The effects of exercise on depressed individuals have been investigated extensively. Too often, the conclusions drawn from these studies have gone beyond what the actual data can defend. Consider a pair of meta-analyses of published studies from groups in Australia and the UK—a meta-analysis involves compiling results from other investigations.
The authors of the Australian study examined data from 97 studies that included over 128,000 participants. That sounds impressive; however, the devil is in the details that the popular media completely ignored. For example, only 11 of those studies were focused on depression.
The authors concluded that exercise had a medium effect on depression. It is impossible to know how a “medium” effect compares with drug therapy since the studies were not head-to-head comparisons. The study also reported that exercise benefited many other health conditions, including HIV or kidney disease, various mental disorders, and cancers. Furthermore, any level of exercise was effective! If this sounds too good to be true, it is.
One point often ignored is that the benefits of exercise described in many studies decreased over time. The authors also noted that patients who exercised less often each week reported greater benefits than patients who exercised more often each week. This is the exact opposite of the typical dose-benefit relationship that should be observed. When a treatment works well for many conditions, and the benefits wear off over time or do not show a dose-response relationship, you are witnessing the placebo effect. The problem is that it is impossible to control for the placebo effect in studies on physical activity; the control subjects are well aware that they are not exercising.
Many studies of the benefits of exercise often have numerous fatal flaws that undermine confidence in their results. Studies have varied widely in size, type of control group, methodological rigor, length of follow-up, and even the type of exercise modality. Randomized exercise trials have generally ranged in length from six weeks to four months and typically emphasized aerobic exercise, although some studies on resistance training have also been conducted. The many differences in study design have contributed to the current level of confusion and misunderstandings about the benefits of exercise on depression.
The Integrity of Studies
Although many trials have been conducted on adults with major depressive disorder, only a few used high-quality methodologies in which the treatment allocation was concealed. Many of the other studies failed to comply with the standard intention-to-treat analyses. This means that the final analysis included every subject assigned a randomized treatment. This type of analysis ignores noncompliance (the subjects exercised too much or not at all), protocol deviations (they performed the wrong exercise), withdrawal from the study, or any number of potential things that might have happened to the subjects after they had been assigned to their study groups.
The general problem: It is very difficult to convince human subjects to exercise consistently without introducing their own creative changes, which often undermines the integrity of the study. Many past studies’ results were difficult to analyze because they failed to include a control group in the design. Sometimes, the most important variable, the degree of depression experienced by the subjects, was assessed by someone not blinded to treatment. Thus, the studies were vulnerable to experimenter bias and are thus unreliable.
In studies that used a standard head-to-head comparison, exercise was no better or worse than standard cognitive behavioral therapy for treating depression—that is, just talking to someone was as effective as exercising. When researchers compared the effectiveness of exercise on adults who were given a standard antidepressant drug therapy, they found no significant benefit of any one approach.
The second meta-analysis from the UK examined 57 trials with 2189 participants and reported a modest benefit of exercise; unfortunately, most of the studies examined had no treatment or a control intervention. When only the seven trials (447 participants) with adequate controls were examined, the benefits of exercise were much smaller. When data from a group of nine trials (405 participants) that had a long‐term follow‐up were considered, the results provided very uncertain evidence about the effect of exercise on depressive symptoms. Finally, when compared with pharmacological treatment, exercise produced little to no effect on depressive symptoms.
The authors of the UK analysis also acknowledged multiple sources of bias. The problem with exercise as an intervention is that blinding of those receiving and those delivering the interventions is inherently impossible. They concluded that, based on a few small trials, exercise appears to be no more or less effective than psychological or pharmacological treatments. Overall, the benefits of physical exercise for depression are subtle but real. These analyses clearly demonstrate that although exercise is not better than drugs, it should be considered an important adjunct therapy, combined with talk therapy, to any treatment plan for depressed patients.