DOI: 10.1542/peds.2010-1800; originally published online December 13, 2010; 2011;127;119Pediatrics

Gregory E. Tasian and Hillary L. CoppSystematic Review and Meta-analysis

Diagnostic Performance of Ultrasound in Nonpalpable Cryptorchidism: A  

  http://pediatrics.aappublications.org/content/127/1/119.full.html

located on the World Wide Web at: The online version of this article, along with updated information and services, is

 

of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2011 by the American Academy published, and trademarked by the American Academy of Pediatrics, 141 Northwest Pointpublication, it has been published continuously since 1948. PEDIATRICS is owned, PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

by guest on May 9, 2013pediatrics.aappublications.orgDownloaded from

Diagnostic Performance of Ultrasound in NonpalpableCryptorchidism: A Systematic Review and Meta-analysis

abstractCONTEXT: Ultrasound is frequently obtained during the presurgicalevaluation of boys with nonpalpable undescended testes, but its clini-cal utility is uncertain.

OBJECTIVE: To determine the diagnostic performance of ultrasound inlocalizing nonpalpable testes in pediatric patients.

METHODS: English-language articles were identified by searchingMedline, Embase, and the Cochrane Library. We included studies ofsubjects younger than 18 years who had preoperative ultrasound eval-uation for nonpalpable testes and whose testis position was deter-mined by surgery. Data on testis location determined by ultrasoundand surgery were extracted by 2 independent reviewers, fromwhich ultrasound performance characteristics (true-positives, false-positives, false-negatives, and true-negatives) were derived. Meta-analysis of 12 studies (591 testes) was performed by using a random-effects regression model; composite estimates of sensitivity,specificity, and likelihood ratios were calculated.

RESULTS: Ultrasound has a sensitivity of 45% (95% confidence interval[CI]: 29–61) and a specificity of 78% (95% CI: 43–94). The positive andnegative likelihood ratios are 1.48 (95% CI: 0.54–4.03) and 0.79 (95% CI:0.46–1.35), respectively. A positive ultrasound result increases andnegative ultrasound result decreases the probability that a nonpal-pable testis is located within the abdomen from 55% to 64% and 49%,respectively. Significant heterogeneity limited the precision of theseestimates, which was attributable to variability in the reporting ofselection criteria, ultrasound methodology, and differences in the pro-portion of intraabdominal testes.

CONCLUSIONS: Ultrasound does not reliably localize nonpalpable tes-tes and does not rule out an intraabdominal testis. Eliminating the useof ultrasound will not change management of nonpalpable cryp-torchidism but will decrease health care expenditures. Pediatrics2011;127:119–128

AUTHORS: Gregory E. Tasian, MD, MSc, and Hillary L.Copp, MD, MS

Department of Urology, University of California, San Francisco,California

KEY WORDScryptorchidism, undescended testes, nonpalpable, ultrasound,diagnostic performance

ABBREVIATIONSTP—true-positiveFP—false-positiveTN—true-negativeFN—false-negativeCL—confidence limitQUADAS—Quality Assessment of Diagnostic Accuracy StudiesCI—confidence interval

www.pediatrics.org/cgi/doi/10.1542/peds.2010-1800

doi:10.1542/peds.2010-1800

Accepted for publication Sep 13, 2010

Address correspondence to Hillary L. Copp, MD, MS, Departmentof Urology, 400 Parnassus Ave, Suite A-610, Box 0738, SanFrancisco, CA 94143-0738. E-mail: copphl@urology.ucsf.edu

PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).

Copyright © 2011 by the American Academy of Pediatrics

FINANCIAL DISCLOSURE: The authors have indicated they haveno financial relationships relevant to this article to disclose.

Funded by the National Institutes of Health (NIH).

REVIEW ARTICLES

PEDIATRICS Volume 127, Number 1, January 2011 119 by guest on May 9, 2013pediatrics.aappublications.orgDownloaded from

Cryptorchidism (undescended testis)is the most common congenital genito-urinary anomaly in boys and has aprevalence of 1% to 3% in term and15% to 30% in premature male infants;of these testes, 10% to 20% are nonpal-pable.1,2 Cryptorchidism is associatedwith impaired fertility, inguinal hernia,and increased risk of testis cancer.3–5

Primary care providers usually diag-nose cryptorchidism during routinecheckups, at which time the child isreferred to a pediatric urologist orsurgeon who performs surgical cor-rection (orchiopexy), ideally before thechild is 12 months of age.6,7

The operative approach is based onthe palpability of the testis. When thetestis is palpable, an inguinal or pres-crotal orchiopexy is performed. On theother hand, a nonpalpable testis maybe present in the inguinal-scrotal re-gion or within the abdominal cavity, orit may be entirely absent. Surgical ex-ploration is compulsory to localize thetestis when present or confirm an ab-sent testis by revealing blind-endingspermatic vessels or a nonviable nub-bin. However, accurate presurgical di-agnosis of an absent testis wouldspare a child an operation, and correctlocalization of a testis could limit theextent of surgery. For this reason, amajority of pediatricians obtain ultra-sounds to locate undescended testes,especially when the testis is nonpal-pable.8 Observational studies con-ducted with pediatric patients over thelast 25 years have revealed limited andsometimes conflicting diagnostic per-formance characteristics, such assensitivity and specificity, but have notrigorously evaluated the clinical utilityof ultrasound in localizing nonpalpabletestes.9,10

Given the frequent use but uncertainutility of ultrasound in evaluating boyswith cryptorchidism, we performed asystematic review and meta-analysisof published studies to determine if

preoperative ultrasound evaluation issufficient to direct surgical manage-ment of boys with nonpalpable testes.

METHODS

Data Sources

We searched Medline, Embase, and theCochrane Library from their incep-tions to March 2010 for published arti-cles on ultrasound evaluation of chil-dren with nonpalpable undescendedtestes. In consultation with a researchlibrarian experienced in systematic re-views, articles were identified by usingthe following search concepts: “cryp-torchidism,” “nonpalpable testes,” “or-chiopexy,” and “ultrasonography.” Theexplosion feature of each databasewas used, the search was limited toEnglish-language publications, and an-imal studies were excluded (see Ap-pendix for search-strategy details). Inaddition, the bibliographies of all po-tentially relevant primary articles (n�66) and review articles (n� 5) identi-fied in the search were read to identify

other relevant articles not detected inthe database search.

Study Selection

Articles were independently reviewedand selected by both of us. We includedstudies of subjects between 0 and 18years of age who had ultrasound eval-uation for nonpalpable testes andwhose testis position was definitivelydetermined by surgery. We excludedcase reports, reviews, expert opinions,editorials, and studies in which testislocation determined by ultrasoundand/or surgery was not sufficiently de-scribed (Fig 1). To reduce selectionbias, we excluded studies forwhich thereported outcome was the result ofsurgical exploration for subjects whoall had ultrasounds that did not local-ize a testis (n� 1).11

Data Extraction

Working independently, we abstracteddata from the studies that met eligibil-ity criteria (n � 15). Disagreement

FIGURE 1Selection process for study inclusion.

120 TASIAN and COPP by guest on May 9, 2013pediatrics.aappublications.orgDownloaded from

was resolved by discussion, afterwhich consensus was achieved in allcases. Our outcome measures weretestis location as determined by ultra-sound and surgery. For each study, weextracted data on testis location as de-termined by ultrasound and surgeryand entered ultrasound performancecharacteristics (true-positives [TPs],false-positives [FPs], false-negatives[FNs], and true-negatives [TNs]) into2� 2 tables. The location of the testisat the time of surgery was treated asthe gold standard and was the refer-ence test against which ultrasoundperformance was measured. For ex-ample, a testis that was detected byultrasound in the inguinal-scrotal re-gion and was found to be in that loca-tion during surgery was recorded as aTP, whereas a testis that was not visu-alized by ultrasound but was found tobe intraabdominal during surgery wasrecorded as an FN. Testes at or justproximal to the internal inguinal ring(peeping testes) were recorded as in-traabdominal. These tables were con-structed for the overall data set and,when possible, for subgroups. The sub-group analysis was performed to de-termine the sensitivity and specificityof ultrasound in localizing nonpalpableinguinal-scrotal (n � 4 studies) andintraabdominal (n� 5 studies) testes.

Statistical Analysis

For each study, the sensitivity andspecificity of ultrasound in localizingnonpalpable testes were calculated.Pooled estimates of sensitivity, speci-ficity, and positive and negative likeli-hood ratios for the overall data set andthe subgroup analysis were calculatedby using a random-effects regressionmodel using the method of DerSimo-nian and Laird.12 We calculated sensi-tivity as TP/(TP� FN), specificity as TN/(FP � TN), the likelihood ratio for apositive test result as (TP/[TP� FN])/(FP/[FP� TN]), and the likelihood ratiofor a negative result as (FN/[TP� FN])/

(TN/[FP� TN]). The Cain et al,13 Graif etal,14 and Elder9 studies were excludedfrom the regression model, because 0values for 2 or more individual perfor-mance characteristics (eg, FP and TN)precluded calculation of likelihood ra-tios. Interstudy heterogeneity was as-sessed with the I2 score; we consid-ered an I2 score of�30% indicative ofsignificant heterogeneity.15

We used Bayes theorem to calculatethe posterior probability of testis loca-tion after ultrasound evaluation. Pos-terior probability is the probability ofan event (testis location) taking intoaccount other relevant evidence (ul-trasound findings). The pretest proba-bility of testis location was set at themean anatomic distribution of nonpal-pable testes reported in publishedstudies of consecutive subjects inwhom testis location was prospec-tively recorded at the time of surgery.The pretest probabilities were 30% foringuinal-scrotal testes, 55% for intra-abdominal testes, and 15% for absenttestes.16–18 These probabilities wereconverted to pretest odds. The pretestodds and likelihood ratios were usedto determine the posttest odds withthe following formula: posttest odds�pretest odds� likelihood ratio. Poste-rior probability was generated by sub-sequent conversion of the posttestodds to a probability. Because inter-study heterogeneity limited the reli-ability of the likelihood-ratio esti-mates, posterior probabilities werecalculated by using both the point esti-mate and the upper and lower confi-dence limits (CLs) for the positive andnegative likelihood ratios, respectively.This analysis was performed to correctfor the possibility that interstudy hetero-geneity caused the model to underesti-mate ultrasound performance.

Working independently, we assessedthe quality of individual studies by us-ing the Quality Assessment of Diagnos-tic Accuracy Studies (QUADAS) crite-

ria.19 We then performed univariatemeta-regression to explore the a priorihypothesis that interstudy differencesin the QUADAS criteria could accountfor the observed heterogeneity. Eachof the 14 QUADAS criteria was catego-rized as a dichotomous variable (yesversus no/unclear) and included in themeta-regression model. We also deter-mined whether the proportion ofinguinal-scrotal and intraabdominaltestes (determined by surgery) foreach study was within 10 percentagepoints of the proportions reported inthe literature and included this covari-ate in the meta-regression. The afore-mentioned pretest probabilities wereused as the reference values for thisdetermination.16–18 Categorization oftesticular location was performed be-cause inguinal-scrotal testes aremorelikely and intraabdominal testes areless likely to be identified by ultra-sound; therefore, studies with mark-edly different proportions of inguinal-scrotal or intraabdominal testeswould affect the reported sensitivity.

Studies with sensitivity and specificityestimates that were outliers were iden-tified by using the bag plot, as describedby Rousseeuw et al.20 A sensitivity analy-sis was then performed by removing theoutlying studies from the meta-analysisregression model to determine if theyhad a significant effect on the compositeestimates of the performance parame-ters. Publication bias was assessed byfunnel-plot analysis.21 Tests were2-sided, and P � .05 was used as thethreshold for statistical significance.Analyses were performed by using Stata11 (Stata Corp, College Station, TX).

RESULTS

Literature Search

The search yielded 750 unique refer-ences. No additional studieswere iden-tified from review of article refer-ences. No previous systematic reviewswere identified through search of Med-

REVIEW ARTICLES

PEDIATRICS Volume 127, Number 1, January 2011 121 by guest on May 9, 2013pediatrics.aappublications.orgDownloaded from

line, Embase, or the Cochrane Library.Of the 750 articles, 66 (8.8%) were po-tentially relevant. A total of 15 (2%)studies met our eligibility criteria,from which data on 696 nonpalpa-ble testes were extracted (Table1).9,10,13,14,22–32 There were no discrepan-cies in the extracted data or in thequality assessment of the studies be-tween the two of us. Eight (53%) stud-ies were prospective, 6 (40%)were ret-rospective, and not enough detailswere provided by 1 study (7%) to deter-mine study design. A pediatric urolo-gist or pediatric surgeon performedthe preoperative physical examinationof the included subjects in all studies.The examination was performed in theoffice in all studies; however, an exam-ination under anesthesia was also per-formed in the Atlas and Stone,25 Ismailet al,32 and Nijs et al31 studies.

Determination of DiagnosticPerformance

Meta-analysis of 12 studies (591 tes-tes) was performed after the Cain etal,13 Graif et al,14 and Elder9 studieswere excluded because of insufficient

data necessary to calculate likelihoodratios. The composite sensitivity andspecificity of ultrasound in localizing anonpalpable undescended testis were45% (95% confidence interval [CI]: 29–61) and 78% (95% CI: 43–94), respec-tively. The composite likelihood ratiosfor a positive and negative ultrasoundresult were 1.48 (95% CI: 0.54–4.03)and 0.79 (95% CI: 0.46–1.35) (Fig 2), re-spectively. Ultrasound differed in itsability to reliably localize testes in dif-ferent anatomic sites. Ultrasound de-tected 88 (97%) viable inguinal testes,15 (30%) nonviable inguinal “nubbins,”and 21 (38%) intraabdominal testes.The sensitivity and specificity of ultra-sound in detecting nonpalpableinguinal-scrotal testes were 52% (95%CI: 27–75) and 88% (95% CI: 33–99), re-spectively. The sensitivity and specific-ity of ultrasound in detecting nonpal-pable intraabdominal testes were 44%(95% CI: 22–68) and 93% (95% CI: 34–100), respectively.

The pretest probability that a nonpal-pable testis was within the abdomenwas �55%.16–18 Using the likelihood-

ratio point estimates, the posteriorprobability that a nonpalpable testiswas intraabdominal after a positiveand negative ultrasound result was64% and 49%, respectively. When us-ing the upper CL of the positive like-lihood ratio, the posterior probabil-ity that a nonpalpable testis wastruly intraabdominal after an ultra-sound localized a testis within theabdomen was 83%. When using thelower CL of the negative likelihoodratio, the posterior probability that anonpalpable testis was intraabdomi-nal after an ultrasound did not local-ize any testis was 36% (Fig 3).

The sensitivity analysis identified 3 outly-ing studies: Liu et al,29 Shah and Shah,30

and Ismail et al.32 Removing these stud-ies produced similar performance pa-rameters to the complete data-set esti-mates: the composite sensitivity was44% (95% CI: 27–66), and specificity was82% (95% CI: 67–91). These valuesyielded positive and negative likelihoodratiosof2.21 (95%CI: 1.24–3.93)and0.66(95% CI: 0.46–0.96), respectively. Usingthe point estimates from the sensitivity

TABLE 1 Characteristics of Included Studies That Assessed Diagnostic Performance of Ultrasound in Localizing Nonpalpable Undescended Testes

Study Patients,n

Testes,n

Sensitivity,SDa

Specificity,SDa

Inguinal-ScrotalTestes, n (%)

AbdominalTestes, n (%)

AbsentTestes, n (%)

QUADAS CriteriaMet, n (SD)b

Kullendorff et al22 (1985) 12 12 0.83 0.67 5 (42) 1 (8) 6 (50) 9Malone and Guiney23 (1985) 11 14 0.15 1.00 6 (43) 7 (50) 1 (7) 10Weiss et al24 (1986) Unknown 21 0.13 0.85 5 (24) 3 (14) 13 (62) 10Graif et al14 (1990)c Unknown 8 1.00 0.00 NA 2 (25) NA 10Atlas and Stone25 (1992) 3 3 0.00 0.50 1 (33) 0 (0) 2 (67) 11Maghnie et al26 (1994) 17 21 0.75 0.80 10 (48) 6 (28) 5 (24) 10Al-Shareef et al27 (1996) 19 24 0.19 1.00 5 (21) 16 (67) 3 (12) 12Cain et al13 (1996) 64 74 0.64 —d 63 (85) 11 (15) 0 (0) 12Yeung et al28 (1999) 21 23 0.41 1.00 18 (78) 4 (17) 1 (4) 12Liu et al29 (2002) 150 170 0.77 1.00 122 (72) 23 (13) 25 (15) 12Elder9 (2002)c Unknown 21 0.00 1.00 NA 10 (48) NA 11Kanemoto et al10 (2005) 46 55 0.57 1.00 48 (87) 3 (5) 4 (7) 11Shah and Shah30 (2006) 23 23 0.27 0.10 10 (43) 12 (52) 1 (4) 10Nijs et al31 (2007) 135 152 0.71 0.75 108 (71) 35 (23) 9 (6) 12Ismail et al32 (2009) 64 75 0.29 0.10 17 (23) 43 (57) 15 (20) 11Total — 696 0.44 (0.33) 0.70 (0.37) 418 (60) 176 (25) 96 (14) 10.87 (0.99)

NA indicates not applicable.a The sensitivity and specificity values shown are exact calculations.b There are a total of 14 QUADAS criteria, and the number of criteria met by each study are shown to provide a general idea of the quality of the included studies. The QUADAS criteria havenot been validated as a numerical score given that the emphasis placed on each criterion will vary depending on the diagnostic test being assessed.c The surgical locations were not given for all testes, which accounts for the missing data in the columns for inguinal-scrotal and absent testes. The studies did specify the number of TPs,FPs, FNs, and TNs, which allowed for calculation of the overall ultrasound-performance parameters.d Specificity could not be calculated because, no FPs or TNs were reported.

122 TASIAN and COPP by guest on May 9, 2013pediatrics.aappublications.orgDownloaded from

analysis, the posterior probability that anonpalpable testis was intraabdominalafter a positive and negative ultrasoundresult was 73% and 45%, respectively.Using the upper CL of the positive likeli-hood ratio to provide an estimate of ul-trasound’s best performance ability, theposterior probability that a nonpalpable

testis was truly intraabdominal after anultrasound localized a testis within theabdomen was 83%. Using the lower CLof the negative likelihood ratio, theposterior probability that a nonpal-pable testis was intraabdominal af-ter ultrasound did not localize anytestis was 36%.

Study Quality, Heterogeneity, andPublication Bias

The included studies met a median of11 of the 14 QUADAS criteria (range:9–12). The quality items that weremost often not included or not ade-quately reported by the included stud-

FIGURE 2Forest plot of positive (A) and negative (B) likelihood ratios. The dot refers to the point estimate for the likelihood ratio, and the lines represent the 95% CLs.Box size is proportional to the weight given to the study and is based on the study sample size and variance. The vertical line denotes the summary likelihoodratio, and the diamond represents the associated 95% CLs. Note that weights are from random-effects analysis.

REVIEW ARTICLES

PEDIATRICS Volume 127, Number 1, January 2011 123 by guest on May 9, 2013pediatrics.aappublications.orgDownloaded from

ies were descriptions of subject-selection criteria and ultrasoundmethodology (eg, specification of ul-trasound transducer frequency) andhaving the physical examination find-ings available to the ultrasonogra-pher, which is information that shouldbe available in clinical practice.

There was significant interstudy het-erogeneity in the summary positive(I2� 81.7%) and negative (I2� 89.3%)likelihood-ratio estimates. A meta-regression model was used to identifyspecific aspects of the included stud-ies that might have accounted for theobserved heterogeneity. The model in-cluded 15 different study characteris-tics, of which 2 were found to have astatistically significant effect on the

pooled sensitivity estimate: (1) differ-ences between the proportion of intra-abdominal testes (� � .16; 95% CI:0.01–0.3; P � .03); and (2) variabilityin the adequacy of ultrasound-methodology description (� � .64;95% CI: 0.45–0.84; P � .03). One studycharacteristic had a significant effecton the pooled specificity estimate: vari-ability in the adequacy of subject-selection criteria reporting (� � .91;95% CI: 0.76–1.00; P� .03). In the sen-sitivity analysis, heterogeneity waseliminated for the positive likelihoodratio (I2 � 0%) and reduced for thenegative likelihood ratio (I2 � 67.6%).There was no evidence of publicationbias in the included studies (P� .55).

DISCUSSION

Summary of Findings

In this systematic review of 15 studiesand meta-analysis of 12 studies thatassessed the diagnostic performanceof preoperative ultrasound in localiz-ing nonpalpable testes, we found thatultrasound performs poorly as a diag-nostic test. Even using the most gener-ous estimates of ultrasound perfor-mance, ultrasound does not reliablylocalize nonpalpable testes. Ultra-sound cannot differentiate nonviablenubbins from surrounding inguinal tis-sue, and bowel gas often precludes lo-calization of intraabdominal testes. Wecannot fairly evaluate whether ultra-sound changed the operative ap-proach in the studies included in themeta-analysis, because in all studies,except that of Nijs et al,31 the same op-erative approach was performed re-gardless of the ultrasound findings.However, results of the analysis per-formed clearly show that the change inthe probability of the location of a non-palpable testis conferred by ultra-sound was small, and there was still asignificant chance that a testis waspresent after a negative ultrasound re-sult. From this result we conclude that

abdominal-scrotal ultrasound is un-necessary in the preoperative evalua-tion of boys with nonpalpable testis,because it would not change the surgi-cal management of boys with thecondition.

We found significant heterogeneityamong the included studies, whichwas a result of differences in the qual-ity and subject characteristics be-tween studies. This heterogeneity lim-ited the reliability and precision of thecalculated performance-criteria esti-mates, including the likelihood ratios.Indeed, meta-analysis of the 12 in-cluded studies produced CIs for thepositive and negative likelihood-ratioestimates that crossed 1. For this rea-son, we also calculated the posteriorprobability of testis location after ul-trasound by using the likelihood-ratioCLs, which was done tomaximize ultra-sound performance and control forthe possibility that the likelihood-ratiopoint estimates underestimated thetrue performance of ultrasound. Wealso tested the durability of our resultsby performing a sensitivity analysis inwhich the outlying studies were re-moved from the model. When compar-ing these analyses, the posterior prob-abilities were similar when using thelikelihood-ratio point estimates andidentical when using the likelihood-ratio CLs. All analyses revealed that ul-trasound changes the probability oftestis location to a degree far less thanis needed to make surgical decisions.However, given the significant hetero-geneity of the data, we caution againstascribing any specific value to the abilityof ultrasound to increase or decreasethe likelihood of the true location of anonpalpable testis. In addition, althoughwe used 3 representative studies to cal-culate the pretest probability of testis lo-cation, using the proportions of testislocation from other well-designedstudies will produce different poste-rior probabilities.33

FIGURE 3Effect of ultrasound on the probability of testislocation. The pretest probability that a nonpal-pable testis is within the abdomen is 55%. Usingthe positive likelihood-ratio point estimate(solid red line) and upper CL (dashed red line),an ultrasound that localizes a nonpalpable tes-tis within the abdomen increases the probabil-ity that the testis is truly in the abdomen to64% and 83%, respectively. Using the negativelikelihood-ratio point estimate (solid blue line)and lower CL (dashed blue line), an ultrasoundthat does not visualize a nonpalpable testis de-creases the probability that the testis is truly inthe abdomen to 49% and 36%, respectively.

124 TASIAN and COPP by guest on May 9, 2013pediatrics.aappublications.orgDownloaded from

In our exploration of the causes of thisheterogeneity, we found that the re-porting of selection criteria, descrip-tion of ultrasound techniques, anddifferences in the proportions of ab-dominal testes had effects on the sen-sitivity and specificity estimates. Be-cause the reporting of ultrasoundtechnique was poor, it is possible thatdifferent-frequency ultrasound probeswere used, which may have affectedthe ability of ultrasound to differenti-ate a testis from a lymph node. It ispossible that unaccounted differencesin the study subjects, such as obesityor the presence of contralateral testishypertrophy, which suggests an ab-sent or nonviable nonpalpable testis,contributed to the heterogeneity giventhat the selection criteria and subjectcharacteristics were often poorly re-ported.34,35 In addition, the proportionsof subjects with intraabdominal tes-tes reported in the included studiesvaried and, most often, were lowerthan the reference values. Furtherinvestigation revealed that of the 10studies with a proportion of abdom-inal testes that differed from that ofthe reference value, 9 (90%) re-ported proportions of intraabdomi-nal testes of �40%. As found in thesubgroup analysis, abdominal testesare less likely to be identified by ul-trasound. Therefore, the true sensi-tivity of ultrasound in localizing non-palpable testes may be less than theestimates we reported. Finally, althoughnot quantified, it is also likely that ultra-soundoperator technique accounted forsome of the interstudy heterogeneitygiven the operator-dependent nature ofultrasound.

Strengths and Limitations

We used expansive search strategiesto identify all relevant studies that as-sessed the diagnostic performance ofultrasound in localizing nonpalpableundescended testes. To our knowl-edge, this is the first systematic review

of this literature and the first study toquantify, in a statistically rigorousfashion, the diagnostic performanceand clinical utility of ultrasound in lo-calizing nonpalpable testes. As men-tioned, themain limitation is the signif-icant heterogeneity of the includedstudies, which limits the reliability ofthe calculated pooled performancecriteria. However, our regressionmodel accounted for heterogeneity,we explored causes of interstudyvariability, and our analysis met therecommended standards for meta-analyses of observational studies.36

Furthermore, excluding the Liu et al,29

Shah and Shah,30 and Ismail etal32studies eliminated the heterogene-ity in the positive likelihood ratio andreduced the heterogeneity in the neg-ative likelihood ratio. Although the re-maining studies were indeed morealike with respect to the study charac-teristics identified as significant fac-tors in causing the observed heteroge-neity, they were not identical, whichsuggests that other unidentified studycharacteristics contributed to the re-ported sensitivity and specificity be-ing significantly higher in the Liu etal29 study and significantly lower inthe Shah and Shah30 and Ismail et al32

studies than the other studies. Sec-ond, we intentionally excluded meet-ing abstracts from our search strat-egy to reduce data heterogeneitybecause of the high prevalence of in-adequate reporting in abstracts andinconsistencies between abstractsand subsequent full-length articlesof observational studies.37 Also, welimited the search to English-language articles; however, it is un-likely that including articles printedin other languages would have af-fected our results.38 An additionallimitation is the incomplete adher-ence of the included studies to theQUADAS criteria.

Relation to Clinical Practice

Although patients with nonpalpabletestes and ambiguous genitalia shouldhave diagnostic imaging evaluation,the results of our analysis do not sup-port the routine use of ultrasound inthe evaluation of boys with nonpal-pable testes.39 Physical examination,even for obese children, is the mostimportant aspect of the presurgicalassessment of a boy with an unde-scended testis.34 All boys with nonpal-pable undescended testes require sur-gical exploration to bring a viabletestis down to the scrotum, removenonviable testicular tissue identified inthe exploration, or to confirm the con-genital absence of a testis. To accom-plish these goals, the urologist or sur-geon will either perform diagnosticlaparoscopy or an inguinal explora-tion, depending on surgeon prefer-ence.40–44 However, if ultrasound couldreliably determine the presence andlocation of a nonpalpable testis, a childcould be spared an operation (in thesetting of an absent testis) or couldundergo a more limited operation re-stricted to where the testis was seenon ultrasound. We found that ultra-sound did not have the diagnosticpower necessary to change the stan-dard surgical algorithm.

Indeed, reliance on ultrasound find-ings can have severe and deleteriousconsequences. If a urologist decidesnot to operate on a child with a nonpal-pable testis that was not visualized byultrasound, there is still, with the mostconservative estimate (Fig 3), a 36%probability that the testis is within theabdomen. Should this testis bepresent, not operating potentially in-creases the risk of testicular carci-noma, which, given the intraabdominallocation of the testis, places the childat a higher risk for presentation withadvanced disease because of the in-ability to perform routine screeningphysical examinations.5,45

REVIEW ARTICLES

PEDIATRICS Volume 127, Number 1, January 2011 125 by guest on May 9, 2013pediatrics.aappublications.orgDownloaded from

Implications for the Health CareSystem

Among Medicare beneficiaries with can-cer, the cost of diagnostic imaging is ris-ing 5% to 10% per year, which is morethan double the total rate of increaseof Medicare expenditures.46 Althoughlarge-scale studies on the global cost ofdiagnostic imaging in the pediatric pop-ulation have been lacking, imaging is notless expensive per pediatric patient. Atour institution, an abdominal-scrotal ul-trasound to evaluate undescended tes-tes costs $2194. The lack of population-level data makes it difficult to generatean average cost of abdominal-scrotal ul-trasound, because charges vary accord-ing to patient insurance status, region ofthe country, and where the examinationwas performed (hospital outpatient ver-sus freestanding imaging center). How-ever,wedoknow that ultrasound is usedheavily in the presurgical evaluation ofboys with cryptorchidism.8 Given thehigh utilization rate of ultrasoundandanestimated cost per abdominal-scrotal ul-trasound of $500 to $2000, the yearlycost of ultrasound evaluation of unde-scended testes would likely be along theorder of tens of millions of dollars.

The US Department of Health and Hu-man Services recently reported that

health care expenses will increase af-ter implementation of the Patient Pro-tection and Affordable Care Act.47 Wemust develop new ways to decreasecosts and improve resource utiliza-tion. Just as physicians have a duty toprovide diagnostic tests and therapiesthat bring maximum benefit to pa-tients, we also are obligated to identifythose health care resources that con-fer limited benefit to patients but addsignificant cost to the health care sys-tem. Recently, Brody48 proposed a“top-5 list” in which he called for allmedical specialties to identify and rec-ommend against using heavily usedand expensive diagnostic tests that of-fer little benefit for whom they are or-dered. Our findings demonstrate thatultrasound used to evaluate cryp-torchidism meets these criteria. Thus,we recommend against using ultra-sound to evaluate children with cryp-torchidism. Embracing the opportunityto eliminate unnecessary utilization ofhealth care resources will allow physi-cians to continue to advocate for theirpatients and also act as stewards for afragile health care system.

CONCLUSIONS

Ultrasound has no clinical utility inthe routine presurgical evaluation of

boys with nonpalpable testes. Elimi-nating the use of ultrasound in thissetting will not deleteriously affectthose with cryptorchidism and willnot change management of the con-dition, but it will decrease healthcare expenditures.

ACKNOWLEDGMENTSDr Copp is supported by the Univer-sity of California, San Francisco(UCSF) KURe Career Develop-ment Program (NIH grant 1-K12-DK-083021), which did not have any rolein the design or conduct of the study,analysis of data, or drafting of themanuscript. Drs Tasian and Copphad full access to all data in thestudy and take responsibility for theintegrity of the data and accuracy ofthe data analysis. Stephen Bent, MD(assistant professor of medicine,UCSF), provided expert assistance inthe review and editing of the manu-script. Ms Gloria Y. Won, MLIS (librar-ian at the H. M. Fishbon Memorial Li-brary at UCSF Medical Center atMount Zion), assisted in designingand conducting the search strate-gies. Neither Dr Bent nor Ms Wonreceived compensation for theircontributions.

REFERENCES

1. Cryptorchidism: a prospective study of 7500consecutivemale births, 1984–8. John Rad-cliffe Hospital Cryptorchidism Study Group.Arch Dis Child. 1992;67(7):892–899

2. Acerini CL, Miles HL, Dunger DB, Ong KK,Hughes IA. The descriptive epidemiology ofcongenital and acquired cryptorchidism ina UK infant cohort. Arch Dis Child. 2009;94(11):868–872

3. Trussell JC, Lee PA. The relationship of cryp-torchidism to fertility. Curr Urol Rep. 2004;5(2):142–148

4. Lee PA, O’Leary LA, Songer NJ, Coughlin MT,Bellinger MF, LaPorte RE. Paternity after bi-lateral cryptorchidism: a controlled study.Arch Pediatr Adolesc Med. 1997;151(3):260–263

5. Pettersson A, Richiardi L, Nordenskjold A,Kaijser M, Akre O. Age at surgery for un-

descended testis and risk of testicularcancer. N Engl J Med . 2007;356(18):1835–1841

6. Kogan SJ, Tennenbaum S, Gill B, Reda E, Lev-itt SB. Efficacy of orchiopexy by patient age 1year for cryptorchidism. J Urol. 1990;144(2pt 2):508–509, discussion 512–503

7. Park KH, Lee JH, Han JJ, Lee SD, Song SY.Histological evidences suggest recom-mending orchiopexy within the first year oflife for children with unilateral inguinalcryptorchid testis. Int J Urol. 2007;14(7):616–621

8. Tasian GE, Copp HL. Imaging utilization andcryptorchidism: determinants of practicepatterns. J Urol. 2011;185: In press

9. Elder JS. Ultrasonography is unnecessaryin evaluating boys with a nonpalpable tes-tis. Pediatrics. 2002;110(4):748–751

10. Kanemoto K, Hayashi Y, Kojima Y, MaruyamaT, Ito M, Kohri K. Accuracy of ultrasonogra-phy andmagnetic resonance imaging in thediagnosis of non-palpable testis. Int J Urol.2005;12(7):668–672

11. Miyano T, Kobayashi H, Shimomura H,Yamataka A, Tomita T. Magnetic resonanceimaging for localizing the nonpalpable un-descended testis. J Pediatr Surg. 1991;26(5):607–609

12. DerSimonian R, Laird N. Meta-analysis inclinical trials. Control Clin Trials. 1986;7(3):177–188

13. Cain MP, Garra B, Gibbons MD. Scrotal-inguinal ultrasonography: a technique foridentifying the nonpalpable inguinal testiswithout laparoscopy. J Urol. 1996;156(2 pt2):791–794

14. Graif M, Czerniak A, Avigad I, Strauss S, Wolf-

126 TASIAN and COPP by guest on May 9, 2013pediatrics.aappublications.orgDownloaded from

stein I, Itzchak Y. High-resolution sonogra-phy of the undescended testis in childhood:an analysis of 45 cases. Isr J Med Sci. 1990;26(7):382–385

15. Higgins JP, Thompson SG. Quantifying het-erogeneity in a meta-analysis. Stat Med.2002;21(11):1539–1558

16. Merguerian PA, Mevorach RA, Shortliffe LD,Cendron M. Laparoscopy for the evaluationand management of the nonpalpable testi-cle. Urology. 1998;51(5A suppl):3–6

17. Smolko MJ, Kaplan GW, Brock WA. Locationand fate of the nonpalpable testis in chil-dren. J Urol. 1983;129(6):1204–1206

18. Denes FT, Saito FJ, Silva FA, Giron AM,Machado M, Srougi M. Laparoscopic diag-nosis and treatment of nonpalpable testis.Int Braz J Urol. 2008;34(3):329–334; discus-sion 335

19. Whiting P, Rutjes AW, Reitsma JB, BossuytPM, Kleijnen J. The development of QUADAS:a tool for the quality assessment of studiesof diagnostic accuracy included in system-atic reviews. BMC Med Res Methodol. 2003;3:25

20. Rousseeuw PJ, Ruts I, Tukey JW. The bag-plot: a bivariate boxplot. Am Stat. 1999;53(4):382–387

21. Deeks JJ, Macaskill P, Irwig L. The perfor-mance of tests of publication bias and othersample size effects in systematic reviews ofdiagnostic test accuracy was assessed. JClin Epidemiol. 2005;58(9):882–893

22. Kullendorff CM, Hederstrom E, Forsberg L.Preoperative ultrasonography of the unde-scended testis. Scand J Urol Nephrol. 1985;19(1):13–15

23. Malone PS, Guiney EJ. A comparison be-tween ultrasonography and laparoscopy inlocalising the impalpable undescended tes-tis. Br J Urol. 1985;57(2):185–186

24. Weiss RM, Carter AR, Rosenfield AT. Highresolution real-time ultrasonography in thelocalization of the undescended testis. JUrol. 1986;135(5):936–938

25. Atlas I, Stone N. Laparoscopy for evaluationof cryptorchid testis. Urology. 1992;40(3):256–258

26. Maghnie M, Vanzulli A, Paesano P, et al. Theaccuracy of magnetic resonance imagingand ultrasonography compared with surgi-cal findings in the localization of the unde-

scended testis. Arch Pediatr Adolesc Med.1994;148(7):699–703

27. Al-Shareef ZH, Al-Shlash S, Koneru SR, TowuE, Al-Dhohayan A, Al-Brekett K. Laparoscopicorchidopexy: one-stage alternative for non-palpable testes. Ann R Coll Surg Engl. 1996;78(2):115–118

28. Yeung CK, Tam YH, Chan YL, Lee KH, Me-treweli C. A new management algorithm forimpalpable undescended testis with gado-linium enhanced magnetic resonance an-giography. J Urol . 1999;162(3 pt 2):998–1002

29. Liu CS, Chin TW, Wei CF. Impalpable cryp-torchidism: a review of 170 testes. Zhong-hua Yi Xue Za Zhi (Taipei). 2002;65(2):63–68

30. Shah A, Shah A. Impalpable testes: is imag-ing really helpful? Indian Pediatr. 2006;43(8):720–723

31. Nijs SM, Eijsbouts SW, Madern GC, LeymanPM, Lequin MH, Hazebroek FW. Nonpalpabletestes: is there a relationship between ul-trasonographic and operative findings? Pe-diatr Radiol. 2007;37(4):374–379

32. Ismail K, Ashour M, El-Afifi M, et al. Laparos-copy in the management of impalpabletestis: series of 64 cases. World J Surg.2009;33(7):1514–1519

33. Cisek LJ, Peters CA, Atala A, Bauer SB, Dia-mond DA, Retik AB. Current findings in diag-nostic laparoscopic evaluation of the non-palpable testis. J Urol. 1998;160(3 pt 2):1145–1149; discussion 1150

34. Breyer BN, DiSandroM, Baskin LS, HsiehMH.Obesity does not decrease the accuracy oftesticular examination in anesthetized boyswith cryptorchidism. J Urol. 2009;181(2):830–834

35. Belman AB, Rushton HG. Is an empty lefthemiscrotum and hypertrophied right de-scended testis predictive of perinatal tor-sion? J Urol. 2003;170(4 pt 2):1674–1675;discussion 1675–1676

36. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies inepidemiology: a proposal for reporting:Meta-analysis of Observational Studies inEpidemiology (MOOSE) group. JAMA. 2000;283(15):2008–2012

37. Bhandari M, Devereaux PJ, Guyatt GH, et al.An observational study of orthopaedic ab-stracts and subsequent full-text publica-

tions. J Bone Joint Surg Am. 2002;84-A(4):615–621

38. Moher D, Pham B, Klassen TP, et al. Whatcontributions do languages other than En-glish make on the results of meta-analyses?J Clin Epidemiol. 2000;53(9):964–972

39. Kaefer M, Diamond D, Hendren WH, et al. Theincidence of intersexuality in children withcryptorchidism and hypospadias: stratifica-tion based on gonadal palpability andmeatal position. J Urol. 1999;162(3 pt 2):1003–1006, discussion 1006–1007

40. Gapany C, Frey P, Cachat F, et al. Manage-ment of cryptorchidism in children: guide-lines. Swiss Med Wkly. 2008;138(33–34):492–498

41. Moore RG, Peters CA, Bauer SB, Mandell J,Retik AB. Laparoscopic evaluation of thenonpalpable tests: a prospective assess-ment of accuracy. J Urol. 1994;151(3):728–731

42. Lotan G, Klin B, Efrati Y, Bistritzer T. Laparo-scopic evaluation and management of non-palpable testis in children. World J Surg.2001;25(12):1542–1545

43. Papparella A, Parmeggiani P, Cobellis G, etal. Laparoscopic management of nonpal-pable testes: a multicenter study of the Ital-ian Society of Video Surgery in Infancy. JPediatr Surg. 2005;40(4):696–700

44. Snodgrass WT, Yucel S, Ziada A. Scrotal ex-ploration for unilateral nonpalpable testis.J Urol. 2007;178(4 pt 2):1718–1721

45. Walsh TJ, Dall’Era MA, Croughan MS, CarrollPR, Turek PJ. Prepubertal orchiopexy forcryptorchidism may be associated withlower risk of testicular cancer. J Urol. 2007;178(4 pt 1):1440–1446, discussion 1446

46. Dinan MA, Curtis LH, Hammill BG, et al.Changes in the use and costs of diagnosticimaging among Medicare beneficiarieswith cancer, 1999 –2006. JAMA. 2010;303(16):1625–1631

47. Foster R. Estimated financial effects of the“Patient Protection and Affordable Care Actof 2009,” as proposed by the Senate Major-ity Leader on November 18, 2009. Availableat : www.cms.gov/ActuarialStudies/Downloads/S_PPACA_2009-12-10.pdf. Ac-cessed April 27, 2010

48. Brody H. Medicine’s ethical responsibilityfor health care reform: the top five list. NEngl J Med. 2010;362(4):283–285

REVIEW ARTICLES

PEDIATRICS Volume 127, Number 1, January 2011 127 by guest on May 9, 2013pediatrics.aappublications.orgDownloaded from

APPENDIX Representative Search Strategy:Medline (Using PubMed)

1. Cryptorchidism2. Cryptorchism3. Undescended testis[tw]4. Undescended testes[tw]5. Maldescended testis[tw]6. Maldescended testes[tw]7. Nonpalpable testis[tw]8. Nonpalpable testes[tw]9. Non-palpable testis[tw]10. Non-palpable testes[tw]11. Impalpable testis[tw]12. Impalpable testes[tw]13. Orchiopexy14. Orchidopexy15. 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10 or11 or 12 or 13 or 14

16. Ultrasonography17. Ultrasound18. Ultraso*19. Sonogra*20. Echogra*21. 16 or 17 or 18 or 19 or 2022. 15 and 2123. 22 and eng[LA]24. 23 not (animals[mh] not humans[mh])

Note that the same search strategy was used to searchEmbase and the Cochrane library.

128 TASIAN and COPP by guest on May 9, 2013pediatrics.aappublications.orgDownloaded from

DOI: 10.1542/peds.2010-1800; originally published online December 13, 2010; 2011;127;119Pediatrics

Gregory E. Tasian and Hillary L. CoppSystematic Review and Meta-analysis

Diagnostic Performance of Ultrasound in Nonpalpable Cryptorchidism: A  

ServicesUpdated Information &

mlhttp://pediatrics.aappublications.org/content/127/1/119.full.htincluding high resolution figures, can be found at:

References

ml#ref-list-1http://pediatrics.aappublications.org/content/127/1/119.full.htat:This article cites 46 articles, 3 of which can be accessed free

Rs)3Peer Reviews (PPost-Publication

  http://pediatrics.aappublications.org/cgi/eletters/127/1/119

R has been posted to this article: 3One P

Subspecialty Collections

cticehttp://pediatrics.aappublications.org/cgi/collection/office_praOffice Practicethe following collection(s):This article, along with others on similar topics, appears in

Permissions & Licensing

mlhttp://pediatrics.aappublications.org/site/misc/Permissions.xhttables) or in its entirety can be found online at: Information about reproducing this article in parts (figures,

Reprints http://pediatrics.aappublications.org/site/misc/reprints.xhtml

Information about ordering reprints can be found online:

rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.Grove Village, Illinois, 60007. Copyright © 2011 by the American Academy of Pediatrics. All and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elkpublication, it has been published continuously since 1948. PEDIATRICS is owned, published, PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

by guest on May 9, 2013pediatrics.aappublications.orgDownloaded from