Variation in the fitted filtration efficiency of disposable face masks by sex

The study addresses the variability in fitted filtration efficiency (FFE) of commonly worn disposable ear loop masks (KN95, surgical, KF94) compared to NIOSH-certified N95 respirators. Contextually, public use of masks expanded during COVID-19 and with increasing wildfire smoke events, yet public access to fit testing and standardized efficacy data for ear loop masks is limited. The research questions were: (1) How does FFE vary across commonly worn disposable masks, and is sex a determinant of fitted performance? (2) Does a simple ear loop clip fastened behind the head improve FFE, and does its effect differ by sex? The study’s purpose is to fill data gaps on real-world fitted performance and to evaluate a simple modification with potential public health relevance.

Prior work and standards indicate N95 respirators provide ≥95% filtration efficiency under certification protocols (NIOSH-42CFR84), but ear loop mask performance varies with design, materials, and user-specific factors such as training, facial hair, and craniofacial variation. During COVID-19, KN95 and surgical masks were the most commonly worn by the public in the U.S., with KF94 and N95 less common. Various fit modifications (“hacks”) emerged to reduce faceseal gaps. International standards (e.g., FFP1/2/3) rate material filtration but do not capture fitted performance on individuals. Existing studies have evaluated filtration efficiencies of mask materials, the impact of double masking and modifications, and effects of facial hair on fit, underscoring the need for data on fitted performance in diverse users.

Design: Experimental study with two experiments. Experiment 1 quantified baseline FFE across mask types; Experiment 2 assessed FFE with an ear loop clip modification. Participants: Healthy adults (N=100; 50 females, 50 males; average age 32.3 ± 9.2 years; BMI 25.5 ± 3.4). Inclusion/exclusion: Excluded BMI <19.0 or >33.4; history of cardiovascular or chronic respiratory disease, cancer, uncontrolled hypertension (≥140/90 mmHg), or diabetes. All participants were clean shaven. Demographics, height, weight, blood pressure were recorded. Ethical approvals obtained from UNC Chapel Hill IRB and EPA Human Subjects Safety Review; clinicaltrials.gov registration NCT05388201. Masks: 3M N95 Aura 9205+ (tri-fold), KN95 (Zhongshan Saifute, China), 3‑ply surgical (Hannah Linen, USA), KF94 size large (Dr. Puri, KM Corporation, Republic of Korea). Testing protocol: Modified OSHA Ambient Aerosol CNC Quantitative Fit Testing Protocol (Table A-2-RESPIRATORS) in an aerosol-supplemented chamber using NaCl particles (mean ~50 nm; TSI 8026 generator). Particle counts measured with paired condensation particle counters (TSI 3775) simultaneously in ambient chamber and behind-the-mask via an aluminum sampling port with Luer lock and conductive tubing, at 1-second resolution. Each test lasted ~2–3 minutes. Donning assistance was provided to correct visible gaps before testing. Clip modification: Plastic clip (5.3 cm × 1.5 cm) used to tighten ear loops, fastened centrally behind head between ears at nearest linkage points. Outcomes: FFE computed as 100 × [1 − (behind-mask particle count / ambient particle count)]. Experiment 1 measured baseline FFE for all masks. Experiment 2 measured baseline and clip-modified FFE for ear loop masks; delta defined as (clip-modified − baseline). Statistical analysis: Mixed effects linear regression with participant as random intercept. Experiment 1 fixed effects: mask type (N95, KN95, surgical, KF94), sex (−1 male, 1 female), and masksex interactions; post-hoc Tukey tests for sex differences by mask. Sensitivity analyses included age, BMI, and race/ethnicity. Experiment 2 fixed effects: mask type (KN95, surgical, KF94), clip (1 unmodified, −1 modified), sex, and clipsex interaction; sensitivity analyses as above; paired t-tests for sex differences in delta by mask. Additional model of delta included baseline FFE and quartile indicators (Q2–Q4). Analyses performed in R with lme4 and emmeans; BIC for model selection; p<0.05 significant. Results presented as means ± SD and effect estimates with 95% CIs.

• N95 respirator achieved the highest baseline FFE and lowest variability: 97.8% ± 3.2.
• Baseline FFE varied significantly by mask; compared to overall mean FFE, all masks differed significantly except KN95 (p=0.43).
• Sex differences at baseline: males had higher FFE than females for ear loop masks. Estimated female–male mean differences (females lower): KN95 −7.4% (95% CI −13.6, −1.1), surgical −10.1% (−16.3, −3.8), KF94 −15.2% (−21.5, −9.0); N95 difference −0.8% (−7.1, 5.4) not significant.
• Overall, males averaged about 11% higher baseline FFE than females for KN95, surgical, and KF94.
• Clip modification significantly increased FFE overall: estimated mean +6.4% (95% CI 5.7, 7.1; p<0.0001).
• Clip-modified absolute FFE (mean ± SD) and average delta: KN95 80.8% ± 12.0 (delta +11.4%), surgical 66.4% ± 9.4 (delta +8.9%), KF94 73.1% ± 13.9 (delta +17.9%).
• Sex-specific clip effects: Females showed larger improvements than males across all masks. Female deltas: KN95 +18.8%, surgical +13.7%, KF94 +26.1%. Male deltas: KN95 +4.0%, surgical +4.1%, KF94 +9.7%. Interaction term indicated males’ improvement with clip was −13.6% (95% CI −16.4, −10.7) lower than females (p<0.0001).
• Quartile analysis: Clip benefit was greatest among participants in the lowest baseline FFE quartile; model indicated decreasing delta with higher baseline FFE (estimated delta = 31.58 − 0.24×baseline FFE, with negative coefficients for quartiles 2–4).
• Some individuals (primarily males) experienced reduced FFE with the clip, despite average improvement across groups.
• With clip, female KN95 averaged 84.5%, exceeding the 80% FFP1 material standard reference, whereas ear loop masks otherwise less reliably exceeded this threshold.

Findings demonstrate that fitted performance of commonly worn ear loop masks is highly variable across individuals and that sex is a major determinant of baseline FFE. N95 respirators provided consistently high FFE with low variability, confirming their benchmark status. For ear loop masks (KN95, surgical, KF94), males exhibited higher baseline FFE than females, likely reflecting sexual dimorphism in craniofacial dimensions and mask–face interactions. A simple ear loop clip substantially improved FFE for females, often enabling them to match or exceed males’ FFE, while males experienced smaller gains and occasional decrements, suggesting modification benefits diminish as fitted performance approaches mask-specific limits. Quartile analyses support that those with lower baseline FFE benefit most from fit modification. Public health implications are significant: fit modifications can narrow sex-based disparities in protection and improve real-world performance of widely used masks during infectious disease outbreaks and air quality emergencies. The results emphasize the need to consider wearer characteristics and fit enhancements when selecting masks for community protection.

This study quantifies sex-based differences in fitted filtration efficiency for commonly used ear loop masks and demonstrates that a simple ear loop clip can substantially improve FFE, particularly for females, effectively reducing observed disparities. N95 respirators remain the most reliable option for high fitted performance. These insights can guide public mask selection and the adoption of fit modifiers to enhance protection against airborne contaminants. Future research should include broader and more diverse populations (age, BMI, health status), multiple manufacturers and sizes (especially for KN95/KF94), and detailed assessments of craniofacial morphology to optimize mask selection and fit strategies at the individual level.

• Only a single model/manufacturer per mask type was tested; results may not generalize across brands and sizes (notably for KN95 and KF94 which lack universal sizing).
• Study population was restricted to healthy adults with limited age and BMI ranges for safety reasons, which may limit generalizability.
• All participants were required to be clean shaven; results may not apply to individuals with facial hair.
• Testing duration and intensity constraints limited broader sampling.
• Some individuals experienced decreased FFE with clip modification, indicating variability in response.