Correlations between bedside and instrumental endoscopic parameters in determining severity of dysphagia: an integrated clinical evaluation of safety and efficiency
Interaction between bedside and endoscopic parameters is of great interest in the management of patients with swallowing disorders. Our aim is to document if and how bedside parameters correlate with severity using endoscopic assessment. 556 consecutive patients (318M/238F, mean age 65.56±10.36years, range 18-91), were evaluated in our Swallowing Centre during 2008. All underwent bedside evaluation and beroptic endoscopic evaluation of swallowing (FEES), considering the pooling score (p-score) and the pooling sensation, collaboration and age score (p-SCA score) to express criteria of clinical severity of dysphagia. The correlation between the two tests (Spearman correlation coefcient) and their agreement to classify severity (Cohen’s kappa) was dened. After dichotomisation (cut-off: no risk/any kind of risk of aspiration), values of sensitivity and specicity were obtained after comparison with FEES results (gold standard). A close and signicant correlation between the p-score and p-SCA score was found (rho=0.88; p<0.001). The agreement among scores in attributing the categories of risk is moderate (Cohen’s Kappa=0.46; p<0.001).The p-score had a sensitivity of 96% and specicity of 60%, while the p-SCA score has a sensitivity of 98% and specicity of 40%. Our results suggest that including even a few parameter from bedside evaluation to an endoscopic score, the level of severity expressed by the latter, decreases. The evaluation of patients with swallowing disorders should consider as many elements as possible, deriving from non-instrumental and instrumental evaluation (integrated clinical evaluation).
Swallowing is a complex neuromuscular act, which requires sensory input and automated neuromuscular activities under real-time modulation guaranteed by the central nervous system in all its parts 1. In optimal physiological and functional conditions, all the ingested bolus is transferred to the stomach, for digestive processing, without invasion of the airway (false route) or residue. In dysfunctional or pathological conditions, material pooling or residue can invade the respiratory tract, upwards and/or downwards, leading to respiratory or nutritional complications (ineffective / inefficient swallow or dysphagia), respectively 2. Several conditions, acting as morbidity or comorbidities, can affect the quality of life (QOL) of dysphagic patients, offering a wide variety of events that are able to influence each other in guiding the clinical options adopted by the multidisciplinary team (MDT) that manages the patient 3.
During the processing of data leading to therapeutic planning, conditions of ineffective/inefficient swallowing (i.e. airway invasion/residue) must be documented with instrumental assessment. However, complete strategic planning cannot disregard other clinical not instrumental information, which is no less important in suggesting therapeutic strategies, even when apparently in contrast with instrumental evaluation. Data from the literature show that clinical non-instrumental evaluation alone tends to underestimate the risk of aspiration, while endoscopic evaluation tends to overestimate it 4. A reasonable balance between non-instrumental and instrumental evaluation seems to be the best combination that can offer the optimal strategy to the MDT, in order to obtain the best compliance of patients and caregivers. The balance mentioned above also seems to be the best way to achieve a concrete and reasonable evaluation of severity of impaired swallowing and the risk of complications 5. In addition, the real and logistical aspects of clinical practice must also be considered, as well as the availability of local resources 3.
In our Swallowing Center all patient referring swallowing disorders are routinely submitted to Bedside Swallowing Evaluation (BSE) 6 and endoscopic evaluation of the upper aerodigestive tract, performing a dynamic test with bolus for the study of swallowing (fiberoptic endoscopic evaluation of swallowing; FEES) 7. The evaluation of the oral and oesophageal phases of swallowing (O-FEES and E-FEES, respectively) 5 is performed considering the patient’s specific complaints or to answer to specific questions that the clinician poses. These approaches, only recently introduced in clinical practice, are not yet adequately supported by the international literature, but their value is purely local and far from any research intent.
The pooling score (p-score) and the pooling Sensation, Collaboration and Age score (p-SCA score) 8 are simple scores used in our endoscopic practice to define a criterion of severity of impaired swallowing, considering the material pooling/residue and the risk of aspiration. Material pooling/residue is evaluated in the hypopharynx/laryngotracheal cavities, considering specific anatomical landmarks, but also considering its amount and management (Tab. I).
After these preliminary considerations, the aims of this study are to evaluate: 1) how the severity of dysphagia, endoscopically defined, changes when BSE parameters are considered together with FEES parameters; and 2) the weight of BSE parameters in determining severity, as mentioned above.
Materials and methods
In this prospective study a sample of 556 consecutive patients (318 M/238 F, mean age 65.56 ± 10.36 years, range 18-91 years) seen at our Swallowing Centre during 2008 was considered. All patients underwent BSE performed alternately by two speech-language pathologists (SLPs) and a phoniatrician. During BSE a preliminary collection of information on pathologies, interventions and drugs was made. Subsequently, cognitive and language skills were assessed, observing the patient’s facies and carrying out manual exploration of the mouth and oropharynx. Finally, assessment of swallowing skills was performed with boluses of different consistency and volume, verifying the appearance of cough, throat clearing and modification of vocal quality.
After BSE, all patients were evaluated by a phoniatrician and submitted to FEES according to a protocol in use at our centre 3. Endoscopic evaluation was performed with a Storz endoscope (model 11101RP2, 30 cm long, 3.5 mm in diameter) and recorded with a workstation (Richard Wolf GmbH, Knittlingen, Germany). The patients were given three trials of different consistencies: 5 cc pureed, 5 cc liquid dyed with 5% methylene blue and 1/4 cracker. After each consistency the p-score and the p-SCA score were obtained (Tab. I). The p-score expresses a continuum of severity, summarised in a simple number ranging from 4 to 11, and clinically distributed over 4 levels of severity. The p-SCA score 8 is the p-score enriched with simple information achieved from BSE, namely sensation, collaboration and age: it ranges from 3 to 16, with the same clinical application on 4 levels of severity. The inter-rater and intra-rater reliability of the p-score was recently determined 8. The p-score considers the risk of aspiration occurring over the time (before, during and after swallowing) as an interaction between material pooling and false route (ineffective/inefficient swallowing). The correlation between material pooling and false route, i.e. safety and efficiency, has been recently evaluated 9,10.
The correlation between the p-score and p-SCA was determined with Spearman’s correlation coefficient. The agreement between the two scores was calculated (Cohen’s Kappa) considering the categories of risk corresponding to the total scores (no dysphagia, mild, moderate, severe). The categories of risk individualised with the two scores was studied with the aim of underlining possible systematic divergences in the attribution of the severity to individual cases. Subsequently, the p-score and the p-SCA score were dichotomised, dividing patients without risk from those with middle and high risk of aspiration. By comparison of dichotomic scores with the results of FEES (considered as the gold standard), the values of sensitivity and specificity were obtained.
All patients were over 18 years of age and consenting, in accordance with the Declaration of Helsinki. The study was approved by the local Ethical Research Committee. Statistical analysis was performed with Intercooled STATA 8.0 for Windows software.
A sample of 556 consecutive patients (318M/238F, mean age 65.56 ± 10.36 years, range 18-91 years) was evaluated. A close correlation between the p-score and p-SCA score in determining dysphagia severity (rho = 0.88) was found (Figs. 1A, B). The agreement among scores regarding the categories of risk showed moderate correlation (Cohen’s Kappa = 0.46; p < 0.001). Table II shows how the two scores classified patients in different categories of risk. Table III summarises the percentage of patients classified by the two scores: the two scores are correlated (Wilcoxon signed-rank test, p < 0.001). Subsequently, the judgement expressed by the scores was dichotomised, setting the cut-off point between patients without risk and those with any risk, with the purpose of comparing the scores with the results of FEES in terms of “aspiration” and to obtain, for both, values of sensitivity and specificity. The values of sensitivity and specificity are reported in Table IV.
In daily clinical practice the possibility of correlating signs and symptoms with residue/aspiration in patients with deglutition disorders due to different aetiologies is an important goal, and is needed to obtain better guidance for MDTs and prevent complications 10. The interaction between signs (instrumentally documented) and symptoms is a complex and intriguing relationship, but nonetheless is likely the best way to offset the trend of instrumental assessments to overestimate risks. With the contribution of instrumental evaluation, the risk of lost episodes of silent aspiration/penetration at the bedside is less, but the risk in generalising pathological random or extraordinary airway invasion events is higher 11,12. Considering our data, the p-score and p-SCA score are both useful tools to define severity, and both show the same statistical significant trend (Figs. 1A, B). However, considering the categories of risk attributed, the two scores seem to work in a different way. The p-SCA score tends to increase severity in the category with lower risk, while in those with higher risk it tends to be more cautious, attributing a category with lower severity in comparison to p-score (Tab. II). Overall, the patients classified as at risk of aspiration by the p-score are 50%, while the p-SCA score considers 67% of patients to be at risk (Tab. III). After dichotomisation (cut-off: no risk/any risk class) and comparison with the gold standard (aspiration documented during FEES), the two scores still show a different trend, offering different levels of sensitivity and specificity (Tab. IV). It is as if the p-SCA score recognises more patients at risk, more sensitive than the p-score, but also less specific with a greater risk of identifying false positive patients. In other words, both have high sensitivity to identify patients with any risk of aspiration, from minimal to high, but the p-score is more specific, and better at recognising false positives and therefore more reliable in correctly classifying patient without dysphagia and those with a risk of swallowing disorders. The low specificity of the p-SCA score also expresses a low ability of the score to identify patients who do not have swallowing disorders or who have low swallowing concerns, despite the more precise contribution of endoscopy results.
It is worth noting, as previously mentioned, that BSE alone, even if well conducted, underestimates the risk of silent aspiration, while endoscopy leads to generalisation of occasional episodes of false routes and residues. In addition, aspiration, as a marker of impaired swallowing, it is the most significant but not the only one, capable of determining the clinical severity of dysphagia and suitable for guiding the treatment plan and the activities of the MDT.
The simplest conclusion of the current experience, in accordance with our aims, seems to be that, by including even a few factors from the BSE in an instrumental score, the level of severity expressed by the latter decreases. Namely, the simple evaluation of sensation, collaboration and age tends to mitigate the judgment of severity expressed by the p-score, putting patients back into categories with lower risk: not only, but the greater impact of this contamination affects the ability of a “hybrid” score to correctly identify false negative patients, with a decrease in specificity. This experience also statistically quantified the weight of BSE parameters in reducing the specificity of a FEES parameter, such as the p-score.
The most relevant clinical implication of our work seems to be that the evaluation of patients with swallowing disorders should consider as many elements as possible, deriving from both non-instrumental and instrumental clinical evaluation: this could be considered as integrated clinical evaluation 5.
The main limitation of this preliminary work is that the p-SCA score considers a limited number of BSE parameters, although the logic supporting it is strong 10.
Figures and tables
|Pooling||Endoscopic landmarks||Sensation||Bedside parameters Collaboration||Age (years)|
|Site||Vallecule: 1 Marginal zone: 1 Pyriform sinus: 2 Vestibule/vocal cords: 3 Lower vocal cords: 4||Presence = -1 Absence = +1||Presence = -1 Absence = +1||+1 (< 65) +2 (65-75) +3 (> 75)|
|Amount||Coating: 1 Minimum: 2 Maximum: 3|
|Management||< 2 2 > 5 > 5|
|Score||p: 4-11||p-SCA: 3-16|
|No||163 61.98||100 38.02||0 0.00||0 0.00||263 100.00|
|Mild||11 7.33||139 92.67||0 0.00||0 0.00||150 100.00|
|Moderate||0 0.00||43 55.18||35 44.87||0 0.00||78 100.00|
|Severe||0 00.0||12 36.36||20 60.61||1 3.03||33 100.00|
|Total||174 33.21||294 55.61||55 10.50||1 0.19||524 100.00|
|No dysphagia At risk||264 261||50.29 49.71|
|No dysphagia At risk||174 350||33.21 66.79|
|Sensitivity||Specificity||Correctly classified||ROC Area [95% Conf. Interval]|
|P-score||95.56%||59.77%||65.90%||0.7766 (0.73908 0.81205)|
|P-SCA score||97.78%||39.63%||49.62%||0.6870 (0.64539 0.72653)|