СТАТИСТИЧЕСКИЕ МЕТОДЫ ДЛЯ АНАЛИЗА ЗНАЧИМОСТИ ФАКТОРОВ РИСКА.pdf (233.38 kB)
DOCUMENT
СТАТИСТИЧЕСКИЕ МЕТОДЫ ДЛЯ АНАЛИЗА ЗНАЧИМОСТИ ФАКТОРОВ РИСКА (общ).pdf (3.16 MB)
1/0
2 files
STATISTICAL METHODS FOR ANALYSIS OF THE SIGNIFICANCE OF RISK FACTORS OF POSTOPERATIVE MORTALITY IN CARDIAC SURGERY PATIENTS UNDER ARTIFICIAL CIRCULATION
Bulletin of Transplantology and Artificial Organs, Volume 14. Materials of the 6th All-Russian Congress of Transplantology, 2012. pp. 234-235. ISSN 1995-1191
(Вестник трансплантологии и искусственных органов, Том 14. Материалы 6 всероссийского съезда трансплантологов, 2012. Стр 234-235. ISSN 1995-1191)
Purpose.
Determine the most significant risk factors for predicting the likelihood of death after cardiac surgery performed with the use of artificial circulation (CI).
Material and methods.
Age over 60 years, bacterial endocarditis, chronic foci of infections and a history of previous cardiac surgery. Intraoperative risk factors were considered the duration of cardiopulmonary bypass more than 180 minutes, blood loss of more than 500 ml, and rethoracotomy. Factors the risk of the early postoperative period included the development of multiple organ failure during the first days after surgery, the duration of ventilation during 2 or more days. The critical value of the level of statistical significance when testing null hypotheses was taken equal to 0.05. If the achieved significance level was exceeded, the null hypothesis was accepted. To compare the central parameters, parametric and nonparametric methods were used: analysis of variance with the criterion Kruskal – Wallis and Wilcoxon rank marks, median test and Wang test der Waerden. The relationship between pairs of discrete features was investigated using the φ coefficient, the contingency coefficient and the Cramer V-coefficient. To analyze the relationship between one qualitative trait acting as a dependent, the resulting indicator, and a subset of quantitative and qualitative features, a logistic regression model with a stepwise algorithm for the inclusion and exclusion of predictors was used. The ranking of the predictors was carried out according to the modulus of the standardized regression coefficients.
Results.
The results of evaluating the logistic regression equations are represented by a set of regression coefficients, the achieved significance levels for each coefficient, as well as
assessment of the indicator of agreement of the patient's actual belonging to one or another of the groups and theoretical affiliation obtained from the logit regression equation. Several tens of logit regression equations were obtained, and during the selection, equations with values equal to more than 80-90% for the probability of death. It is shown that the most a significant predictor of death is the early (within the first days after surgery) development of multiple organ failure, followed by the duration of artificial
ventilation of the lungs - two or more days and the duration of the IC.
Conclusion.
As a result of using various options for statistical processing of clinical and laboratory data, reliable predictors of postoperative mortality in patients with cardiac surgery were established, and logistic regression equations were drawn up with a high value of agreement of the studied parameters.