Predictive Models of Weight and Height for the Evaluation of the Growth of Children and Adolescents With Cerebral Palsy

Overview

Cerebral palsy (CP) is the most frequent disability in children. The vast majority of these patients are malnourished. In this population, there are practical difficulties to perform a nutritional and growth assessment which makes it difficult to treat and follow up, because of the lack of reference growth in Argentina, and the difficulty in taking anthropometric measurements of weight and height because of their motor compromise, posture and muscle tone. The main objective is to design and validate predictive models for the nutritional and growth assessment of children and adolescents with CP and instruments for estimating weight and height from body segments, in order to improve care, quality of life of these patients to promote their social inclusion. Material and method: It will be an observational, descriptive and cross-sectional study. There will be two parts of the study, in the first part the population will be healthy children from 2 to 18 years old from Cordoba, Argentina. The sample size was calculated based on growth WHO standards data, for α=0.05 and 1-β=0.80, creating an stratified sampling divided in 16 age groups for each age. This first part will help to establish which body segments to use. In the second part, the population will be children and adolescents from 2 to 18 years old with diagnosis of CP from Córdoba, Argentina. A stratified sequential sampling shall be performed. The sample size will be 192 patients, 12 per age stratum. The variables studied will be: weight, height, body segments, sex, age, CP type, feeding path and type of feeding. For the analysis of the data the normal continuous variables will be described in means with their respective standard deviations and those of non-normal distribution in medians with their ranges. For the development of the predictive equations using body segments measures, a generalizable linear regression model will be used. The correlation coefficient r, determination R2 and test of F will be calculated with p \<0.05. To generate predictive growth models, the percentiles from 3 to 97 will be calculated, using the LMS method and a q-q graph.

The aim is to design predictive models of weight and height, through predictive equations using segmental measurements and to develop specific growth patterns, for the nutritional and growth evaluation of children and adolescents with cerebral palsy from different provinces of Argentina. This research is a cross-sectional study, which consisted of two stages. First, data were collected from children and adolescents aged 2 to 18 years with typical development from which the anthropometric variables associated with weight and height were determined and then used for the analysis of the population with CP. A stratified sampling was made with 17 strata according to age for each sex, pre-establishing a minimum of 20 children for each stratum. Data were collected transversely in two hospitals and two schools in the City of Córdoba until all strata were completed. In a second stage, data is collected from children and adolescents diagnosed with CP between the ages of 2 and 19 years. Data were included from 17 rehabilitation centers and therapeutic educational centers in 5 provinces of the country (Córdoba, Buenos Aires (CABA), Jujuy, Santiago del Estero and Catamarca). A sequential non-probabilistic sampling was made, including all possible cases of each participating institution. The variables studied were demographic variables, anthropometric measures, nutritional status, type of malnutrition and in the case of the population with PC also variables such as data on access to health, diagnosis, co-morbidities and feeding were included. Normal continuous variables were described as mean and standard deviations, while those of non-normal distribution in medians with their interquartile ranges. Variables are compared by means of t-test or Mann-Whitney as appropriate. Categorical data are expressed in percentages with a 95% confidence interval \[95% CI\]. To analyze association between variables, Chi square of Mantel-Haentzel was calculated, and OR with its 95% CI. For the development of predictive equations, a generalizable linear regression model will be used. The correlation coefficient r, determination R2 and F test were calculated with a p \<0.05. Growth patterns were made using the Generalized Additive Models for Location Scale and Shape (GAMLSS) method. Nutritional status was assessed using the WHO Anthro Plus V1.0.4 program using the WHO standards (2007). The approval of the local ethics committee was obtained and written informed consent was obtained from the participants. The study of nutrition in patients with PC is an emerging field, for which this work plan seeks to develop methodologies that promote social inclusion by making substantial and necessary contributions to improve the treatment of this pathology.