CONNECTIONS THAT COUNT: A NETWORK ANALYSIS OF MULTIMORBIDITY AND GAIT SPEED IN OLDER ADULTS

multimorbidity;multiple chronic conditions;phenotypic network analysis;gait speed;Epidemiology

Burke, Bridget TeevanRoth, David L. ;

Johns Hopkins University

关键词: multimorbidity;    multiple chronic conditions;    phenotypic network analysis;    gait speed;    Epidemiology;   

Others  :  https://jscholarship.library.jhu.edu/bitstream/handle/1774.2/58642/BURKE-DISSERTATION-2017.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: JOHNS HOPKINS DSpace Repository

【 摘 要 】

Identifying older adults at risk for functional impairment is an important goal in multimorbidity research. Disruption in multiple physiologic systems is common to both multimorbidity and slow gait speed with aging, but the temporal relationship between these factors is not well-elucidated. Moreover, there is currently no consensus on which chronic conditions are most relevant to multimorbidity research or how to best operationalize the measurement of multimorbidity. The objectives were to: 1) evaluate associations between multimorbidity and four-year incident dismobility, very slow gait speed, or mortality in community-dwelling older adults; 2) evaluate the ability of counts of chronic diseases to predict four-year incident dismobility or mortality, and to quantify the extent to which symptoms and geriatric syndromes improves that prediction; 3) estimate and evaluate a phenotypic network of chronic diseases, symptoms, and geriatric syndromes using data from a nationally-representative cohort study of older adults; and 4) demonstrate the predictive validity of this network by evaluating associations among communities of chronic diseases, symptoms and geriatric syndromes and four-year incident dismobility or mortality. Data for this dissertation came from the Health and Retirement Study (HRS), a population-based cohort of U.S. older adults. Multivariable logistic regression was used to evaluate associations between multimorbidity and four-year incident dismobility or mortality. Network analysis was used to estimate and evaluate a network of self-reported chronic diseases, symptoms, and geriatric syndromes. Multiple logistic regression was used to evaluate associations between communities of chronic diseases, symptoms and geriatric syndromes and four-year incident dismobility or mortality. Counts of self-reported chronic diseases had an association with four-year incident dismobility or mortality, controlling for age. Counts of symptoms and geriatric syndromes had an association that was as strong or stronger than counts of chronic diseases, particularly in men. Classification accuracy for incident dismobility or mortality was best for the model that incorporated age, counts of chronic diseases, symptoms and geriatric syndromes, and sociodemographic and health risk factors. A nationally-representative ;;map” of phenotypic relationships was estimated. Communities of commonly co-occurring chronic diseases, symptoms and geriatric syndromes included: musculoskeletal, pulmonary, psychological/somatic, cardiovascular, sensory, cognitive/edentulism, and diabetes/hypertension. The psychological/somatic community was strongly linked to other communities, underscoring its interrelatedness with other diseases, symptoms, and geriatric syndromes. Associations were identified between the cognitive/edentulism, pulmonary, psychological/somatic and cardiovascular communities and cancer with incident dismobility or mortality. Reporting hypertension or diabetes was associated with incident dismobility or mortality in females. Symptoms, geriatric syndromes and sociodemographic factors in older adults indicate vulnerability that is not captured in counts of chronic diseases. We show that the use of network analysis facilitates extensions of multimorbidity beyond counts of co-occurring chronic diseases to encompass dynamic relationships among chronic diseases, symptoms and geriatric syndromes in older adults.