Diagnostic Performance of Conventional X-ray for Detecting Foreign Bodies in the Upper Digestive Tract: A Systematic Review and Diagnostic Meta-Analysis

Foreign body (FB) ingestion is a common clinical problem in acute settings. Detecting FBs in the upper digestive tract is challenging. The conventional X-ray is usually the first-line imaging tool to detect FBs. However, its diagnostic performance is inconsistent in the literature. In this study, we performed a systematic review and meta-analysis to determine the diagnostic performance of the conventional X-ray for detecting FBs in the upper digestive tract. We conducted a systematic search of PubMed, Embase, Cochrane Library, Web of Science, and Scopus until 1 August 2020. Prospective or retrospective studies investigating the diagnostic accuracy of conventional X-rays for detecting FBs in the upper digestive tract in patients of all ages were included. The Quality Assessment of Studies of Diagnostic Accuracy-2 tool was used to review the quality of included studies. We used a bivariate random-effects model to calculate diagnostic accuracy parameters. Heterogeneity was assessed using I2 statistics. We included 17 studies (n = 4809) in the meta-analysis. Of the 17 studies, most studies were rated as having a high risk of bias. Conventional X-rays had a pooled sensitivity of 0.58 (95% confidence interval [CI] = 0.36–0.77, I2 = 98.52) and a pooled specificity of 0.94 (95% CI = 0.87–0.98, I2 = 94.49) for detecting FBs in the upper digestive tract. The heterogeneity was considerable. The area under the summary receiver operating characteristic curve was 0.91 (95% CI = 0.88–0.93). Deek’s funnel plot asymmetry test results revealed no significant publication bias (p = 0.41). The overall sensitivity and specificity of conventional X-rays were low and high, respectively, for detecting FBs in the upper digestive tract. Hence, conventional X-rays to exclude patients without upper FBs in the digestive tract are not recommended. Further imaging or endoscopic examinations should be performed for at-risk patients.


Introduction
Foreign body (FB) ingestion is a common clinical problem in acute settings. More than 93,000 cases of FB ingestion were reported in the United States in 2018 [1]. Most cases Diagnostics 2021, 11, 790 2 of 12 of FB ingestion occur in the pediatric population, with the highest incidence observed between the age of 6 months and 6 years [2]. In adults, FB ingestion is more frequently observed in elder persons with impaired swallowing controls, individuals with underlying psychiatric diseases, or those with alcohol intoxication [3]. Most ingested FBs obstruct the oropharynx and hypopharynx; obstruction of the esophagus by FBs is less common [4]. In regards to the management of ingested FBs, approximately 80-90% of ingested FBs pass through the gastrointestinal system and are excreted without any intervention being required, 10-20% require endoscopic removal, and less than 1% require surgical removal [5]. In addition, a timely and accurate detection of FBs in the upper digestive tract is crucial because undetected FBs present in the esophagus might increase morbidity and even mortality [6]. Therefore, FBs should be removed within 24 h of ingestion because the risk of complications substantially increases over time [5].
Detecting FBs present in the upper digestive tract is challenging for clinicians. The conventional X-ray is usually used as the first-line imaging modality to detect FBs. Although several studies [4,[7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22] have investigated the use of conventional X-rays for detecting FBs, they have reported inconsistent results in terms of diagnostic performance, with sensitivity ranging from 15% to more than 90% [7,19] and specificity ranging from 50% to 100% [7,8]. The American Society for Gastrointestinal Endoscopy and the European Society for Gastrointestinal Endoscopy (ESGE) guidelines strongly recommend using conventional X-rays to detect the presence, location, size, configuration, and the number of ingested FBs if the ingestion of radiopaque objects is suspected or if the type of object is unknown. In addition, the ESGE guideline does not recommend radiological evaluation for patients with nonbony food bolus impaction without complications [2,5]. However, the quality of evidence for these recommendations is low.
Because inconsistent results have been reported in the literature and these findings have not been previously synthesized through a meta-analysis, we performed a systematic review and meta-analysis to evaluate the diagnostic performance of conventional X-rays for detecting FBs in the upper digestive tract.

Materials and Methods
The protocol of this systematic review and meta-analysis is registered on PROSPERO (PROSPERO ID: CRD42020201034). This study adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement [23]. We systematically searched the following databases from their inception until 1 August 2020: PubMed, Embase, Cochrane Library, Web of Science, and Scopus. We used the following keywords to search for relevant studies: X-ray, plain radiography, plain film, FB ingestion, and FB obstruction. The details of our search strategy are listed in Supplementary Table S1.
We included both prospective or retrospective studies examining the diagnostic accuracy of conventional X-rays for detecting FBs in the upper digestive tract in patients of all ages. We excluded reviews, case series, case reports, conference proceedings, and animal studies. No language restriction was imposed. Two reviewers (T.W.Y. and Y.C.Y.) independently screened all titles and abstracts to identify potentially eligible studies. The full text of potentially eligible articles was retrieved and checked for inclusion by the two reviewers. If no consensus was reached between the two reviewers, a third reviewer (Y.P.H.) made the final decision. We conducted a study selection using EndNote version 17 (Thomson Research Soft, Stamford, CT, USA). Finally, we checked the reference lists of all included studies to identify additional relevant studies.
Two investigators (K.C.W.C. and S.C.H.) independently extracted data from the included studies. The following data were extracted from each selected study: the name of the first author; publication year; study design; country; inclusion and exclusion criteria; sample size; participants' age and sex; characteristics of the index test; reference standard; the number of true positive, false positive, false negative, and true negative cases.
Two researchers (T.W.Y. and Y.C.Y.) used the Quality Assessment of Studies of Diagnostic Accuracy-2 (QUADAS-2) to independently assess the quality of included studies [24]. This tool has four domains: patient selection, index test, reference standard, and flow and timing. The risk of bias and concerns regarding the applicability, except for timing domains, and the flow were assessed and rated as low, high, and unclear risk. We summarized the results using the Review Manager version 5.3 (The Nordic Cochrane Center, The Cochrane Collaboration, Copenhagen, Denmark). Disagreements were resolved through a discussion.

Subgroup Analysis
We performed a subgroup analysis based on potential factors that may affect the diagnostic accuracy. The results are summarized in Table 2. No significant differences in the sensitivity of conventional X-rays were observed among studies conducted in different continents, those including different types of FBs, those conducted in different settings, and those with different designs. However, the sensitivity of conventional X-ray was significantly higher when used to detect FBs in the esophagus than when used to detect FBs in not only esophageal locations (esophageal FBs = 0.85 and not only esophageal FBs = 0.35; p = 0.01). No significant subgroup differences in the specificity of conventional X-rays were noted for the aforementioned factors. Moreover, for the subgroup based on the age, nine studies [4,7,8,[10][11][12]14,17,21] focused on adults, one study [19] focused on pediatrics, and seven studies [5,6,9,13,15,16] focused patients of all ages. However, in those studies that focused on patients of all ages, no subgroup data based on adults or pediatrics were reported. Therefore, only results on the adult subgroup were pooled. The results indicated that the pooled sensitivity and specificity of conventional X-ray were 0.48 and 0.88, respectively. In contrast, the result from the only study evaluating pediatrics by Wai Pak et al. [19] showed the sensitivity and specificity of conventional X-rays were 0.16 and 0.99, respectively.

Subgroup Analysis
We performed a subgroup analysis based on potential factors that may affect the diagnostic accuracy. The results are summarized in Table 2. No significant differences in the sensitivity of conventional X-rays were observed among studies conducted in different continents, those including different types of FBs, those conducted in different settings, and those with different designs. However, the sensitivity of conventional X-ray was significantly higher when used to detect FBs in the esophagus than when used to detect FBs in not only esophageal locations (esophageal FBs = 0.85 and not only esophageal FBs = 0.35; p = 0.01). No significant subgroup differences in the specificity of conventional X-rays were noted for the aforementioned factors. Moreover, for the subgroup based on the age, nine studies [4,7,8,[10][11][12]14,17,21] focused on adults, one study [19] focused on pediatrics, and seven studies [5,6,9,13,15,16] focused patients of all ages. However, in those studies that focused on patients of all ages, no subgroup data based on adults or pediatrics were reported. Therefore, only results on the adult subgroup were pooled. The results indicated that the pooled sensitivity and specificity of conventional X-ray were 0.48 and 0.88, respectively. In contrast, the result from the only study evaluating pediatrics by Wai Pak et al. [19] showed the sensitivity and specificity of conventional X-rays were 0.16 and 0.99, respectively.  We performed a sensitivity analysis by excluding five studies [8,10,13,17,21] with a sample size of <100 to determine whether the results were influenced by the potential overestimate of the diagnostic performance from the studies with small sample sizes. We observed no significant effect of the sample size on results, with a pooled sensitivity of 0.62 (95% CI = 0.30-0.86) and a pooled specificity of 0.92 (95% CI = 0.84-0.96). The results are summarized in Supplementary Figure S3.

Discussion
To our knowledge, this is the first systematic review and meta-analysis to investigate the diagnostic performance of conventional X-rays for detecting ingested FBs in the upper digestive tract. The results of our meta-analysis demonstrated that the conventional X-ray has a sensitivity of 58% and a specificity of 94% for detecting FBs in the upper digestive tract. The area under the SROC curve was 0.91.
The sensitivity of a tool is to measure the proportion of positives that are correctly identified. A high sensitivity test is reliable when its result is negative since it rarely misdiagnoses those who have the disease. In our study, we found that the overall sensitivity of the conventional X-ray for detecting FBs in the upper digestive tract was low. This low sensitivity of conventional X-ray can be attributed to multiple factors. First, the sensitivity may be affected by the FB type, which varies among age groups. In children, coins are the most commonly ingested FB [26]. However, in adults, the most commonly ingested FBs are fish bones (9-45%), other bones (8-40%), and dentures (4-18%) [27]. Second, many fish bones were detected in patients using normal X-ray, but the radiopacity of fish bones is poor in certain fish species [28]. In our study, we explored this factor by subgrouping patients with only fish bones or those with fish bones and others. The sensitivity of conventional X-ray was low in both these subgroups, with no difference noted between the subgroups. In addition, we examined whether the sensitivity of conventional X-ray is affected by the inclusion of patients from different continents, different study settings, and the study design. We did not observe differences between studies including patients from Asia or those including patients from continents other than Asia, between studies including patients from the ED and those not only including patients from the ED, or between prospective and non-prospective studies.
FBs may not be observed when viewed against a bone or a dense soft tissue in the background, such as in the oropharynx and hypopharynx. In other words, the sensitivity of conventional X-rays could be affected by the location of FBs. In our subgroup analysis, we found that conventional X-ray showed higher sensitivity when used to detect esophageal FBs (sensitivity of 85%) than when used to detect no only esophageal FBs. However, this finding should be interpreted with caution, considering that most of these patients underwent conventional X-ray after the ENT consultation, and the flexible endoscopy result was negative [9][10][11][12]19,21]. In addition, these patients may have developed persistent or more severe symptoms, and hospitalization might have been arranged to perform rigid esophagoscopy under general anesthesia. In these patients, FB-related radiographic signs may frequently occur, including the presence of radiopaque density, air accumulation, and soft tissue swelling and loss of cervical lordosis. Luo et al. reported that the sensitivity of conventional X-rays increased with the number of signs combined and interpreted together [7]. On the basis of these findings, we do not suggest using conventional X-rays to exclude patients without FBs in the upper digestive tract.
Ruling in the presence of FBs in the upper digestive tract of patients is also crucial. Multiple overlapping structures of the soft tissue structures and variable patterns of laryngeal cartilage calcification can masquerade as FBs in the upper digestive tract [29]. By contrast, because most FBs, except for some radiolucent materials such as the bones of certain fish species, wood, and plastics, have higher densities than soft tissues and absorb more X-ray photons, they are more radiopaque [30]. Our study results showed that conventional X-ray has a specificity of 94% for detecting FBs in the upper digestive tract. This high specificity may be attributed to the type of FB, which most clinicians can correctly identify. In addition, the results implied that most attending physicians, otolaryngologists, and radiologists rarely fall into the common pitfall, originating from the normal variation of laryngeal cartilage calcification. Our subgroup analysis revealed that the specificity of the conventional X-ray was high, irrespective of whether the included FB was a fish bone, the location of the FB was the esophagus, the patients were from different continents or different study settings, and the studies were prospective or non-prospective. The overall pooled PLR was 10.1, indicating that clinicians could confirm the presence of FBs in the upper digestive tract when conventional X-ray showed positive results.
For clinical application, FBs are commonly lodged in the oropharynx and posterior hypopharynx, which may be detected through laryngoscopy. In most related studies, conventional X-ray was used after the FB was not detected through laryngoscopy as concern regarding the presence of FBs has persisted [7,[9][10][11]17,[19][20][21]. Therefore, applying this strategy can be reasonable. The main clinical problem in our study was that we observed 42% falsely negative cases after conventional X-ray examinations. Considering that 80-90% of FBs pass through the gastrointestinal tract spontaneously and that potential anesthetic risks and discomfort arise when further esophagoscopy or endoscopy is employed, the observation of the clinical course for a short period was reasonable. Further esophagoscopy or endoscopy should be reserved for patients with persistent or deteriorated clinical symptoms and when suspicion arises of the ingestion of sharp or pointed FBs that can increase the risk of perforation. In addition, studies have reported that computed tomography (CT) has satisfactory sensitivity, ranging from 85.7% to 100%, and specificity, ranging from 66.7% to 100% [10,11,13,17,21]. However, its high cost and radiation exposure limit its application as an initial screening tool. CT can provide clear information regarding the location of FBs and the related complications they cause. Therefore, it can be used as the second step and should be urgently performed when patients have symptoms and signs that suggest perforation or other complications that may require surgery.
This study had several limitations that should be addressed. Firstly, of the 17 studies, 11 [8][9][10][11][12][13][14][15]17,18,22] were retrospective, which may bias the estimate. However, we found that our findings were not affected by the study design through the subgroup analyses. Secondly, most of the included studies were rated as having a high risk of bias because patients were not enrolled consecutively or randomly, and no clear information regarding the cutoff value of the index test or the use of multiple reference standards was provided. Thirdly, we found considerable heterogeneity in pooled sensitivity and specificity, and this could attribute to the different types of FBs, different locations of FBs, and different disease spectrum of patients included in our meta-analysis. Fourthly, the influence of the technical details of the image, the density of FB, and the sizes of FB cannot be determined since most studies did not provide any information. Fifthly,14 [4,7,9,[11][12][13][14][15][16][17][18][19]21,22] of the 17 studies were conducted in Asia. Therefore, whether the findings of this study are applicable to patients from other continents should be investigated in future studies. Finally, only one study evaluating pediatric patients by Wai Pak et al. [19] showed the sensitivity and specificity of conventional X-rays were 0.16 and 0.99, respectively. Of note, the sensitivity and specificity of conventional X-ray remained low and high, respectively. Altogether, further well-designed prospective studies are warranted to clarify these limitations.
Overall, despite the aforementioned limitations, our study results support the recommendations of the ESGE guideline [5] for using a conventional X-ray to detect ingested FBs. Another strength of our study is that it is, to the best of our knowledge, the first metaanalysis to include a large sample size (N = 4809) and investigate this crucial clinical issue. In addition, no publication bias was detected from Deeks' funnel plot asymmetry test.

Conclusions
The overall sensitivity and specificity of conventional X-rays were low and high, respectively, when used to detect FBs in the upper digestive tract. Thus, we recommend not using a conventional X-ray to exclude patients without FBs in the upper digestive tract. Additional imaging studies or endoscopy examinations should be performed for at-risk patients.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/diagnostics11050790/s1. Table S1: Search strategy, Table S2: Type and location of foreign bodies in the included studies, Table S3: Summary of the performance estimates of diagnostic parameters, Figure S1: The methodological quality of the included 17 studies, Figure S2: Deeks' funnel plot asymmetry test for publication bias, Figure S3: Sensitivity analysis after the exclusion of five studies with a sample size of <100.