Long-term cardiovascular risk following COVID-19 hospitalisation in a large US health system: temporal and prediction modelling
G Fontana, G Nadimpalli, S Arnold Egloff, D Lakkireddy, S Kar, S Horr, C Martin, L Daniel, A Braisted, H McDonald, L Stergar, K Sands, S ManoukianAbstract
Background
COVID-19 (C19) hospitalisation has been associated with acute cardiovascular (CV) complications; however, long-term CV risk remains incompletely defined.
Purpose
To evaluate sustained 1-yr CV risk following C19 hospitalisation, characterise temporal patterns across C19 variants, including omicron, and identify independent predictors to inform post-discharge risk for CV events.
Methods
We conducted a retrospective observational study of adults hospitalised with C19 across a large US health system between 2020 and 2023 [Fig1A]. Patients were followed for 365 days post-discharge. The primary endpoint was first readmission within 1-yr for any CV event excluding hypertension (composite outcome). Secondary analyses evaluated high morbidity CV events, individually, such as myocardial infarction (MI), pulmonary embolism (PE), and stroke. Autoregressive integrated moving average (ARIMA) models were used to assess expected vs. observed temporal trends by using pre-pandemic rates (2016-2019). Multivariable logistic regression identified independent predictors, and accuracy was evaluated using the area under the receiver operating characteristics curve (AUC). The parsimonious model was used to construct a nomogram for 1-yr CV risk.
Results
Among 195,978 C19 hospitalisations, 43,401 (22.1%) experienced at least 1 major CV readmission within 1-yr. Of those, the majority were AF (42.4%), followed by MI (12.5%), stroke (6.8%), and PE (5.5%) [Fig1B]. CV readmissions were most frequently observed in the first 30 days (8.1%) but began increasing again at 90 days, reaching 3.7% at 365 days, demonstrating sustained risk [Fig1C].Temporal analyses showed that observed rates exceeded expected rates during early pandemic waves and matched the amplitude of C19 hospitalisations across all variant epochs [Fig2A]. MI and PE rates remained elevated during omicron despite its decreased virulence compared to C19 predecessors [Fig2A]. Independent predictors of 1-yr CV readmission included older age, Black race, elevated troponin, lower albumin, diabetes, coronary artery disease, heart failure, chronic kidney disease, chronic obstructive pulmonary disease, and prior or concurrent CV diagnoses during index admission [Fig2B, Any CV AUC >0.80]. Notably, omicron did not reduce long-term CV risk as it was associated with higher odds compared to prior C19 variants (OR 1.38,1.33–1.44). Prior or concurrent PE were strong predictors of recurrent thromboembolic events, while concurrent hypertension was associated with increased MI risk. The nomogram derived from the parsimonious model enabled individualised 1-yr CV risk estimation but merits external validation.
Conclusion
C19 hospitalization is associated with persistent, long-term CV risk. Divergence between observed and expected rates, coupled with robust model performance, underscores the need for individualized risk stratification at C19 discharge to enable targeted CV surveillance and prevention.