A Better Result for Depression Does Not Cover Every Outcome

The aerobic-exercise synthesis becomes more useful when its outcomes are separated. This chapter examines what the pooled depression result, its subgroup patterns, the small anxiety evidence base, and the broader quality-of-life result each support. The aim is not to decide whether exercise is “good” in general, but to match the strength and meaning of a claim to the outcome that was actually measured—and to show why no single volume can optimize every postpartum goal.

Depression: the average effect is conditional

The review’s most developed result concerns depressive symptoms. Sixteen studies contributed depression outcomes, and the random-effects meta-analysis found that aerobic exercise reduced symptom scores relative to inactive or minimal controls. The standardized mean difference was −0.37, with a 95% confidence interval from −0.60 to −0.14. Because lower depression scores are better, the negative direction favors exercise. The interval does not include zero in this analysis, so the synthesis provides evidence of an average difference between groups.

That sentence is already more precise than “exercise cures postpartum depression.” The measure was generally symptom severity on a validated scale, not necessarily a diagnostic assessment. The estimate describes how the average exercise group differed from the average comparison group under the included program conditions. It does not mean that every participant improved, that each improvement was clinically important, or that exercise alone is sufficient for someone with a diagnosed or persistent depressive disorder. Nor does it compare exercise directly with psychotherapy or medication.

The size of the average effect is also not the same as the size of an individual response. One participant may improve substantially, another may show little change, and another may stop attending before the program ends. A group-level standardized difference compresses that variation into one estimate. It is valuable for judging whether a pattern is detectable across trials, but it cannot tell a person which side of the average they will occupy. This is why the headline result needs the conditions around it.

The subgroup findings supply some of those conditions. Larger improvements appeared in studies involving women with baseline depression, with an SMD of −0.64 and a 95% confidence interval from −1.12 to −0.16. By contrast, no statistically significant improvement was observed in the subgroup without baseline depression. One plausible explanation is room for improvement: people who begin with more symptoms have more symptom burden that a program could reduce, while those who begin near the low end of a scale have less room for a large downward change. That explanation is reasonable, but it is not the same as proving that baseline status alone determines who will benefit.

The difference between a treatment sample and a preventive sample changes the question as well. A prevention study may ask whether exercise helps people at risk avoid a later increase in symptoms. A treatment study may ask whether participants already experiencing depression improve. A small change in a low-risk sample and a large change in a symptomatic sample should not be judged by the same expectation.