Concurrent and construct validity of the Pediatric Evaluation of Disability Inventory.

A major component of a comprehensive pediatric evaluation is the assessment of functional abilities. Development of standardized functional assessments for young children is needed to enhance the quality of physical therapy evaluation of young children. (1,2) The purpose of this study was to determine the concurrent and construct validity of a new instrument for the assessment of pediatric functional ability, the Pediatric Evaluation of Disability Inventory (PEDI).

Functional assessment instruments measure the actual ability of the child to perform necessary daily activities. (3-5) The focus of functional assessments is to determine the extent of independence and the maximization of function achieved within the limitation imposed by physical or cognitive deficits. (6) Comprehensive functional assessment instruments contain mobility, transfer, self-care, and social function items (5); include measurement dimensions of assistance and adaptive equipment (4-7); and incorporate developmental stages of functional skill attainment. (8) Functional assessment may provide the pediatric physical therapist with essential information for treatment and educational planning. (9-10)

Despite the agreement that functional assessments for infants and young children are valuable, few standardized instruments have been developed for this age group. Physical therapists have often relied on developmental milestone scales for indications of functional status in young children. However, inadequate sampling of functional and adaptive content, (5) deficiencies in standardization on disabled subject samples, (6-11) and lack of sensitivity to functional change (12-15) limit the usefulness of developmental milestone scales for functional assessment. A number of other adaptive instruments are available for examining functional status, but are diagnosis specific, (16) known to be clinically useful but not standardize, (17) modified from adult functional assessment instruments, (10) or developed primarily for social and adaptive skill evaluations of older children. (18,19) Recently, new instruments have become available for the functional assessment of young children, (20,21) although conclusive technical data supporting their clinical use have not been reported.

The PEDI is a new functional assessment instrument for chronically ill and disabled children from 6 months through 7 years of age. Item content includes self-care, bowel and bladder control, mobility and transfers, communication, and social function. The PEDI was developed to 1) provide a global evaluation tool for inpatient pediatric rehabilitation programs; 2) serve as an evaluation instrument for outpatient therapy services, school programs, and community agencies serving pediatric clients; and 3) provide a uniform mechanism for reporting functional disability for data registries and health policy data banks. The PEDI is designed to identify the child's functional ability along three scales: 1) typical functional skill level, 2) modifications or adaptive equipment used (ie, braces, motorized wheelchair), and 3) physical assistance typically required of the care-giver.

With the development of any new assessment tool, validity must be determined. Validity is investigated during test development and confirmed through subsequent use. (22) Although all forms of validity are important in determining a test's usefulness, two forms of criterion-related validity will be addressed in this article: 1) concurrent validity and 2) construct validity. Concurrent validity refers to the similarity between the target measure and another measure for which validity is known. (23) Concurrent validity is obtained by correlating two or more measures given to the same subjects at approximately the same time. Construct validity refers to accumulated evidence that a test performs as expected when measuring an underlying trait or concept. In this study, we examined the ability of the PEDI to discriminate between healthy children and a group of children it was designed to assess (children with functional disabilities). (22)

The Battelle Developmental Inventory Screening Test (BDIST) (24) was chosen as a comparison measure to the PEDI. The BDIST includes items from five domains: 1) personal-social, 2) adaptive, 3) motor, 4) communication, and 5) cognitive. Although the BDIST has a developmental framework, it samples essential functional content such as dressing, toileting, and mobility. The BDIST is appropriate to use in this study as the PEDI was designed to assess similar adaptive content areas. In contrast to the PEDI, the BDIST assesses the child's skill level and does not include a scale that takes into account the need for adaptive equipment or physical assistance. The purposes of this study were to examine the concurrent validity of the PEDI with an already validated test instrument, the BDIST, in groups of disabled and nondisabled children and to determine the construct validity of the PEDI by examining its ability to discriminate between groups. Specifically, we sought to determine the concurrent correlations between the PEDI and the BDIST scores. Furthermore, we sought to determine whether the PEDI outcome scores would be significantly different between the disabled and nondisabled groups and, if so, which test instrument best classifies the children according to group status.

Method

Subjects

The subjects were 40 children between the ages of 2 and 8 years whose scores on the cognitive domain of the BDIST were no more than 1.50 standard deviations below the mean for their age group. This exclusion criterion eliminated from the sample children with major cognitive impairments. The disabled group (n = 20) included 10 children (7 female, 3 male) with arthritic conditions who were recruited from the Rheumatology Clinic at The Floating Hospital, New England Medical Center (Boston, Mass). The other 10 children (4 female, 6 male) in this group had spina bifida; 9 of these subjects were recruited from the Birth Defects Clinic at The Floating Hospital, and 1 subject was recruited through a Boston-area special needs preschool. The nondisabled group (n = 20) consisted of children with no identified illness or developmental delay who were selected by convenience from the community and were matched for age, sex, and cognitive status with the disabled group subjects. Informed consent was obtained from all subjects' parents prior to data collection.

Varying levels of functional disability were represented in the clinical sample (Tab. 1). Ninety-five percent of the subjects in each group were Caucasian. There were no significant differences

[TABULAR DATA OMITTED]

between the groups in age or in cognitive domain age-equivalent scores on the BDIST (Tab. 1). Families of the disabled group subjects scored significantly lower on socio-economic status than families of the nondisabled group subjects (AB Hollingshead, unpublished manuscript, 1985) (t = 3.62, df = 18, P < .005). As expected, total BDIST age-equivalent scores (t = -4.06, df = 19, P < .001) and BDIST gross motor domain age-equivalent scores (t = -3.83, df = 19, P < .005) were significantly lower in the disabled group than in the nondisabled group.

Test instruments

The PEDI is a parental-report, or structured-interview, instrument used by pediatric physical therapists and other rehabilitation professionals to assess functional abilities of young children. The current version of the PEDI is known as the Development Edition (pilot version). The creation of the Development Edition provides the opportunity to assess initial reliability and validity data prior to formal standardization of the instrument. A formal content-validity study including feedback from over 30 experts in the fields of pediatric rehabilitation and test development has been completed (SM Haley, R Faas, WJ Coster; unpublished data; 1990). Reliability studies are currently underway. Standardization on a normative sample is planned after analyses of the data from the reliability and validity studies of the Development Edition. This preliminary validity study is intended to exaine the usefulness of the PEDI for describing functional deficits in children with disabilities and for examining its ability to discriminate between groups of disabled and non-disabled children.

The items on the PEDI are grouped into three domains: self-care, mobility, and social function (Tab. 3). For each domain, three independent scale scores are calculated: 1) functional skill level, 2) caregiver assistance, and 3) modifications. Total scores are also calculated for each scale caross domains. In the Development Edition, as for the current analyses, scroes are obtained by summing items within domains for each measurement scale. Potential exists for the misinterpretation of scores when ordinal data are added to create summary scores. (25) However, with the lack of a normative sample and with no hierarchical model yet defined, no other summary score method is currently available. Higher

[TABULAR DATA OMITTED]

scores for functional skill level and caregiver assistance indicate better performance and increased independence. Higher scores for the modifications scale denote more extensive modifications are used to perform the skill. No formal reliability checks were conducted on the PEDI items; however, the primary author (ABF) did observe and conduct a number of structured interviews supervised by the developers of the PEDI prior to data collection.

The BDIST was developed from the Battelle Developmental Inventory (BDI), (24) which consists of 341 test items from five domains: 1) personal-social, 2) adaptive, 3) motor, 4) communication, and 5) cognitive. Excellent technical data have been reported for the BDI, including the ability to discriminate between disabled and nondisabled children. (26-28) Major advantages of the BDI over other available developmental tests are the inclusion of relevant functional and adaptive test items and procedures for evaluating children with motor, visual, and hearing impairments. (6) Although the BDI is not commonly used by physical therapists, a recent article in the physical therapy literature related the use of the BDI to examine different delivery models of physical therapy and occupational therapy in a school setting. (29) Limitations of the BDI are its large item pool and the estimated amount of time required to complete the test (2 hours). Because of practical time limitations of administering two concurrent tests, the BDIST was used in this study.

The BDIST consists of 96 items, with a completion time of approximately 20 to 35 minutes (Tab. 3). The BDIST covers the same domains as the BDI and was validated clinically with a sample of healthy children (N = 164), thus making the BDIST a strong predictor of performance on the BDI. The stated purposes of the BDIST are 1) to identify children who are handicapped or developmentally delayed and areas for more comprehensive testing, 2) to identify strengths and weaknesses of normally developing children, and 3) to document the progress of groups of children with disabilities. (24) Similar administration instructions as well as similar scoring procedures to the BDI were used in this study. Items on the BDIST can be scored by structured administration, observation, or caregiver interview.

Interrater reliability of BDIST scores was established prior to data collection. Six children, aged 4 to 6 years, with suspected developmental delay were simultaneously tested by the primary author and an experienced user of the BDIST. The intraclass correlation coefficients (ICC[2,1]) for the BDIST total raw score and domain scores were all above .90. The standard error of measurement of the BDIST total raw score was less than .80; standard errors of measurement of domain scores were proportionally small. Intrarater reliability of the BDIST scores was estimated by testing

Table 3. Domain Content of the Pediatric Evaluation of Disability Inventory (PEDI) and the Battelle Developmental Inventory Screening Test (BDIST)

PEDI Item Content           BDIST Item Content
Self-care domain            Adaptive domain
 Feeding                     Eating
 Grooming                    Attention
 Bathing                     Dressing
 Dressing upper body         Toileting
 Dressing lower body         Personal responsibility
 Bladder management
 Bowel management
 Toileting
 Independent routines
Mobility domain             Motor domain
 Tub transfer                Body coordination/milestones
 Chair transfer              Perceptual motor
 Bed transfer                Locomotion
 Locomotion within a room    Fine motor
 Locomotion between rooms
 Locomotion outdoors
 Stair-climbing ability
Social function domain      Communication domain
 Comprehension               Receptive
 Expression                  Expressive
                            Personal-social domain
 Social interaction          Self-concept
 Play with peers             Adult interaction
                             Peer interaction
                             Coping
                             Social role
                             Expression of feelings
                            Cognitive domain
                             Memory
                             Perspective discrimination
                             Reasoning and academic skills
                             Conceptual development

three healthy children, aged 2, 4, and 6 years, on two test occasions within a 2-week interval. All tests were administered by the primary author. The Pearson product-moment correlation coefficient for intrarater agreement was high (r = .96), and the standard error of measurement was within two points for the total score and proportionally small for the domain scores.

Procedure

The PEDI was administered as a parental-report questionnaire to the parents of all children in the study. Because of time limitations during inpatient clinic visits, the parents of 10 of the subjects took the PEDI home to complete. They were called within 2 to 4 days by the primary author to answer any questions and were asked to return the questionnaire by mail. The parents of the remaining 30 subjects completed the PEDI while the primary author was present, either at the clinic or in the child's home. Questions regarding the PEDI were answered according to instructions and samples provided by the test developers. All parents completed a demographic form as part of the PEDI questionnaire. The BDIST was administered to all of the children according to the standardized procedures in the test manual. The BDIST was administered to the 10 children with arthritic conditions and to 2 of the children with spina bifida while they were waiting at the clinic to be seen by various health care professionals; it was administered to the remaining children with spina bifida and to the 20 nondisabled children at each child's home. The difference in test site administration occurred because of time limitations in the clinic and convenience to the families.

Data Analysis

First, Pearson product-moment correlations to estimate concurrent validity were calculated using summary scores of the PEDI and BDIST for the total group of children (N = 40), for the disabled group (n = 20), and for the nondisabled group (n = 20). Domains on the PEDI were related to domains on the BDIST that were thought to assess similar areas. Second, Student's paired t tests were used to determine whether significant differences in PEDI scores existed between groups. Finally, a discriminant analysis was performed with the following independent variables: PEDI functional skill level, PEDI modifications, PEDI caregiver assistance, and BDIST total score to determine the relative power of each scale to differentiate between the groups.

Results

Concurrent Validity

Correlations were moderately high and positive for both the PEDI functional skill level summary score and the PEDI caregiver assistance summary score with the BDIST total score

[TABULAR DATA OMITTED]

(Tab. 4). The PEDI self-care domain had a moderately high correlation with the BDIST adaptive domain. The PEDI mobility domain had a moderately high correlation with the BDIST gross motor domain except for the nondisabled group of children. The magnitude of correlations of the PEDI social function domain was variable with the personal-social, cognitive, and communication (expressive and receptive) domains of the BDIST. Overall, with the exception of the PEDI social function domain with the BDIST personal-social domain, these results support the concurrent validity of the PEDI with the BDIST.

Construct Validity

Table 5 presents results from a series of Student's paired t tests using the PEDI summary scale and domain scores. The disabled group scored significantly lower than the nondisabled group in the content domain scores of self-care and mobility and in the total scale scores of functional skill level, caregiver assistance, and modifications. This finding supports the construct validity of the PEDI in that the PEDI is able to effectively discriminate between disabled and nondisabled children.

A discriminant analysis using the four summary scale scores--PEDI functional skill level, PEDI modifications,

[TABULAR DATA OMITTED]

PEDI caregiver assistance, and BDIST total--showed the PEDI modifications score to be the best discriminator between groups (disabled and non-disabled) (Tab. 6). When the modifications scale was removed from the analysis, the PEDI functional skill level

Table 6. Results of Discriminant Analysis to Identify Group Status

                              F       P
Independent variables
  PEDI (a) modifications      13.06   <.001
  PEDI functional skill level  1.28    NS
  PEDI caregiver assistance    0.70    NS
  BDIST (b) total score        0.34    NS
Independent variables
 (with modifications scale
 removed)
  PEDI functional skill level 10.21   <.001
  BDIST total score            0.85    NS
  PEDI caregiver assistance    0.00    NS
(a) PEDI = Pediatric Evaluation of Disability
Inventory.
(b) BDIST = Battelle Development Inventory
Screening Test.

score was the next best discriminator between groups. With the PEDI modifications scale entering the equation first, the group status of 72.5% of the children was correctly identified. Nine of the disabled as nondisabled. One of the nondisabled children was incorrectly identified as disabled. When the PEDI modifications scale was removed from the analysis and the PEDI functional skill level score entered the equation first, the group status of 75% of the children was correctly identified. According to the PEDI functional skill level scale, 8 of the disabled children were incorrectly identified as nondisabled and 2 of the nondisabled children were incorrectly identified as disabled. The ability of the PEDI to discriminate between disabled and nondisabled groups of children, in this case better than the BDIST, further supports its construct validity.

Discussion

The purpose of this study was to assess the validity of the Development Edition of the PEDI by examining the relationship between children's scores on a developmental test with adaptive content and a new functional test.

Overall, the results support the concurrent and construct validity of the PEDI.

Moderately high correlations of PEDI scores with the BDIST scores support the concurrent validity of the PEDI. Although moderately high correlations between the PEDI self-care and mobility domains and the BDIST were demonstrated by the total sample and the disabled group, lower correlations were found for the nondisabled group. This finding may be explained by a ceiling effect in the PEDI as many of the nondisabled children scored maximum scores on the PEDI mobility scales, but did not on the BDIST gross motor domain. Apparently the nondisabled children were independent in the most difficult functional items of the PEDI mobility domain (eg, stair-climbing ability), but were unable to perform the hardest items on the BDIST gross motor domain (ie, jump rope three consecutive times, tandem walk 1.83 m [6 ft], and stand on each foot alternately with eyes closed for 3 seconds).

Similar low correlations were noted with the assistance score in the non-disabled children. A ceiling effect and considerable variability of the nondisabled children's mobility assistance scores may be responsible for these results. The amount of assistance given may depend on the parenting style and time available to the parent to allow the child to complete the task without assistance. Many parents expressed concern about their individual parenting style on the amount of assistance given and about their quickness to give assistance when time is limited. These concerns may not be true for disabled children because allowing the child to complete the task, regardless of time, may be essential in a parent-supported intervention program. For the purposes of this study, however, strong correlations in the mobility domain between tests for the disabled group support the validity of the PEDI with this population.

Construct validity of the PEDI was supported by the results of paired t tests. Group domain scores were found to be significantly different for most of the PEDI scales (Tab. 5). As was expected for the disabled group, the most significant differences in PEDI scores were found in the PEDI mobility and self-care domains. There were no differences between the disabled and nondisabled children on the non-motor-based PEDI scale of social function, suggesting these children were no different in social interaction, comprehension, expression, and play skills. Results of this study are limited to this sample of children with neurologic and musculoskeletal involvement and with mild to no cognitive impairments. A sample of more cognitively or severely physically impaired children may have shown important differences in the PEDI social function domain.

Construct validity was further supported by the results of the discriminant analysis. As expected, the PEDI modifications scale was the strongest independent variable to discriminate between the two groups. This difference occurred primarily because nine of the children with spina bifida and two of the children with arthritic conditions needed modifications. With modifications as the first variable entered, children with disabilities who did not use modifications were incorrectly identified as nondisabled. When the modifications scale was removed from the analysis, the PEDI functional skill level scale continued to be a better predictor than the BDIST. With the PEDI functional skill level scale entering the equation first, eight of the disabled children were incorrectly identified as nondisabled (one child in the spina bifida group [diagnosed with lipomeningocele] and seven of the children with arthritic conditions were relatively high functioning).

The ability of the functional skill level scale to better discriminate between the groups than the BDIST can be explained primarily as a difference in test content. Although some content of each test was similar (eg, stairclimbing ability, performance of household chores, use of buttons, knowledge of address), the PEDI functional skill level content includes more items of basic functional ability, such as locomotion and transfers. These items are particularly important for disabled children and may be important indicators of developmental progress. The inclusion of these relevant functional items make the PEDI a useful tool for pediatric physical therapists.

In addition to having different content, the functional skill level scale of the PEDI has finer increments of progress than the BDIST. This finer skill breakdown allows for more sensitivity to detect meaningful clinical changes. Surprisingly, the caregiver assistance scale did not discriminate as well between groups as did the functional skill level or modifications scale; perhaps with a more impaired sample of children, it would act as a stronger discriminant variable. In this study, the discrepancy in caregiver assistance became greater between groups as age increased and as parental expectation for independence became greater.

In general, the disabled and nondisabled groups were well matched in terms of age, sex, and cognitive status. However, the uncontrolled difference in family socioeconomic status between the groups may have contributed to differences in functioning between the groups. Caution must be taken not to generalize the results beyond this sample of children with physical disabilities (arthritic conditions and spina bifida). Future work with the PEDI should include a greater spectrum of disability types and ranges of cognitive and social competence.

It should also be noted that comparisons were made from scores from an administered test with scores from a parental-report questionnaire. The BDIST consisted of professional observation, structured administration, and parent-interview items, whereas the PEDI consisted solely of parental-response items. Several authors [30,31] have reported that parents give higher average scores on their child's developmental achievements than professionals. Therefore, because the PEDI was administered solely by parental report in this study, the scores may be higher than if therapists or teachers were to observe and score the child. All PEDIs were administered by parental report; therefore, no systematic bias is likely to have occurred.

Clinical Implications

Pediatric physical therapists should consider the use of a variety of standardized tests in their clinical practice. The value of the assessment of developmental milestones in an evaluation cannot be overlooked. Often, the inability to perform a milestone skill, such as balance on one leg for 6 seconds, affects the child's ability to perform a functional task, such as climb into the bathtub. However, assessment for treatment planning or for the evaluation of therapy effectiveness for children with disabilities should not be based solely on developmental skills. The content of a physical therapy assessment should be relevant to the educational and treatment goals of the child. Measuring progress in terms of a child's functional status is often in greater accordance with the physical therapist's and the family's goals than discrete skills. The physical therapy intervention should ultimately result in movement changes that lead to independence in functional daily activities, which may or may not include achievement of developmental milestones. [32] The results of this study suggest the PEDI can be a useful standardized assessment for the description of functional status in young children.

Conclusion

This study provides preliminary evidence for concurrent and construct validity of the Development Edition of the PEDI. Moderately high correlations were found with the BDIST for the PEDI self-care and mobility domains and for the PEDI total scale scores of functional skill level, caregiver assistance, and modifications. The PEDI modifications and functional skill level scales were better discriminators of group status than the BDIST. The results support the impetus for further development and standardization of the final version of the PEDI.

Acknowledgments

We acknowledge the Rheumatology Clinic and the Birth Defects Clinic at New England Medical Center, Boston, Mass, and the Anne Sullivan Center in Tewksbury, Mass, for their assistance in subject recruitment. Appreciation is also extended to Ruth Faas for her help in preparing the PEDI and to Scott Osberg for his technical assistance in data reduction and analysis.

References

[1] Campbell SK. Assessment of the child with CNS dysfunction. In: Rothstein JM, ed. Measurement in Physical Therapy. New York, NY: Churchill Livingstone Inc; 1985:207-228.

[2] Campbell SK. Measurement in developmental therapy: past, present and future. Physical and Occupational Therapy in Pediatrics. 1989;9:1-13.

[3] Fuhrer MJ. Overview of outcome analysis in rehabilitation. In: Fuhrer MJ, ed. Rehabilitation Outcomes: Analysis and Measurement. Baltimore, Md: Paul H Brookes Publishing Co; 1987:1-15.

[4] Frey WD. Functional outcome: assessment and evaluation. In: DeLisa JA, ed. Rehabilitation Medicine: Principles and Practice. Philadelphia, Pa: JB Lippincott Co; 1988:158-172.

[5] Haley SM, Hallenborg SC, Gans BM. Functional assessment in young children with neurological impairments. Topics in Early Childhood Special Education. 1989;9:106-126.

[6] Allen D. Measuring Rehabilitation Outcomes for Infants and Young Children: A Family Approach. Baltimore, Md: Paul H Brookes Publishing Co; 1987.

[7] Kaufert JM. Functional ability indices: measurement problems in assessing their validity. Arch Phys Med Rehabil. 1983;64:260-267.

[8] Christiansen CH, Schwartz RK, Barnes KJ. Self-care: evaluation and management. In: DeLisa JA, ed. Rehabilitation Medicine: Principles and Practice. Philadelphia, Pa: JB Lippincott Co; 1988:95-115.

[9] Harris SR. Early intervention: Does developmental therapy make a difference? Topics in Early Childhood Special Education. 1988; 7:20-32.

[10] Law M, Usher P. Validation of the Klein-Bell Activities of Daily Living Scale for children. Canadian Journal of Occupational Therapy. 1988;55:63-68.

[11] Fuchs D, Fuchs L, Benowitz S, Barringer K. Norm-referenced tests: Are they valid for use with handicapped students? Except Child. 1987;54:263-271.

[12] Garwood SG. (Mis)use of developmental scales in program evaluation. Topics in Early Childhood Special Education. 1982;1:61-69.

[13] Montgomery PC, Connolly BH. Norm-referenced and criterion-referenced tests: use in pediatrics and application to task analysis of motor skill. Phys Ther. 1987;67:1873-1876.

[14] Shonkoff JP. The limitations of normative assessments of high-risk infants. Topics in Early Childhood Special Education. 1983; 3:29-43.

[15] Cole KN, Swisher MW, Thompson MD, Fewell RR. Enhancing sensitivity of assessment instruments for children: graded multidimensional scoring. Journal of the Association for Persons with Severe Handicaps. 1985; 10:209-213.

[16] Sousa JC, Telzrow RW, Holm RA, et al. Developmental guidelines for children with myelodysplasia. Phys Ther. 1983;63:21-29.

[17] Coley IL. Pediatric Assessment of Self-Care Activities. St Louis, MO: CV Mosby Co; 1978.

[18] Sparrow SS, Ballo DA, Cicchetti DV. Vineland Adaptive Behavior Scales. Circle Pines, Minn: American Guidance Service; 1984.

[19] Bruininks RH, Woodcock RW, Weatherman RF, Hill BK. Scales of Independent Behavior. Allen, Tex: DLM Teaching Resources; 1984.

[20] Granger CV, Hamilton BB, Kayton R. Guide for the Use of the Functional Independence Measure for Children (Wee-FIM) of the Uniform Data Set for Medical Rehabilitation. Buffalo, NY: Research Foundation, State University of New York; 1988.

[21] Russell DJ, Rosenbaum PL, Cadman DT, et al. The Gross Motor Function Measure: a means to evaluate the effects of physical therapy. Dev Med Child Neurol. 1989;31:341-352.

[22] Dunn WW. Validity. Physical and Occupational Therapy in Pediatrics. 1989;9:149-168.

[23] Payton OD. Reliability and validity. In: Research: The Validation of Clinical Practice. 2nd ed. Philadelphia, Pa: FA Davis Co; 1988:67-81.

[24] Newborg J, Strock J, Wnek L. Battelle Developmental Inventory. Allen, Tex: DLM Teaching Resources; 1984.

[25] Merbitz C, Morris J, Grip JC. Ordinal scales and foundations of misinference. Arch Phys Med Rehabil. 1989;70:308-312.

[26] Guidubaldi J, Perry J. Concurrent and predictive validity of the Battelle Developmental Inventory at the first grade level. Educational and Psychological Measurement. 1984; 44:977-985.

[27] Mott S. Concurrent validity of the Battelle Developmental Inventory for speech- and language-disordered children. Psychology in the Schools. 1987;24-215:220.

[28] McLean M, McCormick K, Bruder MB, Burdg NB. An investigation of the validity and reliability of the Battelle Developmental Inventory with a population of children younger than 30 months with identifying handicapping conditions. Journal for the Division for Early Childhood. 1987;11:238-245.

[29] Cole KN, Harris Sr, Eland SF, Mills PE. Comparison of two service delivery models: in-class and out-of-class therapy approaches. Pediatric Physical Therapy. Summer 1989; 1:49-54.

[30] Gradel K, Thompson MS, Sheehan R. Parental and professional agreement in early childhood assessment. Topics in Early Childhood Special Education. 1981;1:31-39.

[31] Stancin T, Reuter J, Dunn V, Bickett L. Validity of caregiver information on the developmental status of severely brain-damaged young children. Am J Ment Defic. 1984; 4:388-395.

[32] Campbell PH, Stewart B. Measuring changes in movement skills with infants and young children with handicaps. Journal of the Association for the Severely and Profoundly Handicapped. 1986;11:153-161.

A Feldman, MS, PT, is a level-II therapist, Children's Hospital of Los Angeles, 4650 Sunset Blvd, PO Box 54700, Los Angeles, CA 90054-0700. She was a student in the master's degree program, Department of Physical Therapy, Sargent College of Allied Health Professions, Boston University, when this study was completed in partial fulfillment of her degree requirements. Address all correspondence to Ms Feldman at 11640 Kiowa Ave, #206, Los Angeles, CA 90049 (USA).

S Haley, PhD, PT, is Assistant Professor, Tufts University School of Medicine, and Acting Director, Research and Training Center in Rehabilitation and Childhood Trauma, New England Medical Center Hospitals, Boston, MA 02111.

J Coryell, PhD, PT, was Associate Professor, Department of Physical Therapy, Sargent College of Allied Health Professions, Boston University, 1 University Rd, Boston, MA 02215, when this study was conducted.

This study was supported in part by a grant from the US Department of Education, Office of Special Education and Rehabilitation Services, awarded to Boston University; a Clinical Research Grant from the American Physical Therapy Association, Section on Pediatrics, awarded to Ms Feldman; and Grant No. H133G80043 from the US Department of Education, National Institute on Disability and Rehabilitation Research, awarded to Dr Haley.

This study was approved by the New England Medical Center's Human Investigation Review Committee.