Reliability and Validity of the Pediatric Quality of Life Inventory Generic Core Scales, Multidimensional Fatigue Scale, and Cancer Module

Transcription

1 2090 The PedsQL in Pediatric Cancer Reliability and Validity of the Pediatric Quality of Life Inventory Generic Core Scales, Multidimensional Fatigue Scale, and Cancer Module James W. Varni, Ph.D. 1,2 Tasha M. Burwinkle, M.A. 1 Ernest R. Katz, Ph.D. 3,4 Kathy Meeske, R.N., M.N. 3 Paige Dickinson, M.A., M.S. 1 1 Center for Child Health Outcomes, Children s Hospital and Health Center, San Diego, California. 2 Department of Psychiatry, University of California, San Diego School of Medicine, San Diego, California. 3 Center for Cancer and Blood Diseases, Children s Hospital, Los Angeles, California. 4 Department of Pediatrics, University of Southern California School of Medicine, Los Angeles, California. BACKGROUND. The Pediatric Quality of Life Inventory (PedsQL) is a modular instrument designed to measure health-related quality of life (HRQOL) in children and adolescents ages 2 18 years. The PedsQL 4.0 Generic Core Scales are multidimensional child self-report and parent proxy-report scales developed as the generic core measure to be integrated with the PedsQL disease specific modules. The PedsQL Multidimensional Fatigue Scale was designed to measure fatigue in pediatric patients. The PedsQL 3.0 Cancer Module was designed to measure pediatric cancer specific HRQOL. METHODS. The PedsQL Generic Core Scales, Multidimensional Fatigue Scale, and Cancer Module were administered to 339 families (220 child self-reports; 337 parent proxy-reports). RESULTS. Internal consistency reliability for the PedsQL Generic Core Total Scale Score ( 0.88 child, 0.93 parent report), Multidimensional Fatigue Total Scale Score ( 0.89 child, 0.92 parent report) and most Cancer Module Scales (average 0.72 child, 0.87 parent report) demonstrated reliability acceptable for group comparisons. Validity was demonstrated using the known-groups method. The PedsQL distinguished between healthy children and children with cancer as a group, and among children on-treatment versus off-treatment. The validity of the PedsQL Multidimensional Fatigue Scale was further demonstrated through hypothesized intercorrelations with dimensions of generic and cancer specific HRQOL. CONCLUSIONS. The results demonstrate the reliability and validity of the PedsQL Generic Core Scales, Multidimensional Fatigue Scale, and Cancer Module in pediatric cancer. The PedsQL may be utilized as an outcome measure in clinical trials, research, and clinical practice. Cancer 2002;94: American Cancer Society. DOI /cncr KEYWORDS: health-related quality of life, cancer, pediatrics, health status, children, adolescents. Supported by a research grant from Ortho Biotech Products, L.P. The PedsQL is available at Address for reprints: James W. Varni, Ph.D., Center for Child Health Outcomes, Children s Hospital and Health Center, 3020 Children s Way, San Diego, CA 92123; Fax: (858) ; jvarni@chsd.org Received July 5, 2001; revision received October 30, 2001; accepted October 31, Health-related quality-of-life measurement has received ever-increasing recognition as an important health outcome in clinical trials and health services research and evaluation. 1 3 Although health status, functional status, and health-related quality of life (HRQOL) are terms often used interchangeably, a recent meta-analysis suggests that a more definitive distinction between these terms is warranted. 4 Health status and functional status refer to the physical functioning dimension of the broader HRQOL construct, whereas the HRQOL construct in addition includes the psychosocial dimensions of emotional, social, and role functioning, as well as related constructs. Thus, a HRQOL instrument must be multidimensional, consisting at the 2002 American Cancer Society

2 The PedsQL/Varni et al minimum of the physical, mental, and social health dimensions delineated by the World Health Organization. 5 Although the importance of measuring HRQOL in pediatric cancer has been advocated for several years, 6,7 the measurement of fatigue in pediatric cancer is a more contemporary conceptualization. 8 Pediatric HRQOL measurement instruments must be sensitive to cognitive development and integrate both child self-report and parent proxy-report to reflect their potentially unique perspectives. Imperfect concordance between self- and proxy-report, termed cross-informant variance, 9 has been consistently documented among child/adolescent, parent, teacher, and healthcare professionals, reports in the HRQOL assessment of healthy children 10 and children with chronic health conditions, including cancer. 15 Agreement has been demonstrated to be lower for internalizing problems (e.g., depression and pain) than for externalizing problems (e.g., hyperactivity and walking). Because HRQOL derives from an individual s perception, the demonstration of cross-informant variance indicates the need for reliable and valid child self-report HRQOL measurement instruments for the broadest age range possible. The Pediatric Quality of Life Inventory (PedsQL) builds on a programmatic measurement instrument development effort by Varni and colleagues in pediatric chronic health conditions, including pediatric cancer, during the past 15 years. The PedsQL 1.0, 19 originally derived from a pediatric cancer database, was designed as a generic HRQOL instrument to be utilized noncategorically, that is, across diverse pediatric populations. Given that instrument development is an iterative process, the PedsQL 2.0 and 3.0 included additional constructs and items, a more sensitive scaling range, and a broader age range for child self-report and parent proxy-report. The PedsQL 4.0 Generic Core Scales have resulted from this iterative process and include child self-report for ages 5 18 years and parent proxyreport for ages 2 18 years. The PedsQL 4.0 Generic Core Scales were designed to measure the core physical, mental, and social health dimensions as delineated by the World Health Organization, 5 as well as role (school) functioning. In the initial field trial, the PedsQL 4.0 Generic Core Scales (physical, emotional, social, and school functioning) were administered to 1677 families (963 child self-report; 1629 parent proxy-report), recruited from pediatric healthcare settings. 20 Internal consistency reliability for the Total Scale Score ( 0.88 child, 0.90 parent report), Physical Health Summary Score ( 0.80 child, 0.88 parent report), and Psychosocial Health Summary Score ( 0.83 child, 0.86 parent report) were acceptable for group comparisons. The PedsQL 4.0 Generic Core Scales distinguished between healthy children and pediatric patients with acute or chronic health conditions and were related to indicators of morbidity and illness burden. 20 The PedsQL Measurement Model was designed as a modular approach to measuring pediatric HRQOL, integrating the relative merits of generic and disease specific approaches. 19,21 Although disease specific modules may enhance measurement sensitivity for health domains germane to a particular chronic health condition, a generic HRQOL measurement instrument enables comparisons across pediatric populations and facilitates benchmarking with healthy population norms. Given that the PedsQL Measurement Model integrates generic core scales and disease specific modules into one measurement system, the PedsQL 4.0 Generic Core Scales were specifically designed for application in both healthy and patient populations. The PedsQL 3.0 Cancer Module was designed to measure HRQOL dimensions specifically tailored for pediatric cancer. The PedsQL 3.0 Cancer Module, derived from the PedsQL 1.0 Cancer Module, 19 also has resulted from an extensive iterative process, including additional constructs and items, a more sensitive scaling range, and a broader age range for child self-report and parent proxy-report. The PedsQL Multidimensional Fatigue Scale is a newly developed instrument, designed to measure child and parent perceptions of fatigue in pediatric patients. There are several instruments that measure fatigue in adult cancer patients, 22,23 including several that are explicitly multidimensional fatigue measures In pediatric cancer, the measurement of fatigue is exemplified by the work of Hockenberry- Eaton, Hinds, and colleagues. 8,27 In the development of the PedsQL Multidimensional Fatigue Scale, multidimensional constructs were derived from reviews of both the adult and pediatric cancer fatigue literature and integrated into the PedsQL Measurement Model. The current study presents the measurement properties of the PedsQL in pediatric cancer, reporting on the reliability and validity of the Generic Core Scales, Multidimensional Fatigue Scale, and Cancer Module. METHODS Subjects and Settings Oncology sample Subjects were recruited from the Hematology/Oncology Center at Children s Hospital and Health Center, San Diego, and the Center for Cancer and Blood Diseases at Children s Hospital Los Angeles. Participants were 339 families of children ages 2 18 years. Children ages 5 18 years completed the PedsQL child self-re-

3 2092 CANCER April 1, 2002 / Volume 94 / Number 7 port (n 220), and parents of children ages 2 18 years completed the PedsQL parent proxy-report (n 337). In 119 cases, a parent proxy-report was completed, but a child self-report was not completed, and in 2 cases, a child self-report was completed, but a parent proxy-report was not completed. Depending on the scale, both child self-report and parent proxy-report were available on 190 to 218 parent/child dyads. The PedsQL was self-administered for parents and for children ages 8 18 years and interviewer-administered for children ages 5 7 years. Most patients and parents (n 180, 82%, child report; n 276, 82%, parent report) completed the PedsQL in the hospital outpatient clinics. The PedsQL also was administered when a child was an inpatient in the cancer unit (n 15, 7%, child report; n 23, 7%, parent report), or while a child was admitted to the day hospital (n 2, 1%, child report; n 3, 1%, parent report). Some children (n 23, 10%) and parents (n 35, 10%) completed the PedsQL at home, where the parents were given detailed written instructions about how to complete the questionnaire as well as how to administer the questionnaire to their child. The instructions stated that the parent should first complete the parent proxy-report separately from their child, that they should read the instructions and items aloud to a child ages 5 7 years, and that children and adolescents ages 8 18 years should independently complete the child self-report items after reading the instructions. The measures were administered in two languages: English (n 174, 79%, child report; n 223, 66%, parent report) and Spanish (n 46, 21%, child report; n 114, 34%, parent report). The oncology sample included children with acute lymphocytic leukemia (n 171, 50%), brain tumor (n 24, 7%) non-hodgkin lymphoma (n 20, 6%), Hodgkin lymphoma (n 11, 3%), Wilms tumor (n 19, 6%), and other cancers (n 94, 28%). Patients comprised all diagnostic groups, had no comorbid disease or major developmental disorders, and included newly diagnosed on-treatment, recurrent disease on-treatment, in recent remission off-treatment, or long-term off-treatment. On-treatment status was defined as patients who were receiving medical treatment to induce remission (n 183, 54%). Off-treatment status was defined as patients for whom all therapy was completed by the time of assessment. Off-treatment remission was defined by being disease free accompanied by termination of cancer treatment in the past 12 months (n 61, 18%). Off-treatment long-term survivor was defined by being disease free accompanied by termination of cancer treatment for longer than 12 months (n 95, 28%). For all forms combined, the average age of the 196 boys (58%) and 143 girls (42%) was 8.72 years (standard deviation [SD], 4.57) with a range of 2 to 18 years. For child self-report, the average age of the 124 boys (56%) and 96 girls (44%) was 10.9 years (SD, 3.83) with a range of 5 to 18 years. Children identified as being off-treatment long-term survivor were significantly older (mean age, 10.1 years) than children on-treatment (mean age, 8.2 years) and children off-treatment for less than 12 months (mean age, 8.2; P 0.05, based on Tukey post hoc analysis). Comparisons among oncology treatment groups indicate that there was no statistically significant difference in the ratio of males to females. There was, however, a statistically significant difference between the number of males and females in the oncology sample as a whole (P 0.001). The sample was heterogeneous with respect to race/ ethnicity, with 98 (29%) white non-hispanic, 175 (52%) Hispanic, 15 (4%) black non-hispanic, 19 (6%) Asian/Pacific Islander, 2 (1%) American Indian or Alaskan Native, 27 (8%) other, and 3 (1%) missing. Mean socioeconomic status (SES) for the oncology sample was 35, based on the Hollingshead index, indicating on average a low- to middle-class family SES. 28 There were no statistically significant differences in SES between on-treatment and off-treatment groups. Healthy sample: Generic Core Scales Acute Version Subjects were 157 families of healthy children ages 2 18 years. Healthy children ages 5 18 years (n 105) and parents of healthy children ages 2 18 years (n 157) were administered the PedsQL 4.0 Generic Core Scales Acute Version via telephone. This sample was accrued from a list of patients who had attended an orthopedic clinic for broken bones or fractures 6 months before assessment with the PedsQL, and who had been identified by the clinic nurse as having returned to health (i.e., no current problems due to their orthopedic injury). The average age of the 98 boys (62.4 %) and 58 girls (36.9 %; missing, 1; 0.6%) was years (SD, 8.59). For child self-report, the average age of the 72 boys (68.6 %) and 33 girls (31.4 %) was years (SD, 9.47). The sample was heterogeneous with respect to race/ethnicity, with 77 (49%) white non-hispanic, 39 (25%) Hispanic, 6 (4%) black non-hispanic, 5 (3%) Asian/Pacific Islander, 14 (9%) other, and 16 (10%) missing. Mean SES was unavailable for this sample. Comparisons between this healthy sample and the oncology sample indicates that there was no statistically significant difference in the ratio of males to females assessed with the PedsQL 4.0 Generic Core Scales Acute Version. This healthy sample was significantly older (mean age, 11.6 years) than the oncology sample (mean age, 8.7 years) and

4 The PedsQL/Varni et al the healthy sample for the PedsQL Generic Core Scales Standard Version (mean age, 8.9 years; P 0.01, based on Tukey post hoc analysis). Healthy sample: Generic Core Scales Standard Version Participants were 730 families of healthy children ages 2 18 years, whose scores were derived from the PedsQL 4.0 Generic Core Scales field test. 20 Healthy children ages 5 18 years completed the PedsQL child self-report (n 401), and parents of healthy children ages 2 18 years completed the PedsQL parent proxyreport (n 719). Healthy children were those assessed either in their physician s offices during well-child checks or by telephone and whose parents did not report the presence of a chronic health condition. The average age of the 351 boys (48.1 %) and 378 girls (51.8 %; missing, 1) was 8.9 years (SD, 4.24; missing, 8). For child self-report, the average age of the 182 boys (45.4 %) and 219 girls (54.6 %) was 10.9 years (SD, 3.30; missing, 1). The sample was heterogeneous with respect to race/ethnicity, with 211 (28.9 %) white non- Hispanic, 362 (49.6 %) Hispanic, 48 (6.6 %) black non- Hispanic, 20 (2.7 %) Asian/Pacific Islander, 4 (0.5 %) American Indian or Alaskan Native, 47 (6.4 %) other, and 38 (5.2 %) missing. Mean SES was unavailable for this sample. Comparisons between this healthy sample and the oncology sample indicate that the ratio of males to females was different in the two samples with 351 (48.1%) males and 378 (51.8%) females in the healthy sample as compared with the oncology sample with 196 (58%) males and 143 (42%) females (chisquare(1) 8.66; P 0.01). There was no statistically significant difference in age between this healthy sample and the oncology sample. Healthy sample: Multidimensional Fatigue Scale Acute Version Subjects were 102 families of healthy children ages 2 18 years. Healthy children ages 5 18 years (n 52) and parents of healthy children ages 2 18 years (n 102) were administered the PedsQL Multidimensional Fatigue Scale Acute Version via telephone. This sample was accrued from a list of patients who had attended an orthopedic clinic for broken bones or fractures 6 months before assessment with the PedsQL, and who had been identified by the clinic nurse as having returned to health (e.g., no current problems due to their orthopedic injury). The average age of the 69 boys (67.7 %) and 30 girls (29.4 %; missing, 3) was 8.88 years (SD, 10.98). For child selfreport, the average age of the 40 boys (76.9 %) and 11 girls (21.2 %; missing, 1) was years (SD, 14.45). The sample was heterogeneous with respect to race/ ethnicity, with 28 (27.5 %) white non-hispanic, 47 (46.1 %) Hispanic, 6 (5.9 %) black non-hispanic, 2 (2.0 %) Asian/Pacific Islander, 1 (1.0 %) American Indian or Alaskan Native, 14 (13.7 %) other, and 4 (3.9 %) missing. Mean SES was unavailable for this sample. Comparisons between this healthy sample and the oncology sample indicate that the ratio of males to females was different in the 2 samples, with 69 (67.6%) males and 30 (29.4%) females in the healthy sample in comparison to the oncology sample, with 196 (58%) males and 143 (42%) females (chi-square (1) 4.53; P 0.05). There was no statistically significant difference in age between this healthy sample and the oncology sample. Measures The PedsQL 4.0 Generic Core Scales The 23-item multidimensional PedsQL 4.0 Generic Core Scales encompass 4 scales: 1) physical functioning (8 items), 2) emotional functioning (5 items), 3) social functioning (5 items), and 4) school functioning (5 items). The PedsQL 4.0 Generic Core Scales are comprised of parallel child self-report and parent proxy-report formats. Child self-report includes ages 5 7 years (young child), ages 8 12 years (child), and ages years (adolescent). Parent proxy-report includes ages 2 4 years (toddler), 5 7 years (young child), 8 12 years (child), and years (adolescent). The parent proxy-report forms are designed to assess the parent s perceptions of their child s HRQOL. The items for each of the forms are essentially identical, differing in developmentally appropriate language, or first or third person tense. The instructions for the Standard Version ask how much of a problem each item has been during the past 1 month. For the purposes of the current investigation, we utilized the PedsQL Acute Version (past 7 days recall interval). A 5-point Likert response scale is utilized across child self-report for ages 8 18 years and parent proxy-report (0 never a problem; 1 almost never a problem; 2 sometimes a problem; 3 often a problem; 4 almost always a problem). To further increase the ease of use for the young child self-report (ages 5 7 years), the Likert scale is reworded and simplified to a 3-point scale (0 not at all a problem; 2 sometimes a problem; 4 a lot of a problem), with each response choice anchored to a happy to sad faces scale. 29, 30 Parent proxy-report also includes the toddler age range (ages 2 4 years), which does not include a self-report form given developmental limitations on self-report for children younger than 5 years of age, 30, 31 and includes only 3 items for the school functioning scale. Items are reverse-scored and linearly transformed to a0 100 scale (0 100, 1 75, 2 50, 3 25, 4 0), so

5 2094 CANCER April 1, 2002 / Volume 94 / Number 7 TABLE 1 Scale Descriptives for PedsQL 4.0 Generic Core Scales Child Self-Report and Parent Proxy-Report and Comparisons with Healthy Children Scores a Scale No. of items Oncology sample Healthy sample (standard version) Healthy sample (acute version) n Mean SD n Mean SD n Mean SD Self-report Total score Physical health Psychosocial health Emotional functioning Social functioning School functioning Proxy-report Total score Physical health Psychosocial health Emotional functioning Social functioning School functioning SD: standard deviation. a The Standard Version of the PedsQL Generic Core Scales has a 1-month recall interval in the directions for both child self-report and parent proxy-report. The Acute Version of the PedsQL Generic Core Scales has a 7-day recall interval in the directions for both child self-report and parent proxy-report. All scale scores for the Oncology Sample are significant different from the Healthy Sample (Standard Version) and the Healthy Sample (Acute Version) for both child self-report and parent proxy-report (P 0.001). There are no significant differences between the Healthy Sample Standard Version and the Healthy Sample Acute Version, except for parent proxy-report Physical Health Scale Score (P 0.001) and parent proxy-report Social Functioning (P 0.05). TABLE 2 Scale Descriptives for PedsQL Multidimensional Fatigue Scale: Child Self-Report and Parent Proxy-Report and Comparisons with Healthy Children Scores Scale No. of items Oncology sample Healthy sample (acute version) n Mean SD n Mean SD Difference t test Self-report Total fatigue a General fatigue a Sleep/rest fatigue b Cognitive fatigue c Proxy-report Total fatigue a General fatigue a Sleep/rest fatigue a Cognitive fatigue a SD: standard deviation. a P b P c P that higher PedsQL 4.0 scores indicate better HRQOL. Scale scores are computed as the sum of the items divided by the number of items answered (this accounts for missing data). If greater than 50% of the items in the scale are missing, the scale score is not computed. Imputing the mean of the completed items in a scale when 50% or more items are completed is generally the most unbiased and precise method to calculate a scale score. 32 For this study, greater than 99% of child and parent respondents were included in the scale score analyses after imputing missing values. The Physical Health Summary Score (eight items) is the same as the Physical Functioning Subscale. To create the Psychosocial Health Summary Score (15 items), the mean is com-

6 The PedsQL/Varni et al TABLE 3 Scale Descriptives for PedsQL 3.0 Cancer Module (Acute Version) Child Self-Report and Parent Proxy-Report Scale No. of items n Mean SD Self-report Pain and hurt Nausea Procedural anxiety Treatment anxiety Worry Cognitive problems Perceived physical appearance Communication Proxy-report Pain and hurt Nausea Procedural anxiety Treatment anxiety Worry Cognitive problems Perceived physical appearance Communication SD: standard deviation. puted as the sum of the item responses divided by the number of items answered in the Emotional, Social, and School Functioning Subscales. PedsQL Multidimensional Fatigue Scale The 18-item PedsQL Multidimensional Fatigue Scale Acute Version encompasses 3 subscales: 1) general fatigue (6 items), 2) sleep/rest fatigue (6 items), and 3) cognitive fatigue (6 items). The format, instructions, Likert response scale, and scoring method are identical to the PedsQL 4.0 Generic Core Scales Acute Version, with higher scores indicating better HRQOL (fewer problems or symptoms). The PedsQL Multidimensional Fatigue Scale was developed based on our research and clinical experiences in pediatric cancer, and the instrument development literature, which consisted of a review of the extant literature on fatigue in adult and pediatric cancer, patient and parent focus groups and individual focus interviews, item generation, cognitive interviewing, pretesting, and subsequent field testing of the new measurement instrument in two pediatric cancer centers. PedsQL 3.0 Cancer Module The 27-item multidimensional PedsQL 3.0 Cancer Module Acute Version encompasses 8 scales: 1) pain and hurt (2 items), 2) nausea (5 items), 3) procedural anxiety (3 items), 4) treatment anxiety (3 items), 5) worry (3 items), 6) cognitive problems (5 items), 7) perceived physical appearance (3 items), and 8) communication (3 items). The format, instructions, Likert response scale, and scoring method are identical to the PedsQL 4.0 Generic Core Scales Acute Version, with higher scores indicating better HRQOL (fewer problems or symptoms). The PedsQL 3.0 Cancer Module was developed based on our research and clinical experiences in pediatric cancer, and the instrument development literature Our development of the PedsQL 3.0 Cancer Module was informed by our instrument development research with the Pediatric Cancer Quality of Life Inventory, the PedsQL 1.0 Cancer Module, 19 and iterations of the PedsQL pretested in pediatric cancer. PedsQL Family Information Form The PedsQL Family Information Form, completed by parents, contains demographic information on the child and parents. It contains the SES information required to calculate the Hollingshead SES index. 28 Procedure Potential subjects were identified at each site through examination of the hematology/oncology patient master list, inpatient list, and clinic appointment schedules, as well as through discussions with the clinical team. The parents of pediatric cancer patients identified as possible study participants who were newly diagnosed on-treatment or who experienced recurrence on-treatment and were on site regularly were informed of the study by a graduate student research assistant. Alternatively, the parents of the patients meeting inclusion criteria were contacted by mail, with a follow-up telephone call, and asked to participate in the study after a discussion with the research assistant. Long-term survivors off-treatment initially were contacted by mail or during clinic visits. Given the infrequency of their clinic visits at the centers, most of the long-term survivors were contacted by mail with follow-up telephone calls. Written parental informed consent and child assent were obtained before participation in the study. Parents and children completed the PedsQL separately. One parent (85% mothers; 12% fathers; 3% other) completed the proxyreport version. A research assistant was available to answer questions regarding the parent self-administered instruments. A research assistant administered the PedsQL for the young child (ages 5 7 years) and was available to assist the self-administered instrument for the child (ages 8 12 years) and adolescent (ages years) after the instructions had been given and clarified. The PedsQL 4.0 Generic Core

7 2096 CANCER April 1, 2002 / Volume 94 / Number 7 TABLE 4 PedsQL 4.0 Generic Core Scales (Acute Version) Internal Consistency Reliability for Child Self-Report and Parent Proxy-Report by Age and Summary Score/Subscale Age group (yrs) Scale Toddler (2 4) Young child (5 7) Child (8 12) Adolescent (13 18) Total sample Self-report Total score NA Physical health NA Psychosocial health NA Emotional functioning NA Social functioning NA School functioning NA Proxy-report Total score Physical health Psychosocial health Emotional functioning Social functioning School functioning NA: not applicable. TABLE 5 PedsQL Multidimensional Fatigue Scale Internal Consistency Reliability for Child Self-Report and Parent Proxy-Report by Age and Summary Score/Subscale Age group (yrs) Scale Toddler (2 4) Young child (5 7) Child (8 12) Adolescent (13 18) Total sample Self-report Total fatigue NA General fatigue NA Sleep/rest fatigue NA Cognitive fatigue NA Proxy-report Total fatigue General fatigue Sleep/rest fatigue Cognitive fatigue NA: not applicable. Scales Acute Version, the PedsQL 3.0 Cancer Module Acute Version, and the PedsQL Multidimensional Fatigue Scale Acute Version were administered simultaneously, with parents completing the proxy-report version independently and at the same time that their child completed the self-report version. This research protocol was approved by the Institutional Review Boards at Children s Hospital and Health Center, San Diego and Children s Hospital Los Angeles. Statistical Analysis Feasibility or practicality was determined from the percentage of missing values. 36 Scale internal consistency reliability was determined by calculating Cronbach coefficient alpha. 37 Scales with reliabilities of 0.70 or greater are recommended for comparing patient groups, whereas a reliability criterion of 0.90 is recommended for analyzing individual patient scale 38, 39 scores. Construct validity was determined utilizing the known-groups method. The known-groups method compares scale scores across groups known to differ in the health construct being investigated. In this study, PedsQL 4.0 Generic Core Scales scores in groups differing in known health condition (healthy children and children with cancer) were comput-

8 The PedsQL/Varni et al TABLE 6 PedsQL 3.0 Cancer Module (Acute Version) Internal Consistency Reliability for Child Self-Report and Parent Proxy-Report by Age and Subscale Age group (yrs) Scale Toddler (2 4) Young child (5 7) Child (8 12) Adolescent (13 18) Total sample Self-report Pain and hurt NA Nausea NA Procedural anxiety NA Treatment anxiety NA Worry NA Cognitive problems NA Perceived physical appearance NA Communication NA Proxy-report Pain and hurt Nausea Procedural anxiety Treatment anxiety Worry Cognitive problems Perceived physical appearance Communication NA: not applicable. ed, 40,41 using t tests and one-way analysis of variance (ANOVA). The data for the healthy group of children were derived from the initial field trial of the PedsQL 4.0 Standard Version (1-month recall interval), 20 and a healthy sample that completed the Acute Version (7- day recall period). We hypothesized that healthy children would report higher PedsQL 4.0 scores (better HRQOL) than pediatric patients with cancer. In addition, PedsQL Multidimensional Fatigue Scale scores in groups differing in known health condition (healthy children and children with cancer) were computed, 40,41 using t tests and ANOVA. The data for the healthy group of children were derived from a healthy sample that completed the Acute Version (7-day recall period). We hypothesized that healthy children would report higher PedsQL scores (better HRQOL or less fatigue) than pediatric patients with cancer. Analyses of variance also were conducted to examine whether there were differences in PedsQL Generic Core Scales, Multidimensional Fatigue Scale, and Cancer Module scores among children with cancer on-treatment, offtreatment for 12 months or less, and off-treatment for more than 12 months. Construct validity was further examined through an analysis of the intercorrelations among the PedsQL 4.0 Generic Core Total Scale Score with the PedsQL Multidimensional Fatigue Total Scale and Subscales scores. It was hypothesized that higher Fatigue Total Scale and Subscale scores (fewer problems or symptoms) would be correlated with higher Generic Core Total Scale Scores (better overall HRQOL), based on the conceptualization of disease-specific symptoms as causal indicators of HRQOL. 1 Correlation effect sizes are designated as small ( ), medium ( ), and large ( 0.50). 42 Intercorrelations were expected to demonstrate medium to large effect sizes. 1 Parent/child intercorrelations were computed to examine cross-informant variance. Correlation effect sizes are designated as small ( ), medium ( ), and large ( 0.50). 42 Parent/child concordance for the Total Score, Summary Scores, and the same Subscale were expected to demonstrate medium to large effect sizes, but not so large that child and parent reports would be redundant. Statistical analyses were conducted using SPSS for Windows. 43 Response equivalence has been demonstrated previously across language (English vs. Spanish) for the PedsQL by examining the percentage of missing data, floor and ceiling effects, and scale internal consistency across language. 20 Therefore, responses were pooled across languages. Responses also were pooled across the age ranges for both self-report and proxy-report. RESULTS Missing Item Responses To assess the feasibility or practicality of administration for the PedsQL 4.0 Generic Core Scales Acute

9 2098 CANCER April 1, 2002 / Volume 94 / Number 7 TABLE 7 One-Way ANOVA Values Comparing Healthy Children and Children with Cancer (On- or Off-Treatment): PedsQL 4.0 Generic Core (Acute Version) Child Report Scale n Mean SD Difference df f P value Total score a c**, a, b d*** On Tx a Off Tx 12 b Off Tx 12 c Healthy d Physical health a c, d*** On Tx a Off Tx 12 b Off Tx 12 c Healthy d Psychosocial health a, b d***, c d* On Tx a Off Tx 12 b Off Tx 12 c Healthy d Emotional functioning a c*, a d***, b d** On Tx a Off Tx 12 b Off Tx 12 c Healthy d Social functioning a d**, b d*** On Tx a Off Tx 12 b Off Tx 12 c Healthy d School functioning a d*** On Tx a Off Tx 12 b Off Tx 12 c Healthy d ANOVA: analysis of variance; SD: standard deviation; On Tx, on-treatment sample; Off Tx 12: off-treatment 12 mos sample; Off Tx 12: off-treatment 12 mos/long-term survivor sample. * P 0.05, ** P 0.01, *** P based on Tukey Honestly Significantly Different post hoc analysis. Version, the percentage of missing values was calculated. For child self-report and parent proxy-report, the percentage of missing item responses for the oncology sample was 0.4% and 0.3%, respectively, for all scales except the School Functioning Scale. The percentage of missing items for the School Functioning Scale was 15% for child self-report (ages 5 18 years) and 38% for parent proxy-report (ages 2 18 years). In almost all of these cases, all of the items on the School Functioning Scale were not completed, suggesting that the patient did not attend school the previous 7 days (most of the data from the Los Angeles site were collected during summer vacation). For the PedsQL Multidimensional Fatigue Scale, the percentage of missing item responses was 0% for child self-report, and 0% for parent proxy-report. For the PedsQL 3.0 Cancer Module, the percentage of missing item responses was 0.5% for child self-report and 1.0% for parent proxy-report. Means and Standard Deviations Table 1 presents the means and standard deviations of the PedsQL 4.0 Generic Core Scales for children with cancer as a group and the healthy children population group from our previous field trial with the Standard Version (1-month recall interval) 20 and the healthy sample with the Acute Version (7-day recall interval). Table 2 contains the means and the standard deviations of the PedsQL Multidimensional Fatigue Scale for children with cancer as a group and the healthy sample Acute Version. Table 3 contains the means and standard deviations of the PedsQL 3.0 Cancer Module for the pediatric cancer group. Internal Consistency Reliability Internal consistency reliability alpha coefficients for the PedsQL 4.0 Generic Core Scales across all ages are presented in Table 4. The child self-report scales and

10 The PedsQL/Varni et al TABLE 8 One-Way ANOVA Values Comparing Healthy Children and Children with Cancer (On- or Off-Treatment): PedsQL 4.0 Generic Core (Acute Version) Parent Report Scale n Mean SD Difference df f P value Total score a c*, a, b, c d*** On Tx a Off Tx 12 b Off Tx 12 c Healthy d Physical health a c**, a d***, b d** On Tx a Off Tx 12 b Off Tx 12 c Healthy d Psychosocial health a, b, c d*** On Tx a Off Tx 12 b Off Tx 12 c Healthy d Emotional functioning a b*, a c***, a, b, c d*** On Tx a Off Tx 12 b Off Tx 12 c Healthy d Social functioning a, b, c d*** On Tx a Off Tx 12 b Off Tx 12 c Healthy d School functioning a, b, c d*** On Tx a Off Tx 12 b Off Tx 12 c Healthy d ANOVA: analysis of variance; SD: standard deviation; On Tx: on-treatment sample; Off Tx 12: off-treatment 12 mos sample; Off Tx 12: off-treatment 12 mos/long-term survivor sample. * P 0.05, ** P 0.01, *** P based on Tukey Honestly Significantly Different post hoc analysis. parent proxy-report scales approached or exceeded the minimum reliability standard of 0.70, 38 except for 3 young child (ages 5 7 years) self-report scales. The PedsQL 4.0 Generic Core Scales Total Score across the ages approached or exceeded the reliability criterion of 0.90 recommended for analyzing individual patient 38, 39 scale scores. Table 5 contains the internal consistency reliability coefficients for the PedsQL Multidimensional Fatigue Scale. All child self-report and parent proxyreport scales approached or exceeded the minimum reliability standard of The PedsQL Multidimensional Fatigue Scale Total Score across the ages approached or exceeded the reliability criterion of 0.90 recommended for analyzing individual patient scale scores. 38,39 Table 6 contains the internal consistency reliability coefficients for the PedsQL 3.0 Cancer Module. All of the parent proxy-report scales met or exceeded the minimum reliability standard of The child selfreport results were more variable. All of the adolescent (ages years) self-report scales met or exceeded the standard of Six of the 8 child (ages 8 12 years) self-report scales met or exceeded the standard of Two of the young child (ages 5 7 years) selfreport scales met or exceeded the standard. Construct Validity Table 1 demonstrates the comparisons between the PedsQL 4.0 Generic Core Scales for the healthy children groups and children with cancer as a group. For every comparison, there was a statistically significant difference between healthy children and children with cancer as a group. The hypothesis was confirmed that healthy children as a group would manifest higher PedsQL 4.0 Generic Core Scales scores than children with cancer as a group. Tables 7 and 8 display the one-way ANOVA values comparing healthy children

11 2100 CANCER April 1, 2002 / Volume 94 / Number 7 TABLE 9 One-Way ANOVA Values Comparing Healthy Children and Children with Cancer (On- or Off-Treatment): Child Report Multidimensional Fatigue Scale (Acute Version) Scale n Mean SD Difference df f P value Total fatigue a d***, b d* On Tx a Off Tx 12 b Off Tx 12 c Healthy d General fatigue a c*, a d*** On Tx a Off Tx 12 b Off Tx 12 c Healthy d Sleep/rest fatigue a c*, a d* On Tx a Off Tx 12 b Off Tx 12 c Healthy d Cognitive fatigue a d* On Tx a Off Tx 12 b Off Tx 12 c Healthy d ANOVA: analysis of variance; SD: standard deviation; On Tx: on-treatment sample; Off Tx 12: off-treatment 12 mos sample; Off Tx 12: off-treatment 12 mos/long-term survivor sample. * P 0.05, ** P 0.01, *** P based on Tukey Honestly Significantly Different post hoc analysis. with children with cancer on-treatment or off-treatment with the PedsQL Generic Core Scales. For child self-report, the PedsQL 4.0 Generic Core Scales Total Score demonstrated significant differences between the healthy population group and children with cancer on-treatment (Table 7). For parent proxy-report, the PedsQL 4.0 Generic Core Scales Total Score demonstrated significant differences between the healthy population group and children with cancer on-treatment and off-treatment (Table 8). The analyses among children with cancer revealed that the most consistent group differences were observed primarily between children with cancer on-treatment versus off-treatment greater than 12 months (Tables 7 and 8). For the PedsQL Multidimensional Fatigue Scale, the Total Scale Score and the Subscale scores demonstrated significant differences between the healthy population group and children with cancer as a group (Table 2). For every comparison, there was a statistically significant difference between healthy children and children with cancer as a group. The hypothesis was confirmed that healthy children as a group would manifest higher PedsQL scores (better HRQOL or less fatigue) than children with cancer as a group. Tables 9 and 10 display the one-way ANOVAs comparing healthy children with children with cancer on-treatment or off-treatment with the PedsQL Multidimensional Fatigue Scale. For child self-report, the PedsQL Multidimensional Fatigue Scale Total Score demonstrated significant differences between the healthy population group and children with cancer on-treatment (Table 9). For parent proxy-report, the PedsQL Multidimensional Fatigue Scale Total Score demonstrated significant differences between the healthy population group and children with cancer on-treatment and off-treatment (Table 10). The analyses among children with cancer revealed that the most consistent group differences were observed primarily between children on-treatment versus off-treatment greater than 12 months (Tables 9 and 10). For the PedsQL 3.0 Cancer Module Scales, the analyses among children with cancer revealed that for child-self report, group differences were observed on the Nausea, Treatment Anxiety, and Worry Scales between children on-treatment versus off-treatment greater than 12 months (Table 11). For parent proxyreport, group differences were observed on the Pain, Nausea, Procedural Anxiety, Treatment Anxiety, and Worry Scales between children on-treatment versus off-treatment (Table 12). The intercorrelations between the PedsQL 4.0 Generic Core Scales Total Score and the PedsQL Multidimensional Fatigue Total Scale Score and Subscales are shown in Table 13. As anticipated, the correlations are in the medium to large effect size range, with the largest intercorrelations between the Multidimen-

12 The PedsQL/Varni et al TABLE 10 One-Way ANOVA Values Comparing Healthy Children and Children with Cancer (On- or Off-Treatment): Parent Report Multidimensional Fatigue Scale (Acute Version) Scale n Mean SD Difference df f P value Total fatigue a c*, a, b, c d*** On Tx a Off Tx 12 b Off Tx 12 c Healthy d General fatigue a b**, a c, d***, b, c d** On Tx a Off Tx 12 b Off Tx 12 c Healthy d Sleep/rest fatigue a b*, a c, d***, b d** On Tx a Off Tx 12 b Off Tx 12 c Healthy d Cognitive fatigue a, c d***, b d* On Tx a Off Tx 12 b Off Tx 12 c Healthy d ANOVA: analysis of variance; SD: standard deviation; On Tx: on-treatment sample; Off Tx 12: off-treatment 12 mos sample; Off Tx 12: off-treatment 12 mos/long-term survivor sample. * P 0.05, ** P 0.01, *** P based on Tukey Honestly Significantly Different post hoc analysis. sional Fatigue Scale Total Score with the Generic Core Scales Total Score for both child and parent report. The intercorrelations between the Multidimensional Fatigue Scale Total Score and the Cancer Module Scales were also in the medium to large effect size range. Parent/Child Concordance The parent/child concordance intercorrelations matrix is shown in Table 13. Consistent with the extant literature, child self-report and parent proxy-report correlations are in the medium to large effect size range. DISCUSSION This study presents the measurement properties for the PedsQL 4.0 Generic Core Scales, PedsQL Multidimensional Fatigue Scale, and PedsQL 3.0 Cancer Module in pediatric cancer. The analyses support the reliability and validity of the PedsQL as a child self-report and parent proxy-report HRQOL measurement instrument for pediatric cancer. The PedsQL is the only empirically validated generic and cancer specific HRQOL measurement instrument that we are aware of to span this broad age range for child self-report and parent proxy-report while maintaining item and scale construct consistency. Items on the PedsQL had minimal missing responses, suggesting that children and parents are willing and able to provide good quality data regarding the child s HRQOL. The PedsQL self-report and proxyreport internal consistency reliabilities generally exceeded the recommended minimum alpha coefficient standard of 0.70 for group comparisons. Across the ages, the PedsQL 4.0 Generic Core Scales Total Score for both child self-report and parent proxy-report approached or exceeded an alpha of 0.90, recommended for individual patient analysis, 38 making the Total Scale Score suitable as a summary score for the primary analysis of HRQOL outcome in clinical trials and other group comparisons. The Physical Health and Psychosocial Health Summary Scores are recommended for secondary analyses. The Emotional, Social, and School Functioning Subscales may be utilized to examine specific domains of functioning, with the caveat that until further testing is conducted, the Social and School Functioning Subscales should be used only for descriptive or exploratory analyses for young children ages 5 7 years. The PedsQL Multidimensional Fatigue Scale Total and Subscale internal consistency reliabilities demonstrated the recommended minimum alpha coefficient standard of 0.70 for group comparisons for child selfreport ages 5 18 years and parent proxy-report ages

13 2102 CANCER April 1, 2002 / Volume 94 / Number 7 TABLE 11 One-Way ANOVA Values Comparing Children On- or Off-Treatment: Child Report Cancer Module (Acute Version) Scale n Mean SD Difference df f P value Pain and hurt On Tx a Off Tx 12 b Off Tx 12 c Nausea a c*** On Tx a Off Tx 12 b Off Tx 12 c Procedural anxiety On Tx a Off Tx 12 b Off Tx 12 c Treatment anxiety a c* On Tx a Off Tx 12 b Off Tx 12 c Worry a c* On Tx a Off Tx 12 b Off Tx 12 c Cognitive problems On Tx a Off Tx 12 b Off Tx 12 c Perceived physical appearance On Tx a Off Tx 12 b Off Tx 12 c Communication On Tx a Off Tx 12 b Off Tx 12 c ANOVA: analysis of variance; SD: standard deviation; On Tx: on-treatment sample; Off Tx 12: off-treatment 12 mos sample; Off Tx 12: off-treatment 12 mos/long-term survivor sample. * P 0.05, ** P 0.01, *** P based on Tukey Honestly Significantly Different post hoc analysis years. Across the ages, the PedsQL Multidimensional Fatigue Scale Total Score for both child selfreport and parent proxy-report approached or exceeded an alpha of 0.90, recommended for individual patient analysis, 38 making the PedsQL Multidimensional Fatigue Scale Total Score suitable as a summary score for the primary analysis of HRQOL fatigue outcome in clinical trials and other group comparisons. The General Fatigue, Sleep/Rest Fatigue, and Cognitive Fatigue Subscales may be utilized to examine specific dimensions of fatigue, and are recommended for secondary analyses. The PedsQL 3.0 Cancer Module Scales internal consistency reliabilities generally exceeded the recommended minimum alpha coefficient standard of 0.70 for group comparisons for child self-report ages 8 18 years and parent proxy-report ages 2 18 years. For young child self-report ages 5 7 years, only the Procedural Anxiety and Treatment Anxiety Scales met the 0.70 standard for group comparisons. The Perceived Appearance and the Communication Scales did not meet the 0.70 standard for child self-report ages 8 12 years. Until further testing is conducted, the child self-report scales that did not achieve the standard of 0.70 should be used only for descriptive or exploratory analyses. For parent proxy-report, reliability for all Cancer Module Scales met or exceeded the 0.70 standard and for most Scales were in the 0.80 to 0.90 range. The PedsQL 4.0 Generic Core Scales performed as hypothesized utilizing the known-groups method. The PedsQL 4.0 differentiated HRQOL in healthy children as a group in comparison to children with cancer as a group. The PedsQL Multidimensional Fatigue Scale differentiated fatigue in healthy children as a group in comparison to children with cancer as a group. The intercorrelations among the PedsQL Multidimen-