UCSF

Purpose

To evaluate the ocular signs and tests for keratoconjunctivitis sicca (KCS) in the absence of a gold standard.

Methods

Cross-sectional study of participants from the Sjögren's International Collaborative Clinical Alliance (SICCA) registry. Participants had oral/ocular/rheumatologic examinations, blood/saliva samples collected, and salivary gland biopsy. Latent class analysis (LCA) identified clusters of patients based on 3 to 4 predictor variables relating to signs or tests of KCS. The resulting model-based "gold standard" classification formed the basis for estimated sensitivity and specificity associated with these predictors.

Results

A total of 3514 participants were enrolled into SICCA, with 52.9% classified as SS. LCA revealed a best-fit model with 2 groups. For the gold standard-positive group, an abnormal tear breakup time, ocular staining score (OSS), and Schirmer I had a sensitivity of 99.5%, 91.0%, and 47.4%, respectively. For the gold standard-negative group, an abnormal tear breakup time, OSS, and Schirmer I had a specificity of 32.0%, 84.0%, and 88.5%, respectively. OSS components (fluorescein and lissamine staining), exhibited a sensitivity of 82.6% and 90.5%, respectively, in the gold standard-positive group, whereas these signs in the gold standard-negative group had a specificity of 88.8% and 73.0%, respectively.

Conclusions

OSS and its components (fluorescein and lissamine staining) differentiated 2 groups from each other better than other KCS parameters and had relatively high sensitivity and specificity.