Kawakita 2013.

Methods	Study design: randomized, controlled trial Study site(s): not reported if single‐ or multi‐center Number randomized (total and per group): 27 participants Unit of randomization (individual or eye): individual Exclusions after randomization: 1 participant discontinued for a reason other than the present study Losses to follow‐up: none Number analyzed (total and per group): 26 participants in total; 30 eyes of 15 participants in the treatment group; 22 eyes of 11 participants in the control group Unit of analysis (individual or eye): individual (subjective symptoms); eye (TBUT; Schirmer test; fluorescein staining; rose bengal staining) Reported power calculation? (Y/N): N Reported subgroup analysis? (Y/N): N
Participants	Baseline characteristics: Country: Japan Age (mean ± SD, range): 52.5 ± 2.5 years in the treatment group; 51.9 ± 2.2 years in the control group Gender: 5 men and 10 women in the treatment group; 1 man and 10 women in the control group Inclusion criteria: 1. Dry eye defined as (1) presence of symptoms of dry eye, (2) abnormality of tear production as determined by the Schirmer test (< 5 mm/5 min) or presence of tear film instability as determined by TBUT (< 5 s), and (3) positive ocular surface rose bengal score (> 3 points) or fluorescein vital staining (> 3 points) Exclusion criteria: 1. Severe disorder of the ocular surface such as Stevens‐Johnson syndrome or ocular pemphigoid 2. Systemic illness such as diabetes, hypertension, or autoimmune disease 3. Taking supplements containing eicosapentaenoic acid or docosahexaenoic acid 4. Hemophilia, gastrointestinal ulceration, urinary tract hemorrhage, hemoptysis, vitreous hemorrhage, or hemorrhagic tendency 5. Treated by an ophthalmologic surgery within the previous 6 months 6. Treated by punctal plug within the previous 1 month 7. Wearing contact lenses Equivalence of baseline characteristics? (Y/N): not reported
Interventions	Intervention #1 (treatment group): oral soft gel capsule containing eicosapentaenoic acid 83 mg and docosahexaenoic acid 36 mg (Nippon Suisan Kaisha, Ltd.), 3 times daily, 5 capsules at a time (daily dose of eicosapentaenoic acid 1245 mg and docosahexaenoic acid 540 mg) Intervention #2 (control group): placebo capsules containing medium‐chain triglycerides (dose not reported), 3 times daily Length of follow‐up: 12 weeks, and 4 weeks additional after suspension of treatment Notes: taking of supplements other than experimental supplements was prohibited; participants were instructed to maintain their normal food and exercise habits; all usual dry eye treatments such as eye drops, but not supplementation, were maintained during this experiment in all participants
Outcomes	Primary outcome(s): objective and subjective symptoms by the visual analogue scale (VAS) test and TBUT at the study endpoint; change from baseline in fluorescein staining and rose bengal staining Other outcome(s): Schirmer I test without anesthesia at study endpoint Adverse events reported? (Y/N): N Measurement time points (specify intervals at which outcomes were assessed): baseline, 12 weeks of treatment, 4 weeks after suspension of treatment Other issues with outcome assessment (eg, quality control for outcomes if any): this study has a unit of analysis issue, whereby both eyes from participants were included in the analyses of ocular outcomes, without apparent statistical adjustment for within‐person (between‐eye) correlation
Notes	Study dates: not reported Funding source(s): not reported Conflicts of interest: 2 authors were affiliated with the company that produced the fish oil products in the study Publication language: English Registered on clinical trials registry? (Y/N): N
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"All eligible subjects were randomly allocated to the fish oil or placebo group by using a random number table"
Allocation concealment (selection bias)	Unclear risk	Allocation concealment was not reported
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	"Double‐blind" study, but details of masking were not reported
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	"Double‐blind" study, but details of masking were not reported
Incomplete outcome data (attrition bias) All outcomes	High risk	"One subject dropped out because of the reason other than the present experiment, and final data were calculated from 26 subjects" It is not clear which intervention arm the participant was assigned to
Selective reporting (reporting bias)	Unclear risk	No access to study protocol or clinical trials registry
Other bias	High risk	2 authors were affiliated with the fishing company that produced the fish oil products in the study. In addition, the unit of randomization was the individual participant, but each eye of a single participant was separately included in the analysis for the Schirmer test and TBUT, without taking into account non‐independence (ie, unit of analysis error)