Predictors of change in depressive symptoms in older and multimorbid patients: a longitudinal analysis of the multicare cohort.

Background: Depression in older adults is becoming an increasing concern. As depressive symptoms change over time, it is important to understand the determinants of change in depressive symptoms. The aim of our study is to use a longitudinal study design to explore the predictors of change, remission and incident depression in older patients with multimorbidity.
Methods: Data from the MultiCare cohort study were used. The cohort studied 3,189 multimorbid general practice patients aged 65-85. Data were collected during personal interviews. Depressive symptoms were assessed using the Geriatric Depression Scale (GDS-15). Predictors of change in depressive symptoms were determined using multivariate linear regression, while multivariate logistic regression was used to analyze predictors of remission and incident depression. Models included depressive symptoms at baseline and follow-up, socio-demographics and data on health status and social support.
Results: Overall, 2,746 participants with complete follow-up data were analyzed. Mean age was 74.2 years, 59.2% were female, and 11.3% were classified as depressed at baseline. Burden of multimorbidity and social support were statistically significant predictors in all regression analyses. Further predictors of change in depressive symptoms were: income, pain, nursing grade, self-rated health and self-efficacy.
Limitations: The sample size for prediction of remission limited statistical certainty. Assessment of depressive symptoms using GDS-15 differs from routine clinical diagnoses of depression.
Conclusions: Predictors of change in depressive symptoms in older multimorbid patients are similar to those predicting remission and incident depression, and do not seem to differ significantly from other older patient populations with depressive symptoms.