Is it primary neuropsychiatric systemic lupus erythematosus? Performance of existing attribution models using physician judgment as the gold standard.

OBJECTIVES: Models for the attribution of neuropsychiatric manifestations to systemic lupus erythematosus (NPSLE) that incorporate timing and type of manifestation, exclusion/confounding or favouring factors have been proposed. We tested their diagnostic performance against expert physician judgment.

METHODS: SLE patients with neuropsychiatric manifestations were identified through retrospective chart review. Manifestations were classified according to physician judgment as attributed to SLE, not attributed or uncertain. Results were compared against the Systemic Lupus International Collaborating Clinics (SLICC) attribution models A and B, and one introduced by the Italian Study Group on NPSLE.

RESULTS: 191 patients experienced a total 242 neuropsychiatric manifestations, 136 of which were attributed to SLE according to physician. Both SLICC models showed high specificity (96.2% and 79.2% for model A and B, respectively) but low sensitivity (22.8% and 34.6%, respectively) against physician judgment. Exclusion of cases of headache, anxiety disorders, mild mood and cognitive disorders and polyneuropathy without electrophysiologic confirmation led to modest increases in sensitivity (27.7% and 42.0% for SLICC models A and B, respectively) and reductions in specificity (94.8% and 65.5%, respectively). The Italian Group model showed good accuracy in NPSLE attribution with an area under the curve of the receiver operating characteristics analysis of 0.862; values ≥7 showed the best combination of sensitivity and specificity (82.4% and 82.9%, respectively).

CONCLUSIONS: Attribution models can be useful in NPSLE diagnosis in routine clinical practice and their performance is superior in major neuropsychiatric manifestations. The Italian Study Group model is accurate, with values ≥7 showing the best combination of sensitivity and specificity.