Venema, N.J. (2021) Early cardiac tamponade following cardiothoracic surgery: hemodynamic parameters are the best predictors. thesis, Medicine.
Full text available on request.Abstract
Background: Early cardiac tamponade is a relatively common complication of cardiothoracic surgery at the intensive care unit (ICU). It is associated with increased morbidity and is often accompanied with hemodynamic instability. No reliable non-invasive diagnostic tools are available. A predictive statistical model could provide the solution. This study will asses if a model with hemodynamic parameters has the best predictive value for the occurrence of early�tamponade at the ICU. Methods: A single-center retrospective database study of patients admitted to the ICU after cardiothoracic surgery. Patients were grouped by the occurrence of “cardiac compromise by pericardial collection” (CCPC), defined as hemodynamic instability associated with bleeding or early-tamponade between 2 and 48 hours after ICU admittance. Models with different combinations of hemodynamic and non-hemodynamic parameters were developed for patients with a pulmonary artery catheter (PAC) and patients without a PAC. All parameter data was sampled before the first 2 hours of ICU stay. Validation was performed by comparing c-indexes of bootstrapped models. Results: Included were 3.282 cases whereof 89 (3.6%) with a confirmed diagnosis of CCPC. Models including hemodynamic features were superior to models excluding hemodynamic features, for patients with a PAC (c-index 0.78 [0.75-0.80] vs 0.70 [0.67-0.72]; p<0.001) and for patients without a PAC (c-index 0.80 [0.79-0.81] vs 0.74 [0.72-0.75]; p<0.001). Calibrated models for patients with a PAC (c-index 0.775, sensitivity 65.5%, specificity 75.0%) and without a PAC (c-index 0.803, sensitivity 75.0%, specificity 68.9%) were highly dependent on hemodynamic parameters. Conclusion: Hemodynamic features are the best predictors for CCPC. Models lack the accuracy needed for a diagnostic tool, but can alert clinicians for an impending CCPC event.
| Item Type: | Thesis (UNSPECIFIED) |
|---|---|
| Supervisor name: | Boerma, E.C. |
| Faculty: | Medical Sciences |
| Date Deposited: | 06 Jan 2022 11:32 |
| Last Modified: | 06 Jan 2022 11:32 |
| URI: | https://umcg.studenttheses.ub.rug.nl/id/eprint/2974 |
Actions (login required)
 |
View Item |