Measuring Health-Related Quality of Life With the Parental Opinions of Pediatric Constipation Questionnaire

Transcription

1 Journal of Pediatric Psychology, 40(8), 2015, doi: /jpepsy/jsv028 Advance Access Publication Date: 2 April 2015 Original Research Article Measuring Health-Related Quality of Life With the Parental Opinions of Pediatric Constipation Questionnaire Alan H. Silverman, 1 PHD, Kristoffer S. Berlin, 2,3 PHD, Carlo Di Lorenzo, 4 MD, Samuel Nurko, 5 MD, Rebecca C. Kamody, 2 BS, Ananthasekar Ponnambalam, 6 MD, Suzanne Mugie, 4 BS, Christina Gorges, 1 BS, Rina Sanghavi, 7 MD, and Manu R. Sood, 1 MD 1 Medical College of Wisconsin, 2 The University of Memphis, 3 University of Tennessee Health Science Center- Memphis, 4 Nationwide Children s Hospital, 5 Children s Hospital of Boston, 6 University of South Alabama, and 7 UT Southwestern Medical Center All correspondence concerning this article should be addressed to Alan H. Silverman, PHD, Medical College of Wisconsin, 8701 Watertown Plank Road, Suite B610, Milwaukee, WI 53226, USA. asilverm@mcw.edu Received August 8, 2014; revisions received February 26, 2015; accepted March 4, 2015 Abstract Objectives The purpose of this study was to develop a caregiver-completed constipation condition-specific health-related quality of life (HRQL) instrument. Methods 410 caregivers of children aged 2 18 years completed the Parental Opinions of Pediatric Constipation (POOPC), the PedsQL Generic Core Scales, PedQL Family Impact Module, Pediatric Symptom Checklist, the Functional Disability Inventory, the Pediatric Inventory for Parents, and a demographic questionnaire. Exploratory and confirmatory factor analyses were conducted to assess the psychometric properties of the POOPC. Results Analyses yielded four factors called Parental Burden/Distress, Family Conflict, Difficulties with the Medical Team, and Worry about Social Impact that reflect problems in HRQL secondary to constipation and soiling, which were generally more strongly correlated with similar measures relative to a general measure of youth s psychosocial functioning. Conclusion The POOPC is a psychometrically sound measure, which may be useful to clinicians and researchers identifying domains of treatment needs for children and their families. Key words: assessment; elimination disorders; encopresis; health-related quality of life. Health-related quality of life (HRQL) has emerged as an important construct used to assess patient outcomes in clinical trials, practice improvement evaluations, and in health-care services research. Use of HRQL measures in pediatric health-care settings may improve patient provider communication, increase patient/caregiver satisfaction, and enhance clinical decision making (Varni et al., 2005). While general HRQL measures (e.g., Child Health and Illness Profile, Child Health Questionnaire, Pediatric Quality of Life Inventory) may be useful for screening healthy populations and for making comparisons across various condition types, condition-specific measures are uniquely suited to measuring changes within a condition type (Matza, Swensen, Flood, Secnik, & Leidy, 2004). Many condition-specific HRQL measures exist (e.g., asthma, cancer, diabetes). However, a well-validated English-language pediatric constipation HRQL measure does not exist. Thus, the purpose of this research was to develop a reliable and valid constipation and fecal incontinence measure of HRQL. VC The Author Published by Oxford University Press on behalf of the Society of Pediatric Psychology. All rights reserved. For permissions, please journals.permissions@oup.com 814

2 Pediatric Constipation 815 Constipation in childhood affects 3% of children (van den Berg, Benninga, & Di Lorenzo, 2006). The cost of treating childhood constipation in the United States is $3.9 billion each year (Liem et al., 2009). Constipation is often first diagnosed in infancy. Symptoms of constipation include straining to pass stool, lumpy or hard stools, sensation of incomplete evacuation of defecations, sensation of anorectal obstruction or blockage of defecations, manual maneuvers to facilitate defecations (e.g., digital evacuation, support of the pelvic floor), and fewer than three defecations per week (Drossman, 2006). In many children, constipation is accompanied by overflow soiling (Cambell, Cox, & Borowitz, 2009). Symptoms associated with constipation are often prolonged, with 30 50% of children having persistent symptoms after 5 years of initiating medical treatment (Brooks et al., 2000). Chronic constipation and soiling are presumed to have negative effects on HRQL (Treurniet, Essink- Bot, Mackenbach, & van der Maas, 1997), yet there are no well-validated condition-specific measures to study such concerns. Studies which use general measures of HRQL report associations between symptoms of constipation (with and without soiling) and behavioral, social, and emotional problems (Bongers, van Dijk, Benninga, & Grootenhuis, 2009; Cox, Morris, Borowitz, & Sutphen, 2002; Joinson, Heron, Butler, & von Gontard, 2006; Youssef, Langseder, Verga, Mones, & Rosh, 2005). These children also have disproportionally higher rates of these problems than do children diagnosed with other gastrointestinal complaints (Bongers et al., 2009). Surprisingly, past research comparing children who have constipation without soiling to children with constipation and soiling reported no HRQL differences. It should be noted, however, this study failed to use a condition-specific HRQL measure. In the Netherlands, Bongers and colleagues longitudinally studied a cohort of 401 Dutch children who presented with symptoms of constipation. Data were collected at 1, 5, and 10 years after diagnosis. Deteriorating function was reported in 50% of the population at 1 year, 36% of the population at 5 years, and 20% at 10 years, and that 20 35% of middle and high school aged children in this study continued to experience painful and/or infrequent bowel movements and/or having fecal soiling accidents (Bongers, van Wijk, Reitsma, & Benninga, 2010). Citing concerns that general HRQL instruments lacked sensitivity to assess the impact of constipation and fecal incontinence, the investigatory team developed a disease-specific HRQL for Dutch children (Bongers et al., 2009). Using this instrument the investigators reported that childhood constipation has a significant impact on child and family HRQL. Primary factors associated with reduced HRQL among these children included odor associated with stool leakage and are vulnerable to bullying and social isolation at school. To better understand the unique characteristics of constipation on HRQL, in American children a prospective qualitative study of affected children and their caregivers was conducted (Kaugars et al., 2010). Adult caregivers reported higher rates of negative emotion in their children (e.g., worry, anger, and embarrassment) than was reported by affected children. Families varied in their satisfaction with treatment recommendations, and many reported difficulty finding appropriate care. Overall, child social functioning and family functioning were negatively affected by constipation and fecal incontinence. The authors concluded that general measures of HRQL were inadequate for detecting the specific factors that uniquely affect these families. The results of Kaugers and colleagues earlier study provided the impetus for developing a standardized condition-specific instrument for assessment of HRQL for children with constipation. Consistent with the investigative team s previous qualitative research (Kaugars et al., 2010) it was hypothesized that four to five HRQL factors would emerge. The presumed factors would include issues related to treatment (planning, effectiveness, satisfaction, and recommendations), children s social and family functioning, and general emotional impact specific to constipation and fecal incontinence. It was also hypothesized that these subscales would moderately correlate with established measures of parents report of youth HRQL, and to a lesser extent a general measure of youth social-emotional functioning. Method Participants Caregivers of children aged 2 18 years were invited to participate. To be included, families needed to be fluent in English, and the child had to meet the ROME III criteria for constipation and fecal incontinence as according to one of the following criteria: (1) existing diagnosis of constipation and fecal incontinence, (2) constipation predominant irritable bowel syndrome, or (3) nonretentive fecal incontinence ("Guidelines Rome III Diagnostic Criteria for Functional Gastrointestinal Disorders," 2006). Exclusion criteria included (1) caregivers of children with moderate to severe developmental delays, and (2) caregivers of children with associated chronic disease, which may have had an impact on HRQL (e.g., cerebral palsy, spine deformity or malformations, learning difficulties, severe psychiatric illness, celiac disease). Caregivers of children with moderate to severe developmental delays were excluded owing to concerns that caregivers would report lower HRQL ratings

3 816 Silverman et al. based on factors unrelated to constipation, which would in turn make interpretation of data difficult. A total of 468 patients and families were enrolled in the study. Of these families, 410 had sufficient data and met inclusion criteria. Twenty-eight participants were excluded owing to the presence of another chronic medical condition(s). Twenty-eight participants enrolled but had missing or unusable data. One participant withdrew from the study after consent, and one subject was inadvertently enrolled despite not having constipation. The final total sample was composed of 122 participants from Children s Hospital of Wisconsin; 22 participants from Children s and Women s Hospital of South Alabama; 73 participants from Children s Hospital of Boston; 154 participants from Nationwide Children s Hospital in Columbus, Ohio; and 39 participants from the University of Texas Southwestern Medical Center in Dallas Texas. The mean (SD) age of the remaining children was 7.8 (3.5) years with 215 (52%) of the children being male. One hundred eighty-four (45%) subjects were categorized as having functional constipation alone and 226 (55%) subjects were categorized as having functional constipation with fecal incontinence. The distribution of children within the different age subgroups was as follows: 2 4 years: 106 (26%), 5 7 years: 136 (33%), 8 12 years: 133 (32%), and years: 35 (9%). There were no differences between children with constipation without fecal soiling and children with constipation and fecal incontinence on the following variables: age at time of study participation, age of symptom onset, age when medical help was sought, and symptom duration. The means and SDs (in years) for the constipation versus the constipation with fecal incontinence group were as follows: age at time of study: 7.4 (4.08) versus 7.9 (2.82), age of symptom onset: 3 (3.60) versus 3.6 (2.80), age medical help sought: 4 (4) versus 4.4 (2.87), and symptom duration: 4.2 (3.85) versus 4.4 (3.34). The ethnic group composition of the sample was 78% Whites, 9% Blacks, 5% Hispanic, 4% other, and 3% Asian. Among the mothers, 57% had earned a high school or associate degree, 36% had earned a 4-year or advanced college degree, and 7% did not complete high school. The distribution of maternal education was not significantly different between the two groups. Procedure The exploratory and confirmatory analyses are the second phase in the development of the Parental Opinions of Pediatric Constipation (POOPC). The first phase consisted of a qualitative study of families who were seen for treatment of constipation in a multidisciplinary clinic and is published in detail elsewhere (Kaugars et al., 2010). Briefly summarizing the methodology of the first phase, the interdisciplinary treatment team (advanced practice nurse, pediatric gastroenterologist, and pediatric psychologist) developed a semistructured interview to describe factors that families identified as impacting the HRQL of the affected child. Thematic content coding and triangulation of data analyses were conducted independently by each discipline to extract final factors from the interview transcripts. For the present study, questions were generated by the interdisciplinary investigation team through discussion of each of the major codes from the qualitative study (Kaugars et al., 2010). The original questionnaire consisted of 50 items, which the interdisciplinary team believed were representative of the original factors (see Supplementary Material). Written informed consent for participation was obtained from adult caregivers. Data were collected from each of the participating academic medical centers. The original instrument was a caregiver report of constipation-specific factors that affect a child s HRQL (see Supplementary Material). All families completed the original 50-item version of the POOPC for validation purposes. In addition to the POOPC, caregivers of children also completed the following questionnaires: PedsQL TM Generic Core Scales PedsQL TM Generic Core Scales (PedsQL TM ) are used to measure HRQL in children and adolescents aged 2 18 years (Varni, Seid, & Rode, 1999). The PedsQL 4.0 Generic Core Scales are child self-report and parent proxy-report scales developed as the generic core measure to be integrated with the PedsQL Disease- Specific Modules. For our study we used the Parent Reports of the PedsQL TM for toddlers (ages 2 4 years), young children (ages 5 7 years), children (ages 8 12 years), and teens (ages years), which are composed of 23 items containing four dimensions (physical functioning, emotional functioning, social functioning, and school functioning). Each item is rated on a 5-point Likert scale (0 ¼ never, 1¼ almost never, 2¼ sometimes, 3¼ often, 4¼ almost always). Scoring is reversed and transformed to a scale (0 ¼ 100, 1 ¼ 75, 2 ¼ 50, 3 ¼ 25, 4 ¼ 0). Each dimension is scored separately by taking the sum of the items over the number of items answered. The psychosocial health summary score is computed as the sum of the items over the number of items answered in the emotional, social, and school functioning scales. The physical health summary score is computed as the physical functioning scale score, and the total score is the sum of all the items over the number of items answered on all the scales. Higher scores indicated better HRQL. Clinical validity, using the known-groups approach, was demonstrated for patients on-versus

4 Pediatric Constipation 817 off-treatments. Internal consistency reliability of the validation sample ranged from 0.70 to PedsQL TM Family Impact Module The PedsQL TM Family Impact Module (PedsQL TM FIM) was designed to assess the impact of pediatric chronic health conditions on parents HRQL and family functioning (Varni, Sherman, Burwinkle, Dickinson, & Dixon, 2004). The PedsQL TM FIM is composed of 36 items composed of eight dimensions: physical functioning, emotional functioning, social functioning, cognitive functioning, communication, worry, daily activities, and family relationships. Each item is rated on a 5-point Likert scale (0 ¼ never, 1 ¼ almost never, 2 ¼ sometimes, 3 ¼ often, 4 ¼ almost always). Scoring is reversed and transformed to a scale (0 ¼ 100, 1 ¼ 75, 2 ¼ 50, 3 ¼ 25, 4 ¼ 0). Each dimension is scored separately by taking the sum of the items over the number of items answered. The total score is the sum of all 36 items divided by the number of items answered. The parent HRQL summary score (20 items) is computed as the sum of the items divided by the number of items answered in the physical, emotional, social, and cognitive functioning scales. The family functioning summary score (eight items) is computed as the sum of the items divided by the number of items answered in the daily activities and family relationships scales. Higher scores indicated better functioning. Internal consistency reliability of the validation sample ranged from.82 to.97. Pediatric Symptom Checklist The Pediatric Symptom Checklist (PSC) assess child and adolescent psychosocial well-being, as a screening measure to identify individuals who may be in need of further evaluation, or as an indicator of psychosocial well-being before and following intervention or treatment (Jellinek et al., 1988). The PSC consists of 35 items. Each item is rated as Never ¼ 0, Sometimes ¼ 1, or Often ¼ 2. The total score is the sum of all 35 items. A cutoff score of 28 indicated psychological impairment in children aged 6 16 years, with a cutoff score of 24 for children aged 4 or 5 years. The authors report evidence of good sensitivity and specificity, and concurrent and convergent validities. Internal consistency reliability of the validation sample ranged from.80 to.96. Functional Disability Inventory The Functional Disability Inventory (FDI) is a 15-item parent-report measure that assesses the child s difficulty in completing daily activities in four domains: home, school, recreational, and social domains (Walker & Greene, 1991). Each item is rated on a 5- point Likert scale (0 ¼ no trouble, 1¼ a little trouble, 2 ¼ some trouble, 3¼ a lot of trouble, 4¼ impossible). A total score is computed, 0 60, with higher scores indicating greater illness-related disability. Validity was supported by significant correlations between childand parent-report FDI scores with measures of schoolrelated disability, pain, and somatic symptoms. Internal consistency reliability of the validation sample ranged from.86 to.91. Pediatric Inventory for Parents The Pediatric Inventory for Parents (PIP) assesses stress among parents of children with a critical illness (Streisand, Braniecki, Tercyak, & Kazak, 2001). The PIP includes 42 items grouped into four domain scales (communication, emotional distress, medical care, and role function). Parents rate each item along a 5-point Likert scale (1 ¼ Not at all, 2¼ Alittle,3¼ Somewhat, 4 ¼ Very much, 5¼ Extremely) as to both the item s frequency over the last week and level of difficulty associated with it. Frequency and difficulty scores are summed separately for each of the four domain scales. These scale scores are then added together for an overall total frequency score and total difficulty score: higher scores indicate greater frequency and difficulty. The authors report that the PIP is a face-valid instrument that is internally consistent for examining parents report of stress related to caring for a child with an illness. Internal consistency reliability of the validation sample ranged from.80 to.96. Other relevant clinical information such as ethnicity/race of the parent and child, characteristics of constipation including onset of constipation, medications, Bristol stool chart, and stool size was collected using a demographic questionnaire. Analytic Plan To generate and cross validate the most optimal factor structure for the POOPC, a series of exploratory and confirmatory factor analyses (CFAs) were completed. The full sample was randomly split so that 70% (N ¼ 309) of the parents formed the learning sample for the exploratory factor analyses (EFAs), and the remaining 30% (N ¼ 101) formed the cross-validation sample for the CFAs. This ratio was chosen to increase the power of the EFAs, with the downside being that it yielded less power for the CFAs and cross validation. The two subsamples did not differ on age, F(1, 408) ¼ 0.13, p ¼.72, gender, v 2 (1) ¼ 0.01, p ¼.99, race, v 2 (5) ¼ 5.07, p ¼.41, or condition type (with or without incontinence), v 2 (1) ¼ 0.02, p ¼.91. All EFAs and CFAs were conducted with Mplus 7.0 using a robust weighted least squares estimator, which adjusts the chi-square test statistics and standard errors for nonnormality in the ordinal indictors and accounts for any possible nonindependence due to nesting by sites. For the EFAs, an oblique Geomin rotation (Yates, 1987) was used, as it was designed to

5 818 Silverman et al. allow for complex factors (items with cross-factor loadings), yet still provide an interpretable solution (Browne, 2001; McDonald, 2005). To assess the goodness of fit for the EFAs and CFA, the Comparative Fit Index (CFI; Bentler, 1990), the Tucker Lewis Index (TLI), and the root mean square error of approximation (RMSEA; MacCallum, Browne, & Sugawara, 1996) were calculated. The CFI and TLI values ranged from 0 to 1, with values above 0.90 representing an adequate model fit and above 0.95 representing a good model fit (Bollen, 1989). The RMSEA statistic was interpreted as an indication of population error variance, with values of <0.05 interpreted as good, between 0.05 and 0.08 as acceptable, between 0.08 and 0.10 as marginal, and >0.10 as poor (Browne & Cudeck, 1993; Fabrigar, Wegener, MacCallum, & Strahan, 1999; Hu & Bentler, 1995). The factor analyses proceeded in three steps. In Step 1, EFAs specifying one to eight factors were conducted, with the goal of identifying the first four models that provided an acceptable fit to the data. In Step 2, an additional round of EFAs based on Step 1 was performed with the learning sample to decrease the number of items overall and to generate one or more alternative models. In this additional round, items with double loadings >0.30 and loading <0.40 were dropped. In Step 3, CFAs with the cross-validation sample were conducted based on the models identified in Steps 1 and 2. When appropriate, items with double loadings >0.30 (EFAs) and loading <0.40 (EFA and/ or CFAs) were dropped from the factors. When the number of items/indicators per factor was two or fewer, the entire factor was dropped. The final step in determining the most optimal factor structure for the POOPC was to choose a model among those with an acceptable fit with an item/factor content most closely resembling the a priori qualitative coding used by Kaugars and collegues (Kaugars et al., 2010). This final factor structure was then used to determine the convergent and discriminant validity of the POOPC to establish the construct validity of the measure (Campbell & Fiske, 1959). This was done through examination of the correlations between the POOPC total score and subscales with the FDI, PIP, PSC, PedsQL TM -Parent Report for ages 2 4, 5 7, 8 12, and years, and PedsQL TM FIMo determine whether the final factor structure was the same across younger and older youth, measurement invariance analyses were conducted with the combined EWFA/CFA sample, comparing relatively equal sample sizes of younger children (88 months old, n ¼ 196) to older children (>88 months old, n ¼ 192). More specifically configural (same items per factor), metric (equal factor loadings), and scalar invariance (equal factor loadings and item intercepts) were examined using the aforementioned model fit criteria and a criterion of CFI change of < 0.01 as evidence of invariance across increasing restrictive models (e.g., configural to metric, metric to scalar; Cheung & Rensvold, 2002). Readers interested in a more detailed introduction to measurement invariance and a fully worked pediatric psychology example are encouraged to consult (Gregorich, 2006) and (Kamody et al., 2014). Results Factor Analyses Step 1: EFAs were conducted specifying one to eight factors. The fit statistics presented in Table I suggested that models with 4, 5, 6, and 7 factor solutions (Models D to G) provided a good fit to the data without potentially overfactoring. Step 2: The factor loadings from these four models were reviewed, and items with low or double loadings from Step 1 were removed. Four additional EFAs were then conducted. This resulted in either 4 or 5 factor models, labeled D2, E2, F2, and G2, to indicate their source (e.g., Model D), and that they were then altered (version 2 ). Each of these models had acceptable fit indices as shown in Table I (see Step 2, Models D2 G2). Step 3: CFAs of Models D G (from Step 1) and Models Table I. Fit Statistics for Exploratory and Confirmatory Factor Analyses With Cutoff Criteria for Fit Indicated Parenthetically Model v 2 df p (>.05) CFI RMSEA (>0.95) (<0.05) Step 1: Initial EFAs with learning sample A (1 factor) < B (2 factors) < C (3 factors) < D (4 factors) < E (5 factors) F (6 factors) G (7 factors) H (8 factors) Step 2: Secondary EFAs with learning sample D2 (4 factors) < E2 (5 factors) F2 (5 factors) G2 (4 factors) Step 3: CFAs with cross-validation sample D (4 factors) < D2 (4 factors) < D2.1 (4 factors < drop Q10) D2.2 total score < E (5 factors) < E2 (4 Factors) < E2.1 (5 factors < keeping social) F (6 factors) < F2 (3 factors) < G (7 factor) no convergence G2 (4 factors) <

6 Pediatric Constipation 819 D2 G2 (Step 2) with the cross-validation sample were conducted after dropping items using the aforementioned item-culling strategy. Two additional models were evaluated in Step 3, Model D2, which dropped one item (Item 10, I feel like a partner with child s constipation treatment team ) that diverged conceptually from other items in its factor. An additional model (Model E2.1) retained a conceptually important factor reflecting children s social difficulty that was dropped in E2 owing to the item s low factor loading (<0.40). The fit statistics for these models are presented in Table I (see Step 3), and were mixed with regard to model fit and convergence. More specifically, Models D, D2, D2.1, E2, and F had an acceptable fit across all indices. The final step reviewed the five models from Step 3 that had an acceptable fit. The factors (based on item content) from Models D, D2, and D2.1 appeared to reflect problems in quality of life related to parental burden/distress, family conflict, difficulties with the medical team, and worry about social impact. Model E2 and E2.1 were composed of factors in which the item content reflected Parent and Child Distress/ Hopelessness/Isolation (Factor 1), Family Conflict (Factor 2), Pain/Sickness/Behavior Change due to Constipation (Factor 3), Difficulties with the Medical Team (Factor 4), and Worry about Social Impact (Factor 5, model E2.1 only). Discussion among the lead investigators unanimously agreed that the D Models should be pursued over concerns that Models E and E2.1 contained heterogeneous items, which made factor interpretation less clear than for the factors based on the D Models. Conceptually, the D Models also were more closely aligned with the constructs that were extracted from the initial qualitative phase of this project (Kaugars et al., 2010), namely, Parental Burden/Distress, Family Conflict, Difficulties with the Medical Team, and Worry about Social Impact. Closer review of Model D, D2, and D2.1 revealed two items with substantially low factor loadings on the Difficulties with the Medical Team: Q10 ( I feel like a partner with child s constipation treatment team, Model D2), and Q22 ( Following the treatment recommendations has improved daily life, Models D and D2). In light of the most favorable fit statistics and because neither of these problematic items was included in Model D2.1, this model was chosen as the final model based on statistical, conceptual, and practical grounds (the least number of items). One additional model, Model 2.2, was tested based on Model 2.1, in which a total score (second order) was modeled using the four subscales as indicators. This model provided an excellent fit to the data. Each subscale was a significant indicator of its respective factors and correlated with one another, suggesting that a total score could be formed. Table II Table II. Internal Consistency Estimates and Descriptive Statistics for the POOP-C Subscales and Total Score Total scale and subscales M SD Total score (a combined ¼.891, a learning ¼.889, a cross validation ¼.896) Burden/worry (a combined ¼.804, a learning ¼.847, a cross validation ¼.787) Q25- I worry that child has pain when pooping Q26- I worry that child experiences pain because of constipation Q1- I spend a lot of time for child s constipation Q2- I worry that there is something seriously wrong because of unimproved constipation Q6- Child is upset that we are still treating the constipation Q5- I am afraid that child s constipation will never get better Q7- I feel like very few people understand constipation problem Q27- I am concerned that child is ashamed of the constipation Family (a combined ¼.845, a learning ¼.805, a cross validation ¼.854) Q21- Asking child to sit on the toilet causes conflict Q36- Child does not trust my efforts to help him/her Q44- I am upset because my child resists treatment Q38- Family has trouble getting alone because of constipation problem Q37- Child hides info because of the fear of treatment Q16- I worry that the relationship with child is damaged because of treatment Q43- I feel guilty when we cannot follow all treatment recommendations Q33- Child becomes upset when I ask him/her to go to the bathroom Treatment team(a combined ¼.801, a learning ¼.836, a Cross validation ¼.788) Q45- I think constipation treatment team could have been more helpful Q46- I don t think constipation treatment team understands/listens to our problems Q47- I disagree with treatment team on treatment issues Q18- I do not feel well supported by my child s treatment team Q12- I feel confused when information I find myself is different from that giving by treatment team Social (a combined ¼.836, a learning ¼.851, a Cross validation ¼.831) Q50- I am concerned that child does not make friends because of the constipation problem Q48- I worry that child does not get asked to play because the constipation problem Q49- I worry that others will find out about child s constipation problem

7 820 Silverman et al. Q25 Burden/ Worry Q26 Q1 Q2 Q6 Q7 Q Q27 Q Family Q36 Q44 Q Q Q16 Q43 Q Treatment Q45 Q Q Q Q Q50 Social Q Q49 Figure 1. Final confirmatory factor analysis model with standardized values, all paths significant at p <.01. presents the internal consistency estimates and descriptive statistics for the subscales, items, and total score. Figure 1 shows the final CFA model with standardized values. This model reflects the statistical relation between constructs and the underlying items of the final 24-item instrument. Given the wide age range of participants, measurement Invariance analyses were conducted with the combined EFA/CFA sample comparing younger children (88 months old, n ¼ 196) with older children (>88 months old, n ¼ 192). The measure demonstrated scalar invariance (equal factor loadings and

8 Pediatric Constipation 821 item intercept) using CFI change of < 0.01 as evidence of invariance (Cheung & Rensvold, 2002), suggesting that there were no differences in factor structure or a worsening of model fit as a function of participant age. Details of the invariance analyses, including model fit statistics, can be found in the Supplementary Materials. The convergent validity of the POOPC was determined by examining the correlations between the POOPC and the FDI, PIP, PedsQL TM, and PedsQL TM FIM, with the expectation that they would be moderately correlated. Moderate to large correlations (Table III), as defined by Cohen (1988), were found between the POOPC total score and the FDI total score, PIP frequency total score, PIP difficulty total score, PedsQL TM Core Scales total scores (all age versions), and PedsQL TM FIM total score. Correlations with these measures subscales can be found in the Supplementary Materials. The discriminant validity of the POOPC was determined by examining the correlations between the POOPC and the PSC, with the expectation that they Table III. Correlations Between the POOP-C Total Score and Subscales With the Functional Disability Inventory (FDI), Pediatric Inventory for Parents (PIP), Pediatric Quality of Life v (PedsQL TM ), PedsQL TM Family Impact Module (FIM), and PSC Questionnaire POOP-C total POOP-C BW POOP-C FA POOP-C TR POOP-C SO FDI total PIP difficulty total PIP frequency total PedsQL TM total (age: 2 4 years) PedsQL TM total (age: 5 7 years) PedsQL TM total NS.226 (age: 8 12 years) PedsQL TM total NS NS.502 (age: years) FIM total PSC total Note. All values significant at p <.05 unless denoted by NS. NS ¼ nonsignificant. Established measures tend to be non-disease-specific and/or limited their coverage PIP Communication Subscales; PIP MC ¼ PIP Medical Care Subscales; PIP ED ¼ PIP Emotional Distress Subscales; PIP RF ¼ PIP Role Functioning Subscales; PedsQL TM PHY ¼ PedsQL TM Physical Functioning Subscales; PedsQL TM EMO ¼ PedsQL TM Emotional Functioning Subscales; PedsQL TM SOC ¼ PedsQL TM Social Functioning Subscales; PedsQL TM SCH ¼ PedsQL TM School Functioning Subscales; PedsQL TM FIM ¼ PedsQL TM Family Impact Module; PedsQL TM FIM Parent HRQL ¼ PedsQL TM FIM Parent Health Related Quality of Life Summary; PedsQL TM FIM FAM FUNC ¼ PedsQL TM FIM Family Functioning Summary; PedsQL TM FIM COG ¼ PedsQL TM FIM Cognitive Functioning Subscale; PedsQL TM FIM COMM ¼ PedsQL TM FIM Communication Subscale; PedsQL TM FIM WO ¼ PedsQL TM FIM Worry Subscale; PedsQL TM FIM DA ¼ PedsQL TM FIM Daily Activities Subscale; PedsQL TM FIM FR ¼ PedsQL TM FIM Family Relationships Subscale would not be correlated highly with one another. A moderately small correlation was found between the POOPC and the PSC (r ¼.383), which was lower than the correlations between the PSC and other established measures of HRQL-related constructs including the FDI, PIP, PedsQL TM total score (all age versions), and PedsQL TM FIM total score. For example, the POOPC had stronger correlations with the PIP PedsQL TM Core and PedsQL TM FIM total score than with the PSC, which is indicative of both convergent and discriminate validity, and thus, construct validity. The correlations between the POOPC total score and subscales with the FDI, PIP, PedsQL TM, PedsQL TM FIM, and PSC totals are presented in Table III. Discussion This article has provided preliminary evidence that the POOPC is a reliable and valid instrument that can be used to assess the effects of constipation and constipation with fecal incontinence on the HRQL in children with these specific conditions. Although there are general instruments available that measure HRQL in pediatric populations, there is, to our knowledge, no other English language instrument that focuses on the problems that affect HRQL in children who have these conditions. To the best of our knowledge, this is the first attempt to create such a measure. The four subscales of the POOPC, which include Burden/ Worry, Family, Treatment, Social, and a Total Score, should be helpful to both clinicians and researchers alike. The POOPC may be particularly helpful for the early detection and treatment of children and other family members who are having poor adjustment and might benefit from additional services with a pediatric psychologist. As a clinical tool, the POOPC will be useful in identifying specific domains of treatment need for children and their families. Each of the subscales offers distinct information about the concerns of families affected by childhood constipation and fecal incontinence. The Burden/Worry subscale assesses a pattern of caregiver concerns that focus on their child s negative experiences of having constipation (e.g., pain, lack of improvement of symptoms, duration of symptoms, and embarrassment). The Family subscale assesses a pattern of caregiver concerns focusing on conflicts between family members related to following the constipation treatment regimen (e.g., asking the child to use the toilet, child resisting treatment, worry that the relationship with the child is damaged owing to treatment). The Treatment subscale assesses a pattern of caregiver concerns with a focus on a lack of trust or confidence in providers (e.g., I disagree with treatment options, I do not feel well supported ). The Social subscale assesses a pattern of

9 822 Silverman et al. caregiver concerns that focus on peer relationships (e.g., child is not asked to play, concern that others may find out about the problem). The Total Score assesses a pattern of caregiver concerns that focus on HRQL issues affecting families of children who have chronic constipation. This measure may also facilitate research regarding the lives of families of children with constipation and constipation with fecal incontinence. Earlier studies have been limited by either their use of generic HRQL measures and/or a narrower focused on particular domains of respondent functioning. Examples include the PedsQL FIM and PIP focus on parents experiences related to having a child with a nonspecific illness. These scales include on the frequency and difficulty of parental problems with communication, medical care, emotional distress, role functioning (PIP), and physical, emotional, social, cognitive functioning function, and communication, worry, daily activities, and family relationships (PedsQL FIM). In terms of youth QOL, the PedsQL focuses on youth s overall, physical, emotional, social, and school functioning or overall functioning (FDI). In contrast, a strength of the POOPC is that it assesses multiple domains of diseasespecific functioning for both youth- and parent-related (i.e., youth burden, family impact, youth social functioning, and treatment team relationships), which can be shown with future studies to increase this measure s specificity and sensitivity. Future investigators may wish to use the POOPC to resolve the discrepancies of earlier studies. Questions regarding effects of age of onset and duration of and severity of symptoms and/ or patterns of these variables may be clarified (Berlin, Parra, & Williams, 2014; Berlin, Williams, & Parra, 2014). Investigators may also better understand the relative effects constipation has on HRQL by comparing affected children to healthy controls and to children with other gastrointestinal diseases (e.g., inflammatory bowel disease, gastro-esophageal reflux disease). As with all research studies, this study has limitations that should be acknowledged. First, this sample draws from specialty clinics, and thus, participants likely have had long-standing difficulties with constipation and may represent the most severe of the clinical cases. This may have skewed the response pattern to more severe symptoms and may thus have more negative reporting of HRQL. This study does not have a healthy control group, which may be useful in future work to develop a clinical cutoff score. The instrument was developed from caregiver responses only, and there is not, at this time, a child self-report version. Likewise, caregivers of children with other comorbid conditions were excluded, potentially limiting the instrument s usefulness. Furthermore, little is known about the families of those who declined participation. This is a potential source of selection bias and represents a threat to the generalizability of the instrument in its current form. Finally, there may be other HRQL instruments developed for toileting concerns of which the authors are unaware, but may have been useful for the validation process. Future studies of the measure may wish to expand our sample size and increase the regional representation to ensure that all group differences are accounted for in later versions of this instrument and expand the generalizability of our findings. A larger sample may also help future investigators to determine if response patterns to the instrument differ by sex, ethnic, developmental level, and/or racial differences. Efforts to expand the normative sample to include families of children with comorbid conditions including chronic health concerns and/or developmental concerns would enhance our understanding of properties of the instrument when used with special populations. Development of a youth self-report would also be beneficial, as affected children and their caregivers may have different perceptions regarding factors that affect HRQL. Future studies would also benefit from additional information regarding nonparticipants as this would be useful in determining possible bias. Finally, additional validation calculation may be beneficial, as other instruments are identified or are developed. In summary, the POOPC is a reliable and valid measure of HRQL for pediatric constipation with and without fecal incontinence. The POOPC will add precision to our understanding of these conditions, and may offer providers a method of detecting families of children who may benefit from clinical support. This measure may have wide application and shows promise both as a clinical tool and as a research instrument. Supplementary Data Supplementary data can be found at oxfordjournals.org/. Funding Funding support was provided by Takeda Pharmaceuticals. Conflicts of interest: None declared. References Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, Berlin, K. S., Parra, G. R., & Williams, N. A. (2014). An introduction to latent variable mixture modeling (part 2): Longitudinal latent class growth analysis and growth mixture models. Journal of Pediatric Psychology, 39, doi: /jpepsy/jst085 Berlin, K. S., Williams, N. A., & Parra, G. R. (2014). An introduction to latent variable mixture modeling (part 1): Overview and cross-sectional latent class and latent profile

10 Pediatric Constipation 823 analyses. Journal of Pediatric Psychology, 39, doi: /jpepsy/jst084 Bollen, K. A. (1989). Structural equations with latent variables. New York, NY: Wiley. Bongers, M. E., van Dijk, M., Benninga, M. A., & Grootenhuis, M. A. (2009). Health related quality of life in children with constipation-associated fecal incontinence. Journal of Pediatrics, 154, doi: / j.jpeds Bongers, M. E., van Dijk, M. P., Reitsma, J. B., & Benninga, M. A. (2010). Long-term prognosis for childhood constipation: Clinical outcomes in adulthood. Pediatrics, 126(1), e156 e162. Brooks, R. C., Copen, R. M., Cox, D. J., Morris, J., Borowitz, S., & Sutphen, J. (2000). Review of the treatment literature for encopresis, functional constipation, and stool-toileting refusal. Annals of Behavioral Medicine, 22, Browne, M. W. (2001). An overview of analytic rotation in exploratory factor analysis. Multivariate Behavioral Research, 36(1), doi: / S mbr3601_05 Browne, M. W., & Cudeck, R. (1993). Alternate ways of assessing model fit. In K. A. Bollen & J. S. Long (Eds.), Testing structural equation models (pp ). Newbury Park, CA: Sage. Cambell, L. K., Cox, D. J., & Borowitz, S. M. (2009). Elimination disorders: Enuresis and encopresis. In M. S. Roberts & R. G. Steele (Eds.), Handbook of pediatric psychology (4th ed., pp ). New York, NY: Guilford Press. Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant validation by the multitrait-multimethod matrix. Psychological Bulletin, 56, Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling, 9, doi: / S SEM0902_5 Cohen, J. W. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates. Cox, D. J., Morris, J. B. Jr., Borowitz, S. M., & Sutphen, J. L. (2002). Psychological differences between children with and without chronic encopresis. Journal of Pediatric Psychology, 27, Drossman, D. A. (2006). Rome III: The functional gastrointestinal disorders (3rd ed.). McLean, VA: Degnon Associates. Fabrigar, L. R., Wegener, D. T., MacCallum, R. C., & Strahan, E. J. (1999). Evaluating the use of exploratory factor analysis in psychological research. Psychological Methods, 4, doi: / X Gregorich, S. E. (2006). Do self-report instruments allow meaningful comparisons across diverse population groups? Testing measurement invariance using the confirmatory factor analysis framework. Medical Care, 44(Suppl. 3), S78 S94. doi: /01.mlr f Hu, L., & Bentler, P. M. (1995). Evaluating model fit. In R. H. Hoyle (Ed.), Sturctural equation modeling: Concepts, issues, and applicaitons (pp ). Thousand Oaks, CA: Sage Publications. Jellinek, M. S., Murphy, J. M., Robinson, J., Feins, A., Lamb, S., & Fenton, T. (1988). Pediatric symptom checklist: Screening school-age children for psychosocial dysfunction. Journal of Pediatrics, 112, Joinson, C., Heron, J., Butler, U., & von Gontard, A. (2006). Psychological differences between children with and without soiling problems. Pediatrics, 117, doi: /peds Kamody, R. C., Berlin, K. S., Hains, A. A., Kichler, J. C., Diaz-Thomas, A. M., & Ferry, R. J. (2014). Assessing measurement invariance of the Diabetes Stress Questionnaire in youth with type 1 diabetes. Journal of Pediatric Psychology, 39, Kaugars, A. S., Silverman, A., Kinservik, M., Heinze, S., Reinemann, L., Sander, M.,... Sood, M. (2010). Families perspectives on the effect of constipation and fecal incontinence on quality of life. Journal of Pediatric Gastroenterology and Nutrition, 51, doi: /MPG.0b013e3181de0651 Liem, O., Harman, J., Benninga, M., Kelleher, K., Mousa, H., & Di Lorenzo, C. (2009). Health utilization and cost impact of childhood constipation in the United States. Journal of Pediatrics, 154, doi: / j.jpeds MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power analysis and determination of sample size for covariance structure modeling. Psychological Methods, 1, doi: / X Matza, L. S., Swensen, A. R., Flood, E. M., Secnik, K., & Leidy, N. K. (2004). Assessment of health-related quality of life in children: A review of conceptual, methodological, and regulatory issues. Value in Health, 7, doi: /j x McDonald, R. P. (2005). Semiconfirmatory factor analysis: The example of anxiety and depression. Structural Equation Modeling, 12(1), doi: / s sem1201_9 Rome Foundation. (2006). Guidelines Rome III Diagnostic Criteria for Functional Gastrointestinal Disorders. Journal of Gastrointestinal and Liver Diseases, 15, Streisand,R.,Braniecki,S.,Tercyak,K.P.,&Kazak,A.E. (2001). Childhood illness-related parenting stress: The Pediatric Inventory for Parents. Journal of Pediatric Psychology, 26(3), doi: /jpepsy/ Treurniet, H. F., Essink-Bot, M. L., Mackenbach, J. P., & van der Maas, P. J. (1997). Health-related quality of life: An indicator of quality of care? Quality of Life Research, 6, van den Berg, M. M., Benninga, M. A., & Di Lorenzo, C. (2006). Epidemiology of childhood constipation: A systematic review. American Journal of Gastroenterology, 101, doi: /j x Varni, J. W., Burwinkle, T. M., Sherman, S. A., Hanna, K., Berrin, S. J., Malcarne, V. L., & Chambers, H. G. (2005). Health-related quality of life of children and adolescents with cerebral palsy: Hearing the voices of the children. Developmental Medicine and Child Neurology, 47,

11 824 Silverman et al. Varni, J. W., Seid, M., & Rode, C. A. (1999). The PedsQL: Measurement model for the pediatric quality of life inventory. Medical Care, 37, Varni, J. W., Sherman, S. A., Burwinkle, T. M., Dickinson, P. E., & Dixon, P. (2004). The PedsQL Family Impact Module: Preliminary reliability and validity. Health Qual Life Outcomes, 2, 55. doi: / Walker, L. S., & Greene, J. W. (1991). The functional disability inventory: Measuring a neglected dimension of child health status. Journal of Pediatric Psychology, 16, Yates, A. (1987). Multivariate exploratory data analysis: A perspective on exploratory factor analysis. Albany, NY: State University of New York Press. Youssef, N. N., Langseder, A. L., Verga, B. J., Mones, R. L., & Rosh, J. R. (2005). Chronic childhood constipation is associated with impaired quality of life: A case-controlled study. Journal of Pediatric Gastroenterology and Nutrition, 41(1),