ARTICLE IN PRESS

Nomograms for Predicting Annual Resolution Rate of Primary

Vesicoureteral Reflux: Results From 2,462 Children

Carlos R. Estrada, Jr.,* Carlo C. Passerotti, Dionne A. Graham, Craig A. Peters,Stuart B. Bauer, David A. Diamond, Bartley G. Cilento, Jr., Joseph G. Borer,Marc Cendron, Caleb P. Nelson, Richard S. Lee, Jing Zhou, Alan B. Retik andHiep T. NguyenFrom the Department of Urology (CRE, CCP, CAP, SBB, DAD, BGC, JGB, MC, CPN, RSL, ABR, HTN) and Clinical Research Program,Department of Biostatistics (DAG, JZ), Children’s Hospital Boston, Boston, Massachusetts

Purpose: We determined the resolution rate of vesicoureteral reflux and thefactors that influence it to formulate nomograms to predict the probability ofannual resolution for individual cases of reflux.Materials and Methods: We studied 2,462 children with primary vesicoureteralreflux diagnosed between 1998 and 2006. Cox proportional hazards regressionwas used to model time to resolution as a function of statistically significantdemographic and clinical variables. The resulting model was used to constructnomograms predicting the annual cumulative probability of reflux resolution.Results: Multivariate analysis showed that all cases of unilateral reflux resolvedearlier than female bilateral reflux (HR 1.42, p �0.001). Additionally age lessthan 1 year at presentation (HR 1.31, p �0.001), lower reflux grade (2.96,p �0.001 for grade I; 2.28, p �0.001 for grade II; 1.63, p �0.001 for grade III),reflux diagnosed on postnatal evaluation for prenatal hydronephrosis or siblingscreening (1.24, p � 0.002) and single ureter (1.55, p �0.001) were associatedwith significantly earlier resolution of reflux. Specific predicted cumulative prob-abilities of reflux resolution at annual intervals from diagnosis (1 to 5 years) werecalculated for every possible combination of the significant variables.Conclusions: Our analyses demonstrate that resolution of vesicoureteral refluxis dependent on age at presentation, gender, grade, laterality, mode of clinicalpresentation and ureteral anatomy. We constructed nomogram tables containingestimates of annual reflux resolution rate as a function of these variables. Thisinformation is valuable for clinical counseling and management decisions.

Abbreviations

and Acronyms

PNH � prenatal hydronephrosis

RNC � radionuclide cystogram

UDS � urodynamics

US � ultrasound

UTI � urinary tract infection

VCUG � voiding cystourethrogram

VUR � vesicoureteral reflux

Submitted for publication January 27, 2009.Study received institutional review board ap-

proval.* Correspondence: Department of Urology,

Children’s Hospital Boston, 300 Longwood Ave.,HU-370, Boston, Massachusetts 02115 (tele-phone: 617-355-7796; FAX: 617-730-0474; e-mail:carlos.estrada@childrens.harvard.edu).

See Editorial on page 000.

Key Words: nomograms, treatment outcome, vesico-ureteral reflux

MANAGEMENT of vesicoureteral refluxis challenging, and the debate regard-ing its clinical significance underpinsthe lack of consensus on whom, whenand how to treat.1 Reflux is routinelymanaged by prophylactic antibioticsand watchful waiting, although theefficacy of antibiotics has recently beenchallenged in randomized controlled

trials.1,2 Following diagnosis repeat

0022-5347/09/1824-1535/0THE JOURNAL OF UROLOGY®

Copyright © 2009 by AMERICAN UROLOGICAL ASSOCIATION

radiological assessments of reflux areperformed at an interval of 1 to sev-eral years between studies.3,4 This ap-proach is based on the assumption thatmost cases resolve spontaneously withtime.5–7

Three studies from 1987 to 1992were used to formulate the 1997 Pe-diatric Vesicoureteral Reflux Guide-

lines.4,6,8,9 While these guidelines have

Vol. 182, 1535-1541, October 2009Printed in U.S.A.

DOI:10.1016/j.juro.2009.06.053www.jurology.com 1535

NOMOGRAMS FOR PREDICTING RESOLUTION OF PRIMARY VESICOURETERAL REFLUX1536ARTICLE IN PRESS

served as a general reference for predicting the res-olution of VUR, they are based on relatively smallclinical series. Several other studies published re-cently describe the natural history of reflux andprovide data to aid in predicting spontaneous reso-lution.5,10–12 The ability to predict resolution of VURhas become increasingly important due to 1) theavailability of a Food and Drug Administration ap-proved, minimally invasive endoscopic treatmentoption for reflux, which may influence the timing ofsurgical intervention because of its purported sim-plicity and efficacy; 2) the potential but as yet un-proved long-term risks associated with use of pro-phylactic antibiotics, including resistant infectionsand cancer; and 3) the possibility that prophylacticantibiotics are ineffective.1,2,13–16

We sought to define more completely the factorsthat influence the resolution of VUR, and provide ameans to predict reliably the spontaneous resolutionof VUR based on clinical variables that are univer-sally and reliably attainable. This information willaid practitioners and families with clinical manage-ment decisions.

MATERIALS AND METHODS

Institutional review board approval was obtained for thestudy. Patient records were de-identified before data ex-traction.

Patient Database and EvaluationAn electronic longitudinal database of all VUR cases hasbeen maintained at our institution since 1998. All referredcases are entered into the database, and all initial andfollowup data are maintained and managed by a dedicatedclinical research staff member. Inclusion criteria for thestudy were primary VUR and at least 1 followup visit witha cystogram. Primary VUR was differentiated from sec-ondary VUR by exclusion of patients with anatomical orfunctional obstructive processes, eg posterior urethralvalves and spina bifida, that could cause VUR. We did notexclude patients receiving anticholinergic medications,since these data were not readily available.

All patients underwent initial radiological assessmentwith VCUG, and renal and bladder US. VUR grade (I to V)was assigned according to the International Reflux Studyin Children. Annual followup studies included eitherVCUG or RNC and US.

Complete ureteral duplication was defined as evidenceof such on VCUG and US. The radiology department pro-vided all radiological interpretations, which were con-firmed by review of hospital electronic records before anal-ysis of extracted data. If a discrepancy was noted betweendata in the database and hospital records, the study wasreassessed by a urologist and a consensus was reachedamong investigators. In patients with bilateral VUR therecorded grade reflected the more severe side. Patientswith changes in laterality, ie initial presentation of uni-lateral VUR and subsequent discovery of bilateral or ini-

tial bilateral and subsequent unilateral VUR, were clas-

sified as having bilateral VUR. The primary end point wasspontaneous resolution, which was defined as 1 negativeVCUG or RNC. All data were recorded from the annualvisits until resolution of VUR or surgical intervention, andfrom patients still being followed or those with no furtherfollowup after 2 consecutive VUR studies.

Statistical AnalysisBaseline patient and clinical characteristics were summa-rized with counts and percentages for categorical vari-ables, and means and standard deviations for continuousvariables. Univariate Cox proportional hazards modelswere used to estimate the association between time toVUR resolution and each of the demographic and clinicalvariables. Times to resolution for subjects who underwentsurgery were censored at the last visit preoperatively.Similarly for subjects who were still being followed or hadno further followup time to resolution was censored at thelast visit. The assumption of proportional hazards wasconfirmed graphically. Parametric survival models wereused to evaluate the impact of the interval censoring re-sulting from the yearlong intervals between followup vis-its. The parametric model estimates did not differ mate-rially from the proportional hazards model estimates. Inthe proportional hazards models tied events were handledusing previously described methods.17

Univariate and multivariate analyses were conductedwith the variables male vs female gender, age at presen-tation (younger than 1 year vs 1 year old or older), unilat-eral vs bilateral VUR, VUR grade (I to V), ureteral anat-omy (single vs duplicated) and mode of presentation (PNHvs febrile UTI vs sibling VUR). Patients with nonfebrileUTI were excluded from these analyses, since we do notevaluate these patients for VUR. Univariate proportionalhazards analyses revealed that VUR grades IV and Vcould be collapsed together, and that presentation could begrouped as PNH or sibling VUR vs UTI. Results of explor-atory regression tree analyses showed that dichotomiza-tion of age at 0.83 years yielded the categorization of agewith the optimal prediction of time to resolution. There-fore, age at presentation was dichotomized into youngerthan 1 year vs 1 year old or older in the proportionalhazards models. We initially assigned arbitrary agegroupings of less than 1 year (infancy), 1 to 4 years (nottoilet trained) and older than 4 years (toilet trained) butthese were not statistically valid. Other groupings, includ-ing cases divided into quartiles and each year examinedindividually, also failed to demonstrate significance.

We used multivariate Cox proportional hazards modelsto identify independent risk factors for faster time to VURresolution. Variables were retained if their multivariate pvalue remained less than 0.05. Sensitivity analysis of thefinal model indicated that the parameter estimates werenot affected by informative censoring in the group under-going surgery. In other words, the model was tested todetermine whether inclusion of the surgical group affectedthe nomogram values, and its inclusion did not have asignificant effect. The resulting model was used to con-struct nomograms for the cumulative rate of VUR resolu-tion through time for each possible combination of patientand clinical characteristics. Statistical analyses were per-

formed using SAS® software, version 9.1.

NOMOGRAMS FOR PREDICTING RESOLUTION OF PRIMARY VESICOURETERAL REFLUX 1537ARTICLE IN PRESS

RESULTS

A total of 2,462 patients met study inclusion criteria.Patient data were extracted from January 1998through March 2006. VUR resolved in 1,257 patients(51%), 634 patients (26%) underwent antireflux sur-gery and 571 patients (23%) were still being followedor had no further followup after at least 2 consecutiveVUR evaluations. Mean � SD followup times were1.95 � 1.37 years for patients with resolution of VUR,2.3 � 1.6 years for those who underwent surgery and2.6 � 1.6 years for those who were still being followedor had no further followup. Indications for antirefluxsurgery included breakthrough febrile UTI, renal scar-ring and/or function compromise on scintigraphy, non-resolution and parent preference.

Table 1 shows the overall distribution of patientswith observed spontaneous resolution, prior antirefluxsurgery and ongoing followup/no further followup byVUR grade at presentation. Table 2 contains demo-graphic and descriptive clinical data. There were nosignificant differences in age with respect to gender.Similarly there were no differences in VUR grade ormode of presentation between females and males.

Tables 3 and 4 provide univariate hazard ratiosfor each variable as well as results of the multivar-iate Cox proportional hazards model. The modelbuilding process revealed a significant interaction

Table 1. Outcomes by VUR grade

No. Grade (%)No. Spontaneously

Resolved (%) No. Surgery (%)No. Ongoing or No

Further Followup (%)

I 107 (72) 21 (14) 20 (14)II 422 (61) 110 (16) 156 (23)III 612 (49) 333 (26) 314 (25)IV–V 116 (32) 170 (46) 81 (22)

Total No. (%) 1,257 (51) 634 (26) 571 (23)

Table 2. Demographic and clinical characteristicsof study population

No. gender (%):M 565 (23)F 1,897 (77)

Age � SD (yrs) 2.1 � 2.4No. younger than 1 yr (%) 1,155 (47)No. bilat VUR (%) 1,394 (57)No. VUR grade (%):

I 148 (6)II 688 (28)III 1,259 (51)IV 262 (11)V 105 (4)

No. ureteral duplication (%) 234 (10)No. presentation (%):

UTI 1,830 (74)PNH 308 (13)

Sibling VUR 324 (13)

between gender and laterality. All males, regardlessof laterality, and females with unilateral VUR hadsimilar times to resolution. Therefore, the multivar-iate model contained a dichotomous variable indi-cating whether a patient was male or female withunilateral VUR vs female with bilateral VUR. As agroup, all males and females with unilateral diseasehad VUR resolve faster compared to females withbilateral disease. Additionally, age younger than 1year at presentation, lower grade of VUR, PNH orsibling VUR, and single ureter were associated withsignificantly faster times to VUR resolution.

The cumulative rates of spontaneous resolution ofVUR as estimated by the multivariate Cox model foreach possible combination of gender, VUR laterality,VUR grade, age at presentation, ureteral anatomyand mode of presentation are provided in figure 1.These data were also used to formulate nomogramtables providing predicted rates (as a percentage ofcases) of cumulative annual VUR resolution fromyears 1 to 5 after diagnosis (fig. 2).

DISCUSSION

The ability to predict the clinical behavior of primaryVUR impacts decisions regarding which patients totreat and how best to treat them. It has been reportedthat VUR grade and laterality, ureteral anatomy, gen-der, race, age, mode of presentation, presence of renaldamage and bladder dysfunction influence the sponta-neous resolution of VUR.5,6,10,18–23 Using a largegroup of cases we attempted to study the concurrenteffect of the majority of these variables on the sponta-neous resolution of VUR.

Our analyses yielded several interesting findings.Our modeling demonstrated that age at presentation

Table 3. Unadjusted hazard ratios

Variable Hazard Ratio (95% CI) p Value

M gender 1.31 (1.15–1.50) �0.001Unilat VUR 1.34 (1.20–1.50) �0.001Younger than 1 yr 1.31 (1.17–1.46) �0.001Presentation:

PNH or sibling VUR 1.29 (1.14–1.46) �0.001UTI Reference Not applicable

Single ureter 1.56 (1.24–1.94) �0.001

Table 4. Adjusted hazard ratios

Variable Hazard Ratio (95% CI) p Value

Gender/laterality:M or unilat F 1.42 (1.26–1.59) �0.001Bilat F Reference Not applicable

Younger than 1 yr 1.31 (1.16–1.48) �0.001PNH or sibling VUR 1.24 (1.08–1.42) 0.002

Single ureter 1.55 (1.24–1.93) 0.0001

with b

NOMOGRAMS FOR PREDICTING RESOLUTION OF PRIMARY VESICOURETERAL REFLUX1538ARTICLE IN PRESS

could be dichotomized into younger than 1 year vs 1year old or older, and presentation could be grouped asPNH or sibling VUR vs UTI. These findings are con-sistent with previous data suggesting that VUR diag-nosed in the first year of life has especially high ratesof resolution.24 To our knowledge PNH and siblingVUR have not been previously shown to have anequivalent impact on resolution of VUR. However, ithas been reported that reflux in siblings and refluxdiscovered during postnatal evaluation of PNH mayhave more indolent clinical courses than VUR present-ing with a UTI.21 Additionally our analyses revealedthat boys with unilateral or bilateral VUR and girls

Figure 1. VUR resolution plots. Predicted rates of cumulative anbilateral reflux and girls with unilateral reflux (A), and for girls

Figure 2. VUR resolution nomogram tables. Predicted rates ofconfidence intervals tabularized and expressed as percentage of

reflux and girls with unilateral reflux (A), and for girls with bilateral re

with unilateral VUR had equivalent rates of sponta-neous resolution. In comparison, girls with bilateralVUR had slower rates of resolution.

Knudson et al recently reported a multivariateanalysis of predictive factors of VUR resolution in acohort of 324 children.11 In addition to other factorsconsistent with our data, a bladder volume at VURonset of greater than 50% of predicted capacity wasassociated with earlier spontaneous resolution. Theinclusion of this variable addresses a potentially im-portant factor in the pathophysiology of VUR—blad-der dynamics. Several reports have suggested thatvoiding dysfunction or impaired bladder dynamics

ontaneous resolution of primary VUR in boys with unilateral orilateral reflux (B).

ative annual spontaneous resolution of primary VUR and 95%. Resolution rates are shown for boys with unilateral or bilateral

nual sp

cumulcases

flux (B).

NOMOGRAMS FOR PREDICTING RESOLUTION OF PRIMARY VESICOURETERAL REFLUX 1539ARTICLE IN PRESS

negatively impact the resolution of VUR.19,22,25 At ourinstitution the volume at which VUR is observed is notroutinely documented and, thus, we could not includethis variable. In patients with suspected voiding dys-function, ie those with daytime incontinence after toi-let training, we typically combine UDS with RNC toassess objectively the relationship between VUR andbladder dynamics.

Because of its relatively subjective nature, we like-wise did not include a history of voiding dysfunction asa variable in our model. The impact of objective uro-dynamic data on the present analysis was not deter-mined, and is a subject for further study. Also, patientsundergoing UDS/RNC were not analyzed, since ourcriteria excluded children with potential underlyingbladder pathology and consequent secondary VUR.

We observed a negative effect of a duplicated ureteron spontaneous resolution of VUR. This finding con-tradicts the conventional belief that ureteral duplica-tion does not affect VUR resolution.26 The reason forthis finding is unclear but there are no other knownlarge scale studies on this aspect of VUR. Presence ofrenal damage is an additional variable examined inrelation to spontaneous resolution of VUR. Silva et alperformed a multivariate analysis in 506 Brazilianchildren with VUR and found that among other vari-ables the absence of renal scars predicted resolution ofVUR.19 The importance of renal scarring in the treat-ment of VUR is highlighted by the propensity of pa-tients with renal scarring and reflux who suffer frombreakthrough urinary tract infections.27

The accuracy of a VUR resolution computa-tional model as described by Knudson et al hasbeen improved with the addition of renal scintig-raphy data.12,28 However, we do not use renal scin-tigraphy in a standardized fashion and, therefore,these data were not consistently available for inclu-sion in our analysis. These data would represent areproducible and objective variable, and perhapswould also improve the accuracy of our model.

We compared our resolution plots with those fromthe 1997 Pediatric Vesicoureteral Reflux Guidelines.4

In some instances our data appear more optimisticfor severe VUR and less optimistic for mild VUR.For example in the guidelines the overall predictedresolution of bilateral grade IV reflux at 5 yearsafter presentation was approximately 10%. In ourstudy the 5-year resolution rates varied consider-ably depending on the combination of variables. Therate was approximately 25% in girls 1 year old orolder with bilateral grade IV to V reflux and ureteralduplication who presented with a UTI, compared toapproximately 65% in all boys and girls youngerthan 1 year with unilateral VUR and a single ureterwho presented with PNH or sibling VUR. Converselythe guidelines suggest that more than 90% of pa-

tients with grade I reflux will have spontaneous

resolution by 5 years. By comparison, our data sug-gest that a 60% resolution rate at 5 years can beexpected among girls older than 1 year with bilat-eral grade I reflux, ureteral duplication and UTI.

Because of the difference in the number of vari-ables involved, it is impossible to compare these 2data sets exactly. Moreover, unlike the guidelinesplots, our data are derived from a computationalmodel with a relative underrepresentation of pa-tients at the extremes of the variables studied.Nonetheless, spontaneous resolution of VUR ap-pears to be more varied than previously reported,and the cumulative effects of negative and positivefactors are being discovered. Multi-institutionaltesting of these data will serve to validate them andfurther demonstrate their clinical usefulness.

Our study has limitations, including reliance on acomputational model to address every possible combi-nation of clinical variables. The reported confidenceintervals are reassuring and indicate the statisticalvalidity of the analyses. However, the best approachwould be a large, prospective, multi-institutional studywith close followup for a defined period. Also, our se-ries was retrospective in nature. While the data werecollected longitudinally, they were verified retrospec-tively, which could have introduced error. Another limi-tation was the “all or nothing” nature of the nomograms.In other words, we studied whether VUR resolved withtime but did not account for incremental improvementsand how they would affect ultimate resolution. This isthe subject of a future study.

Finally data were lacking regarding dysfunctionalvoiding in our patients. We attempted to exclude pa-tients with dysfunctional voiding by excluding thosewho underwent combined UDS/RNC. However, with-out standardized definitions or clinical approach in ourgroup we could not have excluded all of these patients.Indeed, the prevalence of voiding dysfunction in thepediatric urological population would suggest that thisvariable was present in some percentage of our studycohort. We look forward to using validated question-naires from which data can be objectively derived andincluded in future nomograms.

CONCLUSIONS

Spontaneous resolution of primary VUR in chil-dren is influenced by a combination of factors thatinclude gender, age at presentation, laterality andgrade of reflux, ureteral anatomy and mode ofpresentation. Based on the combination of theseobjective and universally attainable variables, wewere able to construct annual estimates of spon-taneous VUR resolution from years 1 to 5 afterdiagnosis. This information can aid practitionersand families with clinical management deci-

sions.

NOMOGRAMS FOR PREDICTING RESOLUTION OF PRIMARY VESICOURETERAL REFLUX1540ARTICLE IN PRESS

REFERENCES

1. Garin EH, Olavarria F, Garcia Nieto V et al:Clinical significance of primary vesicoureteral re-flux and urinary antibiotic prophylaxis after acutepyelonephritis: a multicenter, randomized, con-trolled study. Pediatrics 2006; 117: 626.

2. Pennesi M, Travan L, Peratoner L et al: Is anti-biotic prophylaxis in children with vesicoureteralreflux effective in preventing pyelonephritis andrenal scars? A randomized, controlled trial. Pedi-atrics 2008; 121: e1489.

3. Thompson M, Simon SD, Sharma V et al: Timingof follow-up voiding cystourethrogram in childrenwith primary vesicoureteral reflux: developmentand application of a clinical algorithm. Pediatrics2005; 115: 426.

4. Elder JS, Peters CA, Arant BS Jr et al: PediatricVesicoureteral Reflux Guidelines Panel summaryreport on the management of primary vesi-coureteral reflux in children. J Urol 1997; 157:1846.

5. Schwab CW Jr, Wu HY, Selman H et al: Spon-taneous resolution of vesicoureteral reflux: a 15-year perspective. J Urol 2002; 168: 2594.

6. Skoog SJ, Belman AB and Majd M: A nonsurgicalapproach to the management of primary vesi-coureteral reflux. J Urol 1987; 138: 941.

7. Silva JM, Santos Diniz JS, Marino VS et al:Clinical course of 735 children and adolescentswith primary vesicoureteral reflux. Pediatr Neph-rol 2006; 21: 981.

8. Arant BS Jr: Medical management of mild andmoderate vesicoureteral reflux: followup studiesof infants and young children. A preliminary re-port of the Southwest Pediatric Nephrology StudyGroup. J Urol 1992; 148: 1683.

9. Tamminen-Mobius T, Brunier E, Ebel KD et al:Cessation of vesicoureteral reflux for 5 years in

EDITORIAL COMMENTS

erences 11, 12 and 28 in article). The

ment. The International Reflux Study in Children.J Urol 1992; 148: 1662.

10. Zerati Filho M, Calado AA, Barroso U Jr et al:Spontaneous resolution rates of vesicoureteralreflux in Brazilian children: a 30-year experience.Int Braz J Urol 2007; 33: 204.

11. Knudson MJ, Austin JC, McMillan ZM et al:Predictive factors of early spontaneous resolutionin children with primary vesicoureteral reflux.J Urol 2007; 178: 1684.

12. Knudson MJ, Austin JC, Wald M et al: Compu-tational model for predicting the chance of earlyresolution in children with vesicoureteral reflux.J Urol 2007; 178: 1824.

13. Stenberg A and Lackgren G: A new bioimplant forthe endoscopic treatment of vesicoureteral reflux:experimental and short-term clinical results.J Urol 1995; 154: 800.

14. Aaronson IA: Does Deflux alter the paradigm forthe management of children with vesicoureteralreflux? Curr Urol Rep 2005; 6: 152.

15. Velicer CM, Heckbert SR, Lampe JW et al: Anti-biotic use in relation to the risk of breast cancer.JAMA 2004; 291: 827.

16. Conway PH, Cnaan A, Zaoutis T et al: Recurrenturinary tract infections in children: risk factorsand association with prophylactic antimicrobials.JAMA 2007; 298: 179.

17. DeLong DM, Guirguis GH and So YC: Efficientcomputation of subset selection probabilitieswith application to Cox regression. Biometrika1994; 81: 607.

18. Afshar K, Papanikolaou F, Malek R et al: Vesi-coureteral reflux and complete ureteral duplica-tion. Conservative or surgical management?

se factors in-

19. Silva JM, Diniz JS, Lima EM et al: Predictivefactors of resolution of primary vesico-uretericreflux: a multivariate analysis. BJU Int 2006; 97:1063.

20. Upadhyay J, McLorie GA, Bolduc S et al: Naturalhistory of neonatal reflux associated with prena-tal hydronephrosis: long-term results of a pro-spective study. J Urol 2003; 169: 1837.

21. Parekh DJ, Pope JC, Adams MC et al: Outcome ofsibling vesicoureteral reflux. J Urol 2002; 167:283.

22. Yeung CK, Sreedhar B, Sihoe JD et al: Renal andbladder functional status at diagnosis as predic-tive factors for the outcome of primary vesi-coureteral reflux in children. J Urol 2006; 176:1152.

23. Upadhyay J, Bolduc S, Bagli DJ et al: Use of thedysfunctional voiding symptom score to predictresolution of vesicoureteral reflux in children withvoiding dysfunction. J Urol 2003; 169: 1842.

24. Papachristou F, Printza N, Kavaki D et al: Thecharacteristics and outcome of primary vesi-coureteric reflux diagnosed in the first year of life.Int J Clin Pract 2006; 60: 829.

25. Van Arendonk KJ, Madsen MT, Austin JC et al:Nuclear cystometrogram-determined bladder pres-sure at onset of vesicoureteral reflux predicts spon-taneous resolution. Urology 2007; 69: 767.

26. Peppas DS, Skoog SJ, Canning DA et al: Nonsur-gical management of primary vesicoureteral re-flux in complete ureteral duplication: is it justi-fied? J Urol 1991; 146: 1594.

27. Mingin GC, Nguyen HT, Baskin LS et al: Abnormaldimercapto-succinic acid scans predict an in-creased risk of breakthrough infection in childrenwith vesicoureteral reflux. J Urol 2004; 172: 1075.

28. Nepple KG, Knudson MJ, Austin JC et al: Addingrenal scan data improves the accuracy of a com-putational model to predict vesicoureteral reflux

infants and children allocated to medical treat- J Urol 2005; 173: 1725. resolution. J Urol 2008; 180: 1648.

Vesicoureteral reflux is over treated. As with manyconditions, the difficulty remains in determiningwhich children will benefit from treatment. An im-portant component of making this determination onan individual basis is an accurate estimate of thechances of spontaneous resolution. The authors areto be congratulated for their efforts to improve pre-dictability of resolution based on analysis of severalfactors maintained in their large database. In addi-tion to the factors noted, work from our institutionpreviously identified several other factors negativelyaffecting the chances of spontaneous resolution (ref-

clude the presence of dysfunctional voiding, refluxoccurring early during the cystogram and scars ordecreased function in the refluxing kidney. Althoughthese patient characteristics are not in the databasedescribed by the authors, a detailed record review ofa subset of their patients may confirm and furtherelucidate such characteristics as being associatedwith reflux resolution.

Christopher S. Cooper

Division of Pediatric UrologyUniversity of Iowa

Iowa City, Iowa

NOMOGRAMS FOR PREDICTING RESOLUTION OF PRIMARY VESICOURETERAL REFLUX 1541ARTICLE IN PRESS

For at least 2 generations urologists have fixated onVUR cessation. This current work represents themagnum opus of the genre devoted to predictingrates of spontaneous resolution. The omission of dys-functional voiding from the analysis is regrettable,as is the lack of information associating concurrentUTI and persistence of VUR. Still, this work is asdefinitive as any previous publication. As such, thetables will have great usefulness when parents askus to predict the future. I am hopeful that as agroup, urologists can now leave this topic behindand move on to more pertinent issues. What we need

are multivariate analyses describing which children

with or without VUR are at risk for renal damagefrom urinary infection. Given the tenor of the cur-rent literature, which is well cited in this article, weneed evidence-based studies to guide our prescribinghabits. Hopefully in the future we will know not onlythe probability of VUR cessation, but more impor-tantly the likelihood of renal harm and how best toavoid it.

George Steinhardt

Department of UrologyWayne State University

Grand Rapids, Michigan