Abstract
Purpose: :
To report the prevalence of dry eye syndrome (DES) in Valladolid, calculate internal validity of two DES screening questionnaires, and correlate DES diagnostic tests.
Methods: :
Subjects older than 50 years of age, randomly selected from the medical network census in Valladolid (SPAIN), answered two questionnaires (Q1 and Q2), followed by an ophthalmic evaluation. Q1 (modified McMonnies) was intended for general practice screening and had 8 dichotomic (yes=1/no=0) items. Q2 (Ocular Surface Index Questionnaire, OSDI), with 12 questions, (score 0–5 for each) was intended for eye care specialists. Diagnostic tests, always performed by the same investigator, were: tear break–up time (BUT; abnormal <10 s), fluorescein (F) and Rose Bengal (RB) stainings, and Schirmer with anesthesia (S, abnormal < 5 mm in 5 min)). Descriptive analysis, test concordance analysis (kappa index), Q1 and Q2 reliability using Cronbach's alpha, area under ROC curve, and factor analysis were performed. Questionnaire validity was measured by true positives defined as Q1 with > 4 affirmative questions and Q2 with score > 20 and S < 5mm in 5 min and BUT < 10
Results: :
A total of 270 subjects (58.2% female; 41.8% male) with an average age of 64.5 years (CI 95%: 63.3–65.7) were enrolled. Mean score was 1.2 (range, 0–8) for Q1 and 1.4 (range,0–3.6) for Q2, meaning that most symptoms occurred "sometimes" or less. Prevalence of DES based on tests was: F: 13%, RB: 15.6%, BUT: 45%, S: 45.6%, S+ RB: 7.4%, and S+ BUT: 24.2%. Cronbach's alpha coefficient for Q1 was 0.507 and 0.583 for Q2, showing poor internal consistency. Test concordance analysis (kappa index) was: S vs F: 0.017, RB vs S: 0.013, BUT vs S: 0.144, BUT vs RB: 0.045, and F vs RB: 0.319. Only F vs RB was significant (p=0.0) Q1 area under the ROC curve was 0.528 for BUT (p=0.4), 0.544 for F (p=0.4), and 0.506 for S (p=0.8)). Q2 area under the curve was >0.5 only for BUT (0.503, p=0.94). Q1 factor analysis yielded 3 independent factors explaining 51.8% of the total variance. In Q2, 5 independent factors explained 64.2% of total variance.
Conclusions: :
F vs RB were the most concordant tests for DES diagnosis. Prevalence of DES in Valladolid based on these two tests (F+RB) ranged from 13 to 15.6%. Q1 and Q2 were unreliable for DES diagnosis. Q2 is less reliable than Q1. No correlation existed between screening questionnaires and objective DES diagnostic tests.
Keywords: cornea: tears/tear film/dry eye • clinical (human) or epidemiologic studies: prevalence/incidence