top of page

Psychiatry's Worst Nightmare: Lifestyle Medicine for Depression

December 06, 2019.

Cardiovascular diseases, diabetes-2, and obesity qualify as lifestyle diseases because they are primarily caused by pernicious habits: physical inactivity, fast-food diet, poor sleep quality, and smoking. Consequently, the medical specialties that deal with those chronic diseases recommend lifestyle modification as the overarching therapy.

For example, as the first line of treatment for heart disease, the American College of Cardiology (ACC) recommends Lifestyle Management (LM) over the ballyhooed statin drugs. That's because statins' effectiveness at preventing heart attacks is significant only for "at risk" adults, e.g., over 40 with high LDL. For the remaining 73% of the adult population, exercise and a Mediterranean-style diet are the best path to heart health. And for those recovering from a recent heart attack, lifestyle modification cuts the risk of recurrence just as much as statins (which is why both are prescribed).

Also, LM is not beset by the downside of drug therapy, i.e., adverse side effects. To the contrary, lifestyle management is a boon to public health because the side effects of improved diet and exercise are overwhelmingly positive. They include a reduced risk of death, diabetes, some cancers, dementia, and even major depression.

Yes, even depression. The accumulation of scientific evidence shows that physical activity, sleep patterns and diet determine the risk of depression just as strongly as they do for heart disease and diabetes. That evidence is summarized in the appendix below.

Psychiatry's profitable ruse: tinkering with brain chemistry is the proper way to treat a lifestyle disease.

Yet the American Psychiatric Association (APA) does not recognize depression as a lifestyle disease; its list of risk factors conspicuously omits lifestyle and ranks "biochemistry" first. So it's no surprise that the APA's treatment guidelines for major depression enshrine antidepressants as the first line of treatment even for mild to moderate cases. A close second is psychotherapy, especially for the purpose of augmenting drug treatment. If those two don't work, the APA recommends zapping the brain with electromagnetic energy. But nowhere are lifestyle factors mentioned as legitimate targets for medical intervention; rather, those matters are relegated to the category of "self-help and coping."

Psychiatry's rationale for targeting brain chemistry instead of lifestyle is patently false.

The APA and BigPharma insist that treatment with modern antidepressants is the most effective option at hand. As evidence, they marshal the gold standard of scientific research: Randomized Control Trials (RCTs) that compare antidepressant treatment with a placebo. In 2018, 522 of these RCTs were summarized in a colossal meta-analysis (henceforth, META). Immediately, the champions of pharmacotherapy boasted that the results of META were definitive proof that antidepressants are truly effective. News releases echoed "the drugs really work," no doubt about it, case closed.

Hogwash! The only thing META really proved was that the observed difference in depression scores between treatment and placebo could not have been produced by random error.

This leaves wide open the possibility that the difference was produced by systematic bias, and META's authors admit as much. META's own assessment reported that only 18% of the original trials qualified as "low risk" of bias. Also, the assessment was confined to what the original manuscripts reveal about trial design and procedures.

Not addressed, for example, is the fact that the use of inert placebos in antidepressant trials made them extremely susceptible to unblinding. (Blinding means that all participants in a trial remain ignorant of who was assigned to which group, treatment or placebo). The problem is that antidepressant drugs have noticeable side effects, but inert placebos don't. So, members of the treatment group are alerted to their group assignment, and the blind is broken. Since the sugar-pill group is not equally alerted, a larger percentage of the treatment group believe they are getting the "real thing." And since believers are more likely to report improvement, the trial's results end up biased in favor of antidepressants.

The obvious safeguard against this kind of unblinding is the use of an active placebo, one that also has "noticeable side effects." But none of the 522 RCTs bothered to use an active placebo. Perhaps it was infeasible or too costly to do so. Regardless of the reasons, this overt failure to equalize the treatment and control groups is the definition of a failed experiment. Those 522 RCTs are not quite the gold standard they claim to be.

Given the precariousness of blinding, you would think that astute researchers would routinely test for blinding as part of the study protocol; and that publishers would favor manuscripts that report such tests. (For example, a test of patients' ability to guess their group assignment during the trial). Sadly, only three out of 522 studies actually published a test for blinding.

Even worse is META's failure to acknowledge the most scandalous source of bias: the lucre BigPharma showered upon the psychiatric establishment: researchers, "thought leaders," ghost writers, and APA officials. When the META team requested supplemental information from the authors of the original RCTs, it could have asked them about their financial ties to BigPharma; but it didn't. Perhaps they were unaware of the empirical evidence that even medical demigods can be influenced by gifts and honorariums. After all, the authors of META are themselves psychiatry insiders.

How much difference do antidepressants really make, according to META?

On a standardized scale, the effect size of antidepressants is -0.3*, which translates into the difference shown on the graph. The average improvement in depression scores was about 15 points for treatment groups and 13 points for placebo groups, a whopping 2-point difference. The problem is, a 2-point difference is too small to be clinically significant. In other words, 2 points has no practical therapeutic significance for either the patient or the physician.

Furthermore, at least part of that difference was produced by systematic bias. So discounting for bias, the real effectiveness of antidepressants is less than 2-points, somewhere between clinically insignificant and zero.

antidepressant treatment vs placebo

The dismal long-run performance of antidepressant medication.

The foregoing summary only applies to clinical trials of short duration. Today, 25% of antidepressant users have been taking them for at least 10 years. When the effectiveness and side-effects of long term treatment are considered, antidepressants appear to do more harm than good. This includes the current generation of Serotonin Re-uptake Inhibitors (SRIs) introduced in the late 1980s.

In APA guidelines, long term treatment is called "continuation" or "maintenance" therapy. In short-term clinical trials, about 47% of patients treated with SRIs experience "remission" (few or no depressive symptoms). To prevent these remitters from relapsing into depression later on, the APA recommends continuation with SRIs. After all, the results of extended RCTs show that SRI remitters who are switched to a sugar pill relapse at a rate of 45%, compared to only 22% of those who continue taking SRIs.

This justification for continuation ignores an inconvenient fact: in the initial phase of those extended RCTs about one-third of the patients assigned to a placebo group also remitted; but their subsequent rate of relapse was only 27%, far lower than the 45% outcome for SRI remitters. In other words, patients who remit while taking SRIs, and then stop, are much more likely to relapse than than remitters who never took the drug in the first place. This suggests that SRI users become dependent on the drug, perhaps through a neurological adjustment of "oppositional tolerance" or "disruption of evolved adaptations." Eventually, the building tolerance manifests as "SRI poop-out" and "treatment resistant depression."

Aside from RCTs, proponents of continuation therapy point to long term cohort studies that report a lower risk of recurrence among patients who stick to the medication. But these older studies have been discredited because of a common design flaw - "immortal and neglected time bias." The newer cohort studies find just the opposite, that SRI continuation results in worse outcomes; it increases the "chronicity" of the disease.

The final nail in the coffin is the current safety profile of SRIs. The updated profile is much more alarming than historical versions because it captures the long term side effects of SRIs. They include weight gain, sleep disturbance, apathy and a high rate of sexual dysfunction (which is why SRIs have been called "anti-aphrodisiacs.)" On top of that, recent cohort studies find that long term use of SRIs, after controlling for confounding variables, increases the risks of falls, fractures, dementia and death. It's no wonder that APA brochures for patients fail to mention side effects.

The real reason for psychiatry's antipathy to Lifestyle Management is, of course, financial self interest.

The pretenders to the throne of medicine rely on a highly lucrative business model based on the myth of neurochemical imbalance. Compared to the other options in psychiatry's sparse medical toolkit, pharmacotherapy is a gravy train. Indeed, SRIs' lack of efficacy is a goldmine in disguise. The 55% of patients who don't respond to initial treatment are likely to have their medications adjusted or supplemented again and again and again, for which the doctor can charge at least $200/hr (Medicare rate, billing code 99214). On top of that is the steady stream of revenue from maintenance therapy for patients who can't get off the drugs without relapsing (often a misdiagnosis of withdrawal symptoms). Today, 7% of American adults have been stuck on antidepressants for at least 5 years. That's a gravy train.

The other options in the psychiatrists toolkit are not nearly as lucrative. The price of psychotherapy is bid downward by competition from psychologists, counselors, and social workers ($118/hr compared to $200 for medication management). Brain-zapping with TMS or ECT are treatments of last-resort and are still too controversial to be milked as maintenance therapy.

The prospect of treating depression as a lifestyle disease is even more threatening to their bottom line. Not only is it a hard way to make a buck from insurance billing, but it would fundamentally undermine the demand for lucrative drug treatment. As an effective preventive, LM obviates the need for any treatment. And as a treatment, it accomplishes the goal of continuation therapy without the risk of lifelong dependence on chemicals with detrimental side effects.

Moreover, as long as LM remains marginalized, the drug therapy business is enhanced by the licensing effect. Cardiology offers the perfect example: "I don't care what I eat, I'm on statins." Likewise, easy to swallow SRIs (and accompanying sales pitch) function as a license to wallow in a pro-depressive lifestyle. But unlike Psychiatry, the Cardiology profession tries to counteract the licensing effect by promoting LM and its central message: for optimal cardiovascular health statins cannot replace diet and exercise.

Psychiatry on the other hand exploits the licensing effect. Patients beguiled by "powerful medication" are more likely to carry on habits that foster depression, such as inactivity, irregular anti-circadian sleep patterns, and a diet that promotes obesity. This perpetuates illness and enhances the demand for medical treatment which, of course, is an integral part of psychiatry's business model.

How long can American Psychiatry sustain this scam?

British Psychiatry is beginning to surrender to the evidence. The latest NICE guidelines do not countenance drugs for treatment of mild and moderate depression, but instead recommend a tailored program of - you guessed it - physical activity. A recent article in BMC Psychiatry has laid out the case that "lifestyle modification should be a routine part of treatment and preventative efforts," and suggested practical steps for doing so.

If America's psychiatric establishment were intellectually honest and not thoroughly corrupted by BigPharma, they would be following that advice. The knee-jerk excuse that patients won't follow lifestyle prescriptions is a red-herring. After all, noncompliance with SRI treatment is well over 50% after 12 weeks. The evidence shows that adherence to physical exercise prescriptions is actually a little better than that.


Appendix: The Evidentiary Basis of Lifestyle Medicine

In 2014 an article in BMC Psychiatry laid out the case for treating depression as a lifestyle disease. This appendix fortifies the case with updated evidence and details not covered in that article.

Physical Activity - Exercise RCTs

Even psychiatry pays lip service to the preventive potential of physical activity, sleep, and prudent ingestion. But to demonstrate that these can compete with SRIs as a treatment requires randomized control trials. Only exercise has been extensively evaluated in experimental trials. Recent reviews of those RCTs find that exercise reduces symptoms by -0.78 standard units*. Since the comparable figure for SRIs is -0.3, it appears that exercise is 2.6 times more effective than drug therapy.

Actually, the superiority factor is greater than 2.6. The reason is the different makeup of the two sets of RCTs. For example, the -0.3 effect-size for SRIs is based entirely on comparisons with inert placebos. But the controls employed in exercise RCTs are rarely inert. Instead, they compare exercise to an "active intervention" such as medication, meditation, occupational therapy, education sessions, CBT, low intensity exercise, relaxation/massage, and so on. The evidence suggests that active interventions are about -0.2 points better at reducing symptoms than an inert placebo. So, a trial that compares exercise to an active intervention is likely to understate the effect of exercise by -0.2 points. Another study found that the type of controls used in exercise trials produced an improvement in symptoms twice as large as the placebos in SRI trials. So, exercise has to jump a hurdle twice as high to demonstrate effectiveness.

The final flaw is that exercise RCTs disregard the law of diminishing marginal returns. Experimental exercise programs represent additional physical activity imposed on top of the subjects' baseline energy expenditure in everyday life. That baseline level is many times greater than the extra exercise. According to diminishing returns, extra exercise should have less effect on people who are already physically active, and more effect on those who are sedentary. This explains why the effects of exercise are larger for folks older than 60 (and often retired) than for younger (and often working) adults. This flaw also applies to cohort studies discussed below.

Physical Activity - Prospective Longitudinal Studies

A review of 49 cohort studies found a rather small difference in depression risk between the highest and lowest levels of physical activity: only -17%. That's because those studies are predisposed by design to understate the effect of PA. Instead of measuring the total expenditure of physical energy, they only solicit data on the smallest portion: Leisure Time Physical Activity (LTPA). Overlooked is the dominant component: Work Time Physical Activity (WTPA). Not surprisingly, the correlation between the two is negative: men who burn lots of calories at work are the least likely to exercise at home. And even when they do, the law of diminishing returns applies: workers who lay bricks all day realize little additional benefit from an evening jog. So, as long as a cohort includes even moderately active workers, a study of LTPA will under-estimate the effect of physical activity. Also, cohort studies understate the true effect of PA because all 49 of them employ a noisy measure of LTPA. That is, they measure LTPA through self-reports which are known to correlate very weakly with objective measures of exercise.

Fortunately, those shortcoming were remedied by a huge cohort study (n=91,000) published in 2019. It succeeds in measuring total physical activity objectively, by means of accelerometers worn by subjects for at least 72 hours. The findings** imply that those who score in the top tier of total energy expenditure per week are 50% less likely to be subsequently diagnosed with major depression than the bottom tier - an effect 3 times greater that that of the aforementioned 49 studies.

Sleep Quality

According to psychiatry's sacred diagnostic manual (DSM-5), sleep problems are a key symptom of major depression. However, the evidence suggests the reverse: that depression is a symptom of chronic sleep disturbance, not the other way around. So its not surprising that eminent insiders have attacked DSM-5 as a "disingenuous categorical system" that is scientifically meaningless.

Sleep duration is often the sole focus of longitudinal studies. These studies find that persistently short sleepers (<=6 hrs) are, on average, 31% more likely to become "depressed." Unfortunately, sleep duration by itself is a lame indicator of sleep quality. That's because most studies rely on self-reports of duration, a measure that correlates weakly with physiologic indicators of sleep time. Moreover, too much sleep (>9 hrs) is just as predictive of depression as too little.

However, when cohort studies measure sleep quality on more dimensions that just duration, such as sleep disturbance and sleep efficiency, the reported effects are much larger. A summary of those studies finds that poor sleep quality more than doubles the risk of subsequent depression. This is confirmed by a recent high quality study of 1215 medical interns. The combination of disturbed sleep and short-sleep at baseline more than doubled the risk of depression three months later.

Drug Use.

A popular narrative is that depression and substance abuse coexist because depression sufferers use illicit drugs to self-medicate. But studies of cocaine and intravenous drug users find that major depressive episodes do not predict subsequent drug abuse. Rather, the evidence shows that the predominant causal arrow runs from drug abuse to depression.

The evidence also suggests that drug use is a significant cause of depression only when it morphs into a lifestyle of drug dependency and addiction. Alcohol use disorder doubles the risk of depression, but the rate of major depression among meth, heroin, and cocaine addicts is 4 to 7 times that of the general population. This suggests that the strong correlation between cigarette smoking and depression is spurious: it arises from the fact that 85% of drug addicts smoke and smokers are 16 times more likely to become drug addicts.

Diet - Randomized Control Trials

The problem with diet RCTs is that they run the gamut of dietary schemes: low fat, weight loss, Mediterranean, and so on. A statistical summary of these studies (n=16) found that dietary treatments outperformed the placebos to the same extent as SRIs, a meager -0.3 points*. However, two of those trials prescribed a Mediterranean style diet (MSD), and they reported a much larger effect on major depression: 2.5 and 4 times that of SRIs.

Diet - Prospective Longitudinal Studies

Cohort studies are similarly bullish on the MSDs. A statistical summary of those studies reported that MDS cuts the risk of subsequent depression by 33%. However, the SUN study is especially compelling for three reasons:

  • Its huge sample of 15,000 depression-free subjects enabled it to measure the subsequent effect of diet on actual clinical depression, not just an arbitrary threshold on a symptom questionnaire.

  • The subjects' food choices were scored on three variants of a healthy diet including MSD. The three are similar in nutritional composition but different in the precision of their scoring method. It turns out, the more refined the scoring, the greater the reduction in depression risk: the crudely measured MSD cut the risk by 25% while the intricate Healthy Eating Index delivered a 33% reduction.

  • This study is one of the few to control for other lifestyle factors. So, the reported effects of diet are independent of subjects' level of physical activity.


* The standard measure of effect size cited here is Cohen's d or Standard Mean Difference.

** Since total energy expenditure is measured on a continuous scale in the 2019 study, I converted its findings to the discrete "high-low" measures used the 2018 review.



bottom of page