IPL, LipiFlow, Rexon-Eye and more: what 47 trials say when you compare them side by side

If you have meibomian gland dysfunction (MGD), the most common driver of evaporative dry eye, you may have heard of in-clinic devices such as intense pulsed light (IPL), LipiFlow, Rexon-Eye and others. How do they compare? A 2026 network meta-analysis put 47 randomised trials side by side. Its findings are useful, but need careful reading.

What the researchers did

Noyman and colleagues searched PubMed, Scopus and Embase up to July 23, 2025, and included 47 randomised controlled trials involving 3581 patients with MGD from 16 countries, including New Zealand. Every trial had a device-based treatment in at least one arm. Individual studies ranged from 14 to 227 patients, and mean ages ranged from 24 to 75 years. The review was registered in advance (PROSPERO), and the authors declared no funding and nothing to disclose.

They looked at results 2 to 4 months after treatment started, across five outcomes: tear break-up time (TBUT, how long the tear film stays stable), Schirmer testing (tear production), corneal staining (surface damage), symptoms, and meibomian gland quality. Fourteen distinct devices were included, alongside non-device options such as meibomian gland expression (MGX) and acupuncture. The main reference point was "conservative treatment", a mixed group covering warm compresses, lid hygiene, massage, artificial tears and combinations of these.

What is a network meta-analysis?

Few trials compare one device directly with another. A network meta-analysis links trials through shared comparators. If one trial compared IPL with warm compresses and another compared LipiFlow with warm compresses, the method can estimate how IPL and LipiFlow compare indirectly. Those indirect estimates are only as good as the trials feeding them.

What a P-score ranking is, and isn't

The review ranked treatments using P-scores, which run from 0 to 1. A P-score is roughly the average certainty that a treatment beats the others in the network. It is a ranking, not a measure of how much better a treatment is. A treatment tested in one small trial, with a very wide margin of uncertainty, can still earn a high P-score.

The authors note that rankings convey relative probabilities, not absolute clinical value. Rankings also shifted a lot by outcome. Quantum molecular resonance (QMR, the technology used in devices such as Rexon-Eye) ranked first for corneal staining but 14th for meibomian gland quality.

The headline results

• Tear break-up time (43 studies): TearCare combined with MGX (P-score 0.924) and IPL combined with a daily heated eye mask (0.898) ranked highest. They were the only treatments significantly better than conservative treatment.
• Schirmer (14 studies): no treatment significantly outperformed conservative treatment.
• Corneal staining (28 studies): QMR was the only option significantly better than conservative treatment, with a standardised mean difference (SMD) of −4.07 (95% CI −6.19 to −1.96). LipiFlow (−0.30; −1.53 to 0.92) and IPL alone (−0.15; −1.65 to 1.35) were not significantly different from conservative treatment.
• Symptoms (43 studies): IPL with a heated eye mask ranked first (P-score 0.910). TearCare with MGX and LipiFlow were significantly better only than the "control" group, not than conservative treatment.
• Gland quality (37 studies): TearCare with MGX ranked first, with an SMD of −18.43 (−21.17 to −15.70). IPL alone (−0.64; −1.69 to 0.41) and LipiFlow (−0.62; −1.39 to 0.15) did not differ significantly from conservative treatment. In direct trials, IPL plus MGX did better than MGX alone (−1.65; −2.84 to −0.46).

An SMD expresses a difference in units of spread, so results measured on different scales can be combined; negative values here favour the device. The 95% confidence interval (CI) is the range of values the data are reasonably compatible with. If it crosses zero, "no difference" remains a plausible explanation.

The paper contains some internal inconsistencies. For example, QMR's corneal staining P-score is 0.977 in the text but 0.971 in the table, and the text lists fewer non-significant QMR comparisons than the table shows.

Why the top rankings need caution

Several top-ranked options rest on a single trial. TearCare with MGX is one trial of 24 patients, all female, funded by the device maker (Sight Sciences). IPL with a heated eye mask is one trial, funded by the hospital, in a young group with mean ages between 28.06 and 29.88 years. TES, ranked first for Schirmer, is one trial of 24 people. QMR rests on two trials, one of which used a low-power QMR "sham".

The SMD of −18.43 for gland quality is far outside the usual range (an SMD above about 2 is unusual), which suggests a data or scale problem.

IPL and LipiFlow, by contrast, each featured in 14 direct comparisons for gland quality. Their middle-of-the-table rankings may partly reflect more data and narrower uncertainty, rather than poorer performance. We've covered thermal pulsation in more depth in our post on a manual device versus LipiFlow trial, and light-based options in our low-level light therapy evidence review.

What the studies can't tell us

• Very high heterogeneity. I², a measure of how much study results disagree beyond chance, was 98.6% for TBUT, 97.2% for corneal staining, 89.8% for symptoms and 97.3% for gland quality.
• Short follow-up. A 2–4 month window says little about how long benefits last, how often retreatment is needed, or cost-effectiveness, as the authors acknowledge.
• Unmasked designs. Participants usually knew which treatment they received, which matters most for symptom scores.
• Industry involvement. 30% of included studies (14) had industry funding or industry-linked authors.
• Safety is unclear. Only 20 studies (42.5%) reported adverse events in any way, so safety could not be compared. "No serious adverse events" reflects limited reporting, not proven safety.
• Mixed comparators and data handling. "Conservative treatment" varied widely, some values were digitised from graphs or imputed, and several LipiFlow trials were in people preparing for cataract surgery.
• No publication-bias assessment was reported, and signs such as TBUT and staining often don't match how eyes feel.

What this means for you

This review doesn't crown a winner. At 2–4 months, most devices didn't clearly beat good conservative care on most measures, and the standout rankings largely come from single small trials. It does suggest that combining approaches, such as IPL alongside gland expression, may help some people, and that the right choice depends on your own glands, lids and tear film.

That's why we focus on treating the cause. You can read about the options on our treatments page and our IPL therapy page. At Rose Optometry in Hamilton, our optometrists, including Jagrut Lallu and Jacqueline Rowe, can assess your meibomian glands and talk through whether any in-clinic treatment may suit you. To get started, book a Dry Eye Evaluation at a member practice.

References

• Noyman DBE, Chan CC, Teichman JC, et al. Technological Interventions for Dry Eye Disease: A Systematic Review and Random-Effects Network Meta-analysis of 3-Month Outcomes. Ophthalmology and Therapy. 2026;15(5):1721-1759. PMID 41933265. DOI: 10.1007/s40123-026-01360-x