Equal treatment but fewer labels: a paired comparison of anxiety and depression diagnosis and treatment in US adults

Abstract Background. Anxiety disorders are reported to be less likely to be diagnosed than depression, but most evidence compares rates across different people and different instruments. Whether the gap reflects differential labeling of the same patient is unknown. Methods. We used the US National Health Interview Survey (NHIS, 2019–2025) to identify 17,901 adults reporting weekly-or-more symptoms of both anxiety and depression, comparing within-person diagnosis and treatment rates. Analyses used survey-weighted marginal contrasts, McNemar’s test, conditional logistic regression, and TOST equivalence testing. Sensitivity analyses included E-value computation. Severity-matched comparisons used GAD-7 and PHQ-8 waves (2019, 2022, 2025; n = 67,233 concordant). Results. Within persons with both symptom types, anxiety was diagnosed 62.2% vs 73.8% for depression (− 11.7 pp, 95% CI − 12.5 to − 10.9; McNemar χ²=1434.9, p < 0.001); the conditional logit OR was 0.225 (0.207–0.245). Treatment rates were formally equivalent (difference 0.25 pp, 90% CI − 0.12 to 0.62; TOST p < 0.001 at ± 1 pp). The gap was severity-dependent (+ 0.3, − 1.3, − 7.3, − 6.1 pp across bands) and present in all three years. It was not explained by comorbidity, insurance, access, treatment measurement, or survey year; the E-value was 3.6 (typical confounders RR 1.2–2.0). Conclusions. Among US adults with equivalent symptom burdens, anxiety is labeled less often than depression while being treated at the same rate. This labeling asymmetry is severity-dependent, stable across survey years, and contributes to the population-level identification gap for anxiety.