Health effects of filtering facepiece respirators: Systematic review of pulmonary and cardiovascular effects

Abstract

Filtering facepiece respirators (FFRs) were introduced to protect the wearer by removing small particles from inspired air. FFRs are now also used to reduce the spread of transmissible agents from the wearer and are worn outside traditional healthcare and other workplaces. The COVID‐19 pandemic increased concerns about potential adverse effects on wearers. A PUBMED query retrieved articles through June 2022. Abstracts and selected full‐text articles were systematically reviewed by the authors. This article focuses upon cardiopulmonary physiologic effects (e.g., ventilation, CO₂ elimination, oxygen uptake, and respiratory control) with emphasis upon current and potential research methods as well as summarizing results. 1985 records were identified, of which only 26% were published before 2020. FFR effects on CO₂ elimination appear more likely to be significant than effects on oxygenation or cardiovascular function. While FFRs appear well tolerated by healthy persons, more research is needed for those with pulmonary or cardiac disorders, and for children. Many traditional pulmonary exercise study methods require special care when applied to filtering facepiece respirators. Studying additional parameters may explain the paradox of many subjective discomfort reports despite very limited physiologic effects.

1. INTRODUCTION

The COVID‐19 pandemic accelerated the use of respiratory protective devices (RPDs) in industrial, commercial, healthcare, and community settings. Filtering facepiece respirators (FFRs), one type of RPD, are commonly used in the United States (e.g., N95), Europe (e.g., FFP2,3), and Asia (e.g., KN95). Published opinions about their adverse effects range from reassurances to grave concerns. Understanding the potential consequences of use is important for medical assessment of individual users, optimizing design, and redefining governmental device certification criteria.

This article has three main goals: (a) To summarize research findings about pulmonary and cardiovascular effects of N95s and similar FFRs, this article complements publications aggregating results with methods commonly applied to randomized clinical trials ¹ , ² , ³ by selecting articles to illustrate the range of results and research methodologies; (b) To systematically describe research methods currently in use or potentially useful in the near future, the article emphasizes the importance of attention to methodologic detail to avoid potential misinterpretations (c) To facilitate discussion of research priorities to improve respiratory protection in workplaces and the community.

Adverse effects may include both inadequate protection and direct adverse effects of the device upon the wearer. While this review focuses on the latter, these are interrelated since a poorly tolerated device is unlikely to be properly used. With modern designs, increments in tolerability may be more important than increments in the efficiency of particle capture at the filtration surface. ⁴ The US National Institute for Occupational Safety and Health (NIOSH)‐sponsored BREATHE consortium emphasized improving multiple aspects. ⁵

Many findings about FFRs apply to other RPDs. Surgical masks (SMs) are now used to protect the wearer as well as in limiting spread of users' respiratory secretions. ⁶ Cloth face coverings are also widely used.

Many user effects have been reported. Judgment is needed to assess their magnitude and situations in which they are particularly relevant. For example, mild discomfort might be acceptable during brief, but not prolonged, use. Users are increasingly heterogeneous as FFRs are recommended for use by the general public, and even children, for protection against air pollution from wildland fires and SARS‐CoV‐2.

This article focuses on respiratory and cardiovascular physiologic effects (e.g., ventilation, CO₂ elimination, oxygen uptake, and respiratory control) because these have been considered the primary adverse effects of respirators for many years, may be objectively ascertained, and constitute the basis for certifying respirator designs. More recently, other effects have received attention and will be addressed separately. This report emphasizes current and potential research methods as well as summarizing results.

2. MATERIALS AND METHODS

Abstracts of relevant articles were retrieved through queries in PubMed, the US National Library of Medicine bibliographic database, if “N95,” “FFP2,” or “FFP3” appeared anywhere in the entry. (This captures “KN95” as well). References without abstracts and non‐English documents were excluded. The authors coded abstracts with a multidimensional rating scheme including relevance, methodologic clarity, significance, study type (laboratory, field, case report, or expert opinion), and effects. Full‐text articles most relevant to pulmonary and cardiac physiologic effects were obtained and summarized. Additional articles were included if suggested by reviewed articles or needed for principles. The authors chose articles to include in this review based on likely results' significance, to describe techniques, and to illustrate the variety of study methods. Therefore, this does not imply that nonincluded articles are of lesser quality. Data were managed using Microsoft Access and Clarivate EndNote. After reviewing the articles, the authors (PH and WB) added commentary as deemed appropriate based upon the overall synthesis.

3. RESULTS

The retrieval and selection results are summarized in Figure 1. Initial record retrieval was conducted in December 2020 and complemented in June 2022. Only 26% of records retrieved were published before 2020 (see Figure 2). Records without an abstract, showing non‐English publication, or focused on retinal electrophysiology were automatically excluded. Of the 1985 records retrieved, 253 were selected for detailed review, and 55 are described in this article. Table 1 summarizes abbreviations and glossary. Many underlying physiologic principles are discussed in a recent review ⁷ and in our Online Supporting Information S1.

Figure 1.

Retrieval, coding, and selection. The figure summarizes the process for retrieving, exclusion, coding, and selection for inclusion.

Figure 2.

Cumulative percentage of publications by year. The figure shows the cumulative percentage of total publications identified from the search by year. Data for 2022 reflect only the first 6 months.

Table 1.

Abbreviations and glossary

ANOVA	Analysis of variance
BP	Blood pressure
CBF	Cerebral blood flow
COPD	Chronic obstructive pulmonary disease
CO2 output	Cumulative product of exhaled CO2 percentage times volume. Also expressed as $\int \left(\text{FiCO}2\right)\mathit{dV}$
CPET	Cardio‐pulmonary exercise testing
F	Frequency of breathing (breaths per minute)
FEV1	Forced expiratory volume in the first second of expiration
FFP1	Filtering facepiece respirator that filters at least 80% of airborne particles (European Union Standard)
FFP2	FFR that filters 94% of airborne particles (European Union standard)
FFP3	FFR that filters 99% of airborne particles (European Union standard)
FFR	Filtering facepiece respirator
FiCO2	Inhaled fraction of carbon dioxide
FiO2	Inhaled fraction of oxygen
FOT	Forced oscillometry technique, a method of measuring lung inertance and resistance
FRC	Functional residual capacity (volume in the lungs at the end of a normal exhalation).
Hb%Sat	Hemoglobin per cent saturation
HCP	Healthcare personnel
HFM	Half facemask respirator, a reusable elastomeric respirator covering nose, mouth and chin
HRV	Heart rate variability, an indicator of heart autonomic control
KN95	FFR that removes 95% of airborne particles, not resistant to oil; China certification
ICU	Intensive care unit
N95	FFR that removes 95% of airborne particles, not resistant to oil; U.S. NIOSH certification
NIOSH	U.S. National Institute for Occupational Safety and Health of the Centers for Disease Control and Prevention
PAPR	Powered air‐purifying respirator
PaCO2	Arterial CO2 pressure
PaO2	Arterial O2 pressure
PACO2	Alveolar 02 pressure
PCO2/P02	Partial pressure of carbon dioxide/oxygen
PETCO2	End tidal carbon dioxide pressure
PtcCO2	Trans‐cutaneous pressure of carbon dioxide
PtcPO2	Trans‐cutaneous pressure of oxygen
Ptp	Transpulmonary pressure (between intrapulmonary airways and pleural space)
RPD	Respiratory protective device
SM	Surgical mask, a medical device originally designed to trap exhaled infectious material
st02	Cerebral oxygenation
Te	Expiratory time (duration in seconds)
Ti	Inspiratory time (duration in seconds)
TLC	Total lung capacity: Volume in the lung at the end of a maximal inhalation.
VD	Dead Space Volume. Volume of air in respiratory system that is without gas exchange
VD/VT	Dead Space to tidal volume ratio. An index of the amount of unperfused lung
V̇E	Minute ventilation, volume of exhaled air per minute
V̇EO2	Ratio of the exhaled volume to the volume of oxygen consumed, reflecting ventilatory efficiency
V̇ECO2	Ratio of the volume of air ventilating the lungs to the volume of C02 produced
V̇O2max	Maximal oxygen consumption during the highest level of exercise
VT	Tidal Volume (average breath volume)
W	Work of breathing

This section synthesizes the results in four sections: ventilation and CO₂ elimination, oxygenation, cardiovascular, and respiratory controls. Each includes potential FFR effects, measurement methods, and summary of available data. In addition, we provide perspective for the broader implications and potential limitations for each set of results. Summaries of key articles are shown in Table 2.

Table 2.

Illustrative studies

Author‐Year	CO2, V̇E	Oxygenation	Cardiovascular	Respiratory Control	Method	Findings	Comment
Ade (2004) 8	x	x			11 volunteers studied in rapid ramp protocol on cycle ergometer. Peak power, work, Hb%Sat, and peri‐oral O2% and CO2% were measured. They also measured across the filter material using the standard pneumotachograph and differential pressure transducer.	FFR did not ↓ peak work or Hb%Sat, but did have a small ↑ effect on peri‐oral CO2% and ↓ effect on O2%. The figures indicate that the N95 had a flow resistance of approximately 9 cm H2O L− 1 s−1 and a pressure drop of about 7 cm H20 at flow rate of 48 L min− 1; pressure drops were about 7 and 17 cm H2O at flow rates of approximately 50 and 100 L min−1.	High resistance and pressure drops reported may be inaccurate since only a small cross‐sectional area of filter material was used.
Anil (2022) 9		x	x		21 subjects studied with FFR and PAPR in short exercise protocol. Measures included heart rate variability and Hb%Sat.	Hb%Sat ↓ with both FFR and PAPR. 9 of 21 had > 5% ↓ Hb%Sat with N95.	In addition to assessing mean values of effects, this study indicates that some persons may be particularly affected.
Bansal (2009) 10	x	x	x	x	56 subjects including normal and mild respiratory impairment (chronic rhinitis, mild COPD, mild asthma) performed simulated work tasks with N95 or half face mask (HFM) respirator. Respiratory variables were measured by respiratory inductive plethysmography.	Both healthy and mildly impaired subjects tolerated work tasks well while wearing FFR. Respiratory timing was significantly affected by HFM but not N95.	Authors conclude that FFRs are well tolerated by healthy and mildly impaired subjects with little physiological effect. Study did not include no‐mask controls.
Bharatndu (2020) 11	x		x		Cerebral hemodynamics were assessed with transcranial Doppler ultrasound flow measurements of the middle cerebral artery (MCA) in 154 HCP complaining of FFR‐ related headache. PETCO2 was also measured.	The MCA peak flow ↑ 12% and pulsatility index ↓ 13%. The PetCO2 ↑ from 37 to 40 mmHg soon after donning.	Δs in cerebral blood flow or volume due to ↑ PaCO2 might account for the frequent headaches reported by FFR users. This study does not directly relate symptoms, PaCO2, and cerebral blood flow in individuals].
Cabanillas‐Barea (2021) 12	x	x		x	50 healthy volunteers performed standard 6 min walk test. Respiratory accessory muscle tone was measured in addition to the standard parameters.	No difference in respiratory muscle tone was observed using the FFP2, and no effect was seen on distance, heart rate or oxygen saturation.	While no effect of FFR was noted in these healthy volunteers, this commercially available device might be useful in studying persons with significant pulmonary disease, many of whom routinely use their accessory respiratory muscles.
Engeroff (2021) 1	x	x	x		Formal meta‐analysis of published crossover or randomized controlled trials. In addition to significance of mean differences with/without device, authors report effect size as standardized mean difference.	14 of 1499 retrieved articles met inclusion criteria, but most had a high risk of bias. Combining surgical masks and FFR data (total n = 246) demonstrated ↓ Hb%Sat with ↑ exercise, ↓ V̇E, with no effect upon HR or V̇CO2	Few studies met rigid criteria. Statistical method of standardized mean differences permits combining data from different studies but may inadequately consider underlying differences in physiologic methodology or clinical significance. The article focuses on average difference and does not consider whether there are more severely affected outliers at significant risk.
Epstein (2021) 13	x	x			16 healthy males performed strenuous cycle ergometer exercise to exhaustion with N95 FFR, surgical mask (SM), and no mask	PETCO2 ↑ with FFR but not SM (e.g., at peak exercise, ↑ from 35 with no mask to 43 mmHg with FFR). No difference in HR, f, BP, Hb%Sat or time to exhaustion with N95 versus no mask.	While well tolerated by healthy subjects, the ↑ PETCO2 might be problematic for impaired persons. Accuracy of measurement with nasal prongs inside FFR should be validated since any entrained dead space air will overestimate PETCO2].
Fikenzer (2020) 14	x	x	x	x	12 healthy adults exercised to exhaustion on cycle ergometer with tight‐fitting anesthesia‐type mask over nose and mouth and FFP2, surgical mask, or no mask. Spirometry was performed through the anesthesia‐type mask. Cardiac output and stroke volume were estimated by impedance cardiography.	FEV1, F, VT, maximum ventilation, maximum power, and peak blood lactate were reduced, discomfort was increased and there was no change in HR, BP or cardiac output with FFP2.	The use of an anesthesia‐type mask ↑ the subjects' dead space ventilation by an unknown amount.].
Fischer (2021) 15			x		Transcranial hybrid near‐infrared spectroscopy measured cerebral blood flow (CBF), cerebral oxygen saturation (StO2), and several other parameters in 13 young volunteers with brief FFP3 use.	Small ↑ in CBF and StO2 were noted.	While authors note the effect is small and of uncertain significance, they suggest the methodology may be applied in larger studies. [PaCO2 is a determinant of cerebral blood flow].
Fischer (2022) 16	x		x		20 patients with cardiovascular disorders (average age = 58 years) and 10 healthy controls did progressive exercise in laboratory with FFP2 or no mask. Measures included electrocardiography, earlobe capillary blood analysis (PCO2, PO2), peak power output.	The FFP2 did not affect BP, HR, or EKG. A ↓ peak power output and ↑ pCO2 (33 to 36 mmHg) immediately post exercise was observed with FFP2.	Hemodynamic effects in cardiovascular disease patients were minimal. The small ↑ in PCO2 is unlikely to be clinically significant. Illustrates clinically focused study of a patient population
Han (2021) 17			x		39 healthy students were exposed to Traffic‐Related Air Pollution (TRAP) for two hours with N95, PAPR with filters, or shams. Heart rate variability (HRV) with multiple individual components, HR, and blood pressure were ascertained during and several hours after exposure.	HRV was generally unaffected by respirator use. However, one component (VLF) had significant ↑ with N95.	Authors suggest N95 did not protect against cardiovascular effects of TRAP. They suggest that anxiety from N95 use may have caused the single significant Δ of HRV.
Harber (2010) 18	x		x	x	Subjects with mild COPD, asthma, and rhinitis were studied with N95 and a half mask elastomeric respirator during a series of simulated work tasks. No periods without a respirator were included.	Comparison of the two respirator types demonstrated differences in respiratory timing, flow rates, respiratory pattern, and respiratory rate. Patients with mild respiratory disorders generally tolerated the respirator use well. There were significant interactions between respirator type and disease category, particularly for persons with asthma.	Effects of the FFR per se cannot be identified since there was no no‐mask control. However, authors suggest the intermittent nature of asthma limits compensation to respirator use and, therefore persons with asthma warrant attention.
He (2014) 19	x			x	With a mannikin (nonhuman) simulated breathing machine, they measured total inward leakage and filter penetration with several combinations of mean inspiratory flow rate and respiratory frequency.	Breathing pattern affected the protective effectiveness of the FFR by varying the facial seal leakage and filter penetration. Higher F had ↑ inward leakage, perhaps by creating a negative pressure pulling the mask against the face.	This study provides another impetus for looking more carefully at the respiratory pattern rather than total ventilation per se. However, machine simulation with sinewave pattern breathing may not fully emulate humans.
Hopkins (2021) 7	x	x	x	x	Detailed discussion of pulmonary and cardiac physiology largely based upon studies not using respirators.	Based on physiologic principles and experimental studies with surrogate loads, they conclude that FFR effects are likely to be minimal	Good background reading for physiologic principles
Kim (2013) 20	x	x	x		Compared four N95 models with/without expiratory valve in 20 subjects in lab.	Dead space was 280‐390 ml. Resistance (pressure) was 8.6–12.2. Wearing N95 ↑ ed HR by 6‐11/min. Small mean ↑ PtcCO2 noted, but 12 of 20 had potentially significant ↑. No effect of expiratory valve, Hb%Sat not affected.	Expiratory valve had no benefit. Study raises concern that despite minimal average Δs, some subjects may develop CO2 retention. [Accuracy of this technique to reflect actual PaCO2 is uncertain. Measuring mask physical “dead space” is useful, but may not accurately reflect functional dead space].
Kim (2015) 21			x		16 pregnant women +16 controls were studied over 1 h of sedentary or moderate exercise. Investigators indirectly estimated cardiac output, stroke volume, and total peripheral resistance based on Δs in one finger. Fetal heart rates were also monitored.	FFR caused ↑ diastolic and mean arterial pressure ↑ (about 8 mmHg) but did not affect systolic pressure, estimated stroke volume, or fetal heart rate. Effects were similar in pregnant and nonpregnant subjects.	A novel finger clamp system estimated stroke volume, cardiac output, and total peripheral resistance. [This commercial device requires further validation but might be useful in the future.]
Koroteeva (2022) 22	x				Infrared imaging was used to study the distribution and velocity of exhaled air using a CO2 specific absorption band.	N95s reduced CO2 elimination near the face with geometry that depended upon breathing pattern.	Infrared imaging (IR) specific for the CO2 absorption band might extend the utility of IR imaging, which has heretofore imaged exhalation based upon temperature/thermal emissivity].
Kyung (2020) 23	x	x	x		Standardized 6 min walk tests used in 97 subjects with advanced COPD with/without N95. Pre‐post BP, HR, f, Hb%Sat PETCO2.	↑ f, Hb%Sat fell average of 2%, ↑ PETCO2 about 5 mmHg. Statistically significant but very small differences in PETCO2 during 6‐min walk test. Lower FEV1 or higher baseline dyspnea scores Identified subjects who could not tolerate a respirator.	N95s may not be tolerated by persons with advanced lung disease. Authors caution about use of N95 if FEV1 < 30% predicted. Note: Despite N95, average PETCO2 was < normal, suggesting possible hyperventilation.
Lee (2011) 24	x				Study (n = 14) used posterior rhinomanometry to measure nasal airflow resistance during 30 s trial of nasal breathing. FFR was under full facemask	Both inspiratory and expiratory resistance during nasal breathing ↑ approximately 25%.	The study demonstrates ability to measure nasal resistance with an FFR. The ↑ nasal resistance was about ¼ experienced by patients with seasonal allergic rhinitis [More study is needed to understand significance since the nasal breathing trial was only 30 s]
Leung (2022) 25	x				The minimal nasal cross‐sectional area was measured in 50 Asian subjects with acoustic rhinometry without mask, with N95, and after adjustment of the N95 position on the nose.	Significant (27%) ↓ in the minimal cross‐sectional area of the nose as well as concomitant nasal symptoms due to N95 were observed. Repositioning the mask led to significant improvement.	Since nasal resistance is a significant proportion of total respiratory resistance and the minimal cross‐sectional area (“nasal valve”) creates turbulent flow, the findings may partially explain many reported symptoms. [If repositioning the mask could be done without reducing protection, improved tolerance could occur].
Litwinowicz (2022) 2	x	x	x		Systematic review and meta‐analysis of physiologic effects using well‐specified criteria common to clinical trials.	Authors identified 26 studies with combined N = 751. They conclude there are small findings: ↑ HR, ↑ PETCO2, and ↑ PtcCO2 at low exertion. ↑ PO2, ↓ V̇O2, and ↓ V̇E were noted with greater exertion.	Useful application of traditional methods for combining RCT data. Analysis is limited by not considering the quality of physiologic measurements or effects other than mean effects.
Lubrano (2022) 26 , 27	X				22 children (median age 7 years) tested in two groups– N95 with and without exhalation valve, with rest and exercise. HR, F, Hb%Sat, PETCO2, were measured with a commercial system.	↑ F and PETCO2 was noted in both groups after walking, but effect was seen at rest only in the no‐ exhalation valve group. Overall, effects magnitudes were small.	Study suggests that valved respirators might be advantageous, but effects were small, and different brands of N95s were used with/without exhalation valve.
Mapelli (2021) 28	x	x			Study of 12 healthy volunteers using standard cardiopulmonary exercise test (CPET) equipment. The FFR was placed within the tightfitting standard exercise laboratory mask.	V̇E, V̇02, V̇CO2, F ↓ at rest, V̇E, V̇02, F, VT ↓ at peak exercise. Inspiratory time ↑ with FFR.	Study illustrates measurement of respiratory timing parameters. [“Standard” clinical exercise test mask and algorithms may need correction for rebreathing within the CPET mask and Δ the airflow dynamics due to overlying standard exercise test mask.]
Morishita (2019) 29			x		50 healthy subjects with/without N95 in clean room and exposed to traffic related air pollution (TRAP). Extensive cardiovascular measures (blood pressure, aortic hemodynamics, heart rate variability, aortic augmentation pressure, and microvascular endothelial dependent vasodilation).	Use of N95 provided some protection against cardiovascular effects of TRAP (improved aortic hemodynamics).	Excellent demonstration of cardiovascular variables that may be feasibly measured during field studies of respirator use. However, their “N95” was very atypical since it included a fan that would affect CO2 and other factors.
Rebmann (2013) 30	x	x			10 ICU nurses were studied during 12 h work shifts, comparing N95 vs N95 + surgical mask. They measured trends over time in PtcCO2 and Hb%Sat	Over the 12 h shift, average PtcCO2 ↑ from 32.4 to 41.0. No significant Δs were seen with the other variables. Greater body mass index (BMI) was associated with greater effects.	Strengths include measurements during real‐life nursing duties. [The longitudinal analysis design is particularly relevant as N95 use throughout shifts is displacing intermittent brief use. The interaction of FFR and of BMI is notable].
Rhee (2021) 31	x				Investigators used near continuous in‐mask CO2 monitoring for KN95 and valved FFR in 11 volunteers at rest over 15 min.	They found a 2.6% and 2.4% ↑ CO2 in‐mask with the KN95 and the valved FFR respectively.	FFR with/without exhalation valves did not differ. [This illustrates need for clarity about how measurements are summarized. Was in‐mask CO2 simply the average of all 760 CO2 measurements made for each subject in each period (15 min at 1 Hz sample. rate)?.]
Roberge (2010) 32	x	x			Small study (n = 10) of healthy HCPs at 3 exercise levels in lab. Measures included transcutaneous CO2 (PtcCO2); In‐mask CO2 and O2 fractions (FO2, FCO2) and Hb%Sat	Face mask CO2 averaged 3.0%. 2 subjects had peak PtcCO2 > 45 mmHg (48, 61 mmHg). Exhalation valve had no benefit. In‐mask O2% was ↓ (16.5‐17%). Hb%Sat not affected.	↑ in‐mask CO2 may be important with impairments, but this study only looked at a small number of healthy volunteers. Authors note that PtcCO2 may overestimate blood PaCO2.
Roberge (2010) 33	x	x			Compared N95 with and without added surgical mask (SM) in10 HCP in lab. PtcCO2, Hb%Sat, V̇E, in‐mask CO2 were measured.	Minimal physiologic effect of adding surgical mask. 2 subjects had significant ↑ TcPCO2 (48, 61). In‐mask ↑ CO2 (2.9%) and ↓ O2 (16.5‐17%) seen with all N95‐SM combinations. N95s with an exhalation valve had a minimal effect.	Both O2 and CO2 in‐mask were affected adversely. Surgical mask on top of an N95 might allow reuse of N95 without concern about fomites on the outer surface. Study showed that 2 of 10 subjects had significantly ↑ PtcCO2, but this method may be unreliable.
Roberge (2010) 34	x				To assess if moisture retention affected resistance, they used a mannikin with a sine wave pattern breathing machine with heated and fully water saturated air to simulate human tracheal air. They studied 9 FFR models and determined time average pressures, estimated mask surface area, and moisture related weight gain over 4 h.	Inhalation resistance, expressed as average pressure, was low (1.45 cm H2O) and had a small statistically significant ↑ over time (0.04). Exhalation resistance ↑ from a mean of 0.71 to 0.73 cm H2O over time. N95s with an exhalation valve had a non‐statistically significant trend to attenuated ↑ in resistance over time.	Authors conclude that moisture retention has a negligible effect upon resistance. [However, moisture might have other effects].
Shectman (2022) 35	x	x	x		Study of 41 HCP in Emergency Department for 4 h shifts, evaluating venous blood gas tensions, end tidal C02, heart rate and oxygen saturation (Hb%Sat). Measurements pre/post workshift with masks with subjects as own controls, but without mask‐free controls	Venous 02 content ↓ about 5 mmHg, PETC02 ↑ 6 mmHg, no significant Δ in venous pH, venous C02, heart rate, Hb%Sat.	↓ venous oxygen tension may indicate ↑ oxygen extraction peripherally. Study design does not distinguish between effects of workshift and of masks. Authors concluded “the clinical significance is unknown and needs to be determined.”
Shaw (2021) 3	x	x	x	x	Meta‐analysis of studies of impact of masks (including N95) during exercise	No effect of N95 on exercise performance; ↑ respiratory perceived exertion, dyspnea, end tidal C02, heart rate, and respiratory rate.	Meta analysis shows small but subclinical effects which do not ↑ risk or diminish performance in healthy adults.
Shui, (2022) 36	x	x	x	x	12 healthy HCP studied at rest and with maximal cycle ergometer exercise with KN95 or no mask. KN95 was placed under tight‐fitting cardio‐pulmonary exercise testing nose‐mouth mask	KN95 produced ↓ breathing frequency, flow, minute ventilation, and inspiratory time. Oxygen uptake and oxygen pulse, were ↓ at rest, warm up, and maximal exercise. Dyspnea ↑, maximal exercise and total exercise time ↓ with KN95. HR was unaffected. BP was not tested.	KN95 affected ventilation and maximal exercise tolerance associated with ↑ dyspnea. [Dead space of the cardio‐pulmonary exercise testing mask and tubing were not measured].
Tong (2015) 37	x	x	x		Studied 28 pregnant nurses in two steps: (a). Measured oxygen consumption (V̇O2) during actual nursing work. (b) Simulated N95 use by covering inlet/outlet of standard clinical exercise laboratory mask with N95 filter medium. Measured multiple physiologic variables. Measured fetal heart rate, cardiotachography for fetal HR pattern).	Significant impacts on ventilation were noted (↓ V̇E, V̇O2, V̇CO2, VT) and ↑ PETCO2 without Δ in in‐mask FCO2. No effects on fetal measures were seen.	Unique strengths include measurement of (V̇O2) during actual nursing work and fetal monitoring. [Study may overestimate effect of FFR's; since the cross‐sectional area of actual N95s is much greater than the small area of the ports on the exercise test mask. Illustrates need to adapt physiologic measures to FFR research.]
Wojtasz, (2022) 38		x	x		23 young healthy ICU workers served as own controls working 3 h shifts with/without FFP3 plus other PPE (coveralls and goggles). Control shifts used FFP1 and lab coats.	Oxygen saturation (Hb%Sat) by pulse oximeter ↓ over workshift by mean 1.43% (maximum ↓ 2.29%). Heart rate did not differ between conditions.	Combination of FFP3, goggles and coveralls associated with slightly lower O2 saturation within normal range over 3 h ICU shift
Xu (2015) 39	x				Mask dead space volumes of 6 FFRs attached to 4 different mannikin head forms were determined by surface scanning/mathematical calculation and by filling with water.	Dead space volumes ranged from 107–167 ml by calculation and from 98–166 ml by water filling measurement. In general, the largest manikin headform had the greatest dead space for each of the FFR models tested.	Results demonstrate that dead space added by the FFR is similar in magnitude to the anatomical dead space. [Perhaps future combining this approach and computational fluid dynamics flow approaches may determine functionally effective mask dead space and CO2 rebreathing].
Yalciner (2021) 40	x	x			15 HCP had venous blood gas tensions and symptom questionnaire assessed before and after a 4 h workshift while wearing FFP3 mask.	There was no statistically significant Δ in venous blood gas tensions.	There were no mask‐free pre/post control sessions. [Venous blood gas measures only roughly approximate arterial O2 and CO2].
Yildiz, (2021) 41			x		In this ophthalmology study, 20 HCP wore FFP3 for 4 h during usual clinical duties. Pre/post measurements made; no mask‐free controls.	Retinal temporal vein diameters were ↑ , and choroidal vascularity ↑ immediately after shift. Mean arterial pressure and mean ocular perfusion pressure were ↓ . Hb%Sat, pulse rate, and systolic and diastolic blood pressure were unchanged.	Effects were subtle, and no follow‐up measurements were made. Authors did not speculate on implications but called for more research.

Note: This table summarizes illustrative studies. Particularly relevant physiologic aspects are shown in columns 2–5. Comments include likely implications and relevant points; comments with square brackets[] are this review authors' perspective. Δ: change/difference. ↑: increase. ↓: decrease. Other abbreviations are shown in Table 1.

3.1. Carbon dioxide elimination and ventilation

3.1.1. Potential FFR effects

Ventilation is normally carefully adjusted to maintain homeostatic arterial CO₂ (PaCO₂₎ levels. FFR's have dual effects. The mask increases dead space due to rebreathing exhaled CO₂, thus increasing the necessary ventilation. The filter airflow resistance increases the work of breathing (W). While the additional CO₂ production from the additional respiratory muscle effort is small as a proportion of total metabolic CO₂ production, maximal exercise may increase metabolic CO₂ production ten‐fold times greater than that at rest. Effects on the respiratory pattern are discussed in a later section.

3.1.2. Measurement

The impact of FFRs may be assessed by directly measuring ventilation, by characterizing the mask (physical dead space and airflow resistance), measuring the added CO₂ load from rebreathing, or change in PaCO₂. Several measures of ventilatory efficiency are discussed in the online supplemental material; these include the dead‐space‐to‐tidal‐volume ratio (Vd/Vt) and ventilatory equivalents for O₂ and CO₂ (see Table 1).

FFRs create unique challenges for measuring ventilation. Most laboratory methods use an exercise physiology mask or mouthpiece connected to airflow sensors such as pneumotachographs, turbines, or anemometers. These methods cannot be used with FFRs, since air flows through the entire FFR surface. Studies with simulated or field work are subject to greater motion artifact than in the clinical exercise laboratory. Since gas volumes increase with temperature, artifacts can be created as air temperature varies by time and mask location as the warm exhaled air cools passing through the mask.

Respiratory inductive plethysmography is particularly useful since it does not require a mask or mouthpiece. ¹⁸ This is often used in clinical sleep laboratories and has been applied to respirator studies in both laboratory ⁴² , ⁴³ and field settings. ¹⁰ , ⁴⁴ Ventilation is estimated from changes in the circumference of the chest and abdomen. Breathing frequency (F) and timing are often more precisely measured than volumes per se. ¹⁰ , ⁴⁵

To permit use of mouthpieces or standard physiology masks, several studies add surrogate resistance or dead space to traditional exercise masks to simulate FFRs. ⁴⁶ However, simply taping filter material across the ports of standard laboratory test masks ⁸ , ³⁷ creates unrealistically high resistance since the cross‐sectional area of the mouthpiece or mask port is considerably smaller than that of an FFR.

Dead space between the user's face and mask may be measured with mannikins ²⁰ or spatial modeling. Since face size and shape varies, some people fill more of the mask space than others, reducing the apparent dead space. Airflow within the mask is not homogeneous, so some peripheral areas may not contribute to rebreathed CO₂. Respiratory pattern and exhalation valves affect airflow distribution. Infrared imaging describes the spatial inhomogeneity based on temperature differences between ambient air and exhaled air or CO₂ absorption bands. ²² , ⁴⁷

Other approaches measure in‐mask CO₂ concentration; methods should be clearly described and validated. Since CO₂ partial pressure and inhaled fraction (FiCO₂) vary throughout the respiratory cycle, an appropriate summary metric is needed; choices include average FiCO₂, maximal FiCO₂, or total inhaled CO₂ volume (volume‐concentration integral). Time‐weighted averaging is misleading because most of the volume is inhaled in the initial high flowrate portion of inhalation; volume rather than time averaging more accurately indicates the inhaled CO₂ burden. The sampling probe location affects results (e.g., in the central air stream vs. peripheral areas). Placing a collecting device completely over the FFR to avoid in‐mask probes may change the flow distribution and CO₂ measures.

While airflow resistance is implicitly considered a physiologically important factor, pressure rather than resistance is generally used in human and breathing machine simulation studies. Resistance is the pressure drop across the mask divided by the airflow rate. The pressure drop is generally nonlinear and disproportionately increases as the ventilation increases. ⁴⁸

Many governmental programs specify the maximal permissible pressure drop at a single constant flow rate during inhalation and exhalation (e.g., inspiratory pressure <35 mm H₂O at 85 L/min). ⁴⁹ However, this does not adequately reflect the actual resistance. Other metrics include peak pressure, average pressure over time, or the pressure‐volume integral (work). Several systems utilize machine breathing with sine wave patterns or even a servo‐controlled flow generator to realistically simulate human breathing patterns.

Measurement of the concentration of CO₂ in the arterial blood (P_aCO₂) is rarely used for respirator assessments because it involves puncturing the radial or other arteries. Capnography measures CO₂ in the exhaled air. The exhaled CO₂ pressure at the end of an exhalation, end‐tidal CO₂ (P_ETCO₂), estimates the PaCO₂ since it comes from the alveoli where there is minimal gradient between alveolar and blood pCO₂. Accurate probe placement is essential to prevent sampling mask dead space air, and pulmonary disorders causing airflow inhomogeneity produce admixture of alveolar with dead space air. Validation comparing P_aCO₂ and P_ETCO₂ has not been performed for respirators. The expired CO₂‐time curve as well as the P_ETCO₂ should be recorded; a late expiratory plateau suggests accuracy. The cumulative product of exhaled CO₂ percentage times volume measures CO₂ output $(\mathrm{i}.\mathrm{e},\int {\text{FeCO}}_2\mathit{dV})$.

Transcutaneous carbon dioxide (PtcCO₂) measures CO₂ that has diffused through a heated area of skin. ²¹ Since PtcCO₂ may overestimate PaCO₂, ³² , ⁵⁰ its main utility may be to track trends rather than determine actual PaCO₂.

3.1.3. Research summary and implications

Multiple studies report increases in CO₂ levels within the mask and small increases in the users' CO₂ levels. For example, CO₂ concentration in‐mask rose as high as 3% at moderate exercise ³² and 8% at maximal exhausting exercise. ⁵¹ Most studies suggest that while reductions of ventilation (V̇_E) and increments in PaCO₂ do occur, they have small magnitude and are unlikely to have serious adverse effects in most normal users.

Some studies report significant increases in PtcCO₂ among some normal subjects (e.g., 12 of 20 ²⁰ ) (although the PtcCO₂ methodology is variable). While FFRs worn by normal individuals typically have insignificant effects, persons with lung disease or respiratory control disorders may be adversely affected. Pregnancy per se did not modify the response effect patterns. ²¹ , ⁵² Subjects with advanced chronic obstructive pulmonary disease (COPD) ²³ had increased P_ETCO₂, and persons with cardiovascular disease had slightly increased capillary CO_2. ¹⁶ Milder COPD, asthma, and rhinitis patients generally tolerated respirators well. Responses differed by disease state. ¹⁸

Further studies with persons with cardiopulmonary disease and with heavy exertion are needed. Concurrently, measuring device characteristics such as mask dead space and resistance and physiologic consequences should guide FFR design to minimize effects upon CO₂ homeostasis. Computational fluid dynamics and other mathematical models may also help optimize design.

3.2. Oxygenation

Oxygen uptake is determined by the fraction of O₂ in the inspired air (FiO₂) and lung function. Delivery to tissues depends upon the cardiovascular system (see Section 3.3).

3.2.1. Potential FFR effects

Since the sum of all gas partial pressures is constant (ambient atmospheric pressure), the fraction of O₂ in the mask (FiO₂) is reduced if CO₂ pressure increases. Increased oxygen consumption by respiratory muscles to overcome resistance probably has limited significance in view of the small load imposed by FFRs.

3.2.2. Measurement

As for CO₂, effects may be assessed by measuring O₂ fraction (FiO₂) within the mask, oxygen uptake (V̇O₂), or by assessing oxygen levels in the user. Measurement of FiO₂ or partial pressure within the mask is subject to the same pitfalls as described for CO₂. Since mask FiO₂ changes throughout a breath cycle and measures are sensitive to the placement of the sampling probe, the lowest of time‐averaged levels may give an impression of falsely low oxygen availability.

Oxygen uptake (V̇O₂) is frequently measured in cardiopulmonary exercise laboratories by subtracting exhaled O₂ volume from the inhaled volume (product of ambient FiO₂ × minute ventilation). These methods may be misleading for FFRs since the in‐mask FiO₂ is not constant and because capture of exhaled gas is challenging without an exercise mask or a mouthpiece.

Tissue oxygenation measurement with transcutaneous O₂ sensors has been used but has wide variability ⁵³ ; the pulse oximeter is more convenient.

Blood oxygenation is most accurately assessed by measuring the arterial oxygen tension (P_aO₂), but this is uncommon in studies because it requires serial arterial punctures or an indwelling arterial catheter. Since PaO₂ is closely linked to hemoglobin saturation (HbSat%), many studies monitor this with a device on the finger or ear (often called a pulse oximeter).

3.2.3. Research summary and implications

Studies show reduction of in‐mask oxygen concentration, typically of about 3%. These reductions have generally been unaccompanied by reductions of blood hemoglobin oxygen saturation percent (Hb%Sat). A single study showed O₂ percentage as low as 16.5% in the mask ⁴⁷ without effect upon the Hb%Sat; the authors note that the US Occupational Safety and Health Administration (OSHA) Immediately Dangerous to Life and Health (IDLH) limit of 19.5% oxygen ³² is based upon workplace rather than in‐mask air. ⁵⁴ The average mixed in‐mask concentration may not reflect the actual inhaled concentration. ⁷

Healthy persons appear to physiologically compensate for FFR effects. For example, while small decreases of in‐mask FiO₂ and increases of in‐mask PCO₂ were observed in a laboratory study of healthcare workers, neither hemoglobin saturation (Hb%Sat) nor transcutaneous carbon dioxide (PtcCO₂) was significantly affected. ³² N95 use did not affect Hb%Sat in 20 healthy adults undergoing mild‐moderate exercise for 1 h. ²⁰ However, considerable variability among normal users has been reported. ⁹

In addition to healthy volunteers, studies included persons with pregnancy or lung disease. Pregnant subjects did not have reductions in Hb%Sat, nor did they differ from nonpregnant women. ⁵² Patients with advanced COPD (average FEV₁ = 57% predicted) performed 6 min walk tests with and without N95s; because small but statistically significant effects on Hb%Sat were seen, the authors recommend caution in using respirators in those with with severe COPD. ²³

In summary, FFRs slightly reduce inspired oxygen content. Healthy and pregnant adults compensate to maintain normal blood oxygen saturation. In patients with severe chronic lung disease, the ability to tolerate these effects should be ascertained.

3.3. Cardiovascular effects

3.3.1. Potential FFR effects

The cardiovascular system includes the heart, pulmonary circulation, and systemic circulation. Resistance breathing may increase pleural pressure swings with inspiration and expiration. However, the magnitude of such pressure changes is far below those due to mechanical ventilation in ICUs. Increased intrathoracic pressure during expiration may reduce venous blood return and cardiac output. Changes to the pulmonary circulation such as hypoxia‐induced vasoconstriction, while hypothetically possible with FFRs, seem unlikely in the absence of significant disease. Effects on blood oxygen (PaO₂) or CO₂ (PaCO₂) might affect cardiac function or vascular resistance.

Respirator resistance might also affect autonomic nervous system control (e.g., adjustment during postural changes ²¹ or heart rate variability ¹⁷ ). Finally, effects may occur directly on peripheral vasculature resistance or capacitance. Anxiety from FFR use ⁵⁵ may affect cardiac rhythm (increased heart rate) or increase blood pressure.

3.3.2. Measurement techniques

Maximal oxygen consumption (V̇O_2‐max) reflects cardiovascular status in normal persons since cardiovascular rather than pulmonary function limits peak exercise. Numerous studies measure blood pressure, electrocardiogram (EKG), heart rate, and rhythm. Echocardiography may be applied to estimate stroke volume, ejection fraction, and pulmonary artery pressure with exercise.

Measuring total oxygen delivery typically requires measuring cardiac output and is not used for FFR studies. Transcutaneous O₂ sensors (PtcO₂) to estimate tissue oxygenation have had more limited use than PtcCO₂ measurement. Tissue oxygenation measurement is potentially imprecise or inaccurate. ⁵³ A few studies estimate peripheral blood flow and resistance by changes in pulsatile extremity volume. ²¹ Heart rate variability (HRV) indirectly reflects autonomic effects upon the cardiac rhythm. ⁹ , ¹⁷ , ²⁹ , ⁵⁶

Other clinical and research methods are invasive or expensive and unlikely to be used to study FFRs; these include cardiac catheterization, dynamic cardiac imaging with magnetic resonance imaging, or positron emission tomography).

3.3.3. Research summary and implications

Laboratory studies of healthy normal subjects during prolonged exercise have shown small increases in blood pressure with respirators at higher levels of exercise with no electrocardiogram changes. ¹⁰ , ²⁰ , ³² In an incremental exercise test to exhaustion, the maximal oxygen consumption (V̇O₂ max) and ventilation were reduced using the N95, but heart rate, blood pressure, and estimated cardiac output were unchanged. ¹⁴

Responses in pregnant women (increase in mean arterial pressure) were similar to those in nonpregnant women. ²¹ Subjects with advanced COPD had small increases in heart rate and diastolic blood pressure. ²³ Studies of whether an FFR attenuates the effect of air pollution upon persons with cardiovascular disorders had mixed results. ¹⁷ , ²⁹

In summary, cardiovascular effects of FFR use by healthy persons appear to have limited significance. However, studies of N95 use by persons with coronary heart disease and congestive heart failure are needed. Applying newer investigative methods may provide useful insights.

3.4. Respiratory work, control, pattern, and sensation: Areas needing research

3.4.1. Potential FFR effects

Respiratory features other than traditional ventilation measures may explain the paradox that FFRs have only limited observed physiological effects but are often reported to be inadequately tolerated. Sensation of muscle work, pressures, and lung/chest wall stretch may affect respirator tolerability. Work of breathing (W) is increased by both the resistance of the filter medium and increased ventilation need from rebreathing CO₂. These also increase inspiratory and expiratory pressure swings and may affect sensation. Pneumothorax due to pressure swings has been raised as a hypothetical possibility but has no supporting data ⁵⁷ despite possible effects on generated pressures. ⁵⁸ Exhalation valves in FFRs constitute a “threshold resistor” since they do not open until sufficient in‐mask pressure is achieved.

Respirators also affect respiratory timing: Inspiratory resistance increases inspiratory time (Ti) and compresses expiratory time (Te) if breathing frequency (F) remains constant. ⁵⁹ Expiratory time compression may be particularly important for persons with COPD or asthma.

FFRs may change absolute lung volumes over which normal ventilation occurs (e.g., the same V_T of 0.5 l may be inhaled with change in absolute volumes from 1.5 to 2.0 or from 3.5 to 4.0 L. Since work of breathing increases at higher volumes, ⁶⁰ this may impact acceptable device tolerance. While not studied with FFR resistances, volume shifts have been well demonstrated when internal resistance is increased by health conditions.

Nasal breathing conditions the temperature and humidity of inhaled air but also increases total respiratory resistance in comparison to oral breathing. FFRs may change the oral‐nasal breathing partition.

Adaptation to respirator stimuli, both physiologic and subjective, may occur over short times (e.g., work shift) or with frequent long‐term use. Therefore, studies of experienced volunteers may underestimate effects, whereas brief laboratory experimental periods may overestimate effects.

3.4.2. Measurement

Measurement of volumes, timing, and pressures were described earlier. Forces stretching the lung and its recoil pressure are not measured directly in respirator studies since determining transpulmonary pressure (Ptp) requires swallowing a catheter to the intrathoracic esophagus. However, external work to overcome the added resistance may be determined as the integral of volume and mouthpiece or mask pressure. Lung compliance (inverse of stiffness) is the change of volume/change in Ptp. Pressures approximately equalize throughout the respiratory system when there is no airflow such as at the end of inhalation. Therefore, mouth pressure during an inspiratory pause with muscle relaxation may be applicable to respirator research.

Whole‐body plethysmography is used in clinical settings to estimate airway resistance and lung volumes but is not easily applicable to respirator studies. The forced oscillometry technique (FOT) is increasingly used in pulmonary medicine to measure pulmonary mechanical properties. FOT is noninvasive and requires no special effort of the subject. FOT superimposes soundwaves at various frequencies on the inspiratory airstream, measuring the instantaneous pressure or flow response. ⁶¹ Using somewhat complex algorithms, FOT determines both airflow and reactance (the ability of the lung to store energy required for passive expiration). The elastance component of the latter is related to compliance. “Off‐the‐shelf” commercially available FOT devices will require modification to be applicable to unique aspects of FFRs, such as not using mouthpieces. These difficulties may be overcome either by using respirator surrogate resistances and dead space loads or by imposing the oscillatory signal on the chest wall rather than upon the airstream. While not previously used in FFR research, FOT holds great promise.

Absolute lung volumes are measured in clinical medicine by whole‐body plethysmography, dilution of a nonabsorbable gas such as helium in a closed circuit, or nitrogen washout methods. The latter two methods or respiratory inductive plethysmography may be adapted to FFR studies.

Nasal resistance, flow, and nasal‐oral partition have been described with rhinomanometry, simulation, and monitoring perinasal temperature. ²⁵ , ⁶² , ⁶³

Sensation of respiratory loads per se is not synonymous with subjective reports of dyspnea and may be assessed by explicit visual analog scales or other well‐defined rating methods. ⁶⁴ Standard clinical tests of CO₂ or O₂ responsiveness among individuals may be measured to identify persons with respiratory control disorders who are at risk.

Respirators change the respiratory timing pattern as anticipated. While the external work added by the respirator and mask/mouth pressures have been measured, the relationship between these measures and subjective tolerance has not been effectively evaluated. Nasal flow resistance is considerably increased. ²⁴ , ²⁵ While psychophysical respiratory load sensitivity differs among normal subjects, studies have not proven the relationship to respiratory pattern or subjective tolerance in FFR users. ⁶⁴ , ⁶⁵

Studies of the variables discussed in this section may complement the “routine” measures (e.g., V_E, V_T, CO₂, and O₂). Respiratory drive and respiratory control may be particularly important as FFRs are increasingly used by individuals with significant pulmonary and cardiac disorders.

4. DISCUSSION

Respiratory and cardiovascular effects of N95 and other FFRs have been extensively studied in healthy adults including pregnant women. The reported effects generally have small magnitude and are likely to be well‐tolerated by normal persons. CO₂ elimination seems most affected, leading to small increases in arterial CO₂ pressure (PaCO₂). Our review leads to several recommendations for future research endeavors (see Table 3).

Table 3.

Recommendations

Aspect	Current studies	Possible improvements
Study participants	Nearly all subjects are healthy volunteers	Include persons with pulmonary and cardiac disorders and/or obesity
		Do not exclude persons reporting poor tolerance
Study conditions	Short experimental periods in research laboratories	Long‐duration use; more field studies (supported by wearable technology)
Measurement systems	Inaccurate use of standard exercise test equipment	Modify and validate systems for use with FFRs.
	Limited variables measured with limited techniques	Include additional methods
		Integrate with subjective, work performance, and other measures.
	Incomplete documentation	Describe methods fully, preferably with validation
Statistical design	Focus upon comparing difference of averages among conditions	Consider variability as well as central tendency. Identify at‐risk outliers.
	Small sample sizes	Larger studies with standardization of methods to allow meaningful meta‐analysis.
	Each variable assessed independently	Examine patterns of response of multiple variables
Metrics	Most parameters collapsed to a single value	Consider most appropriate metric (e.g., total, volume average, peak).
Design certification criteria	Usually pressure at a fixed flowrate with machine test	Assess best measure(s) to reflect tolerance. Add human testing.
Multiple N95, FFP2/3, KN95 devices used	Cannot assume results for one model are generalizable to all	Study multiple well‐characterized available standard models.

A wider range of subjects should be studied. Persons who report poor tolerance or anxiety should not be automatically excluded in studies. More studies of users with heart and lung disease are needed. Obesity may also affect tolerance. ³⁰ Studies in children are urgently needed because N95s are often recommended for wildland fire smoke or community infectious agent exposure.

While many laboratory studies commonly employ short‐duration use, full‐shift use is increasingly common, and effects may increase or decrease during prolonged use. Wearable technology can facilitate extending physiologic to realistic field studies.

To avoid errors, clinical and research cardiopulmonary exercise physiology test (CPET) techniques must be modified since connection to standard measuring devices is not feasible. Inhomogeneity of flow in the mask and surface requires careful placement of sampling probes. Commercial CPET software that integrate CO₂ or O₂ fractions with increments in volume will introduce errors if the placement of the sampling site is imprecise or if calculations assume inspired air has no CO₂.

The “standard physiologic” measures (CO₂, oxygenation, and ventilation) incompletely describe FFR effects. Well‐conducted studies of subjective tolerance, thermal effects, respiratory sensation, and psychological studies are needed. (A separate article will address the current understanding of these effects).

Additional measurement techniques should be validated and applied. These include respiratory inductive plethysmography, capnography (e.g., P_ETCO₂), transcutaneous CO₂ diffusion, and assessment of peripheral resistance by finger plethysmography. Expanding the range of methods may be productive. Exercise electrocardiography and forced oscillometry appear both feasible and applicable. Standardization and full explication in publications are needed for meaningful data, and they should be validated for use in respirator studies.

Several studies found a small proportion of normal subjects had atypical responses. While most analyses compare the mean values of subjects (e.g., by using analysis of variance, ANOVA), greater attention must be paid to describing the full distribution of responses. This requires considerably larger sample sizes or standardization of methods to allow combining data from multiple studies. Most studies statistically analyzed each variable independently, but combining variables into summary indices may show useful patterns of response.

Physiologic effects are often collapsed to a single metric (e.g., V_T, pressure at fixed airflow rate). Alternatives such as peak flow/pressure, volume‐averaged CO₂, total work) should be investigated to determine which most accurately reflects tolerability. Rather than basing respirator design certification upon maximal pressure during arbitrary and nonphysiological breathing simulation, the most critical combination of parameters should be delineated to assure that FFRs are well tolerated.

FFR's differ significantly in design (e.g., cup, duckbill, flat fold, and with/without exhalation valve), and these differences are likely to produce different effects among these designs and among different manufacturers' models within one of these designs (e.g., three designs had mask static dead space measured by water displacement ranging from 210 to 375 ml ²¹ and pressure drops of 8.6–12.2 mm H₂O at flow of 85 L/min. ²⁰ Functional dead space variation may be greater). This heterogeneity might be addressed by agreeing upon several standard models for testing. Individual products may then be compared to the reference standard.

In summary, the COVID‐19 pandemic has led to proliferation of FFR studies with diverse authors' conclusions about adverse impacts on traditional pulmonary and cardiac measures. Many have small nonrepresentative populations or possible measurement errors. A systematic “top‐down” planning approach is urgently needed, systematically identifying key questions and research priorities.

AUTHOR CONTRIBUTIONS

Both authors contributed to each of the following: conceptualization of the work; the acquisition, analysis, and interpretation of data; drafting the manuscript work. Both approve the version to be published and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

CONFLICTS OF INTEREST

The authors declare that there are no conflicts of interest.

DISCLOSURE BY AJIM EDITOR OF RECORD

John Meyer declares that he has no conflict of interest in the review and publication decision regarding this article.

Supporting information

Supporting information.

Harber P, Beckett WS. Health effects of filtering facepiece respirators: systematic review of pulmonary and cardiovascular effects. Am J Ind Med. 2023;1‐18. 10.1002/ajim.23450

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available in PubMed at https://pubmed.ncbi.nlm.nih.gov/.