Testing Running Against Antidepressants — The SMILE Trial and the 25 Years Since

Introduction — A Common Saying and a Rare Experiment

"Try exercising when you feel down." The advice is so common that it is easy to take lightly. But consider what kind of experiment it would take to actually test it, and the story changes. You would need to recruit patients with a real diagnosis of depression, randomly assign one group to exercise and the other to a proven antidepressant, and compare them months later with formal diagnostic tools. This experiment — no small matter in either cost or ethics review — has actually been done.

The ninth installment of Psychology, Straight from the Papers reads the SMILE study (Standard Medical Intervention and Long-term Exercise) by James Blumenthal and his team at Duke University, together with the 25 years of evidence that followed. One thing to make clear before we begin: this post is a paper walkthrough, not medical advice. Depression is a treatable condition, and treatment decisions should be made with a professional.

SMILE, 1999 — Design and Results

Design. 156 adults aged 50 and over, diagnosed with major depressive disorder, were randomly assigned to three groups.

• Exercise group: aerobic exercise 3 times a week, 30 minutes per session (treadmill or stationary bike at 70-85% of maximum heart rate), for 16 weeks
• Medication group: sertraline (Zoloft) — the standard SSRI antidepressant of the day
• Combination group: exercise + sertraline

Results at 16 weeks (including the 2000 report). All three groups improved substantially and were statistically indistinguishable. Remission rates hovered around 60% in all three groups. The medication group improved faster early on, but by week 16 the endpoints were the same.

The 10-month follow-up — the highlight of this study. The twist arrives in the follow-up results reported by Babyak and colleagues in 2000. Among patients who had reached remission at 16 weeks, relapse rates split by group.

The exercise group had the lowest relapse rate, and those who kept exercising on their own during the follow-up had a lower risk of relapse. The researchers cautiously proposed a sense of mastery — "I produced my own recovery" — as a candidate mechanism for the maintenance effect. In the language of self-efficacy, exercise prescribed not just symptom relief but a source of success experiences along with it.

Why Not to Take These Results at Face Value

Following the rules of this series, we look for the holes in the design before getting excited.

There is no placebo control group. The 1999 design had no placebo arm, so we cannot tell how much of the effect in all three treatments came from the mere fact of being treated. That is why the Blumenthal team ran a replication (SMILE-II) in 2007 with 202 participants. This time a placebo pill group was added, and exercise was split into supervised and home-based formats. The result: supervised exercise (remission rate 45%) and sertraline (47%) were still comparable and beat placebo (31%), but the gap was less dramatic than the impression the 1999 study left. The placebo remission rate of 31% is itself a famous reality of depression research.

Randomization is broken in the 10-month follow-up. Exercising during the follow-up period was a personal choice, so the reverse reading — people who recovered enough to keep exercising also relapsed less — cannot be ruled out. The dramatic 8% versus 38% table should be read as observational numbers, not experimental ones.

The sample is particular. These were volunteers aged 50 and over who were willing to exercise. The results cannot be generalized to every age group and severity.

2024 — A Synthesis of 218 Trials

The limits of a single study are filled in by accumulation. The network meta-analysis by Michael Noetel and colleagues, published in BMJ in 2024, pooled 218 randomized trials of exercise for adult depression (more than 14,000 participants). The key points are these.

• Walking and jogging, yoga, strength training, and dance all showed roughly medium-sized reductions in symptoms versus control groups — a level comparable, across several comparisons, to the average effect of antidepressants alone.
• Effects tended to grow with intensity, and with how supervised the program was.
• That said, the risk of bias in many included trials was not low, so the certainty of the evidence was rated low to moderate. The authors conclude not "replace" but include exercise as one of the core treatments.

Major clinical guidelines today sit in much the same place. For mild to moderate depression, exercise is one of the evidence-based first-line options; from moderate severity up, it is a combination partner alongside psychotherapy and medication. "With exercise you do not need medication" is a sentence found nowhere in this literature.

What We Take Away

1. The antidepressant effect of exercise is real, with clinical-trial-grade evidence. It is not a magnitude to be waved off with light phrases like "a change of mood." Aerobic exercise 3 times a week, 30 minutes at a time, was the actual prescription in SMILE — and it overlaps exactly with the Zone 2 protocol introduced in the body comes first.

2. Maintenance is the crux, and that makes design the crux. The real message of SMILE is that exercise gives a maintenance effect to the people who keep doing it. What keeps you doing it is not willpower but systems and friction design — a fixed time slot, gear laid out in advance, someone to run with.

3. When you are ill, do not decide alone. Every sentence in this post carries the same footnote. If you suspect depression at a diagnosable level, the first action is not buying running shoes but booking an appointment with a professional. Exercise is an excellent treatment partner, but it cannot be the attending physician of your treatment plan. If listlessness has lasted more than 2 weeks, read how to read that signal as well.

Guide to Reading the Originals

• Original trial: Blumenthal, J. A., et al. (1999). Effects of exercise training on older patients with major depression. Archives of Internal Medicine, 159(19), 2349-2356.
• 10-month follow-up: Babyak, M., et al. (2000). Exercise treatment for major depression: Maintenance of therapeutic benefit at 10 months. Psychosomatic Medicine, 62(5), 633-638.
• Placebo-controlled replication: Blumenthal, J. A., et al. (2007). Exercise and pharmacotherapy in the treatment of major depressive disorder. Psychosomatic Medicine, 69(7), 587-596.
• Latest synthesis: Noetel, M., et al. (2024). Effect of exercise for depression: Systematic review and network meta-analysis of randomised controlled trials. BMJ, 384, e075847.

Reading tip: for the 1999 and 2007 papers, comparing Table 2 (remission rates by group) alone gets you the core. In the 2024 BMJ paper, the forest plot by type of exercise is the centerpiece — a good figure for building the habit of reading the width of the confidence intervals too. The final installment turns to the smallest and warmest intervention of all, the original gratitude journal paper.