Sealing Properties of Close-Fitting Masks Worn Over Facial Hair

METHODS

Eligible studies evaluated the seal integrity of close-fitting masks or respirators, designed to achieve the highest levels of respiratory protection by applying a seal directly against the skin, in the presence of facial hair. Thus, masks and respirators designed to achieve fit factors no greater than 100, such as disposable dust masks and aerosol filtering facepiece respirators, were excluded. This encompassed many recent studies of mask performance in healthcare settings since the onset of the coronavirus pandemic.^6–8 Studies that compare a close-fitting mask with other types were excluded if seal leaks were not evaluated specifically.⁹ Only studies of mask leakage in human use are included, noting recent work to standardize leak testing with facial hair using a head manikin.¹⁰

Multiple searches for relevant primary sources of data were conducted on Medline, Scopus, and PubMed, and using Google Scholar, employing combinations of the search terms “beard” OR “facial hair” AND “respiratory protection” OR “respirator” OR “face mask” OR “mask seal” OR “mask leak” OR “mask”. Additional searches were conducted in the Aerospace Medical Association’s journal archives, the Defense Technical Information Centre, and past United Kingdom military reports.

Two sources of secondary evidence feature repeatedly in search results.^2,3 In particular, the review by Stobbe et al. is ubiquitous.² Additional searches on Google Scholar and Scopus were conducted for any papers citing either of these reviews and those abstracts were assessed for relevance. It is unlikely that any relevant research would not have referenced one or other of these reviews. Efforts were made to obtain all 12 primary sources of information reviewed by Stobbe et al. and 3 additional references considered by Bolsover³ that were not in the Stobbe et al. review. Of these 15 primary sources, 5 were unavailable. One Swedish technical report was validated by an independent English translation in the United Kingdom and has been retained. Outcomes from the other four could not be verified independently and they have been excluded. One paper, excluded by Stobbe et al. on the basis that it did not include data, did, in fact, include data and has been included in this review.¹¹ On the other hand, a study of disposable dust respirators was included by Stobbe et al., but has been excluded from the current review, as these are not close-fitting masks and do not achieve a skin-tight seal. The study documented comparatively poor mask seals in both bearded and clean-shaven users relative to the protection achieved by close-fitting designs.¹² All available primary sources of information were re-evaluated for this review and their reference lists screened for any additional relevant sources of original data.

All included studies were evaluated for evidence quality, independently by two authors, against the criteria and scoring system shown in Table I. The correct score against each criterion was usually unambiguous, resulting in high levels of interrater agreement. Occasional differences typically reflected errors in understanding or recording by either rater and were easily identified and resolved through discussion. For each criterion, all scores from “0” to “3” were possible and each was achieved by at least one study. Summing these, the lowest possible total score was zero and the highest was 15.

Table I. Study Evaluation Criteria and Scoring System.

View Fullscreen

Rating of study designs accords with accepted convention for hierarchy of research evidence, with cohort studies scoring higher than case-control designs and then cross-sectional studies. The minimum level of relevant evidence was considered to comprise evaluation of a single mask type on a single occasion when bearded and again when clean-shaven, regarded as the simplest cross-sectional study. Case-control designs are rated higher but are clearly susceptible to bias from highly variable facial anthropometry between subjects, a major influence on mask leakage. One effective way to control for this is to evaluate the same faces when clean-shaven and then repeatedly with progressive hair growth, i.e., repeated measures using subjects as their own controls. Such cohort studies involving multiple test points were usually of longitudinal, repeated measures design, evaluating leakage in the same individuals as beard growth progressed. Those that included multiple repeated measurements (replicates) at each test visit, thereby minimizing the influence of within-subject variability in mask fit from one donning to the next, rated highest.

For mask test conditions, studies evaluating a single mask under a single (positive or ambient) supply pressure, on a single occasion (including a single bearded vs. clean-shaven assessment) were rated “0”. If multiple masks were evaluated similarly, each on only one occasion, the study was rated “1”. At the other end of the scale, if multiple masks, or supply pressure conditions, were evaluated on multiple occasions (e.g., with progressive beard growth or using and reporting rigorous qualitative as well as quantitative testing), the study was rated “3”. A single mask evaluated on multiple occasions was rated “2”, considered to provide more compelling evidence than multiple masks that were only assessed once.

Bearded cohorts were generally small. Studies reporting outcomes involving a single bearded individual were rated “0”, while those involving more than 12 scored “3”. Intermediate scores were arbitrarily distinguished according to whether six or more bearded subjects participated.

Challenge agent “sniff” testing, where the ambient concentration is uncontrolled, is a subjective test that is not intended as a research tool. Studies that only used this technique were rated “0”. Only studies that included a quantitative leak measurement method rated a higher score. Measurement of respired gas composition or direct measurement of leak rate both scored “3”, while measurement of ambient and mask cavity concentrations of challenge agents scored “2”, being technically difficult to achieve consistent, well-calibrated data. Invariably, indirect leak indicators are relatively insensitive and were graded “1”.

Finally, regarding representative testing under wear conditions, studies that only evaluated masks at rest were graded “0”. Ratings of “1”, “2”, or “3” were reserved, respectively, for studies that evaluated leakage under one, two, or three specific activities which challenged the integrity of mask seals. These comprise increased pulmonary ventilation (including exercise or deep breathing), movements (especially of the head), and speech.

Studies were stratified based on total score. Consideration was given to possible meta-analysis of data from high quality studies initially, and then to medium quality studies, before drawing qualitative inferences from the reported study outcomes.

RESULTS

Besides the studies evaluated in earlier reviews,^2,3 this review includes an additional nine sources of primary evidence. Table II provides summary information for all these sources in chronological order, identified by the first author’s name, noting those reviewed previously and identifying those that were unavailable. Some sources report relevant evidence from multiple studies; where these use different methods, they are annotated A and B. With reference to the criteria in Table I, Table III lists the scores achieved by each of the studies. The median total score was 9, range 5–13. The top five (24%) studies achieved ≥11 and are rated as high quality, with the remainder divided into eight (38%) rated medium quality, scoring 8–10, and eight (38%) rated low quality, scoring ≤7.

Table II. Overview of Study Designs and Methods for Primary Sources of Data Included in the Current Review in Chronological Order by First Author.

View Fullscreen

Table III. Study Evaluation Scores, in Chronological Order by First Author.

View Fullscreen

Three high quality cohort studies with repeated measures designs achieved scores of 12 or 13.^17,19,28 The largest case-control study was conducted to a high standard and scored 12.²¹ This included rigorous qualitative testing to achieve “best fits” prior to quantitative assessment, reporting the differential outcomes for clean-shaven and bearded subjects. These discrete mask test conditions have contributed to the study’s assessment for mask test schedule. The fifth high quality study had a mixed design whereby clean-shaven and bearded cohorts undertook repeated testing of two masks over several days, with replicates on each day of testing.²⁷ It is scored as a repeated measures cohort study rather than a case-control study.

Of the eight medium quality studies, one primary source was unavailable for detailed review. However, its outcomes are validated by an independent contemporary translation conducted for the Royal Aircraft Establishment at Farnborough, United Kingdom.¹⁸ Its score is based on this translation augmented by data from secondary sources.^2,3

None of the available data from high and medium quality studies are suitable for meta-analysis since they do not report mask leakage using compatible metrics or data collected under directly comparable conditions. Considering the high quality studies, one uniquely reports mask fit factors relative to specific beard lengths,²⁸ one uniquely reports nitrogen washout rates breathing 100% oxygen,¹⁷ one measured mask protection factors for a positive supply pressure system at fortnightly intervals for 8 wk,¹⁹ one is the largest case-control study of respirators without safety pressure,²¹ and the last was the mixed design study that measured leak rates using the controlled negative pressure method.²⁷

The eight medium quality studies are similarly disparate, reporting different measures in a variety of mask types, under ambient or positive supply pressures, at varying stages of beard growth. Two early studies did both measure percentage penetration by challenge agent wearing ambient pressure respirators while clean-shaven and then repeatedly over the first few days of beard growth.^13,15 However, the later report included comparison data from the earlier one, so this is not repeated here.

Study outcomes are summarized concisely in Table IV. Overall, 19 of 21 discrete studies (90%) report degraded mask performance in the presence of facial hair, including all the high and medium quality studies. Only two lower quality studies failed to do so, both after evaluating quick-don aviator masks.^4,5 The earlier (unpublished) 2020 study was conducted in a hypobaric chamber using emergency flow (positive supply pressure).⁵ It did not recognize the significance of transient reduction in peripheral oxygen saturation (S_po₂) to 93% in a bearded subject breathing 100% oxygen at an equivalent pressure altitude of 25,000 ft (7620 m), when S_po₂ should be >99%. This implies a substantial (∼50%) inward leak of ambient air that could be catastrophic at higher altitudes. The more recent 2025 study also relied only on S_po₂ to infer mask leaks while evaluating a mask in Normal (ambient pressure, air dilution) mode in a normobaric hypoxia chamber.⁴ At sea level ambient pressure, even if the mask breathing gas supply was air, the plateau of the hemoglobin oxygen dissociation curve would maintain S_po₂ close to normal saturation as alveolar oxygen tension fell, making the method insensitive to 15–20% of inward leaks of ambient hypoxic gas. While inconsequential at sea level, such leaks may result in considerable physiological hypoxia at high altitude. When oxygen saturation is already compromised above 33,500 ft (10,211 m), small falls in oxygen supply concentration or partial pressure will result in more severe desaturation. If the mask breathing gas supply was oxygen, a demand air mix regulator would typically run oxygen-rich, delivering over 30% oxygen to the mask cavity at sea level ambient pressure. This is likely to promote insensitivity to even greater leaks, even allowing for inward regulator dilution by ambient hypoxic gas. Hence, these studies do not provide reliable evidence that quick-don masks seal satisfactorily in the presence of facial hair. In contrast, one high quality study assessed three designs of quick-don mask by measuring effectiveness of nitrogen washout breathing 100% oxygen.¹⁷ While clean-shaven mask performance was consistently reliable, bearded performance was highly unreliable. There were no other studies of quick-don masks with facial hair and only one small case-control study that reports data on an aviator oro-nasal type oxygen mask, also suggesting that substantial but highly variable leaks can occur with beards.¹¹

Table IV. Overview of Study Outcomes Ranked by Study Score.

View Fullscreen

DISCUSSION

The overwhelming weight of evidence from almost all these studies indicates that facial hair sitting between the skin and the seal of a close-fitting mask can degrade its performance, making it unpredictable and unreliable. At ambient supply pressures, multiple studies indicate that mask seals are compromised early and progressively over just a few days of hair growth.^15,23,25 Some masks may be compromised by early stubble, with one report of impaired fit of three different masks after only 8 h of beard growth.¹⁸

The magnitude of the decrement in mask performance with facial hair can be considerable. At ambient supply pressures, the most compelling data in this regard come from the largest case-control study of half-mask and full-face respirators.²¹ For the half-mask, the normalized cumulative frequency plot of log₁₀ mask fit factors of 54 bearded men was down-shifted by approximately 2 orders of magnitude relative to that of 188 clean-shaven controls. With the full-face mask, the disparity was even greater; while 115 clean-shaven men achieved excellent fit factors (60% >10,000 and 97% >1000), bearded performance was poor, with the best achieving only 75.

Relative to being clean-shaven, estimates of increased leakage with established beards include factors of 10,²⁷ 50–100,¹⁴ and even 500–1000.²⁹ This last Dutch overview is by the author of two primary sources, cited by Stobbe et al. and Bolsover, that were not available for this review. It includes summary data indicating that the best ambient pressure, bearded protection factors were worse than those achievable when clean-shaven, by factors of 5 for half-masks, 500 for full-face masks, and 1000 for compressed air masks.²⁹ To avoid concern over possible exclusion bias, the other primary sources that were unavailable for review also purportedly indicate that facial hair interferes with mask seals.^2,3

Supplying breathing gas under positive pressure will decrease inward leakage, but facial hair may still compromise mask performance under such conditions, although this may take weeks rather than days of growth to become established.^18,19,23 Protection factors achieved by eight subjects with one positive pressure system were consistently >20,000 when clean-shaven, but became highly variable and unreliable with fortnightly measurements during beard growth. Of three replicates at each assessment, the lowest median protection factor of the eight subjects fell from >20,000 to 669 at 2 wk, 75 at 4 wk, 49 at 6 wk, and 25 at 8 wk.¹⁹ In a detailed evaluation of a positive pressure (small, medium, or large) half-mask respirator, six subjects were tested clean-shaven, with 3 d stubble and after 2–3 mo of beard growth.²³ When clean-shaven, average challenge agent penetration stayed below 0.001% at all supply flow rates and was unaffected by heavy treadmill exercise. However, penetration increased at reduced flows in the presence of stubble and greatly so with established beards, even with flow rates still well over 100 L · min⁻¹.

Multiple studies report highly variable performance between different mask (and seal) designs in the presence of facial hair.^14,15,18 The most compelling data in this respect come from a repeated measures study of four individuals who each evaluated three different quick-don aviator masks (partially randomized design), with replicates of each bearded and clean-shaven measurement, at rest and with exercise.¹⁷ The resting data are most revealing. All masks performed satisfactorily when the subjects were clean-shaven. However, one mask was ineffective in all four subjects when bearded, and one subject was unable to achieve an effective washout with any mask while bearded. The data also suggest other idiosyncratic responses with specific subject-mask combinations. Data from this study also highlight the far greater within-subject variability in performance when bearded, from one replicate to another, indicating the unpredictability of achieving an adequate seal with facial hair from one donning to the next.

Besides duration of growth, other beard characteristics may influence leakage. The reduction in bearded mask fit factors has been related to hair length and areal density (hairs/cm²), but not to texture (coarseness).²⁸ Another study relates leakage to hair aspect ratio (diameter over length).²⁵ This multiethnic study noted that while hair characteristics differed between the groups, there was no significant difference in leak rates between them. Another study suggests a likely interaction between facial anthropometry and magnitude of leakage with facial hair, although the number of bearded subjects in this study was small.²⁷

Many of the studies included in this review took care to achieve an acceptable mask fit, if not the best possible, prior to testing. Many also encompassed tests under representative operating conditions, including increased respiratory demand and/or head movements and/or speech. However, none of the studies reviewed here tested for leaks under donning and wear conditions representative of operational use of quick-don masks, i.e., when donning a single-sized mask in haste (less than 5 s) followed by a communications headset. In this respect, the available data concerning leakage in the presence of facial hair could even be over-optimistic. Study methodologies will tend to be biased in favor of minimizing leak detection by assuring good mask fits beforehand. In operational use, mask leaks are likely to be worse than laboratory data suggest, even when clean-shaven, but should be anticipated to be substantially less reliable in bearded use.

Regarding this review’s limitations, many of the included studies do appear dated, but that does not invalidate their data, outcomes, or conclusions. Few of these studies evaluated quick-don aviator systems specifically. Unfortunately, some recent attempts to do so have used designs and methods that do not specifically measure leakage and, therefore, may miss important effects of facial hair.^4,5 The most relevant study of aviator quick-don masks remains the one designed and conducted at the Civil Aeromedical Institute on behalf of the Federal Aviation Administration and which underpins Advisory Circular 120-43 concerning the influence of beards on oxygen mask efficiency.^17,30 While this study is almost 50 yr old, recent work suggests that we are no closer to finding a mask seal design that provides a reliable close fit in the presence of facial hair.³¹ In the absence of contemporary, high quality, and representative evaluation of modern quick-don systems, it remains necessary to draw upon all available data pertaining to bearded use of close-fitting respiratory protection to infer the consequences for leakage with quick-don masks.

This review has focused on inboard mask leakage. If a quick-don mask seal is compromised in emergency mode, the delivery of 100% oxygen under positive (safety) pressure will generate outboard leakage and decrease the endurance of fixed volume oxygen supplies. Leak rates may approach or even exceed 100 L · min⁻¹ (standard temperature and pressure), resulting in early depletion of the flight crew emergency oxygen supply.³²

In conclusion, facial hair that sits between the skin and the seal of a close-fitting mask is liable to degrade the protection that it provides. This is highly variable within and between individuals, and within and between masks, and the magnitude of leakage is unpredictable from one wear to the next, even in the same individual under otherwise identical conditions. Interactions between individuals and different masks are idiosyncratic and likely to change over time as facial hair growth progresses. There is no immediate prospect of developing a more reliable close-fitting seal design that will reliably accommodate facial hair. As a result, it is unlikely ever to be possible to standardize aeromedical certification of aviator masks for bearded use. There are no published data on the effectiveness of pilot quick-don masks after representative rapid donning, either in clean-shaven or bearded use, and this warrants investigation. Meanwhile, the available evidence appears unambiguous. To function optimally in an emergency, quick-don close-fitting oxygen masks require a reliable skin-tight seal to protect against hypoxia post-decompression or from smoke and fumes; this will not reliably be achieved unless the user is clean-shaven where the mask seal contacts the skin.