We conducted a longitudinal, single-center, retrospective cohort study of adult patients who underwent TTE during an ICU admission at the Beth Israel Deaconess Medical Center. Data were extracted from the Multiparameter Intelligent Monitoring in Intensive Care II database (MIMIC II). MIMIC II is freely available in the public domain and contains information derived from the electronic medical records of 32,425 patients admitted to the ICUs at the Beth Israel Deaconess Medical Center between 2001 and 2008. Twenty-eight-day mortality information was obtained from Social Security Death Index records. The creation and use of the MIMIC II database for research was approved by the institutional review boards of both Beth Israel Deaconess Medical Center and the Massachusetts Institute of Technology (Institutional Review Board protocol 2001-P-001699/3). No further patient consent is required for use of this deidentified public database.
All adult patients in the database were screened. Data regarding age, sex, Sequential Organ Failure Assessment score (SOFA) [9], laboratory values, vital signs, diagnosis-related group (DRG) codes, and International Classification of Diseases-Ninth Revision (ICD-9) diagnoses were extracted. Medical comorbidities were represented by the Elixhauser score [10] for 30 comorbidities as calculated using the DRG and ICD-9 codes from the respective hospital admission. The worst values of common pertinent laboratory results were also extracted, including white blood cell count, lactate, and creatinine. Patients with sepsis were identified using the Angus criteria [11].
Bạn đang xem: Hyperdynamic left ventricular ejection fraction in the intensive care unit
For patients with multiple ICU stays, the first ICU admission was used. Patients in the database were admitted to one of the following: medical ICU (MICU), surgical ICU (SICU), cardiac ICU, or cardiac surgical resuscitation unit. The study was limited to MICU and SICU patients to exclude elective admissions. Patients with at least one TTE were included in the cohort. HDLVEF was defined as an ejection fraction (EF) >70 % based on the ACC guidelines, and NLVEF was defined as an EF of 55-70 % [1]. Those with an EF <55 % were excluded from the analysis to minimize confounding of the relationship between HDLVEF and clinical outcomes. EF was determined predominately by visual estimation using two-dimensional imaging with incorporation of fractional shortening in the parasternal long-axis view according to guidelines established by the American Society of Echocardiography [12]. Contrast echocardiography was used when standard imaging was not diagnostic, and this method has been shown to reliably correlate with quantitative measurements [13]. To ensure the quality of using natural language processing for EF categorization, a random sample of 100 TTE reports was manually reviewed. This showed exceptionally high algorithm accuracy.
Xem thêm : How to Remove Shower Calcium Deposits – San Diego
Baseline comparisons were performed using Fisher’s exact test for categorical variables, where counts and percentages were reported. Continuous variables were compared using a two-sample Wilcoxon rank-sum test, also known as the Mann-Whitney U test, and reported as median with interquartile range (IQR). Statistical significance was defined as p values <0.01 for the baseline characteristics shown in Table 1.
A multivariable logistic regression analysis was performed with 28-day mortality as the outcome. Covariates included age, sex, SOFA score, Elixhauser score of comorbidities, mechanical ventilation use, vasopressor use, and the presence of HDLVEF. Mechanical ventilation was represented as a binary variable. Since vasopressors have wide ranges of therapeutic intensity, sensitivity analyses were performed in the way vasopressor use was adjusted for. The analyses were adjusted for vasopressor use as a binary variable (yes or no), maximum number of vasopressors, or maximum vasopressor dose, defined as a fraction of the highest recommended dose for each vasopressor. The odds ratios (ORs) are preserved in the logistic regression model with 95 % confidence intervals (CIs). Two-sided p values <0.05 were considered statistically significant.
Nguồn: https://buycookiesonline.eu
Danh mục: Info
