Diagnostic profile of tear osmolarity and inter‐ocular variability for dry eye disease

Abstract

Tear hyperosmolarity is a central hallmark of dry eye disease, and perpetuates a vicious cycle of ocular surface inflammation and tear film instability. The measurement of tear osmolarity and inter-ocular variability forms part of the global consensus dry eye diagnostic criteria recommended by the Tear Film and Ocular Surface Society Dry Eye Workshop II (TFOS DEWS II). This investigator-masked diagnostic accuracy study sought to evaluate the discriminative ability and optimal thresholds for tear osmolarity and interocular variability in detecting other dry eye signs and symptoms. The study received institutional ethics committee approval and conformed to the tenets of the Declaration of Helsinki. Participants were required to be 16 years or older, with no ophthalmic surgical procedures in the 3 months preceding study participation. Written consent was provided by 866 participants, satisfying diagnostic accuracy power calculations (sample size ≥ 814, estimated prevalence = 40%, anticipated sensitivity = 70%, confidence level = 95%, absolute precision = 5%, power = 80%). The 5-Item Dry Eye Questionnaire and Ocular Surface Disease Index dry eye questionnaires were administered, and right eye ocular surface parameters (Oculus Keratograph 5M) assessed. An independent observer measured tear osmolarity from both eyes (TearLab Osmometer), and the higher reading and inter-ocular difference was recorded. The presence of nonosmolar dry eye signs and symptoms was determined according to the TFOS DEWS II diagnostic criteria (Table 1). The discriminative ability of osmolarity measurements in detecting other dry eye signs and symptoms was determined by the area under the receiver operating characteristic curve (C-statistic) and compared using the paired DeLong test. Youden-optimal diagnostic cut-off sensitivity and specificity values were then calculated. The discriminative ability of tear osmolarity (C-statistic = 0.82; 95% confidence interval [CI], 0.79-0.85) was greater than inter-ocular variability (C-statistic = 0.68; 95% CI, 0.65-0.72; P < 0.0001), although both were significantly greater than chance (both P < 0.0001). The optimal diagnostic cut-off for tear osmolarity was ≥308 mOsm/L,