1. Study design: an assessor-blinded, randomized, controlled trial
2. Target population: MGD patients having mild to moderate MG dropout from Out-clinic at King Chulalongkorn Memorial Hospital (KCMH).
4\) Randomization and blinding Permuted block randomization will be used. After recruiting in this study, participants who meet the eligibility criteria will be divided into 2 groups by research assistant. There is an independent ophthalmologist (A.L) who is blinded to treatment allocation.
5)Data collection Demographic characteristics, numeric rating scale-11 (NRS-11) and ocular surface disease index (OSDI) will be collected by interviewing. Other parameters will be collected by slit lamp examination, specific machines or further investigations. All the data will be filled in the medical record. The participants will be identified on medical record by a unique research number, not by name or identification (ID) number (apart from on the consent form and enrollment log). Patient's ID, name, surname, and phone number will be stored in different file by the research assistant. The data will be locked up and destroyed 5 years after the study ends.
When the patients are recruited in the study, the patients will be provided information sheet and explained about the study. If the patients agree to participate in the study, they will be asked to sign the consent form. Then, they will be appointed to return to the Refractive Surgery Center in the next day for examination and tear collection between 1 pm to 3 pm because some cytokines are independent diurnal rhythms.
All outcome data will be collected by one single masked investigator (A.L). The data will be collected at 4 time points (day 0, week 4, week 8 and week 12). At baseline and 12-week post treatment, participants are required to be at the hospital on time within the period of 1 pm to 3 pm for tear collection. In each visit, the examination must be done in correct order in order not to avoid the results of outcomes. These steps include:
* Tear collection
* Evaluating the ocular symptoms
* Measuring lipid layer thickness
* Slit-lamp examination (tear break up time, lid margin abnormality, meibum expressibility and meibum quality)
* Non-contact infrared Meibography
* Evaluating adverse events
8\) Sample size: There is no previous study showing the difference in Meibomian gland dropout area between these groups. Our expert's opinion about minimal clinical difference is 10%. The standard deviation of meibomian gland dropout area in obstruction MGD patients is 16.7%. We use the formula for a trial with an equal allocation ratio and repeated measurements. This will be inflated for a 10% drop out, giving a total sample size of 32 per group (total sample size of 64).
* Statistical analysis:
Stata (version 17) software are utilized. All statistical analyses were undertaken by a researcher who are blinded to group allocation. The data were analyzed using an intention-to-treat (ITT) framework, which included all participants randomized in the study, regardless of the treatment type or exposure received. Per protocol (PP) is added as a supportive analysis. PP analysis will exclude the participants who do the lid hygiene less than 20 days per month or using any other treatment than advice. A significance level of 5% was used in all statistical tests.
Demographic data will be interpreted by descriptive analysis. The primary outcome is the change in meibomian gland percentages of each eyelid from baseline over all follow-up visits analyzing by linear mixed models with random intercept for patient, adjusting for baseline Meibomian gland percentage in the study participants and modelling time as a discrete variable. Marginal models will be used to predict the discrete changes at each study month. OSDI, tear break up time and lipid layer thickness are analyzed by linear mixed models (LMM). Lid abnormalities, meibum score and MG morphology will be analyzed by generalized linear mixed model (GLMM). Cytokine levels will be analyzed by unpaired t test.
9\) Missing data management: If a patient chooses to withdraw from the study, the reason for withdrawal will be documented. These reasons will be valuable for analyzing the results later on. Multiple imputation method will be used to handle with missing data.
