Elaboration of a new synovial predictive score of septic origin for acute arthritis on the native joint (RESAS).

OBJECTIVE

To establish a new predictive score for the diagnosis of septic arthritis (SA) according to different synovial fluid (SF) variables.

METHODS

First, we analysed the different clinical, biological and SF variables associated with the diagnosis of SA (according to the Newman's criteria) in a monocentric cohort of acute arthritis (<30 days) (n = 233) (SYNOLACTATE cohort). A new score predictive of SA (RESAS) was created using the independent discriminant variables after multivariate analysis. A value was attributed to each variable of the score according to the weighting based on their likelihood ratio for the diagnosis of SA. RESAS performance was then tested on the first cohort (internal validation) and then checked on a second independent cohort (n = 70) (external validation).

RESULTS

After multivariate analysis, four independent variables of the SF were included for RESAS: (i) purulent SF or white blood cells count ≥70 000/mm3; (ii) absence/presence of crystals; (iii) lactate; and (iv) glucose synovial level. RESAS ranged between -4 and +13 points. The performance of RESAS to predicted SA was excellent with area under the curve (AUC)=0.928 (0.877-0.980) in internal validation and AUC=0.986 (0.962-1.00) in external validation. For a RESAS threshold ≥+4, SA was diagnosed with Se=56.0% (0.371-0.733), Sp=98.1% (0.952-0.993), LR+=29.1 (10.4-81.6) in the first cohort and with Se=91.7% (0.646-0.985), Sp=98.3% (0.909-0.997), LR+=53.2 (7.56-373) in the second cohort.

CONCLUSION

RESAS is a new composite score of four SF variables with excellent performance to predicted SA in acute arthritis population.