Abstract 061: Joint Associations of Objectively Measured Physical Activity and Cardiovascular-Kidney-Metabolic Syndrome with All-Cause Mortality in US Adults
Introduction: Physical activity is recommended for slowing the progression of Cardiovascular-Kidney-Metabolic (CKM) syndrome. Guidelines suggest lifestyle modification may be most beneficial for those with advanced CKM (stages 3-4), but few studies have compared the impact of physical activity across the CKM spectrum. Objective: We examined associations between physical activity (PA) patterns and risk of all-cause mortality across CKM stages. Methods: We conducted a weighted, prospective cohort analysis of the 2003-2006 National Health and Nutritional Examination Survey (NHANES), with follow-up until December 31, 2019. We restricted analyses to those adults who were ≥ 20 years old, not pregnant, and able to walk. We classified participants according to CKM stage. We calculated average daily moderate-to-vigorous physical activity (MVPA) using up to 7 days of actigraphy data and the standard cut-point of 2000 activity counts per minute. We visualized the MVPA distributions by CKM stage and calculated the proportion of adults meeting the recommended MVPA target of ≥ 150 minutes a week (≥ 21.4 minutes per day). We fit Cox regression models to estimate relative and absolute mortality risk using a polynomial B-spline of log-MVPA with a single knot at the midpoint. We stratified all analyses by CKM stage and adjusted models for demographics. Results: We included 7,246 adults (weighted mean age 47.6 yrs, 51.5% female). Over a median follow up of 14.4 years, there were 1676 deaths. The percentage of US adults achieving the AHA-recommended MVPA target ranged from 7.3% (CKM stage 3) to 43.1% (CKM stage 0 to 1) ( Fig. 1A ). Greater MVPA was associated with lower hazard of mortality across CKM stages. Hazard ratios of reaching the AHA recommendation, as opposed to being sedentary, ranged from 0.93 (CKM stage 0/1) to 0.52 (CKM stage 4). The minimum MVPA associated with lower mortality ranged from 4.1 minutes per day (CKM stage 3) to 19.9 minutes/day (CKM stage 0/1) ( Fig. 1B ). The associations between greater MVPA and lower absolute mortality risk were more pronounced in later stage CKM ( Fig. 1C ). Conclusion: MVPA is associated with lower mortality along the full spectrum of CKM syndrome. The benefits of MVPA appeared largest in advanced CKM (stages 3-4), where even modest activity potentially confers cardioprotection.
Authors
- Scott L. Zeger (ORCID: https://orcid.org/0000-0001-8907-1603)
- Chiadi E. Ndumele (ORCID: https://orcid.org/0000-0002-1603-282X)
- Mary R. Rooney (ORCID: https://orcid.org/0000-0002-5607-4848)
- Joseph Sartini (ORCID: https://orcid.org/0000-0001-9081-1955)
- Michael Fang (ORCID: https://orcid.org/0000-0003-2849-1780)
- Jennifer A. Schrack (ORCID: https://orcid.org/0000-0001-9244-9267)
- John W. McEvoy (ORCID: https://orcid.org/0000-0001-6530-5479)
- Elizabeth Selvin (ORCID: https://orcid.org/0000-0002-3539-2070)
Institutions
- Johns Hopkins University (US)
- Ollscoil na Gaillimhe – University of Galway (IE)
- Johns Hopkins Hospital (US)
Publication Details
- Journal
- Circulation
- Published
- 2025-03-11
- DOI
- https://doi.org/10.1161/cir.151.suppl_1.061
- Primary Topic
- Cardiovascular Health and Risk Factors
- Type
- article
- Field-Weighted Citation Impact
- 0.00