Objective: To investigate the inter- and intra-observer reliability of both qualitative and quantitative parameters used in the assessment of nail-fold capillaroscopy images.
Methods: Fifty mosaic nail-fold images of healthy controls (n = 10), patients with primary RP (n = 10) and SSc (n = 30) were assessed in random order by two blinded observers on two occasions at centres in Sweden, UK and The Netherlands. Each image was therefore scored by six observers twice.
Results: Inter- and intra-observer reliability of quantitative parameters showed substantial to almost perfect agreement [inter- and intra-observer weighted κ's for the number of widened capillaries was 0.75 and 0.87 and giant capillaries was 0.84 and 0.92, intra-class correlation coefficients (ICCs) for capillary density was 0.87 and 0.92 and total loop width was 0.94 and 0.98, respectively]. Qualitative parameters including architecture, avascularity, haemorrhage, crossed, ramified and bushy capillaries showed moderate to substantial inter-observer reproducibility (weighted κ ranging from 0.47 to 0.73), and substantial intra-observer repeatability (weighted κ ranging from 0.71 to 0.80), whereas the scoring of tortuous and bizarre capillaries showed poor inter-observer and substantial intra-observer agreement (inter-observer weighted κ's was 0.39 and 0.21 and intra-observer weighted κ's was 0.68 and 0.76, respectively).
Conclusion: All quantitative and certain qualitative parameters are highly reliable in terms of inter- and intra-observer agreement. A combination of parameters with the highest reliability should be incorporated into future capillaroscopic scoring systems in studies of prediction and monitoring of SSc spectrum disorders.