Study design

The present study simulates the breathing pattern model that occurs in different circumstances of the daily life of humans, under resting conditions and with progressive increases in the intensity of physical activity. We have used a crossover design, where each facemask was tested on five consecutive occasions, leaving a rest of five minutes between tests, and recording all environmental data, such as temperature, air humidity, atmospheric pressure, and environmental CO2.

A 3 mm thick polymethylmethacrylate box measuring 320 × 300 × 300 mm (height × length × width) was designed and used for the study. Three 30 mm radious apertures were made in it to attach a volume sensor, from a gas analyzer, in each one of them. Three gas analyzers were used to measure the breathing pattern variables analyzed: the Jäeger Oxycon Mobile® (Jäger, Würzburg, Germany), which was placed in the “air input” just behind the calibration pump (Analyzer 1 see Fig. 1), and measured the “air input” to the system (AIRin); the Jäeger Oxycon Pro® (Jäger, Würzburg, Germany), which was placed at the front “air output” of the facemassk (Analyzer 2), and measured the air filtered by the mask (AIRfil); and finally, the Vyntus CPX (Vyaire, Mettawa, Illinois, USA), at the top of the box, which collected the air that was not filtered (AIRunf) (Analyzer 3). The reliability found in the ventilation measurement of the three analyzers placed in line behind the calibration pump showed an intraclass correlation coefficient of R = 0.999 with p < 0.001, a standard error of 1.09 L/min and a percentage error of 2.1% (see Supplementary data).

Figure 1
figure 1

Design of the protocol and measurement equipment, and detail of the filtered air collection procedure.

The air pressure measurement was performed with a digital pressure manometer MAN-37 (Kowloon, Hong Kong) with a differential pressure gauge (Analyzer 4 see Fig. 1), that enabled the pressure data to be exported to a text file with a sampling frequency of 1 Hz.


We tested nine different facemasks from seven different manufacturers. Facemasks description and available information is listed in Table 1. In general, facemasks were made of two external and very thin layers and between then one to three inner layers. One difference observed is that reusable facemasks are stretchable. Therefore, it can be anticipated that any stretching of the ear loops (either during use or during testing) may affect the properties of the facemask. To avoid this, during the tests, an attempt was made to maximize the degree of adequacy of the facemasks with the surface of the face.

Table 1 Description of the facemasks used in this study.

Each mask was covered with a 0.1 mm thick plastic wrap and high adhesion 3 M double-sided tape, leading to a 30 mm diameter cannula through which all the air filtered by the mask was collected (see Fig. 1). The 30 mm cannula was positioned at the exit hole. In the supplementary material, the results of performing five measurements without changing this wrap versus changing the wrap each time can be observed.

Protocol of test

Before each trial, the adjustment of the facemask and all the measuring elements were checked, and the environmental conditions were noted. The protocol was started by simulating ventilation at rest for 60 s (~ 10 L/min)16. Without pauses between any phase, the next phase started pumping about 30 L/min, named the “warm-up phase”, with a duration of 30 s. In the next three phases (“light, moderate and high intensity exercise”), 60, 80 and 120 L/min of air were pumped, respectively, with a duration of 30 s each. Therefore, the total duration of the protocol was 180 s (see Fig. 1). In total, 45 experiments were performed.

The data from each analyzer [ventilation (L/min), tidal volume (inspired/expired) (L), time (inspiratory/expiratory) (s) and breathing frequency (Hz)] were exported breath by breath and stored in separate files. Pressure data from each embolus of the syringe were exported every second (1 Hz) from the manometer. Of these, absolute values above the median were considered as expiratory values, while negative inspiratory values below the median were considered as expiratory values. Then, both gas analyzer and manometer data were combined into a single database for further analysis.

Real bacterial filtration efficiency

Bacterial filtration efficiencies (BFE) were obtained from the technical specifications of each facemask manufacturer. However, to calculate the Real Bacterial Filtration Efficiency (RBFE) of each facemask, the BFE must be corrected by the real filtered air of each facemask (see Eq. 1).

$$ RBFE = \alpha \cdot BFE $$


where α is the correction coefficient obtained as a linear regression (see Table 4 of "Results").

The amount of air filtered in percentage (Fig. 3) by each facemask was calculated as the fraction of “input air” divided by the “output air” multiplied by 100.

Statistical analysis

All values are expressed as mean ± standard deviation for tables and mean ± standard error of mean (SEM) for figures. A three-way ANOVA for repeated measures (5 × 3 × 9) was performed to analyze the effect of the five exercise phases (rest, warm-up, light exercise, moderate exercise and high exercise), three analyzers (AIRin, AIRfil , AIRunf ), and nine types of facemask (Surgical_1 (MVT), Surgical_2 (Radex), Hygienic_1 (Emotion), Hygienic_2 (Elite), Hygienic_3 (LifeStyle), FFP2_Aura, FFP2_Palens, FFP2_Biofield, FFP3_MC002) in the ventilation, tidal volume (inspired/expired), time (inspiratory/expiratory) and breathing frequency.

A three-way ANOVA for repeated measures (5 × 2 × 2) was performed to analyze the effect of the five exercise phases (rest, warm-up, light exercise, moderate exercise and high exercise), two breathing phases (inspired/expired air) and two analyzers (AIRin, AIRfil ) for tidal volume variable. Similarly, a three-way ANOVA for repeated measures (5 × 2 × 4) was performed to analyze the effect of the five exercise phases (rest, warm-up, light exercise, moderate exercise, and high exercise), two breathing phases (inspired/expired air and four type masks (FFP2, FFP3, Hygienic and surgical) in the pressure variables.

To analyze the effect of changing the mask covering plastic on each occasion, the ventilations of the five attempts of the Emotion facemask without changing the plastic were compared to the five attempts where the plastic was changed on each occasion, using a T-Student for independent samples.

Mauchly’s sphericity test was carried out to evaluate whether the sphericity assumption of the variances was violated, in which case the Huynh–Feldt correction was applied. Bonferroni post-hoc tests were conducted, where significant differences were found in any of the analyzed factors.

Intraclass correlation coefficient was used for estimating the reliability with the ventilation measurement of the three analyzers placed in line behind the calibration pump.

A stepwise linear regression analysis was performed, with the dependent variable being filtered air and the independent variables being all the variables of the proposed respiratory model in all intensities (input air volume, output air volume, respiratory times, pressures, etc.).

Effect size (ES) was estimated by partial eta-squared (ηp2) and considering effects > 0.2 small, > 0.5 medium and > 0.8 large. Data were analyzed using the SPSS Statistic software, version 26.0 for Windows (IBM Corporation; Armonk, New York). The significance level was set at p < 0.05.

Source link