Predictive Value Of GWTG-HF Risk Score For All-cause Mortality In Cardiac Intensive Care Unit Patients

Background:The cardiac intensive care unit(CICU)population is a complex population with high mortality,requiring immediate medical attention and urgent therapy.The risk stratification of CICU patients is a pivotal medical task,aimed to improve the outcome of severely illedpatients.The epidemiology of modern CICU population is no longer composed of only patients with acute coronary syndromes,and evolving to include more elderly patientswith heart failure and multiple comorbidities.However,optimal methods of mortality risk stratification in CICU patients remained uncertain.We aimed to determine whether the admission GWTG-HF score was useful for predicting CICU and hospital mortality in unselected cohort of CICU patients as well as across different primary diagnoses,to assess its discriminative performance compared with other three established risk scores applied in the intensive care units,therefore to explore more effective risk reclassification tool among CICU patients.Methods:We retrospectively analyzed patients admitted to CICU between 2008 and 2019 from the Medical Information Mart for Intensive Care(MIMIC-Ⅳ)database.The inclusion criteria were first ICU admission of patients 18 years and older withavailable documentation of baseline information related to GWTG-HF risk score.All risk scores were calculated at admission.The primary outcome was CICU all-cause mortality,and the second outcomeswashospital all-cause mortality.Logistic regression models were used to determine the association between risk scores andtheoutcomes.The area under the receiver operating characteristic curve(AUROC)values were determined for hospital mortality for each risk score,and compared using the DeLong test.Discriminative performance comparisonswere evaluated by ROC analyses,net reclassification improvement(NRI)and decision curve analyses(DCA).Sensitive analyses were conducted according to different diagnostic subgroups and the Asian subgroup.Results:A total of 14,468 patients admitted in CICU(mean age,67.8 years;male,63.7%).Acute coronary syndrome,heart failure and atrialfibrillation diagnoseswere present in 44.9%,32.4%and 36.0%of the population,respectively.Heart failure accounted for 39.7%patients withacute coronary syndrome.CICU mortality occurred in 714(4.9%)patients and hospital mortality occurred in 865(6.0%)patients.Higher CICU and hospital all-cause mortalityin either unselected CICU cohort or subgroups was significantly associated with higher GWTG-HF risk score and three ICU risk scores.Discrimination of the GWTG-HF risk score performed well for CICU mortality(AUC,0.780;95%CI,0.761-0.798),but inferior to all three ICU riskscores(DeLong test,p<0.001),among which SOFA performed the best(AUC,0.843;95%CI,0.828-0.858)with NRI of 10.4 to the GWTG-HF risk score.For hospital mortality,GWTG-HF risk score(AUC,0.786;95%CI,0.769-0.803)performed similarly with SOFA and OASIS,but was inferior to SAPS Ⅱ score according to DeLong test though NRI had shown no significant improvement.In subgroup analyses,AUC values of GWTG-HF score were 0.773(Hosmer-Lemeshow test,p=0.206)and 0.761(Hosmer-Lemeshow test,p=0.001)for CICU mortality among patients with and without heart failure diagnoses respectively.In patients with acute coronary syndrome,AUC of GWTG-HF score was 0.800(95%CI,0.776-0.824)with good calibration.Inpatients with and with out atrial fibrillation,AUC of GWTG-HF score were 0.775(95%CI,0.745-0.806)and 0.785(95%CI,0.761-0.808),respectively.SOFA score was superior to all of other scores for CICU mortality on discrimination performance(AUC,0.81-0.86).When applied to the Asian population,GWTG-HF performed also excellent(CICU mortality:AUC=0.796;hospital mortality:AUC=0.825)and showed no significant difference to the ICU risk scores.Conclusions:The GWTG-HF risk score and ICU risk prediction scores demonstrate good discrimination for CICU and hospital mortality in a heterogeneous cohort of CICU patients.Discrimination of ICU risk scores performed superior to GWTG-HF risk score for CICU mortality in either unselected CICU cohort or diagnostic subgroups,among which SOFA score performed the best.When applied for hospital mortality,GWTG-HF risk score showed no significant difference to the ICU risk scores among patients with heart failure,acute coronary syndromes or atrial fibrillation diagnoses.In Asian prediction,the GWTG-HF model performed similarly to the ICU risk scores.