Dysphagia 13:1–9 (1998)

© Springer-Verlag New York Inc. 1998

Comparing Treatment Intensities of Tactile-Thermal Application*

John C. Rosenbek, PhD,1 JoAnne Robbins, PhD,1 William O. Willford, PhD,2 Gail Kirk, MS,2 Amy Schiltz, MS,3
Thomas W. Sowell, MS,4 Steven E. Deutsch, PhD,5 Franklin J. Milanti, PhD,6 John Ashford, MS,7
Gary D. Gramigna,8 Ann Fogarty, MS,8 Katherine Dong, PhD,9 Marie T. Rau, PhD,10 Thomas E. Prescott, PhD,11
Anna M. Lloyd, MS,12 Marie T. Sterkel, MS,13 and Julie E. Hansen, MS,14
1
William S. Middleton Memorial Veterans Hospital, Departments of Neurology and Medicine, University of Wisconsin School of Medicine,
Madison, Wisconsin; 2Cooperative Studies Program Coordinating Center, VA Medical Center, Perry Point, Maryland; 3Aurora Regional Medical
Center, Aurora, Colorado; 4VA Medical Center, Little Rock, Arkansas; 5VA Medical Center, Long Beach, California; 6Hines VA Hospital,
Hines, Illinois; 7VA Medical Center, Nashville, Tennessee; 8VA Medical Center, Brockton/West Roxbury, Massachusetts; 9VA Medical Center,
North Chicago, Illinois; 10VA Medical Center, Portland, Oregon; 11VA Medical Center, Denver, Colorado; 12Richard L. Roudebush Veterans
Medical Center, Indianapolis, Indiana; 13VA Medical Center, Reno, Nevada; and 14VA Puget Sound Health Care System,
Seattle, Washington, USA

Abstract. The purpose of this study was to investigate
the relationships of four intensities of tactile-thermal ap-
plication (TTA) to changes in duration of stage transition
(DST) and performance on a newly designed scale of
penetration and aspiration by groups of patients made
dysphagic by stroke. Patients were randomly assigned to
receive 150, 300, 450, or 600 trials of TTA during each
of 2 weeks. Data on the time required to provide such
treatment, the actual number of trials clinicians were able
to provide, and on the influence of the four intensities are
provided. No single intensity emerged as the most thera-
peutic. It is suggested that subsequent studies with larger
groups include intensities between 300 and 550.
Key words: Dysphagia treatment — Thermal stimula-
tion — Stroke rehabilitation — Dysphagia — Degluti-
tion — Deglutition disorders.
Recovery from dysphagia secondary to stroke is not
guaranteed. As a result, a variety of treatments have been
developed for improving the dysphagic stroke patient’s
ability to swallow safely, and perhaps even sufficiently
*This work was performed at the 13 VA Medical Centers whose ad-
dresses appear above.
This is publication number 96-26 of the Madison Geriatrics Research
Education and Clinical Center
Correspondence to: John C. Rosenbek, Ph.D., Chief, Audiology and
Speech Pathology, Wm. S. Middleton Memorial Veterans Hospital,
2500 Overlook Terrace, Madison, WI 53705, USA
to allow a return to completely normal eating and drink-
ing. Of greatest concern to physicians referring patients
for these treatments—to the clinicians providing them, to
dysphagic patients, and to those who pay the bills—is the
efficacy of such treatments. One of the most extensively
studied was originally called thermal stimulation [1]. In
its first form, this method required the clinician to stroke
or rub the anterior faucial pillars with a cold probe prior
to having the patient swallow. It is hypothesized that the
touch and cold stimulation increases ‘‘oral awareness’’
and provide ‘‘an alerting sensory stimulus to the cortex
and brainstem such that, when the patient initiates the
oral stage of swallow, the pharyngeal swallow will trig-
ger more rapidly’’ [2].
A decade of research has fueled cautious opti-
mism about the method’s treatment potential. The earli-
est data were supplied by Lazzara et al. [3] based on their
single-session treatment of 25 neurologically impaired
dysphagic persons, including some who had suffered
from one or more strokes. The experimental protocol
comprised three liquid and three semisolid swallows,
with thermal stimulation being provided before the third
swallow of each consistency. They summarize the com-
parison of selected duration measures during the two
untreated and one treated swallow by observing that
stimulation ‘‘improved triggering of the swallowing re-
flex in 23 of these 25 neurologically impaired patients on
swallows of at least one consistency.’’ Results from a
different protocol with a limited number of these same
patients suggested that the effects of a single sensitiza-
tion lasted for two to three swallows.
2 J.C. Rosenbek et al.: Comparing Treatment Intensities
Rosenbek et al. [4] studied swallowing variability
and short-term effects of what they called ‘‘thermal ap-
plication’’ by comparing two durational measures for 10
untreated, semisolid swallows with 10 treated ones in a
crossover design. Twenty-two dysphagic stroke patients
served as subjects. Half began with the 10 untreated
swallows, and half completed the treated swallows first.
Treatment involved rubbing each anterior faucial pillar
three times with a chilled 00 laryngeal mirror, after
which subjects were instructed to swallow. Twenty swal-
lows provided the opportunity to establish the variability
of dysphagic swallowing and to establish the short-term
effects of the treatment.
Two findings emerged: (1) duration of stage tran-
sition (DST) and total swallow duration (TSD) [5] were
highly variable within and across subjects and had dis-
tributions that were abnormal with nonhomogeneous
variances; (2) thermal application reduced DST and
TSD. This reduction was interpreted as a therapeutic ef-
fect. The study was not designed to discover if the treat-
ment effects persisted. Short-term effects are less appeal-
ing than long-term ones, however. Therefore, a natural
development was to research more enduring effects.
Selinger et al. [6] reported the effects of 9 days of
thermal stimulation on a patient whose dysphagia was
the result of a brainstem stroke. Two modified barium
swallow evaluations before treatment confirmed severe,
stable dysphagia. After 26 sessions of thermal stimula-
tion during 9 days, repeat testing revealed no reduction in
aspiration. Swallow durations, the usual measure of im-
proved triggering of the swallow, were not measured nor
were duration measures used as criteria for enrolling this
patient in treatment. It may be, as a result, that this pa-
tient was not an ideal candidate for the method. Dyspha-
gia resulting from a brainstem lesion is usually charac-
terized by limitation in the duration and extent of upper
esophageal sphincter (UES) opening, a condition for
which thermal stimulation is not designed.
Rosenbek et al. [7] completed a treatment trial
with 7 dysphagic patients whose signs included abnor-
mally long DSTs. These subjects, whose dysphagia re-
sulted from multiple ischemic lesions, underwent a 1-
month trial of thermal stimulation (called ‘‘thermal ap-
plication’’ in their study) using a single-subject
withdrawal or ABAB design. After baseline testing, 6 of
the subjects were randomly assigned to begin the study
with a week of thermal application (B stage), and 1 sub-
ject was randomly assigned to begin with a week of no
treatment (A stage). Week-long A and B stages were
then alternated until each subject had completed the 4-
week study. Progress was tested after each stage and at
follow-up 1 month later. Three judges determined the
influence of thermal application on eight duration and
four descriptive measures by visual inspection of data
displays. Two of the judges agreed that 2 subjects dem-
onstrated decreased DST secondary to treatment, without
changes in the occurrence of aspiration or penetration.
The authors concluded that these findings offered weak
support for the method’s efficacy, and called for clinical
trials, as has every other researcher quoted.
Designing a clinical trial to investigate a meth-
od’s efficacy depends upon the availability of answers to
several questions. One of the most critical is about treat-
ment intensity. How much treatment of a particular sort
is to be compared with the no-treatment condition? The
appropriate intensity of tactile-thermal application has
not been established. Useful data in planning a study of
treatment intensity can be extracted from the study by
Rosenbek et al. [7]. The number of trials received varied
from patient to patient. A trial was defined as consisting
of rubbing both anterior faucial pillars three to five times
with a cold probe followed by instructions for the patient
to swallow. Two of those patients had reduction in DST,
and all 7 had at least one change in swallowing physi-
ology judged to be secondary to treatment. One of the 2
patients with decreases in DST had 300 trials per week,
the other had 400. It was therefore decided to design a
study to establish a response curve for TTA using four
treatment intensities: 150, 300, 450, and 600 trials per
week for 2 weeks using hospitalized, dysphagic stroke
patients. This range bracketed those intensities that
promised efficacy.
Materials and Methods
Experimental Design
To establish a response curve for TTA, a multicenter, randomized
group design was employed. A total of 45 male dysphagic, stroke
patients from 12 VA medical centers met enrollment criteria and were
randomly assigned to one of four groups. Patients in Group I received
150 trials per week, Group II patients received 300 trials per week,
Group III patients received 450 trials per week, and Group IV patients
received 600 trials per week. All treatments were provided by Speech-
Language Pathologists. Standardized videofluoroscopic swallowing
examinations using five 3-ml and five 10-ml thin liquid boluses were
completed at intake, 1 week, and 2 weeks. Outcome measures were
DST and penetration-aspiration, measured on an 8-point scale [8].
Baseline data were available on 45 patients; 1 week and 2 week out-
come data were available on 45 and 43 patients, respectively. The
primary statistical analysis was change scores between baseline and 1
week and 2 weeks for each treatment group, and were statistically
analyzed by paired t-tests. Treatment group differences were analyzed
by analysis of (baseline adjusted) covariance, separately at 1 week and
2 weeks. Study duration was 1 year.
Subjects
Male dysphagic, stroke patients (R 4 45: 38 white, 7 African-
American) for whom baseline data were received from 12 VA medical
J.C. Rosenbek et al.: Comparing Treatment Intensities
centers were randomly assigned to one of four groups. Average age was
65.2 years. All subjects met the following criteria: (1) presence of
stroke-caused dysphagia verified by medical chart review and bedside
swallowing examination, (2) duration of dysphagia between 1 and 12
weeks, (3) medical stability as judged by a referring neurologist, (4)
videofluoroscopic evidence of dysphagia characterized by abnormally
long DST, defined statistically [5] on at least 1 swallow in 10 (or on 2
of 10 swallows if DST was prolonged on the first), (5) videofluoro-
scopic evidence of dysphagia characterized by abnormal penetration-
aspiration as documented by a score of 3 or greater on the penetration-
aspiration scale [8] on at least 1 of 10 swallows (or on 2 if penetration-
aspiration performance was abnormal on the first swallow), and (6)
ability to cooperate with the treatment procedure as revealed by a short
period of trial therapy. Patients were excluded if they were (1) trache-
otomized, (2) pregnant or planning pregnancy, (3) suffering from dys-
phagia from a cause other than stroke, as determined by chart review
and consultation with referring neurologists, (4) treated with any ver-
sion of the experimental therapy within 4 weeks of enrollment in the
present study, (5) participating in any other stroke study that might
influence response to the experimental treatment, and (6) incapable of
or unwilling to give informed consent.
A speech-language pathologist with special expertise in dys-
phagia at each cooperating hospital was responsible for (1) completing
a bedside swallowing examination, (2) confirming subject appropriate-
ness on all criteria requiring chart review and medical consultation, (3)
completing three standardized videofluoroscopic swallowing examina-
tions, (4) shipping those examinations to the study coordinator’s office
for evaluation, (5) securing each subject’s informed consent once can-
didacy was established, and (6) providing standardized treatment.
Videofluoroscopic Swallowing Examinations
Standardized videofluoroscopic swallowing examinations were com-
pleted at intake, 1, and 2 weeks. Studies were completed in each hos-
pital’s videofluoroscopic suite using standard radiographic fluoro-
scopic systems with remote monitor and videofluoroscopic capabilities.
Images were stored on either 3/4-inch, U-matic, or 1/2-inch S-VHS or
simple VHS recorders. A video counter, if available, imprinted a time
code (accurate to 0.01 sec) on the tape. If a timer was unavailable, a
time code was dubbed onto the tape at the study coordinator’s office.
Subjects were seated in the lateral plane and given standardized in-
structions. The fluoroscopic camera was focussed on the lips anteriorly,
the spinal column posteriorly, the hard palate superiorly, and just below
the upper esophageal sphincter inferiorly.
Testing consisted of five, 3-ml thin liquid boluses followed by
5, 10-ml thin liquid boluses of standardized viscosity. To preserve
patient safety, testing was discontinued if a subject aspirated more than
a small amount of two consecutive boluses without successfully clear-
ing them from the airway. If two such aspirations occurred on the
smaller boluses, it was left to the tester’s judgment to move to the larger
boluses. If the larger boluses were tested, that testing stopped after the
occurrence of one aspiration not expelled from the airway. Once com-
pleted, each examination was mailed overnight to the study coordina-
tor’s hospital for independent evaluation.
Random Assignment
Once a patient’s eligibility was established, randomization to one of the
four treatment groups was accomplished centrally by a telephone call to
the Perry Point VAMC Cooperative Studies Program Coordinating
Center (CSPCC). Randomization was stratified by center in order to
obtain treatment group balance.
3
Evaluation of Videofluoroscopic
Swallowing Examinations
All videofluoroscopic swallowing examinations were analyzed by a
specially trained speech-language pathologist who did not know the
group assignment of any subject. Training in reliable measurement of
all outcome measures was completed according to a standard protocol
in the VA-UW Swallowing Research Laboratory. Two measurements
were made of all swallows. The first was DST which was selected
because it is sensitive to the experimental treatment [4]. It is defined as
the duration from the time the head of the bolus reaches the posterior
border of the mandibular ramus until the beginning of maximum el-
evation of the hyoid bone. Penetration which occurs when material
enters the larynx but fails to pass below the vocal folds and aspiration
which occurs when material passes below the vocal folds were mea-
sured with the 8-point penetration-aspiration (PA) scale [8]. Penetration
and aspiration were chosen for measurement because they are associ-
ated clinically with a more severe swallowing impairment than are any
other traditional signs of dysphagia and because changes in penetration
and aspiration are especially prized as treatment goals.
Each swallow was viewed at normal, slowed, and frame-by-
frame speeds as many times as necessary for confident judgment. All
data were entered on standardized data sheets and forwarded to the
Cooperative Studies Program Coordinating Center for statistical analy-
sis.
Treatment
The tactile-thermal application procedures were standardized and con-
ducted by a cooperating speech-language pathologist in each hospital.
Treatment was applied with an ice stick made the way a popsicle is
made by freezing a cylinder of water into which a handle has been
inserted. This method rather than the traditional one of using chilled
laryngeal mirrors was chosen for two reasons: it guarantees that the
stimulator is cold when it is applied [9] and the ice stick begins to melt
during application thereby adding fluid to the stimulation and to the
patient’s saliva for swallowing.
One trial of TTA consists of two actions: (1) the clinician rub-
bing first one and then the other anterior faucial pillar with the ice stick
three or more times each and (2) the clinician urging the patient to
swallow ‘‘hard.’’ The rubbing was accomplished firmly but not so
firmly as to cause discomfort. The rubbing extended from as low on the
faucial pillar as it was possible to reach to as high as it was possible to
reach and no effort was made to avoid the tongue. The bolus to be
swallowed consisted of saliva and whatever melting water was avail-
able from the rubbing. Subjects were asked periodically if they could
feel the cold and if it was uncomfortable. They were also instructed to
announce discomfort as soon as it occurred.
The appropriate number of trials per week—150, 300, 450,
600—was to be distributed across 3 to 5 work days, as the clinician
chose. No treatment on day 5, a test day, could occur within 2 h of
progress testing. Trials were done in groups throughout the day de-
pending on the clinician’s and the patient’s wishes.
Results
Study Implementation
The first data to be presented characterize the implemen-
tation of TTA. The group to which a subject was ran-
domized determined the number of TTA trials to be re-
4 J.C. Rosenbek et al.: Comparing Treatment Intensities

Table 1. Number of trials and time (minutes)a spent by speech pathologist in treating patients (mean, standard error)

Week 1
Treatment group (no. of trials)

150 300 450 600

ANOVA
n 12 10 10 13 and value

Number of trials
achieved 150.0 0.0 300.0 0.0 431.3 18.7 542.5 24.8 <0.001
Preparation time 50.8 19.4 46.2 7.1 83.6 21.0 83.1 19.4 0.32
Patient contact time 103.7 9.8 183.3 26.8 226.8 19.1 323.8 42.6 <0.001
Time after treatment 21.8 4.1 27.9 6.2 40.6 12.6 46.8 11.2 0.20

Week 2
Treatment group (no. of trials)

150 300 450 600

ANOVA
n 12 10 10 11 and value

Number of trials
achieved 150.0 0.0 300.0 0.0 441.0 9.0 540.5 25.2 <0.001
Preparation time 26.3 4.9 37.9 5.6 51.1 7.6 59.5 11.6 0.02
Patient contact time 90.1 7.8 174.4 17.9 234.7 20.4 316.4 41.4 <0.01
Time after treatment 19.7 3.3 36.2 10.9 30.5 7.9 32.3 8.0 0.45

ANOVA 4 analysis of variance.

ceived each week. Table 1 shows the mean number of
trials and the standard error of the mean for all groups for
both weeks 1 and 2. The total number of TTA trials per
week was easily obtained for the 150 and 300 trial treat-
ment groups for each week of the trial. However, the
450, and particularly the 600, trial treatment groups did
not receive the prescribed number of trials, reaching an
average of 431.3 and 542.5 strokes, respectively at 1
week and 441.0 and 540.5 strokes at 2 weeks. Table 1
also shows mean and SE data for the preparation (setup
and patient preparation), patient contact (to provide the
treatment), and after treatment (clean up and charting)
times in min for each group for both weeks. Preparation
time and time after treatment at 1 week were roughly
comparable for the 150 and 300 trial groups as well as for
the 450 and 600 trial treatment groups. Patient contact
and after treatment time at 1 week were directly propor-
tional to the number of trials received. Preparation and
patient contact time were directly proportional to number
of trials during week 2 as was after treatment time, ex-
cept for the 300 trial group which inexplicably required
a disproportionate amount of time after treatment. There
was some experience or learning advantage (decrease in
time spent) at 2 weeks compared with 1 week, particu-
larly in preparation time.
DST and PA performance were to be available on
5, 3-ml and 5, 10-ml boluses for each of the three vid-
eofluoroscopic swallowing examinations. The mean
number of swallows and SE of the mean for each bolus
size for each measurement for each of the four groups are
shown in Table 2. The number of swallows available for
DST and PA measurements at the end of each week were
not much different. The mean number of swallows on
which PA measurements could be made, for the most
part, was slightly closer to the maximum of five than was
the mean on which the DST measurements could be
made. In general, fewer trials of 10-ml than of 3-ml
boluses were available for analysis.
This study also yielded information about appro-
priate candidacy for TTA. Recall that subjects to be en-
rolled had to have a delay in DST equal to 2 SDs beyond
the normal mean, as established by Robbins et al [5] on
at least one swallow or two if the delay occurred on the
first of the 10 baseline swallows. They also had to have
at least one penetration or aspiration unless that too oc-
curred on the first swallow of the 10, in which case it had
to occur on at least one more swallow. As is obvious
from the data in Fig. 1, many mildly dysphagic subjects,
as defined by duration and PA performance, were en-
rolled in this study. Statistically significant changes in
performance with treatment are harder to accomplish
with mildly dysphagic patients, depending on their vari-
ability and sample size. It may be that mild subjects
deserve treatment, however, subsequent clinical trials
need to guarantee that a wider range of severities be
represented in the sample. This could be accomplished
by specifying more rigid criteria or by recruiting patients
across the severity range. Requiring that potential sub-
J.C. Rosenbek et al.: Comparing Treatment Intensities 5

Table 2. Number of trials measured for mean outcome (mean, standard error)

Treatment group (no. of trials)

ANOVA
150 300 450 600 analysis of
n variance
(weeks) Duration of stage transition (3 ml liquid) p-value

1 4.83 0.11 4.30 0.40 4.90 0.10 4.31 0.21 0.16

n 4 12 n 4 10 n 4 10 n 4 13
2 4.75 0.18 4.10 0.38 4.90 0.10 4.82 0.18 0.10
n 4 12 n 4 10 n 4 10 n 4 11
Duration of stage transition (10 ml liquid)

1 4.75 0.18 4.00 0.67 4.40 0.40 4.00 0.51 0.60

n 4 12 n 4 10 n 4 10 n 4 13
2 4.67 0.26 3.90 0.66 4.80 0.13 4.36 0.45 0.48
n 4 12 n 4 10 n 4 10 n 4 11
Penetration/aspiration (3 ml liquid)

1 5.00 0.00 4.20 0.39 5.00 0.00 4.39 0.21 0.04

n 4 12 n 4 10 n 4 10 n 4 13
2 5.00 0.00 4.10 0.38 5.00 0.00 4.82 0.18 0.03
n 4 12 n 4 10 n 4 10 n 4 11
Penetration/aspiration (10 ml liquid)

1 4.67 0.33 4.00 0.67 4.50 0.40 4.08 0.51 0.72

n 4 12 n 4 10 n 4 10 n 4 13
2 4.75 0.25 3.80 0.66 4.80 0.13 4.46 0.46 0.36
n 4 12 n 4 10 n 4 10 n 4 11

jects have at least two instances of penetration or aspi-
ration and an abnormally long DST on the same two or
more swallows has several advantages. The method was
designed for such patients [1], who would be more se-
verely dysphagic than some in the present study. Such
patients would more likely be suffering from functional
consequences of their dysphagia, and the momentum in
modern healthcare favors those treatments for conditions
with functional consequences.
Results
Study Outcome Measurements
The goal of this pilot study was to determine if there was
a frequency of TTA trials-response relationship analo-
gous to a drug dose-response relationship. Researchers
agreed that (1) a decrease of 0.35 secs (4.4% of the usual
range of delay of 0.1–8.0 sec) in DST, and (2) a decrease
of 1.5 units (21.4% of the range 1–8) in PA, at 2 weeks
from randomization, would be sufficient to establish
clinical significance between progress and baseline.
These values were based on clinical judgment rather than
on data. Table 3 gives summary change data for the two
measures, two bolus sizes, and four groups at weeks 1
and 2 (see Fig. 1 for the data graphically).
The first conclusion to be drawn from these data
is that no single treatment intensity emerged as superior.
Second, all intensities of treatment, except 150 at both
weeks 1 and 2 and 600 at week 2 appear to have resulted
in improvement represented by positive mean changes
greater than 0.35 sec in DST on 3-ml boluses; and for all
treatment groups combined, mean change averaged 1.14
and 1.17 sec at 1 and 2 weeks, respectively. This effect,
however, is not present for 10-ml boluses partly because
of performance by the 300 trial group. The small sample
sizes made the achievement of statistical significance (p
< 0.05) virtually impossible, however, the p values for
the overall combined groups tended toward significance:
p 4 0.06 at 2 weeks for the 3-ml boluses. Overall, the
DST results for the 3-ml liquid boluses favor the 300 trial
group. The DST results for the 10-ml liquid boluses seem
to slightly favor the 600 trial group. It is to be recalled
that members of this latter group got an average of
slightly fewer than 550 trials per week.
Alternatively (see Table 3 and Fig. 1) positive
mean changes on the PA measures for the 3-ml liquid
boluses averaged only 0.55 and 0.59, at 1 and 2 weeks,
respectively and 0.51 and 0.32 for the 10-ml boluses.
These values are less than one-half of the 1.5 unit change
assumed to represent a clinically significant change.
However, for the combined group analysis these changes
on the 3-ml boluses provided statistical significance, that
is, p 4 0.04 at 1 week and p 4 0.03 at 2 weeks. Changes
on the 10-ml boluses were not significant. Finally, there
6 J.C. Rosenbek et al.: Comparing Treatment Intensities
Fig. 1. Effects of four treatment intensities on DST for 3-ml boluses (A), duration of stage transition for 10-ml boluses (B), penetration-aspiration
for 3-ml boluses (C), and penetration-aspiration for 10-ml boluses (D).
was no hospital effect in that the patients from one hos-
pital did not improve more or less than patients from any
other hospital.
Discussion
This was not an efficacy study: no comparison of treated
and untreated groups was planned or completed. Instead,
the goal was to gather data on four intensities of treat-
ment with the long-range purpose of using the data to
plan a clinical trial in which treatment and no-treatment
groups are compared. Data useful to that planning have
emerged despite the need for more research with larger
number of subjects.
Logemann (personal communication) recom-
mended four or five, 10–15 min sessions per day, a range
of 200–375 min per week. Taking the time spent, data
from Table 1 compared with Logemann’s prescription is
interesting. The average time spent in preparation, deliv-
ery, and follow-up of treatment to provide an average of
542.5 trials during week 1 for the patients in the 600 trial
group is 453.7 mins. Total time spent providing an av-
erage of 540.5 trials during week 2 was 408.2 mins.
These values are beyond the range Logemann specifies
but not extravagently so, and her prescription does not
include preparation and after-treatment time. An average
of 351.0 min were spent providing an average of 431
trials to the 450 trial per week group during week 1 and
316.3 min were spent during week 2 in providing an
average of 441 trials to that same group. These values are
within Logemann’s range.
Bolus size during testing is another important
clinical issue, with clinical judgment favoring the use of
bolus sizes that resemble those that patients actually take
when they are eating, unless the method is being used
with severely impaired patients who are not yet eating.
The trend, therefore, is toward a range of bolus sizes,
including 3- and 10-ml, as were used in the present
study. In general, fewer 10- than 3-ml data points were
available for analysis. The most frequent reason for miss-
ing data was the need to stop testing to preserve patient
J.C. Rosenbek et al.: Comparing Treatment Intensities 7

Table 3. Videofluoroscopic mean change scores from baseline at 1 and 2 weeks

Treatment group (no. of trials)

150 300 450 600 All treatment

groups
Duration of stage transition in seconds (3 ml liquid) combined

Baseline n 12 10 10 13 45
Mean 1.05 5.31 2.85 1.13 2.42
SE 0.22 3.88 1.65 0.36 0.94
Week 1 n 12 10 10 13 45
Mean change −0.29 3.46 1.33 0.54 1.14
SE 0.42 2.93 1.00 0.31 0.71
p-value* 0.51 0.27 0.22 0.11 0.11
Week 2 n 12 10 10 11 43
Mean change −0.04 2.90 1.93 0.21 1.17
SE 0.27 2.22 1.24 0.18 0.60
p-value* 0.88 0.22 0.15 0.27 0.06
Duration of stage transition in seconds (10 ml liquid)

Baseline n 11 8 10 12 41
Mean 0.58 1.48 0.77 1.28 1.01
SE 0.18 0.91 0.22 0.47 0.23
Week 1 n 11 8 10 11 40
Mean change 0.05 −0.04 0.15 0.71 0.23
SE 0.19 0.24 0.18 0.42 0.14
p-value* 0.81 0.86 0.43 0.13 0.11
Week 2 n 11 8 10 10 39
Mean change 0.01 −2.10 0.11 0.53 −0.23
SE 0.16 2.27 0.19 0.29 0.44
p-value* 0.94 0.39 0.57 0.10 0.60
Penetration/aspiration score (3 ml liquid)

Baseline n 12 9 10 13 44
Mean 3.41 3.54 2.84 2.87 3.15
SE 0.47 0.82 0.55 0.51 0.28
Week 1 n 12 9 10 13 44
Mean change 0.88 0.50 0.26 0.49 0.55
SE 0.36 0.66 0.35 0.65 0.26
p-value* 0.04 0.47 0.49 0.46 0.04
Week 2 n 12 9 10 11 42
Mean change 1.11 0.16 −0.09 0.97 0.59
SE 0.30 0.60 0.36 0.72 0.26
p-value* 0.003 0.79 0.82 0.21 0.03
Penetration/aspiration score (10 ml liquid)

Baseline n 11 7 10 12 40
Mean 4.66 3.43 3.49 3.70 3.86
SE 0.61 0.19 0.50 0.48 0.26
Week 1 n 11 7 10 11 39
Mean change 0.42 0.40 0.35 0.83 0.51
SE 0.73 1.04 0.23 0.69 0.33
p-value* 0.58 0.72 0.18 0.26 0.13
Week 2 n 11 7 10 10 38
Mean change 0.72 −0.19 0.18 0.31 0.32
SE 0.90 0.92 0.27 0.46 0.33
p-value* 0.44 0.85 0.52 0.52 0.35

SE 4 standard error.
*p-value from paired t-test.
8 J.C. Rosenbek et al.: Comparing Treatment Intensities
safety. This situation creates an interesting paradox for
the clinician researcher. On the one hand, it is increas-
ingly important that penetration and aspiration events be
changed by treatment and in general the risk of such
events is higher with bigger boluses, although there are
exceptions [10]. On the other hand, complete data sets
can be jeopardized by larger test boluses, as they were in
this study. The best solution may be large sample sizes if
large boluses are to be tested.
Another finding of the present study related to
bolus size is the difference in DST values for the 3- and
10-ml boluses. On average, delays were greater for the
smaller boluses. Also TTA seemed to have a greater
influence on smaller boluses. None of these differences
were treated statistically, however. Their potential sig-
nificance serves as a warning about the complexity of
designing efficacy research. It may well be that aspira-
tion, for example, has different kinematic explanations
for different bolus volumes even in the same patient.
No single intensity emerged from these data as
superior. One problem was with the sample sizes. Eight
to 13 patients were clearly not enough and subsequent
trials will need to aim for groups as large as 85–125
subjects. These sample size calculations are based on
expectations of change in DST of 0.35 sec and of 1.5 on
the PA scale. Those values were not arbitrary but neither
were they data based, and a different group of research-
ers might have offered different values. All groups ex-
cept the 150 trials group and the 600 trials group at week
2 had changes in DST of 0.35 sec or greater and overall
changes of greater than 1 sec were achieved. A more
rigid criterion for clinical significance of changes in DST
is probably realistic. Studies to establish the functional
significance of varying amounts of change in DST by
comparing them with patient-based measures of per-
ceived improvement, perceived comfort or safety of eat-
ing, and time to eat are necessary. The panel of experts
also required a change of 1.5 points in performance on
the eight-point PA scale [8] for clinical significance. No
single group reached this level of change nor did change
overall reach this level. It may be that this requirement is
unrealistically high regardless of the present study’s out-
come. A change of one scale point may be sufficient. A
single point of change on the scale is evidence that either
penetration is not as deep, that the patient has reacted to
either the penetration or aspiration with an attempt to
expel the material (the difference between scores of 7
and 8), or that the patient has been able to move the
material higher into the airway or pharynx once penetra-
tion or aspiration has occurred. Though all are positive
signs and ones that clinicians would probably value, the
relationship of performance as measured by the rela-
tively new PA scale and other swallowing parameters
such as DST and aspects of quality of life requires further
attention for meaningful implementation and interpreta-
tion.
Subjects were enrolled in this study as early as 1
week after stroke, a time that is well within the period of
physiological recovery. The threat to the present design
created by the probability of such recovery was recog-
nized prior to the study’s initiation. Subjects close to
onset of dysphagia were enrolled because acute patients
are standardly treated in the typical clinical practice, and
clinicians want to know about treatment effects for this
group as well as for those with chronic dysphagia. It was
anticipated that all groups would improve and that the
group receiving the most efficacious treatment would
improve the most. That did not occur in any trenchant
way, unfortunately. However, the tendency toward sig-
nificant treatment effects overall is consistent with the
notion that treatment can boost effects occurring sponta-
neously.
One final caveat about these subjects warrants
comment because it can be argued that subject heteroge-
neity rather than, or in addition to, sample size may have
influenced this study’s findings. The dysphagic subjects
shared the features of delayed initiation of the pharyn-
geal stage of swallowing and aspiration. Like the over-
whelming majority of such stroke patients, they had
other signs as well, including pooling in the pharyngeal
recesses. Our goal in subject selection was to pick pa-
tients who seemed clinically to be aspirating because of
a delay. We included no subjects who aspirated only
after the swallow was complete and we included no sub-
jects who had a focal recurrent nerve palsy. It is likely
that the relationship between delay and aspiration is not
linear and that several variables contribute to aspiration
before and during the swallow. A future goal will be to
select subjects who aspirate only on delayed swallows.
Further research with careful attention to such
subject selection criteria is necessary to establish the ap-
propriate dose of TTA before designing an efficacy trial.
The present findings offer some guidance to the design.
Six hundred trials are difficult to achieve and the data
offer no strong evidence that so many trials are neces-
sary. On the other hand, the overall effect of 150 trials is
not striking. Study of intensities between 300 and 550—
the approximate average number of trials the 600 trial
group actually received—would appear appropriate. The
sample size necessary for confident conclusions is a for-
midable but not impossible challenge for centers work-
ing together. The potential benefits for dysphagia treat-
ment are worth the effort.
Acknowledgment. This study was supported by Rehabilitation Research
and Development Grant # C93-689AP.
J.C. Rosenbek et al.: Comparing Treatment Intensities
References
1. Logemann J: Evaluation and Treatment of Swallowing Disor-
ders. College-Hill Press, San Diego; 1983
2. Logemann JA: The dysphagia diagnostic procedure as a treat-
ment efficacy trial. Clin Commun Disord 3:1–10, 1993
3. Lazzara G, Lazarus C, Logemann JA: Impact of thermal stimu-
lation on the triggering of the swallowing reflex. Dysphagia
1:73–77, 1986
4. Rosenbek JC, Roecker EB, Wood JL, Robbins J: Thermal ap-
plication reduces the duration of stage transition after stroke.
Dysphagia 11:225–233, 1996
5. Robbins JA, Hamilton JW, Lof GL, Kempster GB: Oropharyn-
9
geal swallowing in normal adults of different ages. Gastroen-
terology 103:823–829, 1992
6. Selinger M, Prescott TE, McKinley R: The efficacy of thermal
stimulation: a case study. Rocky Mountain J Commun Disord
8:21–23, 1990
7. Rosenbek JC, Robbins J, Fishback B, Levine RL: The effects of
thermal application on dysphagia after stroke. J Speech Hear
Res 34:1257–1268, 1991
8. Rosenbek JC, Robbins JA, Roecker EB, Coyle JL, Wood JL: A
penetration-aspiration scale. Dysphagia 11:93–98, 1996
9. Selinger M, Prescott TE, Hoffman I: Temperature acceleration
in cold oral stimulation. Dysphagia 9:83–87, 1994
10. Robbins J, Sufit R, Rosenbek J, Levine R, Hyland J: A modi-
fication of the modified barium swallow. Dysphagia 2:83–86, 1987