IMPORTANCE

Tear film breakup time assessment is an integral component of dry eye evaluation. To our knowledge, the comparative discriminative ability of the noninvasive Keratograph (Oculus) vs conventional fluorescein method in detecting dry eye is unknown.

OBJECTIVE

To compare tear film stability measurements obtained with an automated noninvasive corneal topographer vs conventional fluorescein methods and evaluate their respective discriminative ability in detecting dry eye.

DESIGN, SETTING, AND PARTICIPANTS: This investigator-masked randomized crossover trial was conducted at a single-center university clinic between May 26, 2016, and October 3, 2016, and included 74 participants 18 years or older. Participants were recruited into 2 equally sized age, sex, and race/ethnicity-matched groups, with and without symptomatic dry eye (Ocular Surface Disease Index ≥13).

INTERVENTIONS

Participants were assigned to receive a noninvasive keratograph evaluation and topical fluorescein instillation in a randomized order.

MAIN OUTCOMES AND MEASURES

Noninvasive keratograph breakup time (NIKBUT) and fluorescein breakup time (TBUT). Area under the receiver operating characteristic curve, Youden-optimal diagnostic cutoff sensitivity, and specificity of NIKBUT and TBUT in detecting dry eye.

RESULTS

Seventy-four participants (74 eyes; 43 women [58.1%]) with a mean (SD) age of 24 (4) years were randomized. Noninvasive keratograph breakup time was significantly longer than TBUT in participants with dry eye (median, 6.3 seconds vs 4.3 seconds [difference, 2.0 seconds]; 95% CI, 1.1-3.4 seconds; P = .003), and healthy participants (median, 11.9 seconds vs 5.0 seconds [difference, 6.9 seconds]; 95% CI, 4.7-7.6 seconds; P < .001). Fluorescein breakup time measurements were more narrowly distributed in both the dry eye (variance, 188 seconds2 vs 27.9 seconds2; P < .001) and control groups (variance, 113 seconds2 vs 13.4 seconds2; P < .001). The discriminative ability of NIKBUT in detecting dry eye (area under the receiver operating characteristic curve, 0.68; 95% CI, 0.56-0.81; P = .007) was greater than that of TBUT (area under the receiver operating characteristic curve, 0.57; 95% CI, 0.44-0.70; P = .31). The optimal diagnostic cutoff for NIKBUT was 9 seconds or less with a sensitivity of 68% (95% CI, 50%-82%), specificity of 70% (95% CI, 53%-84%), positive likelihood ratio of 2.27 (95% CI, 1.32-3.91), and negative likelihood ratio of 0.46 (95% CI, 0.28-0.77). The optimal threshold for TBUT was 5 seconds or less with a sensitivity of 54% (95% CI, 37%-71%), specificity of 68% (95% CI, 50%-82%), positive likelihood ratio of 1.67 (95% CI, 0.96-2.89), and negative likelihood ratio of 0.68 (95% CI, 0.45-1.03).

CONCLUSIONS AND RELEVANCE

Conventional fluorescein tear film breakup time measurements were significantly shorter with narrower distributions, while automated noninvasive keratograph readings displayed superior discriminative ability in detecting dry eye.

TRIAL REGISTRATION

anzctr.org.au Identifier: ACTRN12617001428358.