ORIGINAL CONTRIBUTION
radiographs, children
Plain Abdominal Radiography
in the Detection of
Major Disease in Children: A Prospective Analysis
From the Departments of Emergency Medicine, Orlando Regional Medical Center, Orlando, Florida;* Riverside General Hospital, Riverside, California;t and Loma Linda University Medical Center, Loma Linda, California;¢ and Emergency Physicians Medical Group, Oakland, California.§
Steven G Rothrock, MD, FACEP* Steven M Green, MD, FACEPt§ Christopher B Hummel, MD*
Receivedfor publication August 20, 1991. Revisions received February 17 and April 23, 1992. Acceptedfor publication May 5, 1992. Presented at the American College of Emergency Physicians Scientific Assembly in Boston, Massachusetts, October 1991.
Study objective: To prospectively evaluate previously described high-yield clinical criteria for obtaining plain abdominal radiographs in the emergency evaluation of children.
Design: Prospective, observational study. Setting: Emergency departments of a university medical center and
an affiliated county hospital. P a r t i c i p a n t s : Three hundred fifty-four children 15 years old or younger who underwent plain abdominal radiography during a one-year period.
Methods and measurements: Physiciansordering plain abdominal radiographs completed data forms that included historical and physical examination information before viewing films. At a later date, records of all patients were reviewed for radiologist interpretation and final diagnosis. The data were analyzed to determine the sensitivity, specificity, and predictive values of previously described high-yield criteria (from a retrospective series) in detecting radiographs that were diagnostic or suggestive of "major" abdominal disease.
Main results: Sixty-one patients (17%) had major diseases potentially requiring procedural intervention (eg, appendicitis, ingested foreign bodies, and intussusception),whereas 296 patients (83%) had minor diseases not requiring procedural intervention (eg, gastroenteritis and nonabdominal diagnoses). The presence of any of the following features--prior abdominal surgery, foreign body ingestion, abnormal bowel sounds, abdominal distention, or peritoneal signs--was 93% sensitive and 40% specific in detecting diagnostic or suggestive radiographs in patients with major disease. Positive and negative predictive values were 11% and 99%, respectively. If only these criteria had been used to obtain radiographs, 38% of films would have been omitted (at an estimated savings of $20,000) with only two suggestive radiographs missed. Conclusion: Our results suggest that restricting abdominal radiographs to patients with at least one of these five high-yield clinical features will detect most diagnostic and suggestive radiographs in children with major abdominal diseases.
[Rothrock SG, Green SM, Hummel CB: Plain abdbminal radiography in the detection of major disease in children: A prospective analysis. Ann ErnergMed December 1992;2I :1423-1429.]
DECEMBER1992 21:12 ANNALS0FEMERGENCYMEDIC]NE
1423/1£
ABDOMINAL
RADIOGRAPHY
Rothrock, Green & Hummel
INTRODUCTION
Plain abdominal radiographs often are recommended routinely in the evaluation of children with abdominal complaints. 1-6 However, recent studies suggest that 30% to 48% of plain abdominal films obtained in children may be unnecessary.7, a Although this modality may be overused in the pediatric population, no prospective studies have defined a subpopulation for which films routinely should be obtained or excluded. A recent retrospective study has suggested that most diagnostic radiographs in children with acute abdominal processes are found in children with one of the following Table 1.
Usefulness of radiography by final diagnosis Final Diagnosis
No.
Radiographic Interpretation Normal or Diagnostic Suggestive Incidental Misleading
Major Foreign body Appendicitis Ventriculoperitoneal shunt malfunction Large-bowel obstruction Small-bowel obstruction Strangulated hernia Intussusception Pyloric stenosis Intra-abdominal trauma Meckel's diverticulum Spontaneous peritonitis Urolithiasis Subtotal
15 15
15 I
0 1
0 11
g 1
10
0
0
10
0
1
0
1
0
0
5 2 4 3
5 0 0 0
0 0 1 3
O 2 3 0
0 0 0 0
2
1
0
1
0
1
O
0
1
0
2 2 61
1 0 23(38%)
0 0 6(10%)
1 2 31 (50%)
0 0 1 (2%)
57
1
25 71 28
1 4 0
22
2
15 16
0 0
Minor Pain of uncertain etiology 58 0 0 Nonabdominal diagnoses* 28 1 1 Gastroenteritis 77 0 2 Constipation 40 8 4 Normal ventriculoperitoneal shunt 24 0 0 Foreign body ruled out 15 0 O Abdominal trauma t 16 0 0 Gastrostomy tube placement 5 1 O Pneumonia 11 0 0 Urinary infection 16 0 0 Ileus 2 0 0 Ovarian cyst 1 0 0 Subtotal 293 10 (3%) 7 (2%) Total 354 33{9%) 13 (4%) *Excluding pneumonia. tNo evidence of intra-abdominalorgan injury in any patient.
20/ 1424
4 10 16 2 I 267 (91%) 298(84%)
0 1 0 0 0 9 (3%) 10(3%)
features: p r i o r abdominal surgery, foreign body ingestion, abdominal distention, peritoneal signs, or a b n o r m a l bowel sounds. 7 Our study attempted to validate prospectively this previously described decision rule. Secondary study goals were to assess the relationship between the ordering physician's suspicion of " m a j o r " disease and the presence of a b n o r m a l radiographs and to compare the reasons for obtaining films with the presence of abnormal radiographs. MATERIALS
AND
METHODS
A prospective, observational study was conducted in the emergency departments of a university medical center and an affiliated county hospital between April 1, 1990, and March 31, 1991. The combined census of both EDs during the study period was 100,000 visits p e r y e a r with approximately 25% of these patients u n d e r the age of 16 years. All encounters in which a supine abdominal view, acute abdominal series (upright and supine views of the abdomen and upright chest), two views of the abdomen (supine and lateral decubitus), or a combined supine chest and abdomen view were obtained in a child under the age of 16 were eligible for inclusion in the study. Emergency attending physicians or residents at the secondor t h i r d - y e a r level completed d a t a forms after each patient evaluation but before viewing the radiographs. This form included the reason for ordering the film, data from the history and physical examination, and the physician's estimate of the probability of m a j o r disease. No effort was made to standardize the presence of subjective physical examination features. Specifically, physicians were instructed to not alter their normal practice p a t t e r n and to use their normal Table 2.
Misleading radiographs Patient No. Radiologic Diagnosis 79 124 143 153
Final Diagnosis
Ventriculoperitoneal shunt disconnection Necrotizing enterocolitis Pyloric stenosis Abdominal mass
Gastroenteritis* Gastroenteritis t Gastroenteritis* Pain of uncertain etiology § 207 Ventricu]operitoneal shunt disconnection Constipation* 237 Constipation Appendicitis 246 Bowel obstruction Pneumonia* 304 Necrotizing enterocolitis Spider bite t 333 Intussusception Gastroenteritis II 354 Intussusception 6astroenteritisll *After computedtomography scan and neurosurgery consultation, both patients were thought to have functioning ventriculoperitoneal shunts. tBoth children lacked risk factors for necrotizing enterocolitis and had symptornatology consistent with gastroenteritis and benign follow-up. *Both patients had presentations consistent with their final diagnosis and were able to take fluids orally and pass stool although no specific gastrointestinal workup was performed. §Mass suspected on initial film was not seen on subsequent ultrasound and KUB. I[ Both had normal barium enemas and clinical course consistent with gastroenteritis (continued fever and diarrhea after initial evaluation),although spontaneousreduction of the intussusception before barium enema cannot be excluded.
ANNALS OF EMERGENCY MEDICINE 21:12 DECEMBER 1992
ABDOMINAL RADIOGRAPHY
Rothrock, Green & Hummel
physical examination techniques to determine if abnormal bowel sounds or peritoneal signs were present. Board-certified radiologists' reports were used to establish the radiographic interpretation of films. Final diagnosis, def'med as the condition responsible for the presenting symptoms, was determined from records of all ED visits and hospitalizations. Final diagnoses were designated as "major" or "minor" before data analysis. Major diagnoses were defined as those that might require acute procedural intervention, including surgery, endoscopy, cystoscopy, or paracentesis. Examples include appendicitis, mechanical bowel obstruction, intussusception, pyloric stenosis, foreign body ingestion, blunt intra-abdominal trauma, malfunctioning ventriculoperitoneal shunt, peritonitis, necrotizing enterocolitis, nephrolithiasis, and toxic megacolon. Minor diagnoses, defined as those not requiring acute procedural intervention, included constipation, nonobstructive ileus, urinary tract infection, gastroenteritis, and nonabdominal medical conditions. Two board-certified emergency physicians reviewed each case and categorized the radiographic interpretation as diagnostic, suggestive, normal, incidental, or misleading with respect to each patient's final diagnosis. Diagnostic was defined as a radiographic finding pathognomonic of a disease (eg, dilated loops of small bowel with multiple air fluid levels on upright or decubitus views and no air noted in the colon in a patient with small bowel obstruction). Suggestive was defined as a radiographic finding associated with but not pathognomonic of a disease (eg, right lower quadrant ileus in appendicitis). Normal was defined as a radiograph lacking any abnormality. Incidental was defined as a radiographic Table 3.
Association of diagnostic, suggestive, normal, and misleading radiographs with reasonsfor obtainingfilms
Indications for Obtaining Radiographs
Diagnostic Diagnostic or or Suggestive Suggestive No. of of Major of Minor Patients Disease Disease Normal
Pain of uncertain etiology 96 Suspect appendicitis 40 Suspect obstruction 37 Evaluate ventriculoperitoneal shunt 34 Suspect foreign body 31 Constipation 22 Vomiting 19 Blunt abdominal trauma 14 6astrointestinal bleeding 12 Suspect intussusception 9 Suspect ureteral stone 6 Other indications 34 Total 354
DECEMBER1992
Misleading
2 1 5
3 1 2
87 38 29
4 O 1
0 15 0 1
0 0 5 g
32 16 17 17
2 0 0 1
0
0
14
0
1
1
9
1
0
0
9
O
0 4 29(8%)
g ,5 17(5%)
6 25 298(84%)
21:12 ANNALS OF EMERGENCYMEDICINE
0 1 10 (3%)
finding thought to be unrelated to the patient's symptoms (eg, bony abnormality in a patient with appendicitis). Misleading was defined as a radiographic finding suggesting significant pathology that was not present (eg, appendicolith and ileus in a patient without appendicitis). The data were analyzed to determine if clinical features were predictive of diagnostic and suggestive radiographs in patients with major disease. Sensitivity, specificity, and positive and negative predictive values were determined for each clinical feature and for combined clinical features from a previous retrospective study. 7 The relationship between the physician's pretest estimate of major disease (expressed as a percentage) and the presence of diagnostic or suggestive radiographs was evaluated using the Pearson product moment correlation. A receiver operating curve was constructed for different values of the physician's pretest estimate of major disease. With this curve, the optimum set point for screening (for diagnostic and suggestive radiographs) was determined. The optimum set point was defined as the lowest cutoff where the sensitivity (of the physician's pretest estimate of major disease) still increased more rapidly than the falsepositive rate. RESULTS Of the 373 eligible patients identified during the study period, charts or data forms from 19 were either incomplete or unavailable, leaving 354 patients for analysis. Mean patient age was 5.2 + 4.6 years. A supine abdominal view was obtained in 173 (49%), acute abdominal series in 118 (33%), two views of the abdomen in 27 (8%), and a single view of the abdomen and chest in 36 (10%). Sixty-one patients (17%) were classified as having major disease, and 293 (83%) were classified as having minor disease. Specific diagnoses in each category are listed (Table 1). With regard to the final diagnosis, 33 radiographs (9%) were categorized as diagnostic, 13 (4%) as suggestive, 298 (84%) as normal or incidental, and ten (3%) as misleading (Table 1). The ten radiographs categorized as misleading were from one patient with major and nine patients with minor disease (Table 2). The number of diagnostic, suggestive, misleading, and normal radiographs are listed with each specified reason for obtaining the radiograph (Table 3). The sensitivity, specificity, and positive and negative predictive values of clinical features in detecting diagnostic or suggestive radiographs in patients with major disease are listed (Table 4). There was a statistically significant correlation between the strength of the physician's suspicion of major disease and the presence of diagnostic and suggestive radiographs (r = .703, P = .023) (Figure 1). A receiver operating curve for the physician's varying estimates of major disease presence is shown (Figure 2). The optimum set point for screening was a more than 40% suspicion of major disease. This cutoff had a sensitivity of 70% and a false-positive rate of 39% for predicting diagnostic or suggestive radiographs.
1425/21
ABDOMINAL RADIOGRAPHY Rothrock, Green & Hummel
DISCUSSION
Plain abdominal radiographs are often used in the evaluation of children presenting to EDs with abdominal complaints. Reported indications for obtaining plain films in children vary from all children with abdominal pain1, 3 to only those with persistent tenderness or distention. 6 Unfortunately, these recommendations a p p e a r to be the opinions of various authors and are not based on published studies. In addition, most children who present to an ED with abdominal pain have a nonspecific or medical cause for their symptoms, making routine abdominal r a d i o g r a p h y potentially unnecessary in a large n u m b e r of patients.9,10 Most studies detailing the usefulness of plain abdominal films have been limited to adults. In a retrospective series of more than 5,000 consecutive patients with acute abdominal pain presenting to an ED in the United Kingdom, Campbell et al found that 60% of films obtained were entirely normal and that positive findings in the remaining 40% were more likely to be misleading than to be helpful, n Brewer et al looked at 1,000 consecutive cases of abdominal pain presenting to a university hospital ED and found that plain radiographic findings never changed a clinician's first diagnostic impression. 12 Several authors have attempted to develop high-yield criteria for obtaining plain abdominal radiographs in adults. Eisenberg et al reviewed 1,780 patients who had undergone abdominal r a d i o g r a p h y and found that limiting radiographs to patients with moderate-to-severe pain, suspected mesenteric ischemia, obstruction, t r a u m a , renal calculi, or gallTable 4. Usefulness of clinical features in detecting radiographs diagnostic or suggestive of major disease Predictive Values (%) Sensitivity Specificity Positive Negative
bladder disease would have decreased the number of ordered films by more than 50% while identifying all serious pathology. 13 A more recent prospective study of 100 consecutive ED patients found that t r a u m a , flank pain with hematuria, ingestion of foreign bodies, previous surgery, and severe acute abdominal pain were all high-yield indications for ordering plain abdominal films. 14 Only one p r i o r study has attempted to define criteria for obtaining plain abdominal radiographs in the pediatric population. 7 Indications for obtaining plain abdominal radiographs are likely to differ between adults and children p r i m a r i l y because acute abdominal conditions differ between age groups. Diseases such as pyloric stenosis, midgut volvulus, and intussusception are more common in children, whereas cholecystitis, mechanical bowel obstruction, abdominal aneurysms, and mesenteric ischemia are more common in adults. P r i o r studies addressing plain abdominal r a d i o g r a p h y in children have suggested that plain radiographs often are overused.7,8 A recent review of 431 consecutive pediatric plain abdominal radiographs found that hmiting plain abdominal r a d i o g r a p h y to patients with either p r i o r abdominal surgery, suspected foreign body ingestion, abdominal distention, peritoneal signs, or abnormal bowel sounds was 100% sensitive and 53% specific in detecting diagnostic radiographs in children with m a j o r abdominal disorders. 7 When the same criteria were applied prospectively to our population, the sensitivity and specificity were 93% and 40%, respectively. Of the 29 patients with m a j o r disease discernible radiographically, only two had none of these five criteria. One of these two patients h a d a typical presentation for appendicitis: fever, vomiting, and right lower q u a d r a n t tenderness. His r a d i o g r a p h revealed an appendicolith. The other patient was a 5-week-old infant with persistent vomiting secondary to pyloric stenosis and gastric distention noted radiographically.
History Age < 2 months Prior abdominal surgery* Foreign body ingestion (or radiopaque substance)* Abdominal pain Vomiting Bilious vomiting •~ Flank pain ~ Hematuria
19 26
92 74
17 8
93 92
56 22 48 15 4 0
93 38 48 96 95 99
39
O
96 86 92 93 92 92
19
50
3
88
30
4 4 27 22 29 0
83 95 81 68 90 ,90
2 7 11 5 14 14
91 92 93 91 95 95
2O
*Any one of five combined criteria 93
40
11
99
3 7 25
6
Physical Examination Features Temperature > 38.3 C Abdominal tenderness Right lower quadrant tenderness Peritoneal signs* Abdominal distention* Altered bowel sounds* 6uaic-positive stool Rectal tenderness
22/1426
Figure 1. Relationship of pretest probability of major disease to the presence of diagnostic or suggestive radiographs % Diagnostic or suggestive of major disease
50
°°t 10
0
10
20 30 40 50 60 70 80 Pretest probability of major disease (%)
90
100
ANNALS OF EMERGENCYMED/CINE 21:12 DECEMBER1992
ABDOMINAL RADIOGRAPHY Rothrock, Green & Hummel
Although the combined criteria were highly sensitive in detecting diagnostic and suggestive radiographs, single features were less valuable. This may be due in part to the large variety of diseases in the group designated as having major disease. The use of a history of prior abdominal surgery in detecting diagnostic and suggestive radiographs appears self-evident. Abdominal surgeries often are associated with obstruction, and bowel obstructions were the second most common major diagnosis seen in our study. Although abdominal radiography has been recommended routinely for all children who have ingested a foreign body,15,16 this approach recently has been questioned, a,17-19 A recent study by Caravati et al indicated that asymptomatic children who have ingested foreign bodies can be followed on an outpatient basis without radiography. 17 Others have recommended routine radiography of the chest and neck but not of the abdomen for all patients with a history of foreign body ingestion. 4,18,19 This approach allows for identification of foreign bodies lodged in the esophagus, trachea, or hypopharynx that can lead to perforation of the esophagus or airway obstruction. Abdominal foreign bodies, however, rarely cause problems in asymptomatic patients, and routine abdominal radiography in these patients may be unnecessary.4,19 We found the presence of abdominal distention, abnormal bowel sounds, and peritoneal signs to be useful for detecting diagnostic or suggestive radiographs in patients with major diseases. This may be related to the high number of patients with mechanical bowel obstructions in addition to functional obstructions (eg, intussusception and pyloric stenosis) in our study. Cases of appendicitis and spontaneous peritonitis also accounted for patients with peritoneal signs. A statistically significant correlation between the ordering physician's pretest suspicion of major disease and the incidence of diagnostic or suggestive radiographs was noted (Pearson's r = .703, P = .023). This finding suggests that the strength of Figure 2. Receiver operating curve
100
% Sensitivity ...->
90
10
. . . - > 40- -~ 30""
80 70 60
.'¢ 70
/;'80
50 40
i
30
J ,~, 90
20 10 i
0 0
i
10
20
I
I
I
i
i
30
40
50
60
70
% False-Positives
DECEMBER 1992
21:12
ANNALS OF EMERGENCY MEDICINE
the physician's suspicion of major disease would be useful in deciding when to order radiographs. Unfortunately, the use of physician suspicion of disease presence is a highly subjective parameter. It is probable that physician interobserver variability would be high with this parameter and not reproducible when applied to other settings. The optimum cutoff for obtaining radiographs was a more than 40% suspicion of major disease. This set point had a 70% sensitivity and 61% specificity in detecting diagnostic or suggestive radiographs in patients with major disease. This cutoff still would have missed 30% of diagnostic and suggestive radiographs, and the false-positive rate (39%) would have been too high to be of any clinical use. Thus, physician pretest estimate of major disease presence was thought to be of little use in predicting diagnostic or suggestive radiographs in patients with major disease. Another limitation of using the physician's estimate of major disease presence for detecting diagnostic or suggestive radiographs is that many major diseases have few or no radiographic findings. For example, Campbell and Gunn found that more than 90% of radiographs in patients with appendicitis were normal or showed incidental findings and that the remainder were more hkely to be misleading than helpful.ll Commonly described positive radiographic findings such as rightward scoliosis, soft-tissue masses, localized ileus, and free peritoneal fluid are als O nonspecific and generally not useful in radiographic diagnosis of appendicitis.2°-23 Even the reported pathognomonic finding of a right lower quadrant appendicolith can be found in patients without appendicitis and thus cannot be depended on to diagnose appendicitis.2 ~,22,24 O t h e r major diseases also have significant rates of normal plain radiography. Radiographs in patients with bowel perforation are normal in 33% to 51% of cases. 25-27 Normal radiographs in patients with necrotizing enterocolitis also approach 50%.28.29 A classic radiographic finding in necrotizing enterocolitis--pneumatosis intestinalis--also can appear, persist, worsen, disappear, or be absent independent of the time course of the disease, thus limiting the sensitivity of radiographs in this disease. 29 Patients with intussusception exhibit normal radiography in only 11% to 25% of cases. 30-32 However, commonly reported positive radiographic findings such as sparse intestinal gas, sparse fecaloid content, a discernible liver edge, or a gas-fluid level in dilated or undilated bowel are not useful in distinguishing between patients with and without intussusception. 3° The sensitivity of plain films in intussusception can be improved if a cross-table supine abdominal view is obtained. 31 However, interpretation of this film requires a radiologist skilled in interpreting this technique. 31 Three of four films in patients with intussusception were normal in our study, although no patient had a cross-table supine view obtained. Pyloric stenosis is another major disease with a substantial incidence (35%) of normal radiography. 33 Riggs and
1 4 2 7 / 2 3
A B D O M I N A L RADIOGRAPHY Rothrock, Green & Hummel
Long examined plain film findings in pyloric stenosis and f o u n d that no single feature was specific for diagnosing this disease, although an algorithm was developed using combinations of seven radiographic c r i t e r i a to separate patients into high-, moderate-, and low-probability categories of pyloric stenosis. 33 If these criteria had been applied to our patient population, the three patients with pyloric stenosis would have been categorized into the m o d e r a t e - p r o b a b i l i t y group. Our rate of 50% normal radiographs in patients with m a j o r disease is consistent with these p r i o r studies. Plain films in patients with abdominal pain of uncertain etiology were n o r m a l in 91% of cases in our study and twice as likely to be misleading as diagnostic of m a j o r disease (Table 3). These findings suggest that plain films often are useless and even confusing in a significant n u m b e r of patients with major disease and undifferentiated abdominal pain. Reliance on plain film r a d i o g r a p h y to exclude m a j o r abdominal diseases can be hazardous and cause serious diagnostic errors. Alternate modalities such as ultrasonography should be considered in evaluating children for specific diseases such as appendicitis, pyloric stenosis, and intussusception. 34-36 The findings of our study are limited by several factors. The classification of diseases as major and minor implies that the radiographic findings of minor disease (eg, constipation or ileus) are not useful to the clinician. However, we believe that major diseases are more important to detect than minor diseases because m a j o r diseases may require acute p r o c e d u r a l intervention. If some clinicians believe that minor diseases are important to identify radiographically, then our criteria would not be applicable to their practice. It also can be argued that certain m a j o r diseases (eg, gastrointestinal foreign bodies) often are insignificant a n d require no treatment. However, we thought that it was more a p p r o p r i a t e to overclassify diseases as being major if there was any possibility that the patient might require a proced u r a l intervention. In the instance of abdominal foreign bodies, the majority will pass through the gastrointestinal tract without causing problems. However, 5% to 10% will require endoscopic or surgical removal.4dsd 9 Thus, foreign body ingestion was categorized as a major disease. Data acquisition for specific physical examination features was not standardized. Physicians were told to use their own interpretation of generally accepted criteria to determine if a b n o r m a l bowel sounds and peritoneal signs were present. We believed that requiring physicians to alter their normal physical examination techniques to comply with data acquisition also might alter results. No standardized radiographic examination was used in this study. The type of film o r d e r e d (eg, supine abdominal view versus three-view examination and so on) was left to the discretion of the physician. Thus, radiographic abnormalities that might require more than a supine abdominal view for diagnosis (eg, small-bowel obstruction or bowel
24/1428
perforation) may have been missed by our study. This factor could not be controlled without obtaining a three-view examination of all patients, exposing them to unnecessary cost and radiation. Our results also might be limited by the patient population. The study hospitals included a tertiary-care center for pediatrics with most subspecialties, including pediatric surgery, critical care, gastroenterology, and neurosurgery, and a large county hospital. Thus, both institutions may treat disproportionately more children with complex abdominal disorders and advanced disease states. No cases of Hirschprung's disease, midgnt volvulus, necrotizing entercolitis, or bowel perforation were present in our study. It is unknown whether the combined criteria would have detected diagnostic or suggestive radiographs in children with these disorders. There also were no cases of biliary tract disease in our study. In adults, radiographs rarely are helpful when biliary tract disease is suspected, although their use in children with biliary disease is uncertain. 37 Due to study limitations, our criteria may not be applicable to all ED settings. However, our results suggest that limiting radiographs to patients with at least one of the five high-yield criteria (ie, p r i o r abdominal surgery, suspected foreign body ingestion, abnormal bowel sounds, abdominal distention, or peritoneal signs) will detect most diagnostic and suggestive radiographs in patients with m a j o r abdominal diseases. If only these criteria h a d been used to obtain radiographs, then 93% of diagnostic and suggestive radiographs would have been detected while eliminating 38% of films ordered. This would have saved a significant amount of time and money (approximately $20,000) and limited radiation exposure to patients. CONCLUSION
A prospective study was u n d e r t a k e n to assess the usefulness of previously described high-yield clinical criteria in detecting diagnostic and suggestive radiographs in children with major abdominal diseases. Abdominal radiographs were normal or misleading in 87% of patients and useful diagnostically (diagnostic or suggestive) in 13% of patients. Fifty percent of abdominal radiographs in patients with maj or abdominal diagnoses also were normal. Physician estimate of the presence of m a j o r disease was found to have poor utility in detecting diagnostic or suggestive radiographs. However, limiting radiographs to children with a history of abdominal surgery, suspicion of foreign body ingestion, a b n o r m a l bowel sounds, abdominal distention, or peritoneal signs would have detected 93% of diagnostic and suggestive radiographs in patients with m a j o r disease while eliminating 38% of films. Although clinical criteria alone cannot replace clinical judgment in the evaluation of patients in the ED, physicians should consider the presence of these five high-yield criteria when determining whether to obtain a plain abdominal radiograph in a child with acute abdominal complaints.
ANNALS OF EMERGENCY MEDICINE 21:12 DECEMBER1992
ABDOMINAL RADIOGRAPHY Rothrock, Green & Hummel
REFERENCES 1. Barkin RM: Abdominal pain in children: Clues to identifying nontraumatic causes. EmergMed Rep 1988;9:9-16. 2. Buchert GS: Abdominal pain in children: An emergency practitioner's guide. Emerg Med Clin North Am 1989;7:497-517.
28. Frey EE, Smith W, Franken EA, et al: Analysis of bowel perforation in necrotizing enterocolitis. Pediatr Radial 1987;17:380-382. 29. Rabinowitz JG, Siegle RL: Changing clinical and roentgenographic patterns of necrotizing enterocolitis. Am J Roentgeno11976;126:560-566.
3. Martin DF: The acute abdomen in childhood. CurrProb Diag Radial 1986;15:340-394.
30. Eklef O, Hartelius H: Reliability of abdominal plain film diagnosis in pediatric patients with suspected intussusception. Pediatr Radial 1980;9:199-206.
4. 0zenoff MB: Emergency radiology in childhood. Emerg Med Clin North Am 1985;3:563-584.
31. Johnson JF, Woisard KK: Ileocolic intussusception: New sign on the supine cross table lateral radiograph. Radiology1989;170:483-486.
5. Singleton EB, Wagner ML: The acute abdomen in the pediatric age group. Semin Roentgeno11973;8:339-356.
32. LeVine M, Schwartz S, Katz I, et al: Plain film findings in intussusception. Br J Radial 1964;37:678-681.
6. Wolf SA, Shermeta DW: Abdominal pain, in Ehrlich FE, Heldrich FJ, Tepas JJ (eds): Pediatric EmergencyMedicine. Rackville, Maryland, Aspen Publishers I nc, 1987, p 213-220.
33. Riggs W, Long L" The value of the plain film roentgenogram in pyloric stenosis. Am J Roentgeno11971;112:77-82.
7. Rothrock SG, Green SM, Harding M, et al: Plain abdominal radiography in the detection of acute medical and surgical disease in children: A retrospective analysis. Pediatr Emerg Care 1991;7:281-285.
34. Schwerk WB, Wichtrup B, Rothmund M, et al: Ultrasoncgraphy in the diagnosis of acute appendicitis: A prospective study. Gastroenterology 1989;97:630-639. 35. Hailer JO, Cohen HL: Hypertrophic pyloric stenosis: Diagnosis using US. Radiology 1986;161:335-339.
8. Cronan K, Shaw K, Bellah R: The utility of abdominal radiography in a pediatric emergency department (abstract). Ann Emerg Med 1990;19:476.
36. See CC, Glassman M, Berezin S, eta[: Emergency ultrasound in the evaluation of acute onset abdominal pain in children. PediatrEmerg Care 1988;4:169-171.
9. Reynolds SL, Jaffe DM: Children with abdominal pain: Evaluation in the pediatric emergency department. PediatrEmerg Care 1990;6:8-12.
37. Rothrock SG, 6oerhuis H, Howard RM: Efficacy of plain abdominal radiography in patients with biliary tract disease. J Emerg Med 1989;8:271-275.
10. Stevenson RJ: Abdominal pain unrelated to trauma. Surg Clin North Am 1985;65:1181-1215. 11. Campbell JPM, Gunn AA: Plain abdominal radiographs and acute abdominal pain. Br J Surg 1988;75:554-556. 12. Brewer RJ, Golden 6% Hitch DC, et al: Abdominal pain: An analysis of 1,000 consecutive cases in a university hospital emergency room. Am J Surg 1976;131:219-223. 13. Eisenberg RL, Heineken P, Hedgcock MW, et aL Evaluation of plain abdominal radiographs in the diagnosis of abdominal pain. Ann Surg 1983;197:464-467.
Address for reprints: Steven G Rothrock, MD Department of Emergency Medicine Orlando Regional Medical Center 1414 Kuhl Avenue Orlando, Florida 32806
14. McCookTA, Ravin CE, Rice RP: Abdominal radiography in the emergency department: A prospective analysis. Ann Emerg Med 1982;11:7-8. 15. Schnaufner L, Mahboubi S: Abdominal emergencies, in Fleischer 6, Ludwig S (eds): Pediatric EmergencyMedicine. Baltimore, Maryland, Williams & Wilkins, 1988, p 936-965. 16.Wesley JR: Managing foreign bodies in various orifices. Emerg Med Rep 1984;5:109-116. 17. Caravati M, Bennett DL, McEIwee NE: Pediatric coin ingestion: Prospective study on the utility of routine roentgenograms. Am J Dis Child 1989;143:549-551. 18. Schunk JE, Corneli H, Bolte R: Pediatric coin ingestions: A prospective study of coin location and symptoms. Am J Dis Child 1989;143:546-548. 19. Hedge D, Techlenburg F, Fleischer 6: Coin ingestion: Does every child need a radiograph? Ann Emerg Mad 1985;14:443-446. 20. Bakhda RK, McNair MM: Useful radiological signs in acute appendicitis in children. Clin Radial 1977;28:193-196. 21. Hatten LE, Miller RC, Hester CL, et al: Appendicitis and the abdominal raentgenogram in children. South MedJ1973;66:803-80& 22. Graham AD, Johnson HF: The incidence of radiographic findings in acute appendicitis compared to 200 normal abdomens. MilitMed 1966;131:272-276. 23. Isdale JM: The radiological signs of acute appendicitis in infancy and childhood. South Aft Med J 1978;53:363-364. 24. Sorer CS: The contribution of the radiologist to the diagnosis of acute appendicitis. Semin Roentgeno11973;8:375-388. 25. Felson B, Wiot JF: Another look at pneumoperitoneum. Semin Roentgenol 1973;8:437-443. 26. Rice RP, Thompson WM, Gedgaudas RK: The diagnosis and significance of extraluminal gas in the abdomen. Radial Clin North Am 1982;20:819-837. 27. Roh J J, Thompson JS, Harned RK, et al: Value of pneumoperitoneum in the diagnosis of visceral perforation. Am J Surg 1983;t46:830-833.
DECEMBER1992
21:12
ANNALS OF EMERGENCY MEDICINE
1429/25
radiographs, children
Plain Abdominal Radiography
in the Detection of
Major Disease in Children: A Prospective Analysis
From the Departments of Emergency Medicine, Orlando Regional Medical Center, Orlando, Florida;* Riverside General Hospital, Riverside, California;t and Loma Linda University Medical Center, Loma Linda, California;¢ and Emergency Physicians Medical Group, Oakland, California.§
Steven G Rothrock, MD, FACEP* Steven M Green, MD, FACEPt§ Christopher B Hummel, MD*
Receivedfor publication August 20, 1991. Revisions received February 17 and April 23, 1992. Acceptedfor publication May 5, 1992. Presented at the American College of Emergency Physicians Scientific Assembly in Boston, Massachusetts, October 1991.
Study objective: To prospectively evaluate previously described high-yield clinical criteria for obtaining plain abdominal radiographs in the emergency evaluation of children.
Design: Prospective, observational study. Setting: Emergency departments of a university medical center and
an affiliated county hospital. P a r t i c i p a n t s : Three hundred fifty-four children 15 years old or younger who underwent plain abdominal radiography during a one-year period.
Methods and measurements: Physiciansordering plain abdominal radiographs completed data forms that included historical and physical examination information before viewing films. At a later date, records of all patients were reviewed for radiologist interpretation and final diagnosis. The data were analyzed to determine the sensitivity, specificity, and predictive values of previously described high-yield criteria (from a retrospective series) in detecting radiographs that were diagnostic or suggestive of "major" abdominal disease.
Main results: Sixty-one patients (17%) had major diseases potentially requiring procedural intervention (eg, appendicitis, ingested foreign bodies, and intussusception),whereas 296 patients (83%) had minor diseases not requiring procedural intervention (eg, gastroenteritis and nonabdominal diagnoses). The presence of any of the following features--prior abdominal surgery, foreign body ingestion, abnormal bowel sounds, abdominal distention, or peritoneal signs--was 93% sensitive and 40% specific in detecting diagnostic or suggestive radiographs in patients with major disease. Positive and negative predictive values were 11% and 99%, respectively. If only these criteria had been used to obtain radiographs, 38% of films would have been omitted (at an estimated savings of $20,000) with only two suggestive radiographs missed. Conclusion: Our results suggest that restricting abdominal radiographs to patients with at least one of these five high-yield clinical features will detect most diagnostic and suggestive radiographs in children with major abdominal diseases.
[Rothrock SG, Green SM, Hummel CB: Plain abdbminal radiography in the detection of major disease in children: A prospective analysis. Ann ErnergMed December 1992;2I :1423-1429.]
DECEMBER1992 21:12 ANNALS0FEMERGENCYMEDIC]NE
1423/1£
ABDOMINAL
RADIOGRAPHY
Rothrock, Green & Hummel
INTRODUCTION
Plain abdominal radiographs often are recommended routinely in the evaluation of children with abdominal complaints. 1-6 However, recent studies suggest that 30% to 48% of plain abdominal films obtained in children may be unnecessary.7, a Although this modality may be overused in the pediatric population, no prospective studies have defined a subpopulation for which films routinely should be obtained or excluded. A recent retrospective study has suggested that most diagnostic radiographs in children with acute abdominal processes are found in children with one of the following Table 1.
Usefulness of radiography by final diagnosis Final Diagnosis
No.
Radiographic Interpretation Normal or Diagnostic Suggestive Incidental Misleading
Major Foreign body Appendicitis Ventriculoperitoneal shunt malfunction Large-bowel obstruction Small-bowel obstruction Strangulated hernia Intussusception Pyloric stenosis Intra-abdominal trauma Meckel's diverticulum Spontaneous peritonitis Urolithiasis Subtotal
15 15
15 I
0 1
0 11
g 1
10
0
0
10
0
1
0
1
0
0
5 2 4 3
5 0 0 0
0 0 1 3
O 2 3 0
0 0 0 0
2
1
0
1
0
1
O
0
1
0
2 2 61
1 0 23(38%)
0 0 6(10%)
1 2 31 (50%)
0 0 1 (2%)
57
1
25 71 28
1 4 0
22
2
15 16
0 0
Minor Pain of uncertain etiology 58 0 0 Nonabdominal diagnoses* 28 1 1 Gastroenteritis 77 0 2 Constipation 40 8 4 Normal ventriculoperitoneal shunt 24 0 0 Foreign body ruled out 15 0 O Abdominal trauma t 16 0 0 Gastrostomy tube placement 5 1 O Pneumonia 11 0 0 Urinary infection 16 0 0 Ileus 2 0 0 Ovarian cyst 1 0 0 Subtotal 293 10 (3%) 7 (2%) Total 354 33{9%) 13 (4%) *Excluding pneumonia. tNo evidence of intra-abdominalorgan injury in any patient.
20/ 1424
4 10 16 2 I 267 (91%) 298(84%)
0 1 0 0 0 9 (3%) 10(3%)
features: p r i o r abdominal surgery, foreign body ingestion, abdominal distention, peritoneal signs, or a b n o r m a l bowel sounds. 7 Our study attempted to validate prospectively this previously described decision rule. Secondary study goals were to assess the relationship between the ordering physician's suspicion of " m a j o r " disease and the presence of a b n o r m a l radiographs and to compare the reasons for obtaining films with the presence of abnormal radiographs. MATERIALS
AND
METHODS
A prospective, observational study was conducted in the emergency departments of a university medical center and an affiliated county hospital between April 1, 1990, and March 31, 1991. The combined census of both EDs during the study period was 100,000 visits p e r y e a r with approximately 25% of these patients u n d e r the age of 16 years. All encounters in which a supine abdominal view, acute abdominal series (upright and supine views of the abdomen and upright chest), two views of the abdomen (supine and lateral decubitus), or a combined supine chest and abdomen view were obtained in a child under the age of 16 were eligible for inclusion in the study. Emergency attending physicians or residents at the secondor t h i r d - y e a r level completed d a t a forms after each patient evaluation but before viewing the radiographs. This form included the reason for ordering the film, data from the history and physical examination, and the physician's estimate of the probability of m a j o r disease. No effort was made to standardize the presence of subjective physical examination features. Specifically, physicians were instructed to not alter their normal practice p a t t e r n and to use their normal Table 2.
Misleading radiographs Patient No. Radiologic Diagnosis 79 124 143 153
Final Diagnosis
Ventriculoperitoneal shunt disconnection Necrotizing enterocolitis Pyloric stenosis Abdominal mass
Gastroenteritis* Gastroenteritis t Gastroenteritis* Pain of uncertain etiology § 207 Ventricu]operitoneal shunt disconnection Constipation* 237 Constipation Appendicitis 246 Bowel obstruction Pneumonia* 304 Necrotizing enterocolitis Spider bite t 333 Intussusception Gastroenteritis II 354 Intussusception 6astroenteritisll *After computedtomography scan and neurosurgery consultation, both patients were thought to have functioning ventriculoperitoneal shunts. tBoth children lacked risk factors for necrotizing enterocolitis and had symptornatology consistent with gastroenteritis and benign follow-up. *Both patients had presentations consistent with their final diagnosis and were able to take fluids orally and pass stool although no specific gastrointestinal workup was performed. §Mass suspected on initial film was not seen on subsequent ultrasound and KUB. I[ Both had normal barium enemas and clinical course consistent with gastroenteritis (continued fever and diarrhea after initial evaluation),although spontaneousreduction of the intussusception before barium enema cannot be excluded.
ANNALS OF EMERGENCY MEDICINE 21:12 DECEMBER 1992
ABDOMINAL RADIOGRAPHY
Rothrock, Green & Hummel
physical examination techniques to determine if abnormal bowel sounds or peritoneal signs were present. Board-certified radiologists' reports were used to establish the radiographic interpretation of films. Final diagnosis, def'med as the condition responsible for the presenting symptoms, was determined from records of all ED visits and hospitalizations. Final diagnoses were designated as "major" or "minor" before data analysis. Major diagnoses were defined as those that might require acute procedural intervention, including surgery, endoscopy, cystoscopy, or paracentesis. Examples include appendicitis, mechanical bowel obstruction, intussusception, pyloric stenosis, foreign body ingestion, blunt intra-abdominal trauma, malfunctioning ventriculoperitoneal shunt, peritonitis, necrotizing enterocolitis, nephrolithiasis, and toxic megacolon. Minor diagnoses, defined as those not requiring acute procedural intervention, included constipation, nonobstructive ileus, urinary tract infection, gastroenteritis, and nonabdominal medical conditions. Two board-certified emergency physicians reviewed each case and categorized the radiographic interpretation as diagnostic, suggestive, normal, incidental, or misleading with respect to each patient's final diagnosis. Diagnostic was defined as a radiographic finding pathognomonic of a disease (eg, dilated loops of small bowel with multiple air fluid levels on upright or decubitus views and no air noted in the colon in a patient with small bowel obstruction). Suggestive was defined as a radiographic finding associated with but not pathognomonic of a disease (eg, right lower quadrant ileus in appendicitis). Normal was defined as a radiograph lacking any abnormality. Incidental was defined as a radiographic Table 3.
Association of diagnostic, suggestive, normal, and misleading radiographs with reasonsfor obtainingfilms
Indications for Obtaining Radiographs
Diagnostic Diagnostic or or Suggestive Suggestive No. of of Major of Minor Patients Disease Disease Normal
Pain of uncertain etiology 96 Suspect appendicitis 40 Suspect obstruction 37 Evaluate ventriculoperitoneal shunt 34 Suspect foreign body 31 Constipation 22 Vomiting 19 Blunt abdominal trauma 14 6astrointestinal bleeding 12 Suspect intussusception 9 Suspect ureteral stone 6 Other indications 34 Total 354
DECEMBER1992
Misleading
2 1 5
3 1 2
87 38 29
4 O 1
0 15 0 1
0 0 5 g
32 16 17 17
2 0 0 1
0
0
14
0
1
1
9
1
0
0
9
O
0 4 29(8%)
g ,5 17(5%)
6 25 298(84%)
21:12 ANNALS OF EMERGENCYMEDICINE
0 1 10 (3%)
finding thought to be unrelated to the patient's symptoms (eg, bony abnormality in a patient with appendicitis). Misleading was defined as a radiographic finding suggesting significant pathology that was not present (eg, appendicolith and ileus in a patient without appendicitis). The data were analyzed to determine if clinical features were predictive of diagnostic and suggestive radiographs in patients with major disease. Sensitivity, specificity, and positive and negative predictive values were determined for each clinical feature and for combined clinical features from a previous retrospective study. 7 The relationship between the physician's pretest estimate of major disease (expressed as a percentage) and the presence of diagnostic or suggestive radiographs was evaluated using the Pearson product moment correlation. A receiver operating curve was constructed for different values of the physician's pretest estimate of major disease. With this curve, the optimum set point for screening (for diagnostic and suggestive radiographs) was determined. The optimum set point was defined as the lowest cutoff where the sensitivity (of the physician's pretest estimate of major disease) still increased more rapidly than the falsepositive rate. RESULTS Of the 373 eligible patients identified during the study period, charts or data forms from 19 were either incomplete or unavailable, leaving 354 patients for analysis. Mean patient age was 5.2 + 4.6 years. A supine abdominal view was obtained in 173 (49%), acute abdominal series in 118 (33%), two views of the abdomen in 27 (8%), and a single view of the abdomen and chest in 36 (10%). Sixty-one patients (17%) were classified as having major disease, and 293 (83%) were classified as having minor disease. Specific diagnoses in each category are listed (Table 1). With regard to the final diagnosis, 33 radiographs (9%) were categorized as diagnostic, 13 (4%) as suggestive, 298 (84%) as normal or incidental, and ten (3%) as misleading (Table 1). The ten radiographs categorized as misleading were from one patient with major and nine patients with minor disease (Table 2). The number of diagnostic, suggestive, misleading, and normal radiographs are listed with each specified reason for obtaining the radiograph (Table 3). The sensitivity, specificity, and positive and negative predictive values of clinical features in detecting diagnostic or suggestive radiographs in patients with major disease are listed (Table 4). There was a statistically significant correlation between the strength of the physician's suspicion of major disease and the presence of diagnostic and suggestive radiographs (r = .703, P = .023) (Figure 1). A receiver operating curve for the physician's varying estimates of major disease presence is shown (Figure 2). The optimum set point for screening was a more than 40% suspicion of major disease. This cutoff had a sensitivity of 70% and a false-positive rate of 39% for predicting diagnostic or suggestive radiographs.
1425/21
ABDOMINAL RADIOGRAPHY Rothrock, Green & Hummel
DISCUSSION
Plain abdominal radiographs are often used in the evaluation of children presenting to EDs with abdominal complaints. Reported indications for obtaining plain films in children vary from all children with abdominal pain1, 3 to only those with persistent tenderness or distention. 6 Unfortunately, these recommendations a p p e a r to be the opinions of various authors and are not based on published studies. In addition, most children who present to an ED with abdominal pain have a nonspecific or medical cause for their symptoms, making routine abdominal r a d i o g r a p h y potentially unnecessary in a large n u m b e r of patients.9,10 Most studies detailing the usefulness of plain abdominal films have been limited to adults. In a retrospective series of more than 5,000 consecutive patients with acute abdominal pain presenting to an ED in the United Kingdom, Campbell et al found that 60% of films obtained were entirely normal and that positive findings in the remaining 40% were more likely to be misleading than to be helpful, n Brewer et al looked at 1,000 consecutive cases of abdominal pain presenting to a university hospital ED and found that plain radiographic findings never changed a clinician's first diagnostic impression. 12 Several authors have attempted to develop high-yield criteria for obtaining plain abdominal radiographs in adults. Eisenberg et al reviewed 1,780 patients who had undergone abdominal r a d i o g r a p h y and found that limiting radiographs to patients with moderate-to-severe pain, suspected mesenteric ischemia, obstruction, t r a u m a , renal calculi, or gallTable 4. Usefulness of clinical features in detecting radiographs diagnostic or suggestive of major disease Predictive Values (%) Sensitivity Specificity Positive Negative
bladder disease would have decreased the number of ordered films by more than 50% while identifying all serious pathology. 13 A more recent prospective study of 100 consecutive ED patients found that t r a u m a , flank pain with hematuria, ingestion of foreign bodies, previous surgery, and severe acute abdominal pain were all high-yield indications for ordering plain abdominal films. 14 Only one p r i o r study has attempted to define criteria for obtaining plain abdominal radiographs in the pediatric population. 7 Indications for obtaining plain abdominal radiographs are likely to differ between adults and children p r i m a r i l y because acute abdominal conditions differ between age groups. Diseases such as pyloric stenosis, midgut volvulus, and intussusception are more common in children, whereas cholecystitis, mechanical bowel obstruction, abdominal aneurysms, and mesenteric ischemia are more common in adults. P r i o r studies addressing plain abdominal r a d i o g r a p h y in children have suggested that plain radiographs often are overused.7,8 A recent review of 431 consecutive pediatric plain abdominal radiographs found that hmiting plain abdominal r a d i o g r a p h y to patients with either p r i o r abdominal surgery, suspected foreign body ingestion, abdominal distention, peritoneal signs, or abnormal bowel sounds was 100% sensitive and 53% specific in detecting diagnostic radiographs in children with m a j o r abdominal disorders. 7 When the same criteria were applied prospectively to our population, the sensitivity and specificity were 93% and 40%, respectively. Of the 29 patients with m a j o r disease discernible radiographically, only two had none of these five criteria. One of these two patients h a d a typical presentation for appendicitis: fever, vomiting, and right lower q u a d r a n t tenderness. His r a d i o g r a p h revealed an appendicolith. The other patient was a 5-week-old infant with persistent vomiting secondary to pyloric stenosis and gastric distention noted radiographically.
History Age < 2 months Prior abdominal surgery* Foreign body ingestion (or radiopaque substance)* Abdominal pain Vomiting Bilious vomiting •~ Flank pain ~ Hematuria
19 26
92 74
17 8
93 92
56 22 48 15 4 0
93 38 48 96 95 99
39
O
96 86 92 93 92 92
19
50
3
88
30
4 4 27 22 29 0
83 95 81 68 90 ,90
2 7 11 5 14 14
91 92 93 91 95 95
2O
*Any one of five combined criteria 93
40
11
99
3 7 25
6
Physical Examination Features Temperature > 38.3 C Abdominal tenderness Right lower quadrant tenderness Peritoneal signs* Abdominal distention* Altered bowel sounds* 6uaic-positive stool Rectal tenderness
22/1426
Figure 1. Relationship of pretest probability of major disease to the presence of diagnostic or suggestive radiographs % Diagnostic or suggestive of major disease
50
°°t 10
0
10
20 30 40 50 60 70 80 Pretest probability of major disease (%)
90
100
ANNALS OF EMERGENCYMED/CINE 21:12 DECEMBER1992
ABDOMINAL RADIOGRAPHY Rothrock, Green & Hummel
Although the combined criteria were highly sensitive in detecting diagnostic and suggestive radiographs, single features were less valuable. This may be due in part to the large variety of diseases in the group designated as having major disease. The use of a history of prior abdominal surgery in detecting diagnostic and suggestive radiographs appears self-evident. Abdominal surgeries often are associated with obstruction, and bowel obstructions were the second most common major diagnosis seen in our study. Although abdominal radiography has been recommended routinely for all children who have ingested a foreign body,15,16 this approach recently has been questioned, a,17-19 A recent study by Caravati et al indicated that asymptomatic children who have ingested foreign bodies can be followed on an outpatient basis without radiography. 17 Others have recommended routine radiography of the chest and neck but not of the abdomen for all patients with a history of foreign body ingestion. 4,18,19 This approach allows for identification of foreign bodies lodged in the esophagus, trachea, or hypopharynx that can lead to perforation of the esophagus or airway obstruction. Abdominal foreign bodies, however, rarely cause problems in asymptomatic patients, and routine abdominal radiography in these patients may be unnecessary.4,19 We found the presence of abdominal distention, abnormal bowel sounds, and peritoneal signs to be useful for detecting diagnostic or suggestive radiographs in patients with major diseases. This may be related to the high number of patients with mechanical bowel obstructions in addition to functional obstructions (eg, intussusception and pyloric stenosis) in our study. Cases of appendicitis and spontaneous peritonitis also accounted for patients with peritoneal signs. A statistically significant correlation between the ordering physician's pretest suspicion of major disease and the incidence of diagnostic or suggestive radiographs was noted (Pearson's r = .703, P = .023). This finding suggests that the strength of Figure 2. Receiver operating curve
100
% Sensitivity ...->
90
10
. . . - > 40- -~ 30""
80 70 60
.'¢ 70
/;'80
50 40
i
30
J ,~, 90
20 10 i
0 0
i
10
20
I
I
I
i
i
30
40
50
60
70
% False-Positives
DECEMBER 1992
21:12
ANNALS OF EMERGENCY MEDICINE
the physician's suspicion of major disease would be useful in deciding when to order radiographs. Unfortunately, the use of physician suspicion of disease presence is a highly subjective parameter. It is probable that physician interobserver variability would be high with this parameter and not reproducible when applied to other settings. The optimum cutoff for obtaining radiographs was a more than 40% suspicion of major disease. This set point had a 70% sensitivity and 61% specificity in detecting diagnostic or suggestive radiographs in patients with major disease. This cutoff still would have missed 30% of diagnostic and suggestive radiographs, and the false-positive rate (39%) would have been too high to be of any clinical use. Thus, physician pretest estimate of major disease presence was thought to be of little use in predicting diagnostic or suggestive radiographs in patients with major disease. Another limitation of using the physician's estimate of major disease presence for detecting diagnostic or suggestive radiographs is that many major diseases have few or no radiographic findings. For example, Campbell and Gunn found that more than 90% of radiographs in patients with appendicitis were normal or showed incidental findings and that the remainder were more hkely to be misleading than helpful.ll Commonly described positive radiographic findings such as rightward scoliosis, soft-tissue masses, localized ileus, and free peritoneal fluid are als O nonspecific and generally not useful in radiographic diagnosis of appendicitis.2°-23 Even the reported pathognomonic finding of a right lower quadrant appendicolith can be found in patients without appendicitis and thus cannot be depended on to diagnose appendicitis.2 ~,22,24 O t h e r major diseases also have significant rates of normal plain radiography. Radiographs in patients with bowel perforation are normal in 33% to 51% of cases. 25-27 Normal radiographs in patients with necrotizing enterocolitis also approach 50%.28.29 A classic radiographic finding in necrotizing enterocolitis--pneumatosis intestinalis--also can appear, persist, worsen, disappear, or be absent independent of the time course of the disease, thus limiting the sensitivity of radiographs in this disease. 29 Patients with intussusception exhibit normal radiography in only 11% to 25% of cases. 30-32 However, commonly reported positive radiographic findings such as sparse intestinal gas, sparse fecaloid content, a discernible liver edge, or a gas-fluid level in dilated or undilated bowel are not useful in distinguishing between patients with and without intussusception. 3° The sensitivity of plain films in intussusception can be improved if a cross-table supine abdominal view is obtained. 31 However, interpretation of this film requires a radiologist skilled in interpreting this technique. 31 Three of four films in patients with intussusception were normal in our study, although no patient had a cross-table supine view obtained. Pyloric stenosis is another major disease with a substantial incidence (35%) of normal radiography. 33 Riggs and
1 4 2 7 / 2 3
A B D O M I N A L RADIOGRAPHY Rothrock, Green & Hummel
Long examined plain film findings in pyloric stenosis and f o u n d that no single feature was specific for diagnosing this disease, although an algorithm was developed using combinations of seven radiographic c r i t e r i a to separate patients into high-, moderate-, and low-probability categories of pyloric stenosis. 33 If these criteria had been applied to our patient population, the three patients with pyloric stenosis would have been categorized into the m o d e r a t e - p r o b a b i l i t y group. Our rate of 50% normal radiographs in patients with m a j o r disease is consistent with these p r i o r studies. Plain films in patients with abdominal pain of uncertain etiology were n o r m a l in 91% of cases in our study and twice as likely to be misleading as diagnostic of m a j o r disease (Table 3). These findings suggest that plain films often are useless and even confusing in a significant n u m b e r of patients with major disease and undifferentiated abdominal pain. Reliance on plain film r a d i o g r a p h y to exclude m a j o r abdominal diseases can be hazardous and cause serious diagnostic errors. Alternate modalities such as ultrasonography should be considered in evaluating children for specific diseases such as appendicitis, pyloric stenosis, and intussusception. 34-36 The findings of our study are limited by several factors. The classification of diseases as major and minor implies that the radiographic findings of minor disease (eg, constipation or ileus) are not useful to the clinician. However, we believe that major diseases are more important to detect than minor diseases because m a j o r diseases may require acute p r o c e d u r a l intervention. If some clinicians believe that minor diseases are important to identify radiographically, then our criteria would not be applicable to their practice. It also can be argued that certain m a j o r diseases (eg, gastrointestinal foreign bodies) often are insignificant a n d require no treatment. However, we thought that it was more a p p r o p r i a t e to overclassify diseases as being major if there was any possibility that the patient might require a proced u r a l intervention. In the instance of abdominal foreign bodies, the majority will pass through the gastrointestinal tract without causing problems. However, 5% to 10% will require endoscopic or surgical removal.4dsd 9 Thus, foreign body ingestion was categorized as a major disease. Data acquisition for specific physical examination features was not standardized. Physicians were told to use their own interpretation of generally accepted criteria to determine if a b n o r m a l bowel sounds and peritoneal signs were present. We believed that requiring physicians to alter their normal physical examination techniques to comply with data acquisition also might alter results. No standardized radiographic examination was used in this study. The type of film o r d e r e d (eg, supine abdominal view versus three-view examination and so on) was left to the discretion of the physician. Thus, radiographic abnormalities that might require more than a supine abdominal view for diagnosis (eg, small-bowel obstruction or bowel
24/1428
perforation) may have been missed by our study. This factor could not be controlled without obtaining a three-view examination of all patients, exposing them to unnecessary cost and radiation. Our results also might be limited by the patient population. The study hospitals included a tertiary-care center for pediatrics with most subspecialties, including pediatric surgery, critical care, gastroenterology, and neurosurgery, and a large county hospital. Thus, both institutions may treat disproportionately more children with complex abdominal disorders and advanced disease states. No cases of Hirschprung's disease, midgnt volvulus, necrotizing entercolitis, or bowel perforation were present in our study. It is unknown whether the combined criteria would have detected diagnostic or suggestive radiographs in children with these disorders. There also were no cases of biliary tract disease in our study. In adults, radiographs rarely are helpful when biliary tract disease is suspected, although their use in children with biliary disease is uncertain. 37 Due to study limitations, our criteria may not be applicable to all ED settings. However, our results suggest that limiting radiographs to patients with at least one of the five high-yield criteria (ie, p r i o r abdominal surgery, suspected foreign body ingestion, abnormal bowel sounds, abdominal distention, or peritoneal signs) will detect most diagnostic and suggestive radiographs in patients with m a j o r abdominal diseases. If only these criteria h a d been used to obtain radiographs, then 93% of diagnostic and suggestive radiographs would have been detected while eliminating 38% of films ordered. This would have saved a significant amount of time and money (approximately $20,000) and limited radiation exposure to patients. CONCLUSION
A prospective study was u n d e r t a k e n to assess the usefulness of previously described high-yield clinical criteria in detecting diagnostic and suggestive radiographs in children with major abdominal diseases. Abdominal radiographs were normal or misleading in 87% of patients and useful diagnostically (diagnostic or suggestive) in 13% of patients. Fifty percent of abdominal radiographs in patients with maj or abdominal diagnoses also were normal. Physician estimate of the presence of m a j o r disease was found to have poor utility in detecting diagnostic or suggestive radiographs. However, limiting radiographs to children with a history of abdominal surgery, suspicion of foreign body ingestion, a b n o r m a l bowel sounds, abdominal distention, or peritoneal signs would have detected 93% of diagnostic and suggestive radiographs in patients with m a j o r disease while eliminating 38% of films. Although clinical criteria alone cannot replace clinical judgment in the evaluation of patients in the ED, physicians should consider the presence of these five high-yield criteria when determining whether to obtain a plain abdominal radiograph in a child with acute abdominal complaints.
ANNALS OF EMERGENCY MEDICINE 21:12 DECEMBER1992
ABDOMINAL RADIOGRAPHY Rothrock, Green & Hummel
REFERENCES 1. Barkin RM: Abdominal pain in children: Clues to identifying nontraumatic causes. EmergMed Rep 1988;9:9-16. 2. Buchert GS: Abdominal pain in children: An emergency practitioner's guide. Emerg Med Clin North Am 1989;7:497-517.
28. Frey EE, Smith W, Franken EA, et al: Analysis of bowel perforation in necrotizing enterocolitis. Pediatr Radial 1987;17:380-382. 29. Rabinowitz JG, Siegle RL: Changing clinical and roentgenographic patterns of necrotizing enterocolitis. Am J Roentgeno11976;126:560-566.
3. Martin DF: The acute abdomen in childhood. CurrProb Diag Radial 1986;15:340-394.
30. Eklef O, Hartelius H: Reliability of abdominal plain film diagnosis in pediatric patients with suspected intussusception. Pediatr Radial 1980;9:199-206.
4. 0zenoff MB: Emergency radiology in childhood. Emerg Med Clin North Am 1985;3:563-584.
31. Johnson JF, Woisard KK: Ileocolic intussusception: New sign on the supine cross table lateral radiograph. Radiology1989;170:483-486.
5. Singleton EB, Wagner ML: The acute abdomen in the pediatric age group. Semin Roentgeno11973;8:339-356.
32. LeVine M, Schwartz S, Katz I, et al: Plain film findings in intussusception. Br J Radial 1964;37:678-681.
6. Wolf SA, Shermeta DW: Abdominal pain, in Ehrlich FE, Heldrich FJ, Tepas JJ (eds): Pediatric EmergencyMedicine. Rackville, Maryland, Aspen Publishers I nc, 1987, p 213-220.
33. Riggs W, Long L" The value of the plain film roentgenogram in pyloric stenosis. Am J Roentgeno11971;112:77-82.
7. Rothrock SG, Green SM, Harding M, et al: Plain abdominal radiography in the detection of acute medical and surgical disease in children: A retrospective analysis. Pediatr Emerg Care 1991;7:281-285.
34. Schwerk WB, Wichtrup B, Rothmund M, et al: Ultrasoncgraphy in the diagnosis of acute appendicitis: A prospective study. Gastroenterology 1989;97:630-639. 35. Hailer JO, Cohen HL: Hypertrophic pyloric stenosis: Diagnosis using US. Radiology 1986;161:335-339.
8. Cronan K, Shaw K, Bellah R: The utility of abdominal radiography in a pediatric emergency department (abstract). Ann Emerg Med 1990;19:476.
36. See CC, Glassman M, Berezin S, eta[: Emergency ultrasound in the evaluation of acute onset abdominal pain in children. PediatrEmerg Care 1988;4:169-171.
9. Reynolds SL, Jaffe DM: Children with abdominal pain: Evaluation in the pediatric emergency department. PediatrEmerg Care 1990;6:8-12.
37. Rothrock SG, 6oerhuis H, Howard RM: Efficacy of plain abdominal radiography in patients with biliary tract disease. J Emerg Med 1989;8:271-275.
10. Stevenson RJ: Abdominal pain unrelated to trauma. Surg Clin North Am 1985;65:1181-1215. 11. Campbell JPM, Gunn AA: Plain abdominal radiographs and acute abdominal pain. Br J Surg 1988;75:554-556. 12. Brewer RJ, Golden 6% Hitch DC, et al: Abdominal pain: An analysis of 1,000 consecutive cases in a university hospital emergency room. Am J Surg 1976;131:219-223. 13. Eisenberg RL, Heineken P, Hedgcock MW, et aL Evaluation of plain abdominal radiographs in the diagnosis of abdominal pain. Ann Surg 1983;197:464-467.
Address for reprints: Steven G Rothrock, MD Department of Emergency Medicine Orlando Regional Medical Center 1414 Kuhl Avenue Orlando, Florida 32806
14. McCookTA, Ravin CE, Rice RP: Abdominal radiography in the emergency department: A prospective analysis. Ann Emerg Med 1982;11:7-8. 15. Schnaufner L, Mahboubi S: Abdominal emergencies, in Fleischer 6, Ludwig S (eds): Pediatric EmergencyMedicine. Baltimore, Maryland, Williams & Wilkins, 1988, p 936-965. 16.Wesley JR: Managing foreign bodies in various orifices. Emerg Med Rep 1984;5:109-116. 17. Caravati M, Bennett DL, McEIwee NE: Pediatric coin ingestion: Prospective study on the utility of routine roentgenograms. Am J Dis Child 1989;143:549-551. 18. Schunk JE, Corneli H, Bolte R: Pediatric coin ingestions: A prospective study of coin location and symptoms. Am J Dis Child 1989;143:546-548. 19. Hedge D, Techlenburg F, Fleischer 6: Coin ingestion: Does every child need a radiograph? Ann Emerg Mad 1985;14:443-446. 20. Bakhda RK, McNair MM: Useful radiological signs in acute appendicitis in children. Clin Radial 1977;28:193-196. 21. Hatten LE, Miller RC, Hester CL, et al: Appendicitis and the abdominal raentgenogram in children. South MedJ1973;66:803-80& 22. Graham AD, Johnson HF: The incidence of radiographic findings in acute appendicitis compared to 200 normal abdomens. MilitMed 1966;131:272-276. 23. Isdale JM: The radiological signs of acute appendicitis in infancy and childhood. South Aft Med J 1978;53:363-364. 24. Sorer CS: The contribution of the radiologist to the diagnosis of acute appendicitis. Semin Roentgeno11973;8:375-388. 25. Felson B, Wiot JF: Another look at pneumoperitoneum. Semin Roentgenol 1973;8:437-443. 26. Rice RP, Thompson WM, Gedgaudas RK: The diagnosis and significance of extraluminal gas in the abdomen. Radial Clin North Am 1982;20:819-837. 27. Roh J J, Thompson JS, Harned RK, et al: Value of pneumoperitoneum in the diagnosis of visceral perforation. Am J Surg 1983;t46:830-833.
DECEMBER1992
21:12
ANNALS OF EMERGENCY MEDICINE
1429/25
