Efficacy of eyelid warming devices as first-step treatment in meibomian gland dysfunction: A systematic review with meta-analysis

Efficacy of eyelid warming devices as first-step treatment in meibomian gland dysfunction: A systematic review with meta-analysis Antonio Ballesteros-S´ anchez a,b,* , Carlos Rocha-de-Lossada c,d,e,f , Jos´ e-María S´ anchez-Gonz´ alez b a School of Optics and Optometry, Universitat Polit` ecnica de Catalunya, Terrassa, Spain b Department of Physics of Condensed Matter, Optics Area, University of Seville, 41012, Seville, Spain c Ophthalmology Department, VITHAS Malaga, 29016, Malaga, Spain d Regional University Hospital of Malaga, Hospital Civil Square, 29009, Malaga, Spain e Qvision, Ophthalmology Department, VITHAS Almeria Hospital, 04120, Almeria, Spain f Surgery Department, Ophthalmology Area, University of Seville, Doctor Fedriani, 41009, Seville, Spain ARTICLE INFO Keywords: Dry eye disease Meibomian gland dysfunction Eyelid warming compress Moisture chamber goggles Warm towel compress ABSTRACT Purpose: To investigate the efficacy of eyelid warming devices as first-step treatment in patients with meibomian gland dysfunction (MGD). Methods: A systematic review with meta-analysis of RCTs, reporting the effects eyelid warming devices in 3 databases, PubMed, Scopus and Web of Science, was performed according to the PRISMA statement. Results: Seven studies including 367 patients, and 440 eyes were analysed. The overall efficacy did not significantly favor either group when comparing eyelid warming devices to the control groups or warm towel compresses, nor between moisture chamber devices and warm compress devices. However, the change in OSDI questionnaire (SMD 0.91; 95 % CI: 0.44 to 1.39; P =0.0002) and NIBUT (SMD 1.10; 95 % CI: 0.61 to 1.59; P < 0.0001) were significantly favorable for eyelid warming devices compared to the control groups. Similar results were obtained for tear film stability (SMD 0.97; 95 % CI: 0.32 to 1.61; P =0.003) when comparing eyelid warming devices to warm towel compresses. Specifically, the sensitivity analysis of these groups revealed that changes in OSDI questionnaire (MD 9.41; 95 % CI: 1.70 to 17.13; P =0.02; I 2 =49 %) and NIBUT (MD 2.24; 95 % CI: 1.20 to 3.28; P <0.0001; I 2 =71 %) were significantly favorable for eyelid warming devices. When comparing moisture chamber devices and warm compress devices, only the change in TBUT (SMD 0.75; 95 % CI: 0.23 to 1.28; P =0.005; I 2 =30 %) were significantly favorable for moisture chamber goggles. Conclusions: Despite their limited overall efficacy, eyelid warming devices significantly reduce OSDI questionnaire and improve NIBUT compared to controls or warm towel compress groups. Evidence remains insufficient to confirm whether moisture chamber devices provide better outcomes than warm compress devices. Further welldesigned RCTs are needed to confirm these findings. 1. Introduction Meibomian gland dysfunction (MGD) is a chronic, diffuse abnormality of the meibomian glands (MGs), which may lead to alterations in the tear film, symptoms of ocular irritation, inflammation and ocular surface disease [1,2]. Several studies have reported that the prevalence of MGD in patients with dry eye range from 3.5 % to 70 % [3,4]. Consequently, MGD is often regarded as the leading cause of evaporative dry eye, which is recognized as the most prevalent subtype of DED [5–7]. The primary cause of MGD is thought to be terminal duct obstruction of the meibomian glands (MGs), resulting from hyperkeratinisation and/ or qualitative/quantitative changes in meibum secreted by MGs [8,9]. Meibum is a complex mixture of various polar and nonpolar lipids, including wax esters, diesters, triacylglycerols, free cholesterol, free fatty acids, and phospholipids [10,11]. Due to its composition, the melting point of meibum is not consistent [12,13]. Evidence suggests that meibum melts within a range of 10 ◦C (healthy meibum) to 45 ◦C (toothpaste-like meibum) [12–16]. Consequently, while healthy meibum remains fluid at temperatures below normal body temperature * Corresponding author. Reina Mercedes Street, University of Seville, Seville, Spain. E-mail addresses: [email protected], [email protected] (A. Ballesteros-S´ anchez). Contents lists available at ScienceDirect The Ocular Surface journal homepage: www.elsevier.com/locate/jtos https://doi.org/10.1016/j.jtos.2025.02.008 Received 27 October 2024; Received in revised form 29 January 2025; Accepted 19 February 2025 The Ocular Surface 37 (2025) 33–46 Available online 21 February 2025 1542-0124/© 2025 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/bync-nd/4.0/ ). ≈37 ◦C [16], the melting point of toothpaste-like meibum frequently exceeds this threshold, which could explain the solid state of meibum in MGD [16,17]. Currently, new in-office treatments have emerged as promising therapeutic options for MGD, such as vectored thermal pulsation (VTP) [18], intense pulsed light (IPL) [19,20], low-level light therapy (LLLT) [21] and quantum molecular resonance (QMR) [22]. These treatments have been shown to improve the MGs function by reducing inflammation and promoting meibum liquefaction [23,24]. As a result, there is an increase in the lipid layer thickness that reduces tear film evaporation, which helps to alleviate the signs and symptoms of dry eye [23,24]. However, the adoption of these treatments is sometimes limited, as they are not frequently covered by most insurance plans [25,26]. In addition, these treatments often require maintenance, typically on an annual or bi-annual basis, which generates a cost that is not accessible to the entire population [27]. Moreover, a recent meta-analysis comparing the efficacy of VTP with warm compresses has determined that the long-term efficacy appears to be comparable between both treatments [28], raising questions about the cost-effectiveness of in-office therapies for MGD. Therefore, warm compress therapy remains the first-step treatment of MGD for many patients, as recommended by the tear Film and Ocular Surface Society (TFOS) Dry Eye Workshop (DEWS) II management and therapy report [29]. A standard treatment recommendation involves soaking a face towel in hot water and applying it to the eyelid margins [30]. However, patients often needed to change the towel during treatment to ensure that the temperature was maintained between 40 and 45 ◦C [17,30]. To address this limitation, several medical devices, have been developed by maximizing heat retention and delivery to the eyelids [29,31,32]. These devices mainly consist of moisture chamber goggles [33] or warm compress [34]. Overall, several studies have determined that these devices have significant positive effect in the treatment of MGD [35–41]. However, the lack of a clear strategy regarding their application combined with the recent introduction of lipid-based artificial tears, has limited their broader use [17,42,43]. Moreover, according to the best of our knowledge, there is no systematic review that had analysed their efficacy. Therefore, this study aims to determine the efficacy of eyelid warming devices in the management of MGD. For this purpose, these devices were compared with control and warm towel compress groups. Additionally, different device designs were evaluated to determine the most effective option. 2. Methods 2.1. Data sources and search strategy This systematic review with meta-analysis (PROSPERO ID: CRD42024599131) was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [44,45]. A total of 122 articles published before October 15, 2024, were identified through the following databases: PubMed, Scopus, and Web of Science. The data search strategy with Boolean operators was as follows: (Moisture chamber goggles OR Warm compress OR Eyelid warming devices OR EyeGiene mask OR MGDRx EyeBag OR Therapearl OR Bruder moist heat compress OR Meibopatch OR Warm towel OR Hot towel) AND (Dry eye disease OR DED OR Evaporative dry eye OR EDE OR Meibomian gland dysfunction OR MGD). The references of the retrieved articles were reviewed to identify other related studies if they met the inclusion criteria. 2.2. Study selection All those 122 articles identified through the search strategy were considered and analysed. Duplicate studies were removed by Mendeley Reference Manager, version 2.122.1 (Elsevier Ltd., Amsterdam, Netherlands) [46]. The remaining studies underwent additional screening stages, which included title screening, abstract screening, and full-text screening. Studies unrelated to the topic were excluded from the review during title and abstract screening. The screening of the full-text studies was performed by one investigator (ABS), who selected them according to the inclusion and exclusion criteria. The inclusion criteria were as follows: full-length prospective, randomized controlled trials (RCTs) in humans that reported on the efficacy of eyelid warming devices in the treatment of MGD. Exclusion criteria included publications in languages other than English and/or non-indexed journals, studies where eyelid warming devices were applied to healthy patients and/or those with discomfort related to contact lens wear, as well as studies comparing eyelid warming devices with vectored thermal pulsation. There were no restrictions placed on the country in which the study was performed, the follow-up period, the sample size or results of the studies. The study selection process of this systematic review is presented with a flowchart diagram in Fig. 1. 2.3. Data extraction and quality assessment The data from each study were collected and summarized independently in tables designed by one researchers (JMSG). The following information was obtained from each article: (1) author and date of publication (year); (2) study design; (3) mean follow-up of all patients in the whole procedure (expressed in months); (4) number of patients; (5) mean age of the patients (expressed in years); (6) patient sex (male/female), (7) number of eyes involved; (8) inclusion criteria of the studies; (9) study group intervention; (10) control group intervention and (11) conflicts of interest. Regarding the results of the studies, the following date were collected: (1) dry eye symptoms, including ocular surface disease index (OSDI) questionnaire and symptom assessment in dry eye (SANDE) questionnaire; (2) tear film stability, including tear film breakup time (TBUT) with fluorescein and non-invasive tear film break-up time (NIBUT); (3) tear film break-up time with fluorescein (TBUT); (4) Schirmer test without anesthesia (ST); (5) Corneal fluorescein staining (CFS); and (6) MG dropout. The literature that remained after full-text screening was examined to assess the quality of the studies. To avoid the risk of bias, one dependable authors (CRDL) created a synopsis based on the Cochrane risk of bias tool [47], which includes the following items: (1) random sequence generation, (2) allocation concealment, (3) masking of participants and personnel, (4) masking of outcome assessment, (5) incomplete outcome data, (6) selective reporting and (7) other sources of bias. A third non-masked assessor (ABS) decided the quality of the studies when disagreements occurred between the two assessors. 2.4. Data synthesis and analysis The data were categorized into 3 sections: (1) RCTs comparing the efficacy of eyelid warming devices versus controls; (2) RCTs comparing the efficacy of eyelid warming devices versus warm towel compresses; and (3) RCTs comparing the efficacy of different eyelid warming devices. In terms of study outcomes, both intra-group and inter-group results were reported within each section, indicating whether the differences were statistically significant based on the statistical analysis performed by the authors of each study. Intra-group outcomes were presented as “Last visit (LV) – Baseline (B) differences”, while inter-group outcomes were reported as follows: “Eyelid warming device (Last visit – Baseline) – Control group (Last visit – Baseline)” (Section 1); “Eyelid warming device (Last visit – Baseline) – Warm towel compress (Last visit – Baseline)” (Section 2); and “Moisture chamber devices (Last visit – Baseline) – Eyelid warm compress devices (Last visit – Baseline)” (Section 3). A meta-analysis was conducted for each section to synthesize the inter-group outcomes of the studies included in this systematic review, using Review Manager Web (RevMan Web), version 5.7 (The Cochrane Collaboration, Oxford, UK) [48]. For data pooling, the results reported by the studies were estimated as mean ±standard deviation (SD), A. Ballesteros-S´ anchez et al. The Ocular Surface 37 (2025) 33–46 34 following the guidelines outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Supplementary Digital Content 1) [49]. During the analysis, if there were different data measurement methods among the studies, standardized mean differences (SMD) were calculated to analyse the continuous parameters of dry eye symptoms, NIBUT, TBUT, ST, CFS and MG dropout [50]. SMD is a measure of the size of the intervention effect in each study with respect to the variability within the study, which allows to analyse the results on a uniform scale [50]. On the other hand, if there were no different data measurement methods among the studies, the mean difference (MD) was calculated [50]. The absolute value was interpreted together with the P value and 95 % confidence intervals (CI) presented on forest plots. P <0.05 was considered statistically significant. Heterogeneity of the included studies was analysed together with the Cochrane Q-statistics chi-square (chi 2 ) test and I-square (I 2 ) test, which was graded as low (50 %) [51]. If there was any significant heterogeneity between studies (I 2 test ≥50 % or chi 2 test with a P <0.1) a random effects model was performed to pool the data, otherwise a fixed effects model was conducted [52,53]. In addition, sensitivity analyses were conducted to evaluate the robustness of the meta-analyses. These analyses were applied in meta-analyses that included at least 3 studies, ensuring that the exclusion of any single study or adjustment of assumptions did not compromise the reliability of the overall results or the assessment of variability. Overall, the measurement methods for dry eye symptoms and tear film stability differed among the studies. Therefore, sensitivity analyses were conducted to include a consistent symptom assessment method, specifically the OSDI questionnaire, as well as invasive or non-invasive variables to evaluate tear film stability, such as TBUT and NIBUT, respectively. More detailed information on the data synthesis and analysis is provided in Fig. 2. 3. Results 3.1. Study characteristics This systematic review included 7 RCTs [35–41] published between 2014 and 2024, involving 440 eyes from 367 patients with a mean age of 46.4 ±11.3 years. The sex distribution was 264 females (71.9 %) and 103 males (28.1 %). Patient follow-up, expressed in months, ranged from 1.5 to 3 months, with a mean follow-up of 1.8 ±0.7 months. Regarding study group intervention, all studies applied eyelid warming devices, which included the MGDRx Eyebag® (The EyeBag Company, West Yorkshire, UK) [35,38,39], the EyeGiene® mask (Eyedetec Medical, CA, USA) [36], the Bruder® Eye Hydrating Compress (Bruder Healthcare Company, GA, USA) [37], the OPTASE® Moist Heat Mask (Scope Ophthalmics Ltd., Dublin, Ireland), the Warmyou® Steam Eye Mask (Shanghai Warmyou Industry Co., Shanghai, China) [41], Therapearl® eye mask (Bausch & Lomb, NY, USA) and Blephasteam® (Thea Pharmaceuticals, Newcastle, UK) [36,40]. However, the control group received no intervention [38], non-heated compress [35] or warm towel compress [36,37,39,41]. Four studies had conflicts of interest as they were supported by different pharmaceutical companies [37–40]. More detailed characteristics of the studies are listed in Tables 1–3. 3.2. Eyelid warming devices Vs. controls Regarding the measurement methods, Bilkhu et al. [35] and Ngo et al. [38] utilized the OSDI questionnaire and NIBUT to assess dry eye symptoms and tear film stability, respectively. However, Bilkhu et al. [35] assessed MG dropout using the Tomlinson scale [54], while Ngo et al. [38] used the Arita scale [55]. Intra-group and inter-group efficacy outcomes are shown in Table 4. Fig. 1. Flowchart study selection process according to the PRISMA statement. A. Ballesteros-S´ anchez et al. The Ocular Surface 37 (2025) 33–46 35 Regarding intra-group outcomes, Bilkhu et al. [35] reported that the application of the MGDRx eyebag® led to improvements in dry eye symptoms (−23.2 ±28.1 points; P <0.05), NIBUT (1.9 ±2.3 s; P < 0.05) and MG dropout (−0.8 ±0.9 points; P <0.05). In contrast, the control group showed only slight improvements in dry eye symptoms (−3.2 ±3.8 points; P >0.05), whereas the rest of the variables remained unchanged. In the study performed by Ngo et al. [38], improvements were also observed only in dry eye symptoms (−11.4 ±8.9 points; P = 0.02) following the application of the MGDRx eyebag®, while the control group showed a smaller improvement in dry eye symptoms (−5.7 ± 4.7 points; P =0.22). Regarding inter-group outcomes, Bilkhu et al. [35] and Ngo et al. [38] showed favorable results for the MGDRx Eyebag® in alleviating dry eye symptoms, with values of −20 points (P <0.001) and −5.7 points (P =0.68), respectively. In addition, Bilkhu et al. [35] reported that the MGDRx Eyebag® was more effective in enhancing NIBUT and MG dropout, with values of 1.8 (P <0.001) and −0.7 points (P <0.05), respectively. 3.2.1. Meta-analysis Forest plots showing the efficacy of eyelid warming devices Vs. control is presented in Fig. 3. Two studies were included in the metaanalysis [35,38]. The overall efficacy did not favor either group However, only the change in OSDI questionnaire (SMD 0.91; 95 % CI: 0.44 to 1.39; P =0.0002; I 2 =0 %) and NIBUT (SMD 1.10; 95 % CI: 0.61 to 1.59; P <0.0001; I 2 =0 %) indicated that the MGDRx Eyebag® had significantly better outcomes than the control group. Fig. 2. Flowchart of data synthesis and analyses. Table 1 Summary of included RCTs comparing eyelid warming devices Vs. control. Author (date) Design F/U a Patients Age b Sex (F/M) Eyes Inclusion criteria Eyelid warming device Control CoI Bilkhu et al. [35] 2014 MN SM 2 25 28.7 ±7.8 19/6 50 •OSDI score ≥13 points •NIBUT score <10 s •ST >5.5 mm after 5 min •MG quality and expressibility score ≥1 MGDRx eyebag (Twice a day 5 min, temperature not reported) Non-heated compress (Twice a day during 5 min) No Ngo et al. [38] 2018 MN SM 8 25 38 ±15 18/7 25 •OSDI score ≥23 points •MG score ≤9 points MGDRx EyeBag (Twice a day 10 min, 38.1 C◦) No eyelid warming device (Remain own DED treatment) Yes CoI =Conflict of interest; SM =Single-masked; F =Female; F/U =Follow-up; M =Male; MG =Meibomian gland; MN =Monocenter; NIBUT =Non-invasive tear film break-up time; OSDI =Ocular surface disease index; RCTs =Randomized controlled trials; ST =Schirmer test without anesthesia. a Expressed as weeks. b Expressed as mean ±SD (standard deviation). A. Ballesteros-S´ anchez et al. The Ocular Surface 37 (2025) 33–46 36 3.3. Eyelid warming devices vs. warm towel compresses Regarding the measurement methods, Murphy et al. [39] (2019 A), Murphy et al. [39] (2019 B), and Wang et al. [41] utilized the OSDI questionnaire to evaluate dry eye symptoms, whereas Sim et al. (2014 A) [36] and Sim et al. [36] (2014 B) employed the SANDE questionnaire. All the mentioned studies assessed ST without anesthesia, reporting the results as the wetting length of the strip over a 5-min period. In addition, they also applied the Oxford grading scale to evaluate CFS. In terms of tear film stability, Tan et al. [37], Murphy et al. [39] (2019 A), and Murphy et al. [39] (2019 B) performed NIBUT, while Sim et al. and Wang et al. [41] used TBUT. Intra-group and inter-group efficacy outcomes are summarized in Table 5. Regarding intra-group outcomes, Sim et al. [36] (2014 A) reported that the application of Blephasteam® resulted in improvements in dry eye symptoms (−18.5 ±21.9 points; P not reported), ST (1.3 ± 1.5 mm; P not reported), and CFS (−1 ±1.2 points; P not reported). Conversely, the warm towel compress group showed smaller improvements in dry eye symptoms (−13.8 ±16 points; P not reported), with similar improvements in ST (1.8 ±2.1 mm; P not reported), while the reduction in CFS was greater (−2.5 ±2.9 points; P not reported). Similarly, Sim et al. [36] (2014 B) found that the use of the EyeGiene® mask improved dry eye symptoms (−6.7 ±6.5 points; P not reported), ST (3.1 ±3 mm; P not reported), and CFS (−2.5 ±2.4 points; P not reported). However, the warm towel compress group exhibited greater improvements in dry eye symptoms (−13.8 ±16 points; P not reported), but smaller improvements in ST (1.8 ±2.1 mm; P not reported), while the reduction in CFS was similar (−2.5 ±2.9 points; P not reported). Table 2 Summary of included RCTs comparing eyelid warming devices Vs. warm towel compress. Author (date) Design F/U a Patients Age b Sex (F/M) Eyes Inclusion criteria Eyelid warming device Warm towel compress CoI Sim et al. [36] 2014A MN SM 12 45 53.6 ±11.5 34/14 45 •Dry eye symptoms •At least one MG opening with pouting •At least one observable plugged MG Blephasteam (Twice a day 10 min, temperature not reported) Warm towel compress (Twice a day 10 min) No Sim et al. [36] 2014B MN SM 12 39 54.7 ±10.2 29/12 39 •Dry eye symptoms •At least one MG opening with pouting •At least one observable plugged MG EyeGiene (Twice a day 10 min, temperature not reported) Warm towel compress (Twice a day 10 min) No Tan et al. [37] 2017 MN SM 1 31 26.1 ±10 21/10 31 •OSDI score ≥13 points •At least one observable plugged MG Bruder moist heat eye compress (Only once 5 min, 40 C◦) Warm towel compress (Only once 5 min) Yes Murphy et al. [39] 2019A MN SM 8 26 55.7 ±17 18/8 26 •MGD grade ≥1 OPTASE moist heat eye mask (Twice a day 10 min, temperature not reported) Warm towel compress (Twice a day 10 min) Yes Murphy et al. [39] 2019B MN SM 8 28 58.6 ±15.1 19/9 28 •MGD grade ≥1 MGDRx eyebag (Twice a day 10 min, temperature not reported) Warm towel compress (Twice a day 10 min) Yes Wang et al. [41] 2024 MT UM 12 134 41.6 ±13.7 100/34 134 •OSDI score ≥13 points •NIBUT score <5 s •ST ≥5 mm after 5 min •MG quality and expressibility abnormalities Warmyou Steam eye mask (Twice a day 12 min, 41 C◦) Warm towel compress (Twice a day 12 min) No CoI =Conflict of interest; SM =Single-masked; F =Female; F/U =Follow-up; M =Male; MG =Meibomian gland; MN =Monocenter; MT =Multicenter; NIBUT =Noninvasive tear film break-up time; OSDI =Ocular surface disease index; RCTs =Randomized controlled trials; ST =Schirmer test without anesthesia; UM =Unmasked. a Expressed as weeks except for Tan et al., 2017 [37], which was expressed in hours. b Expressed as mean ±SD (standard deviation). Table 3 Summary of included RCTs comparing moisture chamber goggles Vs. warming compress devices. Author (date) Design F/U a Patients Age b Sex (F/M) Eyes Inclusion criteria Eyelid warming device Control CoI Sim et al. [36] 2014C MN SM 12 41 52 ±10.7 25/16 41 •Dry eye symptoms •At least one MG opening with pouting •At least one observable plugged MG Blephasteam (Twice a day 10 min, temperature not reported) EyeGiene (Twice a day 10 min, temperature not reported) No Olafsson et al. [40] 2021 MT UM 24 48 55.4 ±8.2 33/15 96 •NIBUT score <10 s •ST ≥5 mm after 5 min •MG quality and expressibility score ≥1 Blephasteam (Once a day 10 min, temperature not reported) Therapearl eye mask (Once a day 10 min, temperature not reported) Yes CoI =Conflict of interest; SM =Single-masked; F =Female; F/U =Follow-up; M =Male; MG =Meibomian gland; MN =Monocenter; MT =Multicenter; NIBUT =Noninvasive tear film break-up time RCTs =Randomized controlled trials; ST =Schirmer test without anesthesia; UM =Unmasked. a Expressed as weeks. b Expressed as mean ±SD (standard deviation). A. Ballesteros-S´ anchez et al. The Ocular Surface 37 (2025) 33–46 37 Both studies reported no significant changes in TBUT across the two groups. However, Tan et al. [37] reported that the application of the Bruder® moist heat eye compress resulted in an improvement in TBUT (3.3 ±3.8 s; P not reported), while it remained unchanged in the warm towel compress group. Murphy et al. [39] (2019 A) indicated that the use of the OPTASE® moist heat mask led to improvements in dry eye symptoms (−12.5 ±10.8 points; P =0.022) and CFS (−0.3 ±0.2 points; P =0.035), while ST decreased (−5 ±4.3 mm; P =0.570) and TBUT remained unchanged. By contrast, the warm towel compress group showed smaller improvements in dry eye symptoms (−9.1 ±7.2 points; P =0.038) and worsening in TBUT (−1 ±0.8 s; P =0.954) and ST (−2 ±1.6 mm; P =0.209), although the reduction in CFS was greater (−0.8 ±0.7 points; P =0.043). Murphy et al. [39] (2019 B) reported that the application of the MGDRx Eyebag® improved dry eye symptoms (−23.1 ±21.6 points; P =0.023), TBUT (1.7 ±1.6 s; P =0.870), and CFS (−0.7 ±0.6 points; P =0.007), while ST decreased (−5.7 ±5.3 mm; P = 0.935). In comparison, the warm towel compress group demonstrated smaller improvements in dry eye symptoms (−9.1 ±7.2 points; P = 0.038), worsening of TBUT (−1 ±0.8 s; P =0.954), and a similar reduction in CFS (−0.8 ±0.7 points; P =0.043), along with a decrease in ST (−2 ±1.6 mm; P =0.209). Wang et al. [41] observed that the application of the Warmyou® steam eye mask led to improvements in dry eye symptoms (−20.7 ±42.1 points; P <0.05), TBUT (2.3 ±4.7 s; P <0.05), ST (1.2 ±2.4 mm; P <0.05), and CFS (−1.3 ±2.6 points; P < 0.05). In contrast, the warm towel compress group exhibited smaller improvements in dry eye symptoms (−6.7 ±13.8 points; P <0.05) and CFS (−0.5 ±1 points; P <0.05), while ST decreased (−1.2 ±2.5 mm; P >0.05) and TBUT remained unchanged. Table 4 Intra-group and inter-group differences outcomes of eyelid warming devices Vs. control. Author (Date) Eyelid warming devices Control Inter-group differences b OSDI NIBUT MG dropout OSDI NIBUT MG dropout OSDI NIBUT MG dropout Bilkhu et al. [35] 2014 Baseline 43.9 ±13.4 9.5 ±3.7 2.9 ±1.3 43 ±14.4 9.2 ±3.6 2.8 ±1.2    Last visit 20.7 ±8.7 11.4 ±3.5 2.1 ±1.2 39.8 ±12.7 9.3 ±3.5 2.7 ±1   Difference a, c ¡23.2 ± 28.1*1.9 ± 2.3*¡0.8 ± 0.9*¡3.2 ± 3.8 0.1 ± 0.1 ¡0.1 ± 0.1 ¡20*1.8* ¡0.7* Ngo et al. [38] d 2018 Baseline 39.1 ±12.6 2.9 ±1.3 2.2 ±1.4 41.3 ±15.4 3.1 ±1.2 2.2 ±1.6     Last visit 27.7 ±14.6 3.0 ±0.9 2.2 ±1.4 35.6 ±23.2 3.1 ±0.9 1.8 ±1.4    Difference a, c ¡11.4 ± 8.9*0.1 ± 0.07 0 ± 0.1 ¡5.7 ± 4.7 0 ± 0.1 ¡0.4 ± 0.3 ¡5.7 0.1 0.4 CFS =Corneal fluorescein staining; LLT =Lipid layer thickness; MG =Meibomian gland; NIBUT =Non-invasive tear film break-up time. * Statistical significance reported by studies with P-value <0.05. a Defined as “Last visit – Baseline.” b Defined as “Eyelid warming device (Last visit – Baseline) – Control group (Last visit – Baseline) .” c The standard deviation of the intra-group mean difference was estimated following the guidelines of the Cochrane Handbook for Systematic Reviews of Interventions [49]. d The standard deviation of baseline and last visit outcomes was estimated following the guidelines of the Cochrane Handbook for Systematic Reviews of Interventions [49], based on the 95 % confidence intervals reported by Ngo et al. [38]. Fig. 3. Overall efficacy of eyelid warming devices compared to the control groups. Forest plot showing the standardized mean difference (SMD), 95 % confidence intervals (CI) and P value for the change dry eye symptoms, non-invasive tear film break-up time (NIBUT) and meibomian gland (MG) dropout. A random effects model was performed, revealing no statistically significant difference between eyelid warming devices and the control group. A. Ballesteros-S´ anchez et al. The Ocular Surface 37 (2025) 33–46 38 Table 5 Intra-group and inter-group differences outcomes of eyelid warming devices Vs. warm towel compress. Author (Date) Eyelid warming devices Warm towel compress Inter-group differences b Dry eye symptoms Tear film stability ST CFS Dry eye symptoms TBUT ST CFS Dry eye symptoms TBUT ST CFS Sim et al. [36] d 2014 A Baseline 41.1 ±27.6 2 ±0.4 13 ±4.1 1.5 ±0.7 52.4 ±20.1 1.9 ±0.3 8.7 ±4.4 3.2 ±0.7     Last visit 22.6 ±17.7 2.8 ±0.5 14.3 ±11.2 0.5 ±0.3 38.6 ±20.5 2.3 ±0.4 10.5 ±11 0.7 ±0.2     Difference a, c ¡18.5 ± 21.9 0.8 ± 0.6 1.3 ± 1.5 ¡1 ± 1.2 ¡13.8 ± 16.3 0.4 ± 0.6 1.8 ± 2.1 ¡2.5 ± 2.9 ¡4.7 0.4 ¡0.5 1.5 Sim et al. [36] d 2014 B Baseline 27.2 ±20.9 2.3 ±0.4 9.3 ±4.7 3.4 ±0.9 52.4 ±20.1 1.9 ±0.3 8.7 ±4.4 3.2 ±0.7     Last visit 20.5 ±18.4 2.5 ±0.3 12.4 ±10.5 0.9 ±0.4 38.6 ±20.5 2.3 ±0.4 10.5 ±11 0.7 ±0.2     Difference a, c ¡6.7 ± 6.5 0.2 ± 0.5 3.1 ± 3¡2.5 ± 2.4 ¡13.8 ± 16.3 0.4 ± 0.6 1.8 ± 2.1 ¡2.5 ± 2.9 7.1 ¡0.2 1.3 0 Tan et al. [37] 2017 Baseline NR 8.7 ±8.1 NR NR NR 8.5 ±5.2 NR NR     Last visit NR 12.0 ±9.4 NR NR NR 8.7 ±6.6 NR NR     Difference a, c - 3.3 ± 3.8 - - - 0.2 ± 0.1 - - - 3.1 - - Murphy et al. [39] d 2019 A Baseline 39 ±20.3 4.1 ±0.9 19.3 ±6.9 0.5 ±0.6 24. 2 ±15.3 6.2 ±3.3 12.1 ±3.3 1 ±0.6     Last visit 26.5 ±17.2 4.6 ±0.6 14.3 ±3.2 0.2 ±0.2 15.1 ±12.8 5.2 ±1.9 9.4 ±1.8 0.2 ±0.2     Difference a, c ¡12.5 ± 10.8*0.5 ± 0.4 ¡5 ± 4.3 ¡0.3 ± 0.2*¡9.1 ± 7.2*¡1 ± 0.8 ¡2 ± 1.6 ¡0.8 ± 0.7*¡3.4 1.5 ¡3 0.5 Murphy et al. [39] d 2019 B Baseline 39.8 ±23.2 6.2 ±2.4 27.5 ±6.8 0.9 ±0.5 24. 2 ±15.3 6.2 ±3.3 12.1 ±3.3 1 ±0.6     Last visit 16.7 ±13.1 7.9 ±3.2 21.8 ±4.8 0.2 ±0.2 15.1 ±12.8 5.2 ±1.9 9.4 ±1.8 0.2 ±0.2     Difference a, c ¡23.1 ± 21.6*1.7 ± 1.6 ¡5.7 ± 5.3 ¡0.7 ± 0.6*¡9.1 ± 7.2*¡1 ± 0.8 ¡2 ± 1.6 ¡0.8 ± 0.7*¡14 2.7 ¡3.7 0.1 Wang et al. [41] 2024 Baseline 27 ±13.7 2.7 ±1.7 11.3 ±9 1.8 ±2.4 25.9 ±13.3 2.9 ±1.7 10.6 ±9.3 1.9 ±3.0     Last visit 6.3 ±5.5 5 ±3 12.5 ±8.8 0.5 ±1.3 19.2 ±11.5 3.2 ±2.4 9.4 ±7 1.4 ±2.5     Difference a, c ¡20.7 ± 42.1*2.3 ± 4.7*1.2 ± 2.4 ¡1.3 ± 2.6*¡6.7 ± 13.8*0.3 ± 0.6 ¡1.2 ± 2.5 ¡0.5 ± 1*¡14*2*2.4 ¡0.8* CFS =Corneal fluorescein staining; OSDI =Ocular surface disease index; ST =Schirmer test. * Statistical significance with a P value <0.05. a Defined as “Last visit – Baseline.” b Defined as “Eyelid warming device (Last visit – Baseline) – Warm towel compress (Last visit – Baseline).” . c The standard deviation of the intra-group mean difference was estimated following the guidelines of the Cochrane Handbook for Systematic Reviews of Interventions [49]. d Baseline and last visit mean ±standard deviation was estimated following the guidelines of the Cochrane Handbook for Systematic Reviews of Interventions [49], based on the median and interquartile range reported by Sim et al. [36] and Murphy et al. [39] for TBUT, ST, and CFS. A. Ballesteros-S´ anchez et al. The Ocular Surface 37 (2025) 33–46 39 Regarding inter-group outcomes, all studies showed favorable results for eyelid warming devices in alleviating dry eye symptoms and improving TBUT, except Sim et al. [36] (2014 B). Specifically, Sim et al. [36] (2014 A) obtained a reduction of −4.7 points (P =0.021) and increase of 0.4 s (P =0.612) in dry eye symptoms and TBUT, respectively. Similarly, Murphy et al. [39] (2019 A) reported a reduction of −3.4 points (P not reported) in dry eye symptoms and increase of 1.5 s (P not reported) in NIBUT. In contrast, Murphy et al. [39] (2019 B) and Wang et al. [41] showed larger improvements, both indicating a reduction of −14 points (P not reported for Murphy et al. [39]; P <0.05 for Wang et al. [41]) in dry eye symptoms and increases of 2.7 (P not reported) and 2 s (P <0.05) in TBUT, respectively. In addition, Tan et al. [37] also reported favorable results for the eyelid warming devices in enhancing TBUT, with a value of 3.1 s (P >0.05). However, contrasting results were observed for ST and CFS between the studies. 3.3.1. Meta-analysis Forest plots showing the efficacy of eyelid warming devices Vs. warm towel compresses is presented in Fig. 4. Four studies were included in the meta-analysis [36,37,39,41]. The overall efficacy did not favor either group. However, only the change in tear film stability (SMD 0.97; 95 % CI: 0.32 to 1.61; P =0.003; I 2 =83 %) indicated that the eyelid warming devices had significantly better outcome than the warm towel compress group. In the sensitivity analyses, overall efficacy was also not in favor of either treatment. However, the change in OSDI questionnaire (MD 9.41; 95 % CI: 1.70 to 17.13; P =0.02; I 2 =49 %) and NIBUT (MD 2.24; 95 % CI: 1.20 to 3.28; P <0.0001; I 2 =71 %) indicated that the eyelid warming devices had significantly better outcomes than the warm towel compresses, while TBUT, ST and CFS showed no statistical significance between the two groups. More detailed results of the sensitivity analyses are presented in the Supplemental Digital Content 2 and 3. Fig. 4. Overall efficacy of eyelid warming devices compared to the warm towel compress groups. Forest plot showing the standardized mean difference (SMD), 95 % confidence intervals (CI) and P value for the change dry eye symptoms, tear film break-up time (TBUT), Schirmer test (ST) and corneal fluorescein staining (CFS). A random effects model was performed, revealing no statistically significant difference between eyelid warming devices and warm towel compress. A. Ballesteros-S´ anchez et al. The Ocular Surface 37 (2025) 33–46 40 3.4. Moisture chamber devices vs. warm compress devices Regarding the measurement methods, Sim et al. [36] (2014 C) utilized the SANDE questionnaire, whereas Olafsson et al. [40] employed the OSDI questionnaire. Both studies evaluated ST without anesthesia, with results reported as the wetting length of the strip over a 5-min period. Additionally, the Oxford grading scale was applied to assess CFS. In term of tear film stability, both studies performed NIBUT. Intra-group and inter-group efficacy and safety results are displayed in Table 6. Regarding intra-group outcomes, Sim et al. [36] (2014 C) reported that the application of Blephasteam® resulted in improvements in dry eye symptoms (−18.5 ±21.9 points; P not reported), ST (1.3 ± 1.5 mm; P not reported), and CFS (−1 ±1.2 points; P not reported). However, the EyeGiene® mask showed smaller improvements in dry eye symptoms (−6.7 ±6.5 points; P not reported), while ST (3.1 ±3 mm; P not reported) and the reduction in CFS was greater (−2.5 ±2.4 points; P not reported). TBUT remained unchanged in both groups. Similarly. Olafsson et al. [40] found that the use of Blephasteam® improved dry eye symptoms (−15.5 ±18.8 points; P not reported), TBUT (3.8 ±4.6 s; P not reported) and ST (4.6 ±5.6 mm; P not reported), while CFS remained unchanged. Conversely, the Therapearl® eye mask obtained lower improvements in dry eye symptoms (−11.7 ±13.5 points; P not reported) and TBUT (1.9 ±2.2 s; P not reported), while ST (−1.8 ±2.1 mm; P not reported) and CFS (1 ±1.2 points; P not reported) showed a worsening. Regarding inter-group outcomes, both studies showed favorable results for the Blephasteam® in alleviating dry eye symptoms and enhancing TBUT. Specifically, Sim et al. [36] (2014 C) obtained a reduction of −11.8 points (P not reported) in dry eye symptoms and increase of 0.6 s (P not reported) in TBUT, while Olafsson et al. [40] reported a reduction of −3.8 points (P =0.92) in dry eye symptoms and increase of 1.9 s (P =0.89) in TBUT. However, contrasting results were observed for ST and CFS between the 2 studies. 3.4.1. Meta-analysis Forest plots showing the efficacy of moisture chamber devices Vs. warm compress devices is presented in Fig. 5. Two studies were included in the meta-analysis [36,41]. The overall efficacy did not favor either group. However, only the change in TBUT (SMD 0.75; 95 % CI: 0.23 to 1.28; P =0.005; I 2 =30 %) indicated that Blephasteam® had significantly better outcomes than the warm compress eyelid devices group. 3.5. Risk of bias The risk of bias summary of the included studies is presented in Fig. 6A. Risk of bias assessment was classified into three evidence level groups: (1) studies with a low risk of bias (Bilkhu et al. [35], Sim et al. [36], Ngo et al. [38] and Murphy et al. [39]) (2) studies with an unclear risk of bias (Olafsson et al. [40] and Wang et al. [41]) and (3) studies with a high risk of bias (Tan et al. [37]). The overall risk of bias summary of the domains used in each study is presented in Fig. 6B. The items used to assess the risk of bias showed an overall risk of bias that was low at > 50 %. The Robvis tool (NIHR, Bristol, UK) was used to create risk of bias assessment figures [47]. 4. Discussion MGD is the most frequent pathogenic mechanism underlying evaporative dry eye, which is recognized as the most prevalent variant of DED [5–7]. New in-office treatments have emerged to improve dry eye signs and symptoms. However, due to the chronicity of this disease and the elevated costs of these treatments [25,26], warm compress therapy remains as a first-step treatment [29]. This systematic review with meta-analysis aimed to investigate the effectiveness of eyelid warming devices in the treatment of MGD. To achieve this goal, eyelid warming devices were compared to control groups and warm towel compresses. Table 6 Intra-group and inter-group differences outcomes of moisture chamber devices Vs. warm compress devices. Author (Date) Moisture chamber devices Warm compress devices Inter-group differences b Dry eye symptoms TBUT ST CFS Dry eye symptoms TBUT ST CFS Dry eye symptoms TBUT ST CFS Sim et al. [36] 2014C Baseline 41.1 ±27.6 2 ±0.4 13 ±4.1 1.5 ±0.7 27.2 ±20.9 2.3 ±0.4 9.3 ±4.7 3.4 ±0.9     Last visit 22.6 ±17.7 2.8 ±0.5 14.3 ±11.2 0.5 ±0.3 20.5 ±18.4 2.5 ±0.3 12.4 ±10.5 0.9 ±0.4     Difference a, c ¡18.5 ± 21.9 0.8 ± 0.6 1.3 ± 1.5 ¡1 ± 1.2 ¡6.7 ± 6.5 0.2 ± 0.5 3.1 ± 3¡2.5 ± 2.4 ¡11.8 0.6 ¡1.8 1 Olafsson et al. [40] d 2021 Baseline 38.7 ±20.4 4.7 ±2.1 17.7 ±10.5 1.50 ±0.7 32.9 ±19.2 4.9 ±2.5 18.9 ±11.3 1.00 ±0.01     Last visit 23.2 ±21.7 8.5 ±4.5 22.3 ±12.5 1.50 ±1.8 21.2 ±14.8 6.8 ±3.7 17.1 ±9.5 2.00 ±1.2     Difference a, c ¡15.5 ± 18.8 3.8 ± 4.6 4.6 ± 5.6 0 ± 0.1 ¡11.7 ± 13.5 1.9 ± 2.2 ¡1.8 ± 2.1 1 ± 1.2 ¡3.8 1.9 6.4 ¡1* CFS =Corneal fluorescein staining; LLT =Lipid layer thickness; MG =Meibomian gland; TBUT =tear film break-up time with fluorescein; ST =Schirmer test without anesthesia. * Statistical significance reported by the studies with P-value <0.05. a Defined as “Last visit – Baseline.” b Defined as “Moisture chamber devices (Last visit – Baseline) – Eyelid warm compress devices (Last visit – Baseline).” . c The standard deviation of the intra-group mean difference was estimated following the guidelines of the Cochrane Handbook for Systematic Reviews of Interventions [49]. d The standard deviation of baseline and last visit outcomes was estimated following the guidelines of the Cochrane Handbook for Systematic Reviews of Interventions [49], based on the 95 % confidence intervals reported by Olafsson et al. [40]. A. Ballesteros-S´ anchez et al. The Ocular Surface 37 (2025) 33–46 41