P141 - ASSOCIATIONS BETWEEN WRIST-WORN ACCELEROMETRY BIOMARKERS AND MORTALITY IN HEART FAILURE

Background: Heart failure (HF) is associated with reduced functional capacity, which is a negative prognostic marker in HF patients. While typically measured in the clinic, functional capacity may be estimated at home using accelerometers (ACC) such as those found in smartwatches. However, aggregate measures such as steps per day have shown limited prognostic value. We propose that novel ACC metrics capturing the central tendency of movement may provide better prognostic information. Here, we investigate the relationship between a novel ACC biomarker and all-cause and cardiovascular (CV) mortality in patients with HF.

METHODS AND RESULTS: Participants in the 2011–2012 and 2013–2014 NHANES waves wore the Actigraph GT3X+ on their wrist for 7 days and were followed up to confirm mortality status and cause of death (ICD-10) in 2019. The sum of vector magnitudes (SVM), a measure of average acceleration over a rolling time window validated in HF cohorts, was calculated for each 5-minute period. For each participant, SVM values were ranked, and the 90th percentile (SVM-90) was captured. Logistic regression models evaluated associations between SVM-90 and all-cause and CV mortality in HF patients. SVM models were compared with those using the previously described Monitor-Independent Movement Summary (MIMS) metric. Univariate and adjusted models (age, sex, BMI) were employed.
291 participants with self-reported HF (50.9% female, mean (SD) age 66.8 (12.6)) were included in the final analysis (Table 1). Univariate logistic regression models showed significant associations of all-cause mortality with SVM-90 (OR 0.61 [0.47,0.79], p< 0.001) and MIMS (OR 0.69 [0.54,0.89], p=0.004; Fig.1a). In models adjusted for age and sex, only SVM-90 was associated with increased all-cause mortality (OR 0.76 [0.57,0.998], p=0.048) (Fig.1b). Univariate models assessing odds of CV mortality found a significant effect of SVM-90 only (OR 0.68 [0.49,0.94], p=0.019; Fig.1c) while multivariate models adjusted for age and sex did not reveal significant effects of any accelerometry measure (Fig.1d). Neither SVM-90 nor MIMS were associated with all-cause or CV mortality in models adjusted for age, sex, and BMI (SVM-90: OR 0.79 [0.59,1.08], p=0.14 for all-cause; OR 0.83 [0.55,1.26], p=0.37 for CV mortality).

Conclusion: We present a novel ACC biomarker, SVM-90, more robustly associated with all-cause and CV mortality in HF patients than a previously described ACC measure. While no ACC biomarker remained associated with mortality in fully-adjusted models, future dedicated studies investigating ACC labeling techniques may improve prognostic value and use in clinical studies and clinical trials in HF.