Longitudinal Changes in Health-Related Quality of Life in Patients With Atrial Fibrillation

Abstract

Background Optimizing health-related quality of life (HRQoL) is an important aim of atrial fibrillation (AF) treatment. Little is known about patients' long-term HRQoL trajectories and the impact of patient and disease characteristics. The aim of this study was to describe HRQoL trajectories in an observational AF study population and in clusters of patients with similar patient and disease characteristics. Methods and Results We used 5-year follow-up data from the Swiss-Atrial Fibrillation prospective cohort, which enrolled 2415 patients with prevalent AF from 2014 to 2017. HRQoL data, collected yearly, comprised EuroQoL-5 dimension utilities and EuroQoL visual analog scale scores. Patient clusters with similar characteristics at enrollment were identified using hierarchical clustering. HRQoL trajectories were analyzed descriptively and with inverse probability-weighted regressions. Effects of postbaseline clinical events were additionally assessed using time-shifted event variables. Among 2412 (99.9%) patients with available baseline HRQoL, 3 clusters of patients with AF were identified, which we characterized as follows: "cardiovascular dominated," "isolated symptomatic," and "severely morbid without cardiovascular disease." Utilities and EuroQoL visual analog scale scores remained stable over time for the full population and the clusters; isolated symptomatic patients showed higher levels of HRQoL. Utilities were reduced after occurrences of stroke, hospitalization for heart failure, and bleeding, by -0.12 (95% CI, -0.18 to -0.06), -0.10 (95% CI, -0.13 to -0.08), and -0.06 (95% CI, -0.08 to -0.04), respectively, on a 0 to 1 utility scale. Utility of surviving patients returned to preevent levels 4 years after heart failure hospitalization; 3 years after bleeding; and 1 year after stroke. Conclusions In patients with prevalent AF, HRQoL was stable over time, irrespective of baseline patient characteristics. Clinical events of hospitalization for heart failure, stroke, and bleeding had only a temporary effect on HRQoL.

Keywords: atrial fibrillation; health‐related quality of life.

Figure 1. Unadjusted trajectories of utilities and EQ VAS scores.

A, Unadjusted trajectories of utilities. Follow‐ups were performed in yearly steps. B, Unadjusted trajectories of EQ VAS scores. Visits were performed in yearly steps. Numbers of observation are indicated above the visit numbers. EQ VAS indicates EuroQoL visual analog scale.

Figure 2. Predicted utility values by cluster.

Visits occurred in yearly steps. Utilities were predicted on the basis of an inverse probability‐weighted model, including an interaction term between visit number and cluster.

Figure 3. Predicted EQ VAS scores by cluster.

Visity occurred in yearly steps. EQ VAS scores were predicted on the basis of an inverse probability‐weighted model, including an interaction term between visit number and cluster. EQ VAS indicates EuroQoL visual analog scale.

Figure 4. Predicted utitlity values according to cluster, dependent on time since diagnosis.

Utilities were predicted on the basis of an inverse probability‐weighted model, including an interaction term between time since diagnosis and cluster.

Figure 5. Predicted EQ VAS scores according to cluster, dependent on time since diagnosis.

Utilities were predicted on the basis of an inverse probability‐weighted model, including an interaction term between time since diagnosis and cluster. EQ VAS indicates EuroQoL visual analog scale.

Figure 6. Effects of follow‐up events on utility.

The effects were predicted on the basis of an inverse probability‐weighted model. The time point 0 indicates the follow‐up visit at which the first follow‐up event of the relevant type was recorded. For patients without a follow‐up event, all visits were denoted as −1. Time point −1 was set as the reference time point. Effects were adjusted for cluster and for age at study visit.