Related Experiment Video
Updated: Sep 4, 2025
Predicting Amputation using Local Circulating Mononuclear Progenitor Cells in Angioplasty-treated Patients with Critical Limb Ischemia
Published on: September 22, 2020
Multimorbidity and Mortality Models to Predict Complications Following Percutaneous Coronary Interventions
Mandeep Singh1, Rajiv Gulati1, Bradley R Lewis2
1Department of Cardiovascular Medicine (M.S., R.G., M.A., P.F., M.R.B.), Mayo Clinic, Rochester, MN.
Insights
New percutaneous coronary intervention (PCI) risk models predict in-hospital mortality, bleeding, acute kidney injury, and stroke. These validated models offer valuable prognostic information for patients and physicians, improving PCI care.
Area of Science:
- Cardiovascular Medicine
- Medical Informatics
- Health Outcomes Research
Background:
- Existing percutaneous coronary intervention (PCI) risk models often focus on single outcomes, limiting their clinical utility.
- There is a need for comprehensive risk assessment tools that predict multiple adverse events following PCI.
- This study addresses the limitations of single-outcome models by developing multi-outcome risk prediction tools.
Purpose of the Study:
- To develop and validate contemporary risk models for percutaneous coronary intervention (PCI).
- To predict in-hospital mortality, post-procedure bleeding, acute kidney injury, and stroke using a common set of variables.
- To provide clinicians with improved prognostic and therapeutic information for PCI patients.
Main Methods:
- Logistic regression models were developed using data from the Mayo Clinic registry (2000-2016) for patients undergoing their first PCI.
- Two sets of models were created: one using only precatheterization variables (Model B) and another including procedural and angiographic variables (Model A).
- Model performance was assessed using C statistics and validated through bootstrap analysis and calibration curves across various patient subgroups.
Main Results:
- The developed risk models demonstrated strong predictive accuracy for in-hospital death (C-statistics 0.90-0.92), bleeding (0.67-0.70), acute kidney injury (0.76-0.77), and stroke (0.71).
- Bootstrap analysis confirmed the models were not overfit, and calibration curves showed good agreement between predicted and observed outcomes.
- The models proved robust across diverse patient populations, including those with acute coronary syndrome, cardiogenic shock, diabetes, women, and the elderly.
Conclusions:
- New risk scoring models, utilizing both precatheterization and procedural/angiographic variables, accurately predict multiple adverse outcomes after PCI.
- These models offer a significant advancement over single-outcome predictors, providing comprehensive risk assessment.
- The ease of application of these models will enhance prognostic guidance and therapeutic decision-making for both patients and physicians.
Background:
Previous percutaneous coronary intervention risk models were focused on single outcome, such as mortality or bleeding, etc, limiting their applicability. Our objective was to develop contemporary percutaneous coronary intervention risk models that not only determine in-hospital mortality but also predict postprocedure bleeding, acute kidney injury, and stroke from a common set of variables.
Methods:
We built risk models using logistic regression from first percutaneous coronary intervention for any indication per patient (n=19 322, 70.6% with acute coronary syndrome) using the Mayo Clinic registry from January 1, 2000 to December 31, 2016. Approval for the current study was obtained from the Mayo Foundation Institutional Review Board. Patients with missing outcomes (n=4183) and those under 18 (n=10) were removed resulting in a sample of 15 129. We built both models that included procedural and angiographic variables (Models A) and precatheterization model (Models B).
Results:
Death, bleeding, acute kidney injury, and stroke occurred in 247 (1.6%), 650 (4.3%), 1184 (7.8%), and 67 (0.4%), respectively. The C statistics from the test dataset for models A were 0.92, 0.70, 0.77, and 0.71 and for models B were 0.90, 0.67, 0.76, and 0.71 for in-hospital death, bleeding, acute kidney injury, and stroke, respectively. Bootstrap analysis indicated that the models were not overfit to the available dataset. The probabilities estimated from the models matched the observed data well, as indicated by the calibration curves. The models were robust across many subgroups, including women, elderly, acute coronary syndrome, cardiogenic shock, and diabetes.
Conclusions:
The new risk scoring models based on precatheterization variables and models including procedural and angiographic variables accurately predict in-hospital mortality, bleeding, acute kidney injury, and stroke. The ease of its application will provide useful prognostic and therapeutic information to both patients and physicians.
Related Concept Videos
01:14Multicompartment Models: Overview
These models offer a more comprehensive representation of drug behavior in the body than one-compartment models. They accommodate the complexity of drug distribution,...
01:27Coronary Artery Disease V: Interprofessional Care
01:05Two-Compartment Open Model: Overview
The...
01:30Acute Coronary Syndrome III: Diagnostic Studies
01:21Mechanistic Models: Overview of Compartment Models
01:20Pharmacokinetic Models: Overview
There are three primary types of models: empirical, compartment, and physiological. Empirical models, with minimal...