To demonstrate the accuracy of handheld metal detectors (HHMDs) for identification and localization of ingested metallic foreign bodies when used by experienced and inexperienced investigators.
Prospective study comparing HHMD scanning with radiography.
Setting and Patients
A consecutive sample of all eligible patients (N=176) presenting to the emergency departments of Children's Medical Center of Dallas, Dallas, Tex, and Children's Hospital of the King's Daughters, Norfolk, Va, who were known or suspected to have ingested a metallic foreign body.
Each patient underwent HHMD scanning and radiographic evaluation.
Main Outcome Measures
Statistical evaluation compared HHMD scanning with radiography and experienced vs inexperienced investigator HHMD scanning to determine the accuracy of the screening tool and investigators.
Experienced investigators performed HHMD scans on 140 subjects; inexperienced investigators scanned all subjects. Disease was defined as a foreign body in the esophagus on radiograph. The 3 experienced investigators demonstrated sensitivity of 100% (95% confidence interval [CI], 94%-100%), specificity of 92.4% (95% CI, 84.2%-97.2%), positive predictive value (PPV) of 90.9% (95% CI, 81.3%-96.6%), and negative predictive value (NPV) of 100% (95% CI, 95.1%-100%); the proportion correct was 95.7% (95% CI, 90.8%-98.4%). The inexperienced investigators demonstrated sensitivity of 95.7% (95% CI, 88.0%-99.1%), specificity of 81.0% (95% CI, 72.1%-88.0%), PPV of 77.0% (95% CI, 66.8%-85.4%), and NPV of 96.6% (95% CI, 90.4%-99.3%); the proportion correct was 86.9% (95% CI, 80.9%-91.5%). The McNemar test demonstrated no statistically significant difference between HHMD scanning by experienced vs inexperienced investigators.
Handheld metal detector scanning is an accurate, inexpensive, radiation-free screening tool and should be used for evaluation of patients suspected of ingesting coins and coinlike foreign bodies.
CHILDREN INGEST numerous objects besides food, prompting varying degrees of concern on the part of parents and physicians. The most commonly ingested foreign bodies among pediatric patients are coins.1- 3 While most coins pass uneventfully into the stomach, coins that become impacted in the esophagus can cause complications, including esophageal ulceration, esophageal perforation, tracheoesophageal fistula formation, and death caused by tracheal compression and aortoesophageal fistula formation.4- 7 Since clinical signs and symptoms have proven to be unreliable predictors of impaction, routine radiographic evaluation is recommended for localization of ingested coins.2,8,9
However, surveys indicate that radiographs are often not obtained for many reasons, including expense, inconvenience, radiation exposure, and the clinician's perception of coin ingestion as a benign event.9,10 Metal detectors have been suggested as a rapid, inexpensive, and radiation-free alternative for evaluation of these patients.11- 15 Previous studies have shown that handheld metal detectors (HHMDs) are accurate in localizing ingested coins when used by experienced investigators in small numbers of patients.11,14,15
Pediatric emergency medicine practitioners at Children's Medical Center of Dallas, Dallas, Tex, and Children's Hospital of the King's Daughters, Norfolk, Va, remained hesitant to apply this new technology based on 2 limitations of the previous studies: small sample size and use of the HHMD exclusively by experienced investigators. We designed this prospective study to address limited sample size by enrolling a sufficient number of patients to demonstrate sensitivity, specificity, positive predictive value, negative predictive value, and the proportion correct with 95% confidence intervals (CIs). To evaluate the usefulness of HHMD for "amateur" operators, both experienced and inexperienced investigators scanned patients.
From November 1, 1995, through November 30, 1996, all patients suspected of ingesting a metallic foreign body who presented to Children's Medical Center of Dallas were eligible for the study. Three of us (K.S., P.A.P, and B.J.E.) participated in a practice session to gain experience in the use of the HHMD. The session consisted of locating coins taped to or hidden in mannequins or under pillows. One of these experienced investigators was notified to enroll study candidates. After interview and physical examination, the experienced investigator performed an initial HHMD scan. Scan findings were recorded on the case report form, sealed in an envelope, and placed in the study packet. The experienced investigator then recruited an inexperienced investigator for a second HHMD scan. The inexperienced investigators were medical personnel in the emergency department with no previous experience in HHMD use. The inexperienced investigators included residents, fellows, attending physicians, nurses, respiratory therapists, paramedics, and registration personnel who did not perform an HHMD scan on more than 1 patient enrolled in the study. They were provided with an instruction sheet on HHMD use for detecting ingested metallic foreign bodies. Inexperienced investigators were blinded to the scan findings of the experienced investigators. The inexperienced investigator scanned the patient, recorded the findings on the case report form, sealed it in an envelope, and placed it in the study packet.
From February 1, 1996, through September 30, 1996, patients were enrolled at a second study site, Children's Hospital of the King's Daughters. All Children's Hospital of the King's Daughters study participants were enrolled and scanned by an inexperienced investigator only. The HHMD scan findings were handled in the manner previously described.
A Super Scanner metal detector (Garrett Security Systems Inc, Garland, Tex) was used at both study sites. The patients were undressed and their jewelry was removed. Younger and nonambulatory patients were held upright by a parent or guardian at arm's length facing the investigator. Older patients stood for the HHMD scan. All patients were positioned away from possible interfering objects, such as beds, walls, and cabinets. Patients were scanned vertically from under the chin to the umbilicus, then horizontally around the abdomen. Scanning of the back was left to the investigator's discretion. A positive scan was defined as a strong audio signal and a positive visual indication on the HHMD.
All patients underwent radiographic evaluation. Patients arriving at the study sites with previously obtained radiographs were enrolled and scanned by investigators blinded to the radiographic findings.
Disease was defined as a metallic foreign body visualized in the esophagus on plain radiograph. Radiographs demonstrating metallic foreign bodies below the diaphragm or demonstrating no foreign body were considered negative for disease. The HHMD scan results indicating a metallic foreign body above the diaphragm were considered positive tests. Negative tests were those HHMD scans indicating a foreign body below the diaphragm or no foreign body.
Statistical analysis of the data was performed after enrollment of 30 patients with and without disease. At that point, if we found no difference in scan results between experienced and inexperienced investigators, the experienced investigator branch would be closed. For the purpose of increasing enrollment, a second study site enrolled patients scanned only by inexperienced investigators. At the primary enrollment site, 7 patients were scanned only by inexperienced investigators when experienced investigators were unavailable. Interim data analysis indicated a notable difference in sensitivity between the investigator groups; thus, data collection was continued in both groups. A power analysis was not done.
This study was approved by the institutional review boards of both participating centers. Written informed consent was obtained for each patient enrolled. Patients were excluded if they demonstrated respiratory distress, had known cardiac surgery or implanted surgical hardware, or were known to be pregnant.
Statistical analysis of HHMD scanning compared with radiographic evaluation determined sensitivity, specificity, positive predictive value, negative predictive value, and proportion correct with calculated CIs. The McNemar test16 was employed to compare testing by study investigators and volunteer investigators.
There were 176 patients enrolled, 88 girls and 88 boys, with an age range of 6 months to 15 years, 7 months (mean, 42 months). The time of presentation ranged from less than 1 hour to 7 days, 19 hours, after ingestion; 146 patients (83.0%) presented less than 24 hours after ingestion and 21 (11.9%) had unknown time since ingestion. Enrollment included 121 patients with suspected coin ingestion, 33 suspected of ingesting other metallic foreign bodies, and 22 suspected of ingesting unknown foreign bodies. Experienced investigators enrolled and scanned 140 patients; all patients were scanned by an inexperienced investigator.
Table 1 compares HHMD scan findings with radiographic results for experienced and inexperienced investigators. One patient was excluded from statistical analysis because he was found to have a metallic foreign body in his right main bronchus, neither in the esophagus nor below the diaphragm, placing him outside of our disease classification.
The sensitivity, specificity, positive predictive value, negative predictive value, and the proportion correct for experienced vs inexperienced investigators are presented in Table 2. No statistically significant difference was demonstrated between the 2 groups of investigators. Sensitivities and specificities of experienced and inexperienced investigators' HHMD scan results were compared using the McNemar test (Table 3) for subjects with and without disease.
The McNemar statistic examines the difference in the likelihood of a positive determination for experienced and inexperienced investigators. Confidence intervals that include the value of 0 indicate no increase in the likelihood of a positive determination for one method vs the other. For subjects with radiographs positive for a foreign body in the esophagus, the estimate of the difference in the proportion positive is 0.033 (95% CI, −0.012 to 0.079). For both abnormal and normal radiographs, this estimate was −0.014 (95% CI, −0.063 to 0.034). The close proximity of the estimates to 0 and the straddling of 0 by the CIs indicate that there was no significant difference in the likelihood of positive determination. Therefore, HHMD scan by an experienced investigator is as likely to predict an abnormal radiographic finding as HHMD scan by an inexperienced investigator. The McNemar test found no statistically significant difference between experienced and inexperienced investigators.
Historical data obtained from the patient, parent, or guardian revealed that pennies were the most commonly ingested coins (66/124 [53.2%]), and coins were more commonly ingested than other metallic foreign bodies (124/176 [70.5%]). Table 4 lists the types of foreign bodies ingested as determined by history and the numbers of patients with and without disease in each category.
Previous studies have shown HHMDs to be accurate screening tools for patients with suspected metallic foreign body ingestion.11,14,15 Carraccio et al12 reviewed these studies and proposed a rationale for HHMD use by practitioners. Our study strengthened previous findings by demonstrating statistically significant HHMD accuracy when used by trained or untrained investigators. We continued enrollment until 60 patients with and without disease had been scanned by both an experienced and an inexperienced investigator to achieve 95% CIs.
Limitations of the study include variability among both inexperienced and experienced examiners and an inadequate sample size to obtain sufficient power. The wide array of medical personnel used as inexperienced investigators raises concerns about interobserver variability. However, this variability of experience with HHMDs was part of the question to be answered in this study. Although the training session for experienced investigators was brief, they gained experience with the HHMD throughout the study period. No power calculation was performed because the sample size would need to be increased about 7.7 times, for a sample size of 1064, to achieve power of 0.80.
While no statistically significant difference between investigator groups was demonstrated at final data analysis, we feel it is of clinical relevance that inexperienced investigators performed 3 false-negative scans. Patient 89 was aged 7 years, 11 months, and presented 6 hours after ingesting a Susan B. Anthony dollar. The experienced investigator scan indicated a foreign body in the midesophagus, the inexperienced investigator scan indicated a foreign body in the abdomen, and radiographic scan demonstrated a coin just above the lower esophageal junction. Patient 108 was aged 17 months and presented an unknown time after foreign body ingestion. No experienced investigator scan was performed, the inexperienced investigator scan indicated a foreign body in the abdomen, and the radiographic scan demonstrated a coin in the midesophagus. Patient 199 was aged 10 months and presented 11 hours after ingesting a penny. The experienced investigator scan indicated a foreign body at the sternal notch, the inexperienced investigator scan indicated no foreign body, and radiographic scan demonstrated a coin at the sternal notch. In addition, 1 patient was excluded because the small spring found lodged in his right main bronchus on radiograph placed him outside of our disease classification. Neither experienced nor inexperienced investigator scans detected this foreign body of low metal mass.
Prior to enrolling patients, 3 of us (K.S., P.A.P, and B.J.E.) participated in a limited, self-taught session (<1 hour) to learn the basic principles of HHMD function and use. We became more skilled and confident in HHMD scanning with repeated use. Consequently, we believe that practitioners can quickly become competent in this procedure. During the learning phase, radiographic evaluation of the first few patients scanned and any questionable patients thereafter would provide appropriate confirmation of skill level. We recommend use of HHMDs as a screening tool in patients suspected of ingesting coins or foreign bodies of similar or greater metal mass.
Forty percent of study patients had a foreign body lodged in the esophagus. In cataloging the foreign bodies retrieved from patients by endoscopy or balloon catheter extraction, we note that coins of any size present a risk for impaction, as do numerous other foreign bodies (Table 4). Evaluation of these historical and follow-up data demonstrates that screening of children for possible foreign body impaction cannot wisely be based on the size of the foreign body ingested or the age of the patient.
Of note, 1 patient was observed by her father to ingest an aluminum pull tab from a beverage can. The experienced and inexperienced investigators' scans both indicated a foreign body in the left upper quadrant of the abdomen. The initial radiograph was read as negative for a foreign body. Review of the radiograph with a pediatric radiologist informed of the history revealed a faintly visible possible foreign body within the stomach contents. This false-negative radiograph is consistent with previous findings and publications noting the "invisibility" of aluminum on radiograph films because of the very low radiodensity of the metal.17- 19 It is likely that HHMDs are more sensitive than plain radiography for the evaluation of patients suspected of ingesting aluminum foreign bodies. Other types of foreign bodies were quite varied in configuration and metal content (Table 5). The only false-negative scans involving a noncoin foreign body were in the patient with a spring in his right main bronchus. Because of the small number of patients enrolled with a noncoin foreign body lodged in the lower esophagus (n=6), we cannot conclude that HHMD screening is equally accurate in this patient subset. Our practice is to treat coinlike foreign bodies (batteries, game tokens) as we do coins, evaluating by HHMD scan. Evaluation of patients suspected of ingesting other types of foreign bodies must be determined on a case-by-case basis using presumed low metal mass as a primary indicator for radiographic evaluation.
We conclude that HHMDs should be used as a screening tool for patients suspected of coin ingestion, negating the need for radiographs in patients found to have a foreign body below the diaphragm or no foreign body. This screening tool can also be applied to noncoin foreign bodies of similar or greater metal mass. Office-based and emergency department practitioners can become proficient at HHMD scanning with only limited practice. Use of HHMD screening can increase early detection of esophageally impacted foreign bodies while decreasing expense, inconvenience, and radiation exposure to patients.
Accepted for publication January 14, 1999.
Presented at the fall meeting of the American Academy of Pediatrics, Boston, Mass, October 25, 1996.
We thank Donald D. McIntire, PhD, for his statistical advice and Lynn Heise for her assistance with manuscript preparation.
Corresponding author: Patricia Primm, MD, Department of Pediatrics, University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Blvd, Dallas, TX 75235-9063.
Editor's Note: Well, HHMD worked in the hands of "inexperienced" actors on the television program ER, so why shouldn't it work in the hands of "inexperienced" clinicians?—Catherine D. DeAngelis, MD
Seikel K, Primm PA, Elizondo BJ, Remley KL. Handheld Metal Detector Localization of Ingested Metallic Foreign BodiesAccurate in Any Hands?. Arch Pediatr Adolesc Med. 1999;153(8):853-857. doi:10.1001/archpedi.153.8.853