Screening utility of a rapid non-invasive dry eye assessment algorithm

Abstract

PurposeTo evaluate the discriminative ability of a rapid non-invasive dry eye assessment algorithm (global evaluation component: SANDE questionnaire, and non-invasive tear film breakup time; subtype classification testing component: tear film lipid layer grade, and tear meniscus height) in detecting dry eye disease, as defined by the TFOS DEWS II diagnostic criteria. MethodsTwo hundred and thirty-five participants (77 male, 158 female), with a mean ± SD age of 43 ± 17 years, were recruited into a prospective diagnostic accuracy study. OSDI, DEQ-5, and SANDE dry eye symptomology scores; non-invasive tear film breakup time; absolute and inter-ocular differences in tear osmolarity; corneal, conjunctival, and lid margin staining scores; tear film lipid layer, meibum expressibility, meibomian gland orifice plugging, and eyelid margin telangiectasia grades; and tear meniscus height were evaluated in a single clinical session. ResultsThe areas under the ROC curves exceeded 0.80 for all individual components of the rapid non-invasive dry eye assessment algorithm, and the discriminative abilities were significantly greater than chance (all p < 0.001). At the Youden optimal diagnostic thresholds for the global evaluation component of the rapid non-invasive assessment algorithm (SANDE score ≥30, non-invasive tear film break-up time <10 s), the overall sensitivity was 86%, specificity 94%, positive likelihood ratio 15.0, and negative likelihood ratio 0.15. ConclusionsThe abridged non-invasive dry eye assessment algorithm may be a useful rapid screening instrument for the full TFOS DEWS II diagnostic test battery, of particular benefit in resource or time-constrained settings.