Accuracy of Dysphagia Standard Assessment (DSA®) bedside screening test: a flowchart for patient eligibility

The OD was investigated using a double-blind methodology between two operators. FEES was performed by two independent deglutologists. On the same day, a different viscosity screening test was performed at two separate times on all of the subjects, by one speech and language pathologist (SLP) blinded to the results of FEES. The FEES results were considered a reference test to establish the presence and severity of OD. In addition, the complete medical history for all patients was collected.

The study population included twenty healthy volunteers to assess normal swallowing physiology and seventy-four patients with suspected OD, investigated by means of bedside evaluations between June 2018 and July 2019 in different wards of the Policlinico Umberto I Hospital in Rome. Our study included 25 patients who had cerebrovascular disease, 24 patients with neurodegenerative diseases, 11 patients with neurological diseases, and 12 with other diseases (Table 1). Exclusion criteria were patients with Glasgow Coma Scale (GCS) < 14, with psychiatric disorders, and who had head and neck cancer surgery. Two patients with previous glossectomy for oral cancer were also excluded. These 2 cases were excluded because, in accordance with recent evidence, it is believed that post-surgical oral dysphagia should always make use of instrumental evaluation as there is no international consensus on which screening for dysphagia is preferred in patients with head and neck cancer [17].

All patients underwent a volume and viscosity screening test (DSA®—Dysphagia Standard Assessment; Nutrisens* medical) designed to protect the patient from aspiration by administering a bolus progressively increasing or decreasing in volume and viscosity. The purpose of the test was to identify the most suitable consistency for liquids based on the patient’s swallowing ability. The test included the following materials: three sachets of powder to make a gel with water, two spoons (3 ml and 8 ml), four glasses of natural cold water (2 glasses of 120 ml and 2 glasses of 90 ml and 60 ml, respectively), one 100-ml syringe to measure the water, a flexible straw to introduce liquids into the mouth where using a spoon was difficult or impossible, and finally one pulse oximeter. The patient was seated or semi-seated with their head bent slightly forward. The screening test included 4 different liquid volumes and consistencies (grades) ranging from natural water (grade 0) to thickened or gelled water (grade 3) (see Fig. 1). Grade 0 corresponded to performing swallowing (120 ml of water), grade 1 to likely aspiration risks (120 ml of water and one sachet of powder), grade 2 to more significant and frequent aspirations (90 ml of water and one sachet of powder), and grade 3 to significant and frequent aspirations (60 ml of water and one sachet of powder). The procedure would start at grade 0 or grade 3. Starting at grade 0 was suitable for all patients who had difficulty swallowing, whereas patients who already had swallowing disorders started at grade 3.

Starting at grade 0, the patient had to swallow 4 times half of a 3-ml teaspoon of natural water. If the patient coughed at the second swallow of half a 3-ml teaspoon of water, the caregiver continued the test by administering 4 times half of a 3-ml teaspoon of grade 1 water. If there were no problems, they increased the volume of the intake by administering 4 times a whole teaspoon of grade 1. The test was continued with 4 times of an 8-ml spoonful. If the patient was coughing, they would pass to grade 2 by administering 4 times of an 8-ml spoonful. If the patient swallowed correctly, they had to take 4 sips from the glass and if there were no problems the recommendation was for grade 2 (Fig. 1).

If starting at grade 3, the patient had to swallow 4 times half of a 3-ml teaspoon of thickened water with grade 3. If there were no problems, the test continued with the patient taking 4 times of an 8-ml spoon of grade 3. If completed without aspiration, they went to grade 2. If the subsequent swallowing did not give problems, they went to grade 1. If the patient coughed while drinking a spoonful, hydration with grade 2 liquids was recommended (Fig. 2).

During bolus administration, cough, wet voice after swallowing, and a desaturation of O2 ≥ 5% were considered significant enough for the risk of aspiration or penetration and were criteria for stopping the test.

All patients underwent a fiberoptic endoscopic evaluation of swallowing (FEES) by two experienced deglutologists using a flexible fiberscope (EF-N XION Nasopharingoskope, Germany). Each patient was placed in a seated or semi-seated position. As a first step, a bolus of semi-liquid consistency (jam or pureed fruit) was administered and then, secondly, a bolus with a liquid consistency (milk or juices). The patient was asked to hold the bolus in the oral cavity and to swallow it upon the command of the operator. The operation was repeated for the subsequent bolus. Each patient was given a score according to the Penetration Aspiration Scale (PAS) [18]. PAS is an 8-point scale, routinely performed in clinics and used to characterize depth and response to airway invasion during swallowing. The cut-off required to establish OD diagnosis was PAS ≥ 3.

The intra-rater reliability of DSA® and the inter-rater agreement of the evaluations of characteristics of swallowing dysfunction were determined by calculating Cohen’s kappa value. Sensitivity and specificity of DSA® relative to FEES for OD incidence were calculated in the target populations. Bayes’ theorem was used to calculate the positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (LR +), negative likelihood ratio (LR-), pre-test probabilities (PrTP), and post-test probabilities (PoTP) of the DSA® test. The range of confidence interval (CI) was 95%. Protocol study was conducted according to the principles and rules laid down in the Declaration of Helsinki and its subsequent amendments.

Regarding FEES, only about 22% of the patients examined were affected by OD. These data showed that the request for endoscopic evaluation of swallowing is very high compared to the number of cases that would actually need an examination. Instead, these resources could be used in targeted cases if BST was routinely performed in hospital wards. However, the systematic use of a test requires good reliability, consistent replication, and ease of administration. The DSA® test showed a 93% likelihood of identifying patients with dysphagia and a 78% likelihood of identifying healthy patients in a population that is neither homogeneous in terms of age nor in their pathology. Consideration should also be given with regard to PPV, since the prevalence of dysphagia is different between the groups examined. The predictive value is affected by this condition and, in fact, if a disease is considered pronounced within a population, the level of predictability for the same test, with equal sensitivity and specificity, grows when compared to a population where the frequency is lower. Faced with a non-homogeneous sample, the predictive values suggest that the probability of a patient turning out to be grade 0 and being healthy is 97%. Conversely, the probability of a grade of 1, 2, or 3 shows that dysphagia is less likely, at 55%, which explains why these patients should always be examined with FEES. These results lead us to think that the outcomes could be different if alternatively, we decided to apply the BST to only one type of target population. The likelihood ratios show the same trend; it is more likely that the patient who tests negative will be non-dysphagic than that of a patient who tests positive and has an undiagnosed swallowing disorder. The post-test probability of dysphagia with a positive test is, as stated, 55%; the post-test probability of non-dysphagia with a negative test is close to zero.

The DSA® test showed good accuracy (82%) in the whole sample, although it appears to have a lower validity in neurodegenerative diseases than in cerebrovascular diseases. In fact, the prevalence of any disease in the population can influence the predictive capacity of the test, which is why in neurodegenerative diseases where the prevalence of the disease (dysphagia) is higher [21], we see better predictive values.

The volume-viscosity swallowing test (V-VST) was designed in 2005 at Mataró Hospital, and several studies have evaluated the outcomes of the V-VST test in screening and diagnosing OD among different patients [12, 14, 16, 23, 24]. From recent evidence, V-VST had a diagnostic sensitivity for OD of 93.17%, specificity of 81.39%, and inter-rater reliability using Cohen’s kappa value of 0.77. The odds ratios for OD were LR + 5.01 and LR − 0.08 and the diagnostic odds ratio for OD was 51.18 [23]. According to the results in the literature, the DSA® test had a sensitivity of 93% and a specificity of 78%, with LR + 4.37 and LR − 0.08, although FEES was used as a reference test instead of VFSS. Relatively low values ​​of specificity and LR + indicate a high false positive rate and a low PPV value when the prevalence of OD is low. For this reason, the test was a good tool to exclude OD, but not to confirm the diagnosis.

Several studies have investigated oxygen saturation as a standalone tool or in combination with BST to determine the presence of a risk of OD and aspiration, but there are conflicting data on its usefulness [11, 25, 26]. Although the use of oxygen saturation could allow better analysis even of those patients with voluntary hypovalid or deficient cough and/or deficient laryngeal reflex, many other factors can influence the measurement of oxygen saturation [27]. Our results showed that only 50% of patients with aspiration and 25% of patients with penetration had oxygen desaturation > 5% at the test (Table 2). Although, as also found in the study by Clavè et al. [12], both oximetry and speech voice after swallowing did increase the diagnostic sensitivity of the test and the probability of identifying patients with silent aspirations or bolus penetration, however, the test was not effective in detecting a patient false negative with aspiration.

Overall, the tests showed negative outcomes in 13/72 (18%) of patients (1 false negative and 12 false positive). Analyzing individual patients, the possible reasons for false negative test results can be traced back to the onset of raclage during the administration of the test in 7/12 cases, to lack of cooperation from the patient in 4/12 cases, and in one of the twelve cases to a patient with tracheostomy who needed SLP rehabilitation for swallowing after non-commotional head trauma. The false negative patient could be traced back to the absence of laryngeal sensitivity and cough reflex, and it was actually difficult to measure saturation as the patient was not cooperating. In all patients with no cough reflex, the presence of wet voice or oxygen values was supportive, but it is not clear whether one or more of these conditions were missing as test reliability varies. In fact, O2 saturation alone does not have sufficient diagnostic accuracy to support routine use, which is why silent aspirations may be missed by the test [28].

One of the strengths of the DSA® test is the opportunity of not having to follow a fixed administration scheme but to proceed based on the patient’s risk of dysphagia and being able to choose whether to start with the administration of a thickened food (grade 3) or just water (grade 0). Starting from grade 0 is suitable for all patients who may have difficulty swallowing, while starting at grade 3 is appropriate for all patients who already have swallowing disorders. In addition, the ability to monitor oxygen saturation allows the test to be used even on patients at high risk of dysphagia, despite the limitations previously described.

The DSA® test also has its limitations: firstly, the duration of test administration, especially when starting from grade 0. Although choosing different starting degrees of liquid strength can reduce execution times, these are still quite long, on average 15–20 min, both for the preparation of the boluses and for their administration. A further limitation is the inability to positively test for all of the diseases we have investigated.

Overall, this study is a starting point that has allowed us to highlight false positives and negatives, and improved indications to apply the test safely, while excluding those patients who cannot exempt themselves from an instrumental evaluation. Based on our result analysis and direct patient observations, a flowchart has been developed to improve indications for DSA use (see Fig. 4). More in depth patient collaboration is the first point: if the patient is not cooperative, FEES should be considered rather than a DSA® test. If the non-cooperation is caused by GCS < 14, an alternative feeding method, e.g., nasogastric tube or parenteral nutrition should instead be immediately considered. If the patient cooperates, orofacial praxis, lingual motor skills, and voluntary cough should be evaluated; FEES should be preferred if deficient; DSA® test if valid. On the contrary, it is always recommended to perform FEES in patients undergoing head and neck surgery, in neurological patients with pathologies characterized by involuntary movement disorders (e.g., dyskinesias in PD, Huntington’s disease, dystonia), in patients with tracheostomy or with acute respiratory disease, in patients with gastroesophageal reflux disease raclage or smoker’s raclage, and in patients with problems of desaturation or with absence of cough reflex. In other cases, the DSA® test is recommended. However, if the test gives a grade 1–2–3 result, FEES would always be recommended to identify any silent aspirations (Fig. 4). Further studies are needed to understand whether, if applying this flowchart, test failure rates could be lower or if further patient selection evaluations are needed for the safe use of volume viscosity tests as a screening method for OD.