Background: The presence of multiple chronic conditions (multimorbidity) has profound impacts on the health and health care of older adults. As a result, epidemiologic and clinical research on older adults depends on being able to measure multimorbidity in an accurate and conceptually valid manner. Yet, existing methods of measuring multimorbidity have critical gaps. Most existing measures that use commonly-available data sources include only diseases that predict hospitalization and death. These methods are ill-suited to measure other aspects of multimorbidity, including its impact on older adults' ability to function independently in daily life ? an outcome of vital importance to this population. The large and growing body of research and clinical programs that focus on functional outcomes thus lack the proper tools to address multimorbidity. Innovative uses of claims data hold great promise to develop new measures of multimorbidity focused on its impact on functional outcomes. Moreover, these approaches offer a unique opportunity to improve upon existing multimorbidity measures that focus on ?traditional? outcomes such as hospitalization and death.
Aims : (1) To develop and validate claims-based measures of multimorbidity that predict decline in ability to perform basic and instrumental activities of daily living (ADLs, IADLs); (2) Using an expanded range of disease characteristics measurable in claims data, to develop and validate measures of multimorbidity that predict hospitalization and death; (3) To compare the predictive validity of our measures for functional decline, hospitalization, and death with existing measures of multimorbidity such as the Charlson Index. Methods: Using Medicare claims data linked to self-report ADL and IADL data from the Health and Retirement Study, we will build 4 indices of multimorbidity using the framework of prognostic model development. Each index will be developed to predict a separate outcome related to multimorbidity, including decline in ability to perform ADLs (e.g. bathing), decline in ability to perform IADLs (e.g. shopping), hospitalization, and death. Predictors in these models will be disease characteristics that can be measured in claims data including diseases, markers of disease severity, and disease-disease interactions. These prognostic models will be converted into simple indices, where each disease characteristic will be assigned a number of points, and the sum of these points will give a patient-level multimorbidity score. We will internally validate our indices using bootstrapping techniques, and externally validate them in a cohort of older adults from the National Health and Aging Trends Study. Throughout this process, we will consult with a group of expert advisors to ensure that we develop our measures in a manner that is maximally useful to researchers and health systems leaders. Relevance / public health significance: Improved measures of multimorbidity that reflect patient-centered outcomes will be crucial tools for clinical epidemiology, health services research, and clinical programs that seek to improve care for older adults.
Multimorbidity has profound impacts on the health and health care of older adults, yet there are key gaps in currently available tools for measuring multimorbidity. Improved measures of multimorbidity that reflect patient- centered outcomes will be crucial tools for clinical epidemiology, health services research, and clinical programs that seek to improve care for older adults.
