Article Text

Download PDFPDF

26 Prevalence and risk factors for atrial fibrillation in beta-thalassaemia major: UK specialist cardio-haematology clinic experience and development of a predictive af risk scoring system (CH4ADI)
  1. Robert Bell1,
  2. Tsz Chau1,
  3. Munashe Veremu1,
  4. Amna Abdel-Gabir1,
  5. Ratna Chatterjee2,
  6. John Porter1,
  7. Malcolm Walker1
  1. 1University College London
  2. 2University College London Hospital


Introduction The prognosis of Beta-thalassemia major (TM) has been transformed in recent decades, revealing new management challenges including atrial fibrillation (AF). While recognised as a common finding, neither the prevalence nor risk factors for developing AF in this patient cohort are well characterised.

Methods We undertook a retrospective cross-sectional analysis of TM patients under review in our specialist-lead, cardio-haematology clinic at UCLH. Two cohorts were established: 80-patient derivation and a 189-patient validation group. Multivariate analysis was performed, deriving odds ratios (OR) to identify potential risk factors to derive a risk score that was then tested using the Receiver Operator Characteristic (ROC), and verified within the validation cohort.

Results Within the 80 patient derivation cohort, aged 20–57 years (mean 38.3, median 38), 27 (33.8%) had AF, representing 75–240 fold greater prevalence by age group compared to the non-TM population. We identified five significant cardiovascular risk factors: Conduction disease (OR 6.11, 95% confidence interval (CI): 1.08–34.44, p=0.04), history of Heart failure (OR 17.85, CI: 2.06–154.50, p=0.009), Atrial enlargement (OR 3.68, CI: 1.36–9.92, p=0.01), Diabetes (OR 4.25, CI: 1.57–11.54, p=0.005) and myocardial Iron overload (T2*<15ms) (OR 4.01, CI: 1.43–11.21, p=0.008). Atrial area is a surrogate for left ventricular diastolic dysfunction: similar ORs were found in association with the Doppler indices, E/A and E/e’.

Interestingly, we found an age-dependent deterioration of diastolic function that was absent in non-AF patients, suggesting an interaction between AF and progressive diastolic dysfunction which may predict future morbidity in these patients.

Based on the risk-factor odds ratios, we were able to construct a risk score (CH4ADI) with a ROC AUC of 0.823 (CI: 0.730–0.917, p<0.001 figure 1A) that was corroborated in the 189 patient validation cohort with good reproducibility with a ROC AUC of 0.720 (CI 0.560–0.751, p<0.001 figure 1B).

Abstract 26 Table 1

Conclusion We demonstrate an extremely high prevalence of early-onset AF in TM patients, identifying five clearly definable risk factors and constructed a reproducible AF prediction score (table 1). Given the thromboembolic risk associated with AF, we believe that clinicians should consider active rhythm monitoring of patients who we may now be able to identify as at being at high risk of developing AF.

Conflict of Interest None

  • Atrial fibrillation
  • Thalassaemia Major
  • Diastolic heart failure

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.