Vol. 40 No. 4 October 2010 Journal of Pain and Symptom Management 559 Original Article Validation of the Pediatric Functional Assessment of Cancer Therapy Questionnaire (Version 2.0) in Brain Tumor Survivors Aged 13 Years and Older Heejung Yoo, PhD, Dong-Seok Kim, MD, PhD, Hee-Young Shin, MD, PhD, Jin-Shei Lai, PhD, David Cella, PhD, Hyeon-Jin Park, MD, PhD, Young-Shin Ra, MD, PhD, Woo-Chul Kim, MD, PhD, and Yong-Soon Shin, PhD Department of Neurosurgery Ulsan University College of Medicine (H.Y., Y.-S.R., Y.-S.S.), Pediatric Neurosurgery, Severance Children s Hospital, Yonsei University College of Medicine (D.-S.K) and Department of Pediatrics (H.-Y.S.), Cancer Research Institute, Seoul National University College of Medicine, Seoul, Republic of Korea; Medical Social Sciences (J.-S.L., D.C.), Feinberg School of Medicine at Northwestern University, Chicago, Illinois, USA; Center for Pediatric Cancer (H.-J.P.), National Cancer Center, Ilsan; and Department of Radiation Oncology (W.-C.K.), Inha University Hospital, Inchon, Republic of Korea Abstract We evaluated the reliability and validity of the Pediatric Functional Assessment of Cancer Therapy-Childhood Brain Tumor Survivor Questionnaire (pedsfact-brs, Version 2). This was specifically directed to patients aged 13 years and older (adolescents). The pedsfact-brs was translated and cross-culturally adapted into Korean, following standard Functional Assessment of Chronic Illness Therapy methodology. The psychometric properties of the pedsfact-brs in adolescents were evaluated in 161 brain tumor (BT) patients (mean age ¼ 15.53 years). Pretesting was performed in 30 patients, and the results indicated good symptom coverage and overall comprehensibility. In validating the pedsfact-brs for adolescents, we found high internal consistency, with Cronbach s a coefficients ranging from 0.76 to 0.91. The pedsfact-brs for adolescents also demonstrated good convergent and divergent validities when correlated with the Revised Children s Manifest Anxiety Scale and the Kovacs Children s Depression Inventory. The pedsfact-brs for adolescents showed good clinical validity and effectively differentiated between clinically distinct patient groups according to Karnofsky score, type of treatment, and treatment on/off status. This reliable and valid instrument can now be used to properly evaluate the quality of life of Korean adolescent BT patients. J Pain Symptom Manage 2010;40:559e565. Ó 2010 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved. Key Words Quality of life, adolescent brain tumor, pedsfact-brs for adolescents, validity This study was supported by a grant from the national R&D Program for Cancer Control, Ministry of Health, Welfare, and Family Affairs, Republic of Korea (No. 0520300). Address correspondence to: Dong-Seok Kim, MD, PhD, Pediatric Neurosurgery, Severance Children s Ó 2010 U.S. Cancer Pain Relief Committee Published by Elsevier Inc. All rights reserved. Hospital, Yonsei University College of Medicine, 250 Seongsanno, Seodaemoon-Gu, Seoul, 120-752, Korea. E-mail: dskim33@yuhs.ac or dskim33@ yumc.yonsei.ac.kr Accepted for publication: January 27, 2010. 0885-3924/$ - see front matter doi:10.1016/j.jpainsymman.2010.01.024
560 Yoo et al. Vol. 40 No. 4 October 2010 Introduction For quality-of-life (QOL) assessment of various pediatric cancer patients, including pediatric brain tumor (BT) patients, the Pediatric Quality of Life InventoryÔ (PedsQLÔ), 1,2 the Child Health Questionnaire, 3 and the QOL in Childhood Cancer (a Taiwanese instrument) are available for child and adolescent cancer patients. 4,5 These tools assess general well-being of the child cancer patient. However, more diseasespecific measurement tools are required, especially in clinical settings with child BT patients. Palmer et al. 6 developed the 24-item PedsQLÔ BT module, and Lai et al. 7 developed the 34- item Patient Version (ages 7e12 years; grade school) of the Pediatric Functional Assessment of Cancer Therapy-BT Survivor (pedsfact- BrS), which has been extended to adolescents since then. The pedsfact-brs has the advantages of brevity; instrument practicality (e.g., length of questionnaire, language suitability, age appropriateness); and ease of administration, particularly in busy clinical settings. 8 Nevertheless, when completing the pedsfact-brs, the challenges faced by older children are different from those of younger survivors. 7 In several studies on QOL of child cancer patients, 2,9,10 age-specific data were collected with the same instrument. A complicating factor is that overall QOL, and functional and psychosocial capacities, are influenced by age-specific changes occurring during childhood and adolescence. 4,11 As Hinds and Hockenberry-Eaton reported, 12 patients 7e12 years old emphasize physical fatigue, whereas the essential fatigue characteristics described by patients 13e18 years old are not only physical but also mental or emotional exhaustion, or a combination of both forms. It is likely that children and adolescents will have different concerns with respect to health-related QOL. 8 Lai et al. 7 also suggested that future studies were necessary to evaluate pedsfact-brs utility in BT survivors of older ages. Lai et al. 7 have revised the pedsfact-brs items and extended it to adolescents by means of a series of cognitive interviews with 20 patients (five medulloblastoma; three ependymoma; three craniopharyngioma; two low-grade astrocytoma; two germinoma; and one each of the following tumor types: oligoglioma, germ cell BT, hypothalamic astrocytoma, primitive neuroectodermal tumor; and pineal blastoma) and their parents. Average patient age was 14.4 years (standard deviation [SD]: 4.4), and average years since diagnosis was 5.3 years (SD: 3.6). Four items were added specifically for this age group: I worry about being able to go to college because of my illness history, I worry about being able to date because of my illness history, I worry about getting a job to support myself because of my illness history, and I often feel that other people are better than me. The items in other domains are the same between child and adolescent versions. To date, few studies have evaluated the validity of any BT-specific tool assessing the QOL of adolescent BT patients. To examine whether the pedsfact-brs for adolescents is an efficient instrument for this purpose, a validity study was needed. Therefore, we took the first important step toward an improved understanding of adolescent BT patients QOL. First, we crossculturally adapted the pedsfact-brs for adolescents into Korean and then evaluated the reliability and validity of the resulting instrument. Methods Subjects We enrolled 161 patients aged 13e18 years (Table 1). As previous childhood studies contained only small numbers of cases, we decided to perform a multicenter study. This study was approved by the Institutional Review Boards of all the centers. The patients were recruited from Asan Medical Center (70 adolescent patients), Severance Children s Hospital (60 adolescent patients), Seoul National University Hospital (23 adolescent patients), and the National Cancer Center (eight adolescent patients). Informed consent was obtained from the subject(s) and/or guardian(s). All patients returning for routine follow-up appointments were consecutively asked to enroll. In addition, patients who were not scheduled for follow-up appointments during the study period were enrolled initially by mail and subsequently by telephone, between December 2006 and July 2008. Subjects were excluded if they had pre-existing cognitive impairment, psychiatric disorders, or hearing loss. Informed consent was obtained from all
Vol. 40 No. 4 October 2010 QOL Assessment of Adolescent Brain Tumor Patients 561 Table 1 Demographic Characteristics of 161 Adolescent Brain Tumor Patients Variables Mean (SD) Age, mean (SD), years 15.53 (1.95) Education 9.53 (2.34) Age at onset 12.69 (3.36) Age of caregiver (mother) 44.04 (3.73) Years of education (mother) 13.08 (2.11) Number of Variables Cases (%) Gender (patient) Male 94 (58.4) Female 67 (41.6) Pathology Medulloblastoma/Primitive 21 (13.0) Neuroectodermal Tumor (PNET) Malignant glioma 15 (9.3) Low-grade glioma 33 (20.5) Other low-grade neoplasm 48 (29.8) Germ cell tumor 44 (27.3) Treatment received Surgery only 70 (43.5) Surgery þ chemotherapy þ radiotherapy 44 (27.3) Surgery þ radiotherapy 47 (29.2) Location Supratentorial 104 Infratentorial 54 Shunt Yes 8 No 145 Currently receiving treatment Yes 27 (16.8) No 134 (83.2) participants. The frequencies of tumor pathologic subtypes were similar to those reported for the American population 9,13 and for earlier Korean populations, 14 except that the proportion of germ cell tumor patients (27.3%) was particularly high in our sample. Assessment Tool Pediatric Functional Assessment of Cancer Therapy- Childhood Brain Tumor Survivor Questionnaire for Adolescents. The pedsfact-brs for adolescents consists of 37 items: 25 generic concerns (seven on physical well-being [PWB], 13 on emotional well-being and illness experiences [EWB], five on social or family well-being [SFWB]), and 12 BT survivor-specific concerns (the BTS questions). 7 Development of the Korean Pediatric Functional Assessment of Cancer Therapy-Childhood Brain Tumor Survivor Questionnaire for Adolescents. The pedsfact-brs for adolescents was translated using standard FACIT methodology. 15 A provisional version of the instrument was pretested on 30 patients in total, five patients each according to their age between 13 and 18 years during their follow-up clinic visits. Patients were asked to self-administer the questionnaires. Subjects were asked to write their opinions on the following questions, which were asked to clarify what and how many items made patients discomforted. 1. Would you please tell me which items were difficult to understand and why they were difficult? Also, could you suggest a better way to phrase these items? 2. Would you please tell me which items were not relevant or were offensive, and why? Also, could you suggest a better way to phrase these items? 3. Is there anything else that should have been included related to your condition? Would you please tell me what should be added? Reliability Internal consistency coefficients were calculated for each subscale. We performed testretests at intervals of 7e10 days in 30 patients among the 161 subjects who agreed to complete the questionnaire twice. Clinical Validity We determined the discrimination of the pedsfact-brs by comparing the pedsfact- BrS scores with Karnofsky scores, treatment type, and treatment on or off status. Convergent and Divergent Validities Convergent and divergent validities were assessed by comparing patient responses with those on the Revised Children s Manifest Anxiety Scale (RCMAS) 16,17 and Kovacs Children s Depression Inventory (CDI). 18 The RCMAS is a self-reported 37-item questionnaire that assesses anxiety in school-aged children. 16 Cronbach s a was 0.81 in this study. The CDI is a symptom-oriented depression scale suitable for self-reporting by school-aged children (Kovacs, 1983, personal communication). Cronbach s a was 0.72 in this comparison. The RCMAS and CDI scales were used as measures of a related concept (mood). Correlations with depression and anxiety were expected to be higher when EWB answers were
562 Yoo et al. Vol. 40 No. 4 October 2010 evaluated and to be lower when SFWB and PWB data were analyzed, because depression and anxiety are both measures of mood. Procedures The pedsfact-brs was administered verbally by research nurses to ensure that questions were always posed in the same manner. A research assistant was available to answer questions posed by patients completing self-administered instruments. All patients completed the questionnaire within about 15e20 minutes. Statistical Analyses In regression models, age at interview, gender, and age at diagnosis were included as covariates. We calculated descriptive statistical composite scores for the overall questionnaire and all subscales, using the mean score of each scale for missing data and with substitution of a mean subscale score for missing items within a subscale. No more than 20% of items within a subscale were missing. 19 Pearson productmoment correlation was used to determine the convergent and divergent validities of pedsfact-brs, including assessment of the overall scale, subscales, and test-retest reliability. The independent-sample t-test and oneway analysis of variance were used to determine whether the test was able to discriminate across Karnofsky score, type of treatment, and receiving-treatment status. Statistical analysis was conducted using standard SPSS Statistical Analysis System Version 15.0 software (SPSS Inc., Chicago, IL). Results The proportions of missing data were 1.8% overall, 2.2% for anxiety items, and 2.3% for depression questions. No subscale was completely omitted by any patient. Pretesting All 30 patients had reported that there was no difficult, relevant, or offensive item in pedsfact- BrS. They suggested no items to be included related to their conditions. Cronbach s a coefficient showed acceptable internal consistency (0.7), suggesting that translated items performed well in association with other items in the same subscale. Reliability All Cronbach s a coefficients were substantially greater than 0.70, indicating internal consistency and reliability. Test-retest reliability yielded correlation coefficients between 0.81 for the SFWB and 0.94 for the PWB subscales (Table 2). Validity There were no significant differences between patients assessed in the clinic or at home. None of age, time from diagnosis, or gender showed significant associations with total QOL. Neither tumor location nor pathological diagnosis was significantly associated with total QOL. Convergent and Divergent Validities It was hypothesized that the EWB would be significantly correlated with depression and anxiety scores (convergent validity), whereas SFWB and PWB would be associated to a lesser degree with such assessments (divergent validity). As expected, correlations among pedsfact-brs, anxiety, and depression subscales showed that, essentially, the same phenomenon was being measured. In particular, anxiety and depression scores were correlated well with EWB scores (convergent validity) and poorly with SWB and PWB (divergent validity) (Table 3). Table 2 Means, Standard Deviations, Internal Consistencies, and Test-Retest Reliabilities Variables Mean (SD) Range Cronbach s a Test-Retest Reliability PWB (7 questions) 22.10 (5.36) 5e28 0.86 0.88 SFWB (5 questions) 16.25 (4.25) 5e23 0.78 0.81 EWB (13 questions) 33.86 (9.28) 8e43 0.76 0.90 BTS (12 questions) 35.89 (8.67) 7e48 0.83 0.87 PSF (18 questions) 50.11 (11.79) 14e62 0.81 0.83 pedsfact-brs (37 questions) 108.14 (21.98) 31e121 0.91 0.94 PSF ¼ psychosocial function (SFWB þ EWB).
Vol. 40 No. 4 October 2010 QOL Assessment of Adolescent Brain Tumor Patients 563 Table 3 Correlations of pedsfact-brs Answers With Anxiety and Depression Variables SFWB EWB BTS PedsFACT-BrS Anxiety Depression PWB 0.47 0.50 0.53 0.73 0.31 0.24 SFWB 1 0.50 0.72 0.78 0.46 0.34 EWB 1 0.57 0.83 0.51 0.50 BTS 1 0.87 0.49 0.44 pedsfact-brs 1 0.58 0.50 Known-Group Comparison We classified patients into three groups based on Karnofsky score: Karnofsky scores of 100, 90, or 80. Overall questionnaire data (P < 0.001) and subscale results enabled discrimination by Karnofsky score. Analysis by treatment type revealed significant differences in PWB (P < 0.05), EWB (P < 0.05), and overall scores (P < 0.05). A significant difference between treatment-on and treatment-off status was seen in only the PWB subscale (P < 0.001). Discussion The results presented here show the good psychometric properties of the Korean version of the pedsfact-brs for adolescents, including good convergent and divergent validities and known-group validity. Correlations greater than 0.40 have been considered satisfactory for convergent validity. Conversely, for adequate divergent validity, scales measuring different constructs should show low correlations, well less than 0.40. 20 As predicted, correlations between EWB and anxiety (r ¼ 0.55) and depression (r ¼ 0.51) were high, reflecting measurement of the same phenomenon. In contrast, depression (r ¼ 0.24) was low, indicating that different phenomena were being addressed. Also, overall high correlations were observed with anxiety (r ¼ 0.58) and depression (r ¼ 0.50), because the total score contained EWB data. Known-group validities were confirmed by Karnofsky score, type of treatment, and receiving-treatment status (Table 4). With respect to treatment type, the surgery-only group showed the highest scores in the PWB and EWB subscales and on overall pedsfact-brs. The mean scores in the SFWB and BTS subscales tended to be higher in surgery-only patients, whereas surgery and radiotherapy patients had somewhat lower scores, but the difference was not statistically significant. Although our results were similar to those of Bhat et al., 9 we did not have a no-treatment group or a radiotherapy-only group in our Table 4 Differences in pedsfact-brs Scores According to Karnofsky Scores, Treatment Type, and Treatment On/Off Status Variables N PWB SFWB EWB BTS pedsfact-brs Karnofsky 100 50 24.60 (2.98) 17.42 (3.56) 35.48 (9.59) 39.00 (7.21) 116.50 (18.04) 90 36 21.67 (5.09) 16.08 (3.42) 32.89 (8.29) 36.06 (8.91) 105.69 (19.47) 80 25 16.16 (6.14) 15.28 (4.70) 30.84 (9.57) 32.84 (9.75) 95.12 (23.39) F-Value 28.45 c (1 > 2 > 3) 2.97 a (1, 2 > 3) 2.28 5.49 b (1, 2 > 3) 10.13 c (1, 2 > 3) Treatment type Surgery 70 23.51 (4.23) 16.89 (3.67) 36.16 (7.81) 37.49 (7.29) 114.04 (17.44) S þ R 30 21.90 (5.16) 15.87 (4.47) 30.93 (11.57) 35.19 (9.24) 104.86 (22.43) S þ R þ C 59 20.79 (6.14) 15.73 (3.87) 33.02 (8.67) 34.33 (9.35) 102.90 (25.47) F-Value 4.46 a (1 > 2 > 3) 1.39 4.16 a (1, 3 > 2) 1.93 4.38 a (1 > 3, 2) Treatment On 27 19.26 (5.73) 16.22 (4.32) 31.33 (9.17) 36.41 (7.78) 103.22 (22.57) Off 134 21.87 (4.98) 16.29 (4.00) 34.63 (9.03) 35.93 (8.70) 109.72 (21.21) T-Value 3.34 b 0.08 1.72 0.26 1.43 S þ R ¼ surgery þ radiotherapy; S þ R þ C ¼ surgery þ radiotherapy þ chemotherapy. a P < 0.05. b P < 0.01 c P < 0.001.
564 Yoo et al. Vol. 40 No. 4 October 2010 study. We could not compare our results directly with any previous report, because no studies on the QOL of adolescent-only BT patients have been conducted. Therefore, further research is needed to compare the QOL of BT patients according to treatment type (surgery only, radiation only, and other treatments). In this study, receiving-treatment status significantly affected only the PWB subscale. In a recent patient proxy report, child BT patients who had fully recovered from the acute toxic effects of surgery, radiotherapy, and chemotherapy showed dramatic improvements in health-related QOL scores. 1 The number of treatment-on patients in the current study was small (27 patients; 16.8%). If numbers in the treatment-on group were increased and analyzed by duration of treatment, 1,2 we would expect to see significant differences in subscales other than the PWB. The limitation of this study was the higher proportion of germ cell tumor patients (27.3%) in our BT sample than Americans. Primary intracranial germ cell tumors make up approximately 0.1%e3.4% of all intracranial tumors in the West. However, the incidence in the East is 2.1%e9.4%, which is noticeably higher than that in the West. Especially, the incidence of germ cell tumors in Korea and Japan is much higher than that of other Eastern countries. 21 In this study, there was no difference in the mean scores of subscales (PWB: P ¼ 0.06; SFWB: P ¼ 0.23; EWB: P ¼ 0.09; BTS: P ¼ 0.39; and pedsfact-brs total scores: P ¼ 0.10) in each of the five pathology types. Conclusion The present results suggest that the Korean translation of the pedsfact-brs for adolescents is a reliable and valid instrument for measuring the QOL of Korean BT patients. However, there are continuing problems in this area related to data collection and heterogeneity in tumor types, locations, and treatments. 9 We did not evaluate patient sensitivity to QOL changes in long-term follow-up. Further research is needed not only to determine whether the questionnaire remains efficient and sensitive to detect longitudinal QOL changes in adolescent BT patients overall but also whether it is suitable for detecting longitudinal QOL changes within homogeneous patient groups, such as those with medulloblastoma, or according to treatment type. References 1. Meeske K, Katz ER, Palmer SN, Burwinkle T, Varni JW. Parent proxy-reported health-related quality of life and fatigue in pediatric patients diagnosed with brain tumors and acute lymphoblastic leukemia. Cancer 2004;101:2116e2125. 2. Varni JW, Burwinkle TM, Katz ER, et al. The PedsQL in pediatric cancer: reliability and validity of the pediatric quality of life inventory generic core scales, multidimensional fatigue scale, and cancer module. Cancer 2002;94:2090e2106. 3. Russell KMW, Hudson M, Long A, Phipps S. Assessment of health-related quality of life in children with cancer. Cancer 2006;106:2267e2274. 4. Chang PC, Yen CH. Agreement between child self-report and parent proxy-report to evaluate quality of life in children with cancer. Psychooncology 2005;14:125e134. 5. Yeh CH, Chao KY, Hung LC. Construct validity of newly developed quality of life assessment instrument for child and adolescent cancer patients in Taiwan. Psychooncology 2003;12:345e356. 6. Palmer SN, Meeske KA, Katz ER, Burwinkle TM, Varni JW. The PedsQlÔ brain tumor module: initial and validity. Pediatr Blood Cancer 2007;49: 287e293. 7. Lai JS, Cella D, Tomita T, et al. Developing a health-related quality of life instrument for childhood brain tumor survivors. Childs Nerv Syst 2007; 23:47e57. 8. Tao ML. Quality of life assessment in pediatric brain tumor patients and survivors: lessons learned and challenged to face. J Clin Oncol 2005;23: 5424e5426. 9. Bhat SR, Goodwin T, Burwinkle TM, et al. Profile of daily life in children with brain tumors: an assessment of health-related quality of life. J Clin Oncol 2005;23:5493e5500. 10. Yeh CH, Hung LC, Chao KY. The quality of life for cancer children (QOLCC) for Taiwanese children with cancer (Part II): feasibility, cross-informants variance and clinical validity. Psychooncology 2004;13:171e176. 11. Mueller HL, Bruhnken G, Emser A, et al. Longitudinal study on quality of life in 102 survivors of childhood craniopharyngioma. Childs Nerv Syst 2005;21:975e980. 12. Hinds PS, Hockenberry-Eaton M. Developing a research program on fatigue in children and adolescents diagnosed with cancer. J Pediatr Oncol Nurs 2001;18:3e12.
Vol. 40 No. 4 October 2010 QOL Assessment of Adolescent Brain Tumor Patients 565 13. Gurney JG, Smith MA, Bunin GR, Smith MA, Gurney JG, et al. CNS and miscellaneous intracranial and intraspinal neoplasms. In: Ries LAG, ed. Cancer incidence and survival among children and adolescents: United States SEER program 1975-1995. Bethesda, MD: National Cancer Institute SEER Program, 1999: 51e63. 14. Korean Central Cancer Registry. 2002 Annual report of the Korea Central Cancer Registry. Seoul, Korea: Ministry of Health and Welfare, 2003:107, 112. 15. Cella D. Manual of the functional assessment of chronic illness therapy (FACIT scales). Evanston, IL: Center on Outcomes, Research and Education (CORE), Evanston Northwestern Healthcare and Northwestern University, 1997. 16. Reynolds CR, Richmond BO. What I think and feel: a revised measure of children s manifest anxiety. J Abnorm Child Psychol 1978;2:271e280. 17. Choi JS, Cho SC. Assessment of anxiety in children reliability and validity of revised children s manifest anxiety scale. Korean J Neuropsychiatric Assn 1990;29:691e702. 18. Cho SC, Lee YS. Development of the Korean form of the Kovacs children s depression inventory. Korean J Neuropsychiatric Assn 1990;29:943e956. 19. Kaasa S, Bjordal K, Aaronson N, et al. The EORTC core quality of life questionnaire (QLQ-C30): validity and reliability when analyzed with patients treated with palliative radiotherapy. Eur J Cancer 1995;31A:2260e2263. 20. Wirrell E, Blackman M, Barlow K, Mah J, Hamiwka L. Sleep disturbances in children with epilepsy compared with their nearest-aged siblings. Dev Med Child Neurol 2005;47:754e759. 21. Cho JH, Choi JU, Kim DS, Suh CO. Low-dose craniospinal irradiation as a definitive treatment for intracranial germinoma. Radiother Oncol 2009;91:75e79.