LAUSR

BACKGROUND Cardiac arrest in intensive care is a rarely studied type of in-hospital cardiac arrest. OBJECTIVE This study examines the incidence, characteristics, risk factors for mortality as well as long-term prognosis following cardiac arrest in intensive care. DESIGN Retrospective cohort study. SETTING Five noncardiac surgical ICUs (41 surgical and 37 medical beds) at a German university hospital between 2016 and 2019. PATIENTS Adults experiencing cardiac arrest defined as the need for chest compressions and/or defibrillation occurring for the first time on the ICU. MAIN OUTCOME MEASURES Primary endpoint: occurrence of cardiac-arrest in the ICU. Secondary endpoints: diagnostic and therapeutic measures; risk factors and marginal probabilities of no-return of spontaneous circulation; rates of return of spontaneous circulation, hospital discharge, 1-year-survival and 1-year-neurological outcome. RESULTS A total of 114 cardiac arrests were observed out of 14 264 ICU admissions; incidence 0.8%; 95% confidence interval (CI) 0.7 to 1.0; 45.6% received at least one additional diagnostic test, such as blood gas analysis (36%), echocardiography (19.3%) or chest x-ray (9.9%) with a resulting change in therapy in 52%, (more frequently in those with a return of spontaneous circulation vs none, P = 0.023). Risk factors for no-return of spontaneous circulation were cardiac comorbidities (OR 5.4; 95% CI, 1.4 to 20.7) and continuous renal replacement therapy (OR 5.9; 95% CI, 1.7 to 20.8). Bicarbonate levels greater than 21 mmol 1 were associated with a higher mortality risk in combination either with cardiac comorbid-ities (bicarbonate <21 mmol I−1: 13%; 21 to 26 mmolI−1 45%; >26mmolI−1:42%)orwithaSOFA at least 2 (bicarbonate <21 mmolI−1 8%; 21 to 26 mmolI−1: 40%; >26mmolI−1: 37%). "In-hospital mortality was 78.1% (n = 89); 1-year-survival-rate was 10.5% (95% CI, 5.5 to 17.7) and survival with a good neurological outcome was 6.1% (95% CI, 2.5 to 12.2). CONCLUSION Cardiac arrest in ICU is a rare complication with a high mortality and low rate of good neurological outcome. The development of a structured approach to resuscitation should include all available resources of an ICU and adequately consider the complete diagnostic and therapeutic spectra as our results indicate that these are still underused. The development of prediction models of death should take into account cardiac and hepatic comorbidities, continuous renal replacement therapy, SOFA at least 2 before cardiac arrest and bicarbonate level. Further research should concentrate on identifying early predictors and on the prevention of cardiac arrest in ICU.