Eliability and Validity of The Harris Infant Neuromotor Test

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Reliability and validity of the Harris Infant

Neuromotor Test
Susan R. Harris, PhD, PT, and Linda E. Daniels, MA, OTR, OTC
are suspected of having a developmen-
Objective: To examine the reliability and validity of the Harris Infant Neu- tal concern from those who appear to
romotor Test (HINT), a screening tool that can be administered and scored be developing normally.2 Those with
in <30 minutes, with the goal of identifying neuromotor differences in in- suspected neuromotor or cognitive
fants aged 3 to 12 months. concerns are then referred for more
comprehensive developmental assess-
Study design: Infants, aged 3 to 12 months (n = 54), were assessed in 2
ments, such as the Bayley Scales of In-
high-risk infant follow-up programs in Vancouver, British Columbia. Inter-
fant Development-II.3
rater, test-retest, and intra-rater reliability were examined. Concurrent and
predictive validity of the HINT with the Bayley Scales of Infant Develop- BSID Bayley Scales of Infant Development
ment-II (BSID-II) were evaluated by using the Pearson product-moment BSID-II Bayley Scales of Infant Development-II
HINT Harris Infant Neuromotor Test
correlation. ICC Intraclass correlation coefficient
Results: Intraclass correlation coefficients for reliability for the Total
HINT Score ranged from 0.98 to 0.99. Concurrent validity of the HINT Criteria for evaluating screening
with the BSID-II Mental Scale during the first year was r = –0.73 (P < .01), tests include acceptability (to the chil-
and with the BSID-II Motor Scale, r = –0.89 (P < .01). The predictive rela- dren and family being screened, the
tionships between the HINT and the BSID-II at 17 to 22 months were professionals receiving the referrals,
r = –0.11 for the BSID-II Mental Scale and r = –0.49 (P < .01) for the and the community at large), simplici-
ty (in learning the test and in its ad-
BSID-II Motor Scale.
ministration), cost, appropriateness (to
Conclusions: The HINT is reliable for screening infant neuromotor perfor- the population under consideration),
mance and has strong concurrent validity with the Bayley-II Mental and reliability, and validity.4
Motor Scales. HINT scores during the first year accounted for 24% of the The Harris Infant Neuromotor Test
variance of Bayley-II Motor scores at 17 to 22 months. (J Pediatr 2001;139: is being developed as a simple, nonin-
249-53) vasive, reliable, and valid screening
tool for use in either clinical or re-
search settings to identify neuromotor
Although approximately one half of ture birth (ie, two thirds of infants with differences in infants aged 3 to 12
children with cerebral palsy were born later cerebral palsy) or prenatal sub- months. Designed to be administered
prematurely1 and thus are more apt to stance exposure, are seldom screened in <30 minutes, the HINT is intended
be assessed in high-risk infant follow- or tested for developmental concerns. for use by occupational therapists,
up clinics, those infants who are not Most pediatric screening tests are physical therapists, pediatricians, fam-
deemed “at risk,” either from prema- designed to discriminate children who ily physicians, and community health
nurses. Item development of the HINT
From the School of Rehabilitation Services, Faculty of Medicine, University of British Columbia, Vancouver, British
Columbia, Canada.
was based on extensive earlier re-
Supported in part by a grant from the British Columbia Health Research Foundation (No. 146- search on early identification of neu-
95-1). rodevelopmental problems.5-15
Submitted for publication Aug 23, 2000; revisions received Dec 13, 2000, and Feb 23, 2001; The first development edition of the
accepted Mar 15, 2001.
HINT16 was used in a pilot study to
Reprint requests: Susan R. Harris, PhD, PT, School of Rehabilitation Sciences, Faculty of Med-
assess 31 high-risk infants, aged 3 to 9
icine, T325-2211 Wesbrook Mall, University of British Columbia, Vancouver, British Columbia
V6T 2B5, Canada. months.17 A series of 4 questions was
Copyright © 2001 by Mosby, Inc. posed to the primary caregivers about
0022-3476/2001/$35.00 + 0 9/21/115896 how their infants were faring develop-
doi:10.1067/mpd.2001.115896 mentally. Their responses were used to

249
HARRIS AND DANIELS THE JOURNAL OF PEDIATRICS
AUGUST 2001

compare their perceptions of the in- grams in Vancouver, British Columbia, composition were not identical for
fant’s development with test results on Canada. The Infants and Children at each aspect of the study.
the Bayley Scales of Infant Develop- Risk Program at Sunny Hill Health
ment.18 A high degree of concurrence Centre for Children provided longitu- Examiners and Examiner
was found, particularly between the dinal developmental assessments for Training
caregivers’ judgments in response to the infants and young children who were The assessments were conducted by
HINT questions and the therapist’s test at risk because of prenatal exposure to 3 physical therapists and 2 occupation-
results with the Bayley Motor Scale drugs and/or alcohol. The Neonatal al therapists. In addition to pediatric
(sensitivity = 80%; specificity = 90.9%). Follow-Up Programme at British Co- experience, the therapists had training
To examine the content validity of lumbia Children’s Hospital provides and experience in administering and
the HINT, that is, the “representative- systematic, comprehensive develop- scoring the Movement Assessment of
ness” of the HINT items to the concept mental assessments for extremely low Infants5 and the first edition of the
of neuromotor development, the first birth weight infants (<1000 g), as well BSID.18 The senior author trained the
development edition was distributed to as term infants with other biologic risk other testers by assessing 3 typically
an international panel of 26 experts for factors such as neonatal seizures, hy- developing infants (aged 5.5 to 7
assessment, review, and modifications. drocephalus, or intrauterine growth months) while the other 4 examiners
The revised edition of the HINT retardation. Sixty-one percent of the observed and scored the test. In addi-
(1993) was used in the current study.19 infants in the sample were from Sunny tion, training and inter-rater reliability
The 5-item parent/caregiver section Hill and 37% were from British Co- evaluation were conducted on the
is designed to assess the parents’ level lumbia Children’s Hospital. One addi- BSID-II3 by the senior author. Six in-
of concern and the specific areas of tional premature infant was recruited fants, ranging in age from 31⁄2 to 18
concern that they may have about their from the Outreach Therapy Pro- months and without any risk factors,
child. The infant assessment section gramme in Port Alberni, British Co- were assessed on the BSID-II with
(items 1-21) includes motor behaviors lumbia. Children with known congeni- inter-observer reliability conducted by
in a supine-lying position, during tran- tal anomalies were excluded. The each of the other assessors. Each of the
sition to a prone-lying position, in a study was approved by the Clinical 5 assessors served as the primary ex-
prone-lying position, and in supported Research Ethics Board at the Univer- aminer for test administration and
sitting and supported standing posi- sity of British Columbia, as well as by scoring for at least one infant, with the
tions. The items are ordered in a logical the ethics committees of both hospitals. other 4 assessors observing and scor-
sequence, by position, to minimize in- Informed consent was obtained from a ing the test independently. Inter-rater
fant handling. Head circumference is parent or guardian of each infant. reliability for raw scores on the BSID-
measured also, as well as behavioral Nineteen female and 35 male infants II, across the 5 assessors, was obtained
state and stereotypical behaviors. A 2- ranged in chronological age from 3.07 at acceptable levels for the study to
part developmental and qualitative months to 12.30 months (mean = 7.79 proceed: intraclass correlation coeffi-
judgement item (item 22) concludes the months) at the initial assessment, with cient (2,1) = .993 for the Mental Scale;
test and is designed to summarize the age corrected for prematurity in in- ICC (2,1) = .995 for the Motor Scale.
examiner’s overall clinical impressions fants born at <38 weeks’ gestation. Al-
about the developmental appropriate- most two thirds (63.5%) were born be- Instruments
ness of the infant’s neuromotor perfor- fore term (23 to 37 weeks’ gestation), The HINT was administered by 1 of
mance and the quality of the infant’s and 51.9% had low birth weight the 5 assessors during the first year of
movement behaviors.19 The purpose of (<2500 g). Mean birth weight was the infant’s life (3 to 12 months). The
the present study was to examine the 2494 g; mean gestational age was first part of the HINT includes back-
reliability and the criterion-related va- 35.67 weeks. Race of the infant’s ground information on the child and the
lidity of the HINT in assessing infants mother was white (43.1%), Native In- caregiver. The second part is a 5-item
at risk for neuromotor and/or cognitive dian/First Nations (25.5%), Asian parent/caregiver questionnaire that is
difficulties. (13.7%), Hispanic (2%), East Indian designed to assess specific areas of con-
(2%), or other/unknown (13.7%). cern. The third part, the Infant Assess-
METHODS Outcome assessments took place be- ment, includes the 21 items that are ob-
tween 17.53 and 22.23 months of served and scored by the test examiner,
Subjects and Sites chronological age (mean age = 18.95 such as motor behaviors in a supine-
A convenience sample of 54 high- months). Some subjects were not able lying position; motor behaviors during
risk infants was recruited, primarily to keep all assessment appointments, transition from a supine-lying to a
from 2 developmental follow-up pro- so the number of subjects and group prone-lying position; motor behaviors in

250
THE JOURNAL OF PEDIATRICS HARRIS AND DANIELS
VOLUME 139, NUMBER 2

a prone-lying position; motor behaviors Concurrent validity was examined by RESULTS


in a sitting position; observation of loco- comparing the infants’ HINT scores
motion, posture, movements, and be- with the raw scores on the BSID-II The ICC for inter-rater reliability for
havioral state; and measurement of head Mental and Motor Scales, which were the 28 infants was 0.99 for the Total
circumference. After calculation of the administered during the same day in HINT score. Of the 21 individual
infant’s total HINT score based on sum- the infant’s first year of life. Predictive HINT items, the ICC was ≥.80 for 9
ming the results of the 21 assessment validity was evaluated by comparing items (42.8%) and ≥.70 for 18 items
items, the examiner makes two summa- the HINT scores during the infant’s (85.7%). However, for 2 HINT items
ry decisions about the infant’s overall first year with the raw scores on the (eye muscle control and stereotypical
neuromotor performance, one for the in- BSID-II Mental and Motor Scales behaviors), the ICCs were <.15 (com-
fant’s developmental level and one for during the second year of life (17 to 22 plete data available from authors on re-
the quality of motor performance. months). BSID-II raw scores were quest).
In the current study, the Mental and used because they represent the most For the 20 infants who were each test-
Motor Scales of the BSID-II were ad- similar psychometric counterpart to the ed twice within 9 days, the ICC value
ministered. Raw scores are converted HINT total raw scores. To avoid possi- for test-retest reliability of the total
to index scores. The Mental Develop- ble expectation bias, examiners were HINT score was 0.98. For the 21 indi-
mental Index and Psychomotor Devel- blinded to all previous test scores dur- vidual HINT items, the ICC was ≥.80
opmental Index both have means of ing administration of the later BSID-II. for 12 items (57.1%) and ≥.70 for 14
100 and SDs of 15. Although an in- items (66.7%). For 2 HINT items, how-
creasing raw score on the HINT indi- Data Analysis ever, the test-retest reliability was very
cates increasing risk, an increasing raw The ICC (2,1) was used to analyze poor (eye muscle control = 0.000; stereo-
score on the BSID-II represents im- inter-rater, test-retest, and intra-rater typical behaviors = 0.377) (complete
proved or more mature performance. reliability for the total HINT score.20-22 data available from authors on request).
An ICC of ≥0.80 was set for inter-rater, Intra-rater reliability was assessed
Procedures test-retest, and intra-rater reliability as for each of the 5 therapists who twice
To avoid bias, the HINT was adminis- “acceptable,” although ≥0.90 is consid- viewed videotapes of 20 infants, ap-
tered first. Background information was ered desirable.23 proximately 1 month apart in time.
first recorded from the chart. The To examine concurrent and predic- Across the 5 therapists, ICC values
BSID-II Mental Scale was adminis- tive validity of the total HINT score, ranged from 0.98 to 0.99 for the Total
tered next, followed by the Motor Scale. the Pearson r was used to express the HINT score (complete data available
To examine inter-rater reliability, 28 degree of relationship between the on request). Our hypotheses were thus
infants were assessed by 1 of the 5 HINT and BSID-II (Mental and supported; that is, all ICC reliability
therapists, designated as the primary Motor Scales). A level of r ≥ 0.60 was values were ≥.80.
examiner, while a second therapist ob- set for concurrent validity of the The relationships of the Total HINT
served and scored the test. The second HINT with the Mental and Motor score to raw scores on the BSID-II
examiner independently administered Scales of the BSID-II. This level was Mental Scale and BSID-II Motor
the 2 items that involve assessment of set based on previous research exam- Scale were calculated by using the
muscle tone (passive range of motion ining concurrent validity between the Pearson product-moment correlation.
in supine position and passive range of BSID Mental Scale and the Stanford- These 3 scores were available for all 54
motion in prone position). Binet IQ test18 for a sample of 2-year- infants in the sample. Because the
To evaluate test-retest reliability, the olds. A correlation of r = 0.57 was re- HINT is predominantly a neuromotor
primary examiner re-administered the ported as a “substantial degree of screening tool, it is not surprising that
HINT to 30 infants within a period of agreement” between the 2 tests. To ex- it correlated more strongly with the
2 to 9 days without referring to the amine the predictive validity of the BSID-II Motor Scale (r = –0.89, P <
original test sheet and scores. HINT to the later BSID-II, an r value .01) than with the BSID-II Mental
Intra-rater reliability was assessed of ≥0.30 was hypothesized between the Scale (r = –0.73, P < .01).
by videotaping the administration of first-year HINT and the second-year To examine the predictive validity of
the initial HINT assessment for 20 in- BSID-II Mental and Motor Scales. the HINT during year 1 (3 to 12
fants in the sample. The videotapes This level was set based on results of months of age), the relationships of the
were later viewed twice by each of the previous research examining the pre- total HINT scores to the raw scores on
5 therapists at an interval of ≥1 month. dictive validity of the Movement As- the BSID-II Mental and Motor Scales,
The assessors were blinded to their sessment of Infants for BSID scores at as administered during year 2 (17 to 22
previous scores. 1 and 2 years of age.7 months of age), were calculated. The

251
HARRIS AND DANIELS THE JOURNAL OF PEDIATRICS
AUGUST 2001

predictive relationship of the HINT to with each of the BSID-II scales exceed cepted method for examining the
the BSID-II Mental Scale was r = the minimally acceptable level of 0.60, HINT’s predictive validity.29 Because
–0.11. Our hypothesis was not sup- according to Meyer.28 we currently have data only on high-risk
ported because the correlation coeffi- In light of the fact that the HINT is infants, it would be premature to con-
cient was <0.30. In contrast, however, quick, inexpensive, and easy to adminis- duct classification analyses at this point.
the predictive relationship between the ter, it is encouraging that this screening The appropriateness of the HINT to
HINT and the BSID-II Motor Scale test provides reasonably comparable in- the population under consideration was
was r = –0.49 (P < .01). formation to the more time-intensive and also evaluated in the content validity
costly scale. However, one important dif- study.19 Although 100% of the experts
ference in the studies reported in the rated the HINT as average or above av-
DISCUSSION BSID-II manual, as compared with the erage in its ability to screen for early
current study, is that the former studies neuromotor delays, only 26.3% gave
The inter-rater, test-retest, and intra- used index scores, whereas the present similar ratings when asked about the
rater reliabilities of the total HINT study used raw scores. Raw scores were HINT’s ability to screen for early cogni-
score were all ≥0.98, well above the ac- preferable for this study because they tive delays. Results of the present study,
ceptable “benchmark” of 0.80 and also represented a form of measurement par- particularly with regard to the HINT’s
exceeding the desirable level of allel to the HINT raw scores. ability to predict later cognitive perfor-
0.90.3,5,23 Of concern, however, were The predictive validity of the early mance, corroborate these opinions.
the low inter-rater and test-retest relia- HINT (3-12 months) compared with The final two criteria for evaluating
bilities of 2 items: eye muscle control the later BSID-II Motor Scale (17-22 screening tests are reliability and valid-
and stereotypical behaviors. These low months), although relatively modest (r ity.4 The HINT has strong inter-rater,
coefficients were due, in part, to the = –0.49, P < .01), is stronger than the test-retest, and intra-rater reliability, as
fact that these behaviors were seldom relationship between the BSID-II well as strong concurrent validity with
observed among the infants in our Motor Scale administered during the the BSID-II. The HINT’s predictive
sample. Consequently, these items will first year of life and that same scale ad- validity for later motor performance,
most likely be omitted from the final ministered during the infant’s second although modest, appears to be compa-
version of the tool. year of life (r = 0.34, P < .05). Although rable to or stronger than that of other
The moderately strong and signifi- the early HINT accounted for nearly infant neuromotor tests administered
cant relationship between scores on one quarter (24%) of the BSID-II during the first year of life.
the HINT and the BSID-II Mental Motor outcome during year 2, the We thank Dr Jonathan Berkowitz for his su-
Scale suggests that the HINT may be early BSID-II Motor Scale accounted perb guidance and help in data analysis and
tapping early cognitive behaviors, at only for 11.6% of the later BSID-II Kate Junaid, Lynn Rogers, and Liisa Hohlsti
least as they are assessed during the Motor outcome. for their invaluable assistance in data collec-
tion. We also sincerely thank the families and
first year of life. The concurrent validi- The fact that the HINT was superior infants who took part in this study.
ty between the HINT and the BSID- to the BSID-II Motor Scale at predict-
II Motor Scale (r = 0.89) is somewhat ing later motor development on the
lower than the correlation coefficient BSID-II Motor Scale speaks to its fu- REFERENCES
reported by Piper and Darrah24 be- ture utility as a screening test, especial- 1. Cummins SK. Prenatal and perinatal
tween the Alberta Infant Motor Scale ly for infants at risk for neuromotor factors and the epidemiology of cere-
and the first edition of the BSID17 for difficulties. bral palsy. Phys Med Rehabil 1991;5:
403-15.
their sample of 48 at-risk infants (r = After collection of normative data on 2. Kirshner B, Guyatt GH. A method-
0.98). The concurrent validity of the 300 healthy Canadian infants (2000- ologic framework for assessing health
Alberta Infant Motor Scale with the 2002), including BSID-II data on 60 indices. J Chron Dis 1985;38:27-36.
BSID Mental Scale was not evaluated, infants during the first year and at ap- 3. Bayley N. Bayley Scales of Infant De-
so it is impossible to make comparisons proximately 18 months, classification velopment. 2nd ed. San Antonio (TX):
Psychological Corporation; 1993.
with results from the present study. analyses will be conducted to examine 4. Stengler SR, Huber CJ, Routh DK.
The concurrent validity of the the sensitivity, specificity, and positive Screening growth and development of
BSID-II Mental Scale with the HINT and negative predictive values of the preschool children: a guide for test se-
compares favorably with relationships HINT compared with the later BSID-II lection. New York: McGraw-Hill;
between the BSID-II and other scores. Data from the 60 healthy infants 1980. p. 34-60.
5. Chandler LS, Andrews MS, Swanson
well-known, standardized assessment will be combined with those from the 54 MW. Movement assessment of infants:
tools.25-27 Furthermore, the HINT high-risk infants in the present study. a manual. Rolling Bay (WA): Authors;
concurrent validity values compared This will provide another, widely ac- 1980.

252
THE JOURNAL OF PEDIATRICS HARRIS AND DANIELS
VOLUME 139, NUMBER 2

6. Harris SR, Haley SM, Tada WL, tion of infants at risk for cerebral palsy. bility analysis in therapeutic research:
Swanson MW. Reliability of observa- Dev Med Child Neurol 1981;23:705-16. practice and procedures. Am J Occup
tional measures of the Movement As- 14. Harris SR. Movement analysis: an aid Ther 1993;47:10-6.
sessment of Infants. Phys Ther 1984; to early diagnosis of cerebral palsy. 23. Anastasi A. Psychological testing. New
64:471-7. Phys Ther 1991;71:215-21. York: MacMillan Publishing Compa-
7. Harris SR, Swanson MW, Andrews 15. Touwen BCL, Hadders-Algra M. Hy- ny; 1988.
MS, Sells CJ, Robinson NM, Bennett perextension of neck and trunk and 24. Piper MC, Darrah J. Motor assess-
FC, et al. Predictive validity of the shoulder retraction in infancy: a prog- ment of the developing infant.
Movement Assessment of Infants. J nostic study. Neuropediatrics 1983;14: Philadelphia: WB Saunders; 1994.
Dev Behav Pediatr 1984;5:336-42. 202-5. 25. McCarthy D. McCarthy Scales of
8. Haley SM, Harris SR, Tada WL, 16. Harris SR. Development of an infant Children’s Abilities. San Antonio: The
Swanson MW. Item reliability of the neuromotor assessment tool. Mary E. Psychological Corporation; 1972.
Movement Assessment of Infants. Switzer Research Fellowship final re- 26. Wechsler D. Wechsler Preschool and
Phys Occup Ther Pediatr 1986;6: port. Washington (DC): National In- Primary Scale of Intelligence–Revised.
21-39. stitute on Disability and Rehabilitation San Antonio: The Psychological Cor-
9. Harris SR. Early predictors of cerebral Research; 1991. poration; 1989.
palsy in low birthweight infants. Dev 17. Harris SR. Parents’ and caregivers’ per- 27. Elliott CD. Differential ability scales:
Med Child Neurol 1987;29:508-19. ceptions of their children’s development. administration and scoring manual.
10. Harris SR. Early detection of cerebral Dev Med Child Neurol 1994;36:918-23. San Antonio: The Psychological Cor-
palsy: sensitivity and specificity of two 18. Bayley N. Bayley Scales of Infant De- poration; 1990.
motor assessment tools. J Perinatol velopment. New York: Psychological 28. Meyer CR. Measurement in physical
1987;7:11-5. Corporation; 1969. education. New York: The Ronald
11. Harris SR. Early diagnosis of spastic 19. Harris SR, Daniels LE. Content valid- Press Co; 1974. p. 86-9. Cited by: Pal-
diplegia, spastic hemiplegia, and quad- ity of the Harris Infant Neuromotor isano RJ. Concurrent and predictive
riplegia. Am J Dis Child 1989;143: Test. Phys Ther 1996;76:727-37. validities of the Bayley Motor Scale and
1356-60. 20. Bartko J. The intraclass correlation the Peabody Developmental Motor
12. Amiel-Tison C, Korobkin R, Esque- coefficient as a measure of reliability. Scales. Phys Ther 1986;66:1714-9.
vaucouloux M. Extensor hypertonia: a Psychol Rep 1966;19:3-11. 29. Sackett DL, Haynes RB, Guyatt GH,
clinical sign of insult to the central ner- 21. Shrout PE, Fleiss JL. Intraclass corre- Tugwell P. Clinical epidemiology: a
vous system of the newborn. Early lation: uses in assessing rater reliabili- basic science for clinical medicine.
Hum Dev 1977;1/2:188-90. ty. Psychol Bull 1979;86:420-8. 2nd ed. Boston: Little, Brown & Co;
13. Ellenberg J, Nelson K. Early recogni- 22. Ottenbacher KJ, Tomchek SD. Relia- 1991.

253

You might also like