Prediction of depressive symptoms at high age (80+) by psychological, biological and functional factors

Abstract

BackgroundLate-life depression (LLD) is highly prevalent, especially in people aged 80 years and older. We aimed to investigate predictors and their influence on depressive symptoms in LLD. MethodsWe analysed data from the NRW80+ study, a population-based cross-sectional study of individuals aged 80 years and older. Data from n = 926 cognitively unimpaired participants were included. We reduced 95 variables to 21 predictors of depressive symptoms by using a two-step cluster analysis (TSCA), which were assigned to one of four factors (function, values and lifestyle, autonomy and contentment, biological-somatic) according to a principal component analysis. A second TSCA with complete data sets (n = 879) was used to define clusters of participants. Using weighted mean composite scores (CS) for each factor group, binary logistic regression analyses were performed to predict depressive symptoms for each cluster and the total population. ResultsThe second TSCA yielded two clusters (cluster 1 (n = 688), cluster 2 (n = 191)). The proportion of participants with depressive symptoms was significantly higher in cluster 2 compared to cluster 1 (39 % vs. 15 %; OR = 3.6; 95 % CI 2.5–5.1; p < .001). Participants in cluster 2 were significantly older (mean age 88 vs. 85 years; p < .001), with a higher proportion of women (56 % vs. 46 %; OR = 1.5; 95 % CI 1.1–2.0; p = .016), had a higher BMI (p = .017), lower financial resources (OR = 2.3; 95 % CI 1.6–3.5; p < .001), lower educational level (OR = 1.8; 95 % CI 1.2–2.5; p = .002), higher proportion of single, separated or widowed participants (OR = 1.9; 95 % CI 1.3–2.6; p < .001) and a smaller mean social network (p = .044) compared to cluster 1. Binary logistic regression analyses showed that the weighted mean CS including the autonomy and contentment predictors explained the largest proportion of variance (22.8 %) for depressive symptoms in the total population (Nagelkerke's R2 = 0.228, p < .001) and in both clusters (cluster 1: Nagelkerke's R2 = 0.171, p < .001; cluster 2: Nagelkerke's R2 = 0.213, p < .001), respectively. LimitationsThe main limitations are the restriction to cognitively unimpaired individuals and the use of a self-rated questionnaire to assess depressive symptoms. ConclusionsPsychological factors such as autonomy and contentment are critical for the occurrence of depressive symptoms at higher age, independent of the functional and somatic status and may serve as specific targets for psychotherapy.