Reproducibility of capillaroscopic classifications of systemic sclerosis: results from the SCLEROCAP study

Objectives Subgroups of capillaroscopic scleroderma landscape have been correlated with stages of SSc: two groups for Maricq's classification (slow and active), and three for Cutolo's classification (early, active and late). We report inter- and intra-observer agreement for these classifications as a preliminary step in the multicentre prospective SCLEROCAP study, which aims to assess the classification and single capillaroscopic items as prognostic tools for SSc. Methods SCLEROCAP included 385 patients. Agreement was studied in the first 100 patients, who were independently rated twice by two observers, blind to patients' characteristics; 30 of the patients were rated once by six observers. After consensus meetings, these ratings were held again. Kappa and intraclass correlation coefficients were used to assess agreement. Results Interobserver agreement on 100 patients was moderate for Maricq and Cutolo classifications [κ 0.47 (0.28, 0.66) and 0.49 (0.33, 0.65), respectively], and became substantial after consensus meetings [0.64 (0.50, 0.77) and 0.69 (0.56, 0.81)]. Intra-observer agreement between two observers was moderate to substantial: κ 0.54 (0.33, 0.75) and 0.70 (0.57, 0.83) for Maricq's classification; 0.57 (0.38, 0.77) and 0.76 (0.65, 0.87) for Cutolo's. Thirty patients were rated once by each of six observers, and agreement was moderate to substantial: κ 0.57 ± 0.10 (Maricq) and 0.61 ± 0.12 (Cutolo). Agreement was substantial for bushy, giant capillaries and microhaemorrhages, moderate for capillary density and low for oedema, disorganization and avascular areas. Conclusion The moderate reproducibility of Maricq and Cutolo classifications might hamper their prognostic value in SSc patients. Consensus meetings improve reliability, a prerequisite for better prognostic performances. A focus on giant capillaries, haemorrhages and capillary density might be more reliable.