Abstract
BACKGROUND: High serum digoxin concentrations (SDCs) are linked to adverse outcomes in patients with heart failure (HF). However, little is known about the prevalence, predictors and clinical outcomes of elevated SDCs in a real-world cohort of patients with HF. PURPOSE: This study aimed to assess baseline differences between HF patients on digoxin therapy and those not receiving the drug, to ascertain the prevalence and predictors of elevated SDCs, and to explore the relationship between SDC levels and clinical outcomes in digoxin users. METHODS: This was a post-hoc secondary analysis from the prospective, observational A systems BIOlogy Study to Tailored Treatment in Chronic Heart Failure (BIOSTAT-CHF) study. A cohort of 2375 patients were included in the current analysis, among whom 338 (14%) received digoxin, with complete digoxin dosage and SDC data available. Baseline differences were examined among patients receiving digoxin and those not, as well as trends across increasing tertiles of SDC (Tertile 1: 0.19–0.62 ng/mL; T2: 0.63–0.98 ng/mL; T3: 1.00–2.97 ng/mL) in those using digoxin. Multivariable linear regression models were performed to identify predictors of SDC, and COX proportional hazards models to assess associations between SDC and all-cause mortality and HF hospitalization after 2 years. RESULTS: In the present analysis, patients had a mean age of 69 ± 12 years. Seventy nine (23%) and 556 (27%) patients in the digoxin and non-digoxin group were women, respectively. The median follow-up was 21 (IQR 15–27) months. Patients receiving digoxin were more likely to have history of AF (73% vs 39%, p < 0.001), presence of AF (74% vs. 30%, p < 0.001) and a higher NYHA class, compared to patients not receiving digoxin. Seventy (21%) patients were above the ESC guideline recommended SDC level (> 1.2 ng/mL). Higher SDCs were associated with higher NT-proBNP levels (p for trend = 0.004), but this association did not persist following multivariable adjustment. In multivariable linear regression analyses, no significant predictors of SDC were identified, including renal function (eGFR, creatinine) or renal disease (all p ≥ 0.05). A statistically significant association was found between SDC and HF hospitalization (T2 vs T1 aHR 1.83, 95% CI 1.04–3.20; T3 vs T1 aHR 1.62, 95% CI 0.93–2.80; T1 reference group) following multivariable adjustment. However, no significant associations were detected between SDC and all-cause mortality nor the combined endpoint of all-cause mortality and/or HF hospitalization (p ≥ 0.05 for all). CONCLUSIONS: In this real-world cohort of patients with HF, a substantial number of patients presented with SDCs above the guideline-recommended levels. Digoxin use showed association with NT-proBNP levels, but after multivariable analysis we did not find specific predictors of SDC, including kidney function. Higher SDCs were associated with a higher risk of HF hospitalization but not all-cause mortality. [Figure: see text]