Development and validation of segmental length equations to estimate height among cerebral palsy children in southern and central regions of Peninsular Malaysia

Cerebral Palsy (CP) involves a group of chronic movement and posture disorders caused by a non-progressive lesion in an immature brain. 70-80% of CP have spastic clinical features such as increased deep tendon reflexes, tremors, and muscular hypertonia which causes stiffness and walking difficulty. It is the most common cause of physical disability in children, with a worldwide incidence of 2.11 per 1000 live births. Children with cerebral palsy (CP) typically suffer from congenital deformities such as scoliosis and contractures. Therefore, it is a challenge to measure the actual height or stature of CP children in a standing position. Studies have proposed that predictive equations based on segmental length (SL), i.e. knee height (KH), tibia length (TL), and upper arm length (UAL) can be used as an alternative method to measure the actual height. The standing height and recumbent length (RL) has become the gold standard for the actual height measurement. However, the current predictive equation has not been tested among the Malaysian population. Previous studies developed the predictive equations in a certain population, different ages, and did not include all Gross Motor Function System Classification (GMFCS) levels of CP. This study aimed to develop and validate the predictive equations based on KH, TL, and UAL for CP children in outpatient pediatric clinics and community rehabilitation centres (CBR) in the Central and Southern Regions of Malaysia. This cross-sectional study consisted of two phases in which Phase 1 is the evaluation of the existing fourteen equations and the development of new equations. Phase 2 was the validation of the new equations established in Phase 1. The study was conducted from August 2018 to December 2019. The subjects of this study were CP children with GMFCS I – V, aged 2 to 18 years.CP children with acute diarrhea, prolong vomited, and other medical illnesses or disabilities such as down syndrome will be excluded. Subjects were categorized into Phase 1 Equation Development Group (EDG) and Phase 2 Cross Validation Group (CVG). For the EDG, a total of 177 subjects were recruited from six outpatient paediatric clinics. 23.7% were GMFCS level I, 14.1% GMFCS level II, 5.7% GMFCS level III, 5.7% level IV, and 50.8% level V. As for the CVG, 139 subjects were recruited from 18 community-based rehabilitation centres (CBR). The CVG subjects were 15% GMFCS Level I, 23% level II, 14% level III,14% level IV and 73% level V. All the clinics and CBR were sampled from the central and southern regions of Malaysia. The independent variables (IV) in this study were KH, TL, UAL, and age while the dependent variables (DV) in this model were actual height and recumbent length. The standing height was determined using a stadiometer. If CP children were unable to stand, recumbent length was taken using the standard procedure. A flexible Seca measuring tape was used to measure the recumbent and segmental lengths of all participants. In phase 1, 14 existing equations that have been developed from previous studies from the year 1994 to 2003 were tested. The equations have been selected based on knee height, tibia length, and upper arm length. The population of the previous studies was from CP and healthy children in the United States, Australia, China, and Japan. In Phase 1, based on the Bland Altman test, all fourteen existing equations (Eq1-Eq14) showed a weak agreement with the actual height of the CP subjects. Hence, six new predictive equation models were developed using multiple linear regression. This regression model was developed using the segmental length KH, TL, and UAL which were in Model 1 KH, Model 3TL, and Model 5 UAL. Equations that used age as a covariate factor for each segmental length in Model 2 KHA, Model 4 TLA, and Model 6 UALA were also developed. Others covariates such as age and gender not included in the model as the demography data shows not significant with the height measurements among CP population. The new KH-based equations included Model 1KH: Y1 = 22.54+2.679 KH and Model 2KHA: Y2 = 31.5 + 2.11 KH + 1.228A. The new equation model based on TL were Model 3TL: Y3= 32.18+3.139 TL and Model 4TLA: Y4 = 39.905 + 2.417 TL+ 1.31 A. Lastly, Model 5UAL: Y5 = 20.469 +3.83 UAL and Model 6UALA: Y6 = 33.15 + 2.771 UAL + 1.55A were based on UAL. In Phase 2, all six new models were cross-validated to validate the predictive equations. Model 3TL showed the lowest standard error (SEM), i.e. 1.42 compared to other models. Intra-correlation coefficient (ICC) analysis demonstrated a strong agreement between the actual and estimated heights for all models except for Model 1KH and Model 2KH (0.58). Apart from that, Model 3TL: Y3= 32.18+3.139 TL demonstrated a strong correlation with the actual height (R2 = 0.834) and showed a small SEM (1.42) and high ICC (0.929). In conclusion, six new equations have been developed in Phase 1. All new models have been validated in Phase 2 to ensure the accuracy and reliability of the model in the CP population. All new equations were valid to use in CP Malaysia population, however, validation results showed that the Model 3TL was the most suitable segmental length prediction equation to be used in estimating the height of CP children aged two to 18 years. Further testing of the new equation for CP subjects in all regions of Malaysia is needed so that its use can be generalised to all the paediatric CP subjects in the Malaysian population.

Text MAHNON SURIA BINTI MOKHY - IR.pdf Download (1MB)

Actions (login required)

View Item