Positive predictive value of tactile tracheal tube cuff ballotability above the suprasternal notch in detecting the correct positioning of the tracheal tube – A diagnostic accuracy study

ABSTRACT

Background and Aims:

Most endotracheal tube (ETT) depth assessment methods require extra equipment or expertise. Suprasternal tactile cuff ballotability is a bedside technique; however, its performance against a radiographic standard is not well characterised. This study assessed the accuracy of identifying an ETT cuff position above the lower border of the second thoracic vertebral body (T2LB) on fluoroscopy, interobserver agreement, and the relationship with tip-to-carina distance.

Methods:

This prospective observational study enroled adults undergoing neuroradiological procedures under general anaesthesia. After intubation, cuff ballotability was assessed by compressing the pilot balloon while palpating the trachea cephalad to the suprasternal notch. Ballotability, ETT marking at the mouth angle, and tip-to-carina distance were recorded. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) with 95% confidence intervals (CIs) were calculated. Interobserver agreement used Fleiss’ Kappa (k).

Results:

Of 85 screened patients, 76 were recruited [mean age 43.58 (13.47) years], and 72 underwent analysis. The correct placement prevalence was 72%. The sensitivity was 98.08% (94.90%–100%), specificity 60.00% (46.68%–71.32%), PPV 86.44% (78.53%–94.35%), and NPV 92.31% (86.15%–98.46%). The interobserver agreement was excellent (κ = 0.77, P < 0.001). The tip-to-carina distance was greater with ballotability (mean difference 17.9 mm; P = 0.001).

Conclusion:

In this study of non-obese adults under general anaesthesia, suprasternal tactile cuff ballotability had high sensitivity (98.1%) but moderate specificity (60.0%) for identifying the radiographic cuff position at or above T2 vertebral body, with excellent interobserver agreement (κ = 0.77).

INTRODUCTION

Various techniques are used to assess endotracheal tube (ETT) tip position, including auscultation, chest rise, capnography, chest radiography, and bronchoscopy.[1] However, auscultation has moderate accuracy (65–75%) and often misses endobronchial intubation; chest radiography, the reference standard, is limited by radiation, cost, and delays; bronchoscopy is invasive; and ultrasound requires expertise.[2,3] Clinical methods, such as auscultation and chest expansion, are unreliable and miss up to 55% of endobronchial intubations, especially in obesity, chronic obstructive pulmonary disease, or those under personal protective equipment.[4] Ultrasound-based techniques, including pretracheal visualisation of a saline-filled ETT cuff, improve accuracy; however, sonography requires specialised equipment, expertise, and time.[5] Tactile assessment of ETT cuff ballotability is a simple bedside technique without need for additional equipment. Ballotability is palpable cuff movement at the suprasternal notch when the pilot balloon is compressed and released.[6] This reverses Pollard and Lobato’s[7] method, in which suprasternal compression transmitted movement to the pilot balloon to confirm tracheal placement. Pollard’s method may risk tracheal mucosal injury, vascular compression, and bronchospasm.[8] The suprasternal notch corresponds to T1-T2. Detection of cuff movement at or just above the suprasternal notch may correlate with ETT tip placement 2–5 cm above the carina, considered optimal for most adults.[9,10]

Although cuff palpation and ballotability have been described previously,[11,12] prior studies mainly evaluated tracheal placement or used palpation for tube repositioning. This study quantified the accuracy of suprasternal ballotability for a prespecified radiographic cuff position and assessed interobserver agreement among observers with airway management experience. We hypothesised that ballotability above the suprasternal notch would have sensitivity for identifying a cuff at or above the lower border of T2 (T2LB), confirmed by fluoroscopy with the head and neck neutral. The secondary objectives included evaluation of interobserver agreement and ETT-tip-to-carina distance when ballotability was present.

METHODS

This prospective observational diagnostic accuracy study was conducted between May 2022 and January 2023 in adult patients undergoing neuroradiological procedures under general anaesthesia in a digital subtraction angiography (DSA) suite of a tertiary academic hospital in India. The study was registered prospectively in the Clinical Trials Registry-India (CTRI/2022/05/042859, dated 27/05/2022). Institutional Ethics Committee approval was obtained (Reference No: IEC-INT/2022/MD-88). Written informed consent was obtained from all participants a priori. The informed consent, information-to-participants section, included explicit disclosure that a single additional fluoroscopic image would be obtained for the study, with radiation exposure amounting to less than 0.1% of the exposure that they were being subjected to as part of their DSA procedure. The study followed the principles of the Declaration of Helsinki (2013) and Good Clinical Practice guidelines.

Adults aged ≥18 years of either gender with a body mass index (BMI) <30 kg.m^-² scheduled for DSA under general anaesthesia were included. The exclusion criteria were known cervical spine instability, anterior neck mass, prior neck irradiation, inability to visualise the complete ETT cuff on fluoroscopy, and refusal to participate (partial visualisation was included). A convenience sample of consecutively admitted patients was enroled.

All patients underwent pre-anaesthetic evaluation. Demographic data (age, gender, height, and weight), diagnosis, procedure type, and comorbidities were recorded. Standard American Society of Anesthesiologists (ASA) monitoring was applied. General anaesthesia was induced with fentanyl 1 µg·kg^-¹, propofol 1–2 mg·kg^-¹ (titrated to effect), and atracurium 0.5 mg·kg^-¹. After the train-of-four (TOF) count of 0 was achieved, tracheal intubation was performed with an appropriately sized cuffed oral ETT [Portex, Smiths Medical (now ICU Medical), Minneapolis, MN, USA] by an anaesthesiologist with ≥3 years of airway management experience who was not involved in the study assessments. Following confirmation of an end-tidal carbon dioxide (ETCO₂) waveform, the ETT was secured at the angle of the mouth. The cuff was inflated to an intracuff pressure of 30 cmH₂O using a calibrated cuff manometer prior to the ballotability assessment.

With the head in a neutral position (Frankfurt plane perpendicular to the table), ETT cuff ballotability was assessed by palpating the trachea with three fingers across its longitudinal axis, immediately cephalad to the suprasternal notch, while compressing and releasing the pilot balloon with the hand [Figure 1a and Supplementary Video]. Assessment was performed after confirmation of ETCO₂, a TOF count of 0, adequate anaesthetic depth (bispectral index 40–60 under sevoflurane in a 1:1 air–oxygen mixture), and controlled ventilation (tidal volume 6 mL·kg^-1, respiratory rate adjusted to maintain ETCO₂ 35–40 mmHg).

Figure 1.

(a) Clinical photograph demonstrating the operator placing three fingers of the left hand in the suprasternal notch along the long axis of the trachea while compressing the pilot balloon to assess endotracheal tube (ETT) cuff ballotability. (b) Fluoroscopic image of the neck showing the ETT cuff. The dashed red line represents the tangent at the lower border of the cuff

Ballotability was assessed by three anaesthesiologists with ≥1, ≥3, and ≥5 years of airway management experience. These categories were not validated competency thresholds. All observers completed training in at least 30 cases before participating in the study. Ballotability was recorded as “felt” or “not felt”, with the number of fingers over which movement was perceived; the pulsation strength was not graded. Each observer assessed independently and was not shown other observers’ findings; blinding to the ETT depth marking was not implemented.

A fluoroscopic image from the skull base to the xiphisternum was obtained to visualise the ETT cuff. The prespecified radiographic reference standard was the upper cuff border below the glottis and the tangent at the lower cuff border (TC) above the lower border of the T2 vertebral body (T2LB) [Figure 1b]. TC below T2LB was classified as an incorrect cuff position. The distance between the ETT tip and the carina (TTC) was measured. The radiologist interpreting the images was blinded to the ballotability findings. T2LB was selected as a pragmatic radiographic landmark based on descriptions relating the suprasternal notch to mid-tracheal anatomy, ensuring ETT tip placement approximately 2–3 cm above the carina and minimising the risk of endobronchial or subglottic cuff positioning.[9,13] After data collection, the treating anaesthesiologist, who was blinded to the radiological findings until the ballotability assessment and image acquisition were complete, was informed of the ETT position to allow adjustment to an optimal depth (2–5 cm above the carina).

The primary outcome was the sensitivity and positive predictive value (PPV) of cuff ballotability in detecting the correct ETT cuff position (above T2LB). Secondary outcomes included interobserver agreement for ballotability and comparison of TTC distance between ballotable and non-ballotable groups, as stated in the study objectives.

The sample size was calculated a priori, assuming a target sensitivity and specificity of 95% each. Exact 95% confidence intervals (CIs) for sensitivity and specificity were estimated using the Clopper–Pearson method.[14,15] Based on the literature, the pooled prevalence of correct tracheal tube placement was assumed to be 79%.[16,17,18] A sample size of 72 patients was estimated to provide CI widths of 0.13 for sensitivity and 0.29 for specificity. After allowing for a 5% dropout rate, the final target sample size was increased to 76 patients.

Continuous numeric variables were represented as mean (standard deviation) [SD] or median (interquartile range) as appropriate. Comparisons between groups were made using Student’s t-test or the Mann–Whitney U test according to the distribution. Categorical variables were presented as number (percentage) and compared using the Chi-square or Fisher’s exact test. A two-tailed P value <0.05 was considered statistically significant.

For the primary diagnostic accuracy analysis, a 2 × 2 table was constructed between the predictor (ballotability: felt vs. not felt) and the actual ETT cuff position on radiography (correct vs. incorrect). The PPV, negative predictive value (NPV), sensitivity, and specificity with 95% CIs were calculated. The rating provided by the observer with the most experience was used for primary analyses. Interobserver agreement between the three observers for cuff ballotability (felt/not felt) was assessed using Fleiss’ kappa.[19] Agreement for the number of fingers on which ballotability was perceived was also assessed.

The complete-case analysis of 72 patients represented the primary diagnostic accuracy analysis. Four enroled patients had incomplete visualisation of the ETT cuff on fluoroscopy and were not included in the primary 2 × 2 analysis. A best-case/worst-case sensitivity analysis was performed by imputing these four outcomes as correctly or incorrectly placed, respectively. For descriptive purposes, the cohort was divided into “incorrectly placed” versus “correctly placed” ETTs. The manuscript conforms to the Standards for Reporting of Diagnostic Accuracy Studies (STARD) 2015 reporting recommendations for diagnostic-accuracy studies.[20] Data analysis was performed using R version 4.2.3 in RStudio (RStudio Team, 2023; RStudio PBC, Boston, MA).

RESULTS

Of 85 patients screened, 76 were recruited between 29 May 2022 and 15 January 2023; 6 were excluded for BMI >30 kg.m^-², and 3 declined participation. Of the 76 enroled, 4 had ETT cuffs not visualised on fluoroscopy, yielding a primary analysis cohort of 72 patients [Figure 2]. The mean (SD) age was 43.58 (13.47) years. Demographic and comorbidity profiles did not differ significantly between correctly and incorrectly placed groups, except for a higher proportion of ASA II patients in the incorrectly placed group [Table 1].

Figure 2.

Flowchart of the study

Table 1.

Demographic variables and characteristics of correctly and incorrectly placed endotracheal tubes (ETTs)

	Overall (n=76)	Incorrectly Placed ETT (n=20)	Correctly Placed ETT (n=56)	P	SMD
Age (years)*	43.58 (13.47)	44.05 (15.30)	43.35 (13.30)	0.85a	-0.049
Height (cm)*	167.05 (10.92)	167.45 (12.21)	166.56 (10.76)	0.77a	-0.078
Body weight (kg) *	69.83 (13.84)	69.35 (15.84)	69.42 (13.34)	0.98a	0.005
BMI (kg.m-2) *	24.84 (3.49)	24.43 (3.56)	24.89 (3.55)	0.62a	0.131
Gender, n (%)
Female	41 (53.95)	12 (60.00)	26 (50.00)	0.62b	0.202
Male	35 (46.05)	8 (40.00)	26 (50.00)
ASA class, n (%)
I/II/III/IV	28 (36.84)/20 (26.32)/26 (34.21)/2 (2.63)	3 (15.00)/10 (50.00)/6 (30.00)/1 (5.00)	23 (44.23)/9 (17.31)/19 (36.54)/1 (1.92)	0.02b	0.878
Hypertension, n (%)	38 (50.00)	12 (60.00)	24 (46.15)	0.43b	0.280
DM, n (%)	14 (18.42)	6 (30.00)	7 (13.46)	0.17b	0.409
COPD, n (%)	2 (2.63)	1 (5.00)	1 (1.92)	0.48b	0.169
ETT size (mm), mean (SD)	7.71 (0.49)	7.55 (0.43)	7.77 (0.51)	0.07a	0.467
ETT mark at angle of mouth (cm), mean (SD)	21.08 (1.16)	21.15 (1.09)	21.02 (1.23)	0.66a	0.113
ETT tip to carina (mm), mean (SD)	35.21 (15.09)	18.81 (6.02)	41.52 (12.55)	<0.001a	2.308
Length of trachea (mm), mean (SD)	117.06 (11.96)	114.33 (10.14)	118.11 (12.52)	0.19a	0.332

^aTwo-sample t-test, ^bChi square/Fisher-exact test; P<0.05 has been considered significant; Data are represented as mean (SD) or n (%). SMD: Standardised mean difference; SD: Standard deviation; ETT: Endotracheal tube; DM: Diabetes mellitus; COPD: Chronic obstructive pulmonary disease; ASA: American Society of Anesthesiologists; BMI: Body mass index

The ETT cuff was correctly placed (lower border of cuff at or above T2LB on fluoroscopy) in 52 of 72 patients, yielding a prevalence of correct placement of 72%. The 2 × 2 confusion matrix is presented in Table 2.

Table 2.

2×2 table/confusion matrix of the primary outcome (presence of ballotability versus prespecified radiographic ETT cuff position) and diagnostic performance metrics

	Incorrect tracheal tube position	Correct tracheal tube position	Total
Not felt (not ballotable)	12 (60.00%)	1 (1.92%)	13(18.06%)
Felt (ballotable)	8 (40.00%)	51 (98.1%)	59 (81.94%)
Total	20 (100.00%)	52 (100.00%)	72 (100.00%)
Performance metrics
	Original values - Analysis ignoring incomplete data (n=72)	When missing outcomes assumed as correctly positioned (n=76)	When missing outcomes assumed as incorrectly positioned (n=76)
Prevalence	72.22% (61.88%-82.57%)	73.68% (63.78%-83.58%)	62.48% (57.97%-78.87%)
Sensitivity	98.08% (94.90%-100%)	96.43% (92.26%-100%)	98.08% (94.99%-100%)
Specificity	60.00% (46.68%-71.32%)	60.00% (48.99%-71.10%)	54.17% (42.96%-65.37%)
Positive predictive value	86.44% (78.53%-94.35%)	87.10% (79.56%-94.63%)	82.26% (73.67%-90.85%)
Negative predictive value	92.31% (86.15%-98.46%)	85.71% (77.85%-93.58%)	92.86% (87.07%-98.65%)

Values are in percentage and (95% Confidence Limits) for the performance metrics

Cuff ballotability demonstrated a high sensitivity of 98.08% (95% CI, 94.90%–100%) and a PPV of 86.44% (95% CI, 78.53%–94.35%) for identifying a correctly placed ETT. The specificity was 60.00% (95% CI, 46.68%–71.32%), and the NPV was 92.31% (95% CI, 86.15%–98.46%). The positive likelihood ratio was 2.45 (95% CI, 1.43–4.19), and the negative likelihood ratio was 0.03 (95% CI, 0.01–0.23). The area under the receiver operating characteristic (ROC) curve was 0.79.

Best-case and worst-case sensitivity analyses (imputing the four cases with partial cuff visualisation as correctly or incorrectly placed, respectively) did not alter the performance metrics by more than 5% [Table 2]. When ballotability was felt across all three fingers, correct ETT cuff positioning was observed in all cases.

Interobserver agreement for detecting the presence or absence of ballotability was excellent (κ = 0.77, P < 0.001) across the three observers with ≥1, ≥3, and ≥5 years of airway-management experience, respectively. Agreement for the number of fingers on which ballotability was perceived was fair to good (κ = 0.48, P < 0.001). The kappa value for no ballotability (fingers = 0) was 0.77 (P < 0.001), and for ballotability felt on all three fingers, it was 0.52 (P < 0.001). The strength of pulsation/ballotability was not graded; therefore, its relationship with radiographic cuff position could not be evaluated.

The ETT TTC distance was significantly shorter in the incorrectly placed group than in the correctly placed group [mean difference 22.7 mm; 95% CI, 16.9–28.6 mm; P < 0.001; Table 1]. When comparing ballotable and non-ballotable groups, the mean TTC distance was significantly greater when ballotability was present (mean difference 17.9 mm; 95% CI, 12.8–22.9 mm; P = 0.001, two-sample t-test). The TTC distance increased significantly with the number of fingers on which ballotability was perceived [one-way analysis of variance (ANOVA), P < 0.001; Figure 3a]. Figure 3b illustrates the variability in the radiographic cuff level across the three-finger categories; the relationship suggests a graded association rather than a precise one-to-one estimate of cuff depth. None of the ETTs were endobronchial or at the carina in our cohort.

Figure 3.

(a) Boxplot demonstrating that the endotracheal tube tip was farther from the carina as the number of fingers on which ballotability was felt increased (P values are Bonferroni-adjusted). (b) Bar graph showing the relationship between the number of fingers on which ballotability was perceived and the vertebral level at which the lower border of the endotracheal tube cuff was positioned. The within-category variability illustrates that the finger count is associated with, but does not precisely determine, the radiographic cuff depth. ETT: Endotracheal tube

DISCUSSION

Our study evaluated whether suprasternal ETT cuff ballotability could detect an ETT position cephalad to the carina with the cuff infraglottic and tracheal (not bronchial or oesophageal). We found 72% of tracheal tubes were “correctly placed”, with the cuff’s lower border cephalad to the lower border of the T2 vertebra (appropriate depth). ETT cuff ballotability had a high PPV of 86.44% and a sensitivity of 98.08%, with excellent interobserver agreement. We found a negative likelihood ratio of 0.03, indicating that without ballotability, the ETT was 33 times more likely to be at an inappropriate depth. These findings support the recommendation of ballotability as a screening tool for suprasternal and tracheal positioning of the ETT cuff in adults. Ultrasound, fibreoptic bronchoscopy, capnography, and radiographic tip-to-carina assessment provide complementary information on placement and depth.[2,5,21,22] Ultrasound-based methods are accurate but depend on equipment and operator expertise. Although not a head-to-head comparison, ETT cuff ballotability can be an adjunct for detecting appropriate ETT depth without additional equipment. Auscultation remains widely used but has recognised limitations for identifying endobronchial intubation.[22] Ballotability showed higher sensitivity than auscultation estimates, but direct comparison is limited by differences in populations, reference standards, and test execution. Ballotability can therefore complement, rather than replace, capnography and clinical assessment.

Fixed-depth recommendations based on external ETT markings reportedly achieve high rates of correct placement,[23] but in this study, the ETT marking at the mouth angle did not differ significantly between correctly and incorrectly placed tubes. This supports individualised assessment over a single external-depth formula, although this study was not designed to compare these approaches. The physiological basis of cuff ballotability is pressure transmission from the pilot balloon to the inflated cuff. This transmission may be influenced by cuff design, length, geometry, compliance, manufacturer-specific construction, inflation volume and pressure, and cuff position relative to the tube tip. Since this study used a single low-pressure, high-volume cuff design, the observed tactile response may not generalise to other ETT designs, including tapered or higher-pressure cuffs. We also did not grade palpable pulsation strength, which could provide additional information about cuff-to-palpation transmission and should be evaluated in future studies.

The 17.9-mm mean difference in TTC distance between the ballotable and non-ballotable groups supports an association between ballotability and a more caudal ETT tip position. However, this does not guarantee a safety margin for individual patients. Tracheal length and the relationship between the T2 landmark, vocal cords, and carina vary with anthropometric characteristics; therefore, a fixed vertebral landmark cannot ensure an identical glottis-to-carina margin in all patients. These findings should be interpreted in the context of previous tactile techniques. Rezvani Kakhki et al.[11] described tactile assessment for confirming tracheal versus oesophageal placement, while Pattnaik and Mittal et al. described cuff palpation/ballotability in other clinical contexts.[12,24] The distinction of this study is that we quantified the diagnostic performance of suprasternal ballotability against a fluoroscopic cuff-position criterion and examined interobserver agreement. Importantly, absent ballotability cannot distinguish a cuff that is too high from one that is too low and could not be evaluated as a marker of endobronchial intubation because no endobronchial or carinal positions occurred.

Greater ballotability across the three-finger field was associated with greater TTC distance; however, Figure 3b also shows variability in cuff vertebral level within each category. Therefore, finger number should not be interpreted as a substitute for radiographic or other depth assessments. Interobserver agreement for binary ballotability was excellent across observers with different experience levels, supporting reproducibility after standardised training, but not a universal training threshold. The study did not assess competency in untrained operators or compare formal curricula. Clinically, cuff ballotability may serve as an adjunctive screening observation after intubation but does not confirm optimal depth or malposition. Waveform capnography remains essential, while imaging may be required in certain cases.

This study has the following limitations: First, it was a single-centre convenience sample of non-obese adults undergoing neuroradiological procedures under general anaesthesia with neuromuscular blockade and neutral neck positioning. The findings may not apply to emergency airways, critically ill or intensive care unit patients, trauma, cervical-spine immobilisation, obese or paediatric patients, or patients with altered neck anatomy. Second, T2LB on fluoroscopy was a pragmatic reference rather than an accepted gold standard for ETT depth; although TTC distance was measured, bronchoscopy was not used to confirm the glottis-to-carina relationship. Third, tactile assessment is operator-dependent. Although observers received supervised training and interobserver agreement was excellent, no competency threshold was established. Fourth, ballotability strength was not graded, so weak versus strong pulsation could not be assessed. Fifth, all ETTs used a low-pressure, high-volume cuff from a single manufacturer/design, limiting extrapolation to other ETT designs. Sixth, moderate specificity means ballotability can occur despite incorrect cuff position. Finally, the sample size was planned primarily around sensitivity precision, and specificity CI remained wide. Larger multicentre studies with varying ETT designs and patient populations, and comparisons with ultrasound, bronchoscopy, and depth-assessment methods, are warranted.

CONCLUSION

Suprasternal ETT cuff ballotability had high sensitivity (98.08%, 95% CI 94.90–100%) but moderate specificity (60.00%, 95% CI 46.68–71.32%) for identifying ETT position above the border of T2 on fluoroscopy. Interobserver agreement was excellent (κ = 0.77), and ballotability was associated with a greater ETT TTC distance. These findings support cuff ballotability as a screening technique to identify the optimal ETT depth; however, validation across populations and ETT designs is required.

Study data availability

De-identified data may be requested with reasonable justification from the authors (email to the corresponding author) and shall be shared after approval as per the authors’ Institution policy.

Disclosure of use of artificial intelligence (AI)-assistive or generative tools

The authors report that AI-assistive and generative AI was not used in their research or preparation of this manuscript.

Author contributions

VS: Data curation, investigation, methodology, writing - original draft, review and editing, approval of final version of manuscript. VG: Formal analysis, software, methodology, validation, visualisation, writing - original draft, review and editing, approval of final version of manuscript; SV: Project administration, investigation, software, resources; HB: Conceptualisation, project administration, resources, methodology, validation, writing - review and editing, approval of final version of manuscript. LNY: Validation, writing - review and editing.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.