Dry Eye Research in 2022: 32 Notable Human Studies

The 2022 picture

What the research shows

The 2022 evidence points in several directions. Large randomized trials gave useful answers about treatment and prevention, including clearly negative omega-3 prevention and tanfanercept results. Devices and newer drops produced promising sign changes, yet symptoms and prespecified endpoints sometimes told a different story. Population studies broadened the picture of who reports or receives a dry-eye diagnosis, while early diagnostic tools still need independent testing.

Controlled treatments and prevention

These studies test drops, devices, nasal stimulation, and supplements. The larger trials offer stronger comparisons, while a favorable secondary measure does not overturn a missed primary endpoint.

1. Once-daily cyclosporine gel in the COSMO trial

   In a masked phase 3 trial at 37 hospitals in China, 627 adults received 0.05% cyclosporine gel or its vehicle for 84 days. A meaningful improvement in inferior corneal staining occurred in 73.7% of the active group versus 53.2% of controls. Eye pain was reported more often with cyclosporine; the findings apply to the trial's selected moderate-to-severe patients.

2. TRPM8 agonist missed the COMET-1 primary endpoints

   A 12-week, three-arm phase 2b trial compared two concentrations of the cooling-receptor agonist AR-15512 with vehicle. The prespecified co-primary discomfort and anesthetized Schirmer outcomes did not separate significantly, although the higher dose favored treatment on several other tear and symptom measures. Those secondary signals justified further study, not a claim that the trial met its main goals.

3. A nasal spray produced short-term tear and symptom changes

   The masked PEARL phase 2 trial randomized 165 adults to three doses of simpinicline nasal spray or vehicle. Higher doses increased Schirmer tear measurements immediately after treatment and reduced dryness during a controlled-environment challenge. This single-dose experiment demonstrated an acute response; it did not establish sustained benefit from regular use.

4. Long-acting diquafosol improved corneal staining

   A Japanese multicenter phase 3 trial randomized 337 people with aqueous-deficient dry eye to long-acting diquafosol or placebo three times daily for four weeks. Corneal staining met the primary endpoint and conjunctival staining also improved more with active drops; other secondary outcomes did not meet their targets. The shorter dosing schedule is appealing, but adherence in everyday use was not tested.

5. Marine omega-3 did not prevent dry eye in VITAL

   In a prespecified analysis of the randomized VITAL trial, 23,523 U.S. adults without prior dry eye took 1 gram of marine omega-3 daily or placebo for a median 5.3 years. Record-confirmed new dry-eye diagnoses occurred in 2.0% of each group. This strong prevention result does not answer whether a supplement helps a person who already has symptomatic disease.

6. IPL added to gland expression improved some signs and dryness ratings

   Patients with moderate-to-severe meibomian gland dysfunction were randomized to four sessions of intense pulsed light plus gland expression or sham light plus expression. Tear breakup time, several gland measures, and an eye-dryness rating favored active light at follow-up, while the OSDI symptom score did not differ between groups. The sham comparison helps isolate the device effect, but the symptom measures gave mixed results.

7. Two thermal pulsation devices at one year

   A multicenter assessor-masked trial randomized 236 people with gland-related evaporative dry eye to one iLux or LipiFlow treatment. At 12 months, iLux met the prespecified noninferiority margins for gland score, noninvasive tear breakup, and symptom bother. Both arms received an active device, so the study compares systems rather than proving the benefit of either against no procedure.

8. A mixed fatty-acid supplement and a symptom subgroup

   Fifty adults completed a three-month masked trial of an omega-3 and omega-6 supplement versus oil placebo. Symptoms improved in both arms, and a very symptomatic treatment subgroup improved from its own baseline; objective eye measures showed no significant change. The small subgroup and lack of a clear overall between-group symptom result limit any claim of general effectiveness.

9. Comparing two cyclosporine strengths

   A double-masked multicenter trial assigned 450 patients to 0.05% or 0.09% cyclosporine drops for three months. The higher strength showed larger improvements in staining, osmolarity, Schirmer measurements, and reported symptoms, while the lower strength was more comfortable. Neither concentration clearly outperformed on tear breakup time or conjunctival cell measures.

10. Tanfanercept's primary result was not significant

    A masked, placebo-controlled phase 2 study tested the TNF inhibitor tanfanercept in 100 patients for eight weeks, including a controlled dry-eye challenge. The primary inferior corneal staining difference was small and not statistically significant; other between-group signs also missed significance. A subgroup of participants age 35 or older suggested a possible signal, but it needs a new trial designed to test it.

11. TearCare and LipiFlow compared directly

    A randomized, assessor-masked trial assigned 135 patients with gland-related dry eye to a single TearCare or LipiFlow procedure. Tear breakup, gland secretion, and symptoms improved at one month in both arms, without a clear overall between-treatment advantage. An active-comparator trial with short follow-up cannot establish how long either effect lasts.

12. Finger-prick blood drops for severe dry eye

    A two-center, single-masked feasibility trial randomized 60 patients to usual treatment with or without added finger-prick autologous blood drops for eight weeks. Symptom scores improved more with the added drops, while the between-group differences in staining and tear breakup were not significant. The small sample and lack of placebo drops call for a larger masked trial before routine use.

Population patterns and patient experience

Administrative records can follow large groups but depend on diagnosis codes. Surveys capture symptoms in people who may never see a clinician, making their prevalence estimates a different kind of measure.

13. Diabetes drug groups and later dry-eye diagnoses

    Researchers compared Taiwanese records for 10,038 new SGLT2-inhibitor users with 1,077 new GLP-1-receptor-agonist users who had type 2 diabetes. After adjustment, diagnosed dry eye was less frequent in the SGLT2 group, with a hazard ratio of 0.78. Treatment choice and unmeasured patient differences could still explain part of this association; it was not a randomized drug comparison.

14. Dry eye within a large type 2 diabetes cohort

    A Taiwanese health-record study compared 10,029 patients with diagnosed dry eye with 142,491 patients with type 2 diabetes who did not develop that diagnosis. Poorer glucose control, neuropathy, retinopathy, and several eye procedures were associated with dry eye. Medication associations also appeared, but prescribing and disease severity can confound a retrospective comparison.

15. Fifteen years of diagnosed dry-eye incidence in Taiwan

    Using nationwide insurance data, investigators tracked new coded dry-eye cases from 2001 through 2015. Age-adjusted annual incidence rose in both sexes and remained higher among women: from 2.06 to 4.91 per 1,000 women and 0.97 to 2.52 per 1,000 men. Changes in care seeking or coding could contribute to the trend, which measures diagnoses rather than all disease.

16. Dry eye and gland findings after age 85

    The population-based Ural Very Old Study examined about 1,500 adults aged 85 or older. Roughly 14.5% met a combined symptom-and-Schirmer definition of dry eye, while some degree of meibomian gland dysfunction was common. Different case definitions produced different estimates, and the results should not be generalized to younger adults or other settings.

17. A community survey in São Paulo

    A questionnaire survey of 582 adults in eastern São Paulo found that 24.4% reported severe dry-eye symptoms or a previous clinical diagnosis. The combined measure was more common among women; age and hypertension also showed associations in women. A self-reported symptom-or-history definition cannot be read as the prevalence of examination-confirmed dry eye.

18. Dry-eye symptoms among Shanghai high-school students

    A cross-sectional survey of 4,825 students during the COVID-19 period found that 70.5% scored at least 13 on the OSDI symptom scale. Longer screen exposure, poorer sleep, stress, and female sex were associated with higher symptom scores. The striking percentage reflects a questionnaire threshold in a specific setting, not an eye-exam diagnosis or proof that screens caused the symptoms.

19. How patients actually used dry-eye drops

    In a web survey of 2,645 people using dry-eye drops, only 10.2% reported instilling them at the frequency stated in the package instructions. Many used drops when symptoms appeared, and more regular users reported greater symptom improvement. Self-selection and self-report prevent the survey from showing that dosing frequency itself caused the difference.

20. Depression and dry-eye severity in DREAM

    A secondary analysis followed 535 participants in the multicenter DREAM trial. Those screening positive for depression had worse dry-eye symptoms and a modestly worse combined sign score; measured inflammatory markers did not differ clearly. The depression screen was not a clinical diagnosis, and the association does not establish which condition influenced the other.

21. Five statistical dry-eye profiles in DREAM

    Researchers reanalyzed baseline symptoms and signs from 535 DREAM participants and identified five profiles, including severe signs with milder symptoms and the reverse pattern. Omega-3 was not significantly better than placebo in any profile. The subtypes are exploratory groups derived from one trial cohort and need validation in other populations.

22. Health anxiety and emotional symptoms in clinic patients

    Among 206 Chinese clinic patients with dry eye, higher OSDI symptom scores and health-anxiety scores were associated with depression and anxiety questionnaire results. Statistical mediation suggested health anxiety might contribute to the relationship. Because the survey captured all measures at one time, it cannot determine a causal sequence.

Measurement, mechanisms, and surgery-related effects

New imaging and molecular tools can expose useful patterns, but most of these studies are early validation or small comparisons. A reproducible association is still needed before a marker becomes a clinical test.

23. A smartphone app for dry-eye screening

    A Japanese single-center study compared app-based symptom and maximum-blink-interval measures with clinic testing in 82 adults, including 42 with dry eye. The combined app score discriminated the groups well within this sample, with an area under the curve of 0.91. Its lower negative predictive value and lack of external testing limit its role as a stand-alone diagnosis.

24. Incomplete blinking measured by video analysis

    Investigators trained a deep-learning blink model using videos from 50 dry-eye patients and 50 controls. Incomplete blinks were more frequent in the dry-eye group, and analysis at 30 frames per second performed better than lower-rate video. The associations with symptoms and signs were modest; the model needs validation beyond this case-control sample.

25. Dry eye made cataract biometry less repeatable

    Before cataract surgery, researchers measured 114 eyes twice on the same day and compared results with tear and surface findings. Corneal curvature measurements were less repeatable in eyes with dry eye; shorter tear breakup time was associated with greater variation in curvature and axial length. The observational design identifies a planning concern but does not test whether treatment improves final refractive outcomes.

26. Tear microRNAs: an early discovery sample

    Tear extracellular vesicles from five dry-eye patients and five controls were sequenced for microRNAs. The investigators reported 126 differentially expressed candidates, including several linked by pathway analysis to inflammation. With only ten participants and many comparisons, this is a discovery result rather than a validated diagnostic panel.

27. Where the tear film breaks first in Sjögren dry eye

    A keratograph study compared 22 patients with Sjögren-associated dry eye with 22 patients whose dry eye had another cause. First tear-film breakup occurred in an inferior region more often in the Sjögren group, a borderline statistically significant finding based on both eyes of a limited sample. Location alone has not been validated as a way to identify Sjögren disease.

28. Conjunctival bacteria differed across dry-eye groups

    Researchers sequenced conjunctival swabs from 23 Sjögren dry-eye eyes, 36 other dry-eye eyes, and 39 control eyes. Diversity was lower in both dry-eye groups, and some relative bacterial abundances differed. This small cross-sectional microbiome comparison cannot tell whether those differences contribute to disease or follow from it.

29. Tear cytokines after two gland treatments

    A post hoc analysis of a randomized trial measured 20 tear cytokines in 13 patients receiving IPL plus gland expression and 13 receiving warm compresses plus expression. Many markers declined in both groups, with larger reductions in several after IPL. Small groups and multiple biomarker tests make the mechanistic comparison preliminary.

30. LASIK versus SMILE: corneal sensation and symptoms

    Forty myopic patients had one eye randomized to LASIK and the other to SMILE, with follow-up for a year. LASIK eyes lost more corneal sensitivity early, but both procedures returned to similar baseline sensitivity by 12 months; reported dry-eye symptoms did not differ. Whole-person symptom scores cannot be assigned cleanly to one operated eye.

31. An ocular surface analyzer in a clinic sample

    A hospital study compared noninvasive analyzer measurements with clinical meibomian gland assessments in 113 consecutive patients. Gland loss helped distinguish diagnosed gland dysfunction within this sample, with an area under the curve of 0.827. The clinic was not a population sample, and the proposed cutoff needs independent validation.

32. Gland signs and patient symptoms did not track closely

    Investigators reanalyzed baseline data from 122 meibomian gland dysfunction patients recruited for an IPL trial. SPEED symptom scores did not correlate meaningfully with tear breakup, staining, gland secretion, or lid-margin signs, including after adjustment for selected factors. Patients entered a treatment study, so the result may not describe every clinic population.