Thin liquids present a high risk of aspiration and aspiration pneumonia in individuals with neurogenic oropharyngeal dysphagia [1]. Although there are many physiologic reasons for aspiration of thin liquids, rehabilitative techniques to solve these problems are limited [2]. Thus, compensatory techniques that require implementation with every swallow are often recommended. A common compensatory technique to reduce the risk of aspiration of thin liquids, regardless of physiologic etiology, is the use of thickened liquids. However, many patients complain or refuse to consume thickened liquids due to poor taste or texture [2]. Dysphagic stroke patients on thickened liquid diets frequently do not meet their fluid intake requirements [3]. When patients do not drink, they are at increased risk of malnutrition and dehydration. Therefore, management of neurogenic oropharyngeal dysphagia often involves working with patients to maintain adequate nutrition and hydration while decreasing their risk of aspiration and aspiration pneumonia. If a range of palatable liquid taste properties could be identified that reduces the risk of laryngeal aspiration and penetration, the potential benefit to patients could be significant.

The critical research giving impetus to this study was done by Logemann et al. [4] which reported a beneficial effect of sour taste on several swallowing measures in patients with neurogenic oropharyngeal dysphagia. However, all subjects reported the sour stimulus was unpalatable, making its use as a therapeutic strategy limited. Adding a second taste to a mixture will partially inhibit the impression of the first taste. Such a reduction in perceived intensity of a mixture component is called mixture suppression [5,6]. A time-honored strategy for ameliorating the negative sensory impact of strongly acidic foods and beverages is the addition of sweet agents. This strategy has the double action of both inhibiting sourness and adding a second highly palatable taste quality to the item [7,8]. Although it is not known whether the reduction in sour impact by the addition of a sweet taste will detract from the swallowing improvements as noted by Logemann et al. [4], it should almost certainly improve the palatability of the stimulus. Recent research has indicated the elderly demonstrate similar patterns of sweet–sour mixture suppression (sucrose–citric acid and aspartame–citric acid mixtures) compared with a young group, and, thus, the elderly may find these mixtures as palatable as the young [9]. Since there is evidence that some neural mechanisms for taste mixture suppression are central, rather than peripheral receptor phenomena [10,11], the possibility exists that the incoming sour signal to the nucleus tractus solitarius (NTS) is yet unattenuated by mixture inhibition and therefore should be just as effective as a pure sour stimulus in affecting swallowing actions.

The aim of this research was to investigate the effect of sour (citric acid) and sweet–sour taste (citric acid–sucrose mixture) on (1) description of swallowing impairment using the penetration–aspiration scale (PAS) (Table 1) [12], (2) stage transition duration (STD), and (3) frequency of spontaneous swallows following taste and water stimuli in nursing home residents with neurogenic oropharyngeal dysphagia. Natural swallowing (no command to swallow) with teaspoon and cup delivery methods was used to aid evaluation of taste stimuli as a clinical strategy. It was hypothesized that a highly acidic tastant (citric acid) and a palatable sucrose–citric acid mixture (similar to lemonade) would elicit a shorter stage transition than water and improve swallowing safety. An increased number of spontaneous swallows was expected after the introduction of taste stimuli due to continued taste stimulation (i.e., residual taste) and increased saliva production. Finally, it was expected that cup delivery would result in more aspiration than teaspoon delivery.

Table 1 Eight-point penetration–aspiration scale by Rosenbek et al. [12]
Full size table

Materials and Methods

Subjects

Nineteen residents in three skilled nursing facilities in central New York were recruited for the study. A resident in the nursing homes, or the person responsible for his/her health care decisions, was approached to participate in the study if the resident had a neurologic medical diagnosis, was currently taking thickened liquids, and the attending physician approved of the participation. Written informed consent was obtained. The Cornell University Human Subjects Committee and the medical director, director of nursing, and administrator of every nursing home approved the study. If the resident desired, a comprehensive FEES® was performed as compensation for their participation at the end of the study protocol.

In order to show improvements from baseline, it was necessary to identify individuals with observable difficulty in swallowing. Seven residents were eliminated because of either resident request to terminate the study prior to its completion (5/7) or there was no observed difficulty safely swallowing water (2/7). One additional subject was eliminated because, except for the initial water trial, no evidence of oropharyngeal dysphagia with thin liquids was observed for the rest of the study or afterward during a comprehensive FEES®. He drank an entire glass of apple juice, using a straw, with no signs or symptoms of oropharyngeal dysphagia. It was concluded this subject could safely manage all thin liquids.

The remaining 11 subjects comprised an older population (mean age = 80.1 years, SD = 13.8, range = 52.6–102.8). They resided in the nursing home for many years (mean = 5.4 years, SD = 4, range = 0.5–13.5) (Table 2). All but one subject (who was developmentally disabled) acquired their neurologic dysphagia due to injury or disease, according to the medical chart. They were currently receiving thickened liquids, based on a previous evaluation by the facility speech-language pathologist. All subjects were in stable medical condition. They were not receiving speech-language pathology services at the time of the study. Six of the 11 subjects were diagnosed with some form of dementia. All subjects were at least one-year post onset of their neurologic accident or disease. Three subjects had a history of aspiration pneumonia. Subjects took an average of 3.8 medications per day (SD = 2.1, range = 1–7).

Table 2 Subject characteristics
Full size table

FEES was conducted in a quiet, private, well-lit room within each nursing home by an experienced endoscopist (CAP). A licensed nurse employed by each nursing home presented the stimuli and assisted in monitoring the subject. Four subjects had family members or staff attend their study and a research assistant (graduate student in food science, sensory evaluation) was present to aid in data collection during the last 13 subjects.

Stimuli

To control water taste, deionized water was the solute for all stimuli. It was produced locally from a Cole–Parmer ultrafiltration system consisting of two Universal ion X changer cartridges (1506-25) and two Research ion X changer cartridges (1506-35). Citric acid monohydrate ACS Reagent 5949-29-1 (Sigma Chemical Co., St. Louis, MO; Lot #37H1006, DOT-E 9275) and sucrose (local grocer) were used to make the stimuli. The stimuli were deionized water, 2.7% w/v citric acid, and a 1.11% w/v citric acid–8% w/v sucrose mixture. No flavoring was added. The high level of citric acid (2.7% w/v) was estimated to be isosour to the acid level presented in the sour bolus described by Logemann et al. [4], as assessed by sensory panel testing [13]. The sweet–sour mixture concentrations were found to be within the “just right” range by consumers in previous sensory testing and resembled commercially available lemonade, but without lemon flavoring [14].

Pilot testing of the water-based stimuli using green food coloring revealed poor light reflection and inadequate coating of the mucosa to identify the path of the bolus during FEES. The addition of white food coloring (titanium dioxide) to the sour and sweet–sour solutions resulted in turbidity and light absorption measurements similar to that of milk (skim and 2% fat), as assessed by a turbidometer (Hach 2100N Turbidometer) and spectrophotometer (Spectronic® 20 Genesys™ spectrophotometer). Clinical trials also confirmed increased visualization of the bolus path given both white and green food coloring. It is believed that white food coloring alone will not allow adequate discrimination of the stimuli from mucous, although this was not pilot tested. Thus, green (3 ml/L, McCormick & Co., #6412CH, #A1029) and white (0.25% w/v, Germantown International Limited, Product #124491000 Enhance Plus) food coloring were added to each stimulus.

Five milliliters of water and sour and sweet–sour samples were premeasured and labeled into 30-ml clear plastic cups with lids. One hundred twenty milliters of water and sour and sweet–sour samples were measured at the nursing home using a graduated cylinder and poured from 1-L clear plastic bottles into labeled Sweetheart® cold drink paper cups (300 ml). All cups were precut into “nosey” cups to aid cup drinking (i.e., reduce head extension) by cutting a 7-cm × 3-cm hemisphere in the rim of the cup. Since a therapeutic beverage would have a chilled serving temperature, all samples were refrigerated overnight, packed in thermal coolers with dry foam, and transported to the nursing homes. Although all samples were initially chilled to 7.1°C or below, the ambient temperature in the nursing homes raised the serving temperatures to an average of 8.9°C (SD = 3.7°C, range = 2.6–13.9°C, n = 8).

Procedure

A Pentax model FNL-10P2 transnasal fiberoptic endoscope and MiniVision™ portable digital recordable endoscopy unit were used to collect the swallowing data for all subjects except the first five evaluations. Those evaluations were conducted with a Stryker Endoscopy 810 Video Camera and were the only evaluations that included audio recordings. All data were recorded or downloaded to S-VHS videotapes using a Panasonic AG1970 videocassette recorder. A SMPTE Time Code generator/reader (Burst Electronics) inserted a continuous display of time (in 1/30-s increments) onto the videotape. Hundredths of a second calculations were based on 1/30-s increments to aid comparison with previous studies.

The stimuli were presented to each subject in a nonrandomized sequence to maximize safety since there are data to presume sour stimuli may improve swallowing but no data exist for sweet–sour stimuli. No command to swallow was given; subjects were instructed to swallow the samples as they normally would. Sixty-second delays were imposed upon insertion of the endoscope and after every sample. An imposed delay between trials allowed time for the peripheral taste stimulation to approximate baseline [15], since rinsing the oral cavity between samples was not possible using FEES. Immediately upon insertion of the endoscope, subjects were informed they could swallow whenever they wished but no command to swallow was given. Subjects spontaneously dry swallowed an average of 0.36 swallow during the first minute after insertion (SD = 0.67, range = 0–2). This is below the average number of dry swallows observed in normals (3/min) in response to the insertion of the endoscope [16]. The endoscope was positioned with the distal tip at the level of the uvula allowing a view of the posterior wall, base of tongue, lateral walls, and endolarynx. This position is sometimes called “home” position. The only exception to home positioning was after each sample was swallowed. At this time, the scope was briefly lowered to the endolarynx to observe for aspiration or penetration.

Each subject was initially presented with water (3 trials of 5 ml via teaspoon followed by 3 trials of cup drinking) until the first incidence of penetration or aspiration occurred or when the initiation of the swallow response appeared to be delayed beyond mean liquid normative value (1.61 s), given no swallow command [17]. After the first observation of penetration, aspiration, or delayed pharyngeal swallow of water, no further water trials were presented to maximize safety. Two 5-ml trials via teaspoon of the high sour liquid (2.7% citric acid) were then presented, followed by two 5-ml-teaspoon trials of sweet–sour mixture (1.11% citric acid–8% sucrose). Two cup trials of sour and sweet–sour liquids, respectively, were presented if safe tolerance at the 5-ml volumes was observed. Safe tolerance was defined as effective expulsion of any aspirated material out of the larynx. Termination of the trials occurred whenever aspirated material was not effectively expelled in two consecutive trials, except if the aspirated material was 5ml of sweet–sour liquid. If this occurred, cup trials of sour were presented, since 5 ml sour trials were previously found to be safe. During cup trials, subjects were instructed to “take a normal sip” from the cup, independently if possible or with assistance from the nurse. A balance was used to weigh each cup sample pre/post consumption. Density of the samples was measured using a hydrometer. The total volume consumed per sip was calculated in milliliters according to sample density and weight per sip.

Data Analysis

All swallows were analyzed at normal speed, slow motion, or frame-by-frame with no audio feedback as many times as necessary to determine the swallowing measures of interest. The following swallowing measures were assessed for each swallow: (1) description of the first swallow of each trial, using the penetration– aspiration scale (PAS) [12]; (2) stage transition duration (first frame of bolus arrival when endoscope was in home position to first frame of complete white out); and (3) number of spontaneous dry swallows observed 60 s after the first swallow of each trial. Dry swallows are defined as nonfeeding swallows, i.e., no stimulus was given. Stage transition duration as calculated during endoscopy is not exactly the same as measured during videofluoroscopic swallowing studies (VFSS) (it will be slightly longer with FEES by a few frames), but it can describe the relative risk of aspiration due to delayed pharyngeal swallow response [16]. Timing to count dry swallows began at the end of the first swallow. One swallow was defined as a complete white out in clear response to laryngeal elevation. White out due to laryngeal pumping (unsuccessful attempts to swallow with partial laryngeal elevation), head/neck movement, or loss of visualization due to secretion adherence to the tip of the scope of the endoscope was not counted as a swallow.

Initial analysis showed difficulty assigning a PAS aspiration rating to the swallows. Several swallows were rapid with no visual residue, with a cough occurring immediately after white out. Therefore, the bolus location that elicited the cough was unavailable to view. Thus, a simpler three-category rating scale that described swallowing behaviors was employed to rate a swallow: (1) No material in the airway (clear airway); (2) Penetration (bolus entered laryngeal vestibule but remained above the true vocal cords), or (3) Aspiration (bolus entered below the true vocal cords). It was hypothesized that taste stimuli might decrease aspiration or penetration in subjects who showed these dysphagic symptoms with water.

Reliability

Interjudge reliability was calculated on 10 random swallows (10% of total) selected by a random number generator and copied onto a S-VHS videotape in that sequence. Another rater with extensive FEES experience judged the swallows and was blind to the type of sample and subject. Raters were allowed to view the swallows in normal motion, slow motion, and frame-by-frame motion as often as desired to determine a measurement and rating.

Interrater reliability showed a strong linear relationship in the two temporal measures judged (first frame of bolus arrival and complete white out) (Pearson’s r = 0.99978). Perfect interrater agreement occurred on 9/20 temporal measures, with a mean difference of 0.22 s noted on 9 remaining measures (range = 0.03–0.53, SD = 0.23). Overall, interrater agreement determining STD and using the three-point category scale was excellent (90% for both). Interrater reliability on the number of swallows counted after the first swallow was good (Pearson’s r = 0.88, p < 0.01). Interrater agreement was lower counting swallows (60%), with all differences ±1 swallow.

Intrajudge reliability was calculated on 15 random swallows (15%) judging three temporal measurements. They were the first frame of bolus arrival, complete white out, and the first frame that white out ended. The number of spontaneous swallows observed after the first swallow was also counted. These ratings occurred approximately 2 months after the initial ratings. The swallows were identified by number only on the videorecording to reduce investigator bias. Ninety-six percent agreement (43/45) of the temporal measurements were within 1 video-generated time frame (1/30 s); the remaining swallows were within two time frames. There was 87% agreement (13/15) in counting the number of spontaneous swallows occurring after the first swallow. The number of swallows was greater by one and less by one in the two disagreements. Intrajudge agreement of all swallows was 95% (97/102).

Statistical Methods

SAS v8.0 (SAS Institute, Cary, NC) was used to perform the statistical analyses. A p-value of 0.05 was considered significant. Multinomial logistic regression was performed with stimuli and delivery method as independent variables (3 × 2 repeated measures factorial design) and the swallowing category ratings as the dependent variable. Odds ratio calculations of these data were performed to aid interpretation.

A mixed analysis of variance (ANOVA) (3 × 2 × 2 factorial design consisting of stimuli × delivery method × repetitions) was performed to investigate the effect of taste stimuli on the number of spontaneous dry swallows that occurred 60 s after a stimulus was swallowed. Bonferroni post hoc analysis was performed to aid interpretation. One data point was omitted from the analysis. One subject coughed immediately and vigorously after the first trial of 5 cc water and continued to cough and swallow 14 times over the next 60 s. This was considered to be an outlier response and was omitted from the ANOVA. It is not possible to ascertain the number of swallows required to clear the oral cavity of a stimulus using FEES. Thus, every swallow observed after the initial stimulus swallow was counted as a dry swallow.

Results

Significant results of the statistical analyses are shown in Table 3. One hundred two swallows were analyzed from 11 subjects. One swallow was not analyzed due to secretions that blocked visualization for several seconds after the swallow, despite numerous clearing attempts. Of the 102 swallows analyzed, 29 were water samples, 39 were acid samples, and 34 were mixture samples. Sixty-three swallows were presented by teaspoon and 39 were by cup delivery. Subject No. 4 used a straw instead of a cup to take a normal sip because she could not drink from a cup and typically used a straw.

Table 3 Summary of significant results
Full size table

All subjects aspirated water except subject No. 12, who penetrated water. Although the study was designed to maximize safety by proceeding to 2.7% citric acid immediately after the first observation of abnormal swallowing with water, this did not occur in two subjects. The endoscopist rated each swallow immediately, without the benefit of videotape review. Thus, two subjects received a second trial of water. In each instance, the second trial of water confirmed aspiration as observed in the first rating.

Results of this analysis showed citric acid improved swallowing (i.e., less aspiration and penetration) compared with water. The palatable mixture was not significantly different from water in improving a swallowing response (z = 1.33, p = 0.18). Each subject’s response to the stimuli is provided graphically (Figs. 1, 2, 3), grouped according to his/her swallowing response to the stimuli and delivery methods. Aspiration was eliminated in 8/10 subjects who aspirated. Subjects were 1.6 times more likely to have a clear airway given a citric acid solution than with water. Inspection of the graphed data suggests that subjects without dementia may be more likely to benefit from taste stimuli than those subjects with dementia (greater than one year post onset). However, statistical analysis could not be performed with dementia as an independent variable because of the low sample size. The temporal pattern of aspiration is shown in Table 4. Silent aspiration occurred in 9/32 swallows. All aspirations that occurred during the swallow elicited an immediate cough after white out, with one exception.

Table 4 Temporal pattern of aspirated swallows
Full size table
Figure 1
figure 1

Taste stimuli not effective in improving swallowing in subjects who aspirated or penetrated water. Swallow rating categories are no trial, clear airway (no material entered airway), penetration, and aspiration (*Dementia present).

Full size image
Figure 2
figure 2

Taste stimuli effective in improving swallowing in subjects who aspirated water. Swallow rating categories are no trial, clear airway (no material entered airway), penetration, and aspiration (*Dementia; **Dementia + CVA).

Full size image
Figure 3
figure 3

Sour taste effective in improving swallowing in subjects who aspirated water, but sweet–sour taste not as effective. Swallow rating categories are no trial, clear airway (no material entered airway), penetration, and aspiration (*Dementia present).

Full size image

Furthermore, subjects were 1.3 times more likely to have a clear airway given teaspoon delivery rather than cup drinking with all stimuli. The mean volume consumed by cup drinking was similar across all stimuli given 39 trials (Table 5). Eight subjects consumed 16 pairs of cup trials, with 4 subjects showing a directional change in volume (one pair increased, one decreased) during cup drinking. Six subjects increased the volume consumed on their second trial if they experienced no material entering the airway on the first trial (mean increase = 3.1 ml, SD = 2.8, range = 0.5–8.3), with all but one maintaining a clear airway (aspiration occurred in one subject). Two subjects increased their volume after aspirating the first cup trial, but their second trial rating improved (increased 0.60 and 6.4 ml with subsequent swallow ratings of penetration and clear airway, respectively).

Table 5 Cup volumes (ml) consumed by subjects
Full size table

Five subjects decreased their second trial volume if they experienced material entering the airway on their first cup trial (mean decrease = 5.9 ml, SD = 5.0, range = 1.4–11.7). One subject decreased the volume of his second cup trial by 5.2 ml even though he had a clear airway on the first trial; he maintained the clear airway on the second trial. Of the remaining four subjects, one improved from aspiration to clear airway, one worsened his performance from penetration to aspiration, and 2 subjects were unchanged (both trials penetration or aspiration).

The taste stimuli (acid and mixture samples) triggered a significant increase in the number of spontaneous swallows after the initial swallow compared with water. Citric acid (2.7%) elicited the greatest number of mean spontaneous swallows after its presentation, followed by the citric acid–sucrose mixture and water (Fig. 4). According to Bonferroni post hoc analysis, the number of spontaneous swallows after a water stimulus was lower than that after citric acid [t(88.3) = −8.48, p < 0.0001] and the citric acid–sucrose mixture [t(88.8) = −3.69, p < 0.0004]. The number of spontaneous swallows after citric acid was higher than that after the acid–sucrose mixture [t(88) = 5.00, p < 0.0001].

Figure 4
figure 4

Mean spontaneous dry swallows 60 s after stimuli (+1 SEM). (No stim = number of swallows after insertion of endoscope).

Full size image

The normative mean liquid STD value, given no swallow command, reported by Palmer et al. [17] is 1.61 s ± 0.55, range = 1.06–2.16 s. An operational definition of delayed swallow response was not reported by Logemann et al. [4]. Ninety percent (92/102) of the swallows were at or below the mean normative STD value (1.61 s) to initiate a liquid swallow (mean = 0.80 s, SD = 0.93 s, range = 0.0–5.50 s). Four of the ten delayed swallows occurred in subject 11. The acid solution shortened STD (Table 6).

Table 6 Stage transitions duration (STD) times and swallow ratings for subject 11
Full size table

Discussion

A high citric acid stimulus (2.7% w/v) significantly improved swallowing by reducing aspiration and penetration in nursing home residents with neurogenic oropharyngeal dysphagia. These results were observed in an elderly, medically compromised nursing home population. Significant improvements in swallowing are not expected in this population. Gustatory stimuli might facilitate swallowing in some patients with neurogenic oropharyngeal dysphagia.

Logemann et al. [4] suggested that sour taste enhanced or provided a different sensory input to the nervous system and, thus, facilitated oropharyngeal swallowing in neurogenic dysphagia. Models of human gustation and swallowing involve overlapping nervous system anatomy, with convergence in some sensory areas that may be critical for eliciting a swallow. Both require peripheral sensory information, with passage through the NTS in the brainstem to subcortical and cortical areas. Mechanosensitive receptors activate the greatest number of neurons in the NTS (via the trigeminal nerve), with chemical stimuli second-best [18,19]. Although cortical regions can modify aspects of a swallow and taste, the brainstem holds the essential interneurons responsible for swallow initiation [18] and taste relay [20].

To trigger a swallow response, it has been proposed that a stimulus must excite multiple peripheral sensory receptors in a dynamic fashion and activate sensory fibers in the NTS [18,21]. To trigger taste perception, a chemical stimulus must activate taste buds and stimulate sensory fibers in the NTS as well [20]. The shared peripheral anatomy and processing to the NTS suggest a possibility that taste may facilitate swallowing by providing additional sensory input to the NTS. The sensory characteristics of a strong acid taste, especially in liquid form, would appear to possess many of the necessary criteria suggested to evoke a pharyngeal swallow [22,23,24].

It is not known what mechanism is responsible for the improved swallowing behaviors observed or if several mechanisms are involved. Improved swallowing behaviors given 2.7% citric acid could be a result of the intense sour taste perception, bad taste (do we swallow pleasant versus unpleasant tastes differently?), and/or trigeminal stimulation (irritation) of the 2.7% citric acid sample. The term “trigeminal stimulation” is commonly used in the sensory science literature to refer to any stimuli that evoke a sensation that is not taste or smell. Although trigeminal receptors do not elicit a taste response, the two sensory systems clearly interact to provide total flavor impact [25].

Increased concentration levels of acid have been found to increase the effectiveness of eliciting a reflex swallow in rats [26] and may be important in humans as well. The lower acid concentration level in the palatable sweet–sour mixture did not significantly improve swallowing behaviors. The concentration levels used in the mixture were chosen because they are typical of palatable beverages, such as lemonade. Given the potential clinical application of this stimulus, we wanted to learn whether a realistic citric acid–sucrose mixture would elicit swallowing improvement. It is not known whether a 1.11% citric acid solution (unmixed) would elicit similar improvements as observed with 2.7% acid, or if a 2.7% acid–sucrose mixture would continue to demonstrate positive effects.

Given the intense sourness and trigeminal stimulation of 2.7% citric acid, additional research is necessary to determine what level of sweetener could be used to suppress the sourness to a palatable level, if possible. If VFSS is used, it would be preferable to use aspartame as the sweetener in the barium mixtures because, unlike sucrose, aspartame does not increase the viscosity or mouthfeel of beverages. This would aid in producing a product as similar as possible in sensory properties to a real beverage, given the necessary addition of barium.

The lack of significant swallowing improvement given the 1.11% citric acid–8% sucrose mixture may be due to the effects of mixture suppression, the lack of sufficient trigeminal stimulation at the lower acid level, or some unknown factor. For instance, it is not known whether hedonics (likes and dislikes) of a food influences swallowing physiology. There is evidence that pleasant tastes stimulate different cortical areas than aversive tastes [27], but the effect of hedonic taste on swallowing physiology has never been investigated.

Mixture Suppression and Swallowing

It is not known how peripheral and central taste perception, or specifically the effect of mixture suppression, may affect a swallow response. Mixture suppression may be a central (CNS) or peripheral (PNS) nervous system phenomenon, or a combination of both. If mixture suppression were purely a central event, the effects of the adapting stimulus would remain the same when it was presented in the same concentration, in an unmixed or mixed condition. However, this has not always been observed. Adaptation was lower when the adapting stimulus was presented in the mixed rather than the unmixed condition at the same concentrations, although the opposite has also been observed (equal efficiency of mixtures and single tastants as adaptors) [28,29,30]. Split-tongue experiments have also been used to investigate the mechanism(s) of mixture suppression, with similar conflicting results as adaptation. Since the tongue has ipsilateral neural connections from the mouth to the thalamus, split-tongue experiments allow inference regarding the locus of mixture suppression by comparing the intensity estimates of mixed and unmixed conditions (same concentration) to each side of the tongue. One study showed evidence of a CNS mechanism only while investigating quinine–sucrose (bitter–sweet) mixture suppression [11], but another suggested a combination of mechanisms (1/3 CNS, 2/3 PNS) for another set of mixtures (quinine HCl–NaCl, bitter–salty) [10]. Possible reasons for differences in the adaptor ability of a mixture versus single tastant may be the nature of the tastants themselves and presentation conditions. Lawless showed that similar adaptation results could be achieved using a mixture or single tastant by restricting the tongue area and reducing the flow rate of the stimulus [30].

Thus, more research is needed to examine the role mixture suppression may play in eliciting improved swallowing behaviors. If VFSS is used to examine the effect of taste stimuli on swallowing behaviors, research is needed to examine the pattern of mixture suppression in barium mixtures. Since taste is an independent variable in this research, it is important to understand how taste may change given the addition of barium. Currently, there are no data on barium mixture suppression patterns.

Stage Transition Duration (Delayed Pharyngeal Swallow)

Increased STD rarely occurred in this study, which is significantly different from the sample described by Logemann et al. [4]. Ninety percent of all swallows (92/102) were not delayed. Penetration and aspiration occurred in 39% of the swallows observed (40/102) because of apparent inadequate airway protection. Given the lack of a swallow command, it is not known whether poor oral control may also be a factor. What constitutes pharyngeal delay and significant deviation from normal swallowing is a subject of controversy in the literature [31]. Although the sample size was low, the absence of increased STD in a very elderly nursing home population was surprising. Perhaps increased STD is not as prevalent in this population as previously thought.

In the one case where increased STD was observed with water, a citric acid stimulus did elicit a quicker swallow, similar to the observation of Logemann et al. [4]. It eliminated aspiration. However, the citric acid–sucrose mixture did not evoke similar improvements in decreasing STD. Citric acid does not adapt as quickly as other taste qualities and can take up to 55 s to decline completely in intensity [32]. It is suggested the longer exposure time required for citric acid adaptation may be related to the cutaneous irritation caused by the acid. Thus, the added trigeminal stimulation provided by the highly sour (acidic) taste may have facilitated the improved swallowing behaviors observed.

Increased Frequency of Spontaneous Swallows after Taste Stimuli

Subjects showed significantly increased spontaneous dry swallows in response to taste stimuli. The reason(s) for this result are not known but may be due to a combination of continued irritation or taste from the acid and increased saliva production. Studies that have examined human gustatory afferent response to taste stimuli report rapid peripheral nerve response to a stimulus, reaching maximum intensity from baseline within 1 s [33]. However, it takes over 1 min to return to baseline [15]. In addition, it takes a long time for gustatory stimulation to be completely removed from the mouth, even after all of the bolus is swallowed. No data are available on how long the removal rate is under natural conditions, but with salt and lithium it can take up to 6 min to return to baseline after adaptation, expectorating every 15 s [34]. Thus, continued stimulation of the taste receptors may have provided additional sensory input after the initial swallow. Repeated sensory stimulation may be important in eliciting cortical facilitation of swallowing by creating a “reflex arc” altering the brainstem threshold for swallowing [18, p. 57].

Increased saliva production does not seem likely to be the cause for increased frequency of swallowing because the amount of saliva produced in 60 s given citric acid is quite low (mean = 0.634 ml/min) [35]. However, patients with hyposalivation may require increased time to complete a swallow, so any stimulus that increases salivation may improve the frequency of swallowing [36]. Furthermore, it is known that medications can cause xerostomia. Whole saliva flow rates can decrease from 1.6 ml/min (stimulated, no medications) to approximately 1 ml/min if four or more medications are taken by older adults [37]. Subjects in this study took an average of four medications per day. Overall, the continued irritation from the acid and increased saliva production may have provided the necessary sensory input to the NTS to lower the swallowing response threshold.

Teaspoon Versus Cup Drinking

Teaspoon delivery of the stimuli resulted in significantly less aspiration and penetration than cup drinking. The mean normal cup sip sizes of these dysphagic subjects were similar to those of nondysphagic individuals reported by others [38,39] and much smaller than the 21 ± 5-ml normal sip volumes reported earlier in the literature [40]. Although increased incidence of aspiration was noted with cup drinking, subjects appeared to spontaneously adjust their cup volumes according to their success in safely managing their first cup trial. This result may have clinical relevance. Clinicians are often worried about allowing patients to drink from a cup, fearing they will take too large a sip and aspirate. However, these results suggest patients naturally regulate their intake in an attempt to maintain swallowing safety. Indeed, in those cases where the first cup trial was aspirated or penetrated, it was often difficult to encourage the subjects to try another sip. Thus, if patients demonstrate safe swallowing using cup drinking, they will probably maintain safety if they are allowed to self-regulate their cup drinking. It would be important that drinking assistance provided by someone else did not interfere with the natural volume the patient wished to consume.

Application of PAS to FEES

Several studies have demonstrated that FEES can identify aspiration, penetration, spillage, and residue comparably to VFSS [41,42,43]. Pharyngeal swallow delay can also be assessed endoscopically [31]. Description of a swallowing impairment can occur using the penetration–aspiration scale (PAS) developed by Rosenbek et al. [12] for use with VFSS. PAS uses an 8-point multidimensional ordinal scale to rate swallowing behaviors. The scale has been shown to approximate interval qualities [44]. When applied to FEES, PAS has adequate interjudge and intrajudge reliability across all categories compared with VFSS (middle categories less reliable than others), with FEES more reliable at evaluating penetration but less reliable at assessing aspiration than VFSS [45]. If continuous extended observation of swallowing behavior is desired to examine the effect of aftertaste and saliva production after a taste sample, FEES is superior to VFSS since it does not need to limit the viewing time due to radiation exposure.

However, it was difficult to assign a PAS rating to several swallows when aspiration occurred during the swallow, with immediate cough after white out and no visual residue. This is probably due in part to the use of the home position only during the evaluation and the fact that white out obscures the view at the height of a swallow. Lowering the scope into the larynx to give a “close view” may have helped determine how low the bolus traveled prior to the onset of the swallow [16]. In addition, despite our efforts to increase visualization of the bolus path using white and green food coloring in the water-based stimuli, these coloring additions may not be sufficient to adequately mark where the material fell.

Limitations of the Research Design

Visual inspection of the data suggests low variability in the swallowing response across repetition of trials. Although two trials of water were not presented to each subject because of safety concerns, it is likely that repeated water trials would have elicited similar swallow ratings. This stable result was observed in the repeated water trials of two subjects. What is not known is whether a “warm up” of swallowing skills occurred during the water trials that facilitated the swallowing improvements noted with sour and sweet–sour stimuli. Although all of the residents were currently recommended to consume thickened liquids only, it is not known whether the swallowing improvements observed with these taste stimuli would also be observed with other tastes. Indeed, two residents were eliminated from the study when initial testing with water showed no difficulty managing thin liquids. A third resident was excluded during analysis when review of his entire FEES exam (which included drinking an entire glass of juice) revealed only one aspiration of water. It is possible that examination of other tastes with the residents may have revealed similar positive results, especially in those residents who showed no problem managing sour and sweet–sour liquids across all trials. If this is true, none of these residents should have been receiving thickened liquids and the positive taste results are unremarkable. These data confirm how difficult it is for speech-language pathologists to determine a swallowing problem from a clinical bedside dysphagia examination. It has been reported that incorrect diagnosis of probable aspiration can occur 40–50% of the time given bedside exams [46]. Thus, these data support the use of instrumental diagnostic procedures to assess swallowing skills.

Interrater reliability measures showed a strong linear relationship for the temporal measures and good reliability when counting swallows, even though their interrater agreement scores were lower. The interrater agreement scores may have been lower than expected due to several reasons. The lack of audio feedback and the presence of excessive secretions hindered visualization at times, preventing raters from knowing exactly when a bolus was presented and distinguishing cough. Laryngeal pumping and extraneous head/neck movement caused partial white out which may have affected assessment of the number of swallows that occurred after the swallow. Audio feedback and external monitors of a swallow (manual assessment, surface electromyography, or measurement of nasal apnea using a nasal cannula coupled to an analog-to-digital airflow sensor thermistor transducer) may improve interrater agreement scores in future studies.

FEES has been shown to assess penetration better than aspiration compared with VFSS [45]. In this study, aspiration was more frequent than penetration. However, FEES was able to provide clinically relevant information regarding the effect of taste on swallowing. Every clinician recognizes the relevance of aspiration and penetration reduction, even though certain swallowing measures observed by VFSS were not obtained [12].

Visual inspection of the graphed individual responses to stimuli suggests that the presence of dementia may hinder the beneficial effect of taste stimuli in neurologically impaired individuals. The small sample size did not allow statistical analysis of this hypothesis. Future research may want to explore this possible trend since there is strong evidence olfaction begins to decline early in Alzheimer’s Dementia (AD) but taste perception remains intact, at least initially, in the disease [47,48 ,49]. It is not known whether demented individuals perceive taste normally as their disease progresses.