Swallowing speed is no adequate predictor of aspiration in Parkinson's disease

There is still a lack of a clinical test to reliably identify patients with Parkinson's disease (PD) being at risk for aspiration.

reaching the trachea), and "aspiration" (ie, bolus reaches the trachea). A severe form of aspiration is the silent aspiration, where the patient does not realize the bolus in the trachea. Food residues play an additional role in the pharynx, due to the danger of later aspiration.
It is of great importance to reliably identify PD patients at risk for aspiration requiring further testing and to exclude PD patients without relevant dysphagia from unnecessary interventions. Such a clinical screening tool should be easy and quick to applicate in daily clinical practice, cost-effective, non-invasive, and safe for the patient. Patients screened at risk for aspiration should then undergo further testing either with flexible endoscopic evaluation of swallowing (FEES) or videofluoroscopic swallowing study (VFSS). The only PD-specific questionnaire-based screening tool for swallowing problems was found to be not sufficiently predictive for FEES-proven aspiration. 5,6 Today, the 3-ounce water swallow test is frequently used to screen individuals with different diseases for aspiration risk-the test required to drink approximately 90 mL of water from a cup without interruption. 7 Criteria for referral for subsequent investigation are an inability to complete the task, coughing or choking as well as hoarse or wet voice either during or within 1 minute of test completion. However, the test is to be prone to over-referral to further examination (eg, by FEES or VFSS) and unnecessary dietary restrictions due to a high false-positive rate in a large heterogeneous collective of patients. 8 Another predictor used for clinically relevant dysphagia has been swallowing speed. Based on correlation with subjective abnormal swallowing, a threshold of 10 mL/s by using a cup was proposed. 9 This threshold was adopted by several studies. [10][11][12] However, a wide range of test volumes (90-150 mL) was applied, which may have significantly affected the resulting swallowing speeds. Furthermore, the study populations were heterogeneous regarding the underlying diseases and inclusion criteria (subjective complaint of abnormal swallowing, clinically suspected swallowing disorder. or unselected cross section design). When considering these four studies together, only four percent of all subjects were known to have PD.
Thus, it remains unclear if this threshold is reliable for PD patients.
Noteworthy, a mean swallowing speed of 7.0 mL/s has been found in 100 PD patients suggesting that a threshold of 10 mL/s might be too high for patients with PD. 13 While it is consensus that average swallowing speed decreases with age in healthy subjects, 14-16 data on gender differences in swallowing speed are controversial with the description of slower 10,15,17 or equal 16 speed in women when compared to men.
The main aim of the present study was to evaluate whether swallowing speed of water, that is, a defined volume of 90 mL water divided by the time needed, is a reliable screening test for FEESproven aspiration in PD patients and which cutoff value might be adequate for PD. We furthermore assessed a potential impact of age, gender, disease duration, disease severity, and cognition on swallowing speed in our PD cohort.

| Study design and ethical approval
The local ethics committee of the Medical Council Hamburg (trial number PV5089) approved this prospective, controlled, cross-sectional study, and all patients gave written informed consent. swallowing times for water could not be measured (premature termination by the examiner because of excessive aspiration of water in two cases, renouncement by the examiner because of preceding aspiration of puree and a teaspoon of water in one instance and premature termination by the patient in one case). Thus 115 patients remained for analysis. Control subjects negated a history of diseases of the central nervous system as well as swallowing problems based on a self-developed 6-item screening questionnaire.

| Assessments
All PD patients were examined during medical "on"-state. Non-motor symptoms, including dysphagia in question 3 (NMS-Quest 3), were evaluated by the non-motor symptoms questionnaire (NMS-Quest).

Key Points
• There is no well-established cutoff value for swallowing speed in Parkinson's disease patients when it comes to detecting aspiration. This study aimed to fill this gap.
• A wide range of potential cutoff values was prone to misallocation of patients and healthy controls.
• Swallowing speed is no useful screening instrument to predict aspiration in Parkinson's disease patients.
teaspoon of green-colored water inconspicuously, he or she was instructed to drink a standardized volume of 90 mL water at room temperature through a straw as quickly as safely possible. We chose a straw instead of a cup as a delivery method because it interferes less with endoscopy. The timer started with the first contact of the water with the lips (by eyesight) and stopped with the end of the last swallow (timed water test) with the lowering of the larynx (as observed during FEES). Swallowing speed was only calculated if the complete volume was applied. Penetration and aspiration of water were assessed according to the eight-step Penetration-Aspiration Scale (PAS) of Rosenbek,18 which is also validated for FEES. 19 Aspiration, that is, water passes below the vocal folds, is indicated by PAS values of 6-8.

| Statistical analysis
Quantitative data were illustrated with means and standard deviation (SD), and differences between groups were analyzed using t test for independent samples. Qualitative data were illustrated with frequencies and analyzed using Fisher's exact test. Correlation coefficients were based on Kendall's tau and are interpreted according to We used Clopper-Pearson confidence intervals (CI) for sensitivity and specificity; those for predictive values were calculated according to Mercaldo et al. 21 There are no widely consented cutoff values for sensitivity and specificity to assess a screening test for aspiration. In our opinion, a pair of 80% sensitivity and 70% specificity could be deemed as acceptable and a pair of 90% sensitivity and 80% specificity as good. All statistical tests were two-tailed, and the alpha level was set to .05. Statistical analyses were performed with SPSS, version 23 (IBM).

| Subject characteristics
Demographic and clinical characteristics of patients and controls are presented in Table 1. Although the quantitative scores for cognitive function (MOCA) and mood (BDI-II) were significantly worse in patients, the two groups did not differ substantially if patients and controls were each categorized into the following clinically relevant groups: cognitive impairment (ie, MOCA score below 26) and degree of depression (ie dependent on BDI-II score no, mild, moderate, or severe depression).

| Results of flexible endoscopic evaluation of swallowing
The main results are shown in Table 2. Leakage was observed regularly in patients, but attained severe extent in only five cases. In contrast, aspiration of water ( Figure 1) occurred in nearly every fourth patient, but in none of the controls. Table 3 illustrates the key findings. Swallowing speeds differed significantly between patients and controls. Though, the absolute difference in means was rather small (2 seconds). A relevant gender effect could only be found within patients with men swallowing faster than women.

| Swallowing speed
A receiver operating characteristic (ROC) analysis was carried out to determine an appropriate cutoff value for swallowing speed to detect aspiration of water reliably (see Figure 2). The points for the formerly proposed cutoff value of <10 mL/s and the point with the shortest distance to the upper left corner of the diagram (and therefore the best compromise of sensitivity and specificity) are tagged. The latter was equivalent to a cutoff value of <5.5 mL/s.
In Table 4  Applying the cutoff values to our healthy controls led to a high false-positive rate, which attained 69% if a threshold of <10 mL/s was used (Table S5).

| Influence of patient characteristics on swallowing speed
A significant correlation between increasing age and decreasing swallowing speed could only be found for men in the patient cohort and exclusively for women in the control cohort (coefficients of −0.25 and −0.72, respectively Figure 3 and Figure S5).
There was a weak to moderate correlation of swallowing speed with disease duration (coefficients of −0.20 for men and of −0.40 for women, Figure S6). We found a significant correlation of swal-

P value (T test)
Swallowing speed (mL/s)

| D ISCUSS I ON
We assessed for the first time swallowing speed of water as a potential predictive parameter for aspiration in PD patients compared with controls. FEES was applied as the gold standard examination to prove aspiration.
The usual cutoff value of 10 mL/s is assumed to indicate dysphagia. We not only found that this usual cutoff value is too high but also that even using an optimized, and almost twofold lower threshold of 5.5 mL/s is not suitable to predict aspiration with reasonable sensitivity and specificity.

| Swallowing speed in PD patients
Searching the literature revealed that the precedent terminology for swallowing speed is heterogeneous and includes particularly the terms "swallowing velocity," "swallowing capacity," "flux of ingestion," or "swallowing flow" as well. 13,14,17,22,23 Considering all these terms, we found five studies assessing swallowing speed primarily in PD patients but none with an objective evaluation of aspiration using either FEES or VFSS. 11,13,[23][24][25] Mean swallowing speed in our PD cohort (6.5 ± 3.9 mL/s) was significantly lower compared with controls (8.5 ± 3.2 mL/s) and in accordance to findings of Kanna and We did not count the number of swallows and therefore could not determine the exact volume per swallow, but prolonged swallowing speed in PD patients might be related to a reduced bolus size as one mechanism of compensation. 26   A prolonged oral phase might negatively influence the speed of swallowing due to its arbitrary innervation. 4 However, our FEES results suggest that the oral phase might be less relevant overall in PD patients as we found severe posterior leakage (as one potential correlate of an affected oral phase) in only five cases but aspiration in twenty-six cases. Thus, the pharyngeal phase seems to be more critical, but it is relatively underrepresented in swallowing times. This is underlined by a videofluorographic study, which found no differences in the duration of the oropharyngeal phase in PD patients with and without aspiration. 28 Besides, 3 of 119 patients (3%) were excluded from the study due to excessive silent aspiration. These patients would have been at relevant risk if a water swallow test with 90 mL had been applied blindly. Silent aspiration (PAS 8) of a lesser degree, which would not have been detected without direct visualization, was encountered frequently (n = 20, 17% of all patients).

| Influence of PD patient characteristics on swallowing speed
We found men swallowing faster than women (7.3 ± 4.2 mL/s vs 5.1 ± 2.8 mL/s) which also fits the results of Kanna and Bhanu 13 (7.2 ± 3.4 mL/s in men and 6.6 ± 2.8 mL/s in women). Noteworthy, we detected a statistically significant lower swallowing speed for women exclusively in the patient cohort but not among controls. The latter result conforms with a large study which found no significant sex-related differences in healthy subjects. 16 Age did not consistently correlate with swallowing speed as a correlation was found only in male PD and female control subjects.
The statistically most robust study found an age-related increase in the duration of swallowing only in the group of older aged participants, that is, 66 years of age or older. 16 Instead of a continuous decrease, these data might suggest a rather exponential decrease of swallowing speed as soon as a cutoff age is reached. Though male and female PD patients did not differ for mean age, subjects with an age of 80 years or older were exclusively found in the male group (n = 9, 12%). This may explain why we found no correlation in female PD patients. Our male control cohort was most likely underpowered (n = 16) to reveal a correlation of swallowing speed with age.
There was a weak to moderate correlation of disease duration with swallowing speed in men and women. This corresponds to the findings of Kanna and Bhanu. 13

| Swallowing speed in the control cohort
Noteworthy, the mean swallowing speed of our healthy controls It is recognized that sipping water through a straw (as our participants were instructed) differs significantly compared to the usual  16 Drinking from a cup resulted in higher swallowing speeds, 29 which might be promoted by a head reclination. However, this maneuver can be dangerous in patients with dysphagia as it promotes aspiration. 30 Hence, using a straw to avoid head reclination is common practice in FEES.
Another reason for slower swallowing speeds in controls may be the verbal instruction. If healthy subjects were instructed to drink in their usual manner, mean swallowing speeds were predominantly below 10 mL/s. 31 The instruction in most prior studies was, however, "to drink as quickly as comfortably possible" respectively "as quickly as possible," 9,11,15 which could have induced them to drink faster than our controls ("to drink as quickly as safely possible").
Several former studies applied higher volumes of up to 150 mL 9,10,15 instead of 90 mL. High amounts of water result in higher swallowing speeds. A study showed that the mean oral-pharyngeal transit time is significantly shorter for forced repetitive swallows compared to a single swallow, 23 that is, that the initiation process accounts for much of the swallowing time. The protocols of several prior studies, unlike our protocol, allowed to calculate swallowing speed even if the volume was drunk only partially. [9][10][11][12] This led, on the other hand, to an underestimation of swallowing speed. Of note, high amounts of water are discussed as critical because of interference with ventilation, which has been proven for a volume of 200 mL. 16

| Limitations of our study
The average volume per swallow was proposed as a more meaningful indicator as, in contrast to swallowing speed, it is discussed to be less prone to the applied volume of water. 24 It is a limitation of our study that we did not measure the number of swallows. Thus, the average volume per swallow, as well as the mean duration of a swallow, could not be calculated. However, reliable measurement of number of swallows needs a second examiner to assess it "hands-on," and to insist on submental EMG might impede the widespread application of this screening tool. Furthermore, we did not perform a reliability check regarding time measurement (eg, by a second examiner).

| CON CLUS IONS
In conclusion, swallowing speed cannot be recommended as a simple bedside test to predict aspiration in PD patients. The widely used threshold of <10 mL/s showed insufficient sensitivity and specificity and resulted in a high number of unnecessary further instrumental investigations. Even an optimized cutoff value of 5.5 mL/s with acceptable sensitivity and specificity is not reliable in terms of aspiration.
Overall, measuring swallowing speed is prone to methodological errors and not suitable as a screening instrument to predict aspiration in PD patients.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N S
CB and CP were responsible for conceptualization, methodology, and project administration; MB performed data curation, formal analysis, and visualization; AN, CB, CP, and MB carried out the investigations; The study was supervised by CB; MB wrote the original draft, which was reviewed and edited by AN, CB, and CP.