own, were included in the meta-analysis. Together, the studies included 806 21

women (410 randomized to NSAIDs and 382 randomized to antibiotics). Compared to the 22

antibiotics group, women randomized to NSAIDs had a higher incidence of incomplete or no 23

symptom resolution at 3 to 4 days (risk difference 18.0%, CI 11.3% to 24.7%; Number 24

Needed to Harm (NNH) 6, CI 4 to 9, Figure 4) and at 7 days (risk difference 9.6%, CI 4.1% to 25

Page 15 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 16 / 26

15.1%, NNH 10, CI 7 to 24). The incidence of pyelonephritis was also higher in women 1

randomized to NSAIDs compared to women randomized to antibiotics (risk difference 2.6%, 2

CI 0.8% to 4.5%, NNH 38, CI 22 to 126). However, antibiotic use was markedly lower in 3

women randomized to NSAIDs compared to women randomized to antibiotics (risk difference 4

-56.1 %, CI -60.8% to -51.4%, Number Needed to Treat 2, CI 2 to 2) and there was little 5

evidence of a difference in the frequency of recurrent UTIs between the groups (Figure 4). 6

Notably, point estimates for the fixed effect meta-analysis and the Bayesian random-effects 7

meta-analysis were similar, although confidence intervals were wider for the random effects 8

model (Figure 5). However, limiting our analysis to large trials increased the precision of the 9

effect estimate. There was little heterogeneity for all outcomes except for use of antibiotics, 10

for which the risk difference was lower in our study compared to the previous trials (Figure 5). 11

12

13

Page 16 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 17 / 26

DISCUSSION 1

In this randomized, double-blind trial in women with uncomplicated lower urinary tract 2

infection, symptomatic treatment with the NSAID diclofenac was inferior to antibiotic 3

treatment with norfloxacin in controlling symptoms. Those treated with diclofenac were 27% 4

less likely to have symptom resolution at day 3 and 12% less likely to have symptom 5

resolution at day 7 after randomization, with higher mean symptom scores, more frequent 6

reconsultations, a higher incidence of clinically diagnosed pyelonephritis and lower patient 7

satisfaction than those in the NSAID group. Conversely, women who received NSAIDs were 8

37% less likely to receive antibiotics until day 30 after randomization. 9

A meta-analysis of five trials concluded that antibiotics are clinically superior to 10

placebo in women with uncomplicated lower UTI.20 Our trial, in conjunction with the 11

accompanying meta-analysis, suggests that antibiotics are also clinically superior to 12

symptomatic treatment with NSAIDs. This contrasts with the conclusions of a small pilot trial 13

by Bleidorn et al,9 which found clinical outcomes of treatment with ibuprofen similar to those 14

of antibiotic therapy with ciprofloxacin. The pilot trial triggered both, our trial and the recently 15

published trial by Gágyor et al.10 Both of these trials were adequately powered, but failed to 16

detect non-inferiority of NSAIDs as compared with antibiotics in terms of symptom control. 17

Importantly, our trial and the accompanying meta-analysis suggest that symptomatic 18

treatment with NSAIDs is associated with an average 2.6% increase in the absolute risk of 19

clinically diagnosed pyelonephritis, which translates into a NNH of 38 as compared with 20

antibiotic treatment. The risk of pyelonephritis with NSAIDs is comparable to the risk of 0.4 to 21

2.6% observed with placebo in two earlier trials,21, 22 despite the previously described 22

antibacterial activity of diclofenac23 and ibuprofen.24 23

On the beneficial side, our meta-analysis suggests that antibiotic use can be halved 24

on average by initial symptomatic treatment with NSAIDs, with a corresponding number 25

needed to treat of 2 to prevent one instant of antibiotic use. Although there was moderate 26

heterogeneity for this outcome, it may relate to methodological differences in the conduct of 27

Page 17 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 18 / 26

our trial compared to prior trials. In particular, we provided women with a rescue antibiotic for 1

discretionary use after completion of study medication and reaching the primary endpoint. 2

This may have facilitated antibiotic use in the NSAID treatment arm of our trial. Nonetheless, 3

given the high global incidence of UTIs, the reduction in antibiotic use is highly relevant and 4

is likely to immediately decrease resistance rates for E. coli and even other microorganisms 5

in the affected population.25 6

Rising antibiotic resistance among uropathogens in general, and Escherichia coli (the 7

most common uropathogen), in particular, is a global concern. Many studies show a clear 8

correlation between antibiotic consumption and rising resistance rates. Accordingly, 9

antibiotics are often withheld in cases of self-limited, benign bacterial diseases as acute otitis 10

media, sinusitis and traveller’s diarrhoea at the cost of a prolongation of symptoms by 11

typically 1-3 days.26-33 Our results in women with uncomplicated lower UTI are well in line 12

with the prolongation of symptoms observed with symptomatic treatment of these conditions. 13

As many women in the NSAID group resorted to antibiotic therapy in our trial, a strategy of 14

selectively deferring rather than completely withholding antibiotic therapy may be preferable 15

for uncomplicated lower UTI.34 This can be achieved through a shared decision making 16

process, during which clinicians inquire about their patient’s ideas and expectations 17

regarding antibiotic treatment for uncomplicated UTI and also explore the option of delaying 18

antibiotic use as a treatment strategy. 19

Subgroup analyses did not provide evidence for any clinically relevant treatment by 20

subgroup interactions. In particular, in contrast to Gágyor et al.10, reduction in antibiotic 21

prescription was comparable in women with and without positive urinary cultures. Testing 22

initial urine samples for other biomarkers associated with UTI, like heparin-binding protein, 23

interleukin-6, acetic acid, trimethylamine, xanthine oxidase, myeloperoxidase, or others,35-37 24

might have resulted in promising treatment by subgroup interactions, but these tests are not 25

yet established in clinical practice, and we are unaware of any evidence to suggest that such 26

interactions would be likely. When comparing women who never had been on antibiotics with 27

women who had used antibiotics until day 30 among those allocated to NSAIDs in post-hoc 28

Page 18 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 19 / 26

analyses, we unsurprisingly found better clinical outcomes in those never on antibiotics, but 1

no relevant differences in baseline characteristics that would allow an early identification of 2

women likely to benefit from NSAIDs alone. However, in further post-hoc analyses, which are 3

purely hypothesis generating, we found that the clinical diagnosis of pyelonephritis was 4

established not earlier than 5 days after symptom onset, and that CRP values above 10 mg/L 5

were more frequent at baseline in women who subsequently were diagnosed with a 6

pyelonephritis. Taken together, these exploratory findings could support a tailored strategy of 7

immediate antibiotic use in women with CRP levels of >10 mg/L and early NSAID use in 8

remaining women for up to 3 to 4 days after symptom onset, followed by deferred, selective 9

antibiotic use in those women who did not show a clear improvement up by then. Naturally, 10

such a tailored strategy would need to be evaluated in an appropriately powered randomized 11

trial. 12

Our trial should be interpreted in view of its strengths and limitations. Strengths are its 13

randomized double blind design with appropriate concealment of allocation, blinding of 14

patients, therapists and outcome assessors, the low loss to follow-up, the robustness of 15

results in a series of sensitivity analyses and the multi-centric primary care setting. The 16

premature termination of patient recruitment before reaching the initially planned sample size 17

is an obvious limitation. However, the decision to stop recruitment was made without 18

inspecting the data and is therefore unlikely to have biased our findings.38 Despite the 19

smaller than originally planned sample size, our results are completely unequivocal. While 20

using slightly different symptom scores in different trials does affect the comparability 21

between trials, this will not affect the validity of our results. Nevertheless, we suggest that 22

future trials should consider to use the symptom score developed by Alidjanov et al.39 Finally, 23

our results are not generalizable to countries and clinical settings with lower rates of 24

susceptibility, which could decrease the effectiveness of antibiotics and render symptomatic 25

treatment with NSAIDs less inferior. 26

In conclusion, symptomatic therapy is inferior to antibiotic therapy for the treatment of 27

women with uncomplicated lower UTI in an ambulatory setting, as it increases median 28

Page 19 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 20 / 26

symptom duration by two days and is likely to be associated with an increased risk of 1

clinically diagnosed pyelonephritis. The observed impressive reduction in antibiotic use, 2

which would likely contribute directly to decreasing resistance rates in the affected population, 3

suggests that alternate approaches of combining symptomatic treatment with deferred, 4

selective antibiotic use should be developed and tested in future trials. 5

6

Page 20 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 21 / 26

Contributors 1

AK and KM were responsible for conception and design and obtained funding for 2

the study. AK, MB, DM, and PF were responsible for the acquisition of data. LB, BC, 3

AL, SR and PJ did the analysis and interpreted it in collaboration with all the 4

remaining authors. BC, AO and PJ performed the meta-analysis. AK and PJ had full 5

access to the final data, co-wrote the manuscript, had final responsibility for content, 6

and the decision to submit for publication. All authors critically revised the paper for 7

important intellectual content and approved the final version. 8

9

Transparency declarations 10

The authors affirm that the manuscript is an honest, accurate, and transparent 11

account of the study being reported; that no important aspects of the study have 12

been omitted; and that any discrepancies from the study as planned (and, if relevant, 13

registered) have been explained. 14

15

Competing interests: All authors declare: no support from any organisation for the 16

submitted work besides the acknowledged financial support. AK has received travel 17

grant and meeting expenses from Gilead, Viofor and the World Health Organisation 18

(WHO), is advisor of the Swiss Federal Office of Public Health concerning antibiotic 19

resistance epidemiology in Switzerland, and provides non-interpreted annual 20

resistance data to LEO pharmaceutic company and the Swiss government. PJ has 21

received research grants to the institution from Astra Zeneca, Biotronik, Biosensors 22

International, Eli Lilly and The Medicines Company for cardiovascular trials, and 23

serves as unpaid member of the steering group of cardiovascular trials funded by 24

Astra Zeneca, Biotronik, Biosensors, St. Jude Medical and The Medicines Company. 25

The remaining authors declare no financial relationships with any organisations that 26

might have an interest in the submitted work in the previous three years and no other 27

relationships or activities that could appear to have influenced the submitted work. 28

29

Page 21 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 22 / 26

Acknowledgments 1

We thank all patients participating in this study, all participating general physicians for 2

their meticulous data collection and the Clinical Trials Unit of the University of Bern 3

(CTU Bern) for the development of the database (Malcolm Sturdy), Monitoring (Lucia 4

Kacina) and telephone interview and data clearance (Madeleine Dähler). We thank 5

the Institute of General Practice BIHAM, for their organisational support and the 6

Hospital Pharmacy of the Inselspital for the production and blinding of the study 7

medications (Marco Eschenmoser). 8

9

Participating physicians. Peter Duner, Eggiwil; Christoph Fry, Belp; Ursula Grob, 10

Herzogenbuchsee; Felix Huber, Zürich; Beat Köstner-Mösching, Neuenegg; Andreas 11

Kronenberg, Bern; Corinna Kronenberg, Stettlen; Danielle Lemann, Langnau; 12

Damian Meli, Huttwil; Gabriele Reinheimer, Worb; Véronique Rigamonti Wermelinger, 13

Bern; Ralf Schäfer, Bern; Urs Schneeberger, Niederönz; Christian Studer, Luzern; 14

Fritz Weber, Buchs; Doris Zundel, Bätterkinden; Anne-Marie Zundel Funk, 15

Zollikofen. 16

17

Funding 18

This study was supported financially by the Swiss National Foundation (SNF project 19

32003B_130867), the Swiss Academy of Medical Sciences, the SwissLife foundation 20

and the Else Kroener-Fresenius foundation. 21

22

23

Figures and Supplements 24

See separate files 25

26

Page 22 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 23 / 26

References 1

1. Organization WH. Global Action Plan on Antimicrobial Resistance. WHO Library 2

Cataloguing-in-Publication Data; 2015. 3

2. Foxman B. Epidemiology of urinary tract infections: incidence, morbidity, and economic 4

costs. Am J Med. 2002; 113 Suppl 1A: 5S-13S. 5

3. Foxman B, Brown P. Epidemiology of urinary tract infections: transmission and risk 6

factors, incidence, and costs. Infect Dis Clin North Am. 2003; 17(2): 227-41. 7

4. Mühlemann K. Surveillance of antibiotic prescription in the outpatient setting using the 8

national Sentinel network (OP-159, http://www.smw.ch/docs/PdfContent/smw-12931.pdf). 9

WONCA World Organization of National Colleges, Academies and Academic 10

Associations of General Practitioners/Family Physicians; 2009; Basel. Switzerland; 2009. 11

5. Lundborg CS, Olsson E, Molstad S. Antibiotic prescribing in outpatients: a 1-week 12

diagnosis-prescribing study in 5 counties in Sweden. Scand J Infect Dis. 2002; 34(6): 13

442-8. 14

6. Farkas A, Alajem D, Dekel S, Binderman I. Urinary prostaglandin E2 in acute bacterial 15

cystitis. J Urol. 1980; 124(4): 455-7. 16

7. Schwarz NT, Jung SY, Kalff JC, Chancellor M, Bauer AJ. Bacterial toxin N-formyl-17

methionyl-leucyl-phenylalanine acutely contracts human and rabbit detrusor through the 18

release of eicosanoids. J Urol. 2002; 167(6): 2603-12. 19

8. Poljakovic M, Svensson ML, Svanborg C, Johansson K, Larsson B, Persson K. 20

Escherichia coli-induced inducible nitric oxide synthase and cyclooxygenase expression 21

in the mouse bladder and kidney. Kidney Int. 2001; 59(3): 893-904. 22

9. Bleidorn J, Gagyor I, Kochen MM, Wegscheider K, Hummers-Pradier E. Symptomatic 23

treatment (ibuprofen) or antibiotics (ciprofloxacin) for uncomplicated urinary tract 24

infection?--results of a randomized controlled pilot trial. BMC Med. 8: 30. 25

10. Gagyor I, Bleidorn J, Kochen MM, Schmiemann G, Wegscheider K, Hummers-Pradier E. 26

Ibuprofen versus fosfomycin for uncomplicated urinary tract infection in women: 27

randomised controlled trial. BMJ. 2015; 351: h6544. 28

11. Little P, Merriman R, Turner S, Rumsby K, Warner G, Lowes JA, et al. Presentation, 29

pattern, and natural course of severe symptoms, and role of antibiotics and antibiotic 30

resistance among patients presenting with suspected uncomplicated urinary tract 31

infection in primary care: observational study. BMJ. 2010; 340: b5633. 32

12. Gupta K, Hooton TM, Roberts PL, Stamm WE. Short-course nitrofurantoin for the 33

treatment of acute uncomplicated cystitis in women. Arch Intern Med. 2007; 167(20): 34

2207-12. 35

Page 23 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 24 / 26

13. McNulty CA, Richards J, Livermore DM, Little P, Charlett A, Freeman E, et al. Clinical 1

relevance of laboratory-reported antibiotic resistance in acute uncomplicated urinary 2

tract infection in primary care. J Antimicrob Chemother. 2006; 58(5): 1000-8. 3

14. Clayson D, Wild D, Doll H, Keating K, Gondek K. Validation of a patient-administered 4

questionnaire to measure the severity and bothersomeness of lower urinary tract 5

symptoms in uncomplicated urinary tract infection (UTI): the UTI Symptom Assessment 6

questionnaire. BJU Int. 2005; 96(3): 350-9. 7

15. Grabe M (chairman) B-JT, Botto H, Çek M, Naber KG, Pickard RS, Tenke P, 8

Wagenlehner F, Wullt B. Guidelines on Urological Infections. 2013 [cited 2015 9

9.7.2015]; Available from: http://uroweb.org/wp-content/uploads/18_Urological-10

infections_LR.pdf 11

16. Chow SC SJ, Wang H Sample Size Calculations in Clinical Research. Second Edition 12

ed: Chapman & Hall/CRC; 2008. 13

17. Chung Y, Rabe-Hesketh S, Choi IH. Avoiding zero between-study variance estimates in 14

random-effects meta-analysis. Stat Med. 2013; 32(23): 4071-89. 15

18. Friede T, Rover C, Wandel S, Neuenschwander B. Meta-analysis of few small studies in 16

orphan diseases. Res Synth Methods. 2017; 8(1): 79-91. 17

19. Higgins JP, Thompson SG, Spiegelhalter DJ. A re-evaluation of random-effects meta-18

analysis. J R Stat Soc Ser A Stat Soc. 2009; 172(1): 137-59. 19

20. Falagas ME, Kotsantis IK, Vouloumanou EK, Rafailidis PI. Antibiotics versus placebo in 20

the treatment of women with uncomplicated cystitis: a meta-analysis of randomized 21

controlled trials. J Infect. 2009; 58(2): 91-102. 22

21. Ferry SA, Holm SE, Stenlund H, Lundholm R, Monsen TJ. Clinical and bacteriological 23

outcome of different doses and duration of pivmecillinam compared with placebo therapy 24

of uncomplicated lower urinary tract infection in women: the LUTIW project. Scand J 25

Prim Health Care. 2007; 25(1): 49-57. 26

22. Christiaens TC, De Meyere M, Verschraegen G, Peersman W, Heytens S, De 27

Maeseneer JM. Randomised controlled trial of nitrofurantoin versus placebo in the 28

treatment of uncomplicated urinary tract infection in adult women. The British journal of 29

general practice : the journal of the Royal College of General Practitioners. 2002; 30

52(482): 729-34. 31

23. Mazumdar K, Dutta NK, Dastidar SG, Motohashi N, Shirataki Y. Diclofenac in the 32

management of E. coli urinary tract infections. In Vivo. 2006; 20(5): 613-9. 33

24. Obad J, Suskovic J, Kos B. Antimicrobial activity of ibuprofen: new perspectives on an 34

"Old" non-antibiotic drug. Eur J Pharm Sci. 2015; 71: 93-8. 35

25. Gottesman BS, Carmeli Y, Shitrit P, Chowers M. Impact of quinolone restriction on 36

resistance patterns of Escherichia coli isolated from urine by culture in a community 37

Page 24 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 25 / 26

setting. Clinical infectious diseases : an official publication of the Infectious Diseases 1

Society of America. 2009; 49(6): 869-75. 2

26. Glasziou PP, Del Mar CB, Sanders SL, Hayem M. Antibiotics for acute otitis media in 3

children. Cochrane Database Syst Rev. 2004; (1): CD000219. 4

27. Falagas ME, Giannopoulou KP, Vardakas KZ, Dimopoulos G, Karageorgopoulos DE. 5

Comparison of antibiotics with placebo for treatment of acute sinusitis: a meta-analysis 6

of randomised controlled trials. Lancet Infect Dis. 2008; 8(9): 543-52. 7

28. Ahovuo-Saloranta A, Borisenko OV, Kovanen N, Varonen H, Rautakorpi UM, Williams 8

JW, Jr., et al. Antibiotics for acute maxillary sinusitis. Cochrane Database Syst Rev. 9

2008; (2): CD000243. 10

29. Del Mar CB, Glasziou PP, Spinks AB. Antibiotics for sore throat. Cochrane Database 11

Syst Rev. 2006; (4): CD000023. 12

30. de Bruyn GH, S. Borwick, A. Antibiotic treatment for travellers' diarrhoea. Cochrane 13

Database of Systematic Reviews, Issue 3, 2009. 2009. 14

31. Genton B, D'Acremont V. Evidence of efficacy is not enough to develop 15

recommendations: antibiotics for treatment of traveler's diarrhea. Clin Infect Dis. 2007; 16

44(11): 1520; author reply 1-2. 17

32. Tan T, Little P, Stokes T. Antibiotic prescribing for self limiting respiratory tract infections 18

in primary care: summary of NICE guidance. Bmj. 2008; 337: a437. 19

33. Rovers MM, Glasziou P, Appelman CL, Burke P, McCormick DP, Damoiseaux RA, et al. 20

Antibiotics for acute otitis media: a meta-analysis with individual patient data. Lancet. 21

2006; 368(9545): 1429-35. 22

34. Little P, Moore MV, Turner S, Rumsby K, Warner G, Lowes JA, et al. Effectiveness of 23

five different approaches in management of urinary tract infection: randomised controlled 24

trial. BMJ. 2010; 340: c199. 25

35. Kjolvmark C, Pahlman LI, Akesson P, Linder A. Heparin-binding protein: a diagnostic 26

biomarker of urinary tract infection in adults. Open Forum Infect Dis. 2014; 1(1): ofu004. 27

36. Lam CW, Law CY, To KK, Cheung SK, Lee KC, Sze KH, et al. NMR-based metabolomic 28

urinalysis: a rapid screening test for urinary tract infection. Clin Chim Acta. 2014; 436: 29

217-23. 30

37. Ciragil P, Kurutas EB, Miraloglu M. New markers: urine xanthine oxidase and 31

myeloperoxidase in the early detection of urinary tract infection. Dis Markers. 2014; 32

2014: 269362. 33

38. Bassler D, Briel M, Montori VM, Lane M, Glasziou P, Zhou Q, et al. Stopping randomized 34

trials early for benefit and estimation of treatment effects: systematic review and meta-35

regression analysis. Jama. 2010; 303(12): 1180-7. 36

Page 25 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 26 / 26

39. Alidjanov JF, Abdufattaev UA, Makhsudov SA, Pilatz A, Akilov FA, Naber KG, et al. New 1

self-reporting questionnaire to assess urinary tract infections and differential diagnosis: 2

acute cystitis symptom score. Urol Int. 2014; 92(2): 230-6. 3

4 5

Page 26 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

Kronenberg A. et al. Symptomatic therapy of uncomplicated lower urinary tract infections (UTI)

in the ambulatory setting. A randomized, double blind trial.

1

Figures for publication

Figure 1: Patient flow

Page 27 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly



2

Figure 2: KM plot for (A) time until definite resolution of symptoms and (B) time until antibiotic up to day 10.

Page 28 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly



3

Figure 3: Stratified analyses of (A) primary outcome and (B) principal secondary outcome. Positive differences indicate an advantage of antibiotics, negative differences an advantage of NSAIDs throughout.

NSAIDs Antibiotics

Risk difference (95% CI) P value for interaction

no. of patients/total no. (%)

A. Resolution of symptoms at day 3

Overall 72/133 (54%) 96/120 (80%)

27% (15 to 38%)

Age (y) 0.39

<45 yr 52/94 (55%) 69/89 (78%) 23% (9 to 37%)

≥45 yr 20/39 (51%) 27/31 (87%) 36% (16 to 55%)

Symptom score 0.36

≤20 71/127 (56%) 93/115 (81%) 26% (14 to 37%)

>20 1/6 (17%) 3/5 (60%) 43% (-9 to 96%)

Symptom duration 0.86

≤3 days 43/80 (54%) 66/83 (80%) 27% (12 to 41%)

>3 days 29/53 (55%) 30/37 (81%) 27% (8 to 45%)

Urine culture 0.84

Negative 21/36 (58%) 28/31 (90%) 30% (9 to 50%)

Positive 50/97 (52%) 69/89 (78%) 26% (12 to 39%)

Urine leucocytes* 0.37

≤2+ 18/32 (56%) 22/30 (73%) 16% (-8 to 40%)

>2+ 54/101 (53%) 75/90 (83%) 30% (17 to 43%)

Documented norfloxacin-susceptible Enterobacteriaceae 0.84

No 25/45 (56%) 34/41 (83%) 27% (7 to 47%)

Yes 47/88 (53%) 63/79 (80%) 26% (12 to 40%)

B. Use of any antibiotic up to day 30

Overall 82/133 (62%) 118/120 (98%) -37% (-46 to -28%)

Age (y) 0.38

<45 yr 60/94 (64%) 87/89 (98%) -34% (-45 to -24%)

≥45 yr 22/39 (56%) 31/31 (100%) -44% (-59 to -28%)

Symptom score 0.82

≤20 78/127 (61%) 113/115 (98%) -37% (-46 to -28%)

>20 4/6 (67%) 5/5 (100%) -33% (-71 to 4%)

Symptom duration 0.89

≤3 days 49/80 (61%) 81/83 (98%) -36% (-48 to -25%)

>3 days 33/53 (62%) 37/37 (100%) -38% (-51 to -25%)

Urine culture 0.25

Negative 19/36 (53%) 31/31 (100%) -46% (-63 to -29%)

Positive 63/97 (65%) 87/89 (98%) -33% (-44 to -23%)

Urine leucocytes* 0.38

≤2+ 17/32 (53%) 30/30 (100%) -44% (-62 to -26%)

>2+ 64/101 (63%) 89/90 (99%) -35% (-44 to -25%)

Documented norfloxacin-susceptible Enterobacteriaceae 0.45

No 26/45 (58%) 41/41 (100%) -42% (-57 to -27%)

Yes 56/88 (64%) 77/79 (97%) -34% (-45 to -23%)

*Post-hoc analysis

Page 29 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly



4

Figure 4: Mantel-Haenszel fixed-effect meta-analysis of randomized double-blind trials comparing non-steroidal anti-inflammatory drugs with antibiotics in women with uncomplicated lower UTI.

Page 30 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly



5

Figure 5: Mantel-Haenszel fixed-effect meta-analysis and Bayesian random effects meta-analysis of all

randomized double-blind trials and large randomized double-blind trials comparing non-steroidal anti-inflammatory

drugs with antibiotics in women with uncomplicated lower UTI

Page 31 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 1 / 19

Supplementary Appendix

This appendix has been provided by the authors to give readers additional information about

their work.

Supplement to: Kronenberg A. et al. Symptomatic therapy of uncomplicated lower urinary

tract infections in the ambulatory setting. A randomized, double blind trial.

Page 32 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 2 / 19

Supplementary Appendix

Table of contents

1. Investigator and committee members ................................................................. 3

2. Collaborators ......................................................................................................... 4

3. Supplementary methods ...................................................................................... 5

4. Supplementary tables and figures ..................................................................... 10

Page 33 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 3 / 19

1. Investigator and committee members

The members of the HWI trial study group are as follows

Steering committee:

K. Mühlemann, A. Kronenberg, P. Jüni

Clinical events committee

K. Mühlemann, A. Kronenberg

Data safety monitoring board

K. Mühlemann, A. Kronenberg, P. Jüni

Study investigators, recruiting patients: (in alphabetical order)

Investigator Study center (practice)

Duner Peter Eggiwil Fry Christoph Belp Grob Ursula Herzogenbuchsee Huber Felix Zürich Köstner-Mösching Beat Neuenegg Kronenberg Andreas Bern Kronenberg Corinna Stettlen Lemann Danielle Langnau Meli Damian Huttwil Reinheimer Gabriele Worb Rigamonti Wermelinger Véronique Bern Schäfer Ralf Bern Schneeberger Urs Niederönz Studer Christian Luzern Weber Fritz Buchs Zundel Doris Bätterkinden Zundel Funk Anne-Marie Zollikofen

Page 34 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 4 / 19

2. Collaborators

Collaborators Function

Nartey Linda Study coordination Heuer Sarah Study coordination Gnahoré Thierry Study coordination Dähler Madeleine Study coordination, telephone interview day 30 Sager Ursina Study coordination, telephone interview day 30 Jäggi Regula Study coordination, telephone interview day 30 Kacina Lucia Monitoring Sturdy Malcolm Development of database Trelle Sven Randomisation Eschenmoser Marco Production of study drugs Droz Sara Microbiological analysis

Page 35 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 5 / 19

3. Supplementary methods

Outcome definitions

Primary outcome

- proportion of patients whose symptoms resolved 3 days after they began treatment.

Main secondary outcome

- proportion of patients who took any antibiotics between the time they were

randomized and day 30 of follow-up

Additional secondary outcomes

- proportion of patients whose symptoms resolved 7 days after treatment began

- proportion of patients who had no symptoms at all on day 3 and on day 7 after

treatment began

- proportion of patients who took single dose fosfomycin as rescue therapy

- proportion of patients whose urinary culture was negative 10 days after treatment

began

- proportion of patients who came back to consult about UTI within 30 days after

treatment began

- mean composite symptom scores 3, 7 and 30 days after treatment began

- time to resolution of symptoms

- proportion of patients who had adverse events

- proportion of patients how had serious adverse events (based on ICH GCP 24)

- quality of life and working days lost on day 3

- overall satisfaction with the way their UTI was managed on day 10

Baseline assessment

At baseline, we determined how long patients had had symptoms before going to the doctor,

and took a capillary blood sample and a mid-stream urinary sample from each participant.

The blood was locally analyzed for C-reactive protein and leukocyte count. The urine was

used for pregnancy testing and dipstick analysis, also locally. The results of the dipstick tests

were semiquantitative categories from 0 to 4+ for erythrocytes and leucocytes, and

dichotome results for nitrites (yes-no). A urine sample (4ml volume) was sent in containers

pre-filled with boric acid preservative (Becton Dickinson, Franklin Lakes, NJ) to the Institute

Page 36 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 6 / 19

for Infectious Diseases, University of Bern, to be cultured. Standard laboratory procedures

were used to analyze and identify all microbiological samples. We used a cut-off of >103

cfu/ml to define a urinary culture as positive, according to European guidelines. Mixed flora

with no predominant microorganism, Lactobacilli and Streptococcus viridans group were

judged negative.

Diary

Patients received a printed paper booklet at the initiation visit. They were told to fill it out

every day at the same time, for 10 days, starting 24 hours after they began therapy. Every

day started on a new page, on which was printed the number of days since they began

therapy, and the calendar date. All pages began with a cross table where participants

rated symptom severity in following six dimensions; dysuria, daytime frequency, night time

frequency, urgency, abdominal pain when passing urine, pain or tenderness in the lower

back or loin. During the initiation visit (day 0) physicians filled out the cross-table with

patients to assess symptom severity before therapy and show patients how to fill out the form.

Day 0 included a question about duration of symptoms and number of UTI during the last 12

months. Pages for day 1 to 3 had blanks where participants fed out the date and time they

took their study medications. Day 3 included a total of four pages; the additional pages

assessed quality of life, recorded their decision to take or not take fosfomycin, and gave

patients the opportunity to note lost working days and side effects. We asked for following

specific side effects: nausea-vomiting, headache, vertigo, tiredness, skin eruptions, difficulty

breathing (e.g. asthma), gastric pain/retrosternal burning, abdominal discomfort/cramps,

diarrhea, or constipation. Three lines were left blank to give patients the opportunity to add

and rate additional side effects. Patients rated severity of each side effect on a Likert scale

from 0 (“no side effects”) to 6 (“worst possible side effects”). At the end of each page there

was a section for free text. Instructions were printed on the last pages. These told patients

that they should now mail the diary, including all unused medications, all empty medication

packages, and a new urine sample to the Institute for Infectious Diseases, University of Bern.

Follow-up

Following UTI symptoms were assessed: dysuria; daytime frequency; night time

frequency; urgency; abdominal pain when passing urine; and, pain or tenderness in the lower

back or loin. The severity of each symptom was recorded on a Likert scale (0 to 6; 0

indicated “normal”, 1 “very little” problems, 2 “slight” problems, 3 “moderately bad”, 4 “bad”, 5

“very bad”, and 6 “as bad as it could be”). We used the mean of daytime and nighttime for

frequency. Quality of Life was assessed using the five dimensions and the visual analogue

Page 37 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 7 / 19

scale (VAS) from the European Quality of Life questionnaire (EQ-5D) 28. General satisfaction

on day 10 was assessed using a numeric scale that ranged from 1 (“not satisfied at all, would

never do this again”) to 10 (“very satisfied”).

Side effects were systematically assessed in the diary on day 3 with a Likert scale

scored from 0 (“no side effects”) to 6 (“worst possible side effects”). We asked them if they

had the following specific side effects: nausea-vomiting, headache, vertigo, tiredness, skin

eruptions, difficulty breathing (e.g. asthma), gastric pain/retrosternal burning, abdominal

discomfort/cramps, diarrhea, or constipation. Three lines were left blank to give patients the

opportunity to add and rate additional side effects. All participating physicians were required

to immediately and systematically report serious adverse events (SAE) and adverse events >

grade 3, according to GCP guidelines.

Additional visits were grouped by the primary investigator, based on the diary and, if

necessary, original documents. Recurrent UTI was defined as additional visits after day 14

because of recurrent UTI symptoms after symptoms had resolved by day 10 (missing data in

2/9), and the physician decided to treat with antibiotics. To assess compliance we used

entries in the diary. If we did not receive information about patient adherence, or if patients

did not document the intake of at least one tablet of study medication, they were excluded

from the per-protocol analysis.

Data management

Baseline urinary samples and baseline local laboratory results on paper forms, second

urinary sample, diaries and unused study medications, were all sent to the Institute for

Infectious Diseases, University of Bern, Switzerland. Urinary samples were analysed locally;

all other material was forwarded to an academic clinical trials unit (CTU Bern, University of

Bern, Switzerland) where data managers fed the data into a central database. The

CTU continuously assessed data quality and completeness, completed missing data as far

as possible, and conducted telephone interviews on day 30. An independent study monitor

from the CTU verified the complete source data for the first two patients enrolled at each

study site, and for 10% of a random sample per study site thereafter. The independent

monitor checked informed consent forms, eligibility, serious adverse events and the primary

endpoint for all included patient.

Statistical analysis

Symptom scores at day 3 were imputed together with baseline symptom scores, based on

patient age, treatment group, baseline laboratory measurements (blood CRP and leucocytes,

urine leucocytes, erythrocytes and nitrite), presence of a positive urinary culture (at baseline

and day 10), duration of UTI symptoms before study start, recurrence of UTI in the last 12

Page 38 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 8 / 19

months, number of working days lost due to UTI. All other outcomes were imputed

separately, based on symptom scores at baseline and day 3 and the same covariates.

Predictive mean matching was used to impute continuous outcomes, binary outcomes were

imputed by using logistic regression. In total, we generated 20 imputed data sets and used

Rubin’s rules to analyzed them, as described above (Rubin DB. Multiple Imputation for

Nonresponse in Surveys. New York: J Wiley & Sons. 1987).

Bayesian Random Effects Meta-analysis

We used a Bayesian approach developed for random effects meta-analysis on the risk

difference scale [Warn et al.]. The model adequately restricts the risk difference (RD) and its

95% credibility interval (95% CrI) within the plausible values of -1 and 1 and accounts for

situations with sparse event data, including zero cells in one or both treatment groups.

Monte-Carlo Markov Chain simulation methods were used to obtain posterior distributions of

the RDs of outcomes of interest and . Pooled RDs were estimated from the median of the

respective posterior distributions. A RD below zero indicates a benefit of NSAIDs as

compared with antibiotics. 95% CrI were obtained from the 2.5th and the 97.5th percentile of

the posterior distribution, which can be interpreted similarly to a conventional 95%

confidence interval if priors are minimally informative. Between-trial heterogeneity was

considered low if the median of the posterior distribution of was around 0.001, which

corresponds to a 95% reference range of RDs from -0.062 to 0.062 if RDs are normally

distributed around 0. Heterogeneity was considered moderate if the median of the posterior

distribution of was around 0.005, which corresponds to a 95% reference range of RDs

from -0.139 to 0.139. Heterogeneity was considered high if the median of the posterior

distribution of was around 0.010, which corresponds to a 95% reference range of RDs

from -0.196 to 0.196 [see Spiegelhalter et al, pages 168 ff for an explanation of the concept].

The prior chosen for treatment effects on an RD scale followed a normal distribution with

mean 0 and standard deviation of 0.53 for all outcomes except pyelonephritis, which resulted

in a 95% reference range from -0.98 to 0.98. The prior for the baseline risk in the control

group followed a beta distribution with shape parameters (1,1) for all outcomes, except

pyelonephritis and antibiotic use, corresponding to a uniform distribution from 0 to 1 and a

95% reference range from 0.02 to 0.98. The prior for followed a beta distribution with shape

parameters (1,9) for all outcomes except pyelonephritis, which resulted in a mode of 0 and

an upper limit of the one-sided 95% reference range of 0.336.

The risk of pyelonephritis is less than 0.026 in women with lower uncomplicated urinary tract

infection,[Falagas et al] therefore RDs more extreme than +/- 0.05 were considered clinically

Page 39 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 9 / 19

implausible, and the corresponding maximally possible would be around 0.03,

corresponding to a of 0.0009. Priors used for pyelonephritis were therefore following a

normal distribution with mean 0 and standard deviation of 0.051 for the treatment effect,

which resulted in a 95% reference range from -0.1 to 0.1, and a beta distribution with shape

parameters (1,40) for and for the baseline risk in the control group, which resulted in a

mode of 0 and an upper limit of the one-sided 95% reference range of 0.072. The probability

to receive antibiotics in control groups that were actually allocated to antibiotics was likely to

be above 0.85 in western settings. We therefore chose a beta distribution with shape

parameters (9,1) for the baseline risk of antibiotic use in the control group, which resulted in

a mode of 1 and a lower limit of the one-sided 95% reference range of 0.72. Models were ran

with 2 chains. Results were obtained after a burn-in of 30000 iterations, retaining every 2nd

out of 50000 iterations. Model convergence was assessed visually using trace plots and

Gelman-Rubin plots. All Bayesian analyses were conducted in R and WinBUGS.

References

Warn DE, Thompson SG, Spiegelhalter DJ. Bayesian random effects meta-analysis of trials

with binary outcomes: methods for the absolute risk difference and relative risk scales. Stat

Med. 2002 Jun 15;21(11):1601-23.

Spiegelhalter D, Abrams K, Myles J. Bayesian approaches to clinical trials and health care

evaluation. Chichester: John Wiley & Sons, 2004.

Falagas ME, Kotsantis IK, Vouloumanou EK, Rafailidis PI. Antibiotics versus placebo in the

treatment of women with uncomplicated cystitis: a meta-analysis of randomized controlled

trials. J. Infect. 2009;58(2):91–102.

Page 40 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 10 / 19

4. Supplementary tables and figures

Supplementary Table 1: Missing data for primary and secondary outcomes

Total (N=253) NSAIDs (N=133) Antibiotics (N=120)

no. and precentage of missing observations


Day 3 (primary outcome) 22 (9%) 14 (11%) 8 (7%)

Day 7 19 (8%) 11 (8%) 8 (7%)

Day 10 21 (8%) 10 (8%) 11 (9%)

Day 30 0 (0%) 0 (0%) 0 (0%)


Day 3 22 (9%) 14 (11%) 8 (7%)

Day 7 19 (8%) 11 (8%) 8 (7%)

Day 10 21 (8%) 10 (8%) 11 (9%)

Day 30 9 (4%) 5 (4%) 4 (3%)


Day 3 27 (11%) 15 (11%) 12 (10%)

Day 7 24 (9%) 12 (9%) 12 (10%)

Day 10 26 (10%) 11 (8%) 15 (13%)

Day 30 24 (9%) 11 (8%) 13 (11%)

Use of any antibiotic

Up to day 3 9 (4%) 4 (3%) 5 (4%)

Up to day 30 (principal secondary outcome) 9 (4%) 4 (3%) 5 (4%)

Use of rescue antibiotic

Up to day 3 12 (5%) 5 (4%) 7 (6%)

Up to day 30 15 (6%) 6 (5%) 9 (8%)

Negative urinary culture at day10 24 (9%) 12 (9%) 12 (10%)

Re-consultations because of UTI 9 (4%) 4 (3%) 5 (4%)

Quality of life (scaled to 0-10)

EuroQol health state 17 (7%) 9 (7%) 8 (7%)

EuroQol visual analogue scale 15 (6%) 8 (6%) 7 (6%)

Patient satisfaction with UTI-management 8 (3%) 5 (4%) 3 (3%)

Number of working days lost due to UTI 19 (8%) 10 (8%) 9 (8%)

Page 41 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 11 / 19

Supplementary Table 2: Number of participants censored until day 10 for time to event curves for resolution of symptoms and use of antibiotics (Corresponding to Figure 2)

Days since

randomisation

No. of patients

censored

A. Definite resolution NSAIDs 0 9

NSAIDs 10 7

Antibiotics 0 7

Antibiotics 4 1

Antibiotics 10 4

B: Antibiotic use NSAIDs 0 4

NSAIDs 10 54

Antibiotics 0 3

Antibiotics 10 2

Page 42 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 12 / 19

Supplementary Table 3: Sensitivity analyses for (A) primary outcome and (B) principal secondary outcome

Analysis Set NSAIDs Antibiotics Risk difference (95%

CI)

P value†

no. of patients/total no. (%)

A. Resolution of symptoms at day 3

Intention-to-treat analysis* 72/133 (54%) 96/120 (80%) 27% (15 to 38%) <0.0001

Complete case analysis 64/119 (54%) 90/112 (80%) 27% (15 to 38%) <0.0001

Per protocol analysis, multiple imputation 68/122 (55%) 91/111 (82%) 26% (15 to 38%) <0.0001

Per protocol analysis, complete case 64/114 (56%) 88/108 (81%) 25% (14 to 37%) <0.0001

Patients with CRP <30 at baseline 68/125 (54%) 91/113 (81%) 27% (15 to 39%) <0.0001

B. Use of any antibiotic up to day 30

Intention-to-treat analysis* 82/133 (62%) 118/120 (98%) -37% (-46 to -28%) <0.0001

Complete case analysis 79/129 (61%) 115/115 (100%) -39% (-47 to -30%) <0.0001

Per protocol analysis, multiple imputation 72/122 (59%) 111/111 (100%) -41% (-50 to -32%) <0.0001

Per protocol analysis, complete case 71/121 (59%) 111/111 (100%) -41% (-50 to -33%) <0.0001

Patients with CRP <30 at baseline 78/125 (62%) 111/113 (98%) -36% (-45 to -27%) <0.0001

* Primary analysis, 14 and 8 patients in NSAID and antibiotic group with multiple imputation for resolution of symptoms, 4 and 5

in NSAID and antibiotic group with multiple imputation for use of any antibiotic.

† P value for superiority, p value for non-inferiority = 0.977 for resolution of symptoms (primary analysis).

Page 43 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 13 / 19

Supplementary Table 4. Baseline characteristics for patients in NSAID group based on antibiotic use.

Never on

antibiotics (N=51)

Use of antibiotics

(N=82) P value

Age - yr 38.4 (14.6) 37.5 (14.1) 0.75

Age <45 yr 34 (67%) 60 (73%) 0.44

Symptom duration - days since UTI onset 3.7 (4.2) 3.5 (2.1) 0.56

Symptom duration ≤3 days 31 (61%) 49 (60%) 1.00

Number of UTIs in the last 12 months 0.5 (0.8) 0.7 (1.3) 0.58


Dysuria 3.0 (1.4) 3.4 (1.2) 0.07

Urgency 3.6 (1.0) 3.8 (1.1) 0.14

Night frequency 2.5 (1.6) 2.6 (1.4) 0.73

Day frequency 3.3 (0.9) 3.6 (1.0) 0.012

Lower abdominal pain while urinating 2.7 (1.5) 2.6 (1.6) 0.74

Back or loin pain 1.3 (1.4) 1.0 (1.4) 0.25

Total symptom score 13.4 (3.6) 13.9 (4.1) 0.53

Symptom score ≤20 49 (96%) 78 (95%) 1.00

Blood tests

CRP - mg/l 7.1 (13.0) 6.4 (8.6) 0.37

CRP >10 mg/ml 7 (14%) 17 (21%) 0.36

Leucocytes - 1000/µl 8.2 (2.3) 8.7 (2.5) 0.43

Urinary dipstick

Nitrites - positive 7 (14%) 10 (12%) 0.80

Erythrocytes - 1 to 4+ (median, IQR) 3+ (1+ to 4+) 4+ (2+ to 4+) 0.07

Leucocytes - 1 to 4+ (median, IQR) 3+ (2+ to 4+) 4+ (3+ to 4+) 0.12

Leucocytes >2+ 37 (73%) 64 (78%) 0.53

Urinary culture* 0.23

Negative 17 (33%) 19 (23%)

Positive 34 (67%) 62 (76%)

Escherichia coli 30 (59%) 52 (63%) 0.71

Other Enterobacteriaceae 3 (6%) 6 (7%) 1.00

Staphylococcus saprophyticus 0 (0%) 3 (4%) 0.29

Enterococcus faecalis 0 (0%) 1 (1%) 1.00

Beta-hemolytic Streptococcus group B 1 (2%) 0 (0%) 0.38

Susceptibility to norfloxacin 32 (63%) 60 (73%) 0.25

Enterobacteriaceae 32 (63%) 56 (68%) 0.57

Susceptibility to fosfomycin† 29 (57%) 55 (67%) 0.27

Continuous variables are presented as mean (sd), binary variables as no. of patients (%).

* Urinary culture missing in 1 patients in “Use of antibiotics” group.

† Enterobacteriaceae only

Page 44 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 14 / 19

Supplementary Table 5. Post-hoc analysis in NSAID group. Positive differences indicate an advantage of “Use of antibiotics”, negative differences an advantage of “Never on antibiotics” throughout.

Outcome

Never on antibiotics

(N=51)

Use of antibiotics

(N=82)

Risk or mean

difference (95% CI)

P value

no. of patients (%) or mean (sd)


Day 3 40 (78%) 32 (39%) -39% (-55 to -24%) <0.0001

Day 7 45 (88%) 66 (80%) -7% (-20 to 5%) 0.28

Day 10 48 (94%) 78 (95%) 1% (-7 to 9%) 0.83

Day 30 48 (94%) 79 (96%) 2% (-6 to 10%) 0.61


Day 3 8 (16%) 2 (2%) -13% (-24 to -2%) 0.006

Day 7 24 (47%) 21 (26%) -21% (-37 to -4%) 0.017

Day 10 30 (59%) 41 (50%) -8% (-26 to 9%) 0.36

Day 30 37 (73%) 64 (78%) 6% (-9 to 21%) 0.44


Day 3 -9.2 (4.3) -6.1 (4.5) -3.1 (-4.7 to -1.6) 0.0001

Day 7 -11.3 (4.6) -10.8 (4.9) -0.4 (-2.1 to 1.2) 0.62

Day 10 -12.2 (4.4) -12.2 (4.3) 0.0 (-1.5 to 1.6) 0.98

Day 30 -12.5 (4.4) -13.3 (4.3) 0.8 (-0.7 to 2.3) 0.30

Negative urinary culture at day 10 33 (65%) 63 (77%) 13% (-4 to 29%) 0.12

Re-consultations because of UTI 1 (2%) 26 (32%) -30% (-41 to -19%) <0.0001

Quality of life (scaled to 0-10)

EuroQol health state 9.3 (1.4) 8.5 (2.4) -0.8 (-1.4 to -0.2) 0.011

EuroQol visual analogue scale 7.8 (1.6) 7.0 (2.1) -0.8 (-1.4 to -0.2) 0.015

Patient satisfaction with UTI-management 7.4 (2.3) 4.7 (3.0) -2.7 (-3.6 to -1.8) <0.0001

Number of working days lost due to UTI 0.6 (0.7) 0.7 (0.5) -0.1 (-0.3 to 0.3)* 0.54

* Relative rate reduction calculated from Poisson regression.

Page 45 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 15 / 19

Supplementary Table 6: Number of participants censored until day 10 for time to event curves for definite resolution of symptoms in the NSAID group (Corresponding to Supplementary Figure 2)

Days since

randomisation

No. of patients

censored

Definite resolution Never on antibiotics 0 1

Never on antibiotics 10 3

Use of antibiotics 0 8

Use of antibiotics 10 4

Page 46 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 16 / 19

Supplementary table 7: Baseline characteristics for patients in the NSAID group with and without pyelonephritis

Pyelonephritis

(N=6)

No pyelonephritis

(N=127) P value

Age - yr 34.2 (17.4) 38.0 (14.1) 0.37

Age <45 yr 5 (83%) 89 (70%) 0.67

Symptom duration - days since UTI onset 2.8 (1.7) 3.6 (3.1) 0.51

Symptom duration ≤3 days 5 (83%) 75 (59%) 0.40

Number of UTIs in the last 12 months 0.2 (0.4) 0.6 (1.1) 0.29


Dysuria 4.2 (0.8) 3.2 (1.3) 0.05

Urgency 4.2 (1.0) 3.7 (1.0) 0.31

Night frequency 3.0 (0.6) 2.6 (1.5) 0.49

Day frequency 4.0 (0.6) 3.5 (0.9) 0.13

Lower abdominal pain while urinating 3.7 (1.2) 2.6 (1.5) 0.10

Back or loin pain 1.7 (2.0) 1.1 (1.3) 0.37

Total symptom score 17.1 (4.0) 13.5 (3.8) 0.041

Symptom score ≤20 5 (83%) 122 (96%) 0.25

Blood tests

CRP - mg/l 14.2 (13.5) 6.3 (10.2) 0.05

CRP >10 mg/ml 3 (50%) 21 (17%) 0.07

Leucocytes - 1000/µl 9.8 (2.2) 8.4 (2.4) 0.13

Urinary dipstick

Nitrites - positive 1 (17%) 16 (13%) 0.57

Erythrocytes - 1 to 4+ (median, IQR) 4+ (3+ to 4+) 3+ (2+ to 4+) 0.30

Leucocytes - 1 to 4+ (median, IQR) 4+ (3+ to 4+) 3+ (2+ to 4+) 0.18

Leucocytes >2+ 6 (100%) 95 (75%) 0.33

Urinary culture* 0.19

Negative 0 (0%) 36 (28%)

Positive 6 (100%) 90 (71%)

Escherichia coli 5 (83%) 77 (61%) 0.41

Other Enterobacteriaceae 1 (17%) 8 (6%) 0.35

Staphylococcus saprophyticus 0 (0%) 3 (2%) 1.00

Enterococcus faecalis 0 (0%) 1 (1%) 1.00

Beta-hemolytic Streptococcus group B 0 (0%) 1 (1%) 1.00

Susceptibility to norfloxacin 6 (100%) 86 (68%) 0.18

Enterobacteriaceae 6 (100%) 82 (65%) 0.10

Susceptibility to fosfomycin† 6 (100%) 78 (61%) 0.08

Continuous variables are presented as mean (sd), binary variables as no. of patients (%).

P values from Wilcoxon rank sum tests for continuous and Fisher’s exact tests for binary variables.

* Urinary culture missing in 1 patients in “No pyelonephritis” group.

† Enterobacteriaceae only

Page 47 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 17 / 19

Supplementary Table 8: Symptom score for patients with and without pyelonephritis in the NSAID group.

Pyelonephritis

(N = 6)

No pyelonephritis

(N = 127) P value

mean (sd)

Symptom score at baseline 17.1 (4.0) 13.5 (3.8) 0.041

Symptom score at day 3 9.0 (3.9) 6.2 (4.6) 0.09




P values from Wilcoxon rank sum tests

Page 48 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 18 / 19

Supplementary Figure 1. Boxplot showing development of symptom score up to day 10 after randomisation in the NSAID and antibiotic group. (Boxes show median and 25 / 75 quartiles, whiskers +/- 1.5 * upper/lower IQR, values outside the whisker-range are depicted as points).

Page 49 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 19 / 19

Supplementary Figure 2. KM plot for time until definite resolution of symptoms in the NSAID group. Observation time was set to 1h for patients with resolution at baseline. Patients with missing observation times are censored after 1h. Only definite resolution is considered an event.

Page 50 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 1 / 23

Symptomatic therapy of uncomplicated lower urinary tract 1

infections in the ambulatory setting. A randomized, double blind 2

trial 3

Andreas Kronenberg,1,2,3 Lukas Bütikofer,4,5 Ayodele Odutayo,6 Kathrin 4

Mühlemann,1,2 � Bruno R. da Costa,7 Markus Battaglia,3 Damian N. Meli,3,7 Peter 5

Frey,7 Andreas Limacher, 4,5 Stephan Reichenbach,5,8 Peter Jüni 6,7 6

1 Institute for Infectious Diseases, University of Bern, Bern, Switzerland 7

2 Department of Infectious Diseases, Bern University Hospital, University of Bern, Bern, Switzerland 8

3 Medix General Practice Network, Bern, Switzerland 9

4 CTU Bern, University of Bern, Bern, Switzerland 10

5 Institute of Social and Preventive Medicine (ISPM), University of Bern, Bern, Switzerland 11

6Applied Health Research Centre (AHRC), Li Ka Shing Knowledge Institute of St. Michael’s Hospital, 12

and Department of Medicine, University of Toronto, Canada 13

7 Institute of Primary Health Care (BIHAM), University of Bern, Bern, Switzerland 14

8 Department of Rheumatology, Immunology and Allergology, Inselspital, Bern University Hospital, 15

University of Bern, Bern, Switzerland 16

17

Corresponding address: Andreas Kronenberg, MD 18

Institute for Infectious Diseases 19

University of Bern 20

Friedbühlstrasse 51 21

3010 Bern 22

Tel: +41 (0)31 632 32 65 / Fax: +41 (0)31 632 49 66 23

e-mail: [email protected] 24

25

26

Running title Symptomatic therapy of cystitis 27

Keywords cystitis, urinary tract infections, antibiotic therapy, anti-inflammatory drugs, 28

symptomatic, ambulatory, outpatient, randomized, double-blind 29

Word count abstract 286378 30

Word count text 31894278 31

Registration: www.clinicaltrials.gov (NCT01039545) 32

33

Page 51 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 2 / 23

ABSTRACT 1

Background: Antibiotic resistance is increasing globally and largely stems from increasing 2

antibiotic use. It has been suggested, that symptomatic therapy with non-steroidal anti-3

inflammatory drugs (NSAIDs) might be non-inferior to antibiotics in treatment of 4

uncomplicated lower urinary tract infection (UTI) in women, thus offering an opportunity to 5

reduce antibiotic use in ambulatory care. 6

Methods: In this randomized, double-blind, non-inferiority trial among 17 general practices in 7

Switzerland, women with uncomplicated lower UTI were randomly assigned 1:1 to 8

symptomatic therapy with diclofenac or antibiotic treatment with norfloxacin. The 9

randomization sequence was computer-generated, stratified by practice, blocked and 10

concealed using sealed, sequentially numbered drug containers. Primary and secondary 11

outcomes were symptom resolution up to day 3 and antibiotic use up to day 30, respectively. 12

Analysis was by intention-to-treat. We also performed a meta-analysis combining the results 13

of the current trial with previously published randomized controlled trials comparing NSAIDs 14

and antibiotic treatment in women with UTIs. 15

Findings: We randomly assigned 253 women to NSAIDs (133) and antibiotics (120). 16

Thereof 72 (54%) and 96 women (80%) experienced symptom resolution at day 3 in the 17

NSAID and antibiotic groups, respectively (risk difference 27%, 95%-CI 15% to 38%, p for 18

non-inferiority 0.98, p for superiority <0.0001). The median time until resolution of symptoms 19

was 4 days in the diclofenac group and 2 days in the norfloxacin group. A total of 82 (62%) 20

and 118 women (98%) used antibiotics up to day 30, respectively (risk difference 37%, 95%-21

CI 28 to 46%, p for superiority <0.0001). Six women in the NSAID group (5%) but none in the 22

antibiotic group received clinical diagnosis of a pyelonephritis (p=0.031). CRP levels > 10 23

mg/L at baseline were observed in 21 women in the NSAID group without pyelonephritis 24

(17%) and in 3 women with pyelonephritis (50%), which resulted in a positive likelihood ratio 25

of 3.02 (CI 1.07 to 5.98, p=0.037). Finally, our results were consistent with two previously 26

published clinical trials when combined in a meta-analysis. 27

Page 52 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 3 / 23

Interpretation: Diclofenac is inferior to norfloxacin in terms of symptom relief, but superior in 1

terms of reducing antibiotic use. Deferring antibiotic therapy in women without CRP elevation 2

and short symptom duration could be justifiable, but should be tested in a randomized trial. 3

Funding: Swiss National Foundation, Swiss Academy of Medical Sciences, SwissLife and 4

Else Kroener-Fresenius Foundation. 5

6

Page 53 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 4 / 23

INTRODUCTION 1

Antimicrobial stewardship aims at reducing antibiotic resistance by optimising or 2

decreasing antibiotic use,1 which includes the prevention of antibiotic treatment in cases of 3

viral infections, the tailored prescription of narrow-spectrum antibiotics, shortening the course 4

of therapy, and deferring therapy for low risk bacterial infections. Urinary tract infection (UTI) 5

is one of the most common bacterial infections in adults, affecting considerably more women 6

than men.2 Approximately half of women have at least one UTI in their lifetime, and 20% to 7

30% have two or more.3 Antibiotic prescriptions for UTI accounts for 10-20% of all antibiotic 8

prescriptions in ambulatory care and is second in number only to respiratory tract infections,4, 9

5. Reducing antibiotic prescriptions for UTI could potentially decrease the risk of antimicrobial 10

resistance. Therefore, the benefit of antibiotic treatment needs to be weigthed against the 11

potential for adverse effects, both at the level of the individual (adverse drug reactions) and 12

population (as a driver of antibiotic resistance). 13

Symptoms of UTI may arise from local increases in pro-inflammatory factors like 14

prostaglandins and non-steroidal anti-inflammatory drugs (NSAIDs) may be useful in 15

alleviating symptoms.6-8 A small randomized pilot trial, which compared the NSAID ibuprofen 16

with the antibiotic ciprofloxacin in 80 women with uncomplicated lower UTI, concluded that 17

symptomatic therapy with NSAIDs may be non-inferior to antibiotics, but suggested 18

confirmation in a larger trial.9 Two adequately powered randomized double-blind trials were 19

therefore initiated simultaneously in February 2012 in Germany and Switzerland. Results of 20

the German trial, which compared ibuprofen with fosfomycin in 494 women, were recently 21

published.10 Here, we report results of the Swiss trial, which compared diclofenac with 22

norfloxacin in 253 women with uncomplicated lower UTI. To put our results into context, we 23

also performed a meta-analysis of all available trials comparing NSAIDs with antibiotics in 24

women with lower UTI. 25

26

Page 54 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 5 / 23

METHODS 1

Study design 2

This randomized controlled patient and assessor blind trial was conducted in 17 general 3

practices in Switzerland. The funding bodies were not involved in the design or the conduct 4

of the trial, nor in the writing of the manuscript or the decision to submit the manuscript for 5

publication. The first and senior author wrote the manuscript, had final responsibility for the 6

decision to submit the manuscript for publication and vouch for the accuracy of the data and 7

analyses and for the fidelity of the trial to the protocol. The funding bodies only had access to 8

the data after finalization of the statistical analysis plan and completion of all analyses. The 9

study was conducted in accordance with the Declaration of Helsinki and was approved by 10

the local research ethics committee and the Swiss Agency for Therapeutic Products 11

Swissmedic. The trial was registered before start of patient recruitment at clinicaltrials.gov 12

(NCT01039545). All women provided written informed consent. 13

14

15

Patients, Randomization and Treatment 16

Women aged 18 to 70, who visited their general practitioners (GPs) because of one 17

or more symptoms or signs typical of acute lower UTI (dysuria, frequency, macrohaematuria, 18

cloudy or smelly urine) or self-diagnosed cystitis were eligible if their urine dipstick was 19

positive for nitrite and/or leucocytes. Women with relevant comorbidities or pregnancy, signs 20

of invasiveness, recurrent UTI, recent antibiotic therapy within the previous 4 weeks or 21

prolonged duration of symptoms (>7 days) were ineligible (for details see Supplementary 22

Appendix).We excluded pregnant women and women with clinical signs of upper UTI such 23

as fever (axillary body temperature >38oC), costovertebral pain or tenderness, rigors, nausea 24

or vomiting. We also excluded women with known or suspected anatomical or functional 25

abnormality of the urinary tract and comorbidities such as diabetes mellitus, active gastric or 26

duodenal ulcer disease or gastrointestinal bleeding, inflammatory bowel disease, severe liver 27

dysfunction (liver cirrhosis and ascites), coagulopathy (including therapy with coumarine 28

Formatted: Normal, Indent: First line: 0",Line spacing: single

Page 55 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 6 / 23

derivates), renal insufficiency grade III or higher (calculated GFR <60 ml/min), known 1

congestive heart failure (NYHA III or higher), psychiatric illness or dementia, inability to to 2

communicate in German or French language and any other serious comorbidity as judged by 3

the treating physician. In addition we excluded women with documented immunosuppression 4

(e.g. prednisone equivalent >10mg per day for >14 days, chemotherapy, radiotherapy, 5

immunmodulators, HIV infection, neutropenia) or hypersensitivity to one of the study 6

medications or history of asthma, urticaria or hypersensitivity-like reactions after consumption 7

of salicylic acid or other non-steroidal anti-inflammatory drugs, as well as women with vaginal 8

symptoms (discharge, irritation), bladder catheter in situ or during the past 30 days, recurrent 9

urinary tract infection (more than 3 infections during the last 12 months), antibiotic treatment 10

during the last 4 weeks or duration of UTI symptoms for more than 7 days before physician’s 11

visit. 12

Women were randomly allocated in a 1:1 ratio to NSAID or antibiotic therapy. The 13

randomization sequence was computer-generated, stratified by practice and blocked with 14

randomly varying block sizes of 4 and 6. Allocation was concealed with sealed, sequentially 15

numbered opaque drug containers of identical appearance that contained opaque hard 16

gelatine capsules of identical size and colour. Women allocated to NSAIDs received 17

capsules containing 75 mg diclofenac retard for three days (Olfen-75 duo release®, Mepha 18

Pharma, Basel, Switzerland),) and women allocated to antibiotics received capsules 19

containing 400 mg norfloxacin for three days (Norfloxacin-Teva® 400mg, Teva Pharma, Tel 20

Aviv, Israel; Norflocin-Mepha® 400mg Lactab, Mepha Pharma, Basel, Switzerland from 21

October 2013 onwards due to delivery restrictions). WomenWe chose norfloxacin because of 22

the high susceptibility rates in Switzerland and diclofenac because of its identical frequency 23

of administration which facilitated patient blinding. Women started therapy immediately after 24

randomization on day 0 and were advised to take 2 capsules per day: one in the morning 25

and one in the evening. All women were given a single open-label package of fosfomycin 26

(Monuril® 3g, Zambon, Cadempino, Switzerland) to be taken as a rescue antibiotic after 27

completion of the study medication on day 3 at their discretion, if symptoms persisted. 28

Page 56 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 7 / 23

Patient involvement 1

No patients were involved in setting the research question or the outcome measures, nor 2

were they involved in developing plans for recruitment, design, or implementation of the 3

study. No patients were asked to advise on interpretation or writing up of results. We plan to 4

disseminate the results of the research to all the scientific community, including study 5

participants. 6

7

Procedures and outcomes 8

The pre-specified primary outcome was symptom resolution up to day 3. (12 hours 9

after intake of the last study medication). In the absence of antimicrobial resistance, the 10

mean duration of symptoms in women with uncomplicated UTIs treated with antibiotics is 3 11

days11. Women rated the severity of five UTI symptoms (dysuria; frequency; urgency; 12

abdominal pain when passing urine; pain or tenderness in the lower back or loin) daily from 13

days 0 (randomization) to 10 by self-report diary and on day 30 by telephone interview on a 14

Likert scale from 0 to 6, with their composite score ranging from 0 to 30; symptom resolution 15

was defined as ≤ 2 points ("slight problems”) on all 5 components.12-14 Complete absence of 16

symptoms was defined as 0 points on all components. The pre-specified principal secondary 17

outcome was the use of any antibiotic (including norfloxacin and fosfomycin as trial 18

medications) up to day 30. The remaining pre-specified outcomes are reported in the 19

Supplementary Appendix. Side effects, quality of life, working days lost and use of the rescue 20

antibiotic were assessed on day 3, general satisfaction on day 10. Serious adverse events, 21

adverse events ≥grade 3 severity, any additional unplanned medical visits, including 22

telephone contacts, and co-medications were assessed by telephone interviews of the 23

treating physician on day 10 and of an independent, blinded interviewer on day 30. The 24

clinical diagnosis of pyelonephritis required the occurrence of loin pain and fever., leading to 25

an unplanned ambulatory visit. Mid-stream urinary samples obtained on days 0 and 10 were 26

processed according to standard laboratory procedures, using a cut-off of >≥103 colony 27

forming units per ml for a urinary culture to be considered positive;1415 mixed flora with no 28

Formatted: English (U.S.)

Page 57 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 8 / 23

predominant microorganism, Lactobacilli or Streptococcus viridans group were considered 1

negative. Details of baseline assessments, diary and scores used are described in the 2

Supplemental Appendix. 3

4

Systematic Review 5

We conducted a systematic search of MEDLINE (inception to April 9th 2017) using a 6

combination MeSH and keyword searching for the following terms: “Anti-bacterial Agents”, 7

“Antibiotics”, “Non-Steroidal Anti-inflammatory Agents”, “Urinary Tract Infections”, and 8

“Cystitis”. We included randomized clinical trials which compared NSAIDs and antibiotics for 9

the treatment of UTIs. No language restrictions were applied. We extracted information on 10

event rates in the NSAID and antibiotic treatment groups for the following outcomes: 11

symptom resolution at day 3 or 4, symptom resolution at 7 days, recurrent UTI, 12

pyelonephritis and use of antibiotics. 13

14

15

Statistical methods 16

We originally planned to recruit 400 women, but recruitment was slow and financial 17

constraints led us to decide in June 2014 to stop patient recruitment by December 2014, 18

when an expected 260 women would be included. The decision was made without inspecting 19

the data and after repeating the power analysis based on a closed-form normal 20

approximation test of proportions, 16 which was less conservative than the simulation-based 21

approach originally used. With the original assumption of 70% of women reaching symptom 22

resolution up to day 3 in both groups and the original, pre-specified non-inferiority margin of 23

15% on a risk difference scale, the projected sample size of 260 women would yield a power 24

of 84% to detect non-inferiority at a one-sided type I error of 5%. 25

The primary outcome was evaluated using a risk difference with a corresponding two-sided 26

95% confidence interval, a one-sided normal approximation test for non-inferiority and a two-27

Formatted: English (U.S.)

Formatted: Indent: First line: 0", Linespacing: Double

Page 58 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 9 / 23

sided Chi-squared test for superiority. Secondary outcomes were compared using 1

conventional two-sided p-values for superiority and corresponding two-sided 95% confidence 2

intervals. We used risk differences with Chi-squared tests for binary data, Poisson regression 3

with robust standard errors for counts, and linear regression with robust standard errors for 4

continuous data. Kaplan-Meier curves accompanied by hazard ratios from Cox models were 5

used to analyse time to symptom resolution and time to antibiotic use. All women were 6

included in the analysis in the groups they were originally allocated to (intention-to-treat 7

analysis), missing values were accounted for by multiple imputation (see Supplementary 8

Appendix). and Supplementary Table 1). We performed pre-specified subgroup analyses of 9

the primary outcome accompanied by Mantel-Haenszel tests for interaction by age (< 45 vs ≥ 10

45 years), symptom severity at baseline (≤ 20 vs > 20), symptom duration (≤ 3 vs > 3 days), 11

and presence of a positive urinary culture at baseline; post-hoc subgroup analyses were 12

performed by urine leucocytes (≤ 2+ vs >2+) and the presence of norfloxacin-resistant 13

Enterbacteriaceae. In per-protocol analyses, we excluded women with protocol deviations 14

who were defined as women with no documented intake of at least one dose of study 15

medication, cross-overs, or women who used rescue antibiotics before day 3. All analyses 16

were pre-specified in a statistical analysis plan before recruitment end and inspection of the 17

data. In post-hoc analyses of women allocated to NSAIDs, we compared baseline 18

characteristics and outcomes between those who used antibiotics until day 30 and those 19

never on antibiotics, determined the time between symptom onset and diagnosis in those 20

clinically diagnosed with pyelonephritis, and explored whether blood or urine findings at 21

baseline were associated with pyelonephritis with clinically relevant positive or negative 22

likelihood ratios above 5 or below 0.2, respectively, which could be used to rule in or rule out 23

future clinical diagnosis of pyelonephritis. 24

In view of the expected low number of trials and the considerable difference in sample size 25

between the initially available pilot trial and subsequent trials, we refrained from performing 26

conventional frequentist random-effects meta-analysis because these methods may yield 27

either overly optimistic or overly conservative results.17-19 Therefore, trial specific risk 28

Page 59 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 10 / 23

differences were combined with a Mantel-Haenszel fixed-effect model and a Bayesian 1

random-effects model. We chose the risk difference as treatment effect estimate as odds-2

ratios were not directly interpretable because some control group event rates were above 3

20%, and risk ratios were heterogeneous because of the variation in control group event 4

rates between trials. The specifications of the Bayesian random-effects model are described 5

in the web-appendix. We conducted two analyses. The first combined all trials and the 6

second was restricted to large trials with more than 100 patients per group; trials which would 7

have more than 50% power to detect a 15% difference in the primary outcome of incomplete 8

or no symptom resolution at day 3 or 4. All analyses were done in 9

Stata Release 14 (StataCorp. 2014. Stata Statistical Software: Release 14. College Station, 10

TX: StataCorp LP), RStudio version 1.0.143 (RStudio: Integrated Development for R. 11

RStudio, Inc., Boston, MA URL http://www.rstudio.com/), and WinBUGS version 1.4.3. 12

13

14

15

16

Formatted: Indent: First line: 0.49"

Page 60 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 11 / 23

RESULTS 1

Between Feb 7, 2012 and Dec 3, 2014, 253 women were included in the trial, 133. 2

Thirty-six patients were recruited in seven single practices, 108 in nine group practices and 3

109 in one big medical centre with 17 physicians in the centre of Bern, Switzerland. One-4

hundred and thirty-three patients were randomly allocated to NSAIDs and 120 to antibiotics. 5

A total of 125 women (94%) received treatment as allocated in the NSAID group, and 118 6

women (98%) in the antibiotic group. 119 (89%) and 112 women (93%) had complete follow-7

up until day 30 (Figure 1). Groups were similar (Table 1), with a mean age of 36.8 years (SD 8

14.1), a mean duration of symptoms of 3.4 days (SD 2.6) and a mean composite symptom 9

score of 13.7 points out of 30 (SD 3.8), and urgency, frequency and dysuria as the most 10

prevalent symptoms. Thirty-five urine samples were nitrite-positive (14%) and 191 had >2+ 11

leucocytes (75%). For 185 urine samples, we found positive urinary cultures (73%), which 12

resulted in a total of 193 isolates, of which 187 isolates (97%) had documented norfloxacin 13

susceptibility. 173 isolates contained Enterobacteriaceae, 160 of which were tested for 14

fosfomycin susceptibility and 158 were found susceptible (99%). 15

Clinical outcomes are presented in Table 2. The primary outcome, resolution of 16

symptoms at day 3, was observed in 72 (54%) and 96 (80%) women in the NSAID and 17

antibiotic groups, respectively (risk difference 27%, 95%-CI 15 to 38%, one-sided p for non-18

inferiority=0.98, two-sided p for superiority in favour of antibiotic group <0.0001). The 19

principal secondary outcome, use of any antibiotic up to day 30, was observed in 82 (62%) 20

and 118 (98%) women, respectively (risk difference -37%, 95%-CI -46 to -28%, p for 21

superiority in favour of NSAID group <0.0001). Among the 82 women in the NSAID group 22

who used antibiotics, 58 (71%) decided to take antibiotics during the first three days and 55 23

of the 58 (95%) took the rescue antibiotic fosfomycin. 24

Figure 2 presents time to event curves for resolution of symptoms (Panel A) and use 25

of antibiotics (Panel B) until day 10. (Figure 2 and Supplementary Table 2). The course of 26

mean symptom scores is shown in Supplementary Figure 1. The median time until resolution 27

Page 61 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 12 / 23

was 4 days in the NSAID vs.versus 2 days in the antibiotic group (HR 1.64, CI 1.26-2.14, 1

p<0.0010001,). The median times until antibiotic use were 05 and 50 days, respectively (HR 2

10.06, CI 6.67 to 15.17, p<0.0010001). Figure 3 shows subgroup analyses for the primary 3

and the principal secondary outcome. Results appeared consistent across all subgroups. 4

Sensitivity analyses revealed consistent results for the primary and main secondary outcome 5

(Supplementary Table 13). 6

Remaining pre-specified outcomes favoured antibiotic therapy except for change of 7

symptom score on day 30 (p=0.88) and working days lost (p=0.18, Table 2). Post-hoc 8

analyses of additional time-points revealed little evidence of a difference between groups for 9

resolution or complete absence of symptoms on day 10 and 30 or change of symptom score 10

on day 10 (p≥0.07) and strong evidence for a difference in favour of the NSAID group for 11

antibiotic use up to day 3 (p<0.0001, Table 2). 12

Supplementary Table 24 shows a post-hoc comparison of baseline characteristics of 13

women who had never been on antibiotics (n=51) with women who had used any antibiotics 14

until day 30 (n=82) among those randomly allocated to NSAIDs. We found little evidence for 15

a difference between groups, except for one component of the symptom composite score. 16

Supplementary Table 35 presents a comparison of outcomes. Resolution and complete 17

absence of symptoms at day 3 were more frequent among women never on antibiotics 18

(p≤0.006), changes of symptom scores were more pronounced at day 3 (p=0.0001), 19

reconsultations were less frequent (p<0.0001), and scores on quality of life and satisfaction 20

with care were higher (p≤0.015). The median time until symptom resolution in women never 21

on antibiotics was 3 days and the median time in women who had used antibiotics was 5 22

days (HR 0.61, CI 0.42 to 0.88, p=0.003, Supplementary Figure 2 and Supplementary Table 23

6). Thirty-four urinary cultures had been positive at baseline among women never on 24

antibiotics, of these 16 had become spontaneously negative on day 10 (47%). 25

Table 3 presents adverse events that resulted in reconsultations: 43 events in 41 26

women in the NSAID group (31%), and 22 events 21 women in the antibiotic group (18%). 27

Adverse events related to UTI were more frequent in the NSAID-group (p=0.012), with 6 28

Page 62 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 13 / 23

cases of clinically diagnosed pyelonephritis in the NSAID group (5%) and none in the 1

antibiotic group (p=0.031) and one woman with clinically diagnosed pyelonephritis in the 2

NSAID group classified to have experienced a serious adverse event since she was 3

hospitalized to receive intravenous antibiotic therapy. The median time from symptom onset 4

to clinical diagnosis of pyelonephritis was 5.5 days (range 5 to 8 days). CRP levels > 10 mg/L 5

at baseline were observed in 21 women in the NSAID group without pyelonephritis (17%) 6

and in 3 women with pyelonephritis (50%), which%, Supplementary Table 7). This resulted in 7

a positive likelihood ratio of 3.02 (CI 1.07 to 5.98, p=0.037) for CRP levels > 10 mg/L. None 8

of the blood or urine findings at baseline were associated with clinically relevant positive or 9

negative likelihood ratios suitable to rule in or out future diagnosis of a pyelonephritis. 10

Although the symptom score for women who developed pyelonephritis was higher at 11

baseline compared to women who did not develop pyelonephritis, there were no differences 12

in symptom score from day 3 to 30 (Supplementary Table 8). 13

14

Systematic Review and Meta-analysis 15

We reviewed 81 studies and excluded 78 in the abstract screen. Of the three articles 16

undergoing full-text review, one article was a published study protocol for an ongoing clinical 17

trial that may be indeed by eligible for inclusion upon completion. Therefore, three trials, 18

including our own, were included in the meta-analysis. Together, the studies included 806 19

women (410 randomized to NSAIDs and 382 randomized to antibiotics). Compared to the 20

antibiotics group, women randomized to NSAIDs had a higher incidence of incomplete or no 21

symptom resolution at 3 to 4 days (risk difference 18.0%, CI 11.3% to 24.7%; Number 22

Needed to Harm (NNH) 6, CI 4 to 9, Figure 4) and at 7 days (risk difference 9.6%, CI 4.1% to 23

15.1%, NNH 10, CI 7 to 24). The incidence of pyelonephritis was also higher in women 24

randomized to NSAIDs compared to women randomized to antibiotics (risk difference 2.6%, 25

CI 0.8% to 4.5%, NNH 38, CI 22 to 126). However, antibiotic use was markedly lower in 26

women randomized to NSAIDs compared to women randomized to antibiotics (risk difference 27

-56.1 %, CI -60.8% to -51.4%, Number Needed to Treat 2, CI 2 to 2) and there was little 28

Page 63 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 14 / 23

evidence of a difference in the frequency of recurrent UTIs between the groups (Figure 4). 1

Notably, point estimates for the fixed effect meta-analysis and the Bayesian random-effects 2

meta-analysis were similar, although confidence intervals were wider for the random effects 3

model (Figure 5). However, limiting our analysis to large trials increased the precision of the 4

effect estimate. There was little heterogeneity for all outcomes except for use of antibiotics, 5

for which the risk difference was lower in our study compared to the previous trials (Figure 5). 6

7

8

Page 64 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 15 / 23

DISCUSSION 1

In this randomized, double-blind trial in women with uncomplicated lower urinary tract 2

infection, symptomatic treatment with the NSAID diclofenac was inferior to antibiotic 3

treatment with norfloxacin in controlling symptoms. Those treated with diclofenac were 27% 4

less likely to have symptom resolution at day 3 and 12% less likely to have symptom 5

resolution at day 7 after randomization, with higher mean symptom scores, more frequent 6

reconsultations, a higher incidence of clinically diagnosed pyelonephritis and lower patient 7

satisfaction than those in the NSAID group. Conversely, women who received NSAIDs were 8

37% less likely to receive antibiotics until day 30 after randomization. 9

A meta-analysis of five trials concluded that antibiotics are clinically superior to 10

placebo in women with uncomplicated lower UTI.20 Our trial, in conjunction with the 11

accompanying meta-analysis, suggests that antibiotics are also clinically superior to 12

symptomatic treatment with NSAIDs. This contrasts with the conclusions of a small pilot trial 13

by Bleidorn et al,9 which found clinical outcomes of treatment with ibuprofen similar to those 14

of antibiotic therapy with ciprofloxacin. The pilot trial triggered both, our trial and the recently 15

published trial by Gágyor et al.10 Both of these trials were adequately powered, but failed to 16

detect non-inferiority of NSAIDs as compared with antibiotics in terms of symptom control. 17

Importantly, our trial and the accompanying meta-analysis suggest that symptomatic 18

treatment with NSAIDs is associated with an average 2.6% increase in the absolute risk of 19

clinically diagnosed pyelonephritis, which translates into a NNH of 38 as compared with 20

antibiotic treatment. The risk of pyelonephritis with NSAIDs is comparable to the risk of 0.4 to 21

2.6% observed with placebo in two earlier trials,21, 22 despite the previously described 22

antibacterial activity of diclofenac23 and ibuprofen.24 23

On the beneficial side, our meta-analysis suggests that antibiotic use can be halved 24

on average by initial symptomatic treatment with NSAIDs, with a corresponding number 25

needed to treat of 2 to prevent one instant of antibiotic use. GivenAlthough there was 26

moderate heterogeneity for this outcome, it may relate to methodological differences in the 27

Page 65 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 16 / 23

conduct of our trial compared to prior trials. In particular, we provided women with a rescue 1

antibiotic for discretionary use after completion of study medication and reaching the primary 2

endpoint. This may have facilitated antibiotic use in the NSAID treatment arm of our trial. 3

Nonetheless, given the high global incidence of UTIs, the reduction in antibiotic use is highly 4

relevant and is likely to immediately decrease resistance rates for E. colicoli and even other 5

microorganisms in the affected population.25 6

Rising antibiotic resistance among uropathogens in general, and Escherichia coli (the 7

most common uropathogen), in particular, is a global concern. Many studies show a clear 8

correlation between antibiotic consumption and rising resistance rates. Accordingly, 9

antibiotics are often withheld in cases of self-limited, benign bacterial diseases as acute otitis 10

media, sinusitis and traveller’s diarrhoea at the cost of a prolongation of symptoms by 11

typically 1-3 days.26-33 Our results in women with uncomplicated lower UTI are well in line 12

with the prolongation of symptoms observed with symptomatic treatment of these conditions. 13

As many women in the NSAID group resorted to antibiotic therapy in our trial, a strategy of 14

selectively deferring rather than completely withholding antibiotic therapy may be preferable 15

for uncomplicated lower UTI.34 This can be achieved through a shared decision making 16

process, during which clinicians inquire about their patient’s ideas and expectations 17

regarding antibiotic treatment for uncomplicated UTI and also explore the option of delaying 18

antibiotic use as a treatment strategy. 19

Subgroup analyses did not provide evidence for any clinically relevant treatment by 20

subgroup interactions. In particular, we were unablein contrast to find the interaction with 21

urine culture suggested by GagyorGágyor et al.10 , reduction in antibiotic prescription was 22

comparable in women with and without positive urinary cultures. Testing initial urine samples 23

for other biomarkers associated with UTI, like heparin-binding protein, interleukin-6, acetic 24

acid, trimethylamine, xanthine oxidase, myeloperoxidase, or others,35-37 might have resulted 25

in promising treatment by subgroup interactions, but these tests are not yet established in 26

clinical practice, and we are unaware of any evidence to suggest that such interactions would 27

be likely. When comparing women who never had been on antibiotics with women who had 28

Page 66 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 17 / 23

used antibiotics until day 30 among those allocated to NSAIDs in post-hoc analyses, we 1

unsurprisingly found better clinical outcomes in those never on antibiotics, but no relevant 2

differences in baseline characteristics that would allow an early identification of women likely 3

to benefit from NSAIDs alone. However, in further post-hoc analyses, which are purely 4

hypothesis generating, we found that the clinical diagnosis of pyelonephritis was established 5

not earlier than 5 days after symptom onset, and that CRP values above 10 mg/L were more 6

frequent at baseline in women who subsequently were diagnosed with a pyelonephritis. 7

Taken together, these exploratory findings could support a tailored strategy of immediate 8

antibiotic use in women with CRP levels of >10 mg/L and early NSAID use in remaining 9

women for up to 3 to 4 days after symptom onset, followed by deferred, selective antibiotic 10

use in those women who did not show a clear improvement up by then. Naturally, such a 11

tailored strategy would need to be evaluated in an appropriately powered randomized trial. 12

Our trial should be interpreted in view of its strengths and limitations. Strengths are its 13

randomized double blind design with appropriate concealment of allocation, blinding of 14

patients, therapists and outcome assessors, the low loss to follow-up, the robustness of 15

results in a series of sensitivity analyses and the multi-centric primary care setting. The 16

premature termination of patient recruitment before reaching the initially planned sample size 17

is an obvious limitation. However, the decision to stop recruitment was made without 18

inspecting the data and is therefore unlikely to have biased our findings.38 Despite the 19

smaller than originally planned sample size, our results are completely unequivocal. While 20

using slightly different symptom scores in different trials does affect the comparability 21

between trials, this will not affect the validity of our results. Nevertheless, we suggest that 22

future trials should consider to use the symptom score developed by Alidjanov et al.39 Finally, 23

our results are not generalizable to countries and clinical settings with lower rates of 24

susceptibility, which could decrease the effectiveness of antibiotics and render symptomatic 25

treatment with NSAIDs less inferior. 26

In conclusion, symptomatic therapy is inferior to antibiotic therapy for the treatment of 27

women with uncomplicated lower UTI in an ambulatory setting, as it increases median 28

Page 67 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 18 / 23

symptom duration by two days and is likely to be associated with an increased risk of 1

clinically diagnosed pyelonephritis. The observed impressive reduction in antibiotic use, 2

which would likely contribute directly to decreasing resistance rates for E. coli in the affected 3

population, suggests that alternate approaches of combining symptomatic treatment with 4

deferred, selective antibiotic use should be developed and tested in future trials. 5

6

Page 68 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 19 / 23

Contributors 1

AK and KM were responsible for conception and design and obtained funding for 2

the study. AK, MB, DM, and PF were responsible for the acquisition of data. BC, AO 3

and PJ performed the meta-analysis. LB, BC, AL, SR and PJ did the analysis and 4

interpreted it in collaboration with all the remaining authors. AK and PJ had full 5

access to the final data, co-wrote the manuscript, had final responsibility for content, 6

and the decision to submit for publication. All authors critically revised the paper for 7

important intellectual content and approved the final version. 8

9

Transparency declarations 10

The authors affirm that the manuscript is an honest, accurate, and transparent 11

account of the study being reported; that no important aspects of the study have 12

been omitted; and that any discrepancies from the study as planned (and, if relevant, 13

registered) have been explained. 14

15

Competing interests: All authors declare: no support from any organisation for the 16

submitted work besides the acknowledged financial support. AK has received travel 17

grant and meeting expenses from Gilead, Viofor and the World Health Organisation 18

(WHO), is advisor of the Swiss Federal Office of Public Health concerning antibiotic 19

resistance epidemiology in Switzerland, and provides non-interpreted annual 20

resistance data to LEO pharmaceutic company and the Swiss government. PJ has 21

received research grants to the institution from Astra Zeneca, Biotronik, Biosensors 22

International, Eli Lilly and The Medicines Company for cardiovascular trials, and 23

serves as unpaid member of the steering group of cardiovascular trials funded by 24

Astra Zeneca, Biotronik, Biosensors, St. Jude Medical and The Medicines Company. 25

The remaining authors declare no financial relationships with any organisations that 26

might have an interest in the submitted work in the previous three years and no other 27

relationships or activities that could appear to have influenced the submitted work. 28

29

Page 69 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 20 / 23

Acknowledgments 1

We thank all patients participating in this study, all participating general physicians for 2

their meticulous data collection and the Clinical Trials Unit of the University of Bern 3

(CTU Bern) for the development of the database (Malcolm Sturdy), Monitoring (Lucia 4

Kacina) and telephone interview and data clearance (Madeleine Dähler). We thank 5

the Institute of General Practice BIHAM, for their organisational support and the 6

Hospital Pharmacy of the Inselspital for the production and blinding of the study 7

medications (Marco Eschenmoser). 8

9

Participating physicians. Peter Duner, Eggiwil; Christoph Fry, Belp; Ursula Grob, 10

Herzogenbuchsee; Felix Huber, Zürich; Beat Köstner-Mösching, Neuenegg; Andreas 11

Kronenberg, Bern; Corinna Kronenberg, Stettlen; Danielle Lemann, Langnau; 12

Damian Meli, Huttwil; Gabriele Reinheimer, Worb; Véronique Rigamonti Wermelinger, 13

Bern; Ralf Schäfer, Bern; Urs Schneeberger, Niederönz; Christian Studer, Luzern; 14

Fritz Weber, Buchs; Doris Zundel, Bätterkinden; Anne-Marie Zundel Funk, 15

Zollikofen. 16

17

Funding 18

This study was supported financially by the Swiss National Foundation (SNF project 19

32003B_130867), the Swiss Academy of Medical Sciences, the SwissLife foundation 20

and the Else Kroener-Fresenius foundation. 21

22

23

Figures and Tables 24

See separate file 25

26 Formatted: Heading 2, Space Before: 22 pt,Line spacing: Multiple 1.15 li

Page 70 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 21 / 23

1. References 1

1. Organization WH. Global Action Plan on Antimicrobial Resistance. WHO Library 2

Cataloguing-in-Publication Data; 2015. 3

2. Foxman B. Epidemiology of urinary tract infections: incidence, morbidity, and 4

economic costs. Am J Med. 2002; 113 Suppl 1A: 5S-13S. 5

3. Foxman B, Brown P. Epidemiology of urinary tract infections: transmission and risk 6

factors, incidence, and costs. Infect Dis Clin North Am. 2003; 17(2): 227-41. 7

4. Mühlemann K. Surveillance of antibiotic prescription in the outpatient setting using the 8

national Sentinel network (OP-159, http://www.smw.ch/docs/PdfContent/smw-12931.pdf). 9

WONCA World Organization of National Colleges, Academies and Academic Associations of 10

General Practitioners/Family Physicians; 2009; Basel. Switzerland; 2009. 11

5. Lundborg CS, Olsson E, Molstad S. Antibiotic prescribing in outpatients: a 1-week 12

diagnosis-prescribing study in 5 counties in Sweden. Scand J Infect Dis. 2002; 34(6): 442-8. 13

6. Farkas A, Alajem D, Dekel S, Binderman I. Urinary prostaglandin E2 in acute 14

bacterial cystitis. J Urol. 1980; 124(4): 455-7. 15

7. Schwarz NT, Jung SY, Kalff JC, Chancellor M, Bauer AJ. Bacterial toxin N-formyl-16

methionyl-leucyl-phenylalanine acutely contracts human and rabbit detrusor through the 17

release of eicosanoids. J Urol. 2002; 167(6): 2603-12. 18

8. Poljakovic M, Svensson ML, Svanborg C, Johansson K, Larsson B, Persson K. 19

Escherichia coli-induced inducible nitric oxide synthase and cyclooxygenase expression in 20

the mouse bladder and kidney. Kidney Int. 2001; 59(3): 893-904. 21

9. Bleidorn J, Gagyor I, Kochen MM, Wegscheider K, Hummers-Pradier E. Symptomatic 22

treatment (ibuprofen) or antibiotics (ciprofloxacin) for uncomplicated urinary tract infection?--23

results of a randomized controlled pilot trial. BMC Med. 8: 30. 24

10. Gagyor I, Bleidorn J, Kochen MM, Schmiemann G, Wegscheider K, Hummers-Pradier 25

E. Ibuprofen versus fosfomycin for uncomplicated urinary tract infection in women: 26

randomised controlled trial. BMJ. 2015; 351: h6544. 27

11. Little P, Merriman R, Turner S, Rumsby K, Warner G, Lowes JA, et al. Presentation, 28

pattern, and natural course of severe symptoms, and role of antibiotics and antibiotic 29

resistance among patients presenting with suspected uncomplicated urinary tract infection in 30

primary care: observational study. BMJ. 2010; 340: b5633. 31

12. Gupta K, Hooton TM, Roberts PL, Stamm WE. Short-course nitrofurantoin for the 32

treatment of acute uncomplicated cystitis in women. Arch Intern Med. 2007; 167(20): 2207-33

12. 34

13. McNulty CA, Richards J, Livermore DM, Little P, Charlett A, Freeman E, et al. Clinical 35

relevance of laboratory-reported antibiotic resistance in acute uncomplicated urinary tract 36

infection in primary care. J Antimicrob Chemother. 2006; 58(5): 1000-8. 37

14. Clayson D, Wild D, Doll H, Keating K, Gondek K. Validation of a patient-administered 38

questionnaire to measure the severity and bothersomeness of lower urinary tract symptoms 39

in uncomplicated urinary tract infection (UTI): the UTI Symptom Assessment questionnaire. 40

BJU Int. 2005; 96(3): 350-9. 41

15. Grabe M (chairman) B-JT, Botto H, Çek M, Naber KG, Pickard RS, Tenke P, 42

Wagenlehner F, Wullt B. Guidelines on Urological Infections. 2013 [cited 2015 9.7.2015]; 43

Available from: http://uroweb.org/wp-content/uploads/18_Urological-infections_LR.pdf 44

16. Chow SC SJ, Wang H Sample Size Calculations in Clinical Research. Second Edition 45

ed: Chapman & Hall/CRC; 2008. 46

17. Chung Y, Rabe-Hesketh S, Choi IH. Avoiding zero between-study variance estimates 47

in random-effects meta-analysis. Stat Med. 2013; 32(23): 4071-89. 48

18. Friede T, Rover C, Wandel S, Neuenschwander B. Meta-analysis of few small studies 49

in orphan diseases. Res Synth Methods. 2017; 8(1): 79-91. 50

19. Higgins JP, Thompson SG, Spiegelhalter DJ. A re-evaluation of random-effects meta-51

analysis. J R Stat Soc Ser A Stat Soc. 2009; 172(1): 137-59. 52

Formatted: Font: 16 pt, Font color: Auto

Page 71 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 22 / 23

20. Falagas ME, Kotsantis IK, Vouloumanou EK, Rafailidis PI. Antibiotics versus placebo 1

in the treatment of women with uncomplicated cystitis: a meta-analysis of randomized 2

controlled trials. J Infect. 2009; 58(2): 91-102. 3

21. Ferry SA, Holm SE, Stenlund H, Lundholm R, Monsen TJ. Clinical and bacteriological 4

outcome of different doses and duration of pivmecillinam compared with placebo therapy of 5

uncomplicated lower urinary tract infection in women: the LUTIW project. Scand J Prim 6

Health Care. 2007; 25(1): 49-57. 7

22. Christiaens TC, De Meyere M, Verschraegen G, Peersman W, Heytens S, De 8

Maeseneer JM. Randomised controlled trial of nitrofurantoin versus placebo in the treatment 9

of uncomplicated urinary tract infection in adult women. The British journal of general 10

practice : the journal of the Royal College of General Practitioners. 2002; 52(482): 729-34. 11

23. Mazumdar K, Dutta NK, Dastidar SG, Motohashi N, Shirataki Y. Diclofenac in the 12

management of E. coli urinary tract infections. In Vivo. 2006; 20(5): 613-9. 13

24. Obad J, Suskovic J, Kos B. Antimicrobial activity of ibuprofen: new perspectives on 14

an "Old" non-antibiotic drug. Eur J Pharm Sci. 2015; 71: 93-8. 15

25. Gottesman BS, Carmeli Y, Shitrit P, Chowers M. Impact of quinolone restriction on 16

resistance patterns of Escherichia coli isolated from urine by culture in a community setting. 17

Clinical infectious diseases : an official publication of the Infectious Diseases Society of 18

America. 2009; 49(6): 869-75. 19

26. Glasziou PP, Del Mar CB, Sanders SL, Hayem M. Antibiotics for acute otitis media in 20

children. Cochrane Database Syst Rev. 2004; (1): CD000219. 21

27. Falagas ME, Giannopoulou KP, Vardakas KZ, Dimopoulos G, Karageorgopoulos DE. 22

Comparison of antibiotics with placebo for treatment of acute sinusitis: a meta-analysis of 23

randomised controlled trials. Lancet Infect Dis. 2008; 8(9): 543-52. 24

28. Ahovuo-Saloranta A, Borisenko OV, Kovanen N, Varonen H, Rautakorpi UM, 25

Williams JW, Jr., et al. Antibiotics for acute maxillary sinusitis. Cochrane Database Syst Rev. 26

2008; (2): CD000243. 27

29. Del Mar CB, Glasziou PP, Spinks AB. Antibiotics for sore throat. Cochrane Database 28

Syst Rev. 2006; (4): CD000023. 29

30. de Bruyn GH, S. Borwick, A. Antibiotic treatment for travellers' diarrhoea. Cochrane 30

Database of Systematic Reviews, Issue 3, 2009. 2009. 31

31. Genton B, D'Acremont V. Evidence of efficacy is not enough to develop 32

recommendations: antibiotics for treatment of traveler's diarrhea. Clin Infect Dis. 2007; 33

44(11): 1520; author reply 1-2. 34

32. Tan T, Little P, Stokes T. Antibiotic prescribing for self limiting respiratory tract 35

infections in primary care: summary of NICE guidance. Bmj. 2008; 337: a437. 36

33. Rovers MM, Glasziou P, Appelman CL, Burke P, McCormick DP, Damoiseaux RA, et 37

al. Antibiotics for acute otitis media: a meta-analysis with individual patient data. Lancet. 38

2006; 368(9545): 1429-35. 39

34. Little P, Moore MV, Turner S, Rumsby K, Warner G, Lowes JA, et al. Effectiveness of 40

five different approaches in management of urinary tract infection: randomised controlled trial. 41

BMJ. 2010; 340: c199. 42

35. Kjolvmark C, Pahlman LI, Akesson P, Linder A. Heparin-binding protein: a diagnostic 43

biomarker of urinary tract infection in adults. Open Forum Infect Dis. 2014; 1(1): ofu004. 44

36. Lam CW, Law CY, To KK, Cheung SK, Lee KC, Sze KH, et al. NMR-based 45

metabolomic urinalysis: a rapid screening test for urinary tract infection. Clin Chim Acta. 46

2014; 436: 217-23. 47

37. Ciragil P, Kurutas EB, Miraloglu M. New markers: urine xanthine oxidase and 48

myeloperoxidase in the early detection of urinary tract infection. Dis Markers. 2014; 2014: 49

269362. 50

38. Bassler D, Briel M, Montori VM, Lane M, Glasziou P, Zhou Q, et al. Stopping 51

randomized trials early for benefit and estimation of treatment effects: systematic review and 52

meta-regression analysis. Jama. 2010; 303(12): 1180-7. 53

39. Alidjanov JF, Abdufattaev UA, Makhsudov SA, Pilatz A, Akilov FA, Naber KG, et al. 54

New self-reporting questionnaire to assess urinary tract infections and differential diagnosis: 55

acute cystitis symptom score. Urol Int. 2014; 92(2): 230-6. 56

Page 72 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

page 23 / 23

1

2

3 Formatted: Indent: Left: 0", First line: 0"

Page 73 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nlyManuscript Title: Symptomatic therapy of uncomplicated lower urinary tract infections in

the ambulatory setting. A randomized, double blind trial

Dear Editor,

Thank you for considering our manuscript for potential publication in the BMJ. The manuscript was

first submitted to the Lancet Infectious Disease and was rejected after review.

To facilitate with the assessment of our manuscript by the BMJ, we enclose the reviewer comments

received from the Lancet Infectious Disease below. We have also provided a point-by-point response

to each of the reviewer comments (lines and pages refer to the TRACK-CHANGE version).

Thank you for your kind consideration.

Prof. Peter Juni

Reviewer Comments Received from Lancet Infectious Disease

Reviewer #1: This paper describes a randomised double-blind trial comparing the effect of a NSAID versus

norfloxacin, with optional rescue antibiotic treatment, on symptomatic resolution in women with

lower UTI.

The majority of the manuscript is written clearly, and the statistical components appear to be

conducted well. However, I have some concerns about some aspects of the study that potentially

accumulate to major revisions. I outline these below:

1. The primary outcome on clinicaltrials.gov is at day 4. In paper (and elsewhere) it is at day 3.

Could the authors clarify this apparent discrepancy between sources?

We agree that this may be a point of confusion for readers. We initially referred to the day of

inclusion and randomization as day one (see secondary outcome on clinicaltrial.gov: “The proportion

of patients ever on antibiotics between randomization at day 1 and follow-up at day 30”). For the

purpose of this report, we refer to the day of randomization as day 0. According to this definition, the

primary outcome was ascertained at day 3, i.e. 72 hours after randomization and start of therapy.

We clarified this (page 7, line 9-12)

2. Can the authors justify why norfloxacin was chosen as the antibiotic of interest? This seems

relevant, given the international readership of the journal. In most places in the manuscript, the

findings were interpreted as "antibiotic therapy" as an umbrella term. I wonder if the choice of

antibiotic may have influenced the outcome, and whether the interpretation should be made more

specific to norfloxacin?

We agree that especially in view of the increasing resistance rates, quinolones are not considered as

first line therapy for uncomplicated lower urinary tract infections and should be used for more

invasive infections such as pyelonephritis. As nitrofurantoin and fosfomycin were not widely used in

Switzerland when our study started, we initially considered trimethoprim-sulfamethoxazole as a

comparator (see history on clinicaltrial.gov). Because of rather high resistance levels (26% in

Switzerland in 2015, www.anresis.ch), we decided to use an antibiotic with lower resistance levels to

ensure our antibiotic therapy was effective and to preserve the validity of our results. Indeed 97% of

all microorganisms detected proved to be susceptible to norfloxacin (page 11, line 13). As shown in

the meta-analysis our results are comparable to other similar studies using different antibiotic

regimens and there is no indication that the choice of antibiotic regimen would influence the results

Page 74 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly(as long as resistance rates are low). In view of the concomitant meta-analysis we retained the

wording as “antibiotic group”.

3. Similarly, can the authors justify the choice of NSAID, and consider the points I raise previously?

We are not aware of any body of literature to argue for or against a specific NSAID for the treatment

of UTI. The decision to select diclofenac was a practical one, because it is given twice daily as the

comparator norfloxacin and could be easily capsuled without any bulging.

We added the following sentence: “We chose norfloxacin because of the high susceptibility rates in

Switzerland and diclofenac because of its identical frequency of administration which facilitated

patient blinding (page 6, line 22-24).

4. For the analysis of urine samples, can the authors clarify how "predominant" was defined?

Predominance was judged by the microbiologist according to the distribution of the different

colonies on the agar plate. There is no clear cut in this definition, but is done in routine in daily

practice. As there is no clear cut definition we did not adapt the text.

5. Can the authors justify their choice of non-inferiority margin and choice of primary time point

(whether it is day 3 or 4)? Given the increasing concerns around the consequences of antibiotic

resistance, these two components of the trial are critical.

The latest meta-analysis available at the time of planning the trial (Falagas et al, J Infect 2009)

indicated that the largest trial comparing antibiotics with placebo (Ferry et al, Scand J Prim Health

Care 2007) found a risk difference of 35% for clinical cure at the end of treatment (25% with placebo

and 60% with antibiotics). The pre-specified non-inferiority margin is below half of this gain over

placebo. In view of a head-to-head trial of two antibiotics by Gupta K et al (Arch Intern Med

2007;167(20):2207-12), which found higher clinical cure rates than Ferry et al (79 and 84%), we

conservatively assumed rates of the primary outcome of symptom resolution to be at 70%. The time-

point was selected to target the end of protocol mandated treatments, before differences would get

diluted by the potential use of rescue antibiotics or other interventions in a large proportion of

patients after completion of study treatments in this non-inferiority trial.

6. In terms of sampling, and the risk of selection bias, can the authors describe how practices were

selected? Along the same lines, there is no description of how many eligible patients presented to

general practice (for example, using screening logs). Could the authors confirm what proportion of

eligible / potentially eligible patients presented, how many of those were approached, and (if

available) whether there were any distinguishing characteristics when comparing those who did

and did not participate in the trial?

There is no established network of primary care physicians interested in scientific collaboration in

Bern. In a first step, we therefore contacted all primary care physicians teaching students in their

practice. All interested physicians were invited to an information session on this study. Physicians

willing to participate had to pass a clinical investigator training (which was the main barrier for most

physicians), which was combined with a training on specific issues of this trial.

In order to minimize the administrative burden, we did not use screening logs. Therefore we are

unable to give you any information on how many patients declined participation in this study. During

the planning of this study, one practice asked about 20 patients, whether they would be willing to

participate (on a theoretical basis). About 60% of these patients declared their hypothetical

willingness. As the trial involved randomization to treatment arms, we do not think, that selection of

patients might have influenced the results.

Reviewer #2:

Page 75 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nlyReview: This is an interesting RCT on non-antimicrobial vs antimicrobial therapy of acute

uncomplicated lower UTI in women. Because antibiotic resistance is increasing, any possibility to

reduce antibiotic consumption is highly appreciated. Women with acute symptoms of uUTI were

randomly assigned to a 3-day therapy either with diclofenac or with norfloxacin. The overall results

show that diclofenac is inferior to norfloxacin in terms of symptom relief, but superior in terms of

reducing antibiotic use.

The study is well designed, performed and presented. The study was closed earlier because of slow

recruitment. There is, however, no hint that the main results would have been altered if more

patients would have been included into the study.

One concern of the non-antimicrobial therapy was the significantly higher number of

pyelonephritis (6 vs 0), which all occurred after day 3, which means which developed as ascending

infection from lower to upper UTI. A similar tendency was seen in an earlier study comparing

ibuprofen with fosfomycin, which is also discussed in the present manuscript. The authors could

found, that CRP as biomarker could be helpful to identify these patients early enough and start

antibiotic therapy delayed but probably early enough to prevent upper UTI (pyelonephritis). This of

course is an interesting aspect, but probably not practical, because acute cystitis in women is

usually treated empirically by oral drugs and monitoring CRP would need an additional blood

sample and additional cost. In general practice short term follow ups are also not performed

regularly. Nevertheless this is an very interesting aspect and opens the question whether other

biomarkers, e.g. obtained from urine, or better analysis of clinical symptoms by validated

questionairs, could be used to detect patients at risk to develop an upper UTI. In this case non-

antimicrobial therapy with the option of delayed antibiotic therapy in patients at risk could be the

best alternative to treat women with acute uUTI safely and reduce antibiotic consumption

significantly.

Thanks for this comment. We agree that uncomplicated lower UTI are usually treated without

additional analyses. Nevertheless, if a marker could have been identified, which would allow sparing

antibiotic therapy, this could even be cost-effective. This is speculative and our trial does not support

such strategies. However, we hope our work will trigger additional interventional studies on this

subject.

Reviewer #3: The authors present the results of an RCT comparing norflaxacin versus diclofenac in

uncomplicated urinary tract infections in women.

Women with typical complaints and a positive dipstick were included.

Overall the article is well written, the hypothesis clearly stated and the results and discussion easy

to read and including the relevant research on the topic. Despite not reaching the originally

intended sample size the article is well conducted and an important piece of work regarding the

discussion about antibiotic therapy in uncomplicated bacterial infections.

Major comments:

1. The effectiveness of both strategies in reducing the symptom burden was measured as the

percentage of patients reaching symptom resolution. This was defined as a symptom score of ≤ 10

points . Patients included had an average symptom score of 13.7 (SD 3.8) on a scale from 0-30.

Therefore the symptom score at the start was already very close to the primary outcome and the

difference between both groups might by very small. Therefore I am missing the discussion of a

minimally clinical important difference (and the necessary data to judge this).

I would ask the authors to present the data on symptom scores in both groups and the

development of the symptom score over the course of time in both groups.

My personal point of view would be that a difference of less than 2 points on a 0-30 scale has no

clinical relevance.

Page 76 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nlyWith respect to the minimally clinically important difference, the latest meta-analysis available at the

time of planning the trial (Falagas et al, J Infect 2009) indicated that the largest trial comparing

antibiotics with placebo (Ferry et al, Scand J Prim Health Care 2007) found a risk difference of 35% for

clinical cure at the end of treatment (25% with placebo and 60% with antibiotics). Consistent with

guidelines for the specification of non-inferiority margins, the pre-specified non-inferiority margin in

our trial is below half of this gain over placebo (U.S. Department of Health and Human Services. Non-

Inferiority Clinical Trials to Establish Effectiveness, Guidance for Industry, Nov 2016)

We thank the reviewer for their comment with respect to the difference in symptom scores but we

respectfully disagree. Symptom resolution was not defined as a total symptom score below 10, but

as “a slight problem”, defined as no more than 2 out of a maximum of 6 points on a Likert scale in

each single symptom dimension (dysuria, urgency, night frequency, frequency, lower abdominal pain

while urinating and back or loin pain). Symptom scores for the different complaint dimensions

differed at baseline, with “frequency” being the most and “back or loin pain” the least prevalent

symptom (table 1). In analogy to the study published by Little et al (BMJ 2010;340:c199,

doi:10.1136/bmj.c199), we selected “slight problems” as primary end point, because it is well known,

that slight problems may persist after clearing of the infection, but usually do not hamper daily

activity. Analyzing complete absence of symptoms (symptom score = 0) did not reveal any different

results (table 2).

As suggested by the reviewer we added an analysis on the development of the symptom score over

the course of time in both groups (Supplementary Figure 1). Because development of symptom

score was not a predefined outcome and results do not alter the messages of this study, we did not

discuss them in detail in the text.

2. In addition to the planned intervention all women were equipped with an antibiotic rescue

medication (fosfomycin) to be taken at the patients discretion (if symptoms persisted). Was there

any further instruction for the patients?

No, there were no additional instructions. We intended to have a low barrier for patients to gain

access to an (additional) antibiotic therapy. On one hand, this explains the higher use of antibiotics in

the control group observed in our study (see meta-analysis. On the other hand, this strengthens our

hypothesis, that antibiotic consumption may be reduced even in settings, where antibiotic therapy is

easily accessible.

3. I do not understand the rationale behind this intervention as from previous observational

studies (Little 2010 BMJ 2010;340:b5633) the mean symptom duration in UTI is slightly longer than

3 days, thus a second antibiotic at day three is likely to bias the results in favour of primary

antibiotic use.

The implications of this early rescue intervention should be discussed.

Thank you for this comment. In the observational study referred to by the reviewer, the overall mean

symptom duration of uncomplicated UTI in the study includes situations where the infectious

organism was resistant to the antibiotic prescribed. Indeed, when there was no antimicrobial

resistance, the duration of symptoms of dysuria, urgency, frequency, and abdominal pain is ≤3 days.

We chose norfloxacin as the primary antibiotic in our trial given the low resistance rates in

Switzerland to this antibiotic and the duration of 3 days is the established treatment regimen for

norfloxacin in uncomplicated UTI.

A potential use of the rescue antibiotic was mainly anticipated after completion of the regular,

protocol mandated regimen, i.e. once the first 3 days of treatment were over. A delay until the

rescue antibiotic was used would not have corresponded to clinical routine and would have been

difficult to defend with IRBs.

Page 77 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

Please note that 9 patients (8%) in the antibiotic group used the rescue antibiotic before day 3, at

their own discretion, and another 9 patients (8%) between days 4 and 30; corresponding numbers in

the NSAID group were 55 (41%) up to day 3, and 18 patients (14%). The trial protocol is therefore

unlikely to have biased results in favor of antibiotics.

4. Patients included only had light to moderate symptom scores (average 13 out of 30). This

suggests a selection bias as already described in the ICUTI study- could you comment on this. Did

you assess non-responders? If so, please give some information on symptom severity or other data

assessed.

We agree, that our study design might have selected for women with less severe symptoms, and as

we did not use screening logs (see last question of reviewer 1), we are not able to quantify this

effect. Although only few women had symptom scores above 20, we were not able to detect any

difference in this sub-group (fig. 1). We agree that our results cannot be extrapolated to women

suffering from very severe symptoms, nevertheless the impact on overall antibiotic prescription,

would be relevant, even if women with very severe symptoms would be treated with antibiotics

primarily.

Non-responders correspond to patients in the NSAID group, switching to antibiotic therapy. This

group was compared to patients in the NSAID group, never treated with antibiotics in a post-hoc

analysis. Results are summarized in the supplementary table 4. We did not find any relevant

differences between these two groups.

5. Patients were included by 17 GP could you give some information about the recruited patients

per practice

Following practice types participated in our study:

- 7 practices with only one physician recruited 1-10 patients each, total 36 patients

- 9 group practices recruiting 3 to 26 patients each, total 108 patients

- One big group practice, with 2 locations in Bern, where 17 practitioners included a total of

109 patients. Patient inclusion in this practice was very high, because the size of the practice,

the location in the center of town (which is very popular for young, otherwise healthy

patients going to work) and the fact, that the first author is working in this practice, working

constantly on motivation of the colleagues. In addition this was the first practice starting with

inclusion of patients.

We added the following sentence in the first paragraph of the results: “Thirty-six patients were

recruited in seven single practices, 108 in 9 group practices and 109 in one big medical centre with 17

physicians in the centre of Bern.” (page 11, line 3-5)

6. The number of women developing pyelonephritis was quite high. Of special interest would be

the symptom score in women developing a pyelonephritis- could you present this data

additionally. Were there any hints on this serious adverse event? Might pain medication lead to

non detection of PN?

How did the symptom score change during the study.

Indeed, it would be desirable if any hints towards a serious adverse event could have been detected,

but unfortunately this was not the case. As stated in the last paragraph of the result section, CRP

level at baseline was the only parameter with a positive likelihood to develop a pyelonephritis. We

looked in detail at the symptom scores, but were not able to detect any additional risk factors. We

added details in Supplementary table 7, a comprehensive summary of these additional analysis in the

result section (page 13, line 11-13) and give data on symptom scores of patients developing

pyelonephritis in the supplementary table 8.

Development of symptom score during the study in depicted in supplementary figure 1 (see answer

1, reviewer 3).

Page 78 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly7. In the discussion I am missing the patient perspective. The results of this and the comparable

trials cited are a valuable basis for shared decision making as, at least from qualitative data

patients are willing to delay or to avoid antibiotic use due to the side effects

We agree with the reviewer that shared decision making probably is appropriate in handling

uncomplicated UTI. Our data help to inform the patient on a scientific basis. Of course in our study,

all participating patients were willing to (possibly) delay or avoid antibiotic therapy, otherwise there

would not have participated. We have added a sentence to our discussion in this regard (page 16,

line 16-19).

All further comments are on minor points only.

8. P 6 Patients

Exclusion criteria were stated, I would suggest to mention fever explicitly here (not only in the

supplement)

As suggested by others we included the complete set of inclusion and exclusion criteria in the main

text (and not in the supplement only, page 5-6)

9. P 7

"general satisfaction on day 10"

Again some information on the method (for example 10 point likert scale) would be helpful,

otherwise the reader would have to look up the supplement

For the symptom score we used a 6 point-likert scale, which is briefly described in procedures and

outcomes (page 7, line 15), Quality of Life was rated from 0-10, which is mentioned directly in table

2.

10. P 7 Cut off >103 cfu is not supported by the citation (EAU guideline) and is different from the

study protocol ( Instead ≥ 103 cfu is mentioned in this guideline and commonly used in practice).

We have adapted this typographical error accordingly (page 7, line 27)

11. P14 "unable to find the interaction with urine culture" to what kind of interaction is this

comment referring to?

As shown in figure 1B we were not able to see any risk difference in antibiotic prescription between

women with and without positive urinary cultures, while Gagyor et al. found a significantly higher

reduction of antibiotic treatment courses in the ibuprofen vs. the fosfomycin group in women with

negative urinary cultures (90.7%, CI 74.3% to 99.9%) than in women with positive urinary cultures

(58.5%, CI 49.8% to 67.0%).

We clarified this sentence: “In particular, in contrast to Gagyor et al. , reduction in antibiotic

prescription was comparable in women with and without positive urinary cultures” (page 16, line 21-

23).

Reviewer #4: The authors present a randomized trial comparing ibuprofen versus norfloxacin in patients with

uncomplicated cystitis.

The topic of the study is timely.

There was some recalcuation of recruitment power due to slow recruitment, however the final

results revealed significant findings.

1. Comments:

Page 79 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nlyThe comparator of this study was norfloxacin, which is in most guidelines not a first choice

antibiotic. There should be some explanations why the authors have choosen norfloxacin in this

study.

Please see answer 2 reviewer 1

2. The primary outcome was symptom resolution at day 3. There should be some note how this

date was selected, why not 7 days or 5 days. Is there any data showing a preference for 3 days?

The measurement of symptoms was performed by a self-reported diary. Is this questionaire a

validated questionaire, which is measuring what it is intended to measure?

In the absence of antimicrobial resistance, the mean duration of symptoms in women with

uncomplicated UTIs treated with antibiotics is 3 days (Little P et al. : BMJ 2010;340:b5633

doi:10.1136/bmj.b5633).

As discussed by Alidjanov et al (Urol Int 2014;92:230-36), there was no single accepted symptom

score measurement for lower UTI. In this review – which was published after the development of our

trial - Adlidjanov et al proposes (and assesses) an 18-item questionnaire including 6 typical symptoms

(“frequency”, “urgency”, “painful urination”, “incomplete emptying”, “suprapubic pain” and

“haematuria”), 4 items concerning potential differential diagnosis as STD or pyelonephritis, 3 items

on Quality of Life and 5 additional questions, which may affect therapy (as pregnancy, diabetes …).

Other studies focused on slightly different typical symptoms as “frequency”, “urgency”, “pain or

burning when passing urine”, “pain or uncomfortable pressure in the lower abdomen/pelvic area”,

“low back pain”, “not being able to empty the bladder completely/passing only small amounts of

urine” and “blood in urine” (Clayson D, et al. BJU Int. 2005 Aug;96(3):350-9); “dysuria”,

“haematuria”, “frequency during day and night”, “smelly urine,” “tummy pain”, “generally feeling

unwell”, and “restriction of daily activities” (Little et al. BMJ 2010;340:c199); “dysuria”, “frequency”

and “low abdominal pain” (Bleidorn et al. BMC Medicine 2010, 8:30); or “dysuria”,

“frequency/urgency of micturition” and “low abdominal pain” (Gagyor et al. BMJ 2015;351:h6544). In

addition Little et al. showed that nocturia and dysuria were the most sensitive symptoms to diagnose

lower UTI, but negative predictive value still was poor (Little et al. Health Technology Assessment

2009; Vol. 13: No. 19).

The selection of our symptom score (“dysuria”, “frequency”, “urgency”, “abdominal pain when

passing urine”, “pain or tenderness in the lower back or loin”) was based on the available literature

at that time and is close to the questionnaire suggested by Alidjanov in 2014. We did not include

“incomplete emptying” and “haematuria”, although the latter was included in the inclusion criteria.

Items concerning differential diagnoses were included in the exclusion criteria and quality of life was

assessed on day 3 only. In addition Alidjanov et al use a 4 point-Likert scale (0-3), while we used a 7

point Likert-scale (0-6). Taking into account the available literature, we still are convinced that our

symptom score addresses the most important symptoms and is valuable in comparing the two

treatment groups. We agree, that using a standardized questionnaire as suggested by Adlidjanov et

al. would ease the comparability between different trials, and should therefore be promoted.

We briefly discuss these considerations in the paragraph on strengths and limitations (page 17, line

20-23).

3. It would be interesting to see a subcohort analysis, which patients would have had a benefit

from ibuprofen and which not. Is this feasable from the data, perhaps from those with lower

symptoms?

This indeed is an interesting point and that is what we intended to do in the post-hoc analysis. The

patients in the NSAIDs group that never used antibiotics can be seen as those that benefitted from

the NSAIDs (in comparison to those that used antibiotics). However, as discussed in the text, we

Page 80 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nlycould not identify baseline characteristics that are clearly different between the two groups

(Supplemental table 4).

4. It seems that the patients were not seen by a physician at follow up visits. This needs to be

clarified. Also if the diagnosis pyelonephritis was only done per telefone interview or if patients

reported to the study centers.

The reviewer is right, a scheduled follow up visit was not pre-planned, in order to keep the procedure

as simple and easy as possible. The follow-up is described in detail in “procedures and outcomes”.

As study physicians were the family physicians, follow-up was very close in case of any problems.

Indeed pyelonephritis was never diagnosed per telephone, all patients diagnosed with pyelonephritis

at least had one unplanned ambulatory visit. In the section “procedures and outcomes” we specified

that “The clinical diagnosis of pyelonephritis required the occurrence of loin pain and fever, leading

to an unplanned ambulatory visit.” (page 7,line 25)

5. It is also debatable, if the rescue antibiotic fosfomycin trometamol that was handed at study

inclusion would have facilitated antibiotic treatment.

We agree that providing the rescue antibiotic has probably facilitated the antibiotic treatment and

this is the most probable explanation for the higher antibiotic prescription in the NSAID group

compared the other studies as depicted in the Meta-analysis.

We have amended the discussion section of our meta-analysis as follows:

“On the beneficial side, our meta-analysis suggests that antibiotic use can be halved on average by

initial symptomatic treatment with NSAIDs, with a corresponding number needed to treat of 2 to

prevent one instant of antibiotic use. Although there was moderate heterogeneity for this outcome,

it may relate to methodological differences in the conduct of our trial compared to prior trials. In

particular, we provided women with a rescue antibiotic for discretionary use after completion of

study medication and reaching the primary endpoint. This may have facilitated antibiotic use in the

NSAID treatment arm of our trial. Nonetheless, given the high global incidence of UTIs, the reduction

in antibiotic use is highly relevant and is likely to immediately decrease resistance rates for E. coli and

even other microorganisms in the affected population.21

”

Reviewer #5:

GENERAL COMMENTS

The study conducted poses a relevant question with appropriate methods to answer. In an era of

increase in antimicrobial resistance primarily driven by increased environmental antibiotic

pressure every effort to improve is most important. Unfortunately, the authors identified an

inferiority of the non-antimicrobial approach compared to antimicrobial approaches. The main

conclusion driving outcome has been 3 day symptom recovery. Another important issue is the risk

of developing more serious UTIs (pyelonephritis) in patients treated for symptoms. As authors,

have highlighted further studies to detect the patients under the risk of transitioning to more

severe infections can now be better justified. This would be useful for individualized treatment

strategies in women with cystitis and ensure a lower antimicrobial pressure.

METHODS:

1. The study has been terminated earlier due to recruitment difficulties. Although, this is a

limitation I agree with the methods of handling the issue. Nevertheless, my interest is related to

with the reasons of low recruitment. Could you please elaborate on the reasons of why the

recruitment targets were not satisfied?

Page 81 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nlyThere are several reasons for the low recruiting. Because we did not analyze this reasons

scientifically, we do not detail them in the revision, nevertheless I give some reasons below:

• So far there was no network of general physicians, interested in scientific collaborations. A

side effect of this study, was to establish this network. However we invested a lot of time in

recruiting physicians, who finally did not include patients or only 1-2. We added some details

about the recruiting physician in the first paragraph or the results.

• The obligations from the Swiss authorities were higher than expected. Every physician had to

perform a clinical investigator training, which withheld some colleagues from participating in

this study.

• Some delay, was due to the long illness and death of Kathrin Mühlemann, the initial primary

investigator.

• In a qualitative survey among clinical investigators during the recruitment phase revealed

two main reasons, for not including more patients: First missing time during daily business to

inform patients properly about this trial and second more patients than expected (see also

reviewer 1, answer 6), denied to participate. Because we did not have screening logs we are

not able to quantify the magnitude of this effect.

2. Similar previous studies cited in the paper have recruited the similar numbers as the established

target in the current study. At this point the results of the trial log-book would be relevant. What is

the rate of willingness to participate in the study from patients approached for the study? If

available please provide in supplement. Did you identify a lack of equipoise in either patients or

clinicians? For those patients who did not participate in the study do you know their treatment

preferences (i.e NSAID vs antibiotics)?

To make recruitment as easy as possible, we did not use screening logs (see answer 1 and answer 6

of reviewer 1) and therefore are not able to give more details on this issue. Because the study was

randomized, these issues should not affect the results of this study.

3. Could you please further elaborate and report the pattern of missing data of questionnaires

(missing completely at random, missing at random and missing not at random).

The proportion of missing data was too low to be able to meaningfully determine whether there are

associations of patient characteristics with data missingness. Therefore, we are unable to come to

any conclusions as to the type of missing data. Since the proportion of missing data was small and

the distribution similar between groups, our estimates are unlikely to be biased even if data had

bveen missing not at random.

To address this reviewer’s comment, we report the numbers of patients with missing data per

outcome in the web-appendix (Supplemental Table 1).

4. Recovery has been described by a symptom score of 2 or less at follow-up. Papers cited for this

purpose are: Gupta et al. 2007, McNulty et al 2006 and Clayson D et al 2005. Nevertheless, I did not

come across with this definition in these papers. For instance, McNulty et al used complete

resolution in follow-up symptom scores as the primary outcome. Could you please further

elaborate on why and how the cut-off value of 2 or less symptom score was selected.

We agree with the reviewer, that different definitions of symptom resolution are used in previous

studies and discuss this issue in answer 1 of reviewer 3. Because different end-points have been

used, we operationalized the outcome of symptom resolution three ways: symptom resolution on

day 3 and 7, complete absence of symptoms at day 3 and 7 and change of symptom score. In

addition we added an analysis of development of symptom score as suggested by two reviewers

(Supplementary Figure 1). All analyses revealed comparable results, reinforcing our results.

5. The primary outcome of the study is 3 day symptom recovery. Why did you prefer to limit this

with 3 days? By selecting this point in time only to compare these management options there is a

preference introduced. The overall effect on symptoms of these treatments (0 to 30 days) with

Page 82 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nlyappropriate methods might summarize the overall situation better for future patients to make

choices.

See also reviewer 1, answer 5. Because we expected, that many patients would use “rescue

antibiotics” on day 3, we chose the primary end point at day 3, directly before the eventual intake of

the rescue antibiotic, which would influence the outcome. The predefined secondary outcome

symptom score on day 7 as well as change of symptom scores at day 3 and 7 did not reveal any

different results. In the supplemental figure 1 we added a boxplot analysis of symptom scores day 0 –

10 (see also reviewer 3, answer 1).

6. Measurement of symptoms was conducted by rating the severity of five UTI symptoms ((i)

dysuria; (ii) frequency, (iii) urgency, (iv) abdominal pain when passing urine, (v) pain or tenderness

in the lower back or loin). The originally developed questionnaire by Clyson D et al. BJUi 2005

determined that there are seven signs and symptoms with three domains. Please justify the

reasons you used a shorter version.

See reviewer 4, answer 2

7. Has the questionnaires used been translated into French and German and if so please cite the

validation study also. This process has been done for EuroQol and necessary citations would be

useful.

We eventually did not include French speaking patients and did not translate the diary or the

informed consent into French.

8. In 2014 a more comprehensive and specific questionnaire was developed for measuring patient

reported outcomes in patients with acute cystitis. This of course was after the study started but I

think it may be relevant for the authors to briefly mention in discussion that future studies could

consider this also. (New Self-Reporting Questionnaire to Assess Urinary Tract Infections and

Differential Diagnosis: Acute Cystitis Symptom Score. Alidjanov J.F.a · Abdufattaev

U.A.a · Makhsudov S.A.a · Pilatz A.c · Akilov F.A.a ·Naber K.G.b · Wagenlehner F.M.c )

We thank the reviewer for this valuable comment. We discuss this issue in detail in answer 2 of

reviewer 4 and briefly in the section on strengths and limitations in the manuscript.

9. Secondary outcome has been specified as use of antibiotics up to 30 days. It is not clear if this

includes the treatment given to the antibiotic group or is it only additional treatments given.

Intuitively it seems like it includes antibiotics given for randomized patients (Norfloxacin).

However, in figure 2b Norfloxacin arm day 1 antibiotic probability is less than 100%. Could you

please explain and make clearer in the manuscript.

In the section procedures and outcomes we describe the secondary outcome as “use of any

antibiotic up to day 30”. We specified now to “use of any antibiotic (including norfloxacin and

fosfomycin as trial medications) up to day 30” (page 7, line 18). As depicted in figure 1, 2/120

patients in the antibiotic treatment arm did not take antibiotics, which translates in probability to

take antibiotics below 100% even in the antibiotic treatment arm in figure 2b.

10. Rescue treatment:

The susceptibility profile of cases with positive urine culture is known at baseline (probably latest

2 days). In cases from the NASID group where fosfomycin resistance was measured were patients

informed? Please give reason if no and if yes the implications (for instance has the rescue

treatment option been changed in these patients?).

Could you explain why was fosfomycin selected as the rescue treatment option?

Patients were not informed on susceptibility results and rescue therapy was not changed for several

reasons: 1) susceptibility to fosfomycin was known to be very high in Switzerland (99% in our trial), 2)

the treating physician was not aware of the treatment group and 3) scheduled follow-up or any

active interaction was not planned during the first 10 days, to make the study as easy as possible.

Page 83 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nlyNevertheless susceptibility results were provided to the treating physician in case of any

complication.

Fosfomycin was selected because of the very high susceptibility rate in Switzerland and the easy

administration.

We did not elaborate on these issues further in the manuscript.

RESULTS

11. table 1 please provide the p values to prove that at baseline groups were similar.

In accordance with the CONSORT guidelines of reporting of randomized trials, we refrained from

performing significance tests of baseline differences between groups. As our study was a randomized

trial, any differences would have occurred by chance and therefore statistically significant p-values

would be unnecessary and potentially misleading to readers.

12. Figure 2A: please also show censored cases on the plot.

In our experience, a clinical readership has difficulty in understanding the ticks indicating censored

cases on time-to-event curves. We therefore indicated the numbers of censored cases in

Supplementary Table 2 and 6.

13. Typo in page 11paragraph 2 last sentence (rescue).

Done

14. Paragraph 3: “Median time until antibiotics use were 0 and 5 days respectively.”. This means

that it was 0 for the NSAID group, in relation to the previous sentence.

Thank you for this comment. You are completely right, we have changed this sentence accordingly.

15. Have you repeated the sensitivity outcomes for other outcomes such as complete resolution

rates, time to complete resolution etc?

No we did not do so. Our sensitivity analysis was done as pre-specified and we did not elaborate

further because there was no hint for any inconsistency.

16. It might be useful to have a more detailed table or plot for susceptibility profiles of pathogens

in each group in the supplement. As it stands it is difficult for a reader interested in details to

follow.

For instance, Enterobacteriaceae was detected in 173 cases. Fosfomycin susceptibility tested for

160 and 158 were sensitive. From supplement table 2 it is understood that urine culture was

negative in 46 (34%) of NSAID group and Enterobacteriaceae was detected in 88 cases (66%).

Susceptibility was only tested for Enterobacteriaceae and 84 were sensitive. This means either 4

cases in NSAID were resistant or not measured for susceptibility.

With respect to my previous question I think this information becomes relevant for rescue

treatment method question above.

The table below lists the detailed information for the 160 isolates tested against fosfomycin. We

believe that an additional table in the supplement to show this data is not needed. We propose to

add an additional comment in the notes of table 1 (“Two isolates from the NSAID group (1 Proteus

mirabilis, 1 Enterobacter cloacae) were not susceptible to fosfomycin”) instead. However, we could

add the table in the supplement, if deemed necessary by the editor.

Page 84 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nlyFosfomycin-susceptibility NSAIDs Antibiotics

E. coli 78/78 68/68

K. pneumoniae 1/1* 1/1

Citrobacter spp. 4/4 1/1

Proteus mirabilis 2/3 1/1*

Pantoea agglomerans 1/1*

Enterobacter cloacae 0/1 1/1

total 85/87 73/73

* in double mixed infection with fosfomycin-susceptible E. coli

17. Although the sample size of patients developing pyelonephritis is low it is very important.

Therefore, it is appreciable that the authors have tried to detect any baseline variables that might

be useful. Was there any other analysis done to determine if post treatment variables such as

response in symptoms, use of rescue treatment etc that could be useful. I appreciate the limited

number of events and difficulty to propose any inferences but it would be a useful hypothesis

generating exercise and help future research.

We have added some additional information in the text and in the supplementary table 7 and 8 (see

answer 6 of reviewer 3)

Discussion is very well written. I only have a few minor comments.

Please try to avoid abbreviations of terms not used before in the text to allow the general to follow

easier.

We replaced “NNT” by “number needed to treat”

In the final conclusion, one potential implication of reduced antibiotic usage is described as future

reduction in E.coli resistance. I would propose a more wider perspective where we shift to a lower

antibiotic pressure environment and ultimately avoid emergence of resistance (not only for E.coli).

A similar argument is made in the discussion section. One of the recent reviews paper to support

this argument is published by Alison Holmes et al: Understanding the mechanisms and drivers of

antimicrobial resistance Lancet 2016; 387

We agree with the reviewer, that reduction in quinolone has to potential to reduce quinolone

resistance not only in E. coli but also in other microorganisms and adapted the corresponding

sentences in the discussion (“…and is likely to immediately decrease resistance rates for E. coli and

even other microorganisms in the affected population” (page 16, line 5)) and the conclusion

(“…decreasing resistance rates for E. coli in the affected population” (page 18, line 3).

Page 85 of 85


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

Documents

Symptomatic therapy of uncomplicated lower urinary tract blind … · 2017-11-10 · Confidential: For Review Only page 1 / 26 1 Symptomatic therapy of uncomplicated lower urinary