ABSTRACT
OBJETIVO: Comparar empiricamente as regressões de Cox, log-binomial, Poisson e logística para estimar razões de prevalência em estudos de corte transversal. MÉTODOS: Foram utilizados dados de um estudo epidemiológico transversal (n=2.072), de base populacional, realizado com idosos na cidade de São Paulo (SP), entre maio de 2003 e abril de 2005. Diagnósticos de demência, possíveis casos de transtorno mental comum e autopercepção de saúde ruim foram escolhidos como desfechos com prevalência baixa, intermediária e alta, respectivamente. Foram utilizadas variáveis de confusão com duas ou mais categorias ou valores contínuos. Valores de referência para estimativas por ponto e por intervalo para as razões de prevalência (RP) foram obtidos pelo método de estratificação de Mantel-Haenszel. Estimativas ajustadas foram calculadas utilizando regressões de Cox e Poisson com variância robusta, e regressão log-binomial. Odds ratios (OR) brutos e ajustados foram obtidos pela regressão logística. RESULTADOS: As estimativas por ponto e por intervalo obtidas pelas regressões de Cox e Poisson foram semelhantes à obtida pela estratificação de Mantel-Haenszel, independentemente da prevalência do desfecho e das covariáveis do modelo. O modelo log-binomial apresentou dificuldade de convergência quando o desfecho tinha prevalência alta e havia covariável contínua no modelo. A regressão logística produziu estimativas por ponto e por intervalo maiores do que as obtidas pelos outros métodos, principalmente para os desfechos com maiores prevalências iniciais. Se interpretados como estimativas de RP, os OR superestimariam as associações para os desfechos com prevalência inicial baixa, intermediária e alta em 13 por cento, quase 100 por cento e quatro vezes mais, respectivamente. CONCLUSÕES: Em análise de dados de estudos de corte transversal, os modelos de Cox e Poisson com variância robusta são melhores alternativas que a regressão logística. O modelo de regressão log-binomial produz e...
Subject(s)
Cross-Sectional Studies , Logistic Models , Prevalence , Statistics as TopicABSTRACT
OBJECTIVE: To empirically compare the Cox, log-binomial, Poisson and logistic regressions to obtain estimates of prevalence ratios (PR) in cross-sectional studies. METHODS: Data from a population-based cross-sectional epidemiological study (n = 2072) on elderly people in Sao Paulo (Southeastern Brazil), conducted between May 2003 and April 2005, were used. Diagnoses of dementia, possible cases of common mental disorders and self-rated poor health were chosen as outcomes with low, intermediate and high prevalence, respectively. Confounding variables with two or more categories or continuous values were used. Reference values for point and interval estimates of prevalence ratio (PR) were obtained by means of the Mantel-Haenszel stratification method. Adjusted PR estimates were calculated using Cox and Poisson regressions with robust variance, and using log-binomial regression. Crude and adjusted odds ratios (ORs) were obtained using logistic regression. RESULTS: The point and interval estimates obtained using Cox and Poisson regressions were very similar to those obtained using Mantel-Haenszel stratification, independent of the outcome prevalence and the covariates in the model. The log-binomial model presented convergence difficulties when the outcome had high prevalence and there was a continuous covariate in the model. Logistic regression produced point and interval estimates that were higher than those obtained using the other methods, particularly when for outcomes with high initial prevalence. If interpreted as PR estimates, the ORs would overestimate the associations for outcomes with low, intermediate and high prevalence by 13%, almost by 100% and fourfold, respectively. CONCLUSIONS: In analyses of data from cross-sectional studies, the Cox and Poisson models with robust variance are better alternatives than logistic regression is. The log-binomial regression model produces unbiased PR estimates, but may present convergence difficulties when the outcome is very prevalent and the confounding variable is continuous.
Subject(s)
Epidemiologic Methods , Mental Disorders/epidemiology , Aged , Aged, 80 and over , Brazil/epidemiology , Dementia/epidemiology , Female , Humans , MaleABSTRACT
BACKGROUND: The aim of this study was to determine the prevalence of dementia in a socioeconomically disadvantaged population of older adults living in the city of São Paulo, Brazil. METHODS: A cross-sectional one-phase population-based study was carried out among all residents aged > or = 65 in defined census sectors of an economically disadvantaged area of São Paulo. Identification of cases of dementia followed the protocol developed by the 10/66 Dementia Research Group. RESULTS: Of 2072 individuals in the study, 105 met the criteria for a diagnosis of dementia, yielding a prevalence of 5.1%. Prevalence increased with age for both men and women after age 75 years, but was stable from 65 to 74 years. Low education and income were associated with increased risk of dementia. CONCLUSIONS: The prevalence of dementia among older adults from low socioeconomic backgrounds is high. This may be partly due to adverse socioeconomic conditions and consequent failure to compress morbidity into the latter stages of life. The increasing survival of poorer older adults with dementia living in developing countries may lead to a rapid increase in the prevalence of dementia worldwide.
Subject(s)
Dementia/epidemiology , Aged , Brazil/epidemiology , Catchment Area, Health , Female , Humans , Male , Prevalence , Socioeconomic FactorsABSTRACT
BACKGROUND: Little is known about the incidence of first-episode psychosis in urban centres of low- or middle-income countries. AIMS: To estimate the incidence of psychosis in São Paulo, a large metropolis of Brazil. METHOD: Prospective survey of first-episode psychosis among residents aged 18-64 years resident in a defined area of São Paulo, over a 30-month period (July 20 2002-December 2004). Assessments were carried out with the SCID-I, and diagnoses given according to DSM-IV criteria. Population at risk was drawn from the 2000 Census data. RESULTS: There were 367 first-episode cases identified (51% women), and almost 40% fulfilled criteria for schizophrenia or schizophreniform disorder. The incidence rate for any psychosis was 15.8/100 000 person-years at risk (95% CI 14.3-17.6). Incidence of non-affective psychoses was higher among younger males. CONCLUSIONS: Incidence of psychosis in São Paulo was lower than expected for a large metropolis.
