Abstract
Background and Aims:
Timed barium esophagram (TBE) is commonly used to evaluate esophageal motor disorders, though its accuracy is limited. The lack of standardized criteria for TBE interpretation, coupled with reliance on single measurements and outdated reference standards, limits its effectiveness. This study aimed to reexamine the accuracy of TBE interpretation using the Standards for Reporting of Diagnostic Accuracy Studies (STARD) approach and updated reference standards.
Methods:
Adult patients with esophageal dysphagia were prospectively enrolled from 2019 to 2022 and underwent motility testing with esophageal manometry, FLIP panometry, and TBE. TBE accuracy for predicting achalasia/FLIP+ esophagogastric junction outflow obstruction (EGJOO), as defined by Chicago Classification 4.0 and FLIP, was assessed using two approaches: barium column height >2 cm at 5 minutes/impacted tablet and a classification tree model.
Results:
The study included 290 participants: 121 (42%) with EGJ outflow disorders, 151 (52%) without, and 18 (6%) with inconclusive results. The optimal classification tree had three levels: maximum esophageal body width at the top, maximum EGJ diameter and barium height at the second level, and tablet passage at the bottom. The TBE column height and tablet approach had a sensitivity of 77.8%, specificity 86.0%, and accuracy 82.2%, while the classification tree model achieved a sensitivity of 84.2%, specificity 92.1%, and accuracy of 88.3%.
Conclusions:
TBE can accurately identify achalasia/FLIP+ EGJOO when using multiple metrics in a classification tree model. This provides a simple, standardized approach to TBE interpretation that is superior to traditional single-metric methods.
Keywords: achalasia, esophageal motility, functional lumen imaging probe (FLIP), manometry, lower esophageal sphincter, dysphagia
Graphical Abstract

LAY SUMMARY
Timed barium esophagram can accurately identify achalasia when interpreted using a multi-metric classification tree model. It offers a standardized approach which outperforms traditional methods relying solely on barium column height.
INTRODUCTION
Timed barium esophagram (TBE) is a diagnostic procedure for evaluating esophageal emptying at varying timepoints. It entails rapidly ingesting 200 mL of barium sulfate, followed by sequential radiographs at 1, 2, and 5 minutes to assess esophageal barium retention.1,2 To improve diagnostic accuracy, the inclusion of a 13 mm barium tablet has been recommended.3 TBE is appealing for the evaluation of esophageal motor disorders due to its broad availability, ease of implementation, and non-invasive nature, although it does require exposure to low dose radiation.
Within the diagnostic framework of the Chicago Classification version 4.0 (CCv4.0), TBE is recommended when high-resolution esophageal manometry (HRM) testing is inconclusive for achalasia or esophagogastric junction outflow obstruction (EGJOO).4 Despite the role of TBE in adjudicating inconclusive diagnoses, there are limited data on normative TBE values.5 Furthermore, data supporting the accuracy of TBE as a screening test for esophageal motility disorders, as well as clinically relevant test thresholds, remains limited.1,6 One study suggested defining an abnormal test as a barium column height > 5 cm at 1 minute, >2 cm at 5 minutes, or an impacted tablet.3
Guidelines from the American College of Gastroenterology consider esophagram to be a suboptimal screening test for esophageal motility disorders in patients with dysphagia. This assessment, however, was based on a study that compared esophagram (without a standardized TBE protocol) to high-resolution manometry (HRM) interpreted using criteria from the Chicago Classification version 2.0. Notably, this version of the Chicago Classification defined “ineffective esophageal motility” (IEM) and “nutcracker esophagus” as esophageal dysmotility, both of which have questionable clinical relevance. These criteria have since been substantially revised in CCv4.0, with nutcracker esophagus no longer considered a motility disorder.1,6
In light of the updated reference standard provided by CCv4.0, we sought to reexamine the accuracy of TBE interpretation. Furthermore, although TBE application often focuses solely on barium column height, we hypothesized that incorporating additional TBE data, such as esophageal anatomy (e.g., width of the esophageal body or esophagogastric junction [EGJ]), and/or integrating multiple TBE metrics into a predictive model, could enhance the diagnostic performance of TBE. Hence, this study aimed to assess the diagnostic accuracy of TBE for identifying disorders of EGJ outflow as defined by HRM/CCv4.0 and functional lumen imaging probe (FLIP) panometry, using the Standards for Reporting of Diagnostic Accuracy Studies (STARD) approach; Table S1.7
METHODS
Subjects
Adult patients (ages 18–89 years) who presented to the Esophageal Center of Northwestern Memorial Hospital for dysphagia evaluation between 2019 and 2022 and were referred for motility testing with upper endoscopy, HRM, FLIP, and TBE, were consecutively enrolled in a prospective NIH-funded study (P01 DK117824). The majority of patients completed a barium esophagram, and those who underwent a TBE were included; Figure 1A. Patients with a history of foregut surgery or esophageal mechanical obstruction (esophageal stricture, eosinophilic esophagitis, severe reflux esophagitis [Los Angeles classification C or D], or hiatal hernia >3 cm) were excluded, as these conditions can cause secondary esophageal motor abnormalities. Written informed consent was obtained from all participants. The study protocol was approved by the Northwestern University Institutional Review Board (IRB #STU00011084).
Figure 1 – STARD Chart for Timed Barium Esophagram Accuracy.

STARD charts illustrating the diagnostic performance of standard TBE interpretation and a classification tree model. Following STARD guidelines, performance metrics are provided with inconclusive results categorized as both “positive” and “negative.”
1A: STARD chart evaluating the diagnostic accuracy of TBE using previously published thresholds. TBE was considered positive for achalasia or FLIP+ EGJOO if the barium column height was >2 cm at 5 minutes or if the tablet was impacted.
1B: STARD chart evaluating the diagnostic accuracy of TBE using a classification tree. CART, classification and regression tree; EGJOO, esophagogastric junction outflow obstruction; FLIP, functional lumen imaging probe; HRM, high-resolution esophageal manometry; TBE, timed barium esophagram
Timed barium esophagram protocol and analysis
During TBE, patients were in the upright position and consumed 200 mL of low-density barium sulfate with anteroposterior images obtained at 1, 2, and 5 minutes.2 If the barium column cleared, patients were also administered a 12.5 mm barium tablet. In 80 studies (including 4 with an inconclusive motility diagnosis), barium clearance did not occur within 5 minutes, and the barium tablet was not given. TBE studies were analyzed by the study team members (LK, NG, DAF, AL), who were blinded to clinical characteristics, including HRM and FLIP findings. Esophagram analyses were performed using Visage software (San Diego, CA, USA) available through the Northwestern electronic medical record. At the 1-, 2-, and 5-minute images, the height of the barium column was measured vertically from the EGJ to the nearest millimeter and the column area (cm2) was measured by outlining the column using the Visage region-of-interest (ROI) tool. Clearance of the barium by a time point was recorded as a column height and column area of 0 cm/cm2. The maximum esophageal body width and EGJ width were measured at the time and location of greatest transverse dimension on review of all available esophagram images. Qualitative impression of the EGJ appearance (beaked, narrowed, or normal) and appearance of the esophagus (corkscrew/sigmoid, megaesophagus) were also noted.
HRM protocol and analysis
HRM studies were completed after a 6-hour fast using a 4.2-mm outer diameter solid-state assembly with 36 circumferential pressure sensors at 1-cm intervals (Medtronic, Shoreview, MN). The HRM assembly was placed transnasally and positioned to record from the hypopharynx to the stomach with approximately 3 intragastric pH sensors. After a 2-minute baseline recording, the CCv4.0 HRM protocol was performed with ten, 5-mL liquid swallows in a supine position and then five, 5-ml liquid swallows in an upright position.4 Studies were analyzed according to the CCv4.0 and blinded to clinical characteristics including TBE findings; Table S2.4 A median integrated relaxation pressure (IRP) of >15 mmHg was considered abnormal for supine swallows; a median IRP of >12 mmHg was considered abnormal for upright swallows.
In scenarios of inconclusive-HRM diagnoses (i.e. EGJOO or absent contractility with IRP at the upper limit of normal), FLIP was applied to determine the reference standard label.4,8,9 An inconclusive-HRM with reduced EGJ opening on FLIP was considered a conclusive disorder of EGJ outflow; an inconclusive-HRM with normal EGJ opening on FLIP was considered normal EGJ outflow; an inconclusive-HRM with borderline EGJ opening on FLIP was considered inconclusive EGJ outflow; Tables S2, S3. Therefore, disorders of EGJ outflow were considered as types I, II, and III achalasia and conclusive EGJOO (i.e. with abnormal FLIP). Disorders of peristalsis (and normal EGJ outflow) were considered as absent contractility, distal esophageal spasm, hypercontractile esophagus, and ineffective esophageal motility (IEM).4
FLIP protocol and analysis
The FLIP study was performed during sedated endoscopy using a 16-cm FLIP catheter (EndoFlip EF-322N; Medtronic) and analyzed as previously described.10–12 Endoscopy was performed in the left-lateral decubitus position with sedation using midazolam and fentanyl or propofol (and possibly other medications) with monitored anesthesia at the discretion of the anesthetist. Sedation choice was shown to not significantly impact FLIP motility findings relative to HRM diagnoses.13 After withdrawal of the endoscope and calibration to atmospheric pressure, the FLIP was placed transorally and positioned in the esophagus with 1–3 impedance sensors beyond the EGJ. This positioning was maintained throughout the FLIP study. Beginning with 40 ml, stepwise 10-ml FLIP distensions were performed increasing to a target volume of 70 ml, with each stepwise distension volume was maintained for 30–60 seconds.
FLIP data were exported to and analyzed with a customized program (available open source at http://www.wklytics.com/nmgi) to generate FLIP Panometry plots for analysis.14 FLIP analysis was performed blinded to clinical characteristics, including HRM and TBE findings. Analysis of EGJ opening omitted areas affected by dry catheter artifact and included the EGJ-distensibility index (DI) at the 60-mL FLIP and the maximum EGJ diameter achieved during the 60- or 70 mL fill volume as previously described; Table S3.14,15
Statistical analysis
The primary outcome was binary: identification of achalasia or FLIP+ EGJOO was classified as “positive”, while the absence of such disorders was classified as “negative”. TBE was the index test, with HRM and FLIP as the reference standards. A positive TBE result was defined according to the criteria recommended by Blonski et al., where a barium column height >2 cm at 5 minutes or an impacted tablet was considered positive.3 A sensitivity analysis using only barium height to identify a positive test was also performed. Additionally, a classification tree, developed as described below, was also applied to identify a positive TBE. A positive reference result was defined according to CCv4.0, with FLIP employed in cases where the HRM diagnosis was inconclusive.4
Prior to developing the TBE classification tree, the accuracy of individual TBE metrics, including column height and area at 1, 2, and 5 minutes, as well as maximum EGJ width and maximum esophageal body width, was evaluated using receiver operating characteristic (ROC) curves, calculation of the area under the ROC curve (AUROC), and construction of interval likelihood ratio (LR) tables. The intervals were selected based on round, interpretable numbers and to ensure adequate sample sizes within each interval.
The TBE classification tree was developed by evaluating models with different combinations of predictors, starting from TBE heights alone and progressively adding variables. Cases with inconclusive reference standards were excluded from the training and testing datasets during model development and selection but were reintroduced in the final performance assessment. Among the top three models with similar performance, the most parsimonious tree was selected, which incorporated barium column heights, tablet passage, maximum esophageal body width, and maximum EGJ width. Model robustness was assessed through bootstrap resampling, with performance evaluated on out-of-bag samples; supplemental methods. Classification was determined using a probability threshold of 0.5.
Index test performance was evaluated using sensitivity, specificity, accuracy, and AUROC. In the primary analysis, inconclusive reference test results were classified as “positive” to reflect clinical management, as these patients often exhibit abnormal esophageal manometry and FLIP findings, even if both are considered “borderline” abnormal. However, as recommended by the STARD guidelines, metrics are also reported with inconclusive results classified as “negative.” Reasons for inconclusive index tests included the absence of a tablet when barium column height was ≤ 2 cm at 5 minutes, lack of a timed component in the esophagram, or technically limited studies; Figure 1. These inconclusive index tests were excluded from the analysis.
All analyses were conducted using R version 4.4.0.16 Results were reported as mean (standard deviation; SD), or median (interquartile range; IQR), depending on data distribution. The 95% confidence intervals were calculated using the Wilson score interval for sensitivity, specificity, and accuracy, and the DeLong method for AUROC.17
RESULTS:
Subjects
The study cohort comprised 290 participants, with a mean age of 54 years; Table 1. Most participants were female (64%), overweight (mean BMI 28 kg/m2, SD 7). The most common HRM diagnosis was normal motility (29%). In total, 121 participants (42%) were diagnosed with achalasia or FLIP+ EGJOO, including 91 patients with achalasia (types I, II, or III) and 30 with EGJOO and a positive FLIP. The median (IQR) interval between endoscopy with FLIP and esophagram was 32 (9, 64) days and the median (IQR) interval between esophagram and HRM was 28 (7, 64) days.
Table 1 -.
Cohort Characteristics
| Overall (n = 290)1 | |
|---|---|
| Demographics | |
| Age | 54 (16) |
| Sex | |
| Female | 186 (64%) |
| Male | 104 (36%) |
| Body Mass Index (kg/m2) | 28 (7) |
| Scleroderma | 16 (5.5%) |
| PPI Use | 147 (55%) |
| Opioid Use | 18 (6.2%) |
| Days Between Studies | |
| Endoscopy/FLIP and HRM | 0 (0, 0) |
| Endoscopy/FLIP and TBE | 32 (9, 64) |
| HRM and TBE | 28 (7, 64) |
| Manometry | |
| Chicago Classification (4.0) Diagnosis | |
| Achalasia I | 27 (9.3%) |
| Achalasia II | 48 (17%) |
| Achalasia III | 16 (5.5%) |
| EGJOO - FLIP+ | 30 (10%) |
| Absent Contractility | 19 (6.6%) |
| Distal Esophageal Spasm | 6 (2.1%) |
| Hypercontractile Esophagus | 6 (2.1%) |
| Ineffective Esophageal Motility | 24 (8.3%) |
| Normal Motility | 84 (29%) |
| EGJOO - Normal FLIP | 12 (4.1%) |
| Inconclusive | 18 (6.2%) |
| IRP, supine, median (mmHg) | 18 (11, 28) |
| Basal EGJ pressure at End-Expiration (mmHg) | 18 (12, 29) |
| Functional Lumen Imaging Probe | |
| EGJ Distensibility Index at 60 mL (mm2mmHg−1) | 2.00 (1.10, 4.40) |
| EGJ Diameter, Max (mm) | 8.30 (6.00, 11.65) |
n (%); Mean (SD); Median (Q1,Q3)
EGJ, esophagogastric junction; EGJOO, esophagogastric junction outflow obstruction; FLIP, functional lumen imaging probe; HRM, high-resolution esophageal manometry; IRP, integrated relaxation pressure; PPI, proton pump inhibitor; TBE, timed barium esophagram
TBE Results by Esophageal Motility Category
Among the 121 participants with achalasia or FLIP+ EGJOO, the most common EGJ appearance was a beaked EGJ (45%), while 81% of the 151 participants without achalasia or EGJOO had a normal EGJ appearance; Table 2. Thirty-three participants with achalasia or FLIP+ EGJOO had an impacted tablet, although 62 did not receive a tablet during their TBE of whom 53 had barium retention at 5 minutes. Individuals with achalasia or FLIP+ EGJOO were observed to have a smaller maximum EGJ width (0.6 cm, IQR 0.5–0.8 vs. 1.1 cm, IQR 0.9–1.3) and a larger maximum esophageal body width (3.6 cm, IQR 2.8–5.0 vs. 2.2 cm, IQR 1.9–2.6) compared to those without these conditions. The distributions of TBE column heights by CCv4.0 motility classification showed greater column heights for all achalasia subtypes and FLIP+ EGJOO compared to other motility disorders. While DES demonstrated significant barium retention at 1 minute, this observation was based on only 6 participants; Figure 2A. Regarding maximum esophageal body width, all achalasia subtypes and FLIP+ EGJOO exhibited greater values compared to other motility disorders; however, absent contractility showed a maximum esophageal body width distribution similar to FLIP+ EGJOO; Figure 2B. Corkscrew esophagus was identified in 15 patients, including 7 with achalasia (3 with type II, 4 with type III), 2 with FLIP+ EGJOO, 1 with normal motility, and 2 with an inconclusive CCv4.0 diagnosis. Regarding FLIP metrics, IRP was greater among patients with achalasia and FLIP+ EGJOO (30.0 mmHg, IQR 22.5–38.5) compared to those without (11.0 mmHg, IQR 7.8–15.0). Meanwhile, EGJ diameter (6.0 mm, IQR 4.9–7.8 vs. 10.8 mm, IQR 8.6–13.0) and distensibility index (1.1 mm2mmHg−1, IQR 0.8–1.6 vs. 4.0 mm2mmHg−1, IQR 2.38–5.9) were smaller in the achalasia and FLIP+ EGJOO group.
Table 2 –
TBE Metrics Used for Classification Tree
| Not Achalasia/EGJOO1 (n=151) |
Achalasia/FLIP+ EGJOO1 (n=121) |
|
|---|---|---|
| EGJ Appearance | ||
| Normal | 122 (81%) | 31 (26%) |
| Narrow | 25 (17%) | 35 (29%) |
| Beaked | 4 (2.6%) | 55 (45%) |
| Tablet Passage | ||
| Passed | 129 (94%) | 26 (44%) |
| Impacted | 8 (5.8%) | 33 (56%) |
| Missing | 14 | 62 |
| Barium Column Height (cm) | ||
| 1 min | 0 (0, 4) | 11 (6, 18) |
| 2 min | 0 (0, 0) | 10 (2, 15) |
| 5 min | 0.0 (0.0, 0.0) | 8.1 (0.0, 12.8) |
| Barium Column Width, Max (cm) | ||
| 1 min | 0.0 (0.0, 1.1) | 2.90 (1.8, 4.6) |
| 2 min | 0.0 (0.0, 0.0) | 2.80 (1.0, 4.4) |
| 5 min | 0.0 (0.0, 0.0) | 2.40 (0.0, 3.9) |
| Barium Column Area (cm2) | ||
| 1 min | 0 (0, 4) | 29 (12, 49) |
| 2 min | 0 (0, 0) | 24 (3, 47) |
| 5 min | 0 (0, 0) | 17 (0, 34) |
| EGJ Width, Maximum (cm) | 1.1 (0.9, 1.3) | 0.6 (0.5, 0.8) |
| Esophageal Body Width, Maximum (cm) | 2.2 (1.9, 2.6) | 3.6 (2.8, 5.0) |
n (%); Median (Q1, Q3)
Classification tree development excluded patients with an inconclusive Chicago Classification diagnosis.
EGJ, esophagogastric junction; EGJO, esophagogastric junction outlet; FLIP, functional lumen imaging probe; IRP, integrated relaxation pressure; TBE, timed barium esophagram
Figure 2 – Timed Barium Esophagram Column Height and Maximum Esophageal Body Width by Esophageal Motility Classification.

2A: Box plots of barium column height at 1, 2, and 5 minutes, stratified by motility disorder.
2B: Box plots of maximum esophageal body width stratified by motility disorder.
Each box represents the interquartile range (IQR), with the median indicated by the central line. Whiskers extend to the most extreme data points within 1.5 times the IQR from the box edges. Points beyond the whiskers represent individual outliers.
EGJOO, esophagogastric junction outflow obstruction; FLIP, functional lumen imaging probe
Evaluation of TBE Metrics
An increase in likelihood of having achalasia or FLIP+ EGJOO (i.e. a LR greater than 1.0) was observed for the 1-minute column height of 8 to 10 cm (LR 2.2) and >10 cm (LR 7.1); Table 3. One-minute column height intervals < 8cm had a LR less than 1, corresponding to a decrease in probability for achalasia or FLIP+ EGJOO. At 2 minutes, only a barium column height between 0 and 2 cm had an LR less than 1. By 5 minutes, a column height between 2 and 5 cm was associated with a LR of 1.04, between 5 and 8 cm 2.81, and 8 to 10 cm LR 5.62. Barium column height at 2 minutes had the largest AUROC of 0.836 (95% CI 0.790–0.883), followed by column height at 1 minute (0.829, 95% CI 0.781–0.878), and column height at 5 minutes (0.795, 95% CI 0.747–0.844). Maximum esophageal body width and maximum EGJ width both demonstrated strong discriminant ability, with AUROCs of 0.867 (95% CI 0.822–0.913) and 0.856 (95% CI 0.811–0.901), respectively; Table 4.
Table 3 –
Interval Likelihood Ratios for Barium Column Height and Area at 1, 2, and 5 Minutes
| 1 Minute | 2 Minutes | 5 Minutes | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Barium Column Height (cm) | Positive Within Interval | Negative Within Interval | Interval LR | Positive Within Interval | Negative Within Interval | Interval LR | Positive Within Interval | Negative Within Interval | Interval LR |
| >10 | 68 | 12 | 7.1 | 62 | 5 | 15.5 | 45 | 1 | 56.2 |
| 8 – 10 | 16 | 9 | 2.2 | 11 | 2 | 6.9 | 18 | 4 | 5.6 |
| 5 – 8 | 8 | 15 | 0.7 | 10 | 8 | 1.6 | 9 | 4 | 2.8 |
| 2 – 5 | 7 | 9 | 1.0 | 6 | 6 | 1.3 | 5 | 6 | 1.0 |
| 0 – 2 | 22 | 106 | 0.3 | 32 | 130 | 0.3 | 44 | 136 | 0.4 |
| Barium Column Area (cm2) | |||||||||
| >50 | 29 | 1 | 36.2 | 24 | 0 | - | 18 | 0 | - |
| 25 – 50 | 38 | 4 | 11.9 | 35 | 1 | 43.7 | 27 | 0 | - |
| 10 – 25 | 25 | 16 | 1.9 | 23 | 10 | 2.9 | 28 | 6 | 5.8 |
| 5 – 10 | 6 | 15 | 0.5 | 7 | 6 | 1.5 | 3 | 8 | 0.5 |
| 0 – 5 | 23 | 115 | 0.2 | 32 | 134 | 0.3 | 45 | 137 | 0.4 |
Patients were classified as positive if they had achalasia or FLIP+ EGJOO, as determined by high-resolution esophageal manometry and, when indicated, functional lumen imaging probe. Likelihood ratios (LRs) indicate the probability of achalasia or EGJOO: LR > 1.0 suggests increased probability, LR < 1.0 suggests decreased probability, and LR = 1.0 indicates no change in probability. To calculate post-test odds, multiply pre-test odds by the appropriate LR.
EGJOO, esophagogastric junction outflow obstruction; FLIP, functional lumen imaging probe; LR, likelihood ratio
Table 4 –
Interval Likelihood Ratios for Maximum Esophageal Body and EGJ Width
| Esophageal Body Width, Maximum (cm) | Positive Within Interval | Negative Within Interval | Interval LR |
|---|---|---|---|
| >4 | 45 | 4 | 14.0 |
| 3 – 4 | 39 | 12 | 4.1 |
| 2 – 3 | 31 | 88 | 0.4 |
| 0 – 2 | 6 | 47 | 0.2 |
| EGJ Width, Maximum (cm) | |||
| >1 | 17 | 89 | 0.2 |
| 0.5 – 1 | 72 | 60 | 1.5 |
| 0 – 0.5 | 32 | 2 | 20.0 |
Patients were classified as positive if they had achalasia or FLIP+ EGJOO, as determined by high-resolution esophageal manometry and, when indicated, functional lumen imaging probe. Likelihood ratios (LRs) indicate the probability of achalasia or EGJOO: LR > 1.0 suggests increased probability, LR < 1.0 suggests decreased probability, and LR = 1.0 indicates no change in probability. To calculate post-test odds, multiply pre-test odds by the appropriate LR.
EGJOO, esophagogastric junction outflow obstruction; FLIP, functional lumen imaging probe; LR, likelihood ratio
Accuracy of TBE Classification Tree
The best-performing, 3-level classification tree constructed using TBE measurements had maximum esophageal body width at the top, maximum EGJ width and barium column height at the second level, and tablet passage at the bottom; Figure 3. Earlier splits are generally considered more important for classification, as they capture the most significant patterns in the data, influencing subsequent splits. Thus, maximum esophageal body width can be considered the most important variable for identifying achalasia or FLIP+ EGJOO, followed by maximum EGJ width, barium column height at 2 minutes, and tablet passage. The overall AUROC of the classification tree, when applied to the test data, was 0.953 (95% CI 0.905–1.00), with a sensitivity of 86.8% (95% CI 72.7–94.2%), specificity of 96.8% (95% CI 83.8–99.4%), and accuracy of 91.3% (95% CI 82.0–96.7%). Model robustness assessed through bootstrap resampling yielded a mean accuracy of 84.8% (95% CI 77.6–91.9%).
Figure 3 – Classification Tree.

Classification Tree for the Interpretation of Timed Barium Esophagram. This tree was constructed using a training set of 203 patients (75% of the total dataset), with inconclusive reference standards excluded. The performance of the classification tree is depicted on the test data (69 patients, 25% of the total dataset). Start at the top and follow the path through subsequent nodes based on the esophagram results until you reach the bottom. Each “leaf” at the bottom of the tree represents a classification outcome: a lighter shade of grey indicates the presence of achalasia or FLIP+ EGJOO, while a darker shade of grey indicates its absence. Within each leaf, the proportion of correctly classified patients for that specific class is displayed.
When applied to the entire dataset using the STARD approach, and classifying inconclusive reference results as “positive”, the classification tree achieved a sensitivity of 84.2% (95% CI 77.2–89.3%), specificity of 92.1% (95% CI 86.6–95.4%), and accuracy of 88.3% (95% CI 84.1–91.5%); Figure 1B, Table S4B. The typical TBE interpretation (barium column height >2 cm at 5 minutes or impacted tablet) had a sensitivity of 77.8% (95% CI 69.8–84.2%), specificity of 86.0% (95% CI 79.4–90.8%), and accuracy of 82.2% (95% CI 77.1–86.3%); Figure 1A, Table S4A. When inconclusive results were classified as “negative”, the classification tree attained a sensitivity of 89.3% (95% CI 82.5–93.6%), specificity of 87.6% (95% CI 81.8–91.7%), and accuracy of 88.3% (95% CI 84.1–91.5%), while the typical TBE interpretation showed a sensitivity of 84.5% (95% CI 76.6–90.1%), specificity of 84.3% (95% CI 77.8–89.1%), and accuracy of 84.4% (95% CI 79.6–88.2%). The sensitivity analysis using only barium column height >2 cm at 5 minutes (without tablet) to determine a positive TBE yielded the following results: when classifying inconclusive reference results as “positive”, the sensitivity was 58.3% (95% CI 50.0–66.1%), specificity was 90.1% (95% CI 84.3–93.9%), and accuracy was 74.8% (95% CI 69.5–79.5%). When classifying inconclusive reference results as “negative”, the sensitivity was 63.6% (95% CI 54.8–71.7%), specificity was 88.8% (95% CI 83.1–92.7%), and accuracy was 78.3% (95% CI 73.2–82.6%); Table S5, Figure S1.
Discussion
The major findings of this study indicate that while reliance on barium column height and tablet passage had limited ability to identify achalasia or FLIP+ EGJOO, a classification tree incorporating multiple TBE metrics was highly effective in identifying these conditions as defined by CCv4.0. Specifically, maximum esophageal body width, maximum EGJ width, barium column height at 2 cm, and tablet passage were sufficient to accurately classify nearly 90% of patients, achieving an AUROC of 0.95. The classification tree was developed using a machine learning method that systematically analyzed all available TBE metrics to identify the most predictive combinations. This provides a practical, standardized framework for clinicians evaluating patients with suspected esophageal motility disorders.
Previous studies have explored the use of esophagram and TBE in predicting esophageal motility disorders. One such study assessed the accuracy of barium esophagram in diagnosing esophageal dysmotility by comparing it to an HRM reference standard in 281 patients.6 They found significant disagreement between studies, with barium esophagram demonstrating a sensitivity of 69% and a specificity of 50% for detecting esophageal dysmotility. Notably, the barium esophagrams were not timed, and dysmotility was assessed subjectively by the interpreting radiologist. These findings were corroborated by another study that similarly demonstrated poor concordance between barium esophagram and HRM, with only 46% of patients showing concordant results.18 Compared to standard esophagrams, TBE offers the advantage of a standardized approach to evaluating esophageal disorders. The administration of a consistent volume and uniform timing for image acquisition enables the generation of quantifiable and reproducible metrics.
A subsequent study evaluated the role of TBE in differentiating achalasia from EGJOO and non-achalasia disorders.3 Their analysis found that a barium column height of >5 cm at 1 minute had a sensitivity of 94% and a specificity of 71%, while a column height of >2 cm at 5 minutes had a sensitivity of 85% and a specificity of 86%. The combination of barium column height and tablet passage increased the diagnostic yield to 100% for achalasia and EGJOO to 60%. This provided the recommended abnormal cutoffs and introduced a simple metric for distinguishing between normal and abnormal TBE. However, the study’s heterogeneous population, including various non-motility causes of esophageal retention (surgical, anatomic, functional), and use of the older Chicago Classification version 3.0 (CCv3.0) as the reference standard, complicate the interpretation. The diagnosis of EGJOO can be challenging, particularly with borderline IRP values, and some achalasia patients may be misclassified as EGJOO under CCv3.0 guidelines. Employing FLIP to clarify uncertain diagnoses, as recommended by CCv4.0, improves the accuracy of the reference diagnosis and strengthens the confidence in our analysis.4 Moreover, relying on a single barium column height cutoff value limits test accuracy. Further, the study only described the 1-minute and 5-minute column heights, whereas the 2-minute column height appeared to be the optimal time point to utilize column height measures to identify achalasia. The interval LR tables from this study demonstrate the pitfalls of using a single cutoff value, as the probability of disease can change exponentially with increasing barium column height.
Column height on TBE is a crucial measure of esophageal function (i.e. esophageal clearance), particularly in predicting treatment outcomes in achalasia.19 However, as a single metric, it cannot distinguish between the causes of esophageal retention, such as EGJ obstruction versus lack of bolus propulsion due to impaired peristalsis. For example, nearly half of the patients with absent peristalsis had a barium column height greater than 5 cm at 1 minute, and all of these patients continued to have a barium column height greater than 2 cm at 5 minutes. In contrast, among patients with FLIP+ EGJOO, half had a barium column height greater than 5 cm at 1 min (similar to absent contractility), and only 20% retained a barium column height greater than 2 cm at 5 minutes. This limitation affects the diagnostic utility for identifying achalasia or FLIP+ EGJOO, as evidenced by the low sensitivity and high specificity observed when relying solely on barium column height with a single value cut-off for TBE interpretation. Adding tablet passage alone to barium column height improves sensitivity, but at the cost of specificity, as demonstrated by our sensitivity analysis; Table S5, Figure S1. Moving from a single metric to the use of LR tables, as provided in this study, improves diagnostic accuracy while quantifying the impact of specific observations on the probability of disease. Moreover, incorporating multiple metrics into a classification tree for TBE interpretation improves discrimination between obstructive and non-obstructive processes, resulting in superior sensitivity, specificity, and accuracy. The metrics identified as important by the classification tree are not surprising, as they capture crucial aspects of esophageal physiology and emptying, specifically, bolus propulsion and EGJ opening. Esophageal body width correlated with peristalsis, with greater body width observed in patients with type I and II achalasia and absent contractility compared to those with disorders of peristalsis; Figure 2B. Barium column height reflects retention, EGJ width assesses EGJ opening, and tablet passage evaluates both bolus propulsion and EGJ opening.
There are several key strengths of this study. First, it utilized a rigorous reference standard, defined by the updated CCv4.0, HRM, and FLIP, and adheres to a standardized approach as outlined by the STARD statement.7 The study also employed an updated, informatics-driven approach to enhance TBE interpretation, incorporating additional and informative TBE metrics. Hence, this study supported using TBE metrics beyond just column height as it demonstrated the benefits of also including esophageal body width and EGJ width measured on esophagram. These additional metrics are easily interpretable and are available on any TBE performed, regardless of clinical setting (i.e. community or academic). Finally, the classification tree that incorporated these multiple TBE metrics is both easily interpretable and implementable, offering a practical alternative when more advanced methods, such as HRM or FLIP, are unavailable.
This study also has several limitations. Notably, there is no definitive gold standard for diagnosing esophageal motility disorders. Although the current state-of-the-art methods, including CCv4.0, HRM, and FLIP, were applied to assign labels independently of the index test (TBE), inconclusive labels were still encountered. These inconclusive results limit the diagnostic evaluation and accuracy testing of both individual TBE metrics and the classification tree model. As previously reported, discordance between HRM, FLIP, and TBE results may occur, underscoring the need for complementary diagnostic tests.20 This study serves as a guide to enhance the application of TBE in clinical management. Furthermore, the patient cohort was prospectively enrolled from an academic esophageal referral center, leading to an overrepresentation of achalasia and EGJOO, which may somewhat limit the generalizability to other clinical practice settings. Overrepresentation may also be due to the profile of patients referred for TBE, HRM, and FLIP, which likely skews towards enrollment of patients with major esophageal motility disorders including achalasia. However, this may also be viewed as a strength, as identifying achalasia is the most important outcome of esophageal motility evaluations, and EGJOO on HRM represents an important, yet often clinically challenging motility pattern. Approximately one-quarter of study participants did not receive a tablet during TBE, primarily due to persistent barium retention at 5 minutes. As a result, fewer achalasia and FLIP+ EGJOO patients were included in the tablet cohort. If all patients had received a tablet, the sensitivity of both the classification tree model and standard TBE interpretation would be expected to be higher. While our results are promising, future development of the classification tree model should include validation in a larger, multi-center cohort of patients to ensure generalizability and robustness.
In conclusion, this study demonstrated that TBE can accurately identify achalasia when analyzed using multiple metrics in a classification tree model. Future studies should explore the use of TBE metrics and models to identify more specific esophageal motor disorders (such as esophageal spasm and absent contractility), as well as validation in a larger, multi-center cohort. Ultimately, this study presented a standardized approach for TBE interpretation that is both easily applicable and superior to traditional single-metric methods, representing a significant advancement in the diagnosis of esophageal motility disorders.
Supplementary Material
WHAT YOU NEED TO KNOW.
Background and Context:
Timed barium esophagram (TBE) is widely used, but its accuracy in identifying achalasia is limited by the lack of standardized interpretation and reliance on outdated reference standards.
New Findings:
A classification tree model significantly improves accuracy in identifying achalasia compared to traditional methods that rely solely on barium column height.
Limitations:
This research used a binary outcome, which does not distinguish between EGJ outflow disorder types. A larger patient population is needed to further validate the classification tree.
Clinical Research Relevance:
TBE can accurately predict achalasia when interpreted using a multi-metric classification tree. This method offers a standardized, easily applicable interpretation and outperforms traditional single-metric approaches. It may serve as a practical alternative when more suitable methods, like HRM or FLIP, are unavailable.
Basic Research Relevance:
The classification tree identifies key metrics for predicting disorders of achalasia. These metrics reflect aspects of esophageal physiology and emptying; specifically, bolus propulsion and EGJ opening.
Financial support/Source of funding:
This work was supported by P01 DK117824 (JEP) and R01 DK137775 (JEP and DAC) from the Public Health service.
Potential competing interests/Disclosures:
JEP: Sandhill Scientific/Diversatek (Consulting, Grant), Takeda (Speaking), Astra Zeneca (Speaking), Medtronic (Speaking, Consulting, Patent, License), Torax/Ethicon (Speaking, Consulting), EndoGastric Solutions (Advisory Board), Phathom (Speaking, Consulting) DAC: Medtronic (Speaking, Consulting, License); Diversatek (Consulting); Braintree (Consulting); Medpace (Consulting); Phathom Pharmaceuticals (Speaking; Consulting); Regeneron/Sanofi (Speaking) Other authors have no conflicts to disclose.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Data Transparency Statement:
Data sources used for this study contain protected patient information and cannot be publicly shared. Select analysis code with patient identifying information removed are available upon request by contacting Ofer Fass directly.
References:
- 1.Gyawali CP, Carlson DA, Chen JW, Patel A, Wong RJ, Yadlapati RH. ACG Clinical Guidelines: Clinical Use of Esophageal Physiologic Testing. Am J Gastroenterol. 2020;115(9):1412–1428. doi: 10.14309/ajg.0000000000000734 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.de Oliveira J 1, Birgisson S, Doinoff C, et al. Timed barium swallow: a simple technique for evaluating esophageal emptying in patients with achalasia. AJR American journal of roentgenology. 1997;169(2):473–479. [DOI] [PubMed] [Google Scholar]
- 3.Blonski W, Kumar A, Feldman J, Richter JE. Timed Barium Swallow: Diagnostic Role and Predictive Value in Untreated Achalasia, Esophagogastric Junction Outflow Obstruction, and Non-Achalasia Dysphagia. Am J Gastroenterol. 2018;113(2):196–203. doi: 10.1038/ajg.2017.370 [DOI] [PubMed] [Google Scholar]
- 4.Yadlapati R, Kahrilas PJ, Fox MR, et al. Esophageal Motility Disorders on High Resolution Manometry: Chicago Classification Version 4.0©. Neurogastroenterol Motil. 2021;33(1):e14058. doi: 10.1111/nmo.14058 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kostic S, Andersson M, Hellström M, Lönroth H, Lundell L. Timed barium esophagogram in the assessment of patients with achalasia: reproducibility and observer variation. Dis Esophagus. 2005;18(2):96–103. doi: 10.1111/j.1442-2050.2005.00460.x [DOI] [PubMed] [Google Scholar]
- 6.O’Rourke AK, Lazar A, Murphy B, Castell DO, Martin-Harris B. Utility of esophagram versus high-resolution manometry in the detection of esophageal dysmotility. Otolaryngology--Head and Neck Surgery. 2016;154(5):888–891. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Cohen JF, Korevaar DA, Altman DG, et al. STARD 2015 guidelines for reporting diagnostic accuracy studies: explanation and elaboration. BMJ open. 2016;6(11):e012799. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Bredenoord AJ, Babaei A, Carlson D, et al. Esophagogastric junction outflow obstruction. Neurogastroenterology & Motility. 2021;33(9):e14193. doi: 10.1111/nmo.14193 [DOI] [PubMed] [Google Scholar]
- 9.Khan A, Yadlapati R, Gonlachanvit S, et al. Chicago classification update (version 4.0): technical review on diagnostic criteria for achalasia. Neurogastroenterology & Motility. 2021;33(7):e14182. [DOI] [PubMed] [Google Scholar]
- 10.Carlson DA, Kou W, Lin Z, et al. Normal values of esophageal distensibility and distension-induced contractility measured by functional luminal imaging probe panometry. Clinical Gastroenterology and Hepatology. 2019;17(4):674–681. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Carlson DA, Baumann AJ, Donnan EN, Krause A, Kou W, Pandolfino JE. Evaluating esophageal motility beyond primary peristalsis: assessing esophagogastric junction opening mechanics and secondary peristalsis in patients with normal manometry. Neurogastroenterology & Motility. 2021;33(10):e14116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Carlson DA, Kahrilas PJ, Lin Z, et al. Evaluation of Esophageal Motility Utilizing the Functional Lumen Imaging Probe. Am J Gastroenterol. 2016;111(12):1726–1735. doi: 10.1038/ajg.2016.454 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Pezzino EC, Pandolfino JE, Toaz E, Kahrilas PJ, Carlson DA. Endoscopic Sedation Type During FLIP Panometry Does Not Significantly Impact FLIP Motility Classification Relative to Manometry. Clin Gastroenterol Hepatol. Published online November 6, 2024:S1542-3565(24)00979-0. doi: 10.1016/j.cgh.2024.09.032 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Carlson DA, Gyawali CP, Khan A, et al. Classifying esophageal motility by FLIP panometry: a study of 722 subjects with manometry. Official journal of the American College of Gastroenterology| ACG. 2021;116(12):2357–2366. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Carlson DA, Prescott JE, Baumann AJ, et al. Validation of clinically relevant thresholds of esophagogastric junction obstruction using FLIP panometry. Clinical Gastroenterology and Hepatology. 2022;20(6):e1250–e1262. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.R Core Team. R: A Language and Environment for Statistical Computing. Published online 2023. https://www.R-project.org/
- 17.DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. Published online 1988:837–845. [PubMed] [Google Scholar]
- 18.Zambito G, Roether R, Kern B, Conway R, Scheeres D, Banks-Venegoni A. Is barium esophagram enough? Comparison of esophageal motility found on barium esophagram to high resolution manometry. The American Journal of Surgery. 2021;221(3):575–577. [DOI] [PubMed] [Google Scholar]
- 19.Vaezi MF, Baker ME, Achkar E, Richter JE. Timed barium oesophagram: better predictor of long term success after pneumatic dilation in achalasia than symptom assessment. Gut. 2002;50(6):765–770. doi: 10.1136/gut.50.6.765 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Reddy CA, Ellison A, Cipher DJ, et al. Frequent discrepancies among diagnostic tests for detecting lower esophageal sphincter-related obstruction. Neurogastroenterology & Motility. 2024;36(3):e14729. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data sources used for this study contain protected patient information and cannot be publicly shared. Select analysis code with patient identifying information removed are available upon request by contacting Ofer Fass directly.
