LINEAR MIXED EFFECT MODELLING OF FACTORS ASSOCIATED WITH ADOLESCENTS QUALITY OF LIFE

Abstract

The world currently consists of 1.2 billion adolescents, which makes 18% of the world's population. Psychosocial or behavioural problem has been attributed to around 20 percent of the world's adolescents. Adolescents have been seen to be affected by depression which is the largest contributor to the global burden of disease, with suicide being one of the three leading causes of mortality in them. This study was conducted to investigate the effect of psychosocial functioning and socio-demographic factors on the quality of life of Adolescents using a linear mixed effect model. A secondary data which was gotten from a cross sectional study conducted among adolescents in Benue state Nigeria was used in this study. The adapted WHO-Quality of Life was administered to 2,095 Secondary school adolescents between the ages of 10-19years to assess their psychosocial functioning and Quality of life of whom 63% were within the ages of 13-17 years. The Adolescents psychosocial functioning inventory (APFI) was used to assess the adolescents' psychosocial functions. Chi square and multinomial logistic regression were used to assess the effect of selected socio demographics on the QOL. Also, linear regression was used to assess the effect of psychosocial functions on the QOL. A linear mixed effects model was used to model the effect of the psychosocial functions on the QOL considering the selected socio demographics, using different methods of estimation and different covariance structures. A complete data of 1963 adolescents was extracted. The participants were 14.71±2.05 years with males being slightly older (14.82±2.19) than females (14.58± 1.86). Majority of the adolescents came from monogamous families (67.9%) with (74.8%) of their parents staying together. The study findings showed that family status (X2 =5.45, P= 0.02) had an effect on the QOL, also, mother's highest level of education had a significant effects on the QOL of adolescents (OR=0.63, 95%C.l=0.40-0.98). Adolescents whose mothers have no formal education were 37% times less likely to have a Poor QOL relative to Adolescents whose mothers have some kind of formal education using High QOL as the reference. Akaike's information criterion was used to select the best fit linear mixed effect model. The best fit model used the restricted maximum method of estimation with the variance component covariance structure. In this model Optimism Coping Strategy (OCS) and General Psychosocial Dysfunctions (GDP) were significant (p=0.01, p= <0.001, AIC= 15386.898) respectively while the Behavior and relationship problems (BRP) was not significant. The results suggested that only OCS, GDP, family status and mother's highest level of education have significantly effects on the QOL of adolescents and that restricted maximum likelihood with variance components provided the best model.