RESUMO
Logistic regression is commonly used to test for treatment effects in observational studies. If the distribution of a continuous covariate differs between treated and control populations, logistic regression yields an invalid hypothesis test even in an uncounfounded study if the link is not logistic. This flaw is not corrected by the commonly used technique of discretizing the covariate into intervals. A valid test can be obtained by discretization followed by regression adjustment within each interval.
Assuntos
Ensaios Clínicos como Assunto/métodos , Modelos Logísticos , Modelos Estatísticos , Resultado do Tratamento , Algoritmos , Análise de Variância , Simulação por Computador , Projetos de Pesquisa Epidemiológica , Humanos , Reprodutibilidade dos Testes , Distribuições EstatísticasRESUMO
Sixteen of 227 patients referred for percutaneous placement of a ureteral stent had impassable stenoses. Stenoses were benign (n = 8) or attributed to malignant retroperitoneal neoplasm (n = 8). Electrocautery was used to create a neotract between the stenosed ureter and the bladder or ileal loop. A double-J stent was placed after dilation of the tract by use of angioplasty. Neotracts were established and stents were placed in all patients. Complications (digestive tract fistulas) developed in two patients. This technique is safe if the electrode is placed close to the bladder or ileal loop. The procedure can be used as an alternative to surgery or permanent nephrostomy or in initial treatment of benign anastomotic stenosis.
