Regional variations in Takotsubo syndrome outcomes: a nationwide Bayesian analysis
E OmerovicAbstract
Background
Swedish healthcare mandates prioritize equity, yet regional disparities in Takotsubo syndrome (TTS) outcomes remain poorly characterized. Frequentist methods often lack sensitivity for rare outcomes in registry data. We used a Bayesian framework to quantify regional disparities in clinical presentation and survival across the healthcare system.
Methods
A cohort of 5,431 patients from the SWEDEHEART registry (2005–2026) was stratified into six healthcare regions (North, Uppsala-Örebro, Stockholm-Gotland [Reference], West, Southeast, and South). These regions are administrative collaborations for specialized care. Baseline imbalances were assessed via Max Standardized Mean Difference (Max SMD). To understand differences in survival rates, we used Bayesian Cox proportional hazards models to look at the risk of dying within 30 days and 1 year, which helped us calculate the likelihood of different risk A Bayesian logistic regression was also used to figure out how likely it was that someone would have cardiogenic shock when they first came in. All models were adjusted for a comprehensive set of baseline covariates, including comorbidities and socioeconomic status, to ensure that regional estimates were independent of demographic and clinical risk profiles. Primary outcomes (30-day/1-year mortality; cardiogenic shock) were analyzed using adjusted Bayesian models.
Results
Baseline risk for 5,255 patients showed significant imbalances in sex (Max SMD: 0.220; 18% males in the South vs. 29% in the West), smoking (0.146), hypertension (0.145), and age (0.118), but not diabetes (0.059). Regional cardiogenic shock risk was relatively uniform, with posterior probabilities of higher risk vs. reference ranging from 11.9% (North) to 57.0% (South). Smoking (95% CrI: 0.13–0.26) and hypertension (0.01–0.25) were strong predictors of shock. Conversely, survival disparities were substantial after adjustment. Uppsala-Örebro and Southeast showed 96.9% and 92.5% probabilities of superior 30-day survival. At 1 year, Uppsala-Örebro maintained a 97.8% probability of lower mortality (aHR: 0.77). Simultaneously, the South region showed an 89.4% probability of higher mortality (aHR: 1.31).
Conclusion
Swedish TTS outcomes exhibit a significant mortality gradient not explained by baseline risk. While shock risk is uniform, survival probabilities (up to 98%) suggest regional variations in longitudinal care delivery. These findings highlight regional inequities and the need for national standardization of care to ensure equitable survival.Figure 1