Random Forest Analysis of Out-of-Pocket Health Expenditures Associated with Cardiometabolic Diseases, Lifestyle, Lipid Profile, and Genetic Information in São Paulo, Brazil

Autor(es)

Leite, Jean Michel R. S. ; Trindade, Lucas A. I. ; Pereira, Jaqueline L. ; de Souza, Camila A. ; Soler, Julia M. Pavan ; Mingroni-Netto, Regina C. ; Fisberg, Regina M. ; Rogero, Marcelo M. ; Sarti, Flavia M.

• Tipo de documento: Artigo Científico
• Fonte: HEALTHCARE; v. 12, n. 22, p. 18-pg., 2024-11-01.

Resumo

Background/Objectives: There is a lack of empirical studies of out-of-pocket health expenditures associated with dyslipidemias, which are major cardiovascular risk factors, especially in underrepresented admixed populations. The study investigates associations of health costs with lipid traits, GWAS-derived genetic risk scores (GRSs), and other cardiometabolic risk factors. Methods: Data from the observational cross-sectional 2015 ISA-Nutrition comprised lifestyle, environmental factors, socioeconomic and demographic variables, and biochemical and genetic markers related to the occurrence of cardiometabolic diseases. GWAS-derived genetic risk scores were estimated from SNPs previously associated with lipid traits. There was phenotypic and genetic information available for 490 independent individuals, which was used as inputs for random forests and logistic regression to explain private quantitative and categorical health costs. Results: There were significant correlations between GRSs and their respective lipid phenotypes. The main relevant variables across techniques and outcome variables comprised income per capita, principal components of ancestry, diet quality, global physical activity, inflammatory and lipid markers, and LDL-c GRS and non-HDL-c GRS. The area under the ROC curve (AUC) of quartile-based categorical health expenditure without GRSs was 0.76. GRSs were not significant for this categorical outcome. Conclusions: We present an original contribution to the investigation of determinants of private health expenditures in a highly admixed population, providing insights on associations between genetic and socioeconomic dimensions of health in Brazil. Ancestry information was also among the main factors contributing to health expenses, providing a novel view of the role of genetic ancestry on cardiometabolic risk factors and its potential impact on health costs. (AU)

• Processo FAPESP: 20/15899-2 - Associação entre dislipidemias e dispêndios em saúde em nível populacional: Escores de risco poligênico e modelagem de microssimulação computacional.
• Beneficiário: Jean Michel Rocha Sampaio Leite
• Modalidade de apoio: Bolsas no Brasil - Doutorado
• Processo FAPESP: 12/22113-9 - Indicadores bioquímicos do estado nutricional de micronutrientes e consumo alimentar habitual de residentes do Município de São Paulo - Estudo ISA 2013
• Beneficiário: Regina Mara Fisberg
• Modalidade de apoio: Auxílio à Pesquisa - Regular
• Processo FAPESP: 17/05125-7 - Estilo de vida, marcadores bioquímicos e genéticos como fatores de risco cardiometabólico: inquérito de saúde na cidade de São Paulo
• Beneficiário: Regina Mara Fisberg
• Modalidade de apoio: Auxílio à Pesquisa - Temático
• Processo FAPESP: 22/14123-6 - Além da predição de custos em saúde relacionados a dislipidemias e outros fatores de risco cardiometabolico: análise explicativa por aprendizado de estrutura e algoritmos de inferência causal
• Beneficiário: Jean Michel Rocha Sampaio Leite
• Modalidade de apoio: Bolsas no Exterior - Estágio de Pesquisa - Doutorado
• Processo FAPESP: 19/23985-9 - Investigação das bases moleculares de doenças crônicas multifatoriais na população de São Paulo por meio de GWAS
• Beneficiário: Jaqueline Lopes Pereira França
• Modalidade de apoio: Bolsas no Brasil - Pós-Doutorado