Interventions for Dry Eye An Overview of Systematic Reviews

Abstract

IMPORTANCE

Dry eye is a common ocular disease that can have substantial morbidity. Systematic reviews provide evidence for dry eye interventions and can be useful for patients, clinicians, and clinical guideline developers. Overviews of reviews use explicit and systematic methods to synthesize findings from multiple systematic reviews, but currently, there are no overviews of systematic reviews investigating interventions for dry eye.

OBJECTIVE

To summarize the results of reliable systematic reviews of dry eye interventions and to highlight the evidence gaps identified.

EVIDENCE REVIEW

We searched the Cochrane Eyes and Vision US satellite database and included reliable systematic reviews evaluating dry eye interventions published from 2016 to 2022. We reported the proportion of systematic reviews that were reliable with reasons for unreliability. Critical and important outcomes from reliable systematic reviews were extracted and verified. Critical outcomes included dry eye–related patient-reported outcome measures. Results were synthesized from reliable systematic reviews to provide summaries of evidence for each intervention. Evidence for each intervention was defined as conclusive or inconclusive depending on whether high-certainty evidence across systematic reviews was available according to Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) criteria and whether findings reached statistical or clinical significance. Recommendations were made for further research.

FINDINGS

Within the Cochrane Eyes and Vision US satellite database, 138 potentially relevant systematic reviews were identified, 71 were considered eligible, and 26 (37%) were assessed as reliable. Among reliable systematic reviews, no conclusive evidence was identified for any dry eye intervention. Inconclusive evidence suggested that environmental modifications, dietary modifications, artificial tears and lubricants, punctal occlusion, intense pulsed light therapy, vectored thermal pulsation therapy (Lipiflow), topical corticosteroids, topical cyclosporine A, topical secretagogues, and autologous serum may be effective. Only unreliable systematic reviews evaluated lifitegrast, oral antibiotics, and moisture chamber devices.

CONCLUSIONS AND RELEVANCE

This overview of systematic reviews found some evidence that dry eye interventions may be effective, but no conclusive evidence was available. The conduct and reporting of most systematic reviews for dry eye interventions warrant improvement, and reliable systematic reviews are needed to evaluate lifitegrast, oral antibiotics, and moisture chamber devices.

Dry eye disease is a multifactorial condition characterized by inflammation and a loss of tear film homeostasis.¹ Symptoms such as dryness, grittiness, burning, and visual fluctuations reduce quality of life and impede productivity.^2,3 Herein, we use the term dry eye to collectively describe all etiologies of dry eye disease.

To inform clinical practice, the 2018 American Academy of Ophthalmology (AAO) Preferred Practice Pattern (PPP) on dry eye recommended treatments including environmental modifications, artificial tears and lubricants, dietary modifications (eg, polyunsaturated fatty acid [PUFA] supplementation), in-office procedures (eg, punctal occlusion, intense pulsed light [IPL] therapy, vectored thermal pulse therapy [Lipiflow]), prescription-only topical therapies (eg, topical corticosteroids, cyclosporine A, lifitegrast, secretagogues), blood-derived serum, and other surgical and nonsurgical approaches.^4,5

Systematic reviews inform trustworthy guideline recommendations.⁶ Nevertheless, in the context of the 2018 AAO PPP on dry eye, the endorsed treatments were underpinned by limited evidence from systematic reviews.^4,7 Subsequent to this publication, more systematic reviews concerning dry eye treatments have emerged. Overviews of systematic reviews use explicit and systematic methodology to identify and summarize evidence from multiple systematic reviews encompassing all interventions for a common condition. Overviews distil evidence into accessible summaries pertinent to health care decisions, sparing evidence users the task of assimilating numerous systematic reviews.⁸

We conducted an overview of systematic reviews concerning interventions for dry eye. We aimed to evaluate their reliability, summarize findings from reliable reviews, map evidence from reliable reviews to 2018 AAO PPP intervention categories and previously prioritized research questions, and propose needs for further primary and secondary research.

Methods

We conducted this overview, registered on PROSPERO (CRD42021279880), from May 1, 2021, to August 1, 2023. The overview adhered to a previously published protocol and contemporary methodology^8–10 and was reported in accordance with the Preferred Reporting Items for Overviews of Reviews (PRIOR) statement.¹¹ Comprehensive methods and protocol deviations are provided in the eMethods in the Supplement.

Eligibility Criteria

We included systematic reviews that investigated interventions for dry eye published after January 1, 2016, in any language. Systematic reviews published more than 5 years ago are unlikely to be up to date.^12,13 We accepted any definition of dry eye as used by review authors. Systematic reviews were defined as full-text reports self-identified as such or as meta-analysis and those aligning with the Institute of Medicine’s criteria.¹⁴ Both systematic reviews of randomized clinical trials (RCTs) and nonrandomized studies of interventions were eligible.¹⁵ Nonrandomized studies of interventions in systematic reviews were only eligible to provide evidence for harms outcomes. We excluded systematic reviews investigating blepharitis, allergic keratoconjunctivitis, and infectious keratoconjunctivitis.

We compared active interventions with no treatment, placebo, standard of care, or other active treatments, as reported in the systematic reviews. We collected critical and important outcomes, including adverse effects (or harms), in prespecified formats or the format described by the systematic reviews (eTable 1 in the Supplement). Our critical outcome was dry eye patient-reported outcome measures because these outcomes are prioritized by patients.¹⁶ Our important outcomes included clinical signs because these are important in evaluating dry eye treatment. We classified important harms findings into standardized, hierarchical terminology using Medical Dictionary for Regulatory Activities (MedDRA).¹⁷ We collected the outcomes within the following time spans: 3 months or less (short term), beyond 3 months to 6 months (medium term), and beyond 6 months (long term).

Search Methods

We searched the Cochrane Eyes and Vision US satellite database of eyes and vision systematic reviews using relevant keywords, controlled vocabulary terms, and search filters.¹⁸This database was most recently updated on August 1, 2022 (eTable 2 in the Supplement).

Selection of Reviews

Two investigators independently evaluated titles and abstracts and then full-text reports for relevance in Covidence.¹⁹ We resolved discrepancies in the screening by consensus. We reported the reasons for full-text exclusions.

Assessment of Reliability

We assessed the reliability of relevant reviews against 5 established methodological criteria (eTable 3 in the Supplement).^7,20–24 If at least 1 of the criteria was not met, we considered the systematic review unreliable and excluded it from the overview synthesis. We used a single assessor with verification by an independent investigator for all assessments and resolved discrepancies by consensus.²⁵

Data Collection

One investigator collected details regarding the descriptive characteristics of each systematic review, risk of bias, and certainty of evidence using Systematic Review Data Repository Plus with verification by another investigator.²⁶ Risk of bias is an assessment of the internal validity of a primary study across several domains. Certainty-of-evidence assessments gauge the level of confidence in the evidence and whether additional primary studies could alter the effect estimates using Grading of Recommendations, Assessment, Development, and Evaluations (GRADE).²⁷

Evidence Synthesis

We used data integration tables to present characteristics of included reviews, description of main features, risk-of-bias assessments, conclusions, certainty of evidence, and results for critical and important outcomes.²⁸ We extracted meta-analysis results from direct pairwise comparisons and described the direction of effect when results demonstrated statistical significance.¹¹ In cases without meta-analysis, we reported the primary study count for each result. We qualitatively presented important results of indirect comparisons from network meta-analysis (NMA) in the narrative synthesis.

We summarized the evidence for each outcome into an evidence sufficiency statement that categorized whether there was sufficient evidence, some evidence, or insufficient evidence of effectiveness between intervention and comparison groups (Table 1).²⁸ We incorporated GRADE-recommended language for the corresponding certainty of evidence within the evidence sufficiency statements.²⁹ We stated whether the evidence for the critical outcome was conclusive or inconclusive for each intervention, with high-certainty evidence required to provide conclusive evidence (Table 1). We stated whether the evidence for each intervention addressed a recommendation from the 2018 AAO PPP on dry eye,⁴ a previously prioritized research question,^16,30 or an intervention not previously prioritized.

Table 1.

Evidence Criteria

Evidence	Criteria
Across systematic reviews a
Conclusive	Consistent high certainty of evidence from GRADE assessments across systematic reviews.
Not conclusive	Inconsistent evidence across systematic reviews or consistent evidence with lower than high certainty of evidence from GRADE assessments.
Within each systematic review b
Sufficient	evidence Evidence sufficiently certain to support the effect of the intervention in relation to a specific outcome. This includes evidence of an effect in terms of (1) benefit or (2) harm. Criteria that need to be met: • High certainty of evidence from GRADE assessment; and • Statistically significant findings of a magnitude considered to be at least of minimalclinically important difference; or • A narrative synthesis of results, with the result qualified according to the review findings,eg, “evidence (eg, 12 studies of 14) reported a positive effect of ….”
Some evidence	Less certain evidence about the effects of a particular intervention in relation to a specific outcome. Criteria that need to be met: • Less than high certainty of evidence from GRADE assessment or GRADE assessment not performed; and • Statistically significant results that may or may not reach minimal clinically importantdifference; or • A narrative synthesis of results, with the result qualified according to the review findings,eg, “some evidence (eg, 5 studies of 9) reported a positive effect of ….”; or • A rating of some evidence may also be based on a statistically significant result obtained ina small number of trials or a statistically significant result obtained from trials with a small number of participants.
Insufficient evidence	Not enough evidence to support the effects of the intervention based on the included studies. This should be interpreted as no evidence of effect rather than evidence of no effect. Where the number of trials and/or participants is small, insufficient evidence may reflect lack of power to be able to detect an effect of the intervention. Where the number of trials and/or participants is large, insufficient evidence may reflect underlying ineffectiveness of the intervention. Criteria that need to be met: • Any level of certainty from GRADE assessment or GRADE assessment not performed; and • Statistically nonsignificant results.
GRADE Working Group grades of evidence
High certainty	Further research is very unlikely to change the confidence in the estimate of effect. GRADE-recommended language example: “X results in a reduction/increase in outcome.”
Moderate certainty	Further research is likely to have an important impact on the confidence in the estimate of effect and may change the estimate. GRADE-recommended language example: “X probably results in a reduction/increase in outcome.”
Low certainty	Further research is very likely to have an important impact on the confidence in the estimate of effect and is likely to change the estimate. GRADE-recommended language example: “X may result in a reduction/increase in outcome.”
Very low certainty	The estimate is very uncertain. GRADE-recommended language example: “X may result in a reduction/increase in outcome, but the evidence is very uncertain.”

Abbreviation: GRADE, Grading of Recommendations, Assessment, Development, and Evaluations.

^aStatements summarizing critical outcomes for each intervention across systematic reviews.

^bStatements summarizing each outcome within a systematic review; no statements made for within-group comparisons with baseline.

We quantified the overlap between systematic reviews, as recommended by the PRIOR statement, by calculating the adjusted corrected covered area.³¹ This measurement quantifies the degree of primary study overlap between systematic reviews and facilitates the exploration of reasons for potentially discordant conclusions between reviews. We displayed overlap using the Graphical Representation of Overlap for Overviews (GROOVE) tool.

Results

Our search yielded 138 unique reports, from which we included 71 eligible reviews for reliability assessment (eFigure in the Supplement). We judged 26 of 71 (37%) reviews to be reliable.^32–57 Reasons for unreliability are shown in eTable 4 in the Supplement. The most common reasons for unreliability were a lack of a comprehensive search (28 of 45 [62%]) and inappropriate meta-analysis (18 of 45 [40%]). Methodological characteristics of reliable systematic reviews are outlined in eTable 5 in the Supplement. GROOVE tools for primary study overlap are in eAppendixes 1–7 in the Supplement.

Summaries of reliable systematic reviews are described in Table 2.^32–57 Data integration tables for each systematic review are outlined in eTable 6 to eTable 31 in the Supplement. Table 3^32–52, 54,55,57 outlines the main findings of the critical outcomes and adverse effects for all dry eye interventions.^4,16 Detailed narrative synthesis of evidence for interventions recommended by the 2018 AAO PPP on dry eye are provided in the eResults in the Supplement.

Table 2.

Characteristics of 26 Reliable Systematic Reviews on the Management of Dry Eye

Source	Population	Intervention	Comparators	Outcomes	No. of studies	No. of databases and platforms searched	No. of participants and eyes	Cochrane RoB tool	≥1 Meta-analysis	Publication bias reported for primary outcome	Certainty of evidence assessed
Byber et al,32 2021	Adults (≥18 y) working in buildings in any occupational sector and in any professional activity, children (preschool and school-age children in an educational setting: kindergarten/preschool/nursery school, daycare centers, primary school), and adolescents and young adults (≤30 y) in an educational setting (college, high school/university)	Environmental humidification	No or alternative intervention	DED PROs, harms (prespecified)	12 Studies (1 RCT and 1NRSI relevant to overview)	9	4551 Participants (807 participants relevant to overview)	1.0	No	No	Yes
Chi et al,36 2019	Patients with typical nonspecific DED	ω-3 ± ω-6 PUFAs	Control	DED PROs, Schirmer test, TBUT, tear osmolarity	13 RCTs	4	1782 Participants	1.0	Yes	Yes, absent	No
Chu et a 1,54 2020	Adult patients with pSS	Immunosuppressive therapy	Placebo or active drug therapy	DED PROs, Schirmer test, harms (prespecified)	32 RCTs (12 RCTs relevant to overview)	3	1988 Participants (1148 participants relevant to overview)	1.0	Yes	No	No
Cote et al,40 2020	Adults (≥18 y) with MGD or evaporative DED	1PL therapy	Standard therapy (eg, warm compresses), placebo therapy (eg, sham IPL), no treatment	DED PROs, Schirmer test, TBUT, ocular surface dye staining, tear osmolarity, meibomian gland function, harms (not prespecified)	3 RCTs	3	114 Participants, 228 eyes	1.0	Yes	No	Yes
de Paiva et al,43 2019	Participants with dry eye (various etiologies, participants of any age, sex, chronicity, severity, or classification of dry eye, such as Sjögren or non-Sjögren dry eye)	Topical formulations of CsA or CsA in combination with artificial tears	Head-to-head comparison with other CsA formulations, artificial tears, placebo therapy, vehicle	DED PROs, Schirmer test, TBUT, ocular surface dye staining, artificial tear use, conjunctival goblet cell density, harms (not prespecified)	30 RCTs	4	4009 Participants	1.0	Yes	No	Yes
Downie et al,37 2019	Populations with dry eye regardless of age, gender, severity of disease, or classification of dry eye (eg, not specifically aqueous deficient, evaporative tear deficiency)	ω-3 ± ω-6 PUFAs in combination with artificial tears or not	Artificial tears; placebo; head-to-head comparisons that compared ω-3 and/or ω-6 PUFA interventions of any form; dose (concentration and frequency), or route of administration (eg, dietary intake, supplements, eye drops) vs other forms, doses, or routes of administration; no treatment	DED PROs, Schirmer test, TBUT, tear osmolarity, artificial tear use, conjunctival goblet cell density, harms (prespecified)	34 RCTs	4	4314 Participants	1.0	Yes	No	Yes
Ervin et al,39 2017	Symptomatic participants diagnosed with aqueous tear deficiency or keratoconjunctivitis sicca (DES); no restrictions in age, sex, comorbidities, or use of adjunctive therapy	Punctal plugs (intracanalicular punctal plugs alone, silicone punctal plugs, collagen ± silicone punctal plugs), punctal plugs ± artificial tears, lower and upper punctal plugs, acrylic and silicone punctal plugs, collagen and silicone punctal plugs	Standard therapy (eg, artificial tears, pilocarpine, cyclosporine, diathermy), placebo therapy (eg, sham treatment), no treatment	DED PROs, Schirmer test, TBUT, ocular surface dye staining, artificial tear use, harms (prespecified)	18 RCTs	5	711 Participants, 1249 eyes	1.0	Yes	No	Yes
Franchini et al,47 2 0 1 9	Ocular surface disease (in systematic review), dry eye (in meta-analysis)	Autologous serum eye drops	Not reported	DED PROs, Schirmer test, TBUT, ocular surface dye staining	19 RCTs (10 RCTs relevant to overview)	4	729 Participants (353 participants relevant to overview)	1.0	Yes	No	Yes
Giannaccare et al,35 2019	Patients having DED of any etiology	Oral supplementation with ω-3 ± ω-6 PUFA	Placebo	DED PROs, Schirmer test, TBUT, ocular surface dye staining	17 RCTs	4	3363 Participants	1.0	Yes	Yes, absent	No
Hao et al49 2020	Women diagnosed with menopause for ≥1 y not taking any HT before the study and no contraindications to HT	Topical or systemic HT containing estrogen or estrogen + progestogen	Placebo	DED PROs, TBUT, Schirmer test, ocular surface dye staining, harms (prespecified)	9 RCTs	5	612 Participants	1.0	Yes	Yes, absent	No
Jongkha-jornpong et al,45 2022	Patients with dry eye aged ≥18y	Biological tear substitutes or topical secretagogues	Placebo, artificial tears, other topical dry eye treatments	DED PROs, TBUT, Schirmer test, harms (not prespecified)	39 RCTs	3	3693 Participants	2.0	Yes	Yes, absent	Yes
Kim et al,55 2018	Patients with DES excluding patients with Sjögren syndrome	Acupuncture ± artificial tears	Artificial tears	DED PROs, Schirmer test, TBUT, ocular surface dye staining, harms (not prespecified)	19 RCTs	10	1126 Participants	1.0	Yes	No	No
Letaief et al,53 2018	Primary Sjogren syndrome	b-DMARDS modulating B cells	Not reported	Schirmer test, harms (not prespecified)	5 RCTs (2 RCTs relevant to overview)	4	573 Participants (152 participants relevant to overview)	1.0	Yes	No	No
Liuetal,42 2021	RCT participants (adult and children) with clinically diagnosed DED regardless of etiology or severity	Topical steroids (betamethasone, clobetasone butyrate, dexamethasone, difluprednate, fluorometholone, loteprednol etabonate, prednisolone)	Artificial tears, placebo therapy, other steroidal or nonsteroidal therapy or a combination of therapies, no treatment	DED PROs, Schirmer test, TBUT, ocular surface dye staining, tear osmolarity, harms (prespecified)	22 RCTs	4	4169 Participants	2.0	Yes	No	Yes
Na et al,57 2021	Adult patients with DES	Acupuncture or its combination with artificial tears	Artificial tears a lone	DED PROs, Schirmer test, TBUT, ocular surface dye staining	21 RCTs	7	1214 Participants, 1542 eyes	1.0	Yes	No	No
Pan et al,46 2017	Adults (>18 y) with dry eye	Application of AS alone or in combination with artificial tears	Artificial tears a lone, saline, placebo, no treatment	DED PROs, Schirmer test, TBUT, ocular surface dye staining, tear osmolarity, impression cytology, nonspecific PROs, harms (prespecified)	5 RCTs	5	92 Participants, 149 eyes	1.0	No	No	Yes
Pang et al,41 2019	Patients with MGD	Vectored thermal pulsation therapy (Lipiflow)	Warm compress treatment	DED PROs, meibomian gland function, Schirmer test, TBUT, ocular surface dye staining, tear osmolarity, harms (not prespecified)	4 RCTs	3	385 Participants	2.0	Yes	No	No
Pucker et a I,33 2016	Adult participants with dry eye (various etiologies)	OTC artificial tears	Another class of OTC artificial tears, placebo therapy (eg, saline or vehicle), no treatment	DED PROs, Schirmer test, TBUT, ocular surface dye staining, tear osmolarity, BCVA, harms (not prespecified)	43 RCTs	4	3497 Participants	1.0	Yes	Yes, present	Yes
Ribeiro et al,34 2019	Individuals aged ≥18 y with dry eye (various etiologies)	Preservative-free artificial tears	Preserved artificial tears	DED PROs, Schirmer test, TBUT, ocular surface dye staining, tear osmolarity, BCVA, impression cytology, harms (not prespecified)	4 RCTs	4	323 Participants	1.0	No	No	No
Singh et al,38 2022	Male or female participants of any age with a diagnosis of CVS	Any intervention for managing signs or symptoms of CVS	Inactive control treatment, placebo treatment, sham treatment, no treatment	DED PROs, harms (not prespecified)	45 RCTs (18 RCTs relevant to overview)	3	4497 Participants (2111 participants relevant to overview)	1.0	Yes	No	Yes
Souza et al,52 2016	Participants aged >18 y and with an established pSS diagnosis according to the 2002 American-European Revised Classification Criteria	Rituximab	Not reported	DED PROs, ocular surface dye staining, Schirmer test, harms (prespecified)	4 RCTs (2 RCTs relevant to overview)	3	276 Participants (150 participants relevant to overview)	1.0	Yes	No	Yes
Tuan et al,44 2020	Patients with DED	Topical CsA or CsA in combination with artificial tears	Artificial tears only	DED PROs, TBUT, Schirmer test, ocular surface dye staining, harms (not prespecified)	11 RCTs	5	1085 Participants	1.0	Yes	No	No
Wang et al,48 2020	Patients with DED	Autologous serum	Artificial tears with or without combination of other therapy	DED PROs, Schirmer test, TBUT, ocular surface dye staining	7 RCTs	5	267 Participants	1.0	Yes	Yes, absent	No
Wang and Deng,50 2020	Patients clinically diagnosed with DED	Androgen replacement therapy	Not reported	DED PROs, Schirmer test, TBUT, ocular surface dye staining, tear osmolarity, harms (prespecified)	RCTs (3 RCTs relevant to overview), 4 NRSIs (0 NRSIs relevant to overview)	3	245 Participants (124 participants relevant to overview)	1.0	No	No	No
Yang et al,56 2020	Patients diagnosed with DED, aged ≥18 y	Qiming granule ± artificial tears	Artificial tears, topical NSAIDs	Schirmer test, TBUT, ocular surface dye staining, harms (not prespecified)	11 RCTs	7	1032 Participants	1.0	Yes	No	No
Zhou et al,51 2018	Patients with dry eye	Pranoprofen monotherapy or combination therapy	Placebo or other positive control (artificial tears, hormones, or NSAIDs)	TBUT, Schirmer test, ocular surface dye staining, harms (not prespecified)	15 RCTs	6	2873 Eyes	1.0	Yes	No	No

Abbreviations: AS, autologous serum; BCVA, best-corrected visual acuity; b-DMARDS, biologic disease modifying antirheumatic drugs; CsA, cyclosporine A; CVS, computer vision syndrome; DED, dry eye disease; DES, dry eye syndrome; HT, hormone therapy; IPL, intense pulse light; MGD, meibomian gland dysfunction

NRSI, nonrandomized study of intervention; NSAIDs, nonsteroidal anti-inflammatory drugs; OTC, over the counter; PROs, primary reported outcomes; pSS, primary Sjogren syndrome; PUFAs, polyunsaturated fatty acids; RoB, risk of bias; TBUT, tear breakup time.

Table 3.

Critical Outcome and Adverse Effects or Harms Results of Direct Comparisons for Dry Eye Interventions

Intervention and source	Date of search	Main findings on the interventions with relevant time points noted	Certainty of evidence for main findings	Research recommendation
Environmental modifications (AAO step 1)		2 Reviews were reliable, and main findings were classified as effective–not conclusively.		New primary research (such as RCTs)
Byber et al,32 2021	December 9, 2020	Indoor air humidification may improve dry eye–specific patient-reported outcome measures (at 6 wk) compared with no humidification. (S)	Low
		Indoor air humidification may cause increased “perception of stuffiness” (at 6 wk) compared with no humidification. (S)	Low
Singh et al,38 2022	December 12, 2019	Moist cool air devices may result in little to no difference in dry eye–specific patient-reported outcome measures (at 5 d) compared with no treatment, but the evidence is very uncertain. (S)	Very low
		Blue light blocking lenses probably result in little to no difference in dry eye–specific patient-reported outcome measures (at 1–5 d) compared with non–blue light blocking lenses. (S)	Moderate
		Visual hygiene exercises (ie, 20–20-20 rule) may result in little to no difference in dry eye–specific patient-reported outcome measures (at 20 d) compared with placebo, but the evidence is very uncertain. (S)	Very low
		Multifocal correction lenses may result in little to no difference in dry eye–specific patient-reported outcome measures (at 1 wk) compared with single vision correction lenses, but the evidence is very uncertain. (S)	Very low
Artificial tears and lubricants (AAO step 1)		2 Reviews were reliable, and main findings were classified as effective–not conclusively.		New primary and secondary research (such as RCTs or systematic reviews investigating preserved vs preservative-free drops and lipid-based artificial tears and lubricants)
Pucker et al,33 2016 (results of 9 select comparisons from 28 total comparisons)	December 4, 2015	0.2% Polyacrylic alcohol (Viscotears) may improve dry eye–specific patient-reported outcome measures (at 3, 4, and 6 wk) compared with 1.4% polyvinyl alcohol. (S)	Not reported
		Carbomer-containing viscous gel (Lubrithal) may slightly improve dry eye–specific patient-reported outcome measures (at 2 wk) compared with 1.4% polyvinyl alcohol. (S)	Not reported
		0.25% Polyethylene glycol 400 may improve dry eye–specific patient-reported outcome measures (at 4 wk) compared with 1% carboxymethylcellulose. (S)	Not reported
		0.18% Sodium hyaluronate may improve dry eye–specific patient-reported outcome measures (at 7 d) compared with 1% carboxymethylcellulose. (S)	Not reported
		1% Carboxymethylcellulose (Refresh Liquigel/Celluvisc Lubricant Ophthalmic solution) may improve dry eye–specific patient-reported outcome measures (at 4 and 8 wk) compared with 0.3% hydroxypropyl methylcellulose. (S)	Not reported
		0.4% Carbomer-based gel may improve dry eye–specific patient-reported outcome measures (at 3 mo) compared with 1% carboxymethylcellulose. (S)	Not reported
		0.4% Hyaluronic acid may improve dry eye–specific patient-reported outcome measures (at 15, 30, and 60 d) compared with 0.3% hydroxypropyl methylcellulose. (S)	Not reported
		Polyethylene glycol 400 may improve dry eye–specific patient-reported outcome measures (at 4 wk) compared with hydroxypropyl-guar. (S)	Not reported
		0.5% Carboxymethylcellulose plus 0.15% hyaluronic acid may improve dry eye–specific patient-reported outcome measures (at 90 d) compared with 0.5% carboxymethylcel-lulose plus 0.1% hyaluronic acid (Optive Fusion, Allergan). (S)	Not reported
		Carbomer-containing viscous gel (Lubrithal) may cause more blurred vision (MedDRA: vision blurred) and discontinuations (MedDRA: treatment withdrawal) compared with 1.4% polyvinyl alcohol.	Not reported
		0.5% Carboxymethylcellulose plus 0.15% hyaluronic acid group may cause more treatment-related adverse events (MedDRA: adverse event) compared with 0.5% carboxymethylcellulose plus 0.1% hyaluronic acid (Optive Fusion, Allergan).	Not reported
Ribeiro et al,34 2019	April 2018	Insufficient evidence that preservative-free artificial tears improve dry eye–specific patient-reported outcome measurements (at 4–12 wk) compared with preserved artificial tears. (S)	Not reported
Dietary modifications (AAO step 1)		4 Reviews were reliable, and main findings were classified as effective–not conclusively.		New primary research (RCTs, NRSIs, or observational primary studies investigating optimal dosages, length of treatment, baseline PUFA intake, and ratios of ω-3 to ω-6 and adverse effects or harms)
Chi et al,36 2019	March 2019	ω-3 ± ω-6 Fatty acid oral supplementation may improve dry eye–specific patient-reported outcome measures (at 1–12 mo) compared with control. (S-L)	Not reported
Giannaccare et al,35 2019	2018	ω-3 ± ω-6 Fatty acid oral supplementation may improve dry eye–specific patient-reported outcome measures (at 1–12 mo) compared with placebo. (S-L)	Not reported
Downie et al,37 2019	February 27, 2018	Oral ω-3 PUFAs plus conventional therapy may improve dry eye–specific patient-reported outcome measures (at 1 mo) compared with conventional therapy alone. (S)	Low
		Oral flaxseed oil (short-chain ω-3 PUFAs) may improve dry eye–specific patient-reported outcome measures (at 12 mo) compared with placebo. (L)	Low
		Combined ω-3 and ω-6 PUFAs may improve dry eye–specific patient-reported outcome measures (at 6 mo) compared with ω-6 PUFAs. (M)	Not reported
		Oral long-chain ω-3 PUFAs (EPA and DHA) probably improve dry eye–specific patient-reported outcome measures (at 45 d to 6 mo) compared with oral ω-6 PUFAs. (S-M)	Moderate
		Insufficient evidence that oral ω-6 PUFAs improve dry eye–specific patient-reported outcome measures (at 45 d) compared with placebo. (S)	Low
		Insufficient evidence that combined oral ω-3 and ω-6 PUFAs improve dry eye–specific patient-reported outcome measures (at 3 mo) compared with placebo. (S)	Low
		Insufficient evidence that oral long-chain ω-3 PUFAs improve dry eye–specific patient-reported outcome measures (at 1–12 mo) compared with oral placebo or no treatment. (S-L)	Low
		Insufficient evidence that combined oral fish oil (long-chain ω-3 PUFAs) and flaxseed oil (short-chain ω-3 PUFAs) improve dry eye–specific patient-reported outcome measures (at 3 mo) compared with combined ω-3 and ω-6 PUFAs. (S)	Not reported
		Insufficient evidence that combined oral fish oil (long-chain ω-3 PUFAs) and flaxseed oil (short-chain ω-3 PUFAs) improve dry eye–specific patient-reported outcome measures (at 60 d) compared with short-chain ω-3 PUFAs. (S)	Not reported
Singh et al,38 2022	December 12, 2019	Oral ω-3 PUFAs may improve dry eye–specific patient-reported outcome measures (at 45 d to 3 mo) compared with placebo. (S)	Low
		Oral berry extract may result in little to no difference in dry eye–specific patient-reported outcome measures (at 4–8 wk) compared with placebo. (S)	Low
		Insufficient evidence that combination of oral supplements improves dry eye–specific patient-reported outcome measures at 4–12 wk compared with placebo. (S)	Very low
		Oral interventions (probiotics) may result in little to no difference in dry eye–specific patient-reported outcome measures (at 8 wk) compared with placebo. (S)	Low
Punctal occlusion (AAO step 2)		1 Review was reliable, and main findings were classified as effective–not conclusively.		New primary research (such as RCTs investigating well-defined aqueous deficient and evaporative dry eye etiologies, and comparisons with topical cyclosporine A and autologous serum, and optimal combination therapies)
Ervin et al,39 2017	December 8, 2016	Punctal plugs may improve patient-reported outcome measures (at 3 mo) compared with artificial tears, but the evidence is very uncertain. (S)	Very low
		Punctal plugs may improve dry eye–specific patient-reported outcome measures (at 3 mo) compared with pilocarpine, but the evidence is very uncertain. (S)	Very low
		Insufficient evidence that punctal plugs improve dry eye–specific patient-reported outcome measures (at 2–12 mo) compared with no punctal plugs. (S-L)	Very low
		Lower punctal plugs may result in little to no difference in dry eye–specific patient-reported outcome measures (at 1 mo) compared with upper ± lower punctal plugs. (S)	Low
		Acrylic punctal plugs may result in little to no difference in dry eye–specific patient-reported outcome measures (at 11 wk) compared with silicone punctal plugs. (S)	Low
		Intracanalicular punctal plugs may result in little to no difference in dry eye–specific patient-reported outcome measures (at 3 mo) compared with silicone punctal plugs, but the evidence is very uncertain. (S)	Very low
		Collagen punctal plugs may result in little to no difference in dry eye–specific patient-reported outcome measures (at 6 mo) compared with silicone punctal plugs, but the evidence is very uncertain. (M)	Very low
		Punctal plugs may cause epiphora (MedDRA: lacrimation increased), “spontaneous plug loss,” and itching (MedDRA: eye pruritus) at the plug placement area compared with no punctal plugs, but the evidence is very uncertain.	Very low
		Punctal plugs may result in little to no difference in punctate epithelial keratopathy (MedDRA: punctate keratitis) (at 2 wk) compared with artificial tears. (S)	Low
Intense pulse light (AAO step 2)		1 Review was reliable, and main findings were classified as effective–not conclusively.		New primary and secondary research (such as RCTs to determine optimal length of treatment periods with appropriate analysis of paired eye data, systematic reviews including 14 ongoing trials)
Cote et al,40 2020	August 1, 2019	Intense pulsed light therapy plus meibomian gland expression may improve dry eye–specific patient-reported outcome measures (at 24 wk) compared with meibomian gland expression alone, but the evidence is very uncertain. (M)	Very low
		Intense pulsed light therapy ± meibomian gland expression may result in little to no difference in dry eye–specific patient-reported outcome measures (at 45 d and 3 mo) compared with control (sham), but the evidence is very uncertain. (S)	Very low
		Intense pulsed light therapy may cause pain-related adverse effects (MedDRA: procedural site pain), but the evidence is very uncertain.	Very low
Vectored thermal pulsation therapy (AAO step 2)		1 Review was reliable, and main findings were classified as effective–not conclusively.		New primary and secondary research (such as RCTs to determine optimal length of treatment periods, systematic reviews that assess certainty of evidence)
Pang et al,41 2019	July 2018	Lipiflow may improve dry eye–specific patient-reported outcome measures (at 2–4 wk) compared with warm compresses. (S)	Not reported
		Lipiflow may cause higher percentage of discomfort/pain during and after the procedure (MedDRA: procedural site pain) compared with warm compresses.	Not reported
Topical corticosteroids (AAO step 2)		1 Review was reliable, and main findings were classified as effective-not conclusively.		New primary and secondary research (such as RCTs to determine optimal sequence of therapy following artificial tears or cyclosporine A, head-to-head comparisons of ester vs ketone steroids in well-defined dry eye etiologies, NRSIs, observational studies, and RCTs of sufficient follow-up to detect adverse effects or harms)
Liu et al,42 2022	August 20, 2021	Topical steroids ± tobramycin probably slightly improve dry eye–specific patient-reported outcome measures (at 1–2 mo) compared with lubricants. (S)	Moderate
		Topical steroids ± cyclosporine A may slightly improve dry eye–specific patient-reported outcome measures (at 8 wk to 3 mo) compared with topical cyclosporine A alone. (S)	Low
		Topical steroids may increase the risk of elevated intraocular pressure (MedDRA: intraocular pressure increased) (at 14 d to 4 wk) compared with lubricants, but the evidence is very uncertain.	Very low
		Insufficient evidence that topical steroids increase the risk of new cataract formation (MedDRA: cataract) (at 14 d) or “serious adverse events” (at 14 d to 4 wk) compared with lubricants. (S)	Very low
		Insufficient evidence that topical steroids ± cyclosporine A increases the risk of elevated intraocular pressure (MedDRA: intraocular pressure increased) (at 8 wk to 3 mo) compared with cyclosporine A alone.	Very low
		Insufficient evidence that topical steroids ± cyclosporine A increases the risk of “serious adverse events” (at 8–12 wk) compared with cyclosporine A alone. (S)	Very low
Topical cyclosporine A (AAO step 2)		2 Reviews were reliable, and main findings were classified as effective–not conclusively.		New primary and secondary research (such as RCTs with adequate reporting of effect sizes, 3-arm RCTs to determine optimal combination therapy ± artificial tears, RCTs to determine optimal treatment durations, optimal dosage of cyclosporine A, and efficacy of cyclosporine A on well-defined dry eye etiologies, systematic reviews that synthesize RCTs with improved conduct and reporting)
de Paiva et al,43 2019	February 16, 2018	Topical cyclosporine A 0.05% plus artificial tears may slightly improve dry eye–specific patient-reported outcome measures (at 6 mo) compared with vehicle plus artificial tears or artificial tears alone. (M)	Low
		Topical cyclosporine A 0.05% may improve dry eye–specific patient-reported outcome measures (at 3–6 mo) compared with placebo or vehicle. (M)	Not reported
		Topical cyclosporine A 1% plus artificial tears may improve dry eye–specific patient-reported outcome measures (at 6 mo) compared with placebo plus artificial tears. (M)	Not reported
		Topical cyclosporine A 2% plus artificial tears may improve dry eye–specific patient-reported outcome measures (at 8 wk) compared with placebo plus artificial tears. (S)	Not reported
		Insufficient evidence that topical cyclosporine A 0.05% plus artificial tears improve dry eye–specific patient-reported outcome measures (within 6 mo) compared with topical cyclosporine A 0.1%, cyclosporine A 0.2%, or cyclosporine A 0.4% plus artificial tears. (M)	Not reported
		Insufficient evidence that topical cyclosporine A 0.1% improves dry eye–specific patient-reported outcome measures (at 2–3 mo) compared with placebo or vehicle plus artificial tears. (S)	Not reported
		Topical cyclosporine A 0.1% cationic emulsion plus artificial tears probably result in little to no difference in dry eye–specific patient-reported outcome measures (at 6 mo) compared with vehicle plus artificial tears. (M)	Moderate
		Topical cyclosporine A 0.05% plus artificial tears may cause burning eyes (MedDRA: eye irritation) and discontinuations (MedDRA: treatment withdrawal) (at 6 mo) compared with vehicle plus artificial tears or artificial tears alone. (M)	Low
		Topical cyclosporine A 0.1% plus artificial tears may cause adverse outcomes including severe stinging sensations (MedDRA: eye pain) and running nose or tearing (MedDRA: lacrimation increased).	Not reported
		Topical cyclosporine A 0.1% cationic emulsion plus artificial tears probably cause more treatment-related ocular adverse events and instillation site pain/irritation (MedDRA: eye pain/irritation) compared with vehicle plus artificial tears.	Moderate
Tuan et al,44 2020	January 25, 2019	Topical cyclosporine A ± artificial tears may improve dry eye–specific patient-reported outcome measures (at 3–12 mo) compared with artificial tears alone. (S-L)	Not reported
		Topical cyclosporine A ± artificial tears may cause more adverse events (MedDRA: adverse event) (at 3–6 mo) compared with artificial tears alone. (S-M)	Not reported
Topical secretagogues (AAO step 2)		1 Review was reliable, and main findings were classified as effective–not conclusively.		New primary and secondary research (such as RCTs with improved conduct and reporting which make head-to-head comparisons between topical secretagogues, autologous serum, and topical immunomodulators, subsequent evidence synthesis in systematic reviews)
Jongkhajornpong et al,45 2022	March 22, 2021	Diquafosol combined with artificial tears may improve dry eye–specific patient-reported outcome measures (at 12 wk) compared with artificial tears alone. (S)	Low
		Diquafosol combined with artificial tears may result in little to no difference in risk of adverse events (MedDRA: adverse event) (at 12 wk) compared with artificial tears alone. (S)	Not reported
		Diquafosol may slightly improve dry eye–specific patient-reported outcome measures (at 12 wk) compared with artificial tears. (S)	Very low
		Diquafosol may result in little to no difference in dry eye–specific patient-reported outcome measures (at 4–12 wk) compared with cyclosporine A. (S)	Not reported
		Diquafosol may increase the risk of adverse events (MedDRA: adverse event) (at 4–12 wk) compared with artificial tears. (S)	Not reported
Autologous/allogeneic serum (AAO step 3)		4 Reviews were reliable, and main findings were classified as effective–not conclusively.		New primary and secondary research (such as RCTs with improved conduct and reporting and appropriate use of parallel designs, longer follow-up time points, well-defined dry eye etiologies, NRSIs, observations studies, or RCTs to investigate potential for complications, intolerance, and adverse effects or harms [eg, from microbial growth in serum], RCTs to make head-to-head comparisons between topical secretagogues or immunomodulators, subsequent evidence synthesis in systematic reviews with appropriate analysis of paired data and crossover designs)
Pan et al,46 2017	July 5, 2016	Autologous serum 20% may improve dry eye–specific patient-reported outcome measures (at 2 wk) compared with artificial tears. (S)	Low
		Autologous serum 20% may result in little to no difference in dry eye–specific patient-reported outcome measures (at 4 wk and 6 mo) compared with normal saline. (S-M)	Not reported
		Insufficient evidence that autologous serum 20% causes more conjunctivitis compared with normal saline.	Not reported
Franchini et al,47 2019	March 30, 2019	Autologous serum 20%-50% may improve dry eye–specific patient-reported outcome measures (at 2–6 wk) compared with artificial tears. (S)	Low
Wang et al,48 2020	December 31, 2018	Autologous serum 20%-50% may improve dry eye–specific patient-reported outcome measures (at 5 wk to 1 y) compared with artificial tears. (S-L)	Not reported
Jongkhajornpong et al,45 2022	March 22, 2021	Autologous serum may improve dry eye–specific patient-reported outcome measures (at 2–4 wk) compared with artificial tears. (S)	Low
Systemic anti-inflammatory and immunosuppressive agents (patient prioritization)		3 Reviews were reliable (2 reviews reported critical outcome), and main findings were classified as effective–not conclusively.		New secondary research (systematic reviews that assess certainty of evidence)
Souza et al,52 2016	December 2015	Rituximab may improve dry eye–specific patient-reported outcome measures (at 24–48 wk) compared with placebo. (M)	Not reported
		Rituximab may result in little to no difference in risk of serious adverse events, serum sickness, infections, or serious infections (at 24 wk) compared with placebo. (M)	Low
Chu et al,54 2020	October 5, 2017	Azathioprine may result in little to no difference in dry eye–specific patient-reported outcome measures (at 6 mo) compared with placebo. (M)	Not reported
		Biologics (IFN-α, rituximab) may result in little to no difference in dry eye–specific patient-reported outcome measures (at 6 mo) compared with placebo. (M)	Not reported
Hormonal therapy (not prioritized)		3 Reviews were reliable, and main findings were classified as effective–not conclusively.		New secondary research (systematic reviews that assess certainty of evidence)
Chu et al,54 2020	October 5, 2017	Systemic steroids may result in little to no difference in dry eye–specific patient-reported outcome measures (at 6 mo) compared with placebo. (M)	Not reported
		Systemic steroids may result in little to no difference in risk of serious adverse events or withdrawals due to adverse effects (at 6 mo) compared with placebo. (M)	Not reported
Hao et al,49 2020	November 2019	Hormonal therapy (topical or systemic estrogen or estrogen plus progestogen) may improve dry eye–specific patient-reported outcome measures (at 1–6 mo) compared with control. (S-M)	Not reported
		Hormonal therapy (topical or systemic estrogen or estrogen plus progestogen) may result in little to no difference in the risk of ocular adverse events at 6 mo to 1 y compared with control. (S-L)	Not reported
Wang and Deng,50 2020	February 2, 2020	Transdermal hormonal therapy (androgen therapy) may improve dry eye–specific patient-reported outcome measures (at 1 mo) compared with placebo.	Not reported
		Transdermal hormonal therapy (androgen therapy) may result in little to no difference in dry eye–specific patient-reported outcome measures (at 2 mo) compared with placebo. (S)	Not reported
		Transdermal hormonal therapy (androgen therapy) may cause more dermatological adverse events (MedDRA: seborrhea/acne) compared with placebo.	Not reported
Traditional Chinese medicine: not prioritized		3 Reviews were reliable (2 reviews reported critical outcome), and main findings were classified as ineffective–not conclusively.		New primary and secondary research (update systematic reviews and assess certainty of evidence)
Kim et al,55 2018	July 2017	Acupuncture ± artificial tears may result in little to no difference in dry eye–specific patient-reported outcome measures compared with artificial tears.	Not reported
Na et al,57 2021	July 2018	Acupuncture ± artificial tears may or may not improve dry eye–specific patient-reported outcome measures compared with artificial tears.	Not reported
Oral antibiotics, topical LFA-1 antagonist drugs (such as lifitegrast), moisture chamber devices (AAO step 2)		No reviews were reliable (5 unreliable reviews available).		New secondary research (reliable systematic reviews)
Patient education, oral secretagogues, therapeutic contact lens options, topical corticosteroid for longer duration, amniotic membrane grafts, surgical punctal occlusion, other surgical approaches (eg, tarsorrhaphy, salivary gland transplantation) (AAO step 1/3/4)		No reliable reviews were available.		New primary research (such as RCTs)
Specific sequences/combinations of treatments (patient prioritization)		No reliable reviews were available.		New primary research (such as RCTs)

Abbreviations: AAO, American Academy of Ophthalmology

DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; NRSI, nonrandomized study of intervention; PUFA, polyunsaturated fatty acid; L, long-term time point; LFA-1, lymphocyte function–associated antigen 1; M, medium-term time point; RCT, randomized clinical trial; S, short-term time point.

In summary, there was some evidence to support most interventions in step 1 to 3 of the 2018 AAO PPP on dry eye (certainty: very low to moderate). However, without high-certainty evidence, the findings were generally inconclusive (Table 3). Indoor air humidification may improve the risk of dry eye symptoms compared with no humidification in the short term (certainty: low). There was insufficient evidence that other environmental modifications improve dry eye symptoms such as moist cool air devices (certainty: very low), visual hygiene exercises (ie, 20–20-20 rule) (certainty: very low), and blue light–blocking lenses (certainty: moderate).^32,38

Artificial tears and lubricants may improve dry eye symptoms from baseline (certainty: not reported). Among 9 comparisons (Table 3), there was some evidence that several over-the-counter artificial tear formulations may improve dry eye symptoms compared with other over-the-counter formulations (certainty: not reported).³³ Of these, 6 comparisons showed some evidence that effective formulations may improve at least 1 other important outcome (certainty: not reported). Among most head-to-head comparisons of 2 active interventions, there was insufficient evidence to support superiority of one formulation over another (certainty: not reported).³³ In addition, there was insufficient evidence of a difference between preserved and preservative-free formulations regarding dry eye symptoms and other important outcomes including ocular surface staining (certainty: not reported).³⁴ However, a lack of expected overlap (adjusted corrected covered area 2.3%) between relevant systematic reviews suggested that the evidence regarding preserved and preservative-free artificial tears may be incomplete.^33,34 Some evidence suggested that higher-viscosity lubricant formulations may cause more treatment-related adverse events (MedDRA: adverse event), including blurred vision (MedDRA: vision blurred) (certainty: not reported).³³

Oral ω-3 with or without ω-6 PUFAs may clinically significantly improve dry eyesymptoms compared with placebo at short- to long-term follow-up (certainty: not reported).^35,36,38 Additionally, ω-3 PUFA supplementation may improve other important outcomes such as tear osmolarity, corneal fluorescein staining, and aqueous production at similar time points (certainty: not reported).³⁶ A Cochrane review stratified the synthesis according to specific PUFA formulations and found that supplementation containing ω-3 PUFAs is probably more effective in improving dry eye symptoms compared with ω-6 PUFA supplementation alone at short- to medium-term follow-up (certainty: moderate or not reported).³⁷ Insufficient evidence was available to determine whether adverse effects were more common with PUFA supplementation compared with control groups, but the most frequently reported adverse effect was gastrointestinal upset (MedDRA: abdominal discomfort).

In-office procedures such as punctal plugs, IPL therapy, and Lipiflow may be effective.^39–41 Specifically, punctal plugs with or without artificial tears may improve dry eye symptoms (certainty: very low) and tear film stability (certainty: moderate) at 3 months compared with artificial tears alone, but there may be little to no difference at 2 weeks.³⁹ Among other clinically relevant comparisons, the evidence was insufficient to support preferred punctal plug placement, material type, or location (certainty: very low to low). Regarding adverse effects, punctal plugs may cause epiphora (MedDRA: lacrimation increased), itching (MedDRA: eye pruritus), and spontaneous plug loss compared with no punctal plugs (certainty: very low). Intense pulsed light therapy may result in little to no difference in dry eye symptoms compared with sham placebo in the short term (certainty: very low).⁴⁰ However, when combined with meibomian gland (MG) expression, IPL therapy may improve dry eye symptoms at medium-term follow-up compared with MG expression alone (certainty: very low).⁴⁰ Additional evidence suggested that IPL therapy with MG expression may improve tear film stability, ocular surface staining, and lipid layer thickness (certainty: very low to low). Insufficient evidence was available to make conclusions regarding adverse effects caused by IPL therapy, but procedural pain (MedDRA: procedural site pain) was reported among some study participants (certainty: very low). Lipiflow may clinically significantly improve dry eyesymptoms, tear film stability, and MG function compared with warm compress treatment at 2 to 4 weeks (certainty: not reported).⁴¹ However, at 3 months, there was conflicting evidence between symptom measurement tools, and there may be little to no difference in other important outcomes.⁴¹ Some evidence suggested that Lipiflow may cause discomfort (MedDRA: procedural site pain) during and after the procedure compared with warm compress treatment (certainty: not reported).

Among prescription-only topical therapies, topical cyclosporine A and topical secretagogues may improve dry eye symptoms, and topical corticosteroids are probably effective.^42–45 Topical corticosteroids with or without tobramycin probably improve dry eye symptoms and corneal fluorescein staining slightly compared with lubricants or tobramycin alone (certainty: moderate).⁴² Topical corticosteroids may increase the risk of elevated intraocular pressure (MedDRA: intraocular pressure increased) at short-term follow-up (certainty: very low).⁴² Topical cyclosporine A, with concentrations ranging 0.05% to 2%, may improve dry eye symptoms at short-to long-term follow-up (certainty: low or not reported).^43,44 Although some evidence suggested that topical cyclosporine A, 0.05%, may clinically significantly improve dry eye symptoms, the results were heavily weighted by 1 small primary study with high risk of bias (certainty: not reported).^44,58 In support of these findings, some additional evidence suggested that topical cyclosporine A, 0.05%, may improve tear film stability and increase conjunctival goblet cell density (certainty: low).^43,44 Consistently, the evidence suggested that topical cyclosporine A may cause more adverse effects, such as burning eyes (MedDRA: eye irritation), stinging eyes (MedDRA: eye pain), tearing eyes (MedDRA: lacrimation increased), and discontinuations (MedDRA: treatment withdrawal), compared with artificial tears (certainty: low to moderate or not reported).^43,44 Evidence from NMA suggested that topical diquafosol may improve dry eye symptoms compared with artificial tears in the short term (certainty: very low)⁴⁵ but may be most effective when used in combination with artificial tears (certainty: low).⁴⁵ Other evidence from indirect NMA comparisons suggested that there may be little to no difference in dry eye symptoms between topical diquafosol and other topical secretagogues (eg, rebamipide) or immunomodulators (eg, cyclosporine A) (certainty: low).⁴⁵ Some evidence suggested that topical diquafosol may increase the risk of adverse events (MedDRA: adverse event) compared with artificial tears (certainty: not reported).

Autologous serum may be effective compared with artificial tears and other interventions. Consistent findings from 4 systematic reviews with very high overlap (adjusted corrected covered area 44.2%) suggested that autologous serum may clinically significantly improve dry eye symptoms compared with artificial tears at short-term follow-up (certainty: low).^45–48 Autologous serum may improve tear film stability compared with artificial tears at short- to long-term follow-up (certainty: low or not reported).^46–48 Indirect NMA comparisons suggested that autologous serum may improve dry eye symptoms compared with topical diquafosol at short-term follow-up (certainty: very low).⁴⁵ Insufficient evidence was available to make conclusions regarding adverse effects or harms caused by autologous serum because few events were reported. However, mixed organism growth in autologous serum containers and instances of conjunctivitis have been reported.

Several treatments not recommended or prioritized were investigated by reliable systematic reviews, including topical pranoprofen, acupuncture, and hormonal therapy (Table 2). Among recommended treatments without reliable systematic review evidence, lifitegrast, moisture chamber devices, and oral antibiotics for MG dysfunction were investigated by unreliable systematic reviews despite the availability of multiple RCTs (eTable 4 in the Supplement).^59–63

Discussion

In this overview, we found that common treatment options from step 1 to 3 of the 2018 AAO PPP on dry eye were supported by lower-certainty evidence. Specifically, the routine recommendation of artificial tears to improve dry eye symptoms was supported by some evidence. However, given the limited reporting of effect sizes, high degree of publication bias, lack of expected overlap across systematic reviews, and few assessments of certainty of evidence, the evidence remains inconclusive. Among environmental modifications, only indoor air humidification showed improvement in dry eye symptoms, and it may be considered a useful short-term behavioral modification particularly for indoor occupational settings. Use of dietary supplementation, specifically ω-3 PUFAs, has been controversial considering evidence from recent RCTs.⁶⁴ Two meta-analyses provided some evidence that supplementation with ω-3 PUFAs may improve dry eye symptoms. In addition, ω-3 PUFAs may improve other important clinical outcomes, including corneal fluorescein staining and tear osmolarity. Wide variability exists in the composition and daily dosages of PUFA supplementation, and optimal doses or ω-3 to ω-6 ratios were not investigated.

Punctal plugs are frequently used in aqueous-deficient dry eyes with some advocating for the use of punctal plugs in both upper and lower puncta for maximum tear preservation. While there was some evidence of dry eye improvement with punctal plugs, there was insufficient evidence to support preferences in punctal plug placement, material type, or location. As punctal plugs with or without artificial tears probably improve tear film stability, they could potentially have utility in evaporative dry eye, but the evidence does not directly address this population. Punctal plugs may need to be in place for at least 3 months before achieving a beneficial effect.

In-office devices (eg, IPL therapy, Lipiflow) are recent technological advances with high interest among practitioners. Both devices showed some evidence that they may improve clinical outcomes, including dry eye symptoms and tear film stability, but all the primary studies were at high risk of bias. Lipiflow, which involves 1 session every 3 months, may provide most of its symptomatic benefit within a month after treatment, becoming less effective as time elapses. A Cochrane review has found conflicting evidence regarding the effectiveness of Lipiflow compared with warm compress treatment after 1 month, but the evidence was very uncertain.⁶⁵ Publication of this Cochrane review will provide a comprehensive evaluation of Lipiflow. In contrast to Lipiflow, IPL therapy, which consists of multiple sessions across several weeks, may take time to become effective and may only be effective when used concomitantly with MG expression. Given the high risk of bias in the primary studies and uncertainty in the evidence, the inconclusive findings for in-office devices should be interpreted very cautiously.

With inflammation characterizing the disease state in dry eye, clinicians have advocated the use of topical immunomodulation. Common clinical practice is to initiate topical corticosteroids in the short term to address inflammation rapidly with concurrent cyclosporine A treatment for long-term management. However, combination therapy has not been the focus of any of the systematic reviews.¹⁶ Available evidence suggests that topical corticosteroids probably improve dry eye symptoms slightly and topical cyclosporine A, 0.05%, may slightly improve dry symptoms.

Topical secretagogues are suggested for evaporative dry eye relating to mucin deficiency. One secretagogue, topical diquafosol, has been shown to improve dry eye symptoms when combined with artificial tears. When compared with autologous serum through indirect NMA, autologous serum may improve dry eye symptoms over topical diquafosol at short-term follow-up. Autologous serum at various concentrations may improve both dry eyesymptoms and tear film stability when compared with artificial tears. However, a recently updated review reported that changes in tear film stability measurements were not clinically important and concluded that there is insufficient evidence to recommend autologous serum use fordryeye.⁶⁶ Variations in eligibility criteria regarding ocular surface disease definitions and the analysis methods implemented to handle cross-over trial results may explain differences in reported summary effect estimates and conclusions across autologous serum reviews.

Implications for Research

The prevalence of unreliable systematic reviews from inadequate searches and inappropriate meta-analysis raises concern. Systematic reviewers should collaborate with information specialists to ensure search quality.^67,68 Within primary studies, significant variability exists in the use of validated instruments and reporting of results limited to P values. Enhanced reporting and utilization of a core outcome set in primary studies would facilitate quantitative synthesis by systematic reviewers. Enrichment of trials with participants exhibiting specific signs and symptoms or disease subtypes may improve the ability to detect treatment efficacy among subpopulations.

High-quality systematic reviews are required to evaluate several recommended treatments.^{59–63,69–72} As novel therapies for dry eye emerge and gain approval, reliable systematic reviews will be necessary to guide clinical recommendations and decisions.^69–72 Systematic reviews aim to identify all the relevant literature, assess their risk of bias, and evaluate the strength of the evidence across trials, offering a thorough and in-depth synthesis of the results that goes beyond merely demonstrating efficacy during regulatory approval processes.

Strengths and Limitations

Our overview holds strengths and limitations. We provided a comprehensive summary of all recent systematic reviews on treatments for dry eye. We collected symptoms, signs, and safety outcomes and noted when patient-reported outcome measures correlated with improvements in clinical signs. We also commented on the clinical significance of the findings. These approaches align with recent draft guidance by the US Food and Drug Administration for evaluating dry eye treatments.⁷³ We did not assess the certainty of evidence de novo for outcomes and only collected these when available. Our overview had limited utility to identify long-term or rare outcomes (eg, harms) because the systematic reviews were largely based on RCTs that are underpowered for these purposes.

Conclusions

Our findings suggest that clinically important and prioritized research questions regarding the efficacy of dry eye treatments are being addressed using systematic reviews. There was some evidence to support most interventions of 2018 AAO PPP on dry eye, but the certainty of evidence was generally low, and the findings were inconclusive. Our overview suggests the conduct and reporting of most systematic reviews for dry eye interventions warrant improvement, and reliable systematic reviews are needed to evaluate lifitegrast, oral antibiotics, and moisture chamber devices.

Key Points.

Question

What does the body of reliable systematic reviews provide regarding evidence for the effectiveness and safety of various interventions for dry eye?

Findings

This overview of reliable systematic reviews finds some evidence for interventions, including environmental and dietary modifications, artificial tears and lubricants, punctal occlusion, intense pulsed light therapy, vectored thermal pulsation therapy, topical corticosteroids/cyclosporine A/secretagogues, and autologous serum. However, certainty of evidence is generally low, rendering findings inconclusive.

Meaning

Most common dry eye interventions may be effective, but the current evidence is largely inconclusive, and most systematic reviews addressing dry eye interventions were not reliable.