Total mortality among levothyroxine-treated women with atrial fibrillation in Swedish primary health care

Abstract

Atrialfibrillation (AF) is equally common amongmen andwomen of Western populations [1], but women develop AF around five years later than men [2], and show less favorable outcomes with higher relative mortality and a higher risk of stroke [3]. Women have a higher prevalence of hypothyroidism and levothyroxine treatment [4], and this treatment could possibly be cardiotoxic in patients with heart disease [5]. Earlier studies have shown conflicting results, with no increased morbidity or mortality [6], increased cardiovascular morbidity [7], and a higher mortality among women with levothyroxine treatment [8]. Our aimwas to study if levothyroxine treatment amongwomen is associatedwithhigher totalmortality or not. The study used individual-level patient data from a total of 75 primary health care centers (PHCC) (n=1,098,420). The data was linked to individual-level data on age, gender, education and marital status for all people registered in Sweden [9], and the Cause of Death Register [10]. The study included women with established atrial fibrillation between 2001 and 2007, identified by the presence of the ICD-10 code for AF in patients' medical records. Cardiovascular disorders and diabetes were used as covariates. Altogether 5642 women≥45 years of age at the time of diagnosis were included, 907 with a concomitant prescription of levothyroxine and 4735 without. Ethical permissionswere achieved at Karolinska Institutet and at the University of Lund. Death was registered from notification of AF diagnosis until 31 December, 2007. Age was dichotomized into 45–79 and ≥80 years. Educational attainment, marital status and drugs prescribed at PHCCs during the assessment period were recorded. Prescription of antithrombotic drugs was coded as no treatment, anti-platelet treatment only, or anticoagulant treatment. Differences in means and distributions were calculated using Student's t-test, chi-square analysis and Fisher's exact test. Follow-up analyses were performed using Cox regression, with time being defined as years from inclusion to death, or to censoring at the end of the followup period. Analyses were performed using separate models for those below 80 and those 80 years of age or older, respectively. In the models we included factors showing significant results in either age-group. A p value b0.05 was considered to be statistically significant. Table 1 shows the characteristics of the sample, divided by the presence of levothyroxine treatment or not. Mean follow-up time was 3.5 years (SD 2.1), in total 3200 person-years. Difference of prescription of pharmacological between the groups was found for anticoagulant agents, specific anti-arrhythmic agents, non-selective beta-blockers, sotalol, and lipid-lowering drugs being more common among younger, and antiplatelet agents, all diuretics and loop-diuretics among older women. Table 2 shows models of Cox' regression, stratified by age group, with diagnoses, prescribed drugs in separate models, and full model with them combined and with also socio-economic factors included. When combining age groups, levothyroxine treatment was associated with a lower mortality when adjusting for co-morbidities, prescribed drugs and socio-economic factors (HR 0.76, 95% CI 0.58–0.99).