In:
European Journal of Cardio-Thoracic Surgery, Oxford University Press (OUP), Vol. 56, No. 6 ( 2019-12-01), p. 1162-1169
Abstract:
OBJECTIVES This study evaluated the various risk factors for chylothorax and persistent serous effusions ( 〉 7 days) after congenital heart surgery and developed equations to calculate the probability of their occurrence. METHODS We performed a retrospective review of different medical databases at the University Hospital of Erlangen between January 2014 and December 2016. Full model regression analysis was used to identify risk factors, and prediction algorithms were set up to calculate probabilities. Discriminative power of the models was checked with the help of C-statistics. RESULTS Of 745 operations on 667 patients, 68 chylothoraxes (9.1%) and 125 persistent pleural effusions (16.8%) were diagnosed. Lowest temperature [P = 0.043; odds ratio (OR) 0.899], trisomy 21 (P = 0.001; OR 5.548), a higher vasoactive inotropic score on the day of surgery (P = 0.001; OR 1.070) and use of an assist device (P = 0.001; OR 5.779) were significantly associated with chylothorax. Risk factors for persistent serous effusions were a given or possible involvement of the aortic arch during the operation (P = 0.000; OR 3.982 and 2.905), univentricular hearts (P = 0.019; OR 2.644), a higher number of previous heart operations (P = 0.014; OR 1.436), a higher vasoactive inotropic score 72 h after surgery (P = 0.019; OR 1.091), a higher central venous pressure directly after surgery (P = 0.046; OR 1.076) and an aortic cross-clamp time 〉 86 min (P = 0.023; OR 2.223), as well as use of an assist device (P = 0.002; OR 10.281). The prediction models for both types of effusions proved to have excellent discriminative power. CONCLUSIONS Persistent serous effusion is associated with a higher vasoactive inotropic score 72 h after surgery, an aortic cross-clamp time 〉 86 min and elevated central venous pressure directly after surgery, which, in combination, potentially indicate cardiac stress. The developed logistic algorithm helps to estimate future likelihood.
Type of Medium:
Online Resource
ISSN:
1010-7940
,
1873-734X
DOI:
10.1093/ejcts/ezz203
Language:
English
Publisher:
Oxford University Press (OUP)
Publication Date:
2019
detail.hit.zdb_id:
1500330-9
