new england

The
journal of medicine
established in 1812 june 19, 2014 vol. 370 no. 25

Antimicrobial Prophylaxis
for Children with Vesicoureteral Reflux
The RIVUR Trial Investigators*

A BS T R AC T

Background
Children with febrile urinary tract infection commonly have vesicoureteral re Members of the writing group (Alejandro
flux. Because trial results have been limited and inconsistent, the use of anti Hoberman, M.D., Childrens Hospital of
Pittsburgh of University of Pittsburgh
microbial prophylaxis to prevent recurrences in children with reflux remains Medical Center, Pittsburgh; Saul P. Green-
controversial. field, M.D., Women and Childrens Hos-
pital of Buffalo, Buffalo, NY; Tej K. Mat-
too, M.D., Wayne State University School
Methods of Medicine, Detroit; Ron Keren, M.D.,
In this 2-year, multisite, randomized, placebo-controlled trial involving 607 chil M.P.H., Childrens Hospital of Philadel-
phia, Philadelphia; Ranjiv Mathews, M.D.,
dren with vesicoureteral reflux that was diagnosed after a first or second febrile or Johns Hopkins School of Medicine, Balti-
symptomatic urinary tract infection, we evaluated the efficacy of trimethoprim more; Hans G. Pohl, M.D., Childrens Na-
sulfamethoxazole prophylaxis in preventing recurrences (primary outcome). Second tional Medical Center, Washington, DC;
Bradley P. Kropp, M.D., University of
ary outcomes were renal scarring, treatment failure (a composite of recurrences and Oklahoma Health Sciences Center, Okla-
scarring), and antimicrobial resistance. homa City; Steven J. Skoog, M.D., Oregon
Health and Science University, Portland;
Caleb P. Nelson, M.D., M.P.H., Boston
Results Childrens Hospital, Boston; Marva Moxey-
Recurrent urinary tract infection developed in 39 of 302 children who received pro Mims, M.D., National Institute of Diabe-
phylaxis as compared with 72 of 305 children who received placebo (relative risk, tes and Digestive and Kidney Diseases,
Bethesda, MD; Russell W. Chesney, M.D.,
0.55; 95% confidence interval [CI], 0.38 to 0.78). Prophylaxis reduced the risk of Le Bonheur Childrens Hospital, Mem-
recurrences by 50% (hazard ratio, 0.50; 95% CI, 0.34 to 0.74) and was particularly phis, TN; and Myra A. Carpenter, Ph.D.,
effective in children whose index infection was febrile (hazard ratio, 0.41; 95% CI, University of North Carolina at Chapel
Hill, Chapel Hill) assume responsibility
0.26 to 0.64) and in those with baseline bladder and bowel dysfunction (hazard for the content of this article. Address re-
ratio, 0.21; 95% CI, 0.08 to 0.58). The occurrence of renal scarring did not differ print requests to Dr. Hoberman at Chil-
significantly between the prophylaxis and placebo groups (11.9% and 10.2%, re drens Hospital of Pittsburgh, General
Academic Pediatrics, 3414 Fifth Ave.,
spectively). Among 87 children with a first recurrence caused by Escherichia coli, the CHOB 3rd Fl., Rm. 321, Pittsburgh, PA
proportion of isolates that were resistant to trimethoprimsulfamethoxazole was 15213, or at hoberman@chp.edu.
63% in the prophylaxis group and 19% in the placebo group. * A complete list of the Randomized In-
tervention for Children with Vesicoure-
teral Reflux (RIVUR) trial investigators
Conclusions
is provided in the Supplementary Ap-
Among children with vesicoureteral reflux after urinary tract infection, anti pendix, available at NEJM.org.
microbial prophylaxis was associated with a substantially reduced risk of recur This article was published on May 4, 2014,
rence but not of renal scarring. (Funded by the National Institute of Diabetes and at NEJM.org.
Digestive and Kidney Diseases and others; RIVUR ClinicalTrials.gov number, N Engl J Med 2014;370:2367-76.
NCT00405704.) DOI: 10.1056/NEJMoa1401811
Copyright 2014 Massachusetts Medical Society.

n engl j med 370;25nejm.orgjune 19, 2014 2367

The New England Journal of Medicine
Downloaded from nejm.org on June 4, 2016. For personal use only. No other uses without permission.
Copyright 2014 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

V
esicoureteral reflux is present in contraindications for the use of trimethoprim
one third of children presenting with fe sulfamethoxazole, or certain medical conditions
brile urinary tract infection and has been (see the Supplementary Appendix, available at
associated with a heightened risk of renal scar NEJM.org). We followed the children for 2 years
ring.1 Early randomized, controlled trials that to ascertain the primary outcome of febrile or
compared antireflux surgery with antimicrobial symptomatic recurrence and secondary outcomes
prophylaxis showed no significant differences in of renal scarring, failure of prophylaxis, and anti
the rates of recurrent urinary tract infection (re microbial resistance. We also assessed bladder and
currences) and renal scarring2-5; however, the lack bowel dysfunction according to a standardized
of a placebo or observation group precluded a measure (see the Supplementary Appendix).12,17,18
determination that either surgery or prophylaxis An NIDDK data and safety monitoring board
was effective. More recent randomized trials, and an institutional review board at each site
most of which were unblinded, have had con approved the protocol. Parents or legal guardians
flicting results regarding the effectiveness of anti of the children provided written informed con
microbial prophylaxis in reducing recurrences.6-11 sent before enrollment.
We designed the Randomized Intervention for
Children with Vesicoureteral Reflux (RIVUR) trial Randomization
to determine whether long-term antimicrobial Randomization was performed in permuted
prophylaxis is effective in preventing febrile or blocks, stratified according to site. Data man
symptomatic recurrences and reducing the likeli agement and statistical analyses were per
hood of renal scarring, and the extent to which it formed by a data-coordinating center at the
contributes to bacterial resistance. University of North Carolina at Chapel Hill. We
randomly assigned children to receive daily tri
Me thods methoprimsulfamethoxazole suspension (3 mg
of trimethoprim plus 15 mg of sulfamethoxa
Study Design and Oversight zole per kilogram of body weight) or placebo
The study was a randomized, double-blind, pla (prepared by UPM Pharmaceuticals) that was
cebo-controlled trial of prophylaxis with trimeth nearly identical in color, smell, taste, and con
oprimsulfamethoxazole (Sulfatrim Pediatric sistency. Four persons at the data-coordinating
suspension from Actavis and Sulfamethoxazole center managed the linkage of study-group as
and Trimethoprim Oral Suspension from Hi-Tech signments with study medications; the data-
Pharmacal) in children with vesicoureteral reflux management software confirmed the eligibility
that was diagnosed after a first or second febrile of each child and generated assignments to
or symptomatic urinary tract infection (index in coded bottles of study medication.
fection). The project steering committee designed
the trial, which was sponsored by the National Study Procedures
Institute of Diabetes and Digestive and Kidney Urine specimens from children who were not
Diseases (NIDDK) and was conducted under an toilet trained were obtained by means of cathe
Investigational New Drug application (number terization or suprapubic aspiration; bag-collected
75,739) to the Food and Drug Administration. specimens were not permitted. Clean, voided
The study rationale and design have been pub specimens were collected from toilet-trained chil
lished previously,12-16 and the protocol is avail dren. Index and outcome infections met stringent
able with the full text of this article at NEJM.org. diagnostic criteria that included evidence of pyuria
The writing group wrote the manuscript and on urinalysis, culture-proven infection, and fever
vouches for the completeness and accuracy of the (38C) or urinary tract symptoms within 24 hours
data and analyses and the fidelity of the study to before or after urine collection.12,14 Renal scan
the protocol. ning with technetium-99mlabeled dimercapto
We enrolled children at 19 clinical sites across succinic acid was performed at baseline and after
the United States who were 2 to 71 months of 1 and 2 years. Two pediatric nuclear-medicine
age and had grade I to IV vesicoureteral reflux. physicians assessed the extent of renal cortical
We excluded children whose index infection oc defects semiquantitatively by dividing the cortex
curred more than 112 days before randomization into 12 segments and determined severity on the
and children with coexisting urologic anomalies, basis of the number of segments affected. Treat

2368 n engl j med 370;25nejm.orgjune 19, 2014

The New England Journal of Medicine

Downloaded from nejm.org on June 4, 2016. For personal use only. No other uses without permission.
Copyright 2014 Massachusetts Medical Society. All rights reserved.
 Prophylaxis for Children with Vesicoureter al Reflux

Figure 1. Enrollment, Randomization, and Follow-up

of Children in the Trial. 10,871 Children were assessed for eligibility
Capture of screening data commenced in November 2007;
children were enrolled at various clinical sites (emergency
departments, radiology departments, and primary care, 9445 Were not eligible
6374 Did not meet UTI criteria
urology, and nephrology offices), resulting in diverse 807 Did not undergo VCUG
criteria for screening (abnormal urinalysis results, posi- 1646 Did not have VUR
tive urine culture, and abnormal result of voiding cysto- 61 Had VUR of grade V
urethrography [VCUG]). Enrollment of children com- 557 Met other exclusion criteria
menced in June 2007; 115 eligible children were enrolled
throughout the study without corresponding screening
data for those not enrolled. Children who were withdrawn 1426 Were eligible
from the study by a parent include some children who
discontinued the intervention because they met criteria
for treatment failure or had a parent who preferred prophy- 819 Did not participate
laxis to be prescribed. TMP-SMX denotes trimethoprim
sulfamethoxazole, UTI urinary tract infection, and VUR
vesicoureteral reflux. 607 Underwent randomization
492 Had screening data available
115 Did not have screening data available
for those not enrolled
ment failure was defined as two febrile recur
rences, one febrile and three symptomatic recur
rences, or four symptomatic recurrences over the
course of the study or as new or worsening scar
ring at 1 year.14 302 Were assigned to receive 305 Were assigned to receive
At baseline, after the occurrence of treatment TMP-SMX placebo

failure, and at 24 months, we obtained rectal

swabs, which were tested in a central laboratory
for resistance of Escherichia coli isolates to trimeth 247 Were assessed at 1-yr visit 262 Were assessed at 1-yr visit
oprimsulfamethoxazole.19 We also assessed, at 261 Were assessed at 2-yr visit 259 Were assessed at 2-yr visit

local laboratories, resistance among pathogens 202 Continued intervention 189 Continued intervention
identified in children with recurrences. 24 Had inconclusive intervention 25 Had inconclusive intervention
status status
76 Discontinued intervention 91 Discontinued intervention
Statistical Analysis 14 Were withdrawn by parent 16 Were withdrawn by parent
12 Met criteria for treatment failure 27 Met criteria for treatment failure
We estimated that a sample of 300 children in 20 Had study medication stopped 18 Had study medication stopped
each study group would provide at least 80% by parent by parent
7 Had adverse drug reaction 6 Had adverse drug reaction
power to detect a reduction in the proportion of 6 Had discontinuation related 7 Had discontinuation related
children with febrile or symptomatic recurrences to UTI to UTI
4 Had parent who preferred 8 Had parent who preferred
during a 2-year follow-up period from 20% in the prophylaxis to be prescribed prophylaxis to be prescribed
placebo group to 10% in the trimethoprimsul 13 Had other reasons 9 Had other reasons

famethoxazole group, at a two-sided alpha level 16 Were withdrawn from study 19 Were withdrawn from study
of 0.05. Our a priori plan called for an intention- by parent by parent

to-treat analysis of recurrences, stratified accord

ing to site, with the use of a CochranMantel
Haenszel test with missing outcomes imputed as 302 Were included in the analysis 305 Were included in the analysis
events, as well as time-to-event analyses with the
use of a stratified log-rank test. We used Kaplan
Meier estimates to describe the proportions of been adjusted for multiple comparisons; a P value
children who had febrile or symptomatic recur of less than 0.046 was considered to indicate sta
rences or who met the criteria for treatment fail tistical significance for the primary outcome.
ure. We used Cox models both in analyses of treat Sensitivity analyses of febrile or symptomatic re
ment effects according to age, sex, severity of currences were performed, first, by including
reflux, and other factors and in analyses that children for whom outcome data were missing
were controlled for multiple factors. We used and classifying them as not having had an event
Fishers exact test to compare adverse events in and, second, by omitting such children from the
the two study groups. Reported P values have not analyses.

n engl j med 370;25nejm.orgjune 19, 2014 2369

The New England Journal of Medicine
Downloaded from nejm.org on June 4, 2016. For personal use only. No other uses without permission.
Copyright 2014 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

R e sult s data for those not enrolled were not available

(Fig. 1). Follow-up was completed in June 2013.
Study Participants Baseline demographic and clinical characteris
We screened and enrolled children between June tics of the trial participants have been reported
2007 and May 2011; however, capture of screen previously.12 Briefly, 554 children (91.3%) were
ing data began in November 2007. Of 10,871 chil enrolled after their first urinary tract infection;
dren with screening data available, 1426 (13.1%) the index infection was febrile in 521 children
met the eligibility criteria, of whom 607 (42.6%) (85.8%). The median age was 12 months, and
were enrolled. For 115 of the 607 children en 558 (91.9%) of the children were girls. A total of
rolled who met all the eligibility criteria in the 484 of 602 children (80.4%) had grade II or III
course of the study, corresponding screening vesicoureteral reflux; 287 of 600 children (47.8%)

Table 1. Baseline Demographic and Clinical Characteristics According to Study Group.*

Trimethoprim
Sulfamethoxazole Placebo Overall
Characteristic (N=302) (N=305) (N=607)
Age
Median mo 12 12 12
Interquartile range mo 531 630 631
Group no. (%)
211 mo 151 (50.0) 147 (48.2) 298 (49.1)
1223 mo 48 (15.9) 59 (19.3) 107 (17.6)
2435 mo 43 (14.2) 36 (11.8) 79 (13.0)
3672 mo 60 (19.9) 63 (20.7) 123 (20.3)
Sex no. (%)
Male
Circumcised 11 (3.6) 7 (2.3) 18 (3.0)
Uncircumcised 14 (4.6) 17 (5.6) 31 (5.1)
Female 277 (91.7) 281 (92.1) 558 (91.9)
Race no./total no. (%)
White 245/298 (82.2) 237/299 (79.3) 482/597 (80.7)
Black 10/298 (3.4) 17/299 (5.7) 27/597 (4.5)
Multiracial 20/298 (6.7) 20/299 (6.7) 40/597 (6.7)
Other 23/298 (7.7) 25/299 (8.4) 48/597 (8.0)
Hispanic ethnic group no./total no. (%) 31/300 (10.3) 46/304 (15.1) 77/604 (12.7)
Educational level of primary caregiver no./total no. (%)
High-school graduate or lower 78/299 (26.1) 80/303 (26.4) 158/602 (26.2)
Some college or 2-yr degree 78/299 (26.1) 78/303 (25.7) 156/602 (25.9)
College graduate or higher 143/299 (47.8) 145/303 (47.9) 288/602 (47.8)
Health insurance no./total no. (%)
Commercial 228/299 (76.3) 211/302 (69.9) 439/601 (73.0)
Public 71/299 (23.7) 91/302 (30.1) 162/601 (27.0)
Index UTI no. (%)
First episode 275 (91.1) 279 (91.5) 554 (91.3)
Second episode 27 (8.9) 26 (8.5) 53 (8.7)
Febrile only 96 (31.8) 100 (32.8) 196 (32.3)
Symptomatic only 49 (16.2) 37 (12.1) 86 (14.2)
Febrile and symptomatic 157 (52.0) 168 (55.1) 325 (53.5)

2370 n engl j med 370;25nejm.orgjune 19, 2014

The New England Journal of Medicine

Downloaded from nejm.org on June 4, 2016. For personal use only. No other uses without permission.
Copyright 2014 Massachusetts Medical Society. All rights reserved.
 Prophylaxis for Children with Vesicoureter al Reflux

Table 1. (Continued.)

Trimethoprim
Sulfamethoxazole Placebo Overall
Characteristic (N=302) (N=305) (N=607)
Susceptibility of index UTI to trimethoprimsulfamethoxazole
no./total no. (%)
Resistant 55/278 (19.8) 65/293 (22.2) 120/571 (21.0)
Sensitive 223/278 (80.2) 228/293 (77.8) 451/571 (79.0)
Bladder and bowel dysfunction no./total no. (%) 34/63 (54.0) 37/63 (58.7) 71/126 (56.3)
Constipation no./total no. (%) 8/65 (12.3) 8/64 (12.5) 16/129 (12.4)
Ultrasound abnormalities no./total no. (%)
Hydronephrosis** 19/300 (6.3) 13/302 (4.3) 32/602 (5.3)
Ureter duplication 18/300 (6.0) 15/302 (5.0) 33/602 (5.5)
Highest degree of vesicoureteral reflux no./total no. (%)
Grade I 35/301 (11.6) 33/301 (11.0) 68/602 (11.3)
Grade II 123/301 (40.9) 131/301 (43.5) 254/602 (42.2)
Grade III 118/301 (39.2) 112/301 (37.2) 230/602 (38.2)
Grade IV 25/301 (8.3) 25/301 (8.3) 50/602 (8.3)
Bilateral vesicoureteral reflux no./total no. (%) 146/300 (48.7) 141/300 (47.0) 287/600 (47.8)
Renal scarring no./total no. (%)
None 280/292 (95.9) 281/290 (96.9) 561/582 (96.4)
Mild 1/292 (0.3) 0/290 1/582 (0.2)
Moderate 3/292 (1.0) 2/290 (0.7) 5/582 (0.9)
Severe 4/292 (1.4) 2/290 (0.7) 6/582 (1.0)
Global atrophy 4/292 (1.4) 5/290 (1.7) 9/582 (1.5)

* There were no significant differences between the study groups in any of the baseline characteristics (P>0.05). UTI denotes urinary tract
infection.
Race and ethnic group were reported by the parent or guardian.
Four children with no insurance were classified as having public insurance.
Included in this category are children whose index infection was caused by enterococcus or pseudomonas, was not tested for suscepti-
bility to trimethoprimsulfamethoxazole, and was assumed to be resistant and children whose index infection was caused by bacteria
that were resistant to trimethoprim but were not tested for susceptibility to sulfamethoxazole.
Bladder and bowel dysfunction was defined as a score of more than 6 for toilet-trained girls and more than 9 for toilet-trained boys on a
modified Dysfunctional Voiding Scoring System.17 Scores range from 0 to 30, with higher scores indicating greater bladder and bowel
dysfunction.
Constipation was defined as two or more conditions according to modified criteria of the Paris Consensus on Childhood Constipation
Terminology Group18 (<3 bowel movements per week, >1 episode of fecal incontinence per week, passage of a large stool that obstructed
the toilet, retentive posture and behavior, and pain during defecation).
** All 32 children had hydronephrosis of less than grade 4 on the Society for Fetal Urology scale.
Seven children with central assessment of no vesicoureteral reflux were included in grade I, and one child with central assessment of
grade V was included in grade IV (with enrollment based on local readings). Data on voiding cystourethrography for five children were
missing because images were not available for central evaluation.
A total of 25 children did not undergo renal scanning with technetium-99mlabeled dimercaptosuccinic acid at baseline. Among 98 chil-
dren (40 in the trimethoprimsulfamethoxazole group and 58 in the placebo group) in whom scanning was performed less than 31 days
after the index UTI, scarring was identified in 1 child in the trimethoprimsulfamethoxazole group. All other instances of scarring were
identified in the 484 children (252 in the trimethoprimsulfamethoxazole group and 232 in the placebo group) in whom scanning was
performed between 31 and 112 days after the index UTI.

had bilateral reflux. Of 126 toilet-trained chil significantly with respect to any baseline charac
dren, 71 (56.3%) had symptoms of bladder and teristic (Table 1).
bowel dysfunction. No significant differences in
sex, race, or ethnic group were apparent between Adherence
enrolled children and those who were screened Parents of 467 of the children (76.9%) reported hav
but not enrolled. The study groups did not differ ing administered the study medication at least 75%

n engl j med 370;25nejm.orgjune 19, 2014 2371

The New England Journal of Medicine
Downloaded from nejm.org on June 4, 2016. For personal use only. No other uses without permission.
Copyright 2014 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

of the time, and parents of 517 children (85.2%) recurrences, 80 (72.1%) were febrile; 2 first re
reported having administered it at least 50% of the
currences took place after 24 months but before
time. Parents of 91 children in the prophylaxis the study exit visit. The risk of febrile or symp
group and of 76 children in the placebo group distomatic recurrences was reduced by half among
continued the study medication. No significant dif
children who received prophylaxis as compared
ference in reported adherence to the study-medicawith those who received placebo, and the differ
tion regimen was noted between the study groups. ence between the two groups widened progres
sively over time (hazard ratio, 0.50; 95% confi
Recurrences of Febrile or Symptomatic dence interval [CI], 0.34 to 0.74) (Fig. 2 and 3).
Urinary Tract Infection The interval between enrollment and a 10% inci
During the trial, 111 children had 171 recurrenc dence of recurrence was 336 days in the prophy
es of urinary tract infection (Table 2, and Table laxis group and 106 days in the placebo group.
S1 in the Supplementary Appendix). Of 111 first Eight children (95% CI, 5 to 16) would have had

Table 2. Clinical Outcomes According to Study Group.

Trimethoprim Absolute Difference in Risk

Outcome Sulfamethoxazole Placebo (95% CI)
no. of children/total no. (%) percentage points
Recurrent febrile or symptomatic UTI*
Children with missing 2-yr data classified 77/302 (25.5) 114/305 (37.4) 11.9 (4.6 to 19.2)
as having had an event (intention-to-
treat analysis)
Children with missing 2-yr data classified 39/302 (12.8) 72/305 (25.4) 12.6 (6.1 to 19.0)
as not having had an event (intention-
to-treat analysis)
Children with missing 2-yr data omitted 39/264 (14.8) 72/263 (27.4) 12.6 (5.7 to 19.5)
Treatment failure 14/302 (5.0) 27/305 (9.6) 4.5 (0.2 to 8.8)
Renal scarring**
Overall 27/227 (11.9) 24/235 (10.2) 1.7 (7.4 to 4.0)
Severe 9/227 (4.0) 6/235 (2.6) 1.4 (4.7 to 1.8)
New 18/220 (8.2) 19/227 (8.4) 0.2 (4.9 to 5.3)
Any cortical defect 29/227 (12.8) 25/235 (10.6) 2.1 (8.0 to 3.7)
Antimicrobial resistance
Resistant Escherichia coli in stool 56/203 (27.6) 41/210 (19.5) 8.1 (16.2 to 0.1)
First recurrent febrile or symptomatic UTI 19/30 (63.3) 11/57 (19.3) 44.0 (64.1 to 24.0)
with resistant E. coli
First recurrent febrile or symptomatic UTI 26/38 (68.4) 17/69 (24.6) 43.8 (61.7 to 25.8)
with any resistant pathogen

* Included are 7 children (3 in the trimethoprimsulfamethoxazole group and 4 in the placebo group) with febrile or
symptomatic UTIs that occurred before a missed 2-year visit. Imputation was applied to 38 children in the trimethoprim
sulfamethoxazole group and 42 children in the placebo group.
P=0.002 by log-rank test stratified according to five core sites.
Percentages are based on KaplanMeier 2-year estimates.
P<0.001 by the log-rank test stratified according to five core sites.
Treatment failure was defined as the occurrence of two febrile UTIs (28 children), one febrile UTI and three symptom-
atic UTIs (2 children), four symptomatic UTIs (2 children), or new or worsening renal scarring (9 children).
P=0.04 by the log-rank test stratified according to five core sites.
** Renal scarring was defined as a decreased uptake of tracer that was associated with loss of contours or the presence
of cortical thinning.
Severe renal scarring was defined as scarring in more than 4 of 12 segments in at least one kidney or global atrophy
characterized by a diffusely scarred and shrunken kidney.
New renal scarring was defined as scarring on the outcome renal scan with technetium-99mlabeled dimercaptosuc-
cinic acid that was not present at baseline.
P<0.001 by the CochranMantelHaenszel test stratified according to five core sites.

2372 n engl j med 370;25nejm.orgjune 19, 2014

The New England Journal of Medicine

Downloaded from nejm.org on June 4, 2016. For personal use only. No other uses without permission.
Copyright 2014 Massachusetts Medical Society. All rights reserved.
 Prophylaxis for Children with Vesicoureter al Reflux

to be treated for 2 years to prevent one case of

febrile or symptomatic urinary tract infection. 40

In our primary intention-to-treat analysis 35

or Symptomatic UTI (%)

with the assumption that children with missing 30

Children with Febrile

primary-outcome data had had an event, a sig 25 Placebo
nificantly smaller proportion of children who
20
received prophylaxis, as compared with those
15
who received placebo, had a febrile or symptom
atic recurrence (25.5% vs. 37.4%; relative risk, 10 TMP-SMX
0.68; 95% CI, 0.53 to 0.87) (Table 2). Similar 5
results were obtained from sensitivity analyses 0
that included children with missing outcome data 0.0 0.5 1.0 1.5 2.0
but assumed they had not had an event (12.8% Years since Randomization
vs. 25.4%; relative risk, 0.55; 95% CI, 0.38 to No. at Risk
0.78) and that omitted children for whom primary- TMP-SMX 302 270 252 244 128
Placebo 305 253 234 214 98
outcome data were missing (14.8% vs. 27.4%;
relative risk, 0.54; 95% CI, 0.38 to 0.77). The Figure 2. Time to First Recurrent Febrile or Symptomatic UTI.
proportion of children meeting the criteria for Shown are KaplanMeier estimates of the cumulative percentage of children
treatment failure in the placebo group was nearly who had a recurrent febrile or symptomatic UTI according to study group.
Fewer children assigned to TMP-SMX prophylaxis had a UTI than children
twice the proportion in the prophylaxis group assigned to placebo (P<0.001 by log-rank test). I bars indicate 95% confi-
(9.6% vs. 5.0%, P=0.035). In separate Cox mod dence intervals.
els, the study-group effect remained unchanged
after we controlled for age, whether the index
urinary tract infection was the childs first or trimethoprimsulfamethoxazole was 63% with
second episode, the degree of baseline reflux, and prophylaxis and 19% with placebo (P<0.001).
the presence or absence of baseline renal scarring
(Table S2 in the Supplementary Appendix). Effect Modifiers
Children with grade III or IV reflux at baseline
Renal Scarring were more likely to have febrile or symptomatic
As reported previously,12 a total of 98 of 582 recurrences than were children with grade I or II
baseline scans were obtained less than 31 days reflux (64 of 280 children [22.9%] vs. 46 of 322
after the index infection (16.8%), and 484 were [14.3%], P=0.003). Hazard ratios for recurrences
obtained between 31 and 112 days after the index consistently favored the prophylaxis group, irre
infection (83.2%). A total of 21 scans (3.6%) were spective of sex, age at study entry, degree of re
interpreted as indicating renal scarring, and 71 flux, whether children had one or two urinary
(12.2%) were interpreted as indicating acute py tract infections before enrollment, whether the
elonephritis. Outcome scans (at the 2-year visit or index infection was caused by a resistant patho
3 to 4 months after the child had met treatment- gen, and whether reflux present at baseline had
failure criteria) showed no significant between- resolved, improved, not changed, or worsened at
group differences in the incidence of renal scar 2 years (Fig. 3). There were significant interac
ring (11.9% in the prophylaxis group and 10.2% tions between study group and the type of index
in the placebo group, P=0.55), severe renal scars infection (P=0.04) and between study group and
(4.0% and 2.6%, respectively; P=0.37), or new re the presence or absence of bladder and bowel
nal scars since baseline (8.2% and 8.4%, respec dysfunction at baseline (P=0.02). The estimated
tively; P=0.94) (Table 2). hazard ratio for a febrile or symptomatic recur
rence in the prophylaxis group relative to the pla
Antimicrobial Resistance cebo group was 0.41 (95% CI, 0.26 to 0.64)
Stool colonization with resistant E. coli was more among children with a febrile index infection
common in the prophylaxis group than in the and 1.28 (95% CI, 0.47 to 3.46) among children
placebo group, but the difference was not sig with a nonfebrile index infection. Corresponding
nificant (Table 2). Among 87 children with a first hazard ratios for children with and those without
febrile or symptomatic recurrence with E. coli, bladder and bowel dysfunction at baseline were
the proportion of isolates that were resistant to 0.21 (95% CI, 0.08 to 0.58) and 1.40 (95% CI, 0.43

n engl j med 370;25nejm.orgjune 19, 2014 2373

The New England Journal of Medicine
Downloaded from nejm.org on June 4, 2016. For personal use only. No other uses without permission.
Copyright 2014 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

No. of Patients Total No. of P Value for

Subgroup with UTI Patients Hazard Ratio (95% CI) Interaction
Overall 111 607 0.50 (0.340.74)
Sex 0.59
Female 107 558 0.51 (0.340.76)
Male 4 49 0.27 (0.032.61)
Age 0.72
<24 mo 65 405 0.52 (0.310.87)
24 mo 46 202 0.45 (0.250.83)
VUR at baseline 0.08
Grade I or II 46 322 0.32 (0.160.61)
Grade III or IV 64 280 0.66 (0.401.09)
VUR at 2 yr 0.93
Resolved 34 218 0.42 (0.200.88)
Improved 17 100 0.52 (0.201.37)
Not improved 25 110 0.49 (0.211.11)
Index UTI 0.11
First episode 94 554 0.43 (0.280.66)
Second episode 17 53 1.01 (0.392.65)
Index UTI type 0.04
Febrile 94 521 0.41 (0.260.64)
Nonfebrile 17 86 1.28 (0.473.46)
Index UTI resistance 0.99
Sensitive to TMP-SMX 77 451 0.49 (0.310.79)
Resistant to TMP-SMX 24 120 0.49 (0.201.20)
BBD at baseline 0.02
Present 25 71 0.21 (0.080.58)
Absent 12 55 1.40 (0.434.58)
BBD during study 0.18
Present 41 154 0.36 (0.180.73)
Absent 24 164 0.76 (0.341.70)
Constipation at baseline 0.78
Present 5 16 0.34 (0.043.31)
Absent 33 113 0.48 (0.230.99)
Constipation during study 0.82
Present 11 51 0.44 (0.121.66)
Absent 58 280 0.52 (0.300.90)
0.01 0.1 1.0 10 100

TMP-SMX Better Placebo Better

Figure 3. Effect of Antimicrobial Prophylaxis on the Risk of Febrile or Symptomatic UTI.

Depicted are hazard ratios (rectangles) and 95% confidence intervals (horizontal lines) for a first recurrent febrile or symptomatic UTI
overall and for subgroups. P values are based on Wald tests for the interaction of subgroup with study-group assignment. Hazard ratios
of less than 1.00 indicate that the risk of a recurrent febrile or symptomatic UTI was lower among the children randomly assigned to
antimicrobial prophylaxis than among those assigned to placebo. BBD denotes bladder and bowel dysfunction.

to 4.58). The study population included too few Adverse Events

boys to evaluate the effect of circumcision, but We found no significant between-group differ
all four boys with a first recurrence were uncir ences in the proportions of children with fever
cumcised at baseline. (14.2% in the prophylaxis group and 18.0% in
the placebo group, P=0.23), otitis media (4.3%
Reflux Status at 2 Years and 7.9%, respectively; P=0.09), diarrhea (3.6%
Voiding cystourethrography was performed at and 6.2%, P=0.19), pharyngitis (6.3% and 4.6%,
2years in 428 children. Reflux was considered to P=0.38), rash (7.6% and 7.5%, P=1.0), or viral
be resolved in 218 children (50.9%), improved in infection (4.6% and 5.2%, P=0.85) (Table S3 in
100 (23.4%), unchanged in 79 (18.5%), and worse the Supplementary Appendix). Hospitalizations
in 31 (7.2%). or emergency room visits also did not differ sig

2374 n engl j med 370;25nejm.orgjune 19, 2014

The New England Journal of Medicine

Downloaded from nejm.org on June 4, 2016. For personal use only. No other uses without permission.
Copyright 2014 Massachusetts Medical Society. All rights reserved.
 Prophylaxis for Children with Vesicoureter al Reflux

nificantly between the prophylaxis and placebo solute risk of symptomatic recurrences of 6 per
groups (41.1% and 48.5%, respectively; P=0.06) centage points (number needed to treat, 14).
(Table S4 in the Supplementary Appendix). However, pyuria was not found at the index in
fection in 32% of the children, only 42% of the
Discussion children had reflux, 41% had normal voiding
studies, 17% did not undergo voiding cystoure
Antimicrobial prophylaxis in children with vesi thrography, and renal scanning with technetium-
coureteral reflux diagnosed after a first or sec 99mlabeled dimercaptosuccinic acid was per
ond urinary tract infection was associated with a formed after 1 year in only 27% of the children.
halving of the risk of febrile or symptomatic re In our trial, all children were younger than 6 years
currences. Differences between the prophylaxis of age, 92% were girls, all were considered to
and placebo groups were apparent early on and have grade I to IV reflux, the diagnosis of infec
increased over a 2-year period. Children with blad tion required the presence of pyuria, renal scan
der and bowel dysfunction at baseline and chil ning was performed after 2 years in 76% of the
dren whose index infection was febrile derived children, and the effect of prophylaxis was sus
particular benefit from prophylaxis, with reduc tained over a 2-year period (number needed to
tions in recurrences of approximately 80% and treat, 8).
60%, respectively. Rates of renal scarring at the Brandstrm et al. randomly assigned 203
outcome visit were low and were not reduced by Swedish children 12 to 23 months of age with
prophylaxis, perhaps because most children were grade III or IV reflux to receive antimicrobial
enrolled after their first infection and because prophylaxis (with trimethoprim, nitrofurantoin,
parents, who were instructed to be vigilant, or cefadroxil), endoscopic treatment, or surveil
sought early medical attention. Not unexpectedly, lance only.20 A total of 37% of the participants
recurrences that did occur in children who re were boys, most of whom were uncircumcised.
ceived prophylaxis were more likely to have been Clinicians and parents were aware of the study-
caused by a resistant pathogen. group assignments, and bag-collected urine spec
We designed our trial to avoid certain limita imens were permitted. Significant differences in
tions of previous studies.2-11,20 We enrolled, from outcome were noted only in girls. Girls in the
a variety of clinical settings, a large, representa surveillance-only group had the highest incidence
tive sample of children with vesicoureteral reflux of febrile recurrence and the shortest time to a
(grade I to IV) that was diagnosed after a uri first recurrence; girls who received antimicrobial
nary tract infection. We applied stringent diag prophylaxis had the lowest incidence of renal
nostic criteria and used standardized scales to scarring at 2 years of follow-up.21
identify predisposing factors.12 Our radiologists Do our results support recent recommenda
applied standardized criteria to enhance the reli tions regarding urinary tract imaging in children
ability of their interpretations of urinary tract with a first febrile urinary tract infection?22
images.15,16 Our trial has certain limitations: our Many factors may influence that decision in in
results may not be applicable to children with dividual children, including the presence or ab
different demographic or clinical characteristics sence of a family history of vesicoureteral reflux,
or children in locales where the choice of tri age, sex, and race, which affect the probability
methoprimsulfamethoxazole may be limited by that vesicoureteral reflux is present; the severity
susceptibility patterns or clinical acceptability, of the clinical course of the infection; the status
and some of our subgroup analyses had limited of bladder and bowel function; the cost of void
statistical power. ing cystourethrography and associated discom
There are both similarities and differences fort; the importance that parents place on pre
between our findings and those of two other venting recurrences (which sometimes result in
trials.10,20 Craig et al. enrolled 576 Australian alarming symptoms, hospitalization, and paren
children under the age of 18 years with stringent tal work loss and which may influence the deci
ly defined urinary tract infection in a placebo- sion to proceed to surgery in children found to
controlled trial of trimethoprimsulfamethoxa have reflux); perceived parental ability to adhere
zole prophylaxis.10 Sixty-four percent of the to a prophylaxis regimen; parental concern about
children were girls. Twelve months of prophy potential long-term effects of prophylaxis, such
laxis was associated with a reduction in the ab as the development of antimicrobial resistance

n engl j med 370;25nejm.orgjune 19, 2014 2375

The New England Journal of Medicine
Downloaded from nejm.org on June 4, 2016. For personal use only. No other uses without permission.
Copyright 2014 Massachusetts Medical Society. All rights reserved.
 Prophylaxis for Children with Vesicoureter al Reflux

and alterations of the microbiome; and most im The content of this report is solely the responsibility of the
authors and does not necessarily represent the official views of
portant, the effectiveness of prophylaxis in pre the National Institute of Diabetes and Digestive and Kidney
venting recurrences. Diseases (NIDDK) or the National Institutes of Health (NIH).
As long as evidence supporting the benefit of Supported by grants from the NIDDK, NIH (U01 DK074059,
U01 DK074053, U01 DK074082, U01 DK074064, U01 DK074062,
prophylaxis was dubious, the recommendation and U01 DK074063); and by Clinical and Translational Science
of a watchful-waiting approach, without perfor Awards from the National Center for Research Resources, now
mance of a voiding cystourethrographic study,22 at the National Center for Advancing Translational Sciences,
NIH (UL1RR024153 and UL1TR000005, to the University of Pitts
seemed reasonable, because the imaging find burgh; and UL1TR000003, to the Childrens Hospital of Phila
ings would not affect the nature of treatment. delphia).
However, our finding that antimicrobial prophy Dr. Hoberman reports receiving fees for providing expert tes
timony in a legal case involving neonatal hypoglycemia. No
laxis was associated with a reduced risk of re other potential conflict of interest relevant to this article was
currence may warrant reconsideration of that reported.
recommendation. Decision analysis and cost- Disclosure forms provided by the authors are available with
the full text of this article at NEJM.org.
effectiveness analysis may help to clarify the We thank Jack L. Paradise, M.D., for editorial suggestions;
clinical and financial trade-offs, which may help and the study participants, their families, and participating
clinicians and families reach more informed deci physicians, investigators, and staff for making this research
possible.
sions about the advisability of imaging in young
children such as those in our study.

References
1. Hoberman A, Charron M, Hickey RW, 8. Pennesi M, Travan L, Peratoner L, et al. 16. Ziessman HA, Majd M. Importance of
Baskin M, Kearney DH, Wald ER. Imag Is antibiotic prophylaxis in children with methodology on (99m)technetium dimer
ing studies after a first febrile urinary vesicoureteral reflux effective in prevent capto-succinic acid scintigraphic image
tract infection in young children. N Engl J ing pyelonephritis and renal scars? A ran quality: imaging pilot study for RIVUR
Med 2003;348:195-202. domized, controlled trial. Pediatrics 2008; (Randomized Intervention for Children
2. Medical versus surgical treatment of 121(6):e1489-e1494. With Vesicoureteral Reflux) multicenter
primary vesicoureteral reflux: report of the 9. Montini G, Rigon L, Zucchetta P, et al. investigation. J Urol 2009;182:272-9.
International Reflux Study Committee. Prophylaxis after first febrile urinary tract 17. Farhat W, Bgli DJ, Capolicchio G, et al.
Pediatrics 1981;67:392-400. infection in children? A multicenter, ran The dysfunctional voiding scoring system:
3. Birmingham Reflux Study Group. Pro domized, controlled, noninferiority trial. quantitative standardization of dysfunc
spective trial of operative versus non-oper Pediatrics 2008;122:1064-71. tional voiding symptoms in children. J Urol
ative treatment of severe vesicoureteric 10. Craig JC, Simpson JM, Williams GJ, 2000;164:1011-5.
reflux in children: five years observation. et al. Antibiotic prophylaxis and recurrent 18. Benninga M, Candy DC, Catto-Smith
Br Med J (Clin Res Ed) 1987;295:237-41. urinary tract infection in children. N Engl AG, et al. The Paris Consensus on Child
4. Smellie JM, Barratt TM, Chantler C, J Med 2009;361:1748-59. [Erratum, N Engl hood Constipation Terminology (PACCT)
et al. Medical versus surgical treatment J Med 2010;362:1250.] Group. J Pediatr Gastroenterol Nutr 2005;
in children with severe bilateral vesico 11. Hoberman A, Keren R. Antimicrobial 40:273-5.
ureteric reflux and bilateral nephropathy: prophylaxis for urinary tract infection in 19. Performance standards for anti
a randomised trial. Lancet 2001;357:1329- children. N Engl J Med 2009;361:1804-6. microbial susceptibility testing: twenty-
33. 12. Carpenter MA, Hoberman A, Mattoo second informational supplement. Wayne,
5. Weiss R, Duckett J, Spitzer A. Results TK, et al. The RIVUR trial: profile and PA: Clinical and Laboratory Standards
of a randomized clinical trial of medical baseline clinical associations of children Institute, 2012.
versus surgical management of infants with vesicoureteral reflux. Pediatrics 2013; 20. Brandstrm P, Esbjrner E, Herthelius
and children with grades III and IV pri 132(1):e34-e45. M, Swerkersson S, Jodal U, Hansson S.
mary vesicoureteral reflux (United States). 13. Chesney RW, Carpenter MA, Moxey- The Swedish reflux trial in children: III.
J Urol 1992;148:1667-73. Mims M, et al. Randomized Intervention Urinary tract infection pattern. J Urol
6. Garin EH, Olavarria F, Garcia Nieto V, for Children With Vesicoureteral Reflux 2010;184:286-91.
Valenciano B, Campos A, Young L. Clini (RIVUR): background commentary of 21. Brandstrm P, Nevus T, Sixt R, Stok
cal significance of primary vesicoureteral RIVUR investigators. Pediatrics 2008;122: land E, Jodal U, Hansson S. The Swedish
reflux and urinary antibiotic prophylaxis Suppl 5:S233-S239. reflux trial in children. IV. Renal damage.
after acute pyelonephritis: a multicenter, 14. Keren R, Carpenter MA, Hoberman A, J Urol 2010;184:292-7.
randomized, controlled study. Pediatrics et al. Rationale and design issues of the 22. Subcommittee on Urinary Tract Infec
2006;117:626-32. Randomized Intervention for Children tion, Steering Committee on Quality Im
7. Roussey-Kesler G, Gadjos V, Idres N, With Vesicoureteral Reflux (RIVUR) study. provement and Management. Urinary tract
et al. Antibiotic prophylaxis for the pre Pediatrics 2008;122:Suppl 5:S240-S250. infection: clinical practice guideline for
vention of recurrent urinary tract infec 15. Greenfield SP, Carpenter MA, Chesney the diagnosis and management of the ini
tion in children with low grade vesicoure RW, Zerin JM, Chow J. The RIVUR voiding tial UTI in febrile infants and children 2 to
teral reflux: results from a prospective cystourethrogram pilot study: experience 24 months. Pediatrics 2011;128:595-610.
randomized study. J Urol 2008;179: with radiologic reading concordance. Copyright 2014 Massachusetts Medical Society.
674-9. J Urol 2012;188:Suppl:1608-12.

2376 n engl j med 370;25nejm.orgjune 19, 2014

The New England Journal of Medicine

Downloaded from nejm.org on June 4, 2016. For personal use only. No other uses without permission.
Copyright 2014 Massachusetts Medical Society. All rights reserved.