The Role of Abdominal Radiography in the Diagnosis of IntussusceptionWhen Interpreted by Pediatric Emergency Physicians

Jessica Morrison, MDCM, Nathalie Lucas, MD, FRCP, and Jocelyn Gravel, MD, FRCP, MSc

Objective To evaluate the sensitivity and specificity of abdominal x-rays in the diagnosis of intussusception wheninterpreted by pediatric emergency physicians.Study design This was a prospective experimental study. Participants were board-certified/eligible pediatricemergency physicians. They evaluated a module containing radiographs of 50 cases of intussusception and 50controls, matched for age and sex. For each x-ray, the physicians stated whether the x-ray increased, decreasedor did not affect suspicion of intussusception. The primary outcome was the percentage of cases for which physi-cians stated that the x-ray increased their level of suspicion (sensitivity). Secondary outcomes included the propor-tion of false-negative results and specificity.Results Fourteen of 15 eligible physicians participated in the study. Overall, abdominal radiography increased theindex of suspicion of intussusception in 48% of cases (sensitivity) and 21% of controls; however, in 11% of cases,the abdominal x-rays were incorrectly interpreted as being reassuring. The specificity was 21%. The radiographswere deemed equivocal for 41% of cases and 58% of controls.Conclusions Abdominal x-rays have a low sensitivity and specificity for diagnosing intussusception when inter-preted by pediatric emergency physicians. (J Pediatr 2009;155:556-9).

Abdominal pain is a common chief complaint in pediatric emergency rooms. Intussusception should be included in thedifferential diagnosis because of its potentially significant morbidity and mortality.1,2 The clinical presentation of in-tussusception is highly variable, making a diagnosis on clinical grounds alone difficult. Less than 40% of children

with intussusception have the classic triad of abdominal pain, vomiting, and bloody mucous stool.3 Abdominal x-rays oftenare the first diagnostic test ordered, because ultrasound, a highly accurate diagnostic modality,4 often is not readily availableto emergency room physicians. A 1999 survey of members of the European Society of Pediatric Radiology revealed that approx-imately 72.5% of participants routinely requested abdominal x-rays in cases of suspected intussusception.5 The diagnosticreliability of abdominal x-rays is a subject of great controversy in the current medical literature, however.2,6-13

The vast majority of existing studies in the literature do not involve emergency room physicians as participants. Radi-ologists are typically those who interpret x-rays to assess sensitivity and specificity. This is of particular importance becausein most cases, a radiologist’s interpretation is not available during the clinical decision making process. Emergency roomphysicians view the initial x-rays and determine appropriate patient management based in part on these images. Our ob-jective in the present study was to evaluate the sensitivity and specificity of abdominal x-rays in the diagnosis of intus-susception when interpreted by pediatric emergency room physicians, and also to investigate the impact of these x-rayson clinical management.

Methods

This was a prospective experimental study that aimed to evaluate the ability of pediatric emergency room physicians todiagnose cases of intussusception based on abdominal radiographs. The study was conducted in 2008 at a tertiary care,university-affiliated pediatric hospital that has 60 000 annual visits to its emergency department. Participants included12 full-time pediatric emergency physicians and 2 pediatric emergency fellows in their final month of training. Allparticipants were either certified by or board-eligible as fellow in Pediatric Emergency Medicine by the Royal Collegeof Surgeons and Physicians of Canada.

Cases of intussusception were selected through the computerized pediatricemergency database, which lists the primary diagnoses of patients seen in theemergency department. Search terms included ‘‘invagination’’ and ‘‘intussuscep-tion.’’ Older cases were obtained from a separate computerized database in thearchives. Intussusception was confirmed in all subjects by ultrasound and/orenema. Once 50 consecutive cases were identified, their x-rays were obtained

From the Pediatrics Residency Program (J.M.) andDivision of Emergency Medicine (N.L., J.G.), Departmentof Pediatrics; CHU Sainte-Justine, University ofMontreal, Montreal, Quebec, Canada

Supported by a grant from the Canadian Association ofEmergency Physicians. The authors declare no conflictsof interest.

0022-3476/$ - see front matter. Copyright � 2009 Mosby Inc.

All rights reserved. 10.1016/j.jpeds.2009.04.006

CI Confidence interval

556

Vol. 155, No. 4 � October 2009

through a computerized database of radiologic images. Thisstudy included cases that presented to our hospital between2004 and 2006. Controls were selected in a similar mannerby using the search term for final diagnosis ‘‘abdominalpain.’’ These were patients who presented to the emergencydepartment with nonspecific abdominal pain unrelated to in-tussusception. All radiology reports of controls were read toensure that no subsequent diagnosis of intussusception wasmade. Cases and controls were matched for age and sex.

ProcedureA work module was created that contained the randomly dis-tributed abdominal x-rays of the 50 cases and 50 controlsmounted on PowerPoint slides. All participants evaluatedthe 100 anonymous abdominal series. They were blinded toclinical history and ultrasound findings and were not in-formed of the number of cases versus controls in the workmodule. For each x-ray, the participants were asked to statewhether the x-ray increased, decreased, or did not affect theirsuspicion of intussusception. The participants also wereasked to specify in which cases they would proceed to ultra-sound, based on their interpretation of the x-rays, in 2 clinicalscenarios, one in which they had a high clinical suspicion ofintussusception and the other in which their level of suspi-cion was low. Our hospital’s Institutional Review Board ap-proved the study design. All physicians were required toprovide informed consent to participate.

The primary outcome measured was the sensitivity of ab-dominal x-rays in diagnosing intussusception. This was cal-culated by determining the percentage of confirmed casesof intussusception for which the participants stated that theassociated x-ray increased their level of suspicion for thiscondition. Many secondary outcomes were evaluated aswell. Specifically, the proportion of false-negative resultswas calculated by taking note of when participants indicatedthat their suspicion of intussusception was decreased when infact intussusception was present. The proportion of ‘‘equivo-cal x-rays,’’ defined as when the radiography had no impacton the clinical index of suspicion, also was calculated. The in-terrater agreement among participants was evaluated by cal-culating the kappa score for multiple raters. Finally, theutility of x-rays was evaluated in 2 clinical scenarios, one inwhich the treating physician had a high clinical suspicionof intussusception and the other in which that level of suspi-cion was low.

Data AnalysisAll data were entered into an Excel database (Microsoft Inc,Richmond, Washington) and analyzed using SPSS version 15(SPSS Inc, Chicago, Illinois). The 95% confidence interval(CI) was determined for each result. Most analyses werereported as descriptive proportions. Previous studies havereported sensitivities ranging from 29% to 89%. The presentstudy aimed to obtain results with a 95% CI width of � 0.09.Based on a conservative estimation of a sensitivity of 50%(worst-case scenario), it was determined that a sample of500 evaluations would provide the desired CI; thus, 50

confirmed cases of intussusception were included in thestudy, and the participation of at least 10 physicians wasdeemed necessary (50 cases � 10 = 500).

Results

Twelve of 13 eligible pediatric emergency physicians partici-pated in the study. The participants had a median of 7 yearsof experience (range, 3 to 28 years) in pediatric emergencymedicine practice. In addition, 2 fellows in pediatric emer-gency medicine in their final month of training were also re-cruited to participate in the study. They were included in thefinal analysis because their results were similar to those of thefull-time emergency physicians.

Overall, abdominal radiography was shown to increasethe index of suspicion of intussusception in 48% of con-firmed cases (95% CI = 44% to 52%) when equivocal x-rays were considered negative (Table I); however, in 11%of cases (95% CI = 9% to 13%), the abdominal x-rayswere incorrectly interpreted as being reassuring. Table IIshows that abdominal x-rays appropriately decreased theprobability of intussusception in 21% of controls (95%CI = 18% to 24%) and wrongly increased the index of sus-picion in another 21% (95% CI = 18% to 24%). The par-ticipants reported that the x-rays neither increased nordecreased their suspicion of intussusception in 41% ofcases (95% CI = 37% to 45%) and 58% of controls(95% CI = 54% to 62%). Tables I and II also showa wide variability in the sensitivity and specificity of the ab-dominal radiographs among the 14 participants. On an in-dividual basis, the sensitivity (22% to 88%), specificity (0to 80%) and false-negative rates (0 to 34%) were quite var-iable among the participants. This high variability is alsoemphasized by the low kappa score (0.1223) for the inter-rater agreement between participants.

Table I. Interpretation of abdominal radiographs byeach participant for the 50 cases

ParticipantsSensitivity*

(95% CI)False-negative†

(95% CI)Equivocalz

(95% CI)

1 0.54 (0.40–0.68) 0.18 (0.07–0.29) 0.28 (0.16–0.40)2 0.42 (0.28–0.56) 0 (0–0.06) 0.58 (0.44–0.72)3 0.40 (0.26–0.54) 0.08 (0.01–0.15) 0.52 (0.38–0.66)4 0.34 (0.21–0.47) 0.16 (0.06–0.26) 0.50 (0.36–0.64)5 0.42 (0.28–0.56) 0.56 (0.42–0.70) 0.02 (0–0.06)6 0.88 (0.79–0.97) 0 (0–0.06) 0.12 (0.03–0.21)7 0.64 (0.51–0.91) 0.08 (0.01–0.15) 0.28 (0.16–0.40)8 0.20 (0.09–0.31) 0 (0–0.06) 0.80 (0.69–0.91)9 0.56 (0.42–0.70) 0.10 (0.02–0.18) 0.34 (0.21–0.47)

10 0.22 (0.11–0.33) 0.04 (0–0.09) 0.74 (0.62–0.86)11 0.32 (0.19–0.45) 0.02 (0–0.07) 0.66 (0.53–0.79)12 0.58 (0.44–0.72) 0.34 (0.21–0.37) 0.08 (0.01–0.15)13 0.70 (0.57–0.83) 0 (0–0.06) 0.30 (0.17–0.43)14 0.50 (0.36–0.64) 0 (0–0.06) 0.50 (0.36–0.54)Total

(n = 700)0.48 (0.44–0.52) 0.11 (0.09–0.13) 0.41 (0.37–0.45)

*Sensitivity when equivocal x-rays were excluded.†Proportion of controls interpreted as intussusception.zProportion of radiographs interpreted as neither clearly positive nor negative for intussus-ception.

557

THE JOURNAL OF PEDIATRICS � www.jpeds.com Vol. 155, No. 4

Figures 1 and 2 show the management approach suggestedby the participants after the interpretation of abdominal ra-diographs given a hypothetical high versus low clinical suspi-cion of intussusception, respectively. Figure 1 shows that foralmost 5% of the situations in which the participant hada high clinical suspicion of intussusception, the abdominalx-ray would have delayed ultrasound and the final diagnosis;however, the number of unnecessary ultrasounds would havedecreased by nearly 10% in the control group after the initialinterpretation of x-rays. For the situations in which theclinical index of suspicion of intussusception was low, theparticipants requested significantly more ultrasounds after

Table II. Interpretation of abdominal radiographs byeach participant for the 50 controls

ParticipantSpecificity*

(95% CI)False-positive†

(95% CI)Equivocalz

(95% CI)

1 0.66 (0.53–0.79) 0.10 (0.02–0.18) 0.24 (0.12–0.36)2 0 (0–0.06) 0.24 (0.12–0.36) 0.76 (0.64–0.88)3 0.20 (0.09–0.31) 0.16 (0.06–0.26) 0.64 (0.51–0.77)4 0.16 (0.06–0.26) 0.40 (0.26–0.54) 0.44 (0.30–0.68)5 0.80 (0.69–0.91) 0.20 (0.09–0.31) 0 (0–0.06)6 0 (0–0.06) 0.54 (0.40–0.68) 0.46 (0.32–0.68)7 0.34 (0.31–0.47) 0.20 (0.09–0.31) 0.44 (0.30–0.69)8 0.04 (0–0.09) 0.02 (0–0.07) 0.94 (0.87–1.0)9 0.26 (0.14–0.38) 0.14 (0.04–0.24) 0.60 (0.46–0.74)

10 0.08 (0.01–0.15) 0.06 (0–0.13) 0.86 (0.76–0.96)11 0 (0–0.06) 0.08 (0.01–0.15) 0.92 (0.84–1.0)12 0.44 (0.30–0.58) 0.36 (0.23–0.49) 0.20 (0.09–0.31)13 0 (0–0.06) 0.16 (0.06–0.26) 0.84 (0.74–0.94)14 0 (0–0.06) 0.32 (0.19–0.45) 0.68 (0.55–0.81)Total

(n = 700)0.21 (0.18–0.24) 0.21 (0.18–0.24) 0.58 (0.54–0.62)

*Specificity when equivocal x-rays were excluded.†Proportion of controls interpreted as intussusception.zProportion of radiographs interpreted as neither clearly positive nor negative for intussus-ception.

Figure 1. Management suggested by study participants afterinterpreting abdominal radiographs for patients with a highclinical suspicion of intussusception.

558

viewing the initial x-rays in cases versus controls (44.7% vs19.2%, a difference of 25.5%; 95% CI = 20.9% to 30.1%).

Discussion

Our data demonstrate that abdominal radiographs inter-preted by pediatric emergency room physicians have an over-all sensitivity of approximately 48% and a false-negative rateof 11%. There is a significant variation in these proportionson an individual basis. These data are compatible with thosefrom several studies in the current literature, where a correctdiagnosis of intussusception was made based on radiographsin approximately half of cases.6,7,12 Reported sensitivitieshave been as high as 89% in other published reports, how-ever.11,13,14

The specificity of abdominal radiography was significantlylower than that reported in the literature. For instance,Meradji et al,13 Smith et al,14 and Sargent et al6 reported spec-ificities of 90%, 58% and 45%, respectively. Moreover, theradiographs were incorrectly interpreted as being normal in11% of cases. This is compatible with the false-negative ratesof 4% to 23% reported in previous studies.15 A possible ex-planation for the low sensitivity and specificity may be relatedto fact that participants had 3 potential interpretations foreach x-ray: increased index of suspicion, decreased index ofsuspicion, and equivocal. We believe that this approachprovides a more realistic picture of emergency room physi-cians’ clinical decision making process.

Figure 2. Management suggested by study participants afterinterpreting abdominal radiographs for patients with a lowclinical suspicion of intussusception.

Morrison, Lucas, and Gravel

October 2009 ORIGINAL ARTICLES

A secondary objective of our study was to determine howabdominal radiography influences subsequent patient man-agement, according to the clinical level of suspicion ofintussusception. Most of the patients at high risk for intussus-ception based on clinical grounds would have received an ul-trasound examination regardless of the abdominalradiography findings; however, in > 4% of the patients, theinterpretation of the x-ray would have delayed the ultrasound(observation).

Unlike in most previous studies, here we evaluated pediat-ric emergency physicians’, not radiologists’, ability to inter-pret abdominal x-rays. This approach is clinically relevant,because it is the emergency physicians who decide whetherfurther investigation is needed, based in part on their evalu-ation of abdominal x-rays. Our results differ from those re-ported in the single previous study that implicatedpediatric emergency physicians. In that study, Smith et al14

reported a sensitivity of 80.5% and a specificity of 58% for6 pediatric emergency physicians evaluating 126 radiographs.Possible explanations for these differences may be related tothe addition of a third possible interpretation (equivocal) inour study, to a disparity in physicians’ skills in evaluatingx-rays, or to differences in the baseline characteristics of thepatients (cases and controls).

Abdominal x-rays may be of benefit if a diagnosis of in-tussusception is considered unlikely. In the present study,when the clinical index of suspicion for intussusceptionwas low, interpretation of abdominal radiographs led tofar fewer requests for ultrasound in the controls than inthe patients with intussusception (19% vs 45%). In addi-tion, fewer patients with intussusception were dischargedcompared with controls (4.3% vs 12.4%). Our interpreta-tion of these results is that a plain x-ray may be usefulto increase the index of suspicion of intussusception, butshould never be used to decrease this index. This conclu-sion is concordant with a recent review of the medical lit-erature that concluded that ‘‘plain abdominal radiographyadds little to the management of patients with suspectedintussusception.’’15

The use of an experimental design may have influenced theparticipants’ interpretations, because they knew that their deci-sions would have no impact on actual patients. In addition,some participants felt uncomfortable evaluating the x-rayswithout having any clinical information relating to the patients.

The study participants were pediatric emergency physi-cians at a single tertiary care center and thus may not reflectthe general population of emergency physicians. This largesample of certified and board-eligible pediatric emergencyphysicians should reflect the standard of care of their prac-tice, however.

The study data indicate that abdominal radiography hasa low sensitivity for intussusception when interpreted bypediatric emergency physicians. A follow-up study in which

The Role of Abdominal Radiography in the Diagnosis of IntussuscEmergency Physicians

pediatric radiologists interpreted the same abdominal filmswould be interesting.

In conclusion, this study demonstrates that abdominalx-rays have a low sensitivity and specificity for intussuscep-tion when interpreted by pediatric emergency physicians.Abdominal radiography should not be performed to excludeintussusception in the emergency department for patientsconsidered at high clinical risk for this pathology. For pa-tients with a low clinical risk, a suggestive radiograph shouldincrease the level of suspicion of intussusception, whereas anormal x-ray should have no influence. Further studiesinvolving pediatric radiologists could explore whether thelow sensitivity is due to suboptimal interpretation of theradiographs by emergency physicians or whether it can be at-tributed to the inherently poor sensitivity of the abdominalx-rays themselves. n

Submitted for publication Dec 16, 2008; last revision received Feb 12, 2009;

accepted Apr 5, 2009.

References

1. McCollough M, Sharieff GQ. Abdominal pain in children. Pediatr Clin

North Am 2006;53:107-37. vi.

2. Hernandez JA, Swischuk LE, Angel CA. Validity of plain films in intus-

susception. Emerg Radiol 2004;10:323-6.

3. Kuppermann N, O’Dea T, Pinckney L, Hoecker C. Predictors of intussus-

ception in young children. Arch Pediatr Adolesc Med 2000;154:250-5.

4. Verschelden P, Filiatrault D, Garel L, Grignon A, Perreault G, Boisvert J,

et al. Intussusception in children: reliability of US in diagnosis. A pro-

spective study. Radiology 1992;184:741-4.

5. Schmit P, Rohrschneider WK, Christmann D. Intestinal intussusception

survey about diagnostic and nonsurgical therapeutic procedures. Pediatr

Radiol 1999;29:752-61.

6. Sargent MA, Babyn P, Alton DJ. Plain abdominal radiography in

suspected intussusception: a reassessment. Pediatr Radiol 1994;

24:17-20.

7. Bolin H. Conventional roentgenography in diagnosis of intussusception

in children. Acta Radiol Diagn (Stockh) 1964;2:32-40.

8. Daneman A, Navarro O. Intussusception, part 1: a review of diagnostic

approaches. Pediatr Radiol 2003;33:79-85.

9. Kirks DR. Diagnosis and treatment of pediatric intussusception: how

far should we push our radiologic techniques? Radiology 1994;191:

622-3.

10. Applegate KE. Intussusception in children: imaging choices. Semin

Roentgenol 2008;43:15-21.

11. Eklof O, Hartelius H. Reliability of the abdominal plain film diagnosis in

pediatric patients with suspected intussusception. Pediatr Radiol 1980;9:

199-206.

12. Levine M, Schwartz S, Katz I, Burko H, Rabinowitz J. Plain film findings

in intussusception. Br J Radiol 1964;37:678-81.

13. Meradji M, Hussain SM, Robben SG, Hop WC. Plain film diagnosis in

intussusception. Br J Radiol 1994;67:147-9.

14. Smith DS, Bonadio WA, Losek JD, Walsh-Kelly CM, Hennes HM,

Glaeser PW, et al. The role of abdominal x-rays in the diagnosis and

management of intussusception. Pediatr Emerg Care 1992;8:325-7.

15. Best evidence topic report: role of plain abdominal radiograph in the di-

agnosis of intussusception. Emerg Med J 2008;25:106-7.

eption When Interpreted by Pediatric 559