📄 medicine

Development and Validation of a Nomogram for Predicting Cardiogenic Shock in Elderly Patients with ST-Segment Elevation Myocardial Infarction: A Retrospective Cohort Study

This retrospective cohort study developed and internally validated a nomogram that effectively predicts admission-time cardiogenic shock risk in elderly STEMI patients, demonstrating that a parsimonious model combining Killip classification and age offers a powerful, practical bedside tool comparable in performance to a more complex model incorporating multiple laboratory and echocardiographic parameters.

Original authors: Zunguo Liu, Shaobing Yang, Yuanyuan Mao, Li Ma, Wanyu Song, Hui Huang

Published 2026-10-18

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Zunguo Liu, Shaobing Yang, Yuanyuan Mao, Li Ma, Wanyu Song, Hui Huang

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Heart attacks are a terrifying emergency where blood flow to the heart muscle is suddenly blocked, causing tissue to die. While modern medicine has become very good at reopening these blocked arteries, a specific and deadly complication can still strike: the heart simply stops pumping with enough force to keep the body alive. This state, known as cardiogenic shock, is the most common cause of death following a heart attack, particularly in older adults. The challenge for doctors is that this condition can develop silently or progress rapidly, and the tools currently used to spot it early are often too blunt to catch the most vulnerable patients before it is too late.

Researchers at the General Hospital of Ningxia Medical University set out to build a better early warning system specifically for elderly patients. They focused on a group of 527 people aged 60 and older who had been admitted with a severe type of heart attack. The team wanted to create a simple visual tool, called a nomogram, that could take a few pieces of information available the moment a patient arrives at the hospital and calculate their specific risk of falling into shock. The goal was to see if combining a classic physical exam with modern blood tests could predict danger more accurately than either method could on its own.

The study began by looking at the records of these patients to see who developed shock and who did not. The researchers gathered a wide range of data, including the patient's age, their blood pressure, and a standard assessment of their heart function called the Killip classification. This classification is a long-standing method where doctors listen to the lungs and check for fluid buildup to grade how well the heart is working. The team also collected detailed blood work, looking for markers of inflammation, blood clotting, and kidney function, as well as measurements of how much blood the heart was pumping with each beat.

When the researchers analyzed the data, two factors stood out as the strongest predictors of shock. The first was the patient's age; for every year older a patient was, their risk increased slightly but significantly. The second, and far more powerful, factor was the Killip classification. Patients who showed signs of severe heart failure or fluid in their lungs upon arrival were vastly more likely to go into shock than those who appeared stable. The study confirmed that these two pieces of information—how old the patient is and how their heart is functioning right now—carry the most weight in predicting the outcome.

The team then tested whether adding specific blood test results could improve the prediction. They looked at markers like D-dimer, which indicates blood clotting activity, and the neutrophil-to-lymphocyte ratio, a measure of the body's inflammation. They also considered glucose levels and the heart's pumping strength. When they combined all these factors into a single complex model, the tool became slightly more accurate at distinguishing between patients who would survive and those who would not. However, the improvement was small. The study found that while the full model with all the blood tests was the most precise, a much simpler version that used only the patient's age and their heart function grade performed almost as well.

This discovery offers a practical path forward for emergency rooms. The researchers created a simplified scoring system that a doctor could use at the bedside without needing a computer or a calculator. If a patient is older and shows signs of heart failure, the risk is high, regardless of what the blood tests say. If the patient is younger and stable, the risk is low. The study suggests that while blood tests provide a more complete picture, they do not change the fundamental risk assessment enough to justify the delay of waiting for results in every single case. The most powerful tool for saving lives in these moments remains a careful physical exam and knowing the patient's age.

The researchers were careful to note that their tool was built and tested on a single group of patients in one hospital. While the results were consistent and the tool performed well in their tests, it has not yet been proven to work in other hospitals or with different patient populations. The study also highlighted a limitation: the tool is best used to assess the overall risk of a patient arriving at the hospital, including those who are already in shock, rather than predicting which stable patients will suddenly develop shock later. Despite these boundaries, the work provides a clear, evidence-based method for doctors to quickly identify the elderly heart attack patients who need the most urgent and aggressive care to prevent a fatal collapse.

Technical Summary: Development and Validation of a Nomogram for Predicting Cardiogenic Shock in Elderly STEMI Patients

Problem Statement
Cardiogenic shock (CS) remains the most lethal complication of ST-segment elevation myocardial infarction (STEMI), with in-hospital mortality rates reaching 40%–50%. Elderly patients (≥60 years) are particularly vulnerable due to complex comorbidities and atypical presentations, yet dedicated prediction tools for CS risk stratification in this specific demographic are limited. While the Killip classification is a standard bedside tool for assessing cardiac function in acute myocardial infarction (AMI), its sensitivity and specificity for predicting CS when used alone are constrained. Previous models have either targeted patients who have already developed CS in the ICU or failed to focus specifically on the elderly population. This study addresses the need for an admission-time risk stratification tool that integrates clinical assessment with laboratory parameters to predict CS in elderly STEMI patients.

Methodology
This single-center retrospective cohort study enrolled 527 patients aged ≥60 years with confirmed STEMI admitted between January 2019 and December 2022. After excluding four patients with missing CS status, 523 patients (39 CS cases, 484 non-CS cases) were analyzed.

Data Collection: Variables included demographics, Killip classification, vital signs, comorbidities, echocardiography (LVEF, LVEDD), coronary angiography, and a comprehensive panel of laboratory tests (e.g., D-dimer, NLR, glucose, NT-proBNP, hemoglobin) obtained within 24 hours of admission.
Variable Selection: Candidate variables were screened using univariate logistic regression ( $P < 0.1$ ). To address multicollinearity, highly correlated variables were selectively removed (e.g., retaining NLR over individual neutrophil/lymphocyte percentages). The final model construction adhered to the events-per-variable (EPV) ratio principle (39 events supporting ~6 variables).
Model Construction: Two primary multivariate logistic regression models were developed:
1. Combined Model: Killip class III–IV, D-dimer, NLR, LVEF, glucose, and age.
2. Laboratory-Only Model: D-dimer, NLR, LVEF, glucose, age, and hemoglobin (excluding Killip).
3. Reduced Model: A parsimonious model restricted to Killip class III–IV and age.
Validation and Evaluation: Model discrimination was assessed via the Area Under the Receiver Operating Characteristic Curve (AUC) with DeLong tests for comparison. Calibration was evaluated using the Hosmer-Lemeshow test. Internal validation was performed using 1,000 bootstrap resamples to calculate optimism-corrected AUC. Sensitivity analyses included Firth penalized logistic regression (for rare events), LASSO regression, and interaction analysis (NLR × Killip). Decision Curve Analysis (DCA) was used to evaluate clinical utility. A simplified integer-based scoring rule was derived for bedside application.

Key Results

Predictors: In the combined model, Killip class III–IV (OR = 13.531, $P < 0.001$ ) and age (OR = 1.076, $P = 0.013$ ) were the only independent predictors of CS. While D-dimer, NLR, LVEF, and glucose were significant in univariate analysis, they did not reach statistical significance in the multivariate combined model.
Model Performance:
- The Combined Model (Killip + labs + age) achieved an AUC of 0.874 (95% CI: 0.819–0.927).
- This significantly outperformed Killip classification alone (AUC = 0.815, DeLong $P = 0.001$ ) and the Laboratory-only model (AUC = 0.775, DeLong $P = 0.014$ ).
- The Reduced Model (Killip + age) achieved an AUC of 0.864, which was not statistically different from the full combined model (Likelihood Ratio Test $P = 0.168$ ; DeLong $P = 0.140$ ).
Validation: The bootstrap-corrected AUC for the combined model was 0.858, indicating good stability with minimal overfitting (optimism = 0.016). Calibration was good (Hosmer-Lemeshow $P = 0.412$ ).
Clinical Utility: Decision Curve Analysis demonstrated a positive net benefit for the combined model across threshold probabilities of 5%–25%. A simplified 25-point bedside scoring rule derived from the model achieved an AUC of 0.846.

Significance and Claims
The authors claim that the developed nomogram effectively predicts admission-time CS risk in elderly STEMI patients. The study highlights two complementary findings:

Incremental Value: Integrating laboratory parameters (D-dimer, NLR, LVEF, glucose) with Killip classification and age significantly improves discrimination compared to using Killip classification alone.
Parsimony: The study modestly concludes that the incremental value of the four laboratory variables over a model containing only Killip class and age is limited and not statistically significant. Consequently, the authors propose that when Killip classification is available, adding age alone constitutes a powerful, parsimonious, and rapid bedside tool for risk stratification. The full nomogram remains preferable when maximal predictive performance is required.

The paper explicitly acknowledges limitations, including the lack of external validation, the single-center retrospective design, and the small number of CS events (39), which may limit statistical power to detect subtle incremental effects of laboratory variables. Furthermore, the authors note that the model addresses "overall admission-time CS risk" (including patients already in shock) rather than predicting de novo CS progression in patients presenting without shock, due to the overlap between Killip class IV and the CS outcome definition. Future work is identified as needing multicenter prospective studies with external validation.

More like this