Search

Background: Predictor analyses of late-life depression can be used to identify variables associated with outcomes of treatments, and hence ways of tailoring specific treatments to patients. The aim of this review was to systematically identify, review and meta-analyse predictors of outcomes of any type of treatment for late-life depression.Methods: Pubmed, Embase, CINAHL, Web of Science and PsycINFO were searched for studies published up to December 2016. Primary and secondary studies reported treatment predictors from randomised controlled trials of any treatment for patients with major depressive disorder aged over 60 were included. Treatment outcomes included response, remission and change in depression score.Results: Sixty-seven studies met the inclusion criteria. Of 65 identified statistically significant predictors, only 7 were reported in at least 3 studies. Of these, 5 were included in meta-analyses, and only 3 were statistically significant. Most studies were rated as being of moderate to strong quality and satisfied key quality criteria for predictor analyses.Limitations: The searches were limited to randomised controlled trials and most of the included studies were secondary analyses.Conclusions: Baseline depression severity, co-morbid anxiety, executive dysfunction, current episode duration, early improvement, physical illnesses and age were reported as statistically significant predictors of treatment outcomes. Only the first three were significant in meta-analyses. Subgroup analyses showed differences in predictor effect between biological and psychosocial treatment. However, high heterogeneity and small study numbers suggest a cautious interpretation of results. These predictors were associated with various mechanisms including brain pathophysiology, perceived social support and proposed distinct types of depressive disorder. Further investigation of the clinical utility of these predictors is suggested. [ABSTRACT FROM AUTHOR]

The aim of this review was to identify and critically appraise predictors and moderators of outcomes of psychological and pharmacological treatments for late-life anxiety disorders. Their identification may guide the development of personalised treatments for older people with anxiety disorders. Web of Science, PsychINFO, CINAHL, Embase, and Pubmed were searched for studies published up to 12 May 2022. Randomised controlled trials and observational studies reporting treatment predictors and moderators were included. Participants with a diagnosis of any anxiety disorder who were aged over 60 years were included. Treatment outcomes included response, remission, and change in anxiety score. Thirteen studies met the inclusion criteria. Twenty-three out of 49 predictors or moderators assessed at post-treatment, and 14 out of 33 predictors or moderators assessed at follow-up were statistically significant. Only one predictor, baseline worry severity at post-treatment, was reported in at least three studies. Most studies were rated as having a low risk of bias in at least three areas and satisfied important quality criteria for predictor and moderator analyses. Samples were predominantly white, female and highly educated, and most studies were secondary analyses. There is evidence that baseline worry severity appears to predict treatment outcome in late-life anxiety disorders. However, this was only explored in psychological intervention studies and therefore its predictive ability in pharmacotherapy remains unknown. Future research should explore predictors and moderators in a range of anxiety disorders and design methodologically-strong and adequately-powered studies with the primary aim of assessing predictors of treatment outcomes. • Identifying predictors/moderators of treatment outcome may aid clinical decisions. • No previous systematic reviews have examined these in late-life anxiety disorders. • Thirteen studies were identified that assessed treatment predictors/moderators. • Only baseline anxiety severity predicted treatment outcome in ≥three studies. • Further adequately powered studies of treatment predictors/moderators are needed. [ABSTRACT FROM AUTHOR]