Modelo logístico para identificar fatores associados e predição da retinopatia da prematuridade

• Ano de defesa: 2016
• Autor(a) principal: Pereira, Rafaela
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual de Maringá; Brasil; Departamento de Estatística; Programa de Pós-Graduação em Bioestatística; UEM; Maringá, PR; Centro de Ciências Exatas
• Palavras-chave em Português: Modelo logístico; Retinopatia da prematuridade; Bioestatística - Brasil; Ciências Exatas e da Terra; Estatística
• Link de acesso: http://repositorio.uem.br:8080/jspui/handle/1/4365

Retinopathy of prematurity is a multifactorial disease that affects the retina of premature infants. It is a major cause of blindness in children in the world. The objective was to set a logistic model to identify associated factors and to assist in predicting the diagnosis of retinopathy of prematurity. We conducted a retrospective cross-sectional study of admitted newborns in the program Mãe Canguru of the University Hospital of Maringá-PR from January 2006 to November 2015. The response variable was the diagnosis of retinopathy classified as absent or present in the newborn. Univariate analysis with the predictor variables and the categorization made by the likelihood ratio test was performed. For the multiple logistic model was verified the quality of the adjustment by Hosmer and Lemeshow test, deviance and Pearson statistics and to assess the assumptions of the model was carried out the residuals analysis and diagnosis.The ROC curve was constructed to evaluate the predictive ability of the model. In the study included 324 newborns admitted in the program Mãe Canguru, 33 (10.18%) children had retinopathy. The variables that were presented as risk factors for retinopathy multiple logistic regression were gestational age 30 weeks (OR = 3.6), birth weight 1250 grams (OR = 5.14) and Apgar score at 1 minute <7 (OR = 2.57). The Hosmer and Lemeshow test and Pearson's statistics showed evidence that the model is well fitted to the data, since the deviance of the degrees of freedom pointed underdispersion evidence. In the analysis diagnosis through observation graph of predicted vs. deviance waste variance function proved inadequate. The model discrimination ability given by the area under the ROC curve was 0.8395, and the probability of the point diagnosing infants with retinopathy and newborns without retinopathy was 0.078 with sensitivity of 0.75 and specificity of 0.79. In conclusion, the model can be used to assist in the prediction of retinopathy of prematurity predict since it presents considerable precision. Newborns that are experiencing more likely than 0.078 by logistic model require greater attention from retina specialist and neonatologists during the screening programs for the prevention of blindness caused by retinopathy