PHQ-9 and GAD-7 in Latino primary care: what the evidence says

The PHQ-9 and GAD-7 are the two most widely used mental-health screeners in US primary care. Both have Spanish versions with published validation in Latino samples — but the confidence intervals around cut-offs are wider than most clinicians assume.

PHQ-9: what the Spanish validation shows

Across studies in Mexican, Puerto Rican, Colombian, and US Hispanic samples, the Spanish PHQ-9 replicates a single-factor structure and internal consistency ≥0.80. Sensitivity at the ≥10 cut-off ranges roughly 74–88%, specificity 88–91%. In practical terms: a score of 10 or higher is a strong signal, but a score of 8–9 is not "safe" — it warrants a clinical interview, especially when item 9 (harm ideation) is endorsed at any level.

GAD-7: fewer studies, consistent signal

Fewer large-sample Spanish validations exist for GAD-7 than for PHQ-9, but existing studies (including Argentine, Spanish, and US Hispanic samples) support the single-factor model and cut-offs of 5 (mild), 10 (moderate), and 15 (severe). The GAD-7 also performs reasonably as a general anxiety screener — a positive result should trigger a differential (panic, social anxiety, PTSD, health anxiety), not a direct diagnosis of GAD.

Three traps to avoid

1. Somatic-item bias. Latino patients may endorse somatic depression symptoms (fatigue, appetite, sleep) at higher rates than affective ones. Do not interpret a PHQ-9 score dominated by somatic items as "less severe" depression — the total is what predicts outcomes.
2. Language mismatch on repeat testing. Retest in the same language the patient completed the baseline. Switching between English and Spanish between visits inflates change scores.
3. Ignoring item 9. Any non-zero response on PHQ-9 item 9 requires a same-visit risk assessment, regardless of total score. This is a screening obligation, not a clinical judgment call.

Documenting the result

Record the language version administered, the total score, item-9 endorsement, and the interpretive category (minimal / mild / moderate / moderately severe / severe). A one-line clinical impression tying the score to the presenting concern makes the record defensible and useful at follow-up.