The Test of Sensory Functions in Infants: Test-Retest Reliability For Infants With Developn1ental Delays
The Test of Sensory Functions in Infants: Test-Retest Reliability For Infants With Developn1ental Delays
The Test of Sensory Functions in Infants: Test-Retest Reliability For Infants With Developn1ental Delays
T
he imegration and organization of sensory infor-
mation has been described as vital to learning
and adaptive behavioral processes (Ayres, 1972).
Sensory imegrarion refers co the overall process by which
the brain receives, registers, and combines sensory input
in order to generate adaptive responses co the environ-
mem (StalJings-Sahler, 1993). Disordered sensory ime-
Tracy L. Jirikowic, MS, OTRlL, is SraffTherapisr, Kwiar Child
Development Center, Seattle, Washington, and Staff Thera- gration was believed co account for some aspects of child-
pist, Fetal Alcohol Syndrome Clinic, Center for Human hood learning and behavior disorders that have been
Development and Disability, University of Washington, identified in preschool and elememary-school-aged chil-
Seattle, Washington. (Mailing address: 727 Bellevue Avenue dren (Ayres, 1972). It is now speculated that certain early
E., Seattle, Washington 98102) sensory processing mechanisms are related to and can
prediCt later learning, behavioral, and developmental dif-
Joyce M. Engel, PhD. OTRlL, FAOTA, is Assistant Professor, De-
ficulties (Greenspan, 1992; Stallings-Sahler, 1993).
partment of Rehabilitation Medicine, Occupational Therapy
The Test of Sensory Functions in Infants (TSFI)
Division, University of WashingtOn, Seattle, WashingtOn.
(DeGangi & Greenspan, 1989) is a 24-item standardized
Jean C. Deirz, PhD, OTRlL, FAOTA, is Associate Professor and assessment that occupational therapists use to identify
Graduate Program Coordinator, Department of Rehabilitation sensory processing deficits in infants. The test examines
Medicine, Occuparional Therapy Division, University of five areas of sensory function: (a) Reactivity to Tactile
Washington, Seattle, Washington. Deep Pressure, (b) Visual-Tactile Integration, (c) Adap-
tive Motor Responses, (d) Ocular-Motor Control, and
This article was acceptedfor publication January 20, 1997.
(e) Reactivity to Vestibular Stimulation. The TSFI was
other 4 fluctuated within the deficient category. Thus, test would reflect a more typical response to tactile input,
despite a wide point spread, it appeared that the classifica- suggesting that some of the infants may have accommo-
tion categories for the total test remained relatively stable. dated to the test stimuli on the second administration.
No specific patterns emerged regarding the subtest perfor- No other obvious patterns emerged regarding the distrib-
mance of these last 4 infants because mild to moderate utions of magnitude of difference between test and retest
variabilities in the point spreads and classification cate- scores for the remaining subtests. Magnitude of differ-
gories were seen in all domains. It also should be noted ence between test and retest scores were relatively evenly
that these 4 infants with the highest variability were distributed in both directions.
reported by their caregivers to be more tired or irritable The findings suggest that the total test classifications
during one of the test sessions. were more stable than the subtest classifications. For the
On the Visual-Tactile Integration subtest, 10 infants total test, 21 infants were classified in the same category
scored one to three points higher between test and retest (i.e., normal, at-risk, deficient) on the test and retest ses-
versus only 2 scoring lower. A higher score on this sub- sions. The percentages of agreement for the sub tests be-
Table 2
Magnitude of Difference Between Test and Retest TSFI Scores
Point Spread
Test < -7 -6 -5 -4 -3 -2 -1 0 2 3 4 6 >7
Reacriviry ro Tactile Deep Pressure
n 3 7 12 2 1 1
% 11.5 26.9 46.2 7.7 3.8 3.8
Adaptive Motor Responses
n 1 2 4 8 6 3 I I
% 3.8 7,7 15.4 30.8 23.1 11.5 3.8 3.8
VisuaJ-Tactile Integration
1 I 14 4 2 4
%" 3.8 3.8 53.8 15.4 7.7 15.4
Ocular-Motor CancroI
I 3 17 5
%
" 3.8 11.5 65.4 19.2
ReaC(iviry ro Vestibular Stimulation
n 2 3 3 13 1 2 1 1
% 7.7 11.5 11.5 50.0 3.8 7.7 3.8 3.8
Total Test
2a I 3 3 3 I 6 4 1 2b
"
% 7.7 38 11.5 11.5 11.5 3.8 23.1 15.4 3.8 7.7
Note. N = 26. TSFI = Test fat Sensory Functions in Infants.
aThe magnitudes of difference fat these scores were -7 and -I 1,
b-rhe magnitudes of difference for these SCates were 9 and 10,
tween test and retest ranged from 58% to 69%. Agree- Additional information regarding the behavior of the
ment was highest for the Reactivity to Vestibular Stimu- infants during the test performance also should be consid-
lation subtest (69%), the Visual-Tactile Integration sub- ered when interpreting results. Infants who were reported
test (65%), and the Ocular-Motor Control subtest (65%), by their caregivers as tired, irritable, or fussy during one of
followed by the Reactivity to Tactile Deep Pressure subtest the test sessions generally performed more poorly on one
(58%) and the Adaptive Motor Responses subtest (58%). session than the other. Despite variabili ty in the actual
Again, subtest scores were somewhat unstable; therefore, total test scores, these infants' classification categories
caution is recommended when using these results. A de- remained relatively stable. No specific diagnostic patterns
tailed analysis of changes between total test and subtest were noted in this group of infants; however, all of them
categories is provided in Table 3. scored in the deficient performance range for the total
In comparison to DeGangi and Greenspan's (1989) test. None of these infants met criteria for exclusion be-
results from preliminary studies with the TSFI, reliability fore testing (i.e., they did not demonstrate inconsolable
coefficients from the current study were slightly lower on crying or fussiness); instead, their irritability and fatigue
the total test, the Reactivity to Tactile Deep Pressure sub- appeared to increase during testing. This suggests that the
test, and the Adaptive Motor Responses subtest. Relia- infants who demonstrated sensory dysfunction were even
bility coefficients from our study were substantially lower less able to cope with the test stimuli and test demands
on the Visual-Tactile Integration and Ocular-Motor Con- when factors such as fatigue or illness were present. There-
trol subtests, and our reliability coefficient for the Reac- fore, clinicians may want to ask caregivers about the in-
tivity to Vestibular Stimulation subtest was higher. When fant's health, behavior, and mood before testing and sched-
comparing the results from these two studies, it is impor- ule testing or provide descriptive information accordingly.
tant to note several differences. DeGangi and Greenspan It also seemed that infants who appeared to have more
examined infants who were typically developing, most of motor deficits performed most poorly on the Visual-
whom were 4 months to 6 months of age; analyzed the Tactile Integration and Adaptive Motor Responses sub-
data with Pearson product-moment correlations; and used tests, both of which require more refined motor responses
more than one examiner to test the infants. (e.g., removing a sticker from the hand or a toy placed on
From their preliminary test-retest studies, DeGangi the stomach). The scores on these subtests with infants
and Greenspan (1989) concluded that decisions regarding wirh mowr deficits may reflect both sensory and motor
infants' vestibular and adaptive motor functioning should performa.nce rather than jusr sensory function, meriting
not be made until test-retest results confirm normalcy or further investigation.
dysfunction. The borderline reliability coefficients and
magnitudes of difference between test and retest scores Limitations
from this study also support these conclusions as well as The limirarions of rhis study included rhe variable narure
suggest that the clinician interpret all subtest scores with of the tested behaviors themselves, which can be affected
caution. In addition, because of the low reliability coeffi- by factors such as the infant's state of arousal, health, and
cients for the subtests, it is recommended that subtest farigue. In addition, rhe small hererogeneous san1ple lim-
scores be used with extreme caution for diagnosric pur- irs the generalizability of rhe results.
poses and documenting change over time. However, the
clinician could have some confidence regarding the stabil- Conclusion
ity of normal, at-risk, or deficient classification decisions Results of rhis srudy suggest thar rhe resr-retest reliability
for rhe toral test. However, funher study is needed In for rhe TSFI roral test scores is borderline and that the
order to evaluate the validity of these decisions. test-retest reliabilities for the subtest scores are low. Per-