ADL-dependent older adults were identified in medico-administrative databases

• Published in: Journal of clinical epidemiology Vol. 139; pp. 297 - 306
• Main Authors: Hucteau, Emilie, Noize, Pernelle, Pariente, Antoine, Helmer, Catherine, Pérès, Karine
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 01.11.2021; Elsevier Limited; Elsevier
• Subjects: Activities of Daily Living; Aged; Aged, 80 and over; Aging; Algorithms; Antidepressants; Antidiabetics; Cardiovascular diseases; Cohort analysis; Cohort Studies; Cohort study; Congestive heart failure; Coronary artery disease; Data Collection - methods; Data Collection - statistics & numerical data; Databases, Factual - statistics & numerical data; Dementia disorders; Dependency; Diabetes mellitus; Dopamine receptors; Epidemiology; Female; Frail Elderly - statistics & numerical data; France; Geriatric Assessment - methods; Geriatric Assessment - statistics & numerical data; Health insurance; Health insurance data; Heart diseases; Humans; Insurance; Insurance, Health - statistics & numerical data; Internal Medicine; Life Sciences; Lipids; Logistic Models; Male; Medical equipment; Older people; Pharmacoepidemiology; Population Surveillance - methods; Regression models; Santé publique et épidémiologie; France; Pharmacoepidemiology; Aged; Activities of daily living; Dependency; Health insurance data; Cohort study
• ISSN: 0895-4356; 1878-5921; 1878-5921
• DOI: 10.1016/j.jclinepi.2021.06.014

•Dependency is not available in the health insurance databases•An algorithm to identify ADL-dependent older adults from health insurance data•It includes medications, long-term diseases, medical devices and acts, age and sex•Estimated prevalence was 9.5% in 2010 first quarter, increasing with age•Dependent subjects were three times more likely to die than the non-dependent ones We aimed to develop an algorithm for the identification of basic Activities of Daily Living (ADL)-dependency in health insurance databases. We used the AMI (Aging Multidisciplinary Investigation) population-based cohort including both individual face-to-face assessment of ADL-dependency and merged health insurance data. The health insurance factors associated with ADL-dependency were identified using a LASSO logistic regression model in 1000 bootstrap samples. An external validation on a 1/97 representative sample of the French Health Insurance general population of Affiliates has been performed. Among 995 participants of the AMI cohort aged ≥ 65y, 114 (11.5%) were ADL-dependent according to neuropsychologists individual assessments. The final algorithm developed included: age, sex, four drug classes (dopaminergic antiparkinson drugs, antidepressants, antidiabetic agents, lipid modifying agents), three type of medical devices (medical bed, patient lifter, incontinence equipment), four medical acts (GP's consultations at home, daily and non-daily nursing at home, transport by ambulance) and four long-term diseases (stroke, heart failure, coronary heart disease, Alzheimer and other dementia). Applying this algorithm, the estimated prevalence of ADL-dependency was 12.3% in AMI and 9.5% in the validation sample. This study proposes a useful algorithm to identify ADL-dependency in the health insurance data.