To evaluate the validity and applicability of the Mini International Neuropsychiatric Interview (MINI) used by family medicine residents in primary health care (PHC) in Brazil.

Training for administrating the MINI was given as part of a broad psychiatry education program. Interviews were held with 120 PHC patients who were at least 15 years old. MINI was administered by 25 resident physicians, while the Structured Clinical Interview for Diagnosis (SCID) was administered by a psychiatrist blind to patients' results on the MINI, and the diagnoses on both interviews were compared. The resident physicians answered questions on the applicability of the MINI.

Concordance levels for any mental disorder, the broader current diagnostic categories and the most common specific diagnoses were analyzed. Kappa coefficients ranged between 0.65 and 0.85; sensitivity, between 0.75 and 0.92; specificity, between 0.90 and 0.99; positive predictive values (PPV), between 0.60 and 0.86; negative predictive values (NPV), between 0.92 and 0.99; and accuracy, between 0.88 and 0.98. The resident physicians considered MINI comprehensibility and clinical relevance satisfactory.

These good psychometric results in a real-world setting may be related to a special training program, which is more frequent, intensive and diversified. In these conditions, the MINI is a useful tool for general practitioners.