What If the Depression Isn’t Why We Drink?

One of the most persistent stories therapists tell about alcoholism is also one of the most psychologically satisfying: people drink because they are depressed, anxious, traumatized, lonely, or otherwise suffering. Alcohol becomes the medication. Treat the underlying emotional disorder, the reasoning goes, and the need for alcohol should diminish.

Sometimes that story is true. But some of the best longitudinal research we have suggests we may frequently have the sequence backward — and newer research adds a variable the old story missed entirely: metabolic dysfunction.

The Harvard Study of Adult Development gave us the first serious challenge to the self-medication story. Beginning in the late 1930s, researchers followed 268 Harvard undergraduates and 456 socially disadvantaged young men from Boston, with psychiatrist George Vaillant eventually examining the natural history of alcoholism across decades.

Depression was certainly associated with alcoholism — severe depression was about five times more common among alcoholic men. But prevalence is not causation. Among the 14 Harvard men who experienced both alcoholism and major depression, only four developed depression first. Similar results held for generalized anxiety, and sustained abstinence frequently reduced depressive symptoms. Vaillant's conclusion was provocative: in many patients, depression and anxiety appeared not as the illnesses from which alcoholism emerged, but as conditions that developed after alcoholism was established.

That conclusion should not be universalized. Vaillant's cohorts were male, overwhelmingly white, drawn from socioeconomic extremes. Later epidemiology showed bidirectionality: alcohol dependence increased subsequent risk of major depression, while depression — particularly among women — could also increase subsequent risk for alcohol dependence. The modern picture is more clinically useful than either extreme: comorbidity does not tell us which disorder came first.

Then metabolism entered the picture. A 2026 study from the National Institute on Alcohol Abuse and Alcoholism examined 1,220 participants in an NIH natural-history protocol, dividing them into healthy controls, metabolic dysfunction alone, alcohol use disorder alone, and both — the last group termed "metAUD." That group showed the greatest abnormalities in liver injury, inflammation, and fibrosis markers. More importantly for clinicians, they carried the greatest psychiatric burden: within the AUD population, increasing metabolic dysfunction tracked with greater anxiety-disorder prevalence and worsening biological markers.

The study was observational — it does not prove metabolic dysfunction causes the psychiatric symptoms. But the clustering is hard to dismiss. The Maastricht Study found that higher fasting glucose, post-load glucose, and HbA1c predicted subsequent depressive symptoms over four years. What emerges is not a neat causal line but a system: alcohol disrupts sleep, glucose regulation, hepatic function, inflammatory signaling, nutrition, autonomic regulation, and brain function. Metabolic dysfunction independently hits many of the same systems. Depression and anxiety exist inside this biological environment rather than floating above it as purely psychological phenomena.

The clinical question therefore changes. Instead of asking only "What emotional pain is this person treating with alcohol?" we might also ask: "What has chronic alcohol exposure done to the organism in which this anxiety and depression are now occurring?"

There is a second lesson in Vaillant's data, about what actually helps. Across decades, formal treatment did not consistently predict long-term recovery. Alcoholics Anonymous was the conspicuous exception — at least as effective as clinic treatment for initiating stable abstinence, and particularly important for maintaining it. A 2020 Cochrane review of 27 studies and 10,000+ participants later confirmed that Twelve-Step Facilitation produced higher continuous abstinence rates than several established interventions, including CBT. The point is not that everyone needs AA. It is that alcoholism behaves like a chronic disorder, and AA provides something psychotherapy often cannot: an enduring behavioral structure organized around not drinking.

This exposes a characteristic clinical error: premature psychological explanation. A person drinks heavily and reports anxiety, shame, loneliness, broken relationships, low motivation, depression. Each may have a genuine psychological history — and each is also entirely compatible with the physiological consequences of prolonged alcohol misuse. If we explain the drinking through the depression, we may be using a downstream symptom to explain the process producing it. A person can drink because alcohol temporarily regulates an intolerable state, and simultaneously become more anxious and depressed because continued drinking disrupts the systems regulating mood. Both can be true.

For clinicians, the formulation has to widen: chronology matters, metabolic health matters, sleep matters, and the response of psychiatric symptoms to sustained abstinence matters. We should be cautious about assuming every depressed alcoholic drinks because he is depressed. Sometimes the more revealing question is the reverse: how much of the person we are treating has been shaped by the alcohol itself? That is not reducing psychotherapy to biology. It is expanding what psychological formulation means.