Stahlhut, Michelle PT, BSc; Christensen, Jette PT, MSc; Aadahl, Mette PT, MPH, PhD
Children with disabilities often experience functional limitations in their motor, cognitive, perceptual, and social development. These limitations influence the child's ability to perform daily and leisure time activities, and the child may be at greater risk of social isolation.1 The consequences of disability during childhood have become evident through epidemiologic studies focusing on the incidence of childhood disability. Newacheck et al2 found that 18% of US children younger than 18 years had a chronic physical, developmental, behavioral, or emotional condition, requiring special health care.
In the rehabilitation of children with disabilities, a general goal is to maximize their opportunities to perform daily activities.3 Thus, it is important to assess and evaluate the functional abilities of a child. The Pediatric Evaluation of Disability Inventory (PEDI) measures functional status in children with disabilities and is frequently used as an assessment instrument within this population.4 In the PEDI, central functional skills (FSs) are described in children aged 6 months to 7½ years.5
The use of a functional assessment instrument such as the PEDI is advantageous because it emphasizes mastery of FSs and independence rather than normality. The PEDI has been found to be consistent with a majority of individual goals made for children with cerebral palsy in the components of activity and participation in the International Classification of Functioning, Disability and Health.6 Furthermore, the PEDI supports family-centered practice by involving parents in the evaluation of the child's functional abilities.7
The PEDI was developed and standardized in the United States in 1992, and subsequently, it has been translated into other languages.8–12 Several studies support the validity, reliability, and responsiveness to change of the PEDI.13–16 In 2005, the PEDI was translated into Danish, and the content validity was satisfactory.17 The Danish PEDI is formulated as an extended questionnaire to allow full parent report or for professionals to perform structured interviews. A small Danish pilot study of interrespondent reliability between 14 mothers of children with disabilities and the children's therapists showed low to moderate agreement.18 During a standardization process, it is also necessary to assess the applicability of an instrument in the cultural and practical context in which it is used.19 Applicability and cross-cultural validation studies of the PEDI have been performed in some European countries, which point to possible cultural differences.8,11,20–22 One could hypothesize that the US reference data from the PEDI are applicable in a Danish population because living conditions are somewhat similar. However, culture-specific child care practices could influence the development of skills in children.23,24 Bornstein et al25 have reported a cross-national study of mothers' self-evaluations in 7 countries in Asia, Europe, North America, and South America. The authors found significant differences in cultural ideas about parenting, which influenced child-rearing activities and parent-child interactions. The main purpose of this study was to analyze the applicability of the US PEDI reference values in a random Danish sample. Moreover, the intrarespondent reliability of the Danish version of the PEDI was examined.
To obtain a representative population sample, a random sample consisting of 605 children aged 0.8 to 5.8 years was drawn electronically from the Danish Civil Registration System. This age range was chosen to account for any delays in data collection. To be included in the applicability analysis, the children had to be within 1.0 to 5.9 years of age, as children without disabilities master most skills in the PEDI at the age of 5.5 to 6.0 years.26 In the reliability analysis, no age limits were set. Children would be excluded if their parents reported a medically verified disability. Demographic characteristics of the sample were obtained from Statistics Denmark. Nationality, geographic region, and maternal education were chosen as demographic characteristics to resemble the original PEDI sampling procedure.
The PEDI measures capability (what the child can do) and performance (what the child does) in daily activities in the domains of self-care, mobility and social function on FS, caregiver assistance (CA), and modifications scales, respectively. In this study, the FS and CA scales are used, corresponding to 6 subscales given that each domain and each scale constitutes a subscale. The PEDI can be administered through structured interview with parents, parent-completed questionnaire, and/or observations made by professionals.
According to the manual, each scale in the PEDI has different scoring criteria.5 Raw scores for each of the 6 subscales can be summarized and transformed into normative standard scores and scaled scores using the PEDI software. The normative standard scores are age adjusted and allow comparison with the US normative sample comprising 412 children without disabilities. The normative standard scores have a mean of 50 and a standard deviation of 10. A normative standard score less than 30 indicates functional delays because children without disabilities generally have scores between 30 and 70 (50 ± 2 standard deviations).5 The scaled scores are distributed along a continuum from 0 to 100, reflecting increasing levels of function and describing the child's performance relative to the maximum scores. The scaled scores are not age adjusted and are useful in the evaluation of the child's progress.
In this study, the Danish PEDI was administered as a parent-completed questionnaire. The reliability of this administration method has not yet been established.5 Thus, a small pilot study was conducted before this study. A convenience sample consisting of 20 children without disabilities aged 1.1 to 5.6 years participated in the pilot study in which the PEDI was administered as a parent-completed questionnaire and a structured interview completed by a physical therapist (PT). The administration method was chosen at random and median time between the 2 administrations was 15 days (range, 6–22 days). The level of agreement between the 2 methods was moderate to high (intraclass correlation coefficient [ICC] = 0.45–0.97). Based on these findings, it was decided to proceed with the parent-completed questionnaire method.
The PEDI questionnaire, a letter of information, and a prepaid envelope were mailed to the parents. Parents were asked to indicate whether they wanted to answer the questionnaire twice within a few weeks. The first questionnaire had to be returned within 14 days. After 3 weeks, a reminder was sent to increase the number of responses. The goal was to obtain at least 15 children in each age group. Based on previous PEDI studies with response rates as low as 17%,27 it was decided to contact approximately 600 families. The Danish Civil Registration System and Statistics Denmark provided information on 605 children. Initially, the questionnaire was mailed to 605 families (see flowchart in Figure 1). Eight questionnaires were returned because of wrong addresses. A total of 260 completed questionnaires were returned, yielding a response rate of 43%.
For the purpose of the applicability analysis, 30 of the returned questionnaires were excluded. Eleven children were excluded because they had a known disability, such as heart disease, neurologic disorder, or psychiatric disorder. Thirteen children were excluded as they fell outside the age limit; however, data from 7 of these children were used in the intrarespondent reliability analysis. Six questionnaires were excluded because the parents had completed less than two-thirds of the questionnaire. Questionnaires with fewer than 5 missing items in total were still included. Accordingly, the random sample of children without disabilities consisted of 230 children aged 1.0 to 5.9 years. In 92% of the cases, the questionnaires were completed by the mother, in 5% of the cases by both parents, and in 3% of the cases by the father.
For the purpose of the reliability analysis, a total of 92 parents returned the questionnaire twice. The PEDI questionnaires were completed by the same respondents twice. As a result, the reliability analysis was performed with data from 92 children without disabilities aged 0.8 to 5.9 years.
The study was approved by the Danish Data Protection Agency (J.nr. 2007-41-0958). All participating parents gave informed consent. Participation was voluntary, and the parents were informed that they could withdraw from the study at any stage. All participating parents received a letter outlining the PEDI results for their child. In the case of a child having low normative standard scores (<30) in more than 2 subscales, parents were contacted to clarify whether the results were based on misinterpretations of the questionnaire or whether the child had unknown functional delays. Four parents were contacted, and none of their children were considered to have functional delays.
Differences in sex, nationality, geographic region, and maternal education between children of respondents and nonrespondents were compared using the χ2 test. Independent t tests and analyses of variance were used to compare the PEDI data from included respondents.
In the PEDI manual, means and standard deviations in the US normative sample are available for the scaled scores. Unfortunately, it was not possible to obtain means and standard deviations for the normative standard scores. A linear regression model was used to determine whether there were significant differences between scaled scores in the US normative sample and the Danish study sample. For each score, a second-degree polynomium of age was fitted separately for Danish and US data, weighted by the inverse variances of the means. The 2 curves were tested to determine whether they were equal, assuming parallel curves.
Moreover, an estimate of item difficulty in the Danish study sample was calculated in logit values. Calculations were based on item responses in the Danish sample. Positive logit values represent increasing item difficulty. The Danish logit values were compared with the US logit values using Pearson's correlation analysis. This comparison gave an indication of similarity in item difficulty.
ICCs were calculated by domain for normative standard scores and scaled scores in the intrarespondent analysis. A P value less than .05 was considered statistically significant. The PEDI software was used to store all raw data from the questionnaires and to transform raw scores into normative standard scores and scaled scores. All statistical analyses were conducted using R (version 2.6.2) software package.
Description of Study Sample
Table 1 describes the demographic characteristics among children of nonrespondents and respondents. Significant differences between the 2 groups were found in nationality and maternal education. In the respondent group, the proportion of children with a nationality other than Danish was smaller. Furthermore, there was a larger proportion of mothers with a graduate education in the respondent group. No additional demographic differences were found between the excluded (n = 30) and included (n = 230) children.
Further analysis of the PEDI results in the 230 included children showed no relationship between the results and the nationality of child (Danish versus other nationality) or the educational level of mother (4 educational levels). Overall, the Danish study sample was found to be representative of a general Danish population with respect to sex, nationality, and geographic region. According to Statistics Denmark, 90.5% of the population are of Danish nationality corresponding to the proportion seen in the study sample. A larger proportion of mothers with graduate education was seen in the study sample compared with the 31.7% in all Danish women.28
Descriptive statistics for the study sample in each of the 6 subscales are given in the Appendix (available online at http://links.lww.com/PPT/A9).
Analysis of Applicability
Means and standard deviations of the scaled scores in the US normative sample and the Danish study sample were plotted against age separately for the US and Denmark in all 6 subscales. This is seen in Figures 2 through 7. A misfit is seen in 3 subscales: FS (social function) and CA (self-care and mobility) where the analysis showed a significant difference with P values less than .05 (Figures 4–6). Accordingly, Danish children seem more capable in social function skills, whereas US children seem more independent in self-care and mobility. The average difference in scaled scores between the Danish and the US children was 2.3 points in FS (social function), 5.5 points in CA (self-care), and 3.8 points in CA (mobility).
An additional analysis was conducted with an applied correction of age in these 3 subscales. This was done to identify possible systematic age differences between the US and Danish children. The PEDI manual provides data for children at 6-month intervals. Accordingly, US children were scored 6 months older in FS (social function) and 6 months younger in CA (self-care and mobility). Significant differences still persisted in FS (social function) but now in favor of the US children (ie, US children were more capable than Danish children). However, no significant differences were seen in the CA (self-care and mobility) subscales after the age correction, indicating that US children on average are 6 months ahead of Danish peers in these 2 subscales.
The applicability of the PEDI was further examined with comparisons of logit values. Based on the logit values in the Danish study sample and the US normative sample, Pearson correlations were calculated. As seen in Table 2, strong correlations were found both in the FS scales and the CA scales. All correlations were more than 0.8, indicating that the PEDI item difficulty is likely to be comparable in the US and Denmark.
Analysis of Intrarespondent Reliability
A total of 92 children (42 males, 50 females) aged 0.8–5.9 years were included in the reliability analysis. Median time between questionnaire responses was 21.5 days (range, 13–36 days). Table 3 reports the ICCs for the normative standard scores and scaled scores. The ICCs indicate that the intrarespondent reliability was good to high.
The US PEDI reference values were only partly applicable in a Danish population. Results showed that the PEDI reference values were not applicable in the subscales of FS (social function) and CA (self-care and mobility). According to the results, Danish children were more capable in social FSs, whereas US children were more independent in self-care and mobility activities. In the analysis of the CA subscales of self-care and mobility, significant differences did not persist when US children were regarded as being 6 months younger. Several other studies have found that the PEDI reference values are not directly applicable in Slovenia, the Netherlands, and Norway.11,20,22 In Norway, Berg et al22 used a similar sampling method as in this study. Norwegian children, however, had significantly lower scores in all subscales of the PEDI, especially pronounced in the self-care domain, which highlighted the need for Norwegian norms. In contrast, a Swedish study showed that the US PEDI reference values were applicable in Sweden for children aged 2.0 to 6.9 years.8
Additional analysis of items in the PEDI showed strong correlations (Pearson r > 0.8) between the US and Danish structure of items. This indicates that the item difficulty is comparable between the 2 countries and, thus, supports the applicability of the PEDI in Denmark. Further item validation, however, using Rasch analysis to confirm the hierarchical structure of the items in Denmark is needed.
Various cultural and social factors influencing child-rearing practices could explain the functional differences found between US and Danish children. Pachter and Dworkin23 showed that maternal expectations and beliefs about infant development vary across different ethnic groups in the US. For instance, European American mothers expected children to take first steps and become toilet trained at a later age than African American mothers. This has implications on child-rearing and handling practices that aid the development of skills.24 Studies of child-rearing often compare parents from Western and non-Western cultures. However, differences might exist among different Western cultures. Tulviste et al29 showed that child-rearing values in Swedish and Finnish mothers vary. Similarly, Bornstein et al25 found significant differences in mothers' self-evaluations among several Western cultures. As such, it is possible that cultural differences are evident between the United States and Denmark. Furthermore, the US PEDI reference values were derived almost 20 years ago; thus, habits and views of child independence may have changed.
In the self-care domain, the results showed that US children become independent at an earlier age. This domain focuses on skills such as bladder/bowel control, dressing, and tooth brushing. A recent study reports an increase in age of attainment of bladder and bowel control from 25 to 27 months in the mid-1980s to 35 to 39 months in the mid-1990s to approximately 45 months in 2002.30 A contributing factor to this trend might be the convenience of new and better disposable diapers.31 The extended use of diapers could also increase the children's need for assistance when dressing the lower body. Regarding tooth brushing, Danish dentists recommend that parents take part and supervise their children until the age of 10 years, whereas the PEDI reference values show that children are expected to brush their teeth thoroughly at the of age 6 years.32 In the returned questionnaires, parents often pointed to the recommendation from the dentist. These aforementioned factors could have increased the Danish children's need for CA in self-care activities.
US children also seemed more independent in mobility than Danish children. The mobility domain involves locomotion and transfers (eg, in/out of the car and bathtub). In the past decades, increasing car safety has entailed automatic door locks and child seats with complicated lock systems.33 Thus, activities such as opening/closing car doors and managing a seat belt will be done by parents. Furthermore, Danish children might not have extensive experience with bathtub transfers because most families only have showers.34 Parents of Danish children might therefore assist the child for safety reasons when they occasionally use a bathtub.
In the social function domain, Danish children mastered more skills at an earlier age. This could be due to the fact that the majority of Danish children younger than 5 years (97%) attend daycare centers.28 Danish children might therefore engage in social interaction and peer play at earlier ages.
The results from the applicability analysis have important implications for the interpretation and clinical use of the PEDI in a Danish rehabilitation setting. In a physical therapy setting, normative standard scores are used to determine whether a child demonstrates functional delays. Given the differences between US and Danish children, PTs must be cautious when interpreting the normative standard scores, especially in the subscales CA (self-care and mobility) where Danish children without disabilities are more dependent on assistance. It might be of practical use to allow the Danish child to be scored in a younger age group (6 months younger) in these 2 subscales when the PT needs an indication of the functional status of the child. Still, this will not allow a fully accurate interpretation of functional delays using the normative standard scores. As the scaled scores of the child are not age adjusted, these scores are, however, very useful when PTs evaluate change and plan treatment.
In the intrarespondent analysis, ICCs showed good to high levels of reliability. In some cases, the ICCs fell below 0.80, which could be explained by a test effect because parents might be more conscious of the capability and performance of their child after the first administration.15 Results from this study are comparable with the reliability levels found in an US and Dutch study of test-retest reliability. Nichols and Case-Smith13 examined test-retest reliability in 23 children with disabilities aged 1 to 6 years and found ICCs between 0.70 and 0.98 in normative standard scores and between 0.74 and 0.98 in scaled scores. Correspondingly, Wassenberg-Severijnen et al15 found ICCs between 0.91 and 0.98 in scaled scores in 53 children with disabilities and 63 children without disabilities aged 6 months to 7½ years.
The random sampling method is considered to be a strength of this study, resulting in the response rate of 43%, which corresponds to those rates found in similar studies.22,35 Children were randomly drawn from the Danish Civil Registration System, yielding a representative study population. Demographic analysis between respondents and nonrespondents showed an overrepresentation of mothers with graduate education and an underrepresentation of children with another nationality than Danish. However, factors such as maternal education and nationality did not affect the PEDI results, which correspond to findings in other studies.5,22 A sample bias might have occurred because parents who feel confident about the development of their child tend to be those who participate in studies.36 This study is limited by the data collection method with parent-completed questionnaires. With the use of mailed questionnaires, parents may misinterpret the PEDI items, complete items inadequately, or give socially desirable responses (ie, their child seems to be more skillful with regard to culturally derived standards). Given the similarities and differences found between US and Danish children, the latter seems to be unlikely. The reliability of the PEDI as a parent-completed questionnaire has not been established. However, a previous pilot study examining the agreement between the PEDI data collected through a structured interview or questionnaire supported the questionnaire method with high ICCs except in the subscale CA (social function). Previous studies have also found the social function domain to be less reliable.5,14
Overall, results showed that the US PEDI reference values were partly applicable in a Danish population. The Danish study sample had significantly lower scores in the subscales CA (self-care and mobility), whereas they had significantly higher scores in the FS subscale (social function). The differences between US and Danish children might be due to cultural factors or social development during the past 20 years. Analyses of item difficulty in the United States and Denmark showed high correlations supporting the applicability of the PEDI in Denmark. Further studies of item validation using Rasch analysis are, however, recommended to determine whether cross-cultural differences exist. The PEDI is a useful assessment instrument in the Danish rehabilitation setting; however, PTs must be careful when interpreting functional delays using the PEDI reference values. The use of PEDI as an evaluation tool is still very advantageous. Furthermore, study results supported the intrarespondent reliability of the Danish version of the PEDI.
The authors thank all the parents and children who participated in this study.
1. McDougall J, King G, de Wit DJ, et al. Chronic physical health conditions and disability among Canadian school-aged children: a national profile. Disabil Rehabil. 2004;26:35–45.
2. Newacheck PW, Strickland B, Shonkoff JP, et al. An epidemiologic profile of children with special health care needs. Pediatrics. 1998:102:117–123.
3. Custers JWH. Pediatric Evaluation of Disability Inventory: The Dutch Adaptation. Utrecht, the Netherlands: Department of Pediatric Physical Therapy, University Medical Center Utrecht; 2001: ch. 1. PhD Dissertation.
4. Law M (co-ed.). Outcome measurement in pediatric rehabilitation. Phys Occup Ther Pediatr. 2003;23:1–4.
5. Haley SM, Coster WJ, Ludlow LH, et al. Pediatric Evaluation of Disability Inventory (PEDI). Version I. Development, Standardization and Administration Manual. Boston, MA: New England Center Hospital; 1992.
6. Engelen V, Ketelaar M, Gorter JW. Selecting the appropriate outcome in paediatric physical therapy: how individual treatment goals for children with cerebral palsy are reflected in GMFM-88 and PEDI. J Rehabil Med. 2007;39:225–231.
7. Helders PJM, Engelbert RHH, Gulmans VAM, et al. Paediatric rehabilitation. Disabil Rehabil. 2001;23:497–500.
8. Nordmark E, Orban K, Hägglund G, et al. The American Paediatric Evaluation of Disability Inventory (PEDI). Applicability of PEDI in Sweden for children aged 2.0–6.9 years. Scand J Rehab Med. 1999;31:95–100.
9. Custers JWH, Wassenberg-Severijnen JE, van der Net J, et al. Dutch adaptation and content validity of the “Pediatric Evaluation of Disability Inventory” (PEDI). Disabil Rehabil. 2002;24:250–258.
10. Berg M, Jahnsen R, Holm I, et al. Translation of a multi-disciplinary assessment—procedures to achieve functional equivalence. Adv Phys. 2003;5:57–66.
11. Srsen KG, Vidmar G, Zupan A. Applicability of the pediatric evaluation of disability inventory in Slovenia. J Child Neurol. 2004;20:411–416.
12. Chen KL, Hsieh CL, Sheu CF, et al. Reliability and validity of a Chinese version of the pediatric evaluation of disability inventory in children with cerebral palsy. J Rehabil Med. 2009;41:273–278.
13. Nichols DS, Case-Smith J. Reliability and validity of the pediatric evaluation of disability inventory. Pediatr Phys Ther. 1996;8:15–24.
14. Berg M, Jahnsen R, Froeslie KF, et al. Reliability of the pediatric evaluation of disability inventory. Phys Occup Ther Pediatr. 2004;24:61–77.
15. Wassenberg-Severijnen JE, Custers JWH, Hox JJ, et al. Reliability of the Dutch “pediatric evaluation of disability inventory” (PEDI). Clin Rehabil. 2003;17:457–462.
16. Haley SM, Coster WJ, Faas RM. A content validity study of the pediatric evaluation of disability inventory. Pediatr Phys Ther. 1991;3:177–184.
19. McKenna SP, Doward LC. The translation and cultural adaptation of patient-reported outcome measures. Val Health. 2005;8:89–91.
20. Custers JW, Hoijtink H, van der Net J, et al. Cultural differences in functional status measurement: analyses of person fit according to the Rasch model. Qual Life Res. 2000;9:571–578.
21. Berg M, Froeslie KF, Hussain A. Applicability of pediatric evaluation of disability inventory in Norway. Scand J Occup Ther. 2003;10:118–126.
22. Berg M, Aamodt G, Stanghelle J, et al. Cross-cultural validation of the pediatric evaluation of disability inventory (PEDI) norms in a randomized Norwegian population. Scand J Occup Ther. 2008;15:143–152.
23. Pachter LM, Dworkin PH. Maternal expectations about normal child development in 4 cultural groups. Arch Pediatr Adolesc Med. 1997;151:1144–1150.
24. WHO Multicentre Growth Reference Study Group. Assessment of sex differences and heterogeneity in motor milestone attainment among populations in the WHO Multicentre Growth Reference Study. Acta Paediatr. 2006;450:66–75.
25. Bornstein MH, Haynes OM, Azuma H, et al. A cross-national study of self-evaluations and attributions in parenting: Argentina, Belgium, France, Israel, Italy, Japan, and the United States. Dev Psychol. 1998;34:662–676.
26. Dumas HM, Haley SM, Fragala MA, et al. Self-care recovery of children with brain injury: descriptive analysis using the pediatric evaluation of disability inventory (PEDI) functional classification levels. Phys Occup Ther Pediatr. 2002;21:7–27.
27. Wassenberg-Severijnen J. Pediatric Evaluation of Disability Inventory (PEDI). Calibrating the Dutch Version. Utrecht, the Netherlands: Department of Pediatric Physical Therapy, University Medical Center Utrecht; 2005. PhD Dissertation.
28. Gunnersen SJ, Bisgaard MP, eds. Statistical Yearbook 2009 [in Danish]. Statistics Denmark, 2009.
29. Tulviste T, Mizera L, De Geer B, et al. Child-rearing goals of Estonian, Finnish, and Swedish mothers. Scand J Psychol. 2007;48:487–497.
30. Schum TR, Kolb TM, McAuliffe TL, et al. Sequential Acquisition of toilet-training skills: a descriptive study of gender and age differences in normal children. Pediatrics. 2002;109:e48.
31. Bakker E, Wyndaele JJ. Changes in the toilet training of children during the last 60 years: the cause of an increase in lower urinary tract dysfunction? BJU Int. 2000;86:248–252.
32. Sundby A. Oral health promotion and prevention of dental diseases in children and adolescents [in Danish]. Tandlaegebladet. 2005;109:458–463.
33. Bull MJ, Sheese J. Update for the pediatrician on child passenger safety: five principles for safer travel. Pediatrics. 2000;106:1113–1116.
34. Quitzau M. Bathroom practices today. Paper presented at the 7th Conference of European Sociological Association, Torun, Poland, September 9–12, 2005.
35. Janson H, Squires J. Parent-completed developmental screening in a Norwegian population sample: a comparison with US normative data. Acta Paediatr. 2004;93:1525–1529.
36. Janus M, Goldberg S. Factors influencing family participation in a longitudinal study: comparison of pediatric and healthy samples. J Pediatr Psychol. 1997;22:245–262.
© 2010 Lippincott Williams & Wilkins, Inc.