作者: Fernando Liaño , Araceli Gallego , Julio Pascual , Florencio García-Martín , José Luis Teruel
DOI: 10.1159/000187139
关键词: Prospective cohort study 、 Population 、 Linear model 、 Medicine 、 Regression analysis 、 Multivariate analysis 、 Acute tubular necrosis 、 Surgery 、 Internal medicine 、 Logistic regression 、 Linear regression
摘要: The ability to predict the outcome in acute tubular necrosis (ATN) remains elusive despite considerable efforts. Accurate prediction is a crucial priority and has large economical ethical implications, mainly judge when treatment futile further efforts only prolong miserable agony. To analyze influence of risk factors prognosis ATN, we applied, an initial phase, prospective protocol demographic data, cause renal failure, diuresis, need dialysis clinical conditions 228 patients using multiple linear logistic regression models. In control phase with 100 consecutive patients, checked accuracy results previously obtained, evaluating overall population 328 synthetic phase. Finally, validation equations obtained was verified 25 from another hospital. As complement this 4-phase study, detailed statistical comparisons between both models were undertaken. Correlation probability death applied real evolution these survival or death, excellent. study revealed coma, assisted respiration, hypotension, oliguria jaundice as having independent positive on mortality nephrotoxic etiology normal consciousness good prognosis. For model, same cut-off point discriminant score (0.9) above which there no chances for could be established 4 phases. With it found at later better than model terms correlation mortality, sensitivity specificity intervals, easier use adjustment distribution standardized residuals expected function. Early ATN possible can given simple features. A allows distinguish without survival. ATN.