Health-related quality of life in children and adolescents with cerebral palsy

At the time of data collection in July 2014, the Slovenian National Cerebral Palsy Registry (SRCP) included 371 children of all ages. They were registered through neurodevelopmental paediatricians in regional outpatient clinics covering around 90% of Slovenian paediatric CP population. Out of original 150 children aged 8-16 years, randomly selected from the SRCP, contact data for 122 families were available; 91 caregivers (of 43 children aged 6-12 years and 48 adolescents aged 13-16 years) were willing to cooperate. Children with tested IQ score >70 or attending regular school were considered cognitively able to self-report, and 48 of them (26 children and 22 adolescents) agreed to do so.

The caregivers of 122 children with CP were contacted by telephone at one or (in the case of first call nonresponders) at two occasions by a single physician. Study aims were explained to them and caregivers were invited to participate. Questionnaires were sent to 103 caregivers who accepted the invitation, together with written instructions and informed consent form.

DISABKIDS instruments were used to assess HRQoL. DISABKIDS Chronic Generic Measure – long version (DCGM-37) – contains 3 domains (Mental, Social and Physical) that are further divided in order to measure the following 6 dimensions: Independence, Emotion, Social inclusion, Social exclusion, Limitation, and Treatment. All items are Likert-scaled and transformed to a scale from 0 to 100, where higher scores indicate a better HRQoL. According to field testing, the instrument shows sound psychometric properties with satisfactory reliability, construct validity, convergent and discriminant validity (15). It is available in self-reported and proxy versions, and it was designed for children aged 8-16 years. DISABKIDS Cerebral Palsy Module (DCSM-CPM) and DISABKIDS Epilepsy Module (DCSM-EM) both address two dimensions, namely: Impact (10 items) and Communication (2 items) for DCSM-CPM, and Impact and Social (both 5 items) for DCSM-EM.

All six questionnaires (self-reported and proxy versions) underwent the translation and validation procedure with the guidance of European DISABKIDS Group and DISABKIDS manual, and were approved for the use in the study (17). Paediatric Quality of Life Inventory 4.0 (PedsQL) Generic Core Scales was used for comparison, as it has already been validated in the Slovenian language. Its concept is very similar to DCGM-37 and consists of 23 items encompassing: Physical functioning, Emotional functioning, Social functioning and School functioning (18).

The assessment bundle included: DCGM-37, DCSM-CPM, DCSM-EM (in cases of comorbid epilepsy) and PedsQL. In order to gain information about clinical usefulness of selected HRQoL measures, we added 5 additional questions asking for personal opinion about the questionnaires. All questionnaires were sent in proxy versions and, if applicable, in self-reported versions.

Items from all questionnaires were scored according to DISABKIDS manual (17). Raw scores were converted into values on a 0-100 scale. A score was calculated if at least 80% of items were answered. Overall scores were calculated as a sum of all separate scores. Each score was expressed as mean and standard deviation (SD).

The effect of various factors on different dimensions and the final score was evaluated by independent samples t-test or one-way ANOVA, as the distribution of results was close to normal. Linear regression model was constructed to determine the most important factors influencing HRQoL. The differences were expressed as mean difference and 95% confidence interval (CI). A value of p<0.05 was considered statistically significant.

The partial and final scores of self-reports and proxyreports were compared using Pearson’s Product Moment Correlation coefficient (Pearson’s r).

The partial and final scores of self-reports and proxyreports were also compared to subscales and the total score of PedsQL measure, where Pearson’s r was used again.

All the analyses were conducted with the SPSS (Statistical Package for Social Sciences Program) version 20.0.

The mean age of all participants with CP was 12 years and 4 months (SD 2.02); 43 (48%) were children (the mean age is 10 years, 6 months, SD 1.2) and 48 (53%) adolescents (the mean age is 14 years, SD 0.9). The sample consisted of 53 (58%) males and 38 (42%) females.

Among 48 self-reporting participants, there were 26 (54%) children (the mean age is 10 years 8 months, SD 1.1) and 22 (46%) adolescents (the mean age is 13 years, 11 months, SD 0.8). Gender distribution across the selfreported sample was balanced with 23 (48%) males and 25 (52%) females.

Compared to the data in the SRCP registry, the study sample shows similar gender, CP type and GMFCS distribution, and was therefore found to be representative of the population of Slovenian paediatric patients with CP (19).

All characteristics of children concerning their demographic parameters, level of impairment, comorbidities, therapeutic interventions and family social status are listed in Table 2 and Table 3.

Caregivers rated their children’s HRQoL as ‘good’ (DCGM-37 mean total transformed score 66.9, SD 18.2). The worse scored domains were Social Inclusion (mean 53.2, SD 24.5) and Independence (mean 58.1, SD 24.4). The highest rated were Emotion (mean 78.3, SD 20.4) and Medication (mean 79.8, SD 21.0). HRQoL of children was scored higher than HRQoL of adolescents (mean 75.1, SD 12.5 versus mean 60.5, SD 19.5; mean difference 14.64, 95% CI 7.21 - 22.07; p<0.001). The largest differences were for Independence (mean 67.79, SD 20.84 versus 50.34, SD 24.54; mean difference 17.45, 95% CI 7.36 - 27.54; p=0.001), Social Inclusion (mean 65.23, SD 19.17 versus 44.02, SD 24.34; mean difference 21.21, 95% CI 11.40 - 31.02; p<0.001) and Social Exclusion domains (mean 82.52, SD 13.65 versus 71.74, SD 22.76; mean difference 10.78, 95% CI 2.23 - 19.34; p=0.014). There were some missing data for each item, and 5 items had a missing data rate equal to 10% or more. In nine questionnaires, scoring was not possible due to too many missing values (>20% of unanswered items).

Seventy-nine DCSM-CPM completed proxy reports with a mean total transformed score 72.6, SD 16.9. The subscale Communication (mean 78.0, SD 29.2) was scored higher than Impact (mean 71.0, SD 17.3). Caregivers of adolescents scored lower (mean 69.2, SD 18.6) compared to caregivers of children (mean 76.8, SD 13.6), but the difference was borderline statistically important (p=0.050).

Nineteen DCSM-EM proxy reports were valid and gave the total transformed score of mean 89.3, SD 14.3.

Among DCGM-37 self-reporters, HRQoL was perceived as ‘good’ (the mean total transformed score is 75.6, SD 15.9). The worse scored subscales were, as in proxy-reports, Social inclusion (mean 65.7, SD 19.5) and Independence (mean 70.4, SD 20.1). The highest score was given to the sub-scale Emotion (mean 85.3, SD 17.7). Children rated their HRQoL better than adolescents (mean 80.8, SD 10.0 versus mean 69.4, SD 19.4; mean difference 11.37, 95% CI 2.57 - 20.17; p=0.01). Between subscales, the biggest difference was seen in Emotion (mean 91.07, SD 9.76 versus mean 78.57, SD 22.43; mean difference 11.37, 95% CI 2.57 - 20.17, p=0.01), Physical (mean 80.29, SD 11.82 versus mean 67.99, SD 23.48; mean difference 12.30, 95% CI 1.74 - 22.85, p=0.02) and Social exclusion domains (mean 84.94, SD 10.54 versus mean 74.62, SD 21.28; mean difference 10.31, 95% CI 0.78 - 19.84; p=0.035). There were no missed items in the sample.

Twenty-three children and 21 adolescents filled out DCSMCPM self-reports. The mean transformed total score of all 44 reports was 78.8, SD 14.7. The Communication domain scored higher (92.3, SD 20.6) than Impact (76.1, SD 15.0). No differences were observed among children and adolescents in DCSM-CPM reports (p=0.14).

Only 6 children and adolescents able to self-report had concomitant epilepsy. They scored their HRQoL through DCSM-EM as very good (80.4, SD 24.1).

The correlation between DCGM-37 proxy and self-reports was good (Pearson r=0.80 for total scores and r=0.59 - 0.80 for separate domains, where only Social Inclusion and Social Exclusion domains resulted in r<0.70). The absolute difference between proxy and self-reported scores was significant, self-reporting participants rating their HRQoL higher than caregivers (mean 75.6, SD 15.9 versus mean 72.3, SD 17.9; the mean difference 3.23, 95% CI 0.03 - 6.43; p=0.048).

The same was found for DCSM-CPM measure with the total transformed score r=0.76 and r=0.73 and r=0.81 for Impact and Communication subscales. There was no difference between proxy and self-reported DCSM-CPM scores (p=0.97).

The mean total score of 78 PedsQL – proxy measures – was 61.5, SD 21.4, and of 41 PedsQL self-reported measures 75.6, SD 19.6. The correlation between DISABKIDS and PedsQL total transformed scores was very good (Pearson r=0.81 for self-reports and r=0.86 for proxy reports). Domains measuring similar concepts showed high correlation as well: r=0.70 for physical domains (self) and r=0.81 for physical domains (proxy), r=0.74 for emotional domains (self) and r=0.64 for emotional domains (proxy) and r=0.75 for social domains (self) and r=0.80 social domains (proxy).

DCGM-37 proxy reports: lower age was found a strong single predictor of better HRQoL (mean 75.1, SD 12.5 versus mean 60.5, SD 19.5; the mean difference 14.64, 95% CI 7.21 - 22.07; p<0.001). There was a negative correlation between HRQoL and disease severity. GMFCS level alone, tested with one-way ANOVA, was negatively associated with HRQoL (p=0.023). So were pain (the mean difference 23.21, 95% CI 14.71 - 31.72; p<0.001), disturbed sleep (the mean difference 17.96, 95% CI 8.88 - 27.05; p=0.001) and cognitive abilities (the mean difference 13.38, 95% CI 5.84 - 20.93; p=0.001). However, the multivariable analysis of the same variables showed that the main factors reducing HRQoL were pain (p<0.001) and disturbed sleep (p=0.002), and not GMFCS itself (p=0.735). Adjusted r score for this model was 0.49 and the strongest standardised regression coefficient was found for pain -0.47, p<0.001. No correlation was found between the number of implemented therapeutic interventions and HRQoL, nor for comorbidities, such as epilepsy, and language, speech and attention disorders. There was also no HRQoL association with parents’ education, employment status, financial support, child cognitive abilities and schooling type.

DCGM-37 self-reports: in the single variable analysis, besides higher age (the mean difference 11.37, 95% CI 2.57 - 20.17; p=0.012), the main factors reducing HRQoL were pain (the mean difference 18.57, 95% CI 6.03 - 31.11; p=0.005) and comorbidities (the mean difference 9.56, 95% CI 0.24 - 18.88; p=0.045). In the multivariable model using these variables, the adjusted r score was 0.27, and the strongest standardised regression coefficient was found for pain (-0.39, p=0.009), whereas age was only borderline significant (p=0.059).

DCSM-CPM proxy reports: in the multivariable model using variables that tested significant in univariate models (pain, disturbed sleep, GMFCS, cognitive impairment, speech impairment and epilepsy), pain was the only factor negatively influencing HRQoL (adjusted r score 0.34, standardized regression coefficient -0.30, p=0.009). DCSM-CPM self-reports: pain was the only significant single factor related to a lower perception of HRQoL (p=0.027).

The caregivers needed, on average, 13.6 minutes to fill out DCGM-37 and DCSM-CPM proxy modules. Forty-six caregivers (54%) thought they were useful and 40 (47%) interesting. Eighty-three (87%) would potentially fill them out again, 50 (58%) would do that gladly.

The self-reported children and adolescents needed, on average, 14.5 minutes to fill out both questionnaires. Twenty-one (48%) of them found the questionnaires interesting and 16 (36%) useful. Only 3 participants (7%) considered the questionnaires stupid. The items made 2 participants (5%) feel uncomfortable, 12 (27%) felt a bit embarrassed, but were not bothered by them, and 30 (68%) were not bothered by them not at all. Forty-one (93%) individuals would potentially fill out the questionnaires again, 20 (45%) would do that gladly.