Study design and calculation
A total of 200 consecutive patients who visited the Optometry Clinic with complaints related to dry eye were recruited for the study. These patients had been referred by their medical doctors or the health service of the institution for an ocular surface examination due to dry eye symptoms. Patients willing to participate were provided with a self-administered electronic questionnaire through a designated mobile form, which took approximately five minutes to complete. The questionnaire gathered information on the patients' demographics, Ocular Surface Disease Index (OSDI) score [11, 12], and whether they had previously received a diagnosis of rheumatoid arthritis (RA) from their medical doctor. Informed consent was obtained from all participants, and the study was conducted in accordance with the Declaration of Helsinki and approved by the institutional Ethics Committee of the university.
The sample size was calculated using PS Power and Sample Size Calculations Version 3.1.2 (Copyright© by William D. Dupont and Walton D. Plummer) based on the TFOS DEWS II Diagnostic Methodology report principles [7, 13, 14]. The literature-reported standard deviations (SDs) for symptomatology status (OSDI score), tear film osmolarity, tear film FBUT, and fluorescein corneal staining were assumed to be 6.7, 4.8 mOsm/L, 2.9 s, and 2, respectively [7, 11]. To achieve a power of 80% (Type II error associated) for a significance level of 𝛼 = 0.05 (Type I error associated) and a confidence level of 95% to detect a clinical difference between non-pathological and pathological participants of 7.3, 5 mOsm/L, 5 s, and 1, respectively, a minimum of 20, 18, 12, and 17 subjects in each group were required to achieve a minimum control/experimental ratio of 1:1. These results were in line with the recommendations of the TFOS DEWS II Diagnostic Methodology report [7]. The highest value among these was used as the reference for the sample size per group (20 participants) to ensure a more reliable study. A larger sample size was recruited to increase the impact of the study results.
Study Design and Diagnostic Criteria
A battery of clinical procedures was conducted in all participants according to the TFOS DEWS II Diagnostic Methodology report [7, 13, 14]. The tests were always performed in the same order, from the least to the most invasive, while the first eye to be measured was randomly selected [7, 13, 14]. The procedures included tear film osmolarity, lipid layer pattern (LLP), tear meniscus height (TMH), tear film break-up time (FBUT), and fluorescein corneal staining.
To diagnose dry eye disease (DED), the cut-off values used were an OSDI score ≥ 13, tear osmolarity ≥ 308 mOsm/L, FBUT < 10 s, and corneal staining (Oxford Scheme) ≥ 2 (Table 1). Only participants who were classified as having DED, with positive symptomatology and positive results on at least one clinical test, were included in the study sample [7, 13, 14]. According to the TFOS DEWS II criteria [7, 8], participants were classified as having aqueous-deficient DED (ADDE) if they obtained a low tear meniscus height (TMH ≤ 20mm), as having evaporative DED (EDE) if they obtained a thin lipid layer pattern (LLP ≤ CM), or as having mixed DED if both criteria were met [8, 15, 16]. Participants classified as mixed DED were not included in the final analysis to avoid possible interferences (Table 1).
Table 1
Summary of the battery of clinical procedures performed [7, 13, 14]. OSDI = ocular surface disease index; FBUT = fluorescein break-up time. TMH = tear meniscus height; LLP = lipid layer pattern. CM = closed meshwork. ADDE = aqueous deficient dry eye. EDE = evaporative dry eye.
Symptoms | Signs | Type |
OSDI score ≥ 13 | Only one condition must be filled: - Osmolarity in one eye ≥ 308 mOsm/l - FBUT in one eye < 10 s - Corneal staining (Oxford Scheme) in one eye ≥ 2 | TMH ≤ 20mm LLP < CM | ADDE |
TMH < 20mm LLP ≤ CM | EDE |
TMH ≤ 20mm LLP ≤ CM | Mixed |
All measurements were performed and recorded by the same examiner who was unaware of the questionnaire results. The measurements were taken in a single session to avoid inter-examiner or intra-session variability. The instruments and materials used, such as the fluorescein strip and TearLab test cards, were kept in a humidity- and temperature-controlled room with a temperature of 20–23°C and humidity of 50–60%, where the study was conducted [17]. Participants were allowed to rest for 5–10 minutes to adapt to the ambient conditions prior to the measurements.
Evaluation Procedures
Symptomatology Assessment
The symptomatology of DED was assessed using a self-administered OSDI questionnaire. Participants scanned a QR code provided prior to the examination and were asked questions related to their symptoms during the previous week using a standardized interview model. The scores obtained were computed by the researchers on a scale of 0 to 100 points according to published guidelines, with higher scores indicating greater disability [18].
Tear Film Osmolarity
The TearLab osmometer (TearLab Corp, San Diego, CA, United States) was used to measure tear film osmolarity [19]. Participants were instructed to sit with their head tilted back and eyes looking up towards the ceiling. The osmometer probe was then placed on the lower tear meniscus until a beep was emitted, indicating that the tear sample had been collected. The osmometer converted the electrical impedance of the sample into osmolarity (mOsm/L), which was displayed on the device screen. To avoid any inter-eye interference, the contralateral eye was measured after a 5–10-minute interval following the same protocol [17]. The system's correct function was regularly verified using a quality control electronic check card, which confirmed that the reading was 334 ± 3 mOsm/L, indicating that the device was working accurately.
Fluorescein Break-Up Time
The FBUT measurement was conducted and videorecorded using the fluorescein function of the Keratograph 5M (Oculus Optikgerate GmbH, Wetzlar, Germany)[20]. To avoid inter-patient evaluation interference, a fluorescein strip hydrated with saline solution was applied to the patient's eye in the upper bulbar conjunctiva area[21]. After applying the dye, participants were instructed to blink normally to ensure even distribution and then to blink three times while keeping their eyes open until the end of the test [22]. The FBUT was defined as the time interval between the last blink and the appearance of the first black spot on the corneal surface. The procedure was repeated three times for each participant [22]. The FBUT was calculated by a second masked examiner using the recorded videos and the open-source software VistualDub64, which improved temporal resolution by converting the video into frames (8 frames = 1s) [22]. The final value was the mean of the two closed measurements [22].
Corneal Staining
Immediately following the FBUT assessment, corneal staining was evaluated and recorded by the Keratograph 5M using the fluorescein function while the participants remained seated in the same position [20]. Participants were instructed to focus on a red target at the center of the device, after which they were asked to look in various directions (right, left, up, and down) to assess the paracentral areas; during the "look down" phase, the upper lid was slightly manipulated to examine the upper part of the cornea [23, 24]. A second masked examiner assessed the recorded videos of this test using the Oxford Scheme, which grades the severity of dry eye as mild (stage 0 or 1), moderate (stage 2 or 3), or severe (stage 4 or 5) [23, 24].
Tear Meniscus Height
TMH was semi-automatically quantified using a previously established protocol [15, 25]. Participants were seated at a Topcon SL-D4 slit-lamp biomicroscope (Topcon Corporation, Japan) and instructed to maintain primary eye gaze with a natural blink while fixating on a target. A Topcon DC-4 digital camera connected to a computer was used to record video of the tear meniscus. A 3x5 mm light beam with moderate illumination was used to avoid reflex tearing and prevent direct light from shining into the pupil during measurements [15, 25]. The central meniscus was captured at the 6 o'clock position without tilting the illumination column, and images of the tear meniscus were extracted from the recording. The images were then measured by a second experienced observer, who was masked to the study groups, using computer-assisted image analysis software (ImageJ software v1.53i, National Institutes of Health, Bethesda, MD; http://imagej.nih.gov/ij/)[15, 25, 26]. The data were converted from pixels to millimetres.
Lipid Layer Interference Pattern
The lipid layer of the tear film was assessed using a Tearscope (Keeler, Windsor, United Kingdom)[27], which was fixed at a constant distance with the chinrest of the Topcon SL-D4 slit-lamp to provide a standardized area. Image acquisition was performed following a previously established protocol[16]. The participant was seated behind a slit-lamp and instructed to look at a target to maintain primary eye gaze, with the lipid layer region of interest centered, while a natural blink was allowed. Throughout the procedure, the illumination was provided by the Tearscope. The tear film was recorded by a digital camera attached to the slit-lamp and stored on a connected computer. Care was taken to ensure that the videos met minimum quality requirements, including being free of blur, the lipid layer being well spread after a complete blink, and being well centered. It is important to note that correct centering and focusing on the LLP videos requires prior training [16]. Images were acquired from recorded videos by a second experienced masked observer [16]. When the LLP evolution was stable with minimal variations, the clearest image was extracted (approximately 1-1.5 seconds after blinking). As the appearance of the lipid layer is not static between blinks, the images were categorized following Guillon's Clinical scheme as open meshwork (OM), closed meshwork (CM), wave (W), amorphous (AM), and color (COL) [16, 28]. Since LLPs do not follow a discrete classification, but rather a continuous evolution of the lipid tear film thickness, when the pictures fell between two grades, the observers classified them as the pattern that was closer, following the Guillon scheme and their own experience.
Statistical Analysis
Data analysis was performed using SPSS statistical software version 25.0 for Windows (SPSS Inc., Chicago, IL, United States). Significance was set at p ≤ 0.05 for all analyses. Prior to analysis, the normal distribution of the data was checked using the Shapiro-Wilk test [29]. Tear osmolarity, FBUT and TMH showed a normal distribution (Shapiro-Wilk, all p > 0.05), while OSDI, corneal staining, and LLP were not normally distributed (Shapiro-Wilk, all p < 0.05). Descriptive statistics were calculated as mean with SD for parametric variables, median and interquartile range (IQR) for non-parametric variables, and minimum and maximum values were reported in all cases. Differences in parameter values between groups or subgroups were analyzed using unpaired t-tests for parametric variables and Mann-Whitney U tests for non-parametric variables.
Odds ratios (OR) along with 95% confidence intervals (CI) were estimated to assess the magnitude of the association between RA and the established DED category [29]. Due to the categorical nature of the data, a chi-squared test was used to compare the outcomes of the risk factor studied across different DED categories. Fisher's exact tests and Cramer's V were performed to evaluate the associations and correlations between categorical data [29].