Professional Documents
Culture Documents
The Test of Masticating and Swallowing Solids (TOMASS) : Reliability, Validity and International Normative Data
The Test of Masticating and Swallowing Solids (TOMASS) : Reliability, Validity and International Normative Data
Research Report
The Test of Masticating and Swallowing Solids (TOMASS): reliability,
validity and international normative data
Maggie-Lee Huckabee†‡, Theresa McIntosh†‡, Laura Fuller†‡, Morgan Curry†‡, Paige Thomas†‡,
Margaret Walshe§, Ellen McCague§, Irene Battel§¶, Dalia Nogueira, Ulrike Frank#, Lenie van den
Engel-Hoek∗∗ and Oshrat Sella-Weiss††‡‡
†Swallowing Rehabilitation Research Laboratory at the Rose Centre for Stroke Recovery and Research, Christchurch,
New Zealand
‡Department of Communication Disorders, The University of Canterbury, Christchurch, New Zealand
§Department of Clinical Speech and Language Studies, Trinity College Dublin, University of Dublin, Dublin, Ireland
¶Fondazione Ospedale di Neuroriabilitazione IRCCS San Camillo, Venice, Italy
Instituto Universitário de Lisboa (ISCTE-IUL), Business Research Unit (BRU-IUL), Lisbon, Portugal
#Department of Cognitive Neurolinguistics, Swallowing Research Lab, University of Potsdam, Potsdam, Germany
∗∗
Donders Centre for Neuroscience, Department of Rehabilitation, Radboud University Medical Center, Nijmegen,
the Netherlands
††Department of Communication Disorders, Ono Academic Collage, Kiryat Ono, Israel
‡‡Department of Communication Disorders, University of Haifa, Haifa, Israel
(Received January 2017; accepted May 2017)
Abstract
Background: Clinical swallowing assessment is largely limited to qualitative assessment of behavioural observations.
There are limited quantitative data that can be compared with a healthy population for identification of impairment.
The Test of Masticating and Swallowing Solids (TOMASS) was developed as a quantitative assessment of solid
bolus ingestion.
Aims: This research programme investigated test development indices and established normative data for the
TOMASS to support translation to clinical dysphagia assessment.
Methods & Procedures: A total of 228 healthy adults (ages 20–80+ years) stratified by age and sex participated in
one or more of four consecutive studies evaluating test–retest and interrater reliability and validity to instrumental
assessment. For each study the test required participants to ingest a commercially available cracker with instructions
to ‘eat this as quickly as is comfortably possible’. Further averaged measures were derived including the number
of masticatory cycles and swallows per bite, and time per bite, masticatory cycle and swallow. Initial analyses
identified significant differences on salient measures between two commercially available crackers that are nearly
identical in shape, size and ingredients, suggesting the need for separate normative samples for specific regional
products. Additional analyses on a single cracker identified that the TOMASS was sensitive at detecting changes
in performance based on age and sex. Test–retest reliability across days and interrater reliability between clinicians
was high, as was validation of observational measures to instrumental correlates of the same behaviours. Therefore,
normative data are provided for the TOMASS from a minimum of 80 healthy controls, stratified by age and sex,
for each of seven commercially available crackers from broad regions worldwide.
Outcomes & Results: Analyses on a single cracker identified Arnott’s Salada, and that TOMASS measures were
sensitive for detecting changes in performance based on age and sex. Interrater and test–retest reliability across
days were high, as was validation of observational measures to instrumental correlates of the same behaviours.
Significant differences were identified between two commercially available crackers, nearly identical in shape, size
and ingredients, thus normative samples for specific regional products were required. Normative data were then
acquired for the TOMASS from a minimum of 80 healthy controls, stratified by age and sex, for each of seven
commercially available crackers from broad regions worldwide.
Address correspondence to: Maggie-Lee Huckabee, The University of Canterbury Rose Centre for Stroke Recovery and Research, Leinster
Chambers, Level One, 249 Papanui Road, Merivale, Christchurch 8042, New Zealand; e-mail: maggie-lee.huckabee@canterbury.ac.nz
International Journal of Language & Communication Disorders
ISSN 1368-2822 print/ISSN 1460-6984 online C 2017 Royal College of Speech and Language Therapists
DOI: 10.1111/1460-6984.12332
TOMASS: test development 145
Conclusions & Implications: The TOMASS is presented as a valid, reliable and broadly normed clinical assessment
of solid bolus ingestion. Clinical application may help identify dysphagic patients at bedside and provide a
non-invasive, but sensitive, measure of functional change in swallowing.
All participants were self-reported to be healthy with and the Osem Golden cracker (6.7 × 4.3 cm, 3.6 g)
no history of dysphagia, head and neck, neurological or are both larger and perceptually similar, while the Carr’s
gastroenterological illness or injury. Informed written Table Water cracker (6 cm diameter; 3.5 g) is smaller
consent was obtained from each participant and all pro- and perceptually drier than the other crackers.
tocols were approved and conducted in accordance with Objective measures of mastication and swallowing
the ethical standards of the relevant University or Re- in study 3 were collected using surface electromyography
gional Human Ethics Committee. (sEMG), acoustic and nasal airflow functions of the Kay
Pentax Digital Swallowing Workstation.
Materials
Data collection
Several commercially available crackers were utilized for
this research programme. The Arnott’s SaladaTM cracker For all studies, including normative database develop-
was used in all studies and is readily available throughout ment, the TOMASS was carried out in the following
New Zealand and Australia. Each cracker weighs 3.75 manner. Participants, seated comfortably, were asked to
eat a single portion of the cracker ‘as quickly as is com-
g with dimensions of 5 cm2 . The Nabisco SaltineTM
fortably possible and when you have finished, say your
cracker was selected for comparison with the Salada.
name out loud’. They were advised not to talk during
This cracker weighs 16 g with dimensions of 5 cm2 and
ingestion. However, speaking their name on comple-
is readily available across North America. Both crackers
tion of the entire cracker was used as a marker of task
contain unbleached flour (wheat flour, niacin, reduced
completion and oral cavity clearance. Participants were
iron, thiamine, mononitrate (vitamin B1), riboflavin
carefully observed and the number of bites was deter-
(vitamin B2), and folic acid, soybean oil, partially hy-
mined by how many discrete segments of cracker the
drogenated cottonseed oil, sea salt, salt, malted barley
participant placed in their mouth, while the number of
flour, baking soda and yeast) and are identical in size,
swallows was recorded based on visual observation of
shape and appearance. Exact proportions of ingredients
movement of the thyroid cartilage. Both measures were
were not supplied by the manufacturer on request. How-
manually recorded on a data-collection sheet. The num-
ever, the Saltine cracker is perceptually slightly more
ber of masticatory cycles was counted through observa-
crumbly and dry. Further normative data were collected
tion of jaw movements; a lap function on a digital stop-
using the Carr’s Table WaterTM cracker (Ireland and the watch was used to mark each masticatory cycle. Timing
UK), Albert Heijn BasicTM cracker (the Netherlands), was initiated when the cracker passed the bottom lip and
DeBeukelaer Tuc ClassicTM cracker (Germany), Gran was stopped when the participant said their name. For
PavesiTM cracker (Italy and Portugal), and the Osem all participants, the above procedure was carried out a
GoldenTM cracker (Israel). The crackers chosen from second time.
the Netherlands and Italy/Portugal were very similar, For the study of reliability, 40 participants ingested
but not identical, in ingredients, size and perceptual the Arnott’s Salada cracker twice in a single session,
characteristics with the previously described Salada. The using the same method of data collection and allowing
DeBeukelaer Tuc Classic cracker (5 × 6.5 cm, 3.75 g) for water ingestion to clear the oral cavity and moisten
148 Maggie-Lee Huckabee et al.
mucosa before and between the two trials. To evaluate compare the data between the two crackers (Arnott’s
test–retest consistency, data collection was repeated on Salada and Nabisco Saltine). An a priori decision was
three consecutive days. During one session only, two made that if no significant difference was identified
raters were present to make independent measures of between crackers, all subsequent analyses would be
participant performance as an assessment of interrater completed on the combined data from both crackers.
reliability. If a significant difference between data on any raw
For the study of validity, 24 participants ingested data measure was identified, subsequent analyses would
the Arnott’s Salada cracker twice in one session. Par- be completed for each cracker independently. A t-test
ticipants then returned after a period of at least 24 h was then conducted to evaluate for a trial effect be-
at which they completed the TOMASS twice more. A tween first and second trials. General linear model, two-
glass of water was offered to participants prior to the way, fixed-factor MANOVAs were then completed to
first trial as well as between the two trials. Objective evaluate the influence of age and gender on all vari-
measures were collected with sEMG electrodes placed ables. The reported p-values represent application of
over the masseter and submental muscles, nasal prongs Bonferonni correction for multiple comparisons when
to detect respiratory phase and a stethoscope secured appropriate.
over the lateral aspect of the thyroid cartilage to de- Cronbach’s alpha and mixed-model intraclass corre-
tect swallowing acoustics. All data recorded using the lation coefficients using single measure methods were
Kay-Pentax Digital Swallowing Workstation. All sen- derived for the raw data only to evaluate test–retest
sors were placed on the right side of the participant’s consistency of performance across the three sessions
face to allow optimal viewing for the researcher who and interrater reliability between two raters in a single
was positioned to the left of the participant. Objective session.
measurement of one masticatory cycle was determined Intraclass correlation coefficients using single mea-
by the point at which the sEMG amplitude for masseter sure methods were calculated to evaluate validity of be-
activity was at maximum and for submental muscles havioural measures when compared with instrumental
was at minimum, followed by a reversal of these signals, assessment. Analysis was also conducted to determine
indicating jaw closure and opening. A swallowing event the interrater reliability between two raters evaluating
was denoted by the presence of swallowing apnoea in the the objective data.
respiratory waveform, accompanied by a peak in the sub- Finally, normative data were established for partici-
mental muscle sEMG activity. The acoustic signal was pants ingesting each cracker, calculated by age and gen-
used as additional confirmation of swallowing; however, der as mean and 95% confidence interval for number
a strong acoustic signal was not clearly detected in all of bites, number of masticatory cycles and number of
participants. The objective measure for time taken was swallows per cracker as well as total time. Further nor-
from the first chew recorded by sEMG until the time in mative data were calculated for the derived measures of
which there was a large acoustic signal indicating that masticatory cycles per bite, swallows per bite, time per
the participant had said their name to indicate that they bite, time per masticatory cycle and time per swallow.
had finished.
Results
Data preparation and statistical analysis
Study 1: Cracker, trial, age and gender effects
In addition to the raw data of number of discrete
Salada versus saltine comparison
bites, masticatory cycles, swallows per cracker and total
time required for ingestion, several additional derived For both the first and second trials of the TOMASS
measures were calculated, similar to those derived for there were significant differences between crackers for
the TWST. These measures included averaged num- most, but not all, measures (table 2). In general, the
ber of masticatory cycles per bite (number of mastica- group ingesting the Salada cracker took more discrete
tory cycles/number of discrete bites), averaged number bites, required more masticatory cycles and swallows,
of swallows per bite (number of swallows/number of and more time to ingest the cracker than the group
discrete bites), averaged time per bite (number of dis- ingesting the Saltine cracker.
crete bites/total time), time per masticatory cycle (total
time/number of masticatory cycles), and time per swal- Trial effect. Paired t-tests compared data from the
low (total time/number of swallows). first trial with data from the second trial on all variables.
Two general linear model one-way, fixed-factor mul- Three of the four raw data measures (discrete bites: t =
tivariate analyses of variance (MANOVAs)—one based –3.29, p < .01; masticatory cycles: t = –2.14, p =
on data from the first trial, and one on data from .035; and swallows per cracker: t = –2.62, p = .01)
the second trial—were conducted on all variables to were significantly different between the first and second
TOMASS: test development 149
trials, with a tendency toward slower, less efficient per-
< .01
< .01
.02
< .01
.01
.04
.06
.36
.95
p
formance on the second trial. No significant differences
Saltine were identified for total time to ingest and the derived
.004
measures (p > .05). However, as a trial effect was evi-
27.63
7.99
5.29
12.07
7.01
4.11
3.66
.86
F
Sex Effect
< .01
< .01
< .01
< .01
< .01
.53
.41
completed only on the first trial.
p
Salada
7.38
7.86
7.53
.39
.69
7.2
13.9
F
< .01
< .01
.02
.33
.50
< .01
.73
sex: F = 3.56, p < .001); but no significant age and
p
23.39
3.68
1.16
.79
4.20
.43
for the four raw data variables of discrete bites, mas-
10.1
F
Table 2. Summary of the statistical output for the evaluation of trial, age and gender effects
Age Effect
< .01
< .01
.99
.14
.75
.59
< .01
7.32
19.97
0.01
1.85
0.41
0.64
4.71
25.0
F
< .01
< .01
.01
< .01
< .01
< .01
.43
35.22
33.24
6.27
13.12
18.71
14.36
.61
2.51
33.94
18.96
1.06
14.68
.79
14.70
42.5
2.5
3.13
62.35
55.38
22.98
19.96
.90
15.54
1.5
Cracker effect
<.01
<.01
< .01
< .01
< .01
<.01
.07
p
14.41
21.22
7.10
7.42
15.51
7.97
3.36
44.68
2.59
2.68
35.80
19.38
1.14
15.34
.82
14.24
3.57
53.39
23.65
1.46
21.17
.90
16.13
#masticatory
masticatory
#swallows
cycles
cycles
swallow
cycle
sec)
bite
Interrater
Note: a Mean values.
Average time
per event
(in sec)
cracker
Measure
Table 3. TOMASS normative data consisting of mean and 95% confidence intervals by age and gender for Arnott’s SaladaTM cracker
Table 4. TOMASS normative data consisting of mean and 95% confidence intervals by age and gender for the Nabisco SaltineTM cracker
10.50–17.16
12.51–25.73
14.13–24.35
16.27–26.84
9.24–13.00
11.46–18.79
12.71–25.29
11.61–21.26
Time per swallow (in
95% C.I.
table 5.
sec)
Test–retest reliability
13.83
19.13
19.24
29.55
11.12
15.12
19.00
16.44
Mean
Test–retest reliability was also very high. Cronbach’s α
95% C.I.
.80–1.08
.82–1.29
.64–.98
.70–.98
.75–1.0
.73–.94
.68–.99
.77–.93
cycle (in sec)
and .98 suggesting a high level of internal consistency.
Data for individual measures are summarized in table 4.
Mean
.81
.84
.87
.83
.84
.94
1.06
.85
Study 3: Validity
11.07–17.70
11.07–25.35
12.75–23.43
17.46–29.93
9.99–14.45
9.27–15.56
10.04–24.25
10.98–16.51
Time per bite (in sec)
95% C.I.
The ICC value between objective and behavioural mea-
sures of the number of masticatory cycles was .99 with
Table 5. TOMASS normative data consisting of mean and 95% confidence intervals by age and gender for Carr’s Table WaterTM cracker
a 95% confidence interval from .98 to .99 (F(d.f. =
14.39
18.21
18.09
23.69
12.23
12.42
17.15
13.75
Mean
95) = 142.26, p < .001). For number of swallows, the
ICC was .85 with a 95% confidence interval from .79
.94–1.16
.81–1.04
.79–1.13
.76–1.65
.98–1.24
.67–1.05
.62–1.34
.72–1.06
95% CI
Swallows per bite
time was .99 with a 95% confidence interval from .91
to 1.0 (F(d.f. = 95) = 634.51, p < .001).
The ICCs for the reliability of two independent
Mean
1.05
.93
.96
1.21
1.11
.86
.98
.89
raters of the instrumental measures were greater than
.95 for the number of masticatory cycles and time taken.
12.53–26.40
14.24–28.67
14.72–27.43
19.97–38.11
12.23–18.56
10.83–15.47
9.81–23.64
13.22–19.03
95% C.I.
The ICC for interrater reliability of the number of swal-
lows recorded by instrumental assessment was .73.
19.47
21.45
21.08
29.04
15.39
13.15
16.73
16.13
Normative data represented by the mean and 95% confi- Mean
17.54–30.54
23.29–33.87
25.61–53.77
42.63–67.42
29.78–38.55
34.06–43.88
37.08–71.38
39.56–62.06
dence interval for the TOMASS during the first trial in-
95% C.I.
Total time (in sec)
Discussion
Mean
1.90
1.83
2.30
2.75
3.17
2.91
3.22
3.44
95% C.I.
20–40
40–60
60–80
Age
80+
Table 6. TOMASS normative data consisting of mean and 95% confidence intervals by age and gender for the Gran PavesiTM cracker
Table 7. TOMASS normative data consisting of mean and 95% confidence intervals by age and gender for the DeBeukelaer Tuc ClassicTM cracker
Males 20–39 1.25 1.04–1.46 31.4 25.23–37.57 1.35 1.04–1.66 22.8 18.72–26.88 27.02 21.07–32.98 1.10 .920–1.28 19.53 15.85–23.20 .75 .68–.83 18.71 15.07–22.37
40–59 1.43 1.09–1.77 34.62 30.4–39.20 1.81 1.22–2.39 27.90 22.72–31.98 27.35 22.72–31.98 1.39 1.03–1.74 21.36 17.86–24.87 .80 .70–.89 18.49 15.01–21.98
60–79 2.57 1.39–3.75 34.90 37.69–42.12 1.48 1.17–1.78 26.38 21.42–31.35 19.95 13.90–26.00 .92 .61–1.22 14.65 10.39–18.53 .78 .68–.88 20.74 15.41–26.08
Females 20–39 1.9 1.422–2.18 29.95 26.44–33.46 1.95 1.62–2.27 27.1 22.89–31.02 19.80 15.19–24.42 1.22 .92–1.28 17.47 13.09–21.85 .90 .82–.98 15.63 11.58–19367
40–59 2.59 2.01–3.16 34.91 31.45–38.37 1.72 1.36–2.09 30.41 26.72–34.10 17.69 12.68–22.69 .87 .56–1.18 15.78 10.94–20.61 .88 .79–.97 21.03 16.56–25.50
60–79 3.27 2.72–3.82 46.5 40.85–52.15 1.68 1.31–2.05 40.32 35.02–46.62 15.83 12.91–18.74 .57 .42–.78 13.71 11.08–16.34 .87 .80–.94 27.18 23.11–33.24
Table 9. TOMASS normative data consisting of mean and 95% confidence intervals by age and gender for the Osem GoldenTM cracker