Assessment of four validated questionnaires for screening of dry eye disease in an African cohort

To evaluate the 5-item dry eye questionnaire (DEQ-5) for measurement precision and unidimensionality, and assess the ability of Ocular Surface Disease Index (OSDI), McMonnies Dry Eye Questionnaire (MQ), Ocular Comfort Index questionnaires (OCI), and DEQ to predict clinical measures of DED in an African population.

Participants (n = 113; mean age 24.7 ± 5.4 years) completed all four questionnaires administered in a randomized order. Internal consistency and test-retest reliability of the instruments were assessed. A masked examiner performed clinical dry eye tests in both eyes of participants and the results used to classify into either "dry" or "non-dry" groups. The diagnostic criteria for dry eye disease were tear break up time (TBUT) < 10 s plus at least one of Schirmer <15 mm in 5 min., > 5 corneal staining spot, or > 9 conjunctival staining spots. Rasch analysis was performed to assess the DEQ for unidimensionality and measurement precision, and receiver operating characteristic curve (AUC) analysis was used to assess the ability of the four questionnaires to predict clinical measures of DED.

The responses from the DEQ-5 met the Rasch analysis criterion of unidimensionality with infit and outfit values of between 0.74-1.43 mean-square and 0.69-1.49 mean-square, respectively, and measurement precision of 2.33. Internal consistency and test-retest reliability of each test was high, with Cronbach alpha >0.70 for the questionnaire and their respective subscales but internal consistency reliability of the MQ was poor (Cronbach's alpha = 0 .29-0 0.52). Each test segregated the "dry" and "non-dry" groups with moderate (all AUC 0.70-0.80) and similar accuracy (P > .05).

The DEQ -5 demonstrated adequate measurement precision and unidimensionality. Each of the four questionnaires showed psychometric validity for clinical use in the African population, and modest and similar accuracy when screening for dry eye signs but the performance of the McMonnies questionnaire was limited in terms of internal consistency reliability.