Comparison of the prognostic performance of cardiogenic shock scores in a low-middle income country

Abstract

Abstract Introduction Mortality in cardiogenic shock (CS) is up to 40% of the patients; as such, risk scores were proposed to stratify and assess the mortality in CS, although they may have an inconsistent performance. Purpose To compare readily available clinical prognostic scores and describe their performance in the first 30 days, and describe the clinical characteristics and management in a real-world low-middle income country. Methods We included 872 patients with cardiogenic shock (systolic blood pressure ≤90 mmHg, need of vasopressor, cardiac index ≤2.2 L/min/m2, or blood lactate ≥2 mmol/l). SCAI (2022 definition), CARDSHOCK, IABP-Shock II, CSS, ALC, GWTG-HF, and ADHERE scores were calculated [1–7]. Cox regression was applied to construct AUC statistic against time; using the IPCW method at 24 h, 72 h, 7, and 30 days globally for in-hospital mortality prediction. AUC differences were computed by Uno's method. This method was applied to assess differences between CS-AMI and non-AMI-CS. Decision curve analysis considering time as a factor was created. Results The cohort was principally constituted by AMI-CS (75.57%), men (70.76%), had a presence of diabetes (DM) in 44.8%, hypertension in 50.3%, chronic HF in 26.6%, with mortality of 67.43%. In AMI-CS, patients: were older; had a more prevalence in men (76.6 vs. 52.28%); had more comorbidities such as DM, hypertension, but less chronic HF history; had higher blood pressure and lower heart and respiratory rates at admission with similar LVEF. At the laboratory data, AMI-CS had higher hemoglobin levels, leukocytes, platelets, glucose, Na, Cl, liver enzymes, and higher eGFR. In contrast, lower levels of Cr and lactate were seen. Higher utilization of dobutamine, levosimendan and number of vasoactives at the shock management level was seen compared to non-AMI-CS. IABP was used more frequently in AMI-CS (48.56%). Mortality was not statistically different between these groups. Regarding cardiogenic shock scores: SCAI had more proportion in stage C with patients in the non-AMI-CS and more D stages in AMI-CS. GWTG-HF, CARDSHOCK, and IABPSHOCK II scores were higher in AMI-CS. Furthermore, ALC, ADHERE, and CSS were higher in non-AMI-CS. When analyzed by the progression of time, significant differences arise in the AUC, suggesting a time-sensitive prediction in mortality in our cohort. C-statistic showed that the highest AUC is for CARDSHOCK score 0.658 (0.633–0.682) followed by SCAI 0.622 (0.599–0.645) (Fig. 1A, 2A). The AMI-CS-related the higher AUC was for CARDSHOCK score (0.671, 0.643–0.699; p<0.001) (Fig. 1B, 2B). In non-AMI-CS, SCAI was the best 0.642 (0.506–0.778, p<0.001) (Fig. 1C, 2C). Conclusion The clinical scores also show a time-sensitive AUC, suggesting that performance could be influenced by time and also the type of CS. Understanding the time influence in the scores could provide a better prognostication and could be a useful tool to escalate treatment in CS. Funding Acknowledgement Type of funding sources: None.