Establishing minimal clinically important differences for scores on the pediatric evaluation of disability inventory for inpatient rehabilitation.

Physical therapists are increasingly directing their interventions at achieving changes in patient or client function. (1) Changes in function during an episode of physical therapy intervention are often characterized by changes on a functional assessment instruments. (2,3) Mean change scores from a functional assessment instrument may give statistically significant results, particularly if large samples are involved, but the interpretation of the clinical meaningfulness of the change may not be known. One approach in which the meaningfulness of change has been examined is by comparing change scores on an instrument on serial testing with an external standard. (4) In our study, we used this criterion-based approach in determining the meaningfulness of clinical changes as indicated by scores on the Pediatric Evaluation of Disability inventory (PEDI) (5) during inpatient rehabilitation by using clinicians' opinions as the external standard. The PEDI (5) is an assessment instrument that was designed to measure functional status in children and youth between the ages of 6 months and 7.5 years in the 3 domains: Self-care, Mobility, and Social Function. Numerous studies have examined the sensitivity of the PEDI to identify change during recovery from injury, (6,7) before and after an intervention, (8,10) and from hospital admission to discharge. (11-15) We use the term "sensitivity" in this context as the ability of a measure to be used to identify change, regardless of whether the change is meaningful. (16) Sensitivity can be estimated by various statistical measures based on group means and standard deviations, (7-20) and it is considered a necessary, but insufficient, condition to establish responsiveness of a measure. Responsiveness to change is the ability of an instrument to measure a meaningful or clinically important change. (16) Data correlating standardized test changes on the PEDI to perceptions of change in patient or client status provided by respondents such as the patient, caregiver, or health care provider are not available for inpatient care. Based on clinical experience, the developers of the PEDI have suggested that a change score of approximately 10 points on its 100-point scale for any of the PEDI scales may have clinical relevance, (5) but no studies have been conducted to validate this contention. The primary purpose of this study was to identify the minimal clinically important difference (MCID) of the PEDI during inpatient rehabilitation using clinician's ratings as a reference.

Jaeschke et al (21) described a "clinically important" change or clinically important difference as the difference in patient function that is perceived as beneficial and that would change the patient's management. They referred to the "minimal clinically important difference" as the smallest change in what is measured that is considered to be worthwhile or important by a respondent. Different respondents such as the clinician, parent or caregiver, payer, or child, if able, all can provide useful information about importance of changes. (16,22) A clinically meaningful improvement may relate to changes that help a child return home safely or changes that alter the focus of the child's intervention and management plan. (16,22) In our study, we chose to use clinicians' ratings as the external standard for the MCID. We used clinicians' reports of functional improvement as the criterion to characterize clinical significance of PEDI change scores. We found it more practical to obtain external reports from clinicians than from other respondents (eg, parents, children). Our choice of using clinicians' opinions (rather than opinions of independent raters) as external criteria also was driven by our interest in determining whether clinicians could feasibly collect MCID information on their own patients for the purposes of improvement of quality of data collection in future prospective clinical documentation.

Methods of determining the MCID from clinicians' ratings may affect the MCID scores. (4) In a study involving the development of a new clinical performance measure for children with brain injury, Bedell et al (23) used a visual analog scale (VAS) to provide an external criterion for ratings from clinicians. The VAS has gained wide application in the measurement of subjective phenomena such as pain (24) and mood states (25) and more recently in the measurement of patient satisfaction (26) and the perception of change in the patient's level of function with intervention. (21) In contrast, in a number of studies in which researchers have examined MCID in scores obtained with health-related quality-of-life instruments, Likert scales were used. (21,22,27,28) These studies, we believe, have provided data that support the utility, of Likert scales in determining both patients' and clinicians' perceptions of the importance of clinical changes. Thus, a secondary purpose of our study was to determine whether we could obtain similar MCIDs using either a Likert scale or a VAS. The, objectives of this study were: (1) to identify change scores on the PEDI during inpatient rehabilitation that corresponded to minimal changes in patient function considered to be important by clinicians and (2) to identify the extent of correspondence between clinicians' ratings of change in patient status on 2 rating scales (a 15-point Likert scale and a 15-cm VAS).

Method

Subjects

Fifty-three children and adolescents (mean age = 9.7 years, SD = 4.5, range = 1.8-19.3), accounting for 55 admission-discharge episodes in the inpatient rehabilitation program at Franciscan Children's Hospital & Rehabilitation Center (FCH) in Boston, Mass, within the previous 18 months (range = 1.4-18.8 months, episodes listed from most recent admission to most distant admission), were included in this study. Children and adolescents between the ages of 1 and 22 years with one or more complete admission and discharge PEDI domain scores were eligible for inclusion in the study. Of the 55 eligible episodes, 53 were complete for the Self-care domain on the Functional Skills scale of the PEDI, 52 were complete for the Self-care domain on the Caregiver Assistance scale of the PEDI and for the Mobility domain on the Functional Skills and Caregiver Assistance scales, and 37 were complete for the Social Function domain on the Functional Skills and Caregiver Assistance scales. Patient characteristics are shown in Table 1. Patient diagnoses included a wide range of pediatric illnesses and injuries, such as traumatic brain injury, spinal cord injury, lower-extremity trauma, nontraumatic brain injury, and congenital and developmental conditions.

Instrumentation

The PEDI is designed to provide measurements that represent function and level of independence in the areas of self-care, mobility, and social function. (5) The PEDI measures capability and performance using 2 different scales. The Functional Skills scales characterize a child's capability to perform basic functional activities that are part of important daily skills. The Caregiver Assistance scales characterize the amount of help required from others during complex functional activities. The Rasch rating scale methodology (29,30) was applied during the construction of the PEDI domain scales, yielding hierarchical, unidimensional, and linear scales for each dimension.

A series of studies over the past decade have shown the reliability and validity of data obtained with the PEDI.(10,31-37) Nichols and Case-Smith (33) examined the test-retest reliability and the interrater reliability of data obtained with the PEDI in a sample of children with varied diagnoses. Intraclass correlation coefficients (ICCs) for test-retest reliability based on parent responses on the PEDI for 23 children on 2 occasions separated by an interval of 1 week ranged from .67 to 1.00 for all of the domains of the PEDI, with an ICC of .98 for the summary scores of the Functional Skills scales in each of the 3 domains. The ICCs for test-retest reliability for the Caregiver Assistance scales ranged from .63 to .98 for scores on all of the domains, with ICCs ranging from .85 to .92 for the summary scores in each domain. The ICCs for interrater reliability examined by analyzing responses obtained from the parents and the rehabilitation professionals of 17 participants in interviews separated by 48 hours ranged from .20 to .93 on the Functional Skills scales for the 3 domains and from .15 to .95 on the Caregiver Assistance scales for the 3 domains. Wright and Boschen (34) evaluated responses from parents, teachers, occupational therapists, and physical therapists in their reliability study sampling 40 children between the ages of 3 and 7 years with varying severity of cerebral palsy. They reported ICCs for the Functional Skills scales of the PEDI within each respondent group (ICC>.95 for the total scores and ICC>.80 for all 3 domains) and between respondent groups (ICC = .85 for total scores, ICC = .87 for the Self-care domain, ICC = .81 for the Mobility domain, and ICC = .72 for the Social Function domain).

The PEDI's designers attempted to examine what they described as content validity of data obtained with the PEDI using responses on a questionnaire from a panel of 31 experts in the areas of pediatric rehabilitation. (35) We believe that results of that study indicate strong content validity of data obtained with the PEDI. Feldman et al (36) attempted to establish what they considered construct validity of data obtained with the PEDI by demonstrating superior capability of the PEDI scores to discriminate between a group of children with disabilities (n = 20) and a group of children without disabilities (n = 20) as compared with scores obtained with the Battelle Development Inventor Screening Test (BDIST). We believe that all of these studies taken together provide support for the reliability and content and construct validity of data obtained with the PEDI and for its use as a discriminative instrument.

We used a 15-point Likert scale and a 15-cm VAS to assess clinicians' perceptions of the magnitude of functional change. The 15-point Likert scale ranged from -7 ("a very great deal worse") to +7 ("a very great deal better"), with 0 representing "no change." A similar Likert scale was used reliably (ICC = .96 for interrater reliability, ICC = .95 for test-retest reliability) in a previous study comparing parents' and clinicians' perceptions of change in the motor ability of children with Down syndrome. (38) The 15-cm VAS used in a study by Russell at al (38) was constructed with "a great deal worse" and "a great deal better" representing the ends of the scale and "no change" in the center. Although the 10-cm VAS has been used widely for research and other purposes, Revill et al (24) reported comparable accuracy for ratings based on the 15-cm VAS. In another application of the VAS scale for pain ratings, (24) a high degree of consistency (r = .98) was reported for subjects' repeated ratings based on the 15-cm VAS.

Procedure

Clinicians were recruited for this study from the physical therapy, occupational therapy, and speech-language-hearing departments at FCH. Two of the authors (LVI and HMD) contacted the clinicians who were on staff at FCH during the previous 18 months to explain the purpose of the study and request their participation. Participants were asked to sign an informed consent from affirming voluntary participation prior to the initiation of the study. The clinicians were asked to review demographic data (name, age, sex, admission date, discharge date, and discharge destination) and the inpatient admission and discharge summaries (which were completed by a child's primary therapist in any of the 3 clinical disciplines [physical therapy, occupational therapy, or speech therapy]) from the medical record. They were then asked to rate their perception of change in function between admission and discharge for each child. Copies medical records were modified such that information regarding any change in a patient's functional status and PEDI scores were masked and not made available to the clinician. The use of the medical records was intended to inform the clinician of the child's functional status at admission and discharge and to guide the clinician's decision making for rating the child's functional change over that period on the 2 rating scales (Likert scale and VAS).

The PEDI is routinely used in the physical therapy, occupational therapy, and speech-language-hearing departments at FCH to generate numerical scores that reflect children's functional change from inpatient admission to discharge. The physical therapists, occupational therapists, and speech-language pathologists were asked to rate changes in mobility, self care, and social function because these are the domains of the PEDI for which they are responsible during a child's rehabilitation hospitalization. The therapists rated the changes for the children for whom they were the primary treating therapist at the time of admission. In the event that the child's primary therapist in any of the 3 clinical disciplines (physical therapy, occupational therapy, or speech therapy) was unwilling to participate or was no longer on staff at FCH, the information wars reviewed by another participating clinician representing the same clinical discipline. Ratings on only 7 cases (4.9%) from a total of 142 cases (53 self-care, 52 mobility, and 37 social function) were provided by clinicians other than the primary therapist. Five physical therapists with 5 months to 8.5 years of experience (X = 5.8, SD = 3.17), 6 occupational therapists with 2.5 to 9.5 years of experience (X = 6, SD = 1.94), and 4 speech-language pathologists with 3 to 20 years of experience (X = 10.13, SD = 1.55 participated in this study.

Three sample cases were given to each therapist prior to data collection. For each of the sample cases, ratings of change on the Likert scale and the VAS were agreed on by a consensus of 3 of the authors (LVI, SMH, HMD). The sample cases were used to determine the therapists' readiness to proceed to rate the study cases. For all therapists, median Likert scale rating category discrepancies between the consensus ratings and the individual therapists' ratings on the PEDI scales were between 0 and 2, and the average differences in VAS scores were between 0.13 and 2.1 cm. For both Likert scale and VAS methods, greater discrepancies were observed for the 2 Social Function domain scales than for the Self-care or Mobility domain scales. When large discrepancies (ie, more than a 2.0-cm deviation on the VAS or more than a 2-point rating category, deviation on the Likert scale) were found for a therapist's ratings on the sample cases, additional explanation of the scales and training were conducted prior to formal data collection. Training sessions typically lasted for a half hour and were conducted immediately prior to actual data collection.

Data for the study were collected during 2 sessions, with each therapist's sessions separated by at least 1 week. The first author (LVI) supervised each session. During the initial data collection session, clinicians reviewed the demographic data and admission and discharge summaries for their children and adolescents and rated the amount of change in the area of self-care, mobility, or social function on 1 of the 2 rating scales (Likert scale or VAS). The Likert scale and the VAS are shown in the Appendix. During the second session, clinicians rated the amount of change using the alternate scale. The order of the presentation of the rating scales was random because we believed this would prevent systematic bias.

Data Analysis

For parsimonious data presentation, and in concordance with previous research, (21) Likert scale ratings were organized into 4 categories, such that "no change" and "worse" categories were grouped into "no change," ratings of +1 to +3 represented "minimal clinically important difference," ratings of +4 to +5 represented "moderate change," and ratings of +6 to +7 represented "large change." The VAS ratings also were grouped into 4 categories, with "no change" at either the midpoint (zero) or left of zero ("worse") and with distances to the right of zero proportionally divided into the categories of "minimal change," "moderate change," and "large change." The rating categories of the Likert scale and VAS representing regression or worsening in patient status were collapsed into the "no change" category, because almost the entire sample (99.3%) made positive functional gains. The mean differences between admission and discharge PEDI scores were calculated for the 4 categories. The average PEDI change score corresponding to clinicians' ratings of minimal change on the Likert scale and the VAS served as the estimate for the MCID for each PEDI scale (study objective 1). For study objective 2, the correspondence between scores on the Likert scale (ordinal) and the VAS was determined by the Kendall tau-b ([tau]), a nonparametric correlation statistic, using the Statistical Program tot the Social Sciences, version 9.0. (39) *

Results

Results of the mean PEDI change scores based on ratings for the Likert scale are presented in Table 2. The MCID values of the PEDI based on the clinicians' ratings for the Likert scale ranged from 8.7 to 14.9 (X = 10.9 for the Functional Skills scales, and X = 11.6 for the Caregiver Assistance scales). The MCID values of the PEDI for the VAS (Tab. 3) ranged from 6.0 to 15.6 (X = 11.6 for the Functional Skills scales, and X = 11.2 for the Caregiver Assistance scales).

For the 6 PEDI scales, the clinician ratings between the Likert scale and the VAS had a degree of correspondence of [tau] = .73 to .80. Correlations of [tau] = .43 to .68 ([X.sub.[tau]] = .59) were found between the PEDI change scores for all 6 scales and the Likert scale change categories, and correlations of [tau] = .55 to .67 ([X.sub.[tau]] = .61) were found between the PEDI change scores for all 6 scales and the VAS change categories (Tab. 4).

Discussion

Developing an estimate of MCID is, we believe, a step toward providing meaning to the concept of functional change. Age-based scoring, such as use of age equivalents, or norm-based scoring to depict change may be suitable in certain pediatric diagnoses (eg, developmental delay) in which approximation to normal functioning may be expected. However, for most children with disabilities, the return to age-expected performance is not a realistic goal. Thus, some other means of describing meaningful clinical change is desirable.

The PEDI scaled scores (0-100 metric) have been developed primarily to evaluate changes over time. Scaled scores provide an indication of the number of newly achieved functional skills or new levels of independence that have been achieved. The PEDI scaled scores are transformed from raw scores and have been developed using a one-parameter Rasch model. (30) The Rasch model places PEDI items on unidimensional, hierarchical scales. This means that items are placed on each scale so that each sequential item is expected to be more difficult than the previous item. As Hays and Woolley (4) noted, the Rasch method is valuable for interpreting the MCID because Rasch methods yield interval-level scales. A change of 11 points at a particular location on a PEDI scale, in theory, is similar to a change of 11 points at any position along the scale.

We used the 15-point Likert scale and the 15-cm VAS to reflect clinicians' perceptions of change in patient status during inpatient rehabilitation. Based on previous research, (19) we chose to divide the Likert scale scores into 4 categories, with +1 to +3 representing the MCID. If we defined MCID more conservatively (ie, using only +2 and +3 for the calculation of MCID and removing the category of +1 ["about the same, hardly any better at all"] in the MCID calculations), would our interpretation of MCID change? When the data were reanalyzed to exclude all ratings of +1, the MCID scores for 4 of the 6 PEDI scales were no more than [+ or -]1 point, which we interpret as essentially the same as when the + 1 category was included. However, the estimate of MCID for the Social Function domain scales increased 5 to 6 points when the +1 ratings were excluded. This increase may have occurred because a greater number of +1 ratings were selected for these scales and generally smaller changes were noted in the Social Function domain scales than in other PEDI scales. In addition, these scales had a smaller sample size than the Self-care and Mobility domain scales; thus, the recoding of data was more likely to affect the MCID calculation. Based on the reanalysis of data, only the MCIDs on the Social Function scales would have changed if the MCID criteria oil the Likert scale were changed so as not to include the +1 category, which suggests to us that the MCID estimates as reported on the Self-care and Mobility scales are quite robust.

In general, the data suggest that a scaled score change of approximately 11 points seems to be sufficient to conclude that an important clinical change occurred. From our experiences, we had previously suggested that a scaled score change of 10 points is consistent with a meaningful change. (5) This suggestion was based on the observation that the 95% confidence intervals around many of the scaled scores on the PEDI (other than the extreme scores) are no more than 10 points.

An 11-point change in PEDI scaled scores as it relates to skills of the patient can be best described by examining the item map provided by the one-parameter Rasch analysis of each PEDI scale. (5) The item map describes probable positions of items along a continuum of easy to difficult. These item locations on the map identify approximate difficulty levels for the items along a 0 to 100 scale. For example, if an item is located on the map at a score level of 40 (eg, on the mobility scale, "walks, but holds on to objects or person"), then an 11-point change would indicate that a child would be able to perform the activity described by an item near a score level of 51 (eg, "able to walk 50-100 feet [15-30 m] alone"). This 11-point change as identified by the MCID data can be traced at any point along the item maps of each of the PEDI scales.

Other researchers (20) have suggested that the standard error of measurement (SEM) should be considered a substitute for the MCID. Although this substitution would eliminate the need for an external criterion as we have used in this research, the SEM is based on statistical thresholds and may not reflect a true meaningful change. In the case of the PEDI, the Rasch model provides estimates of precision (standard errors) at each score point along the scale. (30) The MCID (11 points) appears to be in about the same range as the 95% confidence interval (10 points) around most PEDI scaled scores.

Even with sample sizes of around 50, each category of change represents relatively small numbers; thus, variability and outliers can have a strong influence on any of the PEDI change scores within subcategories. As is demonstrated by the overlapping category ranges shown in Tables 2 and 3, the change scores within subcategories have considerable variability in most of the PEDI scales. As an example, the "no change" category of the Mobility domain on the Functional Skills scales of the PEDI (Tabs. 2 and 3) was negative for both rating scales. On close inspection of the raw PEDI data, it became apparent that an emergency discharge of a single child had caused the PEDI change scores and the mean to be skewed in the negative direction. A further examination of outliers is needed in studies involving a larger subject population with greater possibility of variance in scores.

Measuring and interpreting "clinically important change" presents physical therapists with a number of challenges. The magnitude of change that is considered to be minimally important may differ depending on whether the judge is the clinician, the caregiver, or the child or youth (if able). The magnitude of MCID also can be influenced by where the child is placed along the continuum of function (low to high) on admission. Liang (16) has suggested that people who start with low functional status may require a smaller change in function to be rated as having clinically important change as compared with people with higher skill levels who may need to make a larger change in function in order for clinicians to consider it clinically important. This phenomenon may have been present to some extent in our study, because the vast majority of subjects had relatively low mobility scores on admission (partial criterion for admission to inpatient rehabilitation). We noticed that a number of children with low levels of self-care and mobility skills at the time of admission were consistently rated on the rating scales to have made greater progress than that indicated by changes in scores for the Mobility domain scales of the PEDI. Furthermore, children with decreased cognitive function and more limited participation in therapy (occupational therapy and physical therapy) were rated by clinicians to have made lesser progress on the rating scales as compared with children who had functional changes in self-care and mobility as indicated by PEDI scores. We observed, however, a very wide spread of initial skill levels in the Self-care and Social Function domain scales, and we found little or no effect on the initial status and the magnitude of the MCID on those scales.

The ratings on the Likert scale and the VAS (Tab. 4) produced what we believe is a high degree of correspondence ([tau] = .73-.80); this level of association between scales is comparable to results obtained in previous studies. (25) Strikingly, both the Likert scale and the VAS identified 11 points as the MCID on the PEDI. This finding strengthens the argument that both the Likert scale and the VAS can be validly used to identify the MCID on the PEDI when clinicians are the raters. Other MCID studies (27,28) indicate that the Likert scale is considered easier to administer, its scores are easier to interpret, and it involves less training time as compared with the VAS. We also observed preference for the Likert scale over the VAS by a number of clinicians.

In general, correlations between the rating scale scores and the PEDI change scores were in what we would consider the moderate range for Self-care and Mobility domain scales ([tau] = .59-.68) and a little less for the Social Function domain scales ([tau] = .43-.59). We used a conservative nonparametric statistic (Kendall tau-b) because of what we believe are the noninterval properties of the rating scales; therefore, these correlation coefficients may appear to be artificially low. Yet, it is clear that there are certain aspects of change ratings that are not consistent with PEDI change scores. For some subjects, meaningful changes occurred that were not reflected in the PEDI list of items, and, for other subjects, the changes recorded with the PEDI were not considered important to the clinicians. This occurred in the social area more than in the self-care and mobility areas. The developers of the PEDI considered the Social Function domain scales as more complex and difficult to rate in an inpatient environment as compared with the Mobility and Self-care domain scales. (5) The complexity of the Social Function domain could contribute to decreased correlations between therapists' ratings on the global rating scales and PEDI change scores. Further research is needed to examine the important areas of behavioral and social change that occurs within a hospital setting that provides a child with the likelihood of returning to the home and the community, as well as how those areas can be better incorporated into a functional assessment instrument.

In this study, we attempted to establish initial estimates of the MCID for the PEDI using a retrospective design with clinicians' opinions as the external criterion. These choices were based on the most feasible manner in which we could obtain initial estimates of MCID. Clinicians on-site at FCH were used as raters because it was the most practical alternative for this retrospective clinical study. Using clinicians at FCH as raters also provided us with information on the ability of the clinicians to use the rating scales for future prospective clinical studies. Use of clinicians from a single facility as raters may limit the generalizability of our data to some extent, suggesting a need for further research involving raters from multiple facilities. In previous work on describing outcomes of children with brain injuries, Bedell et al (23) found very dependable and consistent ratings of overall functioning when clinicians were presented with retrospective information from medical records. The results of our study provide initial evidence that an 11-point change in PEDI scores generally can be viewed as clinically important. However, replication of these findings in prospective studies and using other external criteria such as patients and children will be important in the future. It is also important to note that all of the MCID estimates obtained are specific to the subject population included in this study undergoing rehabilitation at one inpatient setting.

We believe our results can assist physical therapists in inpatient rehabilitation settings to examine program outcomes and benchmarks used to monitor quality assessment and improvements in performance. Based on our results, patients or clients who experience clinically important changes in function can be assigned as a positive responder to inpatient services when a minimum gain of 11 points occurs on scales of the PEDI. We contend that the MCID can help physical therapists and other clinicians to determine whether a meaningful functional change is likely to have occurred. This could improve the therapist's ability to interpret the effectiveness of interventions. For those clinicians and researchers using the PEDI in outcome studies or clinical trials, these initial MCID estimates can help interpret the effectiveness of physical therapy interventions on groups or individual children.

Conclusion

We determined that approximately 11 points on a 0 to 100 scaled score metric was the MCID of the PEDI Functional Skills and Caregiver Assistance scales using clinicians as the external criterion in an inpatient rehabilitation setting. The rating scales (Likert scale and VAS) performed equally well, although clinicians indicated a slight preference for using the Likert scale. The results, we contend, provide further evidence that the PEDI is not only a sensitive measure of functional change, but also is responsive to meaningful clinical changes in an inpatient pediatric rehabilitation setting.

Appendix. Rating Scales

15-Point Likert Scale

Please indicate how much this child changed from admission to discharqe in--capability/level of independence in the area of--(self-care, mobility, or social function) by choosing one of the options from the sheet in front of you.

7-A very great deal better 6-A great deal better 5-A good deal better 4-Moderately better 3-Somewhat better 2-A little better 1-About the same, hardly any better at all 0-No change -1-About the same, hardly any worse at all -2-A little worse -3-Somewhat worse -4-Moderately worse -5-A good deal worse -6-A great deal worse -7-A very great deal worse

15-cm Visual Analog Scale

Please indicate how much this child changed from admission to discharqe in--(capability/level of independence) in the area of--(self-care, mobility, or social function) by marking an X at the appropriate distance and direction from the center.

A Great Deal Worse

No Change

A Great Deal Better

Table 1.
Sample Characteristics

Age (y): X = 9.7, SD = 4.5, range = 1.8-19.3

Episode by Age (y) (a)    %   Discharge Setting              %

Under 2 (n = 1)           2   Immediate family              84
2-under 6 (n = 11)       20   Extended family                7
6-under 13 (n = 28)      51   Foster care                    4
13 and older (n = 15)    27   Acute care hospital            6

Sex                      %    Average Length of Stay (d)

Female                   40   X = 65.36
Male                     60   SD = 66.39
                              Range = 6-327

Race and Ethnicity       %    Diagnosis                     %

Asian/Pacific Islander    6   Traumatic brain injury        13
Black                     9   Spinal cord injury             7
Hispanic                 26   Lower-extremity trauma         9
White                    49   Nontraumatic brain injury     67
Other                     9   Congenital                     2
                              Growth/maturation              2
                                 (developmental)

(a) Number of admission-discharge episodes shown in parentheses.


Table 2.
Average Change in Pediatric Evaluation of Disability Inventory (PEDI)
Scaled Scores for Likert Scale Categories

                            No Change (0)
PEDI Scale (a)              N      X         SD       Range

Self-care FS (n = 53)        7       3.2      7.5      -3.9-8.6
Self-care CA (n = 53)        7       3.9      7.3       0.0-20.1
Mobility FS (n = 52)         4     -10.9     21.8     -42.8-5.3
Mobility CA (n = 52)         4       8.6     17.3       0.0-34.5
Social FS (n = 37)           4       5.8      5.1       0.0-11.3
Social CA (n = 37)           5       2.6      4.0       0.0-9.2
Summary means (SD)
   for all PEDI scales       2.2 (6.8)

                            Minimal Change (1-3)
PEDI Scale (a)              N      X         SD       Range

Self-care FS (n = 53)       15      10.1      9.3         0-34.2
Self-care CA (n = 53)       13      10.6     11.3         0-32.3
Mobility FS (n = 52)        12       8.7     11.5     -11.9-30.5
Mobility CA (n = 52)        20      14.9     14.6       0.0-57.4
Social FS (n = 37)          13      13.9     12.1       0.0-37.0
Social CA (n = 37)          17       9.3     12.3       0.0-46.8
Summary means (SD)
   for all PEDI scales      11.3 (2.5)

                            Moderate Change (4-5)
PEDI Scale (a)              N      X         SD       Range

Self-care FS (n = 53)       21      24.3     17.5     4.2-62.5
Self-care CA (n = 53)       18      18.5      8.5     3.4-30.9
Mobility FS (n = 52)        17      28.4     14.3     9.1-53.4
Mobility CA (n = 52)        21      38.9     18.0     3.6-70.5
Social FS (n = 37)          14      22.5     17.7     2.3-53.6
Social CA (n = 37)          11      29.7     22.8     0.0-65.4
Summary means (SD)
   for all PEDI scales      27.1 (7.1)

                            Large Change (6-7)
PEDI Scale (a)              N      X         SD       Range

Self-care FS (n = 53)       10      55.0     22.1     15.7-85.1
Self-care CA (n = 53)       14      60.6     24.1     18.7-100.0
Mobility FS (n = 52)        19      58.7     20.1     17.7-100.0
Mobility CA (n = 52)         7      66.6     22.7     34.9-100.0
Social FS (n = 37)           6      36.5     19.8      7.4-65.1
Social CA (n = 37)           4      29.9     18.4     19.1-57.3
Summary means (SD)
   for all PEDI scales      51.2 (14.6)

(a) FS = Functional Skills, CA = Caregiver Assistance.


Table 3.
Average Change in Pediatric Evaluation of Disability Inventory (PEDI)
Scaled Scores for Visual Analog Scale Categories

                           No Change (0)
PEDI Scale (a)             N      X         SD       Range

Self-care FS (n = 53)       9       6.3     12.4      -3.9-34.2
Self-care CA (n = 53)       8       3.5      7.3       0.0-21.0
Mobility FS (n = 52)        4     -10.9     21.8     -42.8-5.3
Mobility CA (n = 52)        5       8.0     15.0       0.0-34.5
Social FS (n = 37)          4       5.0      5.8       0.0-11.3
Social CA (n = 37)          6       1.5      3.8       0.0-9.2
Summary means (SD)
   for all PEDI scales      2.2 (6.9)

                           Minimal Change (0.1-2.5 cm)
PEDI Scale (a)             N      X         SD       Range

Self-care FS (n = 53)      11       9.2      7.0     1.2-24.1
Self-care CA (n = 53)      11      12.3     11.4     0.0-32.3
Mobility FS (n = 52)       15      14.2     11.9     0.0-38.6
Mobility CA (n = 52)       17      15.6     16.0     0.0-57.4
Social FS (n = 37)         16      11.5     11.1     0.0-37.0
Social CA (n = 37)         16       6.0      7.2     0.0-26.6
Summary means (SD)
   for all PEDI scales     11.5 (3.5)

                           Moderate Change (2.6-5.0 cm)
PEDI Scale (a)             N      X         SD       Range

Self-care FS (n = 53)      14      14.3     10.6       3.1-33.1
Self-care CA (n = 53)      12      16.9     13.7       2.3-53.4
Mobility FS (n = 52)       19      30.3     17.9     -11.9-56.1
Mobility CA (n = 52)       15      30.5     16.5       3.6-58.8
Social FS (n = 37)          9      27.5     17.3       7.6-53.6
Social CA (n = 37)          7      37.5     17.6      12.8-63.3
Summary means (SD)
   for all PEDI scales     25.9 (8.5)

                           Large Change (5.1-7.5 cm)
PEDI Scale (a)             N      X         SD       Range

Self-care FS (n = 53)      19      46.0     22.0     14.8-85.1
Self-care CA (n = 53)      21      46.7     27.4     10.2-100.0
Mobility FS (n = 52)       14      65.4     17.6     45.7-100.0
Mobility CA (n = 52)       15      56.9     21.4     20.3-100.00
Social FS (n = 37)          8      35.8     16.8      7.4-65.1
Social CA (n = 37)          8      31.1     19.1     15.2-65.4
Summary means (SD)
   for all PEDI scales     46.9 (12.8)

(a) FS = Functional Skills, CA = Caregiver Assistance.

Table 4.
Correlations Among Pediatric Evaluation of Disability Inventory (PEDI)
Change Scores and Visual Analog Scale (VAS) and Likert Scale

                                 Correlation (c)
                                 With PEDI
                                 Change Score
                 Average                             Correlation (c)
                 Change in       With                Between VAS
PEDI             PEDI Scaled     Likert     With     Likert and
Scale (a)        Score (b)       Scale      VAS      Scale

Self-care FS     23.3 (23.0)     .59        .63      .80 (d)
Self-care CA     25.9 (26.7)     .66        .60      .80 (d)
Mobility FS      31.9 (28.4)     .68        .67      .74
Mobility CA      31.1 (25.0)     .63        .59      .75 (d)
Social FS        20.0 (17.6)     .43        .55      .73
Social CA        15.8 (18.7)     .53        .59      .75 (d)

(a) FS = Functional Skills, CA = Caregiver Assistance.

(b) PFDI scores are mean (standard deviation).

(c) Kendall tau-b ([tau]).

(d) Significant correlations.

References

(1) Guide to Physical Therapist Practice. 2nd ed. Alexandria, Va: American Physical Therapy Association; 2001.

(2) Schneider JW, Gurucharri LM, Gutierrez AL, Gaebler-Spira DJ. Health-related quality of life and functional outcome measures for children with cerebral palsy. Dev Med Child Neurol. 2001;43:601-608.

(3) McAuliffe CA, Wenger RE, Schneider JW, Gaebler-Spira DJ. Usefulness of the Wee-Functional Independence Measure to detect functional change in children with cerebral palsy. Pediatric Physical Therapy. 1998;10(1):23-28.

(4) Hays RD, Woolley JM. The concept of clinically meaningful difference in health-related quality-of-life research; How meaningful is it? Pharmacoeconomics. 2000;18:419-423.

(5) Haley SM, Coster WJ, Ludlow LH, et al. Pediatric Evaluation of Disability Inventory (PEDI). Boston, Mass: Trustees of Boston University; 1992.

(6) Coster WJ, Haley SM, Baryza MJ. Functional performance of young children after traumatic brain injury: a 6 month follow-up study. Am J Occup Ther. 1994;48:211-218.

(7) Tokcan G, Haley SM, Gill-Body KM, Dumas HM. Item-specific recovery of functional activities for children with acquired brain injury. Pediatric Physical Therapy. 2003;15(1):16-22.

(8) Dudgeon BJ, Libby AK, McLaughlin JF, et al. Prospective measurement of functional changes after selective dorsal rhizotomy. Arch Phys Med Rehabil. 1994;75:46-53.

(9) Ketelaar M, Vermeer A, 't Hart H, et al. Effects of a functional therapy program on motor abilities of children with cerebral palsy. Phys Ther. 2001;81:1534-1545.

(10) Nordmark E. Comparison of the Gross Motor Function Measure and the Pediatric Evaluation of Disability Inventory in assessing motor function in children undergoing selective dorsal rhizotomy. Dev Med Child Neurol 2000;42:24.5-252.

(11) Fragala MA, Haley SM, Dumas HM, Rabin JP. Classifying mobility recovery in children and youth with brain injury during hospital-based rehabilitation. Brain Inj. 2002;16:149-160.

(12) Dumas HM, Haley SM, Ludlow LH, Rabin JP. Functional recovery in pediatric traumatic brain injury during inpatient rehabilitation. Am J Phys Med Rehabil 2002;81:661-669.

(13) Dumas HM, Haley SM, Fragala MA, Steva BJ. Self-care recovery of children with brain injury: descriptive analysis using the Pediatric Evaluation of Disability Inventory (PEDI) functional classification levels. Physical & Occupational Therapy in Pediatrics. 2001;21(2/3):7-27.

(14) Haley SM, Dumas HM, Ludlow LH. Variation by diagnostic and practice pattern groups in the mobility outcomes of inpatient rehabilitation programs for children and youth. Phys Ther. 2001;81:1425-1431.

(15) Dumas HM, Haley SM, Steva BJ. Functional changes during inpatient rehabilitation for children with musculoskeletal diagnoses. Pediatric Physical Therapy. 2002;14(2):85-91.

(16) Liang MH. Longitudinal construct validity: establishment of clinical meaning in patient evaluative instruments. Med Care. 2000;38:84-90.

(17) Lipsey MW. A scheme for assessing measurement sensitivity in program evaluation and other applied research. Psychol Bull. 1983;94:152-165:

(18) McHorney CA, Tarlov AR. Individual-patient monitoring in clinical practice: Are available health status surveys adequate? Qual Life Res. 1995;4:293-307.

(19) Jacobson NS, Roberts LJ, Bern SB, McGlinchey JB. Methods for defining and determining the clinical significance of treatment effects: description, application, and alternatives. J Consult Clin Psychol. 1999;67:300-307.

(20) Wyrwich KW, Nienaber NA, Tierney WM, Wolinsky FD. Linking clinical relevance and statistical significance in evaluating intra-individual changes in health related quality of life. Med Care. 1999;37:469-478.

(21) Jaeschke R, Singer J, Guyatt GH. Ascertaining the minimal clinically important difference. Control Clin Trials. 1989;10:407-415.

(22) Osoba D, Rodrigues G, Myles J, et al. Interpreting the significance of changes in health-related quality--of-life scores. J Clin Oncol. 1998;16:139-144.

(23) Bedell GM, Haley SM, Coster WJ, Smith KW. Developing a responsive measure of change for pediatric brain injury rehabilitation. Brain Inj. 2002;16:659-671.

(24) Revill SI, Robinson JO, Rosen M, et al. The reliability of a linear analogue for evaluating pain. Anaesthesia. 1976;31:1191-1198.

(25) Clarke PRF, Spear FG. Reliability and sensitivity in the self-assessment of well-being. Bulletin of the British Psychological ,Society. 1964;17:55.

(26) Singer AJ, Thode HC Jr. Determination of the minimal clinically significant difference on a patient visual analog satisfaction scale. Acad Emerg Med. 1998;5:1007-1011.

(27) Guyatt GH, Townsend M, Berman LB, Keller JL. A comparison of the Likert and visual analogue scales for measuring change in function. J Chron Dis. 1987;40:1129-1133.

(28) Jaeschke R, Singer J, Guyatt GH. A comparison of seven-point and visual analogue scales: data from a randomized trial. Control Clin Trials. 1990;11:43-51.

(29) Rasch G. Probabilistic Models for Some Intelligence and Attainment Tests. Copenhagen, Denmark: Denmarks Paedgogoiske Institut; 1960.

(30) Wright BD, Masters GN. Rating Scale Analysis. Chicago, Ill: MESA Press; 1982.

(31) Reid DT, Boschen K, Wright V. Critique of Pediatric Evaluation of Disability Inventory (PEDI). Physical & Occupational Therapy in Pediatrics. 1993;13(4):57-93.

(32) Sundberg KB. Interrater Reliability of the Pediatric Evaluation of Disability Inventory Parental and Professional Agreement [master's thesis]. Boston, Mass: Boston University; 1992.

(33) Nichols DS, Case-Smith J. Reliability and validity of the Pediatric Evaluation of Disability Inventory. Pediatric Physical Therapy. 1996;8(1):15-24.

(34) Wright FV, Boschen KA. The Pediatric Evaluation of Disability Inventory (PEDI): validation of a new functional assessment outcome instrument. Can J Rehabil. 1993;7:41-42.

(35) Haley SM, Coster WJ, Faas RM. A content validity study of the Pediatric Evaluation of Disability Inventory. Pediatric Physical Therapy. 1991;3(3/4):177-184.

(36) Feldman AB, Haley SM, Coryell J. Concurrent and construct validity of the Pediatric Evaluation of Disability Inventory. Phys Ther. 1990;70:602-610.

(37) Ziviani J, Ottenbacber KJ, Shephard K, et al. Concurrent validity of the Functional Independence Measure for children (WeeFIM) and the Pediatric Evaluation of Disability Inventory with developmental disabilities and acquired brain injury. Physical & Occupational Therapy in Pediatrics. 2001;21(2-3):91-101.

(38) Russell D, Palisano R, Walter S, et al. Evaluating motor function in children with Down syndrome: validity of the GMFM. Dev Med Child Neurol. 1998;40:693-701.

(39) Statistical Program for the Social Sciences, Version 9.0. Chicago, Ill: SPSS Inc; 1999.

LV Iyer, PT, MS, is Staff Physical Therapist, Warbasse Nursery School, Brooklyn, NY. She was a student in the Post-Professional Graduate Program in Physical Therapy, MGH Institute of Health Professions, Boston, Mass, when this study was conducted.

SM Haley, PT, PhD, is Director of Center of Rehabilitation Effectiveness and Associate Professor of Physical Therapy, Sargent College of Health and Rehabilitation Sciences, Boston University, 635 Commonwealth Ave, Boston, MA 02215 (USA) ([e-mail]). Address all correspondence to Dr Haley.

MP Watkins, PT, DPT, is Professor, MGH Institute of Health Professions.

HM Dumas, PT, MS, is Manager, The Research Center for Children With Special Health Care Needs, Franciscan Children's Hospital & Rehabilitation Center, Boston, Mass.

Dr Haley provided concept/idea/research design. All authors provided writing. Ms Iyer provided data collection, and Ms Iyer, Dr Haley, and Ms Dumas provided data analysis. Ms Iyer and Ms Dumas provided project management. Dr Watkins provided institutional liaisons. Ms Dumas provided subjects, facilities/equipment, and consultation (including review of manuscript before submission). The authors thank Ms Maggie Foley, Ms Tara Carey, Dr Nancy Latham, and the clinical and administrative staff of the physical therapy, occupational therapy, and speech-language-hearing departments at Franciscan Children's Hospital & Rehabilitation Center for their assistance with this project.

This study was undertaken in partial fulfillment of the requirements for Ms Iyer's post-professional degree of Master of Science in Physical Therapy.

The Spaulding Rehabilitation Hospital Institutional Review Board and the Institutional Review Board at Franciscan Children's Hospital & Rehabilitation Center approved this study.