Thirteen RCTs (n approximately 1,264) were included in the review.
All 13 studies reported using random allocation but only six described the method of allocation concealment. Only 5 studies reported blinding of the outcome assessment. Four studies described adequately both allocation concealment and blinding. Intention-to-treat analysis was only explicitly stated in four.
Non-depressed elderly (5 studies): only one of the 5 studies reported a significant benefit of exercise; the exercise was aerobic and the benefit was only seen in men.
Depressed elderly (5 studies): aerobic exercise was reported to reduce self and therapist-reported depressive symptoms (1 study) and somatic symptoms (1 study). Benefits were reported in relation to the response rate for the Hamilton Rating Scale-D (1 study), responses in patients that had not responded to medication (1 study), and responses were shown to be comparable to medication (1 study). T'ai chi gave positive results in depressed or dysthymic out-patients (1 study).
Mixed elderly populations (3 studies): all 3 studies reported positive results in relation to depressive symptoms for exercise groups compared with controls.