Table 6 Summary of PBM performance for included studies

Bartlett et al. (2010) [33]

Adolescents with cerebral palsy

HUI3 vision, cognition and pain dimensions steadily declined as GMFCS level increased, statistical significance not reported.

Examining correlation coefficients, there was no indication that the HUI3 vision (r = 0.01; p = 0.92), cognition (r = − 0.05; p = 0.59) or pain dimensions (r = 0.16; p = 0.07) were determinants of motor capacity, as measured using the GMFM-66.

NA

NA

Individuals with a GMFCS level of V exhibited the largest mean decreases in HUI3 dimension levels over time, ranging from − 0.2 (±1.2) for the HUI3 vision dimension to − 0.3 (±1.3) for the HUI3 cognition and pain dimensions.

Bray et al. (2017) [15]

Children and adolescents with impaired mobility (relevant conditions: cerebral palsy, hemiplegia, muscular dystrophy)

NA

NA

Large variance between mean utility scores derived from different PBMs, ranging from 0.24 (EQ-5D-Y) to 0.53 (HUI2) for child self-reported utility, and from 0.01 (EQ-5D-Y) to 0.49 (HUI2) for parent-reported proxy utility.

A significant respondent type effect was found, with mean child self-reported utility scores significantly (p ≤ 0.021) higher than equivalent proxies on all PBMs.

Significant strong correlations were found between utility scores for children/parent proxies using all measures: EQ-5D-Y (r = 0.67; p = 0.026), HUI2 (r = 0.73; p = 0.005) and HUI3 (r = 0.84; p < 0.001).

Using Bland-Altman plots, sufficient agreement between utility scores for children/parent proxies was found for the HUI2 (CL = 0.22) and HUI3 (CL = 0.22). EQ-5D-Y exhibited clinically important discrepancies between child and parent proxy responses (CL = 1.04).

NA

Burstrom et al. (2014) [11]

Children and adolescents with functional motor, orthopaedic and medical disabilities (relevant conditions: artrogryposis multiple congenital, myelomeningocele, cerebral palsy, orthopaedic lower limb deformities, achondroplasia)

Statistically significant (p ≤ 0.001) differences between case and control groups on all dimensions. Mean dimension scores not reported, proportions of patients choosing each dimension level indicate case group reported more problems on all dimensions than the control group: 83% of case group reported some/a lot of problems on any dimension, compared to 37% of control group; likewise 21% of case group reported extreme problems on any dimension, compared to 2% of control group.

Strong significant correlation was found between the EQ-5D-Y anxiety/depression dimension and the KIDSCREEN-27 psychological well-being dimension (r = − 0.51; p = 0.001); KIDSCREEN-27 physical well-being dimension (r = − 0.53; p = 0.001); and life satisfaction ladder (LSL) (r = − 0.54; p < 0.001). Moderate correlation also found between this dimension and the KIDSCREEN-10 HRQoL index score (r = − 0.50; p = 0.001) and self-reported health (SRH) (0.42; p = 0.007).

The self-care EQ-5D-Y dimension exhibited moderate correlation with the KIDSCREEN-27 physical wellbeing dimension (r = − 0.37; p = 0.028) and the usual activities EQ-5D-Y dimension was moderately correlated with the KIDSCREEN-27 psychological wellbeing dimension (r = − 0.35; p = 0.027). All other correlations were weak or absent.

NA

NA

NA

Cavazza et al. (2016) [34]

Adolescents and adults with Duchenne muscular dystrophy

NA

NA

NA

NA

NA

Christensen et al. (2016) [35]

Children and adolescents with cerebral palsy

NA

Using multivariate linear regression, a significant association was found between the HUI3 pain dimension score at baseline and GMFCS level (b = − 0.11, β = − 0.15; p < 0.036; 95% CI − 0.21 to − 0.01): higher pain score at baseline was associated with greater improvement in pain status in GMFCS level I compared to level V.

NA

NA

Using one-way ANOVA analysis, a significant association was found between physician primary pain aetiology and change in HUI3 pain status (p = 0.001): children with musculoskeletal pain at baseline showed significant improvements (mean change 0.55; 95% CI 0.053 to 1.05) compared to children without pain at baseline (mean change − 0.39; 95% CI − 0.62 to − 0.15) (p = 0.006). No other associations were reported. HUI3 pain dimension scores did not change significantly over time: median score of 2 (out of 5) at both visits. However, 55% of children changed pain status over time: 34% worsening, 21% improving.

Findlay et al. (2015) [36]

Children with cerebral palsy

NA

NA

NA

NA

NA

Hendriksz et al. (2014) [37]

Children and adults with Morquio A syndrome

A significant effect of wheelchair use on utility outcomes was reported; significant differences reported between: adult non-wheelchair users and occasional wheelchair users (p = 0.0115); adult occasional wheelchair users and full-time wheelchair users (p = 0.0007); child occasional wheelchair users and full-time wheelchair users (p = 0.0018); and child non-wheelchair users and full-time wheelchair users (p = 0.0018). Children who occasionally used a wheelchair had a higher mean utility score on average than non-wheelchair users, although both groups had higher average utility scores than full-time wheelchair users.

NA

NA

NA

NA

Karmur and Kulkarni (2018) [38]

Children and adolescents with spina bifida (myelomeningocele) and shunted hydrocephalus

Using multivariate regression analysis, anatomical level of myelomeningocele had a significant effect on utility score (p = 0.01): lower anatomical level of myelomeningocele was associated with a higher utility score.

NA

NA

NA

NA

Kennes et al. (2002) [39]

Children with cerebral palsy

NA

Using Kendall’s tau-b test of association, the HUI3 dimension most associated with GMFCS level was ambulation (tau-b = 0.82; p < 0.01): higher GMFCS level was associated with increased mobility impairment.

Moderate correlations (ranging from tau-b = 0.36 to 0.58; p < 0.01) were found between GMFCS level and the vision, speech and dexterity dimensions. Overall patterns were similar to the ambulation dimension, but correlations were weaker.

Hearing (tau-b = 0.16; p = 0.04) and cognition (tau-b = 0.27; p < 0.01) dimensions had statistically significant but low association with GMFCS level. The emotion (tau-b = 0.03; p = 0.24) and pain (tau-b = 0.07; p = 0.37) dimensions were not significantly associated with GMFCS level.

NA

NA

NA

Kulkarni et al. (2004) [40]

Children with hydrocephalus

NA

Strong correlation was found between utility score and the HOQ overall health (r = 0.81), physical health (r = 0.88), social-emotional (r = 0.56) and cognitive (r = 0.57) scores.

NA

NA

NA

Kulkarni et al. (2006) [41]

Children with hydrocephalus

NA

Correlation between utility score and HOQ overall health score was high (r = 0.81), a scatterplot demonstrated a strong linear relationship. Simple and complex linear regression models both accounted for a large proportion of HUI2 variability (adjusted R² = 0.66 and 0.80 respectively).

NA

NA

NA

Kulkarni et al. (2008) [42]

Children with hydrocephalus

NA

Mean utility score (0.58 ± 0.63) close to cHOQ mean scores for overall health (0.65 ± 0.20), Physical health (0.66 ± 0.25), Cognitive health (0.55 ± 0.28) and Social-emotional health (0.71 ± 0.19). Statistical significance not reported.

NA

NA

NA

Kulkarni et al. (2008) [43]

Children with hydrocephalus

NA

Significant correlation was found between the cHOQ overall health score (self-reported) and utility score (proxy-reported) (r = 0.60; p < 0.001).

NA

NA

NA

Landfeldt et al. (2016) [44]

Children and adolescents with Duchenne muscular dystrophy

Ambulatory class was significantly associated with proxy-reported utility scores (p < 0.001), decreasing from early ambulatory class (mean 0.75) to late non-ambulatory class (mean 0.15). Caregiver assessments of health and mental status were significantly associated with utility score (p < 0.001).

NA

NA

NA

NA

Lindquist et al. (2014) [45]

Adults who experienced hydrocephalus in infancy

The study group had significantly (p ≤ 0.004) lower dimension scores compared to the control group in vision, eating, usual activities, mental function dimensions. The most reported problems were associated with the neuroimpairment subgroup (mean utility score 0.87 compared to 0.94 in the no neuroimpairment subgroup). The subgroup without neuroimpairment were not significantly different to the control in terms of mean utility score (0.94 and 0.95 respectively, p value not reported).

NA

NA

NA

NA

Livingston and Rosenbaum (2008) [46]

Adolescents with cerebral palsy

NA

Disattenuated correlation coefficients between utility scores and QOL Instrument scores demonstrated weak to moderate correlation for the being (r = 0.48); belonging (r = 0.35); becoming (r = 0.29) and overall quality of life (r = 0.35) scales. The two measure shared up to 23% variance.

NA

NA

Generalizability coefficients were calculated to assess variability of scores over time. Dimensions with greater stability (i.e. larger G scores) had less variability between individuals over time. The ambulation dimension (G = 0.94) and overall utility (G = 0.91) were found to be highly stable; while the speech (G = 0.87), vision (G = 0.87); dexterity (G = 0.82), cognition (G = 0.81), and hearing (G = 0.72) dimensions were moderately stable. The pain (G = 0.48) and emotion (G = 0.24) dimensions were found to have low stability.

Lopez-Bastida et al. (2017) [47]

Children with spinal muscular atrophy

Children with Type II spinal muscular atrophy were found to have a lower average utility score (− 0.012) than the combined average of children with all types of spinal muscular atrophy (0.158). Statistical significance was not reported.

NA

NA

NA

NA

Morrow et al. (2011) [48]

Children with chronic conditions (relevant condition: cerebral palsy)

NA

NA

NA

Agreement between respondents was assessed using Cohen’s kappa coefficient. Moderate agreement was found between parents of children with cerebral palsy and doctors for the HUI2 dimensions of sensation (63.6% agreement; Kappa 0.41), cognition (70%; Kappa 0.56) and self-care (100%; Kappa 1). Only the HUI3 ambulation dimension (63.6%; Kappa 0.46) demonstrated moderate agreement. All other dimensions exhibited slight or fair agreement

NA

Penner et al. (2013) [49]

Children and adolescents with cerebral palsy

A significant negative correlation was found between the HUI3 pain dimension and GMFCS level (r = 0.36; p < 0.001).

NA

Good correlation was found between child self-reported pain (Wong-Baker scale) and proxy reported pain (HUI3 pain dimension) (Goodman and Kruskall’s y = 0.57; p < 0.001). Using the HUI3 pain dimension, physicians identified pain in 4.4% of cases where parent proxies did not identify pain. In 17.4% of cases, parent proxies identified pain when physicians did not. Statistical significance not reported.

NA

Perez Sousa et al. (2017) [50]

Children and adolescents with cerebral palsy

NA

NA

NA

Child/father agreement was poor for all EQ-5D-Y dimensions (Kappa range 0.016–0.067; non-significant). Child/mother agreement between dimensions was mostly poor (Kappa range 0.057–0.389; non-significant), however for the mobility dimensions agreement was good (Kappa 0.713; p = 0.000) and for the usual activities dimension agreement was moderate (Kappa 0.436; p = 0.000). Mothers and father both tended to report fewer problems (by proxy) than the child.

NA

Petrou and Kupek (2009) [51]

Children with childhood conditions (relevant conditions: microcephaly, cerebral palsy, spinal muscular atrophy, muscular dystrophy, spina bifida)

Microcephaly: Adjusted disutility from perfect health was estimated to be − 0.820 (95% CI − 0.670 to − 0.970). Adjusted disutility from childhood norms was estimated to be − 0.745 (95% CI − 0.598 to − 0.899).

Cerebral palsy: Adjusted disutility from perfect health was estimated to be − 0.726 (95% CI − 0.607 to − 0.846). Adjusted disutility from childhood norms was estimated to be − 0.652 (95% CI − 0.536 to − 0.775).

Muscular dystrophy and spinal muscular atrophy: Adjusted disutility from perfect health was estimated to be − 0.616 (95% CI − 0.471 to − 0.761). Adjusted disutility from childhood norms was estimated to be − 0.541 (95% CI − 0.400 to − 0.690).

Spina bifida: Adjusted disutility from perfect health was estimated to be − 0.552 (95% CI − 0.404 to − 0.701). Adjusted disutility from childhood norms was estimated to be − 0.478 (95% CI − 0.333 to − 0.630).

NA

NA

NA

NA

Rocque et al. (2015) [52]

Children and adolescents with spina bifida

Diagnosis and type of spina bifida had a significant effect on overall utility and certain dimension scores. Overall utility was found to be significantly lower (p < 0.001) in children and adolescents with myelomeningocele compared to individuals with closed dysraphism. The HUI3 ambulation dimension was significantly lower (p < 0.001) in individuals with open myelomeningocele compared to individuals with closed neural tube defects. The HUI3 cognition dimension was significantly lower (p = 0.039) in individuals with open myelomeningocele compared to individuals with closed neural tube defects

Utility and ambulation scores were significantly associated with bowel/bladder continence (p < 0.05). Those with bowel continence had higher average utility scores.

Utility scores were significantly lower for patients with a history of receiving shunt and/or CM-II decompression interventions (p < 0.005). The dimensions of cognition (p = 0.17) and ambulation (p = 0.002) were found to be significantly lower for individuals with a history of shunting, while the dimensions of speech (p = 0.01) and cognition (p = 0.005) were significantly lower for individuals with a history of CM-II decompression. The results remained significant when controlled for age; shunt status accounted for 10.5% of variability in overall utility scores. Shunt revision status weakly correlated with utility score (r = − 0.197; p = 0.048) and dimension scores for vision, speech, ambulation (coefficients not given; p values ranged from 0.01 to 0.05).

NA

NA

NA

NA

Rosenbaum et al. (2007) [53]

Adolescents with cerebral palsy

GMFCS level was found to have a significant impact on utility score (p < 0.01); average utility scores decreased steadily as GMFCS level increased. Post-hoc Bonferoni correction confirmed significant differences in mean overall utility scores between all GMFCS levels (p values not reported), except between levels II and III (p = 0.82). HUI3 dimension levels were also significantly associated with GMFCS level for all dimensions (p < 0.05).

A strong negative correlation was found between utility score and GMFCS level (r = − 0.81; p value not reported)

Adolescents’ ability to self-report using the QOL Instrument was significantly associated (p < 0.01) with parent-reported HUI3 ratings of speech (tau-b = 0.52); cognition (tau-b = 0.50); dexterity (tau-b = 0.35); ambulation (tau-b = 0.31); vision (tau-b = 0.22) and hearing (tau-b = 0.19).

Utility score was significantly (p value not reported) but weakly correlated with scores on the QOL Instrument for Being (r = 0.37), Belonging (r = 0.17), Becoming (r = 0.20), and Overall quality of life (r = 0.28). Utility scores explained between 2.9% (Belonging) and 14% (Being) of variance in QOL Instrument scores.

NA

NA

NA

Sims-Williams et al. (2016) [54]

Children with spina bifida

Moderate correlation was found between child-reported (r = 0.49) and proxy-reported (r = 0.38) utility scores and child-reported VAS - the authors describe this as a ‘poor’ correlation, p values not reported.

NA

Child self-reported and caregiver proxy utility scores were highly correlated (r = 0.85; p value not reported).

NA

Slaman et al. (2015) [55]

Adolescents and young adults with cerebral palsy

NA

NA

NA

NA

No significant difference between the control and intervention groups at the end of the trial (p = 0.42). QALYs gained equated to 0.78 for the control group, and 0.79 for the intervention group; the incremental difference of 0.013 was not significant (p = 0.76)

Tilford et al. (2005) [56]

Children with spina bifida

A Trend test across the case group revealed that lesion level had a significant effect (at p = 0.01 level) on average utility score, with individuals with thoracic lesions scoring the lowest utility scores on average.

NA

NA

NA

NA

Usuba et al. (2014) [57]

Adolescents and adults with cerebral palsy

NA

NA

Average utility scores varied depending on PBM used; HUI3 derived utility scores were lower at baseline (0.29) and 8-year follow-up (0.29) than equivalent AQoL derived utility scores (0.35 and 0.32 respectively); statistical significance not reported.

NA

The ‘older adult’ group were more likely to report utility deterioration than the ‘younger adult’ group (HUI3: Relative risk [RR] = 1.19; 95% CI 0.66–2.15 / AQoL: RR = 3.17; 95% CI 1.12–9.00).

A significant interaction was found between age group and time of survey using the AQoL (p = 0.002); AQoL derived utility improved over time in the ‘younger adult’ group but the opposite occurred in the ‘older adult’ group.

The distribution of PBM dimension scores was stable across the 8-year follow-up, except for the AQoL social relationships, AQoL independent living and HUI3 ambulation dimensions, which all improved in the ‘younger adult’ group but deteriorated in ‘older adult’ group.

Vitale et al. (2001) [58]

Adolescents with orthopaedic problems (relevant condition: cerebral palsy)

There was found to be a significant difference (p > 0.05) between the mean utility scores of children with cerebral palsy (mean 0.92) and children with scoliosis with comorbidities (mean 0.725), but not children with idiopathic scoliosis (mean 0.889). Justification for these groupings was based on sample size, thus there is no explicit explanation as to why utility differences between these groups would be expected.

NA

NA

NA

NA

Wallander et al. (2009) [59]

Adults treated for CTEV in infancy

Male participants had significantly higher average utility score than the comparable norm group (p = 0.027). Male participants reported significantly less moderate/extreme problems with anxiety/depression than the comparable norm group (6.3% vs 21.2%; p = 0.007).

Female participants had worse utility on average than a comparable norm group, but not significantly. Female participants reported significantly more moderate/extreme problems with mobility (45% vs 12.2%; p < 0.001) and usual activities (35% vs 12.5%; p = 0.006) than a comparable norm group.

NA

NA

NA

NA

Young et al. (2010) [22]

Adolescents and young adults with cerebral palsy

In both the ‘youth’ and ‘adult’ groups, utility scores deteriorated steadily as GMFCS level increased. Statistical significance not reported

Linear regression analysis revealed that GMFCS level in childhood was the most important influence on utility scores; responsible for 53.2% of variance in HUI3 utility outcomes (β = − 0.205; p < 0.001) and 45.2% of variance in AQoL utility outcomes (β = − 0.148; p < 0.001).

The SRH was found to be moderately correlated with utility scores derived from the HUI3 (r = 0.41; p < 0.001) and AQoL (r = 0.41; p < 0.001).

Utility scores derived from HUI3 were on average higher than those derived from AQoL across all GMFCS levels; however strong correlation was found between the HUI3 and AQoL (r = 0.87; p < 0.001).

Proxy utility scores were generally lower (by 0.16) when adjusted for cognition, general health and CP severity. Statistical significance not reported.

NA

Young et al. (2013) [21]

Adolescents and young adults with spina bifida

Utility scores derived from both HUI3 and AQoL varied in the same way according to lesion level, with thoracic lesions associated with lowest utility. Linear regression analysis revealed that the most important single factor contributing to utility outcomes was surgical lesion level; responsible for 40% of variance in HUI3 utility scores and 18% in AQoL utility scores. Both lesion level and age were important when combined, accounting for 48% variance in HUI3 utility scores and 22% variance in AQoL utility scores.

The SRH was found to be moderately/strongly correlated with utility scores derived from the HUI3 (r = − 0.45; p < 0.001) and AQoL (r = − 0.58; p < 0.001). The HAQ was also strongly correlated with utility scores derived from the HUI3 (r = 0.79; p < 0.001) and AQoL (r = 0.70; p < 0.001).

Utility scores derived from the AQoL were higher than from the HUI3 across all lesion level sub-groups; however the AQoL and HUI3 exhibited strong correlation (r = 0.73; p < 0.001).

Mean utility scores were slightly higher in the self-reported group (HUI3 mean + 0.04; AQoL mean + 0.03) however this was based on the comparison of youth and adult data and the regression models remained unchanged.

NA