Prediction of infection risk associated with the dynamic behavior of aerosol particles during dental aerosol generating procedures in dental clinics in the context of the COVID-19 pandemic

Background: The health emergency declaration owing to severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) has drawn attention toward nosocomial transmission. The transmission of the disease varies depending on the environmental conditions. Saliva is a recognized SARS-CoV-2 reservoir in infected individuals. Therefore, exposure to �uids during dental procedures leads to a high risk of contagion. Objective: This study aimed to develop an infection risk prediction model for COVID-19 based on an analysis of the dynamic behavior of aerosol particles generated during dental procedures. Materials and methods: The dispersion of aerosol particles during dental aerosol-generating procedures (AGPs) performed on phantoms was evaluated using colored saliva. The gravity-deposited aerosol particles were registered using �lter paper within the perimeter of the phantom head and the settled particles were recorded in standardized photographs. Digital images were processed to analyze the stained area and the drops dimensions. A logistic regression model was built with the variables ventilation, distance from the mouth, instrument used, area of the mouth treated, and location within the perimeter area. Results: The largest percentage of aerosol particles ranged from 1 to 5 µm. The maximum settlement range from the mouth of the phantom head was 320 cm, with a high-risk cut-off distance of 78 cm. Ventilation, distance, instrument used, area of the mouth being treated, and location within the perimeter showed association with the amount of aerosol particles. These variables were used for constructing a scale to determine the risk of exposure to aerosol particles in dentistry within an infection risk prediction model. Conclusion: Contamination by disseminated aerosol particles represents a risk for the dental staff. Thus, it is advisable to improve ventilation and use biosafety measures. The need to implement new clinical and educational strategies was evident. This model is useful for predicting the risk of exposure to COVID-19 in dental practice.


Introduction
COVID-19 is an infectious disease caused by coronavirus (severe acute respiratory syndrome coronavirus 2 [SARS-CoV-2]) and has spread worldwide in recent months [1].As of now, the great majority of countries have already undergone a second wave of coronavirus transmissions.The nosocomial transmission of this infection has been reported more recently [2,3].Evidence indicates that the infection rate among dental health professionals is up to 10%, probably due to strict biosafety measures and a signi cant decrease in dental consultation rates during the pandemic [4].
The mode of transmission of the disease and the extent of nosocomial infection vary depending on the conditions of the environment [5].To et al. [6] recognized saliva as a reservoir of SARS-CoV-2 in infected individuals; therefore, it is recommended to reduce the number of aerosol-generating procedures (AGPs) during the diffusion phase of COVID-19.
Exposure to saliva produced during the majority of AGPs represents a high risk of contagion in dental schools, clinics, and o ces [7][8][9].Taking into account that the mouth is part of the oronasal pharynx, it also houses viruses and bacteria from the respiratory tract.Therefore, AGPs may produce aerosolized saliva and then lead to airborne spreading of infections such as SARS-CoV-2 [10].
According to the guidelines set by the World Health Organization and various other government agencies for the gradual restoration of health services during the mitigation and control phases of the current health emergency, infection in oral health care settings can be transmitted directly through the inhalation of droplets generated while coughing or sneezing, by the exposure of the mucous membrane to infectious droplets, and by indirect transmission through contaminated surfaces [11,12].Transmission of SARS-CoV-2 may occur when droplets (diameter, > 5 µm) that do not remain suspended in the air are deposited on surfaces at a distance of < 2 m from the individual; contact with these surfaces can contaminate the mucosa of the mouth, nose or conjunctiva [13,14].
Airborne transmission occurs through droplet nuclei (diameter, < 5 µm), which contain viral particles.Droplet nuclei can remain in the air for long periods and reach people more than one meter away [15].Airborne transmission of the SARS-CoV-2 virus may be possible under speci c circumstances and in the locations where AGPs are performed (during manual ventilation before intubation, intubation, tracheostomy, and open airway suction) [16,17].The risk in dentistry is mainly related to the use of high-speed handpieces and ultrasound instruments, which generate aerosol particles often mixed with saliva and blood [18,19].
Even though vaccination has already started in the majority of countries, its effectiveness and the duration of the vaccineinduced immunity are to be established throughout the upcoming months.Thus, infection control measures such as personal protective equipment (PPE) remain necessary in order to protect patients and health care providers.In turn, those measures should be adjusted to meet the needs of each setting.Presuming that COVID-19 will remain a pandemic disease for a signi cant amount of time and that, therefore, there are still several waves to come, it is necessary to effectively increase biosafety at oral health care settings in order to face the infection risk properly.Hence, describing the dispersion of aerosol particles, de ning their settlement patterns, and developing a model for predicting the risk of exposure to SARS-CoV-2 based on the dynamic behaviour of aerosol particles generated during the AGPs remain unavoidable.

Location and distribution of experimental units
After receiving approval from the institutional ethics committee (B-CIEFO-074-2020), two dental units were selected from each one of the six dental clinics from the Faculty of Dentistry of the National University of Colombia.Phantom heads (Bader, Nigran, Pontevedra, Spain) were assembled in each dental unit, and an ANA-4 typodont (Bader) containing 28 teeth was installed in each phantom.The phantoms were placed in two distant areas in the clinics to evaluate the in uence of changes in air ow generated within the clinics (Fig. 1).Regarding ventilation, the most convenient location of the dental unit with respect to the windows and to the access of each clinic (Unit 1) was determined using a portable smoke machine (AGPTEK 500W, NY, USA) with a power 2000 CFM output.A range of 3 m and a fog time duration of 25 s was used, to evaluate the generation of an unidirectional air ow without cloud formation or turbulence.

Aerosol-generating dental procedures
Dental procedures were performed using four aerosol-generating instruments: high-speed handpiece, pneumatic scaler, ultrasonic scaler, and triple syringe (Fig. 2).These instruments were used in a total of 48 procedures, sorted as follows: eight procedures per clinic in six clinics, corresponding to two procedures replicated in two different placements within each clinic, and in two different conditions of ventilation -open window and closed window-.
In the four general dentistry and prosthodontic clinics, a high-speed handpiece was used to make full crown preparations on a given anterior tooth (tooth #8) and full crown preparations on a given posterior tooth (tooth #30).An expert operator performed the procedures in a standardized way with regard to the dimensions and the time required for the treatment.The preparation for a full ceramic crown (1.8 mm deep with a 6° convergence) was performed with a medium-grit, at-end tapered diamond bur (Brasseler, Savannah, GA, USA).Palatal reduction on the anterior tooth was performed using the eggshaped diamond instrument (Brasseler).The nal preparation was re ned with matching nishing diamonds (Brasseler).
Adequate and even tooth reduction was checked with a silicone matrix.In the periodontics clinic, two sextant antero-mandibular scaling procedures were performed to compare the aerosol particles generated by the pneumatic scaler and the ultrasonic scaler.Finally, in the orthodontics clinic, two procedures commonly carried out for assembling and disassembling orthodontic appliances were performed, namely, vestibular resin removal in upper and lower anterior teeth using matching nishing diamonds (Brasseler), and etch, rinse and dry in upper and lower anterior teeth using an air-water syringe (Table 1).The operator and the assistant used the recommended PPE comprising a long-sleeved surgical gown with back closure, disposable cap, latex gloves, N-95 mask, eye protector, face shield, and closed shoes.A nontoxic concentrated arti cial dye (Nature's Flavors, Orange, CA, USA) diluted 1:100 in water was loaded into the accessory tank of two dental units in each clinic.Additionally, arti cial saliva was prepared using propylene glycol, USP grade carboxymethylcellulose, and < 5% bivalent cations, which were mixed with arti cial dye until a density of 1.005 g/mL was reached.The density was measured using a density meter (DMA 4500, Anton Paar, Hidalgo, Mexico) calibrated at 15.56°C.The colored saliva was poured into the posterior oral area of each phantom head to simulate natural saliva.
The gravity-deposited aerosol particles were registered using lter paper placed in plastic Petri dishes (size, 100 × 25 mm).Different areas within the perimeter of the phantom head (up to 320 cm away from each phantom's mouth) were covered to obtain 1256 samples collected from settled aerosol particles in the six clinics (Fig. 3).The papers were held in position for up to 30 min after the end of the procedure.The average ambient temperature in the clinical areas was 20°C, and the relative humidity was 70%.A 2:1 air-water pressure ratio was used in each dental unit.The supply pressures for air and water were calibrated at 80 psi and 40 psi, respectively.

Identi cation of aerosol particles
The dispersion of the dye on the lter paper was recorded using standardized photographs of the aerosol particles.The photos were taken with a Nikon D5500 camera (Nikon Corporation, Tokyo, Japan) and standardized at a focal length of 85 mm | 6000 × 4000 dimensions | 1/ 125 s | f/ 32 | ISO 200.To improve the detection of the particles settled on the lter paper, the colorimetric image was converted to a uorometric image using the Adobe Photoshop program (Adobe Inc., San Jose, CA, USA).The digital images were processed to analyze the stained area using the ImageJ program (National Institutes of Health, Bethesda, Maryland, USA) in a binary system (Fig. 4) in order to determine the number and size of the particles.The sizes of the settled aerosol particles were measured in pixels and converted to microns (µm).

Data analysis
The analyses were carried out using the statistical package SAS 9.3 (SAS Institute, Cary, North Carolina, USA).The Microsoft O ce Excel 2010 software (Microsoft Corporation, Redmond, Washington, USA) was used for the construction of the databases.The independent variables -i.e., ventilation, distance from the mouth, instrument used, region in the mouth to be treated, location within the perimeter area, division between units, and placement of the dental unit with regards to the windows-were analyzed using a bivariate analysis in order to determine their association with the "stained area" outcome (p < 0.05).The infection risk prediction model for exposure to aerosol particles included the variables that showed either statistical association or biological plausibility.Normality of the data on the stained area was determined using the Shapiro-Wilk test.The outcome variable "stain area" was dichotomized from the median (50th percentile) after performing a logarithmic transformation, whereas the "distance" variable was dichotomized according to the cut-off points established by the maximum discriminatory capacity with respect to the stained area.The odds ratio (OR) values were obtained from the logistic regression model and used to determine the scores of the variables used for the construction of an infection risk scale for COVID-19.

Spread of aerosol particles
One thousand two hundred and fty-six sites were evaluated within the perimeters of the phantom heads, of which 739 had a greater or lesser degree of aerosol settlement.The sites with the most signi cant aerosol settlement were at a distance of less than 80 cm, which is where the operator, patient, assistant, and dental unit were located.Settlement levels of aerosol particles were highest in front of the patient and reduced with increasing distance from the mouth.Some aerosol particles reached the maximum distance measured, i.e., 320 cm, although few reached accessory areas, such as adjacent units, sinks, auxiliary tables, and walls.An inverse relationship was observed between the distance and the settlement of the aerosol particles (Fig. 5).
The in uence of ventilation on the number of settled aerosol particles was observed, independent of the location of the dental unit (near or far) from the windows (Fig. 6a and 6b).However, an important decrease in the amount of aerosol particles generated during AGPs performed with the windows opened was observed for the dental unit near the ventilation with respect to the unit far from ventilation (Fig. 6a and 6b).Most of the aerosol particles generated by the high-speed handpiece were droplet nuclei (between 1 and 5 µm), especially in the dental unit close to the ventilation (Fig. 6c and 6d).
The stained areas corresponding to aerosol particles generated by the sonic and ultrasonic devices were compared.An important in uence of ventilation on the stained area was evidenced, especially in the case of aerosol particles generated by an ultrasonic scaler (Fig. 7a and 7b).The predominant size of the aerosolized particles remained stable (1 to 5 µm) independent of the instrument (Fig. 7c and 7d).The Shapiro-Wilk normality test demonstrated a non-parametric distribution of the data (p < 0.0001).A comparison with the Wilcoxon signed-rank test of the two procedures showed differences between both instruments; the stained area generated by the pneumatic scaler was greater than that produced by the ultrasonic scaler (Table 2).The areas stained by the aerosol particles on the lter papers located near the assistant were greater than those near the operator, patient, and dental unit when working in the anterior sextant (upper and lower) with a high-speed piece and triple syringe (Fig. 8a).Air ow through open window ventilation decreased the stained area for both instruments (Fig. 8b).The stained area produced by the triple syringe (air/water spray) was smaller than those produced by the high-speed handpiece.The mostly particles were classi ed as droplet nuclei in both instances (Fig. 8c and 8d).

Characterization of aerosol particles produced during AGPs
The aforementioned AGPs generated a percentage of droplet nuclei higher than those of droplets and drops, regardless of the following variables: instrument used, ventilation, distance from the oral cavity, location of the dental unit, and area of the mouth being treated (Table 3).The aerosol particles were classi ed according to the settlement pattern on the surface.Several shape and size patterns were found, depending on the location of the paper from the phantom's mouth, the instrument used, the angle of entry of the drops toward the surface of the paper, and the possibility of the formation of large drops resulting either from the union of several drops before their settlement or from dripping through the gloves or hoses of the dental unit (Fig. 9).

Risk prediction model
For the construction of the prediction model, the assumptions of normality, independence, and homoscedasticity were veri ed.Seven variables were used (Table 4), out of which ve were selected (ventilation, distance, instrument used, area of the mouth being treated, and location within the perimeter) to be included in the model, based on their association with the stained area outcome and the biological plausibility.For the distance variable, according to its median value, the cutoff point was found at 78 cm away from the mouth.Similarly, the median was obtained as the cut-off point to dichotomize the stained area outcome variable (Table 5).Five variables were nally included in the model using the forward stepwise method.Considering the more parsimonious model and with a higher coe cient of determination (R ^ 2 = 0.3368), the ORs were estimated for each of the variables, as shown in Table 6.Based on the OR values for the risk of infection due to exposure to aerosol particles generated during AGPs, different scores were assigned to construct the scale, in order to obtain a total value of 10 points, as shown in Table 7.The discrimination capacity of the model was evaluated using a ROC curve (Fig. 10).Additionally, the t of the model was evaluated using the Hosmer-Lemeshow test, which demonstrated good discrimination capacity and adequate adjustment, as observed in Table 8.The observed and expected values were reported according to the predictive values obtained by the model for each of the scoring levels.The best cut-off point was ve points, which showed adequate sensitivity and speci city (Table 9).The following scale shows the score assigned to each variable included in the model according to its OR value (Table 10).

Recommendations
The recommendations suggested based on the ndings of the current study are shown in Fig. 11.These recommendations complement those that were widely disseminated in previous reports [11,12,[20][21][22][23].

Discussion
The results of the present study provide evidence that aerosol particles generation is an imminent consequence of carrying out dental procedures and constitutes a potential mechanism for the spread of several infections such as that produced by SARS-CoV-2.The spread of aerosol particles during AGPs represents a signi cant risk of exposure, primarily for dental staff.The variables that were associated with a higher risk of exposure in the prediction model were as follows: a distance of less than 78 cm; low ventilation; the use of a high-speed handpiece or pneumatic scalers (in periodontics); the location of the patient, operator, and assistant; and, to a lesser degree, the intervention of the anterior region of the mouth.The operator, assistant and patient resulted consistently stained by the aerosol particles.The majority of the aerosol particles generated during the AGPs presented sizes ranging from 1 to 5 µm.This size has been previously associated with increased severity, morbidity and fatality in infected patients because droplet nuclei can penetrate the respiratory tract to establish infection in the lower airways [24].In the case of SARS-CoV-2 infection, droplet nuclei seem even more risky insofar as the epithelial cells of the lung alveolar surface abound with the angiotensin-converting enzyme II (ACE2), which works as a receptor for the S protein present on the surface of this virus [25].
The association of these variables should be put into the clinical context, considering the reported transmission routes of SARS-CoV-2, either by direct contact or by airborne transmission [11,12].The proximity between the patient -possibly infected-and the dental staff poses a risk of contagion, which is dependent on the adherence to the biosafety standards and the use of the recommended PPE [9,12,15].Aerosol particles of different sizes, mainly < 5 µm (86%), were produced.Some of them may settle due to gravity, whereas some could remain suspended in the air and enter the respiratory tract [26], which favors the spread of SARS-CoV-2 insofar as its dissemination is not produced exclusively via airborne transmission or droplet mechanisms but by both methods simultaneously [15], with an added risk in dentistry due to the high transmissibility of the virus during the asymptomatic period [27].
The permanence of these aerosol particles suspended in the air depends on the environmental conditions [28].The infectious range depends mainly on the time interval between its presence in the atmosphere until its settlement [19].
Factors such as relative humidity, ambient temperature, and air ow have been closely related to the particle size and the time it takes to settle on a surface [28].During sample collection, conditions of 70% relative humidity and a temperature of 20°C could favor the settlement of the aerosol particles.Previous studies have shown that low relative humidity [29] and high ambient temperature [28] are related to a longer residence time of the droplet nuclei and droplets in the air [30].The two environmental conditions mentioned increase the tendency of the drops to pass to the vapor phase, which tends to decrease their size by drying.This results in an increase in the mobility and circulation of the particles in the air [31], thereby increasing the risk of spreading the infection in the clinical area [32].
Poor ventilation demonstrated a high association with a greater stained area.Additionally, previous reports estimated that better ventilation substantially reduces the suspension time of aerosol particles in the air [33].The positive in uence of ventilation will depend on several conditions: rst, on the amount of outdoor air that is available within the indoor space, de ned as the ventilation rate; second, the direction of air ow from clean areas toward contaminated areas; and, nally, the distribution of air, which must cover all spaces while entering and leaving the clinical area [34].These characteristics will depend on the infrastructure and layout of the area [35].Although in this study, the experiments were carried out in six different clinical situations and twelve different dental unit locations, the extrapolation of the results should be done with caution, protecting the staff from exposure to hazardous conditions using engineering control measures, and without disregarding the particular layout of each setting [36].
The mass of the aerosol particles determines several settlement patterns resulting from different sizes and shapes of the aerosol particles deposited on the surfaces [37].Sedimented particles may facilitate the transmission of infection by fomites [10].Thick drops may be formed by splashes produced by the rebound of the pressurized water on some oral structure or by the accumulation of oversized droplets on the operator's gloves or the patient's face and neck, which means that a mixture of aerosol particles with particles that are not aerosol particles [15], which can contain saliva, blood, and microorganisms [10], might have caused some portion of the stained areas.Furthermore, thick droplets may be formed by the phenomenon of coalescence or aggregation [15,37], de ned as a binary process in which two drops of the liquid merge to form a single drop.The factors that directly in uence drop-drop interactions include Brownian motion, viscosity, density, interfacial contact area, diffusivity, surface tension, and concentration gradients; therefore, this interaction depends on the nature of the liquids [38][39][40].
As reported by Guzman [41], the SARS-CoV-2 viral load required to initiate COVID-19 disease is expected to be below 1,000 particles.In theory, taking into account the size of a SARS-CoV-2 viral particle is in the range of 0.006-0.14µm [42], a 1 µm drop could transport around eight viral particles.Hence, any aerosol particle or set of aerosol particles over 120 µm in size may contain su cient ciral load for infection.Of the 1256 samples obtained in the current study, 664 presented stained areas ≥ 120 µm, which makes transmission via generated aerosol particles biologically plausible during a dental procedure.However, other factors, such as the infectious capacity of the virions in the particles [24,43], the inactivation potential of the virus, the saliva-water dilution ratio that varies between 1:20 and 1:100 [15] the chemical composition of the drops, and the viability on different surfaces [44,45], should be taken into account when evaluating the infectious potential of the aerosol particles.
Our ndings are consistent with those of previous studies carried out with different methods [46,47].Our methods were able not only to characterize the risk of exposure when performing the AGPs by using the settlement patterns of the aerosol particles generated during the procedures, but also to recognize potential contamination sources within a dental care setting, and to delimit critical areas for the settlement of aerosol particles.Our ndings are also useful for guiding the implementation of new clinical techniques in dental operatory and new teaching models in dental schools, as well as for evaluating the effectiveness of ventilation and extraction systems and PPE kits.
The present study has two limitations.First, an in vivo model was not used to determine the amount of viable infectious viruses in the aerosol particles, and second, the model used in this study was sensitive and was able to detect only particles that have the capacity by size and weight to settle during the 30-minute period after the completion of the AGPs, which has been reported as the time through which most of the aerosol particles are likely to settle [3,46,48].Therefore, another model will be necessary to determine the amount of viable infectious viruses remaining in aerosol particles, as well as the amount and size of aerosol particles that remain suspended in the environment for a longer period.
Nonetheless, a signi cant contribution was made to the characterization of the size and settlement patterns of the aerosol particles generated by widely used instruments in dental school clinics and healthcare settings and, consequently, to the determination of speci c biosafety measures proven to be effective for protecting both the dental staff and the patient from the infection risk associated with the dynamic behavior of aerosol particles generated during dental AGPs.Even though the estimated infection rate among dental care workers during the rst waves of the pandemic ranges between 1% and 10% [4], resuming activities at dental care settings is likely to cause this amount to rise in case biosafety measures are overlooked.

Conclusions
Under the limitations of this study, assuming that COVID-19 will remain a pandemic disease for some time due to the variable vaccination rates among countries, it can be concluded that dispersion of aerosol particles poses a high risk for the dental staff, and it is advisable to take biosafety measures.The highest percentage of aerosol particles generated during the dental AGPs ranged from 1 to 5 µm in size.The settlement patterns of the aerosol particles varied depending on the area surrounding the phantom head and the instrument used.Depending on the mass of the aerosol particles, they could be classi ed into ultra ne-ne and thick drops, which showed point, splatter, and spot patterns.The variables associated with the stained area outcome were distance, ventilation, instrument used, location in the perimeter area, and area of the mouth to be treated.Aerosol particles were able to reach a distance of 320 cm away from the mouth of the phantom head, with a cut-off point of 78 cm as a risk factor.Despite the presence of adequate barriers, dental staff may be exposed to signi cant dissemination of aerosol particles.Therefore, the risk of contagion during and after dental AGPs necessitates an improvement in biosafety measures in order to protect the dental staff.Thus, our model is useful insofar as it included several variables to develop a prediction scale for estimating the risk of exposure to an airborne virus like SARS-CoV-2 associated with the dynamic behavior of aerosol particles.Furthermore, our data are useful to guide new dental care strategies and teaching models for future scenarios.
Distribution   Image processing to analyze the stained area and dimensions of the settled particles.The stained area was de ned as the sum of the stained area measurement of each particle settled on the lter paper in µm2.
Settlement patterns of particles.Recommendations based on outcomes.The recommendations above complement those widely reported in the literature and those disseminated by health regulators in different countries.

Figure 5 Distribution
Figure 5

Table 2
Comparative analysis between the stained area for the pneumatic scaler and the ultrasonic scaler *Stained area data in µm2

Table 3
Types and sizes of settled aerosol particles * Mean particle settlement percentage throughout the eight procedures performed per clinic, according to the size of the aerosol particles.

Table 4
a Statistical association of independent variables (p < 0.05).

Table 6
Prediction model of the risk of infection due to exposure to aerosol particles generated during AGPs

Table 7
Score of the variables according to the OR

Table 8
Scale validation in terms of discrimination and goodness of t