2 files

Statistical model for predicting mortality in cardiac surgery patients operated on under cardiopulmonary bypass

conference contribution
posted on 07.05.2021, 05:36 by Dmitry Simankov, И.В. Мелемука, О.А. Савостьянова, Нина Инзаровна Габриэлян, Владимир Евгеньевич Толпекин
XIV All-Russian Congress of Cardiovascular Surgeons.
November 9-12, 2008.
(XIV Всероссийский съезд сердечно-сосудистых хирургов.
C 9 по 12 ноября 2008 года.)

The goal is to determine among the known risk factors the most significant for predicting the likelihood of death after cardiac surgery performed with the use of cardiopulmonary bypass (CI).

Methods - using a tree of classification according to the CRT growth method with a measure of Twoing admixture, 1,760 (169 of them died, 9.6%) of admitted patients were distributed, in the period from 2004 to 2006 at the Research Institute of TIO, according to belonging to a group with a certain set of risk factors ...

Results - The following preoperative risk factors were identified: age over 60 years (34% of admitted patients and 53% of deaths), bacterial endocarditis (10% and 10%), chronic foci of infections (28% and 33%) and a history of previous cardiac surgeries (11% and 24%). Intraoperative risk factors were considered - duration of cardiopulmonary bypass more than 180 minutes (13% and 48%), blood loss of more than 500 ml (11% and 50%), rethoracotomy (7% and 25%). Risk factors for the early postoperative period included the development of multiple organ failure during the first days after surgery (9% and 63%), duration of ventilation for 2 or more days (10% and 60%), intra-aortic balloon placement (9% and 37%). ). To assess the significance of each risk factor, a classification tree was built and an analysis was performed using the appropriate criteria (dependent variable - death and independent risk factors). Studying the contribution of risk factors to the outcome of the operation using the usual tabular method, it is impossible to answer the question of how each of the factors affects the outcome of the operation, since in each case only a combination of factors determines the lethality.

Using the Tree model in SPSS 13 makes it possible to visualize the available data and specifically answer the problem questions. This study has shown that the most significant predictor of death after cardiac surgery is the early (within the first day) development of multiple organ failure, followed by the duration of artificial lung ventilation - two or more days and the duration of CP for more than 180 minutes. In contrast to the widespread point of view that the duration of cardiopulmonary bypass is the only reliable predictor of mortality, the statistical analysis showed that in the presented sample this factor cannot be used as the only reliable criterion for death, as well as risk factors such as the use of balloon counterpulsation and a history of previous heart surgery.

Conclusion - Using the method of statistical data processing, the SPSS 13 program allows to determine with sufficient reliability predictors of postoperative mortality in patients with cardiac surgery.