March 2012
Volume 53, Issue 14
Free
ARVO Annual Meeting Abstract  |   March 2012
Composite Index Scores for Dry Eye Syndrome: a Comparative Approach
Author Affiliations & Notes
  • Craig W. See
    Dept. of Ophthalmology, Columbia University, New York, New York
  • Bozorgmehr Pouyeh
    Dept. of Ophthalmology, Bascom Palmer Eye Institute, University of Miami, Miami, Florida
    Dept. of Ophthalmology, Miami Veterans Affairs Medical Center, Miami, Florida
  • Richard A. Bilonick
    Dept. of Ophthalmology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
    Dept of Biostatistics, University of Pittsburgh Graduate School of Public Health, Pittsburgh, Pennsylvania
  • Anat Galor
    Dept. of Ophthalmology, Bascom Palmer Eye Institute, University of Miami, Miami, Florida
    Dept. of Ophthalmology, Miami Veterans Affairs Medical Center, Miami, Florida
  • Footnotes
    Commercial Relationships  Craig W. See, None; Bozorgmehr Pouyeh, None; Richard A. Bilonick, None; Anat Galor, None
  • Footnotes
    Support  RO1 EY018624-01
Investigative Ophthalmology & Visual Science March 2012, Vol.53, 549. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Craig W. See, Bozorgmehr Pouyeh, Richard A. Bilonick, Anat Galor; Composite Index Scores for Dry Eye Syndrome: a Comparative Approach. Invest. Ophthalmol. Vis. Sci. 2012;53(14):549.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose: : Dry eye syndrome (DES) is a prevalent condition with significant associated morbidity. Studying the condition ischallenging as there is currently no gold standard for the diagnosis and the individual tests used are poorly correlated. Recently, independent component analysis (ICA) has been applied to DES in order to generate a composite score and evaluate the sensitivity/specificity of each individual test based on this score. As the optimal way to combine information from various DES tests is unknown, the purpose of our study was to generate DES composite scores using two methods, ICA and latent class analysis (LCA) and compare the resultant outputs.

Methods: : The Veterans Dry Eye study is an observational study examining DES in male US military veterans. For each individual, the five-item dry eye questionnaire (DEQ5) as well as conjunctival injection, tear osmolarity (TearLAB), tear breakup time, meibomian gland dysfunction, corneal staining (PEE), and Schirmer’s strips, were obtained. ICA, which maximizes independence in the data, and LCA, which allows for estimation of sensitivity and specificity in the absence of a gold standard, were applied to available data. Severity of cases was assigned by linear weighted model (ICA) and a posterior probability of dry eye (LCA). Relative performance was determined by weighting factors (ICA) and by sensitivity plus specificity (LCA).

Results: : 233 individuals were included. ICA found that meibomian gland score performed best, followed by conjunctival staining and tear osmolarity; these had the worst performance according to LCA. LCA found that corneal staining for PEE performed best, followed by tear breakup time and Schirmer’s; these had the worst performance according to ICA. ICA and LCA severity showed moderate correlation, R^2 = 0.55. 63% (37/58) of individuals were assigned to the top quartile by both methods; 67% (39/58) were in agreement in the bottom quartile.

Conclusions: : ICA and LCA are very different approaches to the interpretation and combination of various DES tests. There was poor consistency between ICA and LCA; the best-performing tests in one were the worst-performing in the other. This highlights the need for further research on the optimal approach for combining information in this condition.

Keywords: cornea: tears/tear film/dry eye • clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology • clinical laboratory testing 
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×