Cardiogenic shock (CS) remains a leading cause of in-hospital death after acute coronary syndromes (ACS). Risk prediction in patients who are not yet in overt shock—but may deteriorate during hospitalization—is important for triage, monitoring intensity, and planning escalation of care. The SEX-SHOCK score provides sex-specific estimates of the probability of developing in-hospital CS in adults with ACS who undergo percutaneous coronary intervention (PCI), using presentation and laboratory data, with optional post-procedural angiographic information when the full model is used.

Clinical problem: CS after ACS

ACS encompasses ST-elevation myocardial infarction (STEMI), non–ST-elevation ACS (NSTE-ACS), and related presentations. A subset of patients initially stable from a hemodynamic standpoint may progress to CS during the index hospitalization. Once CS is established, mortality is high and interventions may be applied late. Tools that quantify prospective CS risk in the pre-shock phase can support shared decision-making about level of care, frequency of reassessment, and readiness for vasopressor, inotropic, or mechanical circulatory support pathways—always alongside bedside judgment, imaging, and invasive hemodynamics when indicated.

Why a sex-specific approach matters

Women and men with ACS differ in age distribution, comorbidity burden, presentation patterns (for example, NSTE-ACS is more common in women in many cohorts), delays to revascularization, and use of invasive procedures. Prior risk scores developed in predominantly male populations may underperform in women or mis-calibrate absolute risk. SEX-SHOCK was therefore developed using sex-disaggregated data and separate coefficients for females and males, with the goal of more equitable and accurate risk estimation across sexes.

Relationship to the ORBI score

The ORBI score was an earlier tool focused on predicting in-hospital CS among STEMI patients treated with primary PCI, using presentation and angiographic variables. In large multinational analyses, ORBI showed somewhat lower discriminative performance for CS prediction in female patients than in males. SEX-SHOCK retains a similar overall structure (logistic modeling, integration of clinical and procedural factors) but refines the predictor set: notably, it incorporates C-reactive protein (CRP), creatinine, left ventricular ejection fraction (LVEF) category, and ST-segment elevation, while replacing several ORBI components that were less optimal in sex-stratified modeling. Killip classification in SEX-SHOCK emphasizes Killip class III (rather than scoring Killip II as in ORBI), reflecting how the final logistic models were parameterized in the derivation work.

Derivation and validation at a glance

SEX-SHOCK was developed using large European ACS registries of patients undergoing PCI, excluding those with CS on admission. Development combined traditional logistic regression with machine-learning approaches for variable exploration; the publicly disseminated clinical calculators are based on the logistic formulation with published coefficients. The score was internally validated and assessed in external cohorts from different countries and health systems, with reporting of discrimination (for example, area under the ROC curve), calibration concepts, and measures relevant to reclassification versus ORBI.

As with all registry-derived tools, performance will vary somewhat by case mix, era of care, and data completeness (especially for biomarkers such as CRP, which may not be available for every patient in routine practice).

Model structure: linear predictor and probability

For each sex, the model specifies a linear predictor Y as the sum of an intercept and weighted terms. Continuous inflammatory and renal markers enter as base-2 logarithms: log₂(CRP in mg/L + 1) and log₂(creatinine in µmol/L), reflecting right-skewed laboratory distributions and aligning with the published nomogram methodology. Binary and categorical clinical features contribute sex-specific β coefficients. The predicted probability of in-hospital CS is then:

Probability (%) = 100 / (1 + e^−Y)

Variables typically included

• Laboratory: CRP and creatinine (log₂-transformed as specified).
• ECG / syndrome: ST-segment elevation (ACS presentation).
• LVEF: modeled with a reference category and separate indicators for mid-range and preserved ejection fraction (exact coding follows the published coefficient table).
• Demographics and presentation: age over 70 years, presentation with cardiac arrest, Killip class III, heart rate over 90/min on admission.
• Hemodynamics / metabolic: low systolic blood pressure combined with narrow pulse pressure (as defined in the score), and hyperglycemia above the specified threshold.
• Post-PCI angiography (full model): left main culprit lesion and TIMI flow less than 3 in the infarct-related artery after PCI—used when the “full” model is appropriate.

SEX-SHOCK full versus SEX-SHOCK light

A full model incorporates angiographic predictors after PCI when those data are available. A pre-specified SEX-SHOCK light formulation omits procedural/angiographic terms and uses its own intercept and coefficients for the remaining variables. This supports risk estimation earlier in the course of care or when cath-lab results are not yet finalized. The two formulations answer slightly different timing questions; neither replaces repeat clinical assessment as the patient evolves.

How to use the estimate responsibly

The output is a modeled probability of in-hospital CS, not a diagnosis and not a substitute for hemodynamic monitoring, echocardiography, coronary anatomy, or specialist consultation. Abrupt changes in blood pressure, perfusion, lactate, rhythm, or ischemia should prompt reassessment regardless of a prior probability estimate.

When communicating with patients and families, frame the result as one structured estimate derived from population data, to be integrated with individual context—including frailty, bleeding risk, comorbid conditions, and goals of care.

Limitations

• Population fit: Coefficients reflect the derivation cohorts; transportability to other regions or ethnic groups may differ.
• Missing data: Missing CRP or other inputs can limit real-world use; imputation was used in research datasets but clinical sites must decide how to handle absent labs.
• Timing: The outcome is in-hospital CS during the admission; the exact time-to-event profile may not be captured in a single static calculation.
• PCI requirement: The tool is intended for ACS populations managed with PCI in the development framework; extrapolation to patients not undergoing PCI should be cautious.

Educational overview for the CalcMD SEX-SHOCK calculator. Clinical decisions should follow institutional protocols and current society guidance for ACS and cardiogenic shock.