Cancer threatens the health and life of children worldwide. The latest childhood cancer report showed that more than 380 000 children and adolescents aged 0 to 19 years were diagnosed with cancer annually, and the incidence of cancer was 140.6 per million in children aged 0 to 14 years and 155.8 per million in those aged 0 to 19 years.1 The 2012 cancer registration data in China showed that the morbidity of pediatric cancer was 10.78 per 100 000, lower than 11.35 per 100 000 in 2009.2,3 Cancer itself and the treatment of cancer have a great impact on children's health status and quality of life. Common adverse effects include fatigue, nausea, vomiting, emotional distress, altered sleep quality, and poor social and physical functions.4–6 Patient-reported outcome (PRO) instruments are often the most refined evaluation method of these effects.7–9 Patient-reported outcomes are defined as a “measurement of any aspect of patient's health status that comes directly from the patient, without the interpretation of the patient's responses by physician or others.”9 Soliciting PROs provides valuable information about the patient's treatment and disease-related experiences in pediatric oncology clinical research and practice.7,10 Many studies have confirmed that pediatric patients' self-report should be considered as the standard for measuring perceived symptoms and health-related quality of life (HRQOL), children as young as 8 years old can reliably report on most aspects of their own health status, and younger children have reliably reported on selected health domains.10,11
To advance the use of PROs in pediatric oncology, reliable, valid, and responsive instruments should be available with consideration of the child's cognitive abilities.9–11 Numerous general and disease-specific PROs have been developed, validated, and used in population surveys of children.4,6,10,11 Pediatric instruments are limited in their comparability because they are developed specifically for an age range or diagnosis, conceptual definitions, and content or item format.4,10 In addition, most measures were developed based on classical test theory,4,10,12 rather than modern test theory, and do not apply advanced administration and scoring technology. Modern measurement theories, such as item response theory (IRT), enable the creation of item banks that support fixed-length forms, as well as computerized adaptive testing, often resulting in improved measurement precision and responsiveness.13 So using the modern test theory could enable general instruments to have greater precision, a broader range of measurement, and equivalent or superior test characteristics compared with traditional (ie, disease-specific) instruments.14
Patient-Reported Outcomes Measurement Information System (PROMIS), funded by the National Institutes of Health as part of the National Institutes of Health Roadmap for Medical Research Initiative,15 was developed by using the modern test theory (IRT) to advance the science and application of PROs and provide efficient, reliable, and valid assessments of adult and pediatric self-reported health status. The PROMIS has the potential to transform PRO measurement by creating a shared, unifying terminology and metric for reporting common symptoms and functional life domains. There are 3 key steps for PROMIS International translation culturally adapted for use in adult and children in many countries: a universal translation/cultural adaptation, calibration, and validation.16 Patient-Reported Outcomes Measurement Information System item banks or short forms are developed or in progress in other languages other than English. The PROMIS Pediatric multisite project was developed to sensitively measure self-reported symptoms, quality of life, and function in children and adolescents. This advanced measurement system was used as the same sensitive measures across a variety of chronic illnesses so that possible new knowledge could be gained through direct comparability of a core set of items and scores across diverse pediatric groups. The pediatric PROMIS (Ped-PROMIS) was developed using both qualitative and quantitative methods17–22 and was tested in a variety of settings, enabling a flexible use of short forms of the measures or computerized adaptive testing.22–26
There are limited pediatric measures in pediatric oncology in China, so it reduced the possibility to incorporate children's voices into their treatment and care. The Pediatric Quality of Life Inventory Generic Core Scales (PedsQL 4.0 Generic Core Scales), Cancer Module (PedsQL 3.0 Cancer Module Scale), and Multidimensional Fatigue Scale have been widely used in China and abroad.10,27–29 However, they have the same limitation because they were developed using the classical test theory. To remedy this, with the authorization from PROMIS Health Organization, the Chinese versions of the Ped-PROMIS (C-Ped-PROMIS) measures were developed.30,31 The 8 short forms focus on the symptoms and functioning that are most frequently reported in cancer treatment and survivorship: physical functioning–upper extremity, pain interference, fatigue, depressive symptoms, anxiety, anger, physical functioning–mobility, and peer relationships. The feasibility and known-group validity of these short form measures have been successfully tested in American children in active cancer treatment and survivorship.32,33 As the American PROMIS measures reports and previous reports indicated that children and adolescents in treatment of cancer report significantly worse quality of life than the survivors of childhood cancer who were free of treatment,32,34 we purposefully enrolled the participants in treatment and survivorship and hypothesized that their responses to the C-Ped-PROMIS measures would be significantly different from each other.
The objective of this study was to systematically examine the reliability, validity, and factorial structure of these 8 C-Ped-PROMIS measures, which will provide evidence for the reliable and valid measures for children and adolescents with cancer in China. The specific aims were (1) to estimate item and scale reliability of each C-Ped-PROMIS short form, (2) to estimate known-group validity and concurrent validity of the C-Ped-PROMIS measures, and (3) to examine the factorial structure of the C-Ped-PROMIS measures.
The PROMIS, including Ped-PROMIS, were built on the World Health Organization Health framework of physical, mental, and social health to sensitively measure self-reported symptoms, quality of life, and function domains in the general population, adults, and pediatric populations with a chronic condition(s).15 Our study framework was the theory of unpleasant symptoms.35,36 According to this theory, symptom characteristics (duration, quality, distress, and intensity) and their outcomes are influenced by at least 1 of 3 factors (physiologic, psychological, and situational). Performance outcomes of the troubling symptoms include functional status, cognitive functioning, and physical performance. All framework factors were measured, which were physiologic (health problems, treatment status, and diagnosis), psychological (depressive symptoms, anger, and anxiety), situational (in treatment or in survivorship), resulting symptoms (fatigue and pain), and performance (peer relationships, physical functioning–mobility, and physical functioning–upper extremity). We did not intend to test the framework but instead assess theoretically derived, previously supported relationships using non-PROMIS measures,37 as validity assessments of the C-Ped-PROMIS measures in this patient group, for example, higher fatigue, would be associated with lower physical functioning. The PedsQL 4.0 Generic Core Scales and PedsQL 3.0 Cancer Module were used to measure the physical, psychological, social, and school functions and specialized symptoms of children with cancer.
An instrument methodological study was conducted. The 8 C-Ped-PROMIS short forms and the Chinese version of PedsQL 4.0 Generic Core Scales and PedsQL 3.0 Cancer Module were administered in a cross-sectional design; children and adolescents who were aged 8 to 17 years in treatment of childhood cancer or in survivorship were eligible. Reliability, validity, and factorial structure of the C-Ped-PROMIS short forms were examined in this study.
Sample and Settings
This study was conducted in 3 tertiary children hospitals in Shanghai and Chengdu in China from May 2012 to August 2015. The number of pediatric and adolescent beds in the pediatric oncology department ranges from 60 to 153, and the number of newly diagnosed pediatric oncology patients annually at each site ranges from approximately 200 to more than 500. A convenience sampling method was used to recruit the eligible participants in hospital-based inpatient and outpatient clinics. Children who were (1) aged 8 to 17 years, (2) with a diagnosis of cancer either in treatment or in survivorship, (3) able to speak and read Chinese, and (4) willing to participate in this study and have their parent's permission were included. Anyone who (1) had a diagnosed psychiatric condition or a cognitive or other impairment (eg, visual) that would interfere with participation, (2) declined or had a parent declined to participate, and (3) were receiving end-of-life care (life expectancy < 6 weeks) were excluded in this study. The ethics approvals for this study were given by the institutional review boards in the University and 3 hospitals. Our proposed sample was 300, which was large enough for the reliability and validity analytical plans; for example, it was enough for a well-known rule of thumb for a sample size estimation of CFA models, which was 10 cases per observed variable (eg, item). A total of 304 children and adolescents with cancer were eligible and approached, and 272 (89.5%) valid instrument data were collected in this study; 11 of the eligible participants declined to participate: 5 reported lack of time, 2 indicated that they were not interested, and 4 indicated that their parents agreed but they refused. Twenty-one agreed to participate but did not finish the investigation.
THE C-PED-PROMIS MEASURES
The original English version of the Ped-PROMIS was developed with mixed methods (ie, focus group, cognitive interview, and diverse population testing) to create item banks and short forms to measure the symptoms and quality of life of 8- to 17-year-old children with a variety of chronic illnesses.19–24 Eight non–disease-specific short forms and 1 disease-specific short form have been developed and are well tested in the Ped-PROMIS version 1.0, including physical functioning–upper extremity, pain interference, fatigue, depression, anxiety, anger, physical functioning–mobility, and peer relationships and asthma.23–28 These pediatric measures were well tested in samples of 8- to 17-year-olds experiencing chronic diseases such as kidney disease, cancer, and asthma or who are in need of long-term rehabilitation.20–26,32,33 The C-Ped-PROMIS measures include those most frequently reported in cancer treatment and survivorship (ie, physical functioning–upper extremity, pain interference, fatigue, depression, anxiety, anger, physical functioning–mobility, and peer relationships).The short form measures were developed by using the translation and linguistic validation methodology (FACIT translation methodology), which was an approach focusing on commonalities, rather than differences, across versions developed in regions or countries speaking the same language. Bilingual teams from the United States and China finished the translation (procedures including 2 forward translations, reconciliation, 1 back-translation, and 3 expert reviews) to develop a provisional version, which was then pretested with cognitive interview by probing 10 native Chinese-speaking children aged 8 to 17 years in China to check how the child participant interpreted each item or responded in terms of comprehension of what the item was asking. Most items, response options, and instructions were well understood by the children, and some revisions were made to address patients' comments during the cognitive interview.30 Most short forms have 8 items, with the exception of fatigue that has 10 items and anger that has 6 items. These measures ask the child to respond in relation to the past 7 days and in terms of a 5-point response categories ranging from “never” to “almost always” in most measures. Each completed measure receives a T score with a mean of 50 and a standard deviation of 10 based on the original reference sample. In this manner of scoring,29 70 was considered a high score and 80 was very high, whereas 30 was a low score and 20 was a very low score. Higher scores meant that more of the measured symptoms were experienced.
PEDSQL 4.0 GENERAL CORE SCALE
The PedsQL,10 created by James W. Varni, is a modular instrument designed to measure HRQOL in children and adolescents aged 2 to 18 years. The PedsQL 4.0 Generic Core Scale is a multidimensional scale developed as a generic core measure. The 23-item PedsQL4.0 Generic Core Scales comprises 4 dimensions: physical functioning, emotional functioning, social functioning, and school functioning. The instructions of PedsQL4.0 Generic Core Scales ask the child how much of each item he/she has experienced during the past month. All items use a 5-point Likert category, ranging from 0 (never) to 4 (almost always). Items are reverse-scored and transformed to a score of 0 to 100 (ie, 0 = 100, 1 = 75, 2 = 50, 3 = 25, and 4 = 0), with higher scores indicating better HRQOL. Two pediatric self-report versions of PedsQL 4.0 Generic Core Scales for children (8-12 years old) and adolescents (13-18 years old) were used to test the concurrent validity of C-Ped-PROMIS measures. The reliability and validity were well tested in children with cancer in China. The internal consistency reliability (Cronbach's α) of PedsQL 4.0 Generic Core Scales total scale was .90, and those of the subscales ranged from .81 to .89.27
PEDSQL 3.0 CANCER MODULE SCALE
The PedsQL 3.0 Cancer Module Scale is a multidimensional measure of HRQOL specifically for pediatric cancer. The 27-item scale includes 8 dimensions: pain, nausea, procedural anxiety, treatment anxiety, worry, cognitive problems, perceived physical appearance, and communication. Two pediatric self-report versions for children (8–12 years old) and adolescents (13–18 years old) were used to test the concurrent validity of C-Ped-PROMIS measures. The instructions, time recall period, item responses, and scoring are the same as PedsQL 4.0 Generic Core Scales. The reliability and validity of the Chinese version of PedsQL 3.0 Cancer Module were well tested. The internal consistency reliability of PedsQL 3.0 Cancer Module Scale was 0.94. The model fit index (comparative fit index [CFI]) was 0.96, and the root mean square error of approximation (RMSEA) was 0.074.28
Data Collection Procedures
Eligible participants were recruited during a scheduled inpatient or outpatient admission; the parents or legal guardians were approached by a trained member of the study team and were carefully informed about the study. If the parents gave permission for the team member to approach their child about the study, the team member then asked their child's permission and told the child about the study. Parent and child refusals and agreements were documented. When all permissions have been secured according to ethical standards, the C-Ped-PROMIS measures, PedsQL 4.0 Generic Core Scales, and PedsQL 3.0 Cancer Module Scale were administered to the participants. A study identification number was generated. Parents completed the demographic forms and disease-related questions regarding their child's cancer treatment and the presence and intensity of 3 symptoms (altered appetite, altered sleep, and fatigue) experienced by their child during the previous 7 days, and the children completed the C-Ped-PROMIS measures, PedsQL 4.0 Generic Core Scales, and PedsQL 3.0 Cancer Module. The participants completed survey measures in a private clinic/hospital room.
Statistical analysis was conducted using SPSS 21.0 and Mplus 7.1.38–40
STEP 1: ITEM AND SCALE RELIABILITY
Item-to-total correlation was used to examine item reliabilities in each C-Ped-PROMIS measure. The analysis was conducted to identify problematic items (correlation < 0.30) in each subscale.41,42 Scale reliability of each C-Ped-PROMIS measure was evaluated using Cronbach's α. Scale reliability was also evaluated using the CFA model result.39,40 The test-retest reliability was not explored in this study as the symptom and functioning are changeable in children with cancer, and these 8 measures were designed responsive to the outcome changes and the PROMIS.43
STEP 2: VALIDITY—KNOWN-GROUP VALIDITY AND CONCURRENT VALIDITY
The known-group validity was evaluated by testing whether the measure scores of C-Ped-PROMIS measures were significantly different between children and adolescents in cancer treatment and in cancer survivorship as we hypothesized. The t test or Mann-Whitney U test was used to assess the known-group validity of each C-Ped-PROMIS measure based on the distribution of each group; concurrent validity of the C-Ped-PROMIS measures was evaluated by examining the correlations of each C-Ped-PROMIS measure with the total scores of PedsQL 4.0 Generic Core Scales and PedsQL 3.0 Cancer Module separately. Pearson or Spearman correlation was used based on the distribution of C-Ped-PROMIS measures and PedsQL 4.0 Generic Core Scales and PedsQL 3.0 Cancer Module.
STEP 3: VALIDITY—FACTORIAL STRUCTURE
As each C-Ped-PROMIS measure refers to the measurement of 1 dimension of symptoms (anxiety, depression, fatigue, pain, and anger) or functions (physical functioning–upper extremity, physical functioning–mobility, and peer relationships), single-factor CFA model was applied to evaluate the factorial structure of each C-Ped-PROMIS measure. Model fit was evaluated using fit statistics/indices, such as model χ2 statistic, RMSEA (≤0.08), 90% confidence interval of RMSEA, close-fit test (ie, probability RMSEA ≤ 0.05), P value, CFI (>0.90), Tucker-Lewis index (>0.90), and standardized root mean square residual (≤0.08).39,40
A total of 272 children and adolescents participated in this study. The demographic and treatment information are presented in Table 1. More than half of them (60.3%, n = 164) were male, between the ages of 8 and 12 years (61.4%, n = 167), and diagnosed with leukemia (77.2%, n = 210). Most participants were in active cancer treatment (72.1%, n = 196). Most guardians were mothers (64.3%, n = 175) and married (91.9%, n = 250). The largest group of guardians had junior middle school educational level (40.4%, n = 110). Few children had other health problems (5.7%, n = 16). Patients in the treatment group had greater days of poor appetite (mean [SD], 2.02 [2.06] vs 0.36 [1.07]), poor sleep (mean [SD], 1.20 [1.61] vs 0.29 [0.81]), and tiredness (mean [SD], 1.86 [1.91] vs 0.35 [0.98]), indicating that participants in treatment were more symptomatic than were survivors (Table 2).
The average original scores and T scores were reported in Table 3. All the T scores of the 8 short forms were lower than 50, ranging from 43.06 to 49.21, indicating more distressing symptoms and worse peer relationship and physical function compared with the American reference population. The average symptom short form scores of children in active treatment were higher and the function scores were lower than that of children in survivorship (Table 5).
Item reliability was estimated using item-total correlation. All the items were significantly correlated with their underlying scale score (P < .01). According to Cohen44 (1988), an absolute value of correlation coefficient of 0.1 is considered small, an absolute value of 0.3 is medium, and an absolute value of 0.5 is large. Most of the item-total correlations were large (ie, ≥0.50), ranging from r = 0.527 to 0.834; only two such correlations, associated with two items of mobility measure, were r < 0.50. The Cronbach's α were greater than .70 for all the scales, ranging from .758 to .910 (Table 4). Model-estimated scale reliabilities were also large, ranging from 0.740 to 0.905 (Table 4).
Validity: Known-Group Validity and Concurrent Validity
As hypothesized, the mean values of the 8 C-Ped-PROMIS measures are significantly different between the active treatment group and the survivorship group (P < .01, except anger with P = .018) (Table 5). Participants in the treatment group had worse scores on both the symptom and function measures, compared with those in the survivorship group, indicating good known-group validities for these measures. All the C-Ped-PROMIS measures were significantly correlated with PedsQL 4.0 Generic Core Scales and PedsQL 3.0 Cancer Module (P < .05) (Table 6). The absolute values of the correlation coefficients were shown in Table 6, except a few that were large, ranging from r = 0.519 to 0.655. However, the absolute values of the coefficients of peer relationship with PedsQL 4.0 Generic Core Scales and PedsQL 3.0 Cancer Module were relatively small (r = 0.255 and 0.136, respectively; Table 6). In addition, the correlation of physical function–upper extremity with PedsQL 3.0 Cancer Module was also lower than 0.3.
Validity: Factorial Structure
All the 8 single-factor CFA models fit data well (Table 7): RMSEA ranged from 0.057 to 0.079, CFI ranged from 0.976 to 0.926, Tucker-Lewis index ranged from 0.891 to 0.957, and SRMR ranged from 0.034 to 0.064. The CFA models for 7 C-Ped-PROMIS measures were close-fitting models for which the close fit test P values were greater than .05 expect fatigue (P = 0.024).
The pediatric PROMIS aimed to measure the symptom, function, and quality of life of 8- to 17-year-old children and adolescents with a variety of chronic illnesses. The reliability, validity, and factorial structure of C-Ped-PROMIS short form measures were evaluated in pediatric children with cancer in this study. The result demonstrated good item and scale level reliability, good known-group validity, and concurrent validity of these measures and provided evidence that all the 8 C-Ped-PROMIS measures were unidimensional as designed, and these short forms could be applied in the pediatric cancer care and research in China.
As all item-to-total correlations of the 8 C-Ped-PROMIS measures were higher than 0.30, it indicated the adequacy of their item placement in the scale and further supported the homogeneity of the scale.45 The Cronbach's α coefficients were large (>.70) for all the C-Ped-PROMIS measures indicating satisfactory internal consistency of all short forms, considering that Cronbach's α was a lower bound to the true reliability assuming tau-equivalent or parallel measures and no correlated measurement errors.40 We also estimated scale reliability using CFA model results in this study, which also demonstrated good reliability of each C-Ped-PROMIS measure.
It has been confirmed that quality of life of children with cancer in follow-up was better than that of children in active treatment.34 The results of known-group validity showed that all the short form measures could distinguish the symptom and functioning difference between the children and adolescents in active cancer treatment or in survivorship. The similar result has been demonstrated in American children with cancer, except anger short form.32 The PedsQL is widely accepted and used worldwide10,11,34; the PedsQL 4.0 Generic Core Scales and PedsQL 3.0 Cancer Module are used to measure general physical, emotional, and social functions and cancer-related symptoms. As C-Ped-PROMIS was used to measure physical (physical function–mobility and physical function–upper extremity), emotional (anxiety, depressive symptom, and anger), and social (peer relationship) functions and symptoms (fatigue and pain), we used the PedsQL 4.0 Generic Core Scales and PedsQL 3.0 Cancer Module to test the concurrent validity of each C-Ped-PROMIS measure. All the short forms were significantly correlated with the PedsQL 4.0 Generic Core Scales and PedsQL 3.0 Cancer Module. Depressive symptoms, anger, anxiety, fatigue, and pain interference were strongly correlated with the PedsQL 4.0 Generic Core Scales and PedsQL 3.0 Cancer Module Scale (r > 0.5). Mobility was strongly correlated with PedsQL 4.0 Generic Core Scales and moderately correlated with PedsQL 3.0 Cancer Module, and upper extremity was moderately correlated with PedsQL 4.0 Generic Core Scales. Unexpectedly, peer relationship was weakly correlated with PedsQL 4.0 Generic Core Scales, in which the social and school functioning dimensions were included. As the peer relationship was designed to measure the social function of children and adolescents, the reason for the lower correlation needs to be explored in future studies.
The pediatric PROMIS measures were developed with the World Health Organization Health framework by using the advanced IRT. One of the characteristics of all short forms was the single-factor model,15 which implied that a set of items forming an instrument measure 1 latent variable in common and was an important aspect of construct validity. Confirmatory factor analysis was used in this study to evaluate the single-factor model. Most model fit indices of the C-Ped-PROMIS were acceptable, proving that each short form measured their latent variables in children with cancer in China as theoretically defined.
Implication for Practice
This study provides evidence that the 8 C-Ped-PROMIS short form measures could be used in children and adolescent with cancer in China to self-report their symptoms, function, and quality of life. It will be a reliable and valid instrument for healthcare professionals to solicit or collect child self-reported outcomes in cancer research and future clinical practice, promoting listening to the children's voice in pediatric care in China.
There were several limitations in this study. First, a convenience sampling method was used in this study. Second, the test-retest reliability of the C-Ped-PROMIS measures was not examined in this study, as the symptoms and function could be changed in different treatment statuses of children with cancer, and it should be evaluated in the healthy children population in the future. Third, 1 or more error covariances were set as free parameters in each single-item CFA model for the purpose of model fit improvement. However, setting only 1 error covariance as a free parameter in a model would not violate the assumption of local independence. Finally, measurement invariance of the C-Ped-PROMIS measures across different populations and different samples of the same population over time needs to be examined in future studies.
This is the first study to systematically examine the measurement characteristics of the C-Ped-PROMIS measures in children with cancer in China. Our results provide evidence that the C-Ped-PROMIS measures are reliable and valid and can be readily applied in the Chinese population with pediatric cancer to measure children's symptoms, function, and quality of life. Longitudinal and comparative possibility studies in children with cancer could be conducted in the future.
The authors acknowledge the cooperation and support from the children and parents who participated in this study.
1. Steliarova-Foucher E, Colombet M, Ries LA, et al. International incidence of childhood cancer, 2001-10: a population-based registry study. Lancet Oncol
2. Chen WQ, Zheng RS, Zuo TT, Zeng H, Zhang S, He J. National cancer incidence and mortality in China, 2012. Chin J Cancer Res
3. Chen W, Zheng R, Zhang S, et al. Report of incidence and mortality in China cancer registries, 2009. Chin J Cancer Res
4. Hockenberry MJ, Hinds PS, Barrera P, et al. Three instruments to assess fatigue in children with cancer: the child, parent and staff perspectives. J Pain Symptom Manage
5. Hinds PS, Billups CA, Cao X, et al. Health-related quality of life in adolescents at the time of diagnosis with osteosarcoma or acute myeloid leukemia. Eur J Oncol Nurs
6. Baggott C, Dodd M, Kennedy C, et al. Multiple symptoms in pediatric oncology patients: a systematic review. J Pediatr Oncol Nurs
7. Trotti A, Colevas AD, Setser A, et al. Patient-reported outcomes and the evolution of adverse event reporting in oncology. J Clin Oncol
8. Basch E, Abernethy AP, Mullins CD, et al. Recommendations for incorporating patient-reported outcomes into clinical comparative effectiveness research in adult oncology. J Clin Oncol
9. US Department of Health and Human Services, Food and Drug Administration. Guidance for industry: patient-reported outcomes measures. http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM193282.pdf
. Accessed October 6, 2016.
10. Varni JW, Burwinkle TM, Katz ER, et al. The PedsQLTM
in pediatric cancer: reliability and validity
of the Pediatric Quality of Life InventoryTM
Generic Core Scales, Multidimensional Fatigue Scale, and Cancer Module. Cancer
11. Varni JW, Limbers C, Burwinkle TM. Literature review: Health-related quality of life measurement in pediatric oncology: hearing the voices of the children. J Pediatr Psychol
12. Hinds PS, Hockenberry M, Tong X, et al. Validity and reliability of a new instrument to measure cancer-related fatigue in adolescents. J Pain Symptom Manage
13. Véronique S, Jean-Benoit H, Le NT, et al. Methodological issues regarding power of classical test theory (CTT) and item response theory (IRT)-based approaches for the comparison of patient-reported outcomes in two groups of patients—a simulation study. BMC Med Res Methodol
14. Leung YY, Tam LS, Kun EW, et al. Comparison of 4 functional indexes in psoriatic arthritis with axial or peripheral disease subgroups using Rasch analyses. J Rheumatol
15. Cella D, Yount S, Rothrock N, et al. The Patient-Reported Outcomes Measurement Information System (PROMIS): progress of an NIH Roadmap cooperative group during its first two years. Med Care
. 2007;45(5 suppl 1):S3–S11.
16. Alonso J, Bartlett SJ, Rose M, et al. The case for an international patient-reported outcomes measurement information system (PROMIS®) initiative. Health Qual Life Outcomes
17. Reeve BB, Hays RD, Bjorner JB, et al.; PROMIS Cooperative Group. Psychometric evaluation and calibration of health-related quality of life item banks: plans for the Patient-Reported Outcomes Measurement Information System (PROMIS). Med Care
18. DeWalt DA, Rothrock N, Yount S. PROMIS Cooperative Group. Evaluation of item candidates: the PROMIS qualitative item review. Med Care
19. Walsh TR, Irwin DE, Meier A, Varni JW, DeWalt DA. The use of focus groups in the development of the PROMIS pediatrics item bank. Qual Life Res
20. Irwin DE, Varni JW, Yeatts K, DeWalt DA. Cognitive interviewing methodology in the development of a pediatric item bank: a patient reported outcomes measurement information system (PROMIS) study. Health Qual Life Outcomes
21. Pilkonis PA, Choi SW, Reise SP, et al. Item banks for measuring emotional distress from the Patient-Reported Outcomes Measurement Information System (PROMIS®): depression, anxiety, and anger. Assessment
22. Irwin DE, Stucky BD, Thissen D, et al. Sampling plan and patient characteristics of the PROMIS pediatrics large-scale survey. Qual Life Res
23. Irwin DE, Stucky B, Langer MM, et al. An item response analysis of the pediatric PROMIS anxiety and depressive symptoms scales. Qual Life Res
24. Varni JW, Stucky BD, Thissen D, et al. PROMIS Pediatric Pain Interference Scale: an item response theory analysis of the pediatric pain item bank. J Pain
25. DeWitt EM, Stucky BD, Thissen D, et al. Construction of the eight-item patient-reported outcomes measurement information system pediatric physical function scales: built using item response theory. J Clin Epidemiol
26. Irwin DE, Stucky BD, Langer MM, et al. PROMIS Pediatric Anger Scale: an item response theory analysis. Qual Life Res
27. Lu Y, Tian Q, Hao YT, et al. Reliability and Validity
for Chinese Version of Pediatric Quality of Life Inventory PedsQL4.0. Sun Yat-sen Univ (Med Sci)
28. Zhang JW. Using and Assessment of Reliability and Validity for PedsQLTM3.0 [Dissertation]
. Guangzhou, China: Sun Yat-Sen University; 2009.
29. Wu MS, Xu LH, Zhang BH, et al. The level of cancer related fatigue and its related factor of children with acute leukemia. Chin J Nurs
30. Liu Y, Hinds PS, Wang J, et al. Translation and linguistic validation of the Pediatric Patient-Reported Outcomes Measurement Information System measures into simplified Chinese using cognitive interviewing methodology. Cancer Nurs
31. Liu Y, Wang J, Hinds PS, et al. The emotional distress of children with cancer in China: an item response analysis of C-Ped-PROMIS Anxiety and Depression short forms. Qual Life Res
32. Hinds PS, Nuss SL, Ruccione KS, et al. PROMIS pediatric measures in pediatric oncology: valid and clinically feasible indicators of patient-reported outcomes. Pediatr Blood Cancer
33. Menard JC, Hinds PS, Jacobs SS, et al. Feasibility and acceptability of the patient-reported outcomes measurement information system measures in children and adolescents in active cancer treatment and survivorship. Cancer Nurs
34. Varni JW, Limbers CA, Burwinkle TM. Impaired health-related quality of life in children and adolescents with chronic conditions: a comparative analysis of 10 disease clusters and 33 disease categories/severities utilizing the PedsQL4.0 generic core scales. Health Qual Life Outcomes
35. Lenz ER, Pugh LC, Milligan RA, et al. The middle-range theory of unpleasant symptoms: an update. ANS
36. Myers JS. A comparison of the theory of unpleasant symptoms and the conceptual model of chemotherapy-related changes in cognitive function. Oncol Nurs Forum
37. Hinds PS, Hockenberry MJ, Gattuso JS, et al. Dexamethasone alters sleep and fatigue in pediatric patients with acute lymphoblastic leukemia. Cancer
38. Field A. Discovering Statistics Using SPSS
. 2nd ed. London, England: Sage; 2005.
39. Muthen L, Muthen B. Mplus. User's Guide
. Los Angeles, CA: Muthen & Muthen; 1998-2012.
40. Wang J, Wang X. Structural Equation Modeling With Mplus: Methods and Applications
. New York, NY: John Wiley; 2012:35.
41. Everitt BS. The Cambridge Dictionary of Statistics
. 2nd ed. Cambridge, England: Cambridge University Press; 2002.
42. De Vaus DA. Analyzing Social Science Data
. London, England: Sage; 2002.
43. Reeve BB, Edwards LJ, Jaeger BC, et al. Assessing responsiveness over time of the PROMIS®
, pediatric symptom and function measures in cancer, nephrotic syndrome, and sickle cell disease. Qual Life Res
44. Cohen J. Statistical Power Analysis for the Behavioral Sciences
. 2nd ed. Hillsdale, NJ: Erlbaum; 1988.
45. Wang WR, Thompson DR, Chair SY, Hare DL. A psychometric evaluation of a Chinese version of the Cardiac Depression Scale. J Psychosom Res