*2.1. Ethics*

This study was conducted in accordance with the guidelines of the World Medical Association, Declaration of Helsinki and the Ethical Guidelines for Medical and Health Research Involving Human Subjects in Japan. Subjects received full explanation of the procedures and provided written informed consent before enrollment. The study protocol was approved by the Institutional Review Board of the Clinical Study, Ryogoku Eye Clinic, Tokyo, Japan. The study was registered in a public registration system (University Hospital Medical Information Network registry no. UMIN 000015890) [10].

#### *2.2. Study Population*

We collected data from the DECS-J, which was an observational study conducted in 10 eye clinics in Japan. All investigators were specialists in ocular surface disorders and dry eye disease and belonged to the Japanese Dry Eye Society. To ensure the quality of the survey, two investigators' meetings were held prior to the start of patient enrollment to discuss the study protocols and examination procedures.

We consecutively enrolled outpatients who were at least 20 years old and were newly or previously diagnosed with dry eye disease. The criteria for diagnosis was as follows: (1) at least one abnormal tear examination result [Schirmer I test ≤ 5 mm, tear film break-up time (TBUT) ≤ 5 s]; (2) abnormal results in ocular surface vital staining tests (fluorescein keratoconjunctival staining score ≥ 3); and (3) presence of dry eye symptoms [11]. Subjects who met two or all three criteria were considered to probably or definitely have dry eye disease, respectively, and were included in the study. Up to 50 patients were enrolled at each of the 10 study sites from 1 December 2014 to 28 February 2015.

A total of 449 patients with an average age of 62.6 ± 15.7 years were included in this study. Of these, 386 (86.0%) were females. Table 1 shows the age distribution of the patients in this study.


**Table 1.** Age and sex distribution of the patients in this study.

#### *2.3. Systemic Comorbidities and Oral Medicine*

Patients were asked to provide information regarding systemic comorbidities and oral medicine. Hypertension, insomnia and depression were defined as having hypertension or taking an antihypertensive medication; as having insomnia or taking a sedative; and as having depression or taking an antidepressant, respectively.

#### *2.4. Quality of Life (QOL) and Health Utility Assessment*

QOL and health utility among patients with dry eye disease were evaluated using the DEQS [12] and the Human Utility Index Mark 3 (HUI-3) questionnaires [13], respectively. The DEQS consists of 15 questions and is scored using an overall summary scale. The score derived from this questionnaire is considered to be a quantitative measure of dry eye symptoms, in which 0 indicates the best and 100 indicates the worst. Test/retest reliability and discriminant validity of the DEQS were confirmed in a previous study, showing that the score was significantly higher in patients with dry eye disease than in healthy controls (33.7 vs. 6.0) [9].

The HUI-3 is a standard method for assessing the utility value. It is a multi-attribute utility classification system for preferences associated with generic health states. It consists of 15 questions assessing eight health attributes: vision, hearing, speech, ambulation, dexterity, emotion, cognition and pain. Individual sets of answers were converted to utility values, using the method described by Furlong et al. [14]. A utility value representing the overall health state was derived by applying a weighted scoring algorithm, in which utility is defined on a continuous scale from 0 to 1, where 0 corresponds to the worst possible QOL outcome (equal to death) and 1 corresponds to the best possible one (equal to perfect health). Patients were asked to fill out the questionnaires at home and return them by post.

#### *2.5. Ophthalmic Evaluation*

Ophthalmic examinations included assessment of conjunctival and corneal vital staining with fluorescein sodium, measurement of the TBUT and the Schirmer I test.

Test strips containing fluorescein sodium (Fluores Ocular Examination Test Paper, Ayumi Pharmaceutical Co., Tokyo, Japan) were used for vital staining and TBUT measurement. After the tip of a wet test strip had been applied to the inferior temporal tear meniscus, patients were asked to blink several times to ensure adequate mixing of the fluorescein dye with the tears. The time interval between the last complete blink and the appearance of the first corneal dark spot was timed using a stopwatch. The average of three measurements was recorded as the TBUT. Corneal and conjunctival epithelial damage was then evaluated via corneal fluorescein staining, according to the National Eye Institute grading system [15]. Briefly, corneal staining was graded with a score of 0–3, assigned to each of five corneal zones (superior, nasal, central, inferior and temporal), with a maximum total score of 15. The fluorescein staining score of the keratoconjunctival was determined according to the modified grading system of van Bijsterveld [16], according to which each eye is divided into three sections (temporal conjunctiva, cornea and nasal conjunctiva) and scored from 0 to 3. The final score ranged from 0 (minimum) to 9 (maximum). The Schirmer I test was performed without topical anesthesia after all other examinations had been completed. A Schirmer test strip (Ayumi Pharmaceutical Co.) was placed on the outer one third of the temporal lower conjunctival fornix for 5 min. The strip was then removed, and the length of the wet filter paper was recorded in millimeters. To avoid any effect of keratoconjunctival staining on the Schirmer I test, the tests were performed 15 min apart, at minimum.

For each patient, the eye that met the most criteria for dry eye diagnosis was included in the study. If both eyes met the same number of criteria, the eye with the higher fluorescein staining score and the shorter TBUT was included. If these values were the same for both eyes, the right eye was used.

#### *2.6. Classification of Dry Eye Subgroups*

#### 2.6.1. Aqueous-Deficient (AD)/sTBUT

We classified the subjects into aqueous-deficient (AD) and short TBUT dry eye subgroups. The AD dry eye group was comprised of subjects who fulfilled the following criteria: (1) the presence of dry eye symptoms and (2) decreased tear production (Schirmer value ≤ 5 mm). The short TBUT dry eye subgroup included subjects who met the following conditions: (1) the presence of dry eye symptoms, (2) abnormal tear stability (TBUT ≤ 5 s), (3) no abnormality in tear production (Schirmer value > 5 mm), and (4) no abnormality in the ocular surface vital staining test (keratoconjunctival score < 3) [17].

#### 2.6.2. Meibomian Gland Function (MGD+/−)

We evaluated meibomian gland function (MGD in the central one-third of the upper eyelid using slit-lamp biomicroscopy. The patient was instructed to look down while the investigator gently and partially everted the upper lid to examine the lid margin. Typical pictures of the following signs/findings were distributed to each investigator to aid in the examination: (1) one or more abnormal findings around the meibomian gland orifices, such as vascular engorgement, anterior or posterior displacement of the mucocutaneous junction or irregularity of the lid margin (2) orifice obstruction, such as plugging, pouting and ridges [18]; and (3) hypersecretion or hyposecretion of the meibum [19]. We diagnosed MGD when all the signs and findings listed in the criteria were present [17].

#### 2.6.3. Friction-Related Diseases (FRD+/−)

We considered that subjects had friction-related conditions when any of the following conditions were present: lid wiper epitheliopathy (in the upper and lower lids), conjunctivochalasis (at the central site of the lower eyelid) or superior limbic keratoconjunctivitis [17,20–22].

#### *2.7. Statistical Methods*

We used one eye from each subject for the analysis, as explained above (Section 2.5). Data for all parameters are presented as the mean ± standard deviation. We performed all statistical analyses using SAS software, version 9.4 (SAS Inc., Cary, NC, USA). For comparisons between groups, we used Fisher's exact test for dichotomous variables and Student's *t*-test for continuous variables. We performed a linear model analysis to adjust for the e ffects of age. To estimate the association between a given independent variable and the outcome after adjustment for age, multiple linear regression was used for continuous variables and multiple logistic regression was used for binary outcomes. We considered a *p*-value lower than 0.05 as statistically significant for all analyses.
