A new electrocardiographic left ventricular hypertrophy prognostic score

Area of Science:

• Cardiology
• Predictive Modeling
• Epidemiology

Background:

• The Romhilt-Estes score is a traditional method for diagnosing left ventricular hypertrophy (LVH).
• Optimizing predictive accuracy for cardiovascular mortality is crucial in clinical practice.
• Evaluating novel scoring systems against established criteria is essential for advancing diagnostic tools.

Purpose of the Study:

• To assess if a modified Romhilt-Estes score, incorporating Cox hazard models for component weighting, improves cardiovascular mortality prediction.
• To determine if Cornell voltage criteria should supersede original criteria within the scoring system.
• To compare the predictive performance of the original Romhilt-Estes score against the modified version.

Main Methods:

• A Cox proportional hazards model was used to derive new multipliers for Romhilt-Estes score components, including age and gender.
• Cornell voltage criteria were analyzed for their predictive value, with and without adjustments for demographic factors.
• The modified score, with and without Cornell voltage criteria, was evaluated against the original Romhilt-Estes score using data from 20,903 subjects followed for 17 years.

Main Results:

• The modified score demonstrated superior predictive ability, with an area under the curve (AUC) of 0.7 compared to 0.63 for the original Romhilt-Estes score.
• Patients meeting the "definite" LVH cutoff in the modified score had a significantly higher hazard ratio (HR) for cardiovascular mortality (13.6) versus the original score (5.6).
• All seven components of the new score were significant predictors of cardiovascular mortality, with gender having the greatest HR.

Conclusions:

• The modified Romhilt-Estes score, utilizing Cox hazard-derived multipliers and excluding voltage criteria, significantly outperforms the original score in predicting cardiovascular mortality.
• The study highlights the value of Cox regression in refining existing clinical prediction scores.
• The findings suggest a more accurate and robust method for identifying individuals at high risk of cardiovascular events.