ObjectivesNutrition risk is a key component of frailty and screening, and treatment of nutrition risk is part of frailty management. This study identified the determinants of a 3-year decline in nutrition risk (measured by SCREEN-8) for older adults stratified by risk status at baseline.MethodsSecondary data analysis of the comprehensive cohort sample of the Canadian Longitudinal Study on Aging (CLSA) (n = 5031) with complete data for covariates at baseline and 3-year follow-up. Using a conceptual model to define covariates, determinants of a change in nutrition risk score as measured by SCREEN-8 (lower score indicates greater risk) were identified for those not at risk at baseline and those at high risk at baseline using multivariable regression.ResultsModels stratified by baseline nutrition risk were significant. Notable factors associated with a decrease in SCREEN-8 for those not at risk at baseline were mental health diagnoses (− 0.83; CI [− 1.44, −0.22]), living alone at follow-up (− 1.98; CI [− 3.40, −0.56]), and lack of dental care at both timepoints (− 0.91; CI [− 1.62, −0.20]) and at follow-up only (− 1.32; CI [− 2.45, −0.19]). For those at high nutrition risk at baseline, decline in activities of daily living (− 2.56; CI [− 4.36, −0.77]) and low chair-rise scores (− 1.98; CI [− 3.33, − 0.63]) were associated with lower SCREEN-8 scores at follow-up.ConclusionDeterminants of change in SCREEN-8 scores are different for those with no risk and those who are already at high risk, suggesting targeted approaches are needed for screening and treatment of nutrition risk in primary care.