Watson · The American journal of psychiatry 2021 · prospective cohort study · n=601

Incident Major Depressive Disorder Predicted by Three Measures of Insulin Resistance: A Dutch Cohort Study.

Cited 124 times in the scientific literature.

Level 3 - non-randomized controlled study

Prospective longitudinal cohort study

PubMed 34551583 · doi:10.1176/appi.ajp.2021.20101479 · record verified 2026-08-28

What was done

In the Netherlands Study of Depression and Anxiety (NESDA), 601 participants aged 18–65 without a lifetime history of depression or anxiety were followed prospectively for 9 years. Researchers assessed whether baseline surrogate measures of insulin resistance (triglyceride-to-HDL ratio, fasting plasma glucose, and waist circumference) and the emergence of these markers during the first 2 years predicted incident DSM-IV major depressive disorder using Cox proportional hazards models.

What was found

Over 9 years of follow-up, 14% of the cohort developed major depressive disorder. Incident depression was predicted by baseline triglyceride-to-HDL ratio (hazard ratio=1.89, 95% CI=1.15, 3.11), higher fasting plasma glucose levels (hazard ratio=1.37, 95% CI=1.05, 1.77), and higher waist circumference (hazard ratio=1.11, 95% CI=1.01, 1.21). Developing prediabetes in the first 2 years was also associated with incident depression (hazard ratio=2.66, 95% CI=1.13, 6.27), whereas new onset of high triglyceride-to-HDL ratio or high central adiposity (waist circumference ≥100 cm) in the first 2 years was not significantly associated.

Why it matters

This study shows that markers of metabolic dysregulation and insulin resistance precede the onset of major depression in adults with no psychiatric history. Metabolic profiling may assist in early identification and prevention strategies for individuals at risk of depression.

Limits

The study relied on surrogate markers of insulin resistance rather than direct physiological measures (such as hyperinsulinemic-euglycemic clamps). The cohort was restricted to Dutch adults aged 18–65, limiting generalizability to other populations, and the modest sample size (n=601) led to relatively wide confidence intervals for some risk estimates.

Cited by