Indoor Environmental Quality and Effectiveness of Portable Air Cleaners in Reducing Levels of Airborne Particles during Schools’ Reopening in the COVID-19 Pandemic

Abstract

Educational buildings tend to fail in the contagion containment of airborne infectious diseases because of the high number of children, for several hours a day, inside enclosed environments that often have inadequate indoor air quality (IAQ) conditions. This study aimed to assess indoor environmental quality and test the effectiveness of portable air cleaners (PACs) in alleviating airborne particle levels in schools of Central–Southern Spain during the period of reopening after the lockdown due to the COVID-19 outbreak. To accomplish this, three sampling campaigns were organized from September to December 2020 to consistently monitor temperature and relative humidity, carbon dioxide, and particulate matter in nineteen classrooms (seven school buildings). Results showed that although the recommendation of maintaining the windows open throughout the day seemed to be effective in promoting, in general, proper ventilation conditions (based on CO₂ levels). For the colder campaigns, this practice caused notorious thermal comfort impairment. In addition, a great number of the surveyed classrooms presented levels of PM_2.5 and PM₁₀, attributable to outdoor and indoor sources, which exceeded the current WHO guideline values. Moreover, considering the practice of having the windows opened, the installation of 1 unit of PACs per classroom was insufficient to ensure a reduction in particle concentration to safe levels. Importantly, it was also found that children of different ages at different education levels can be exposed to significantly different environmental conditions in their classrooms; thus, the corrective measures to employ in each individual educational setting should reflect the features and needs of the target space/building.

1. Introduction

A growing body of evidence has recognized the “airborne transmission” or “aerosol transmission”—where persistent aerosols emitted by an infected person can stay airborne and be subsequently breathed in by others—as a major pathway of SARS-CoV-2 transmission [1,2,3]. In addition, some studies have shown that the risk of transmission indoors can be orders of magnitude higher than in outdoor environments [4]. In fact, achieving enhanced air quality by improving ventilation conditions may be critical to avoiding the accumulation of exhaled air and, thus, mitigating the spread of SARS-CoV-2 and other similar viruses in enclosed environments. This is particularly relevant for indoor environments with high occupancy and prolonged stays.

In accordance with the World Health Organization (WHO) recommendations [5], during the COVID-19 pandemic, several national and international bodies adopted preventive plans that included the promotion of air renovation (ventilation) to reduce the risk of SARS-CoV-2 transmission in indoor environments, including educational settings for children [6,7]. In Spain, the main recommendation for the prevention of COVID-19 transmission through the ventilation of classrooms’ environment made by authorities was frequent ventilation with fresh outdoor air and keeping openable windows open as long as possible [8].

Monitoring carbon dioxide (CO₂) has been recognized as a useful proxy of the load of the air exhaled by occupants that is accumulating in indoor air, and consequently of the risk of transmission of SARS-CoV-2 and other airborne infectious diseases in shared built environments [9,10,11]. In fact, some recent studies carried out in recent years in different countries have evaluated indoor air quality (IAQ), based on CO₂ concentrations, as an approach to estimate the risk of COVID-19 infection and/or to identify actions for reducing the risk for airborne transmission in buildings. For example, Di Gilio et al. implemented a surveillance activity focused on the real-time monitoring of CO₂ concentrations as a proxy of SARS-CoV-2 transmission risk in several preschool, primary, and secondary classrooms proposing a four-level-risk classification including specific corrective measures for each level (i.e., <700 ppm for low, 700–800 ppm for moderate, 800–1000 ppm for high, and >1000 ppm for very high COVID-19 transmission risk) [12]. Previously, Allen et al. suggested 700 ppm of CO₂ (a minimum of five air changes per hour (ACH) in classrooms for a US school classroom of 5000 ft³ occupied by 15 students) as a reference for achieving safe indoor conditions under typical school settings [13]. Most of the corrective actions derived from the studies conducted during the COVID-19 pandemic in educational settings were focused on identifying the best practices or technologies to ensure adequate ventilation levels. In this regard, Alonso et al. reported the achievement of a reduction in CO₂ concentrations in classrooms in the South of Spain from the pre-pandemic to the pandemic period of 300 and 400 ppm by improving ventilation using hybrid and natural ventilation, respectively [14]. A study conducted in Poland assessed ventilation strategies in nursery buildings and recommended the use of small ventilation units in each room [15]. Additionally, Monge-Barrio et al. analyzed the impact of prevention protocols on CO₂ levels in schools, reporting reductions of up to 1400 ppm [16]. However, the same authors supported the need to consider mechanical ventilation with heating recovery as a supplement to natural ventilation, at least in the coldest periods, to avoid thermal discomfort. In line with this, another study conducted in central Spain found that university classrooms with mechanical ventilation presented better ventilation conditions as well as better thermal comfort conditions than naturally ventilated secondary school classrooms [17].

Children are widely recognized as a vulnerable group to airborne hazards. However, there was a high degree of uncertainty regarding the role of children and adolescents in the transmission and spread of SARS-CoV-2 during the pandemic period [18,19,20]. Eexisting evidence suggests that children transmit COVID-19 at a lower rate to children than to adults [21]. Despite this, household transmission has been identified as the most prominent source of child-to-child and child-to-adult transmission, with household adults being at the highest risk of transmission from an infected child. In addition, it has been recognized that further research is required to understand better how child transmission of COVID-19 has been impacted by the reopening of schools [21].

In this context, during the reopening of schools (September 2020), after a period of 7 months of lockdown due to the COVID-19 pandemic, our group organized a series of studies aiming at comprehensively assessing the effectiveness of the adopted strategies in promoting adequate indoor environmental quality (IEQ) in a sample of pre-, primary, and secondary schools located in Central–Southern Spain. The preliminary results of the first campaign of sampling work conducted from 30 September to 27 October 2020 showed that, according to the CO₂ levels measured, the ventilation conditions were improved compared with the majority of the pre-COVID-19 reports for naturally ventilated European classrooms [22]. However, indoor PM_2.5 and PM₁₀ concentrations exceeded 25 μg/m³ and 50 μg/m³ in 63% and 32% of the studied classrooms, respectively. These results were in line with findings from broader studies conducted in classrooms of several locations in Europe, highlighting those high levels of particulate matter (PM) in the indoor air of schools as a real concern that needs to be properly addressed [23,24]. The most recent WHO Global Air Quality Guidelines, published in September 2021, announced that new more strict daily limit values for exposure to critical airborne pollutants, including PM_2.5 and PM₁₀ (15 and 45 μg/m³, respectively), should be seriously adopted by nations [25]. Air cleaners based on filtration, namely those using HEPA (High-Efficiency Particulate Air) filters, have been presented as an effective technological solution for removing particles, including virus-laden aerosols, from the indoor air [26,27,28,29].

With all the above in mind, our study that aimed to assess the air quality in schools in Central–Southern Spain [22] immediately after the reopening of schools during the COVID-19 pandemic was extended to two additional sampling campaigns (until 22 December 2020), in order to:

(i)

collect a higher amount of CO₂ and PM data in the classrooms surveyed to allow the comprehensive investigation of statistically significant associations determining relevant influencing factors for the levels assessed;

(ii)

study the impact of increased ventilation through open windows in the thermal comfort for colder months;

(iii)

assess the effectiveness of air cleaners with HEPA filters in decreasing the levels of PM in the indoor air of a subset of classrooms that had installed these devices during this study.

2. Materials and Methods

2.1. Study Design and Context

After the COVID-19 lockdown (starting on 13 March 2020) in the region of Castilla-La Mancha, schools reopened in September 2020. At that time, the contingency plan implemented in Spain was based on the WHO recommendations and included the following measures:

(i)

mandatory use of face masks during the entire duration of the lesson, indoors and outdoors (not applicable for preschool children);

(ii)

implementation of rigorous practices for infection control, including frequent hand hygiene and recurrent cleaning of surfaces;

(iii)

maximizing physical distance to maintain at least 1.5 m distance (whenever possible) through an appropriate spatial distribution of the seats within the classrooms;

(iv)

increasing ventilation rates in the rooms by keeping windows and doors open.

A total of seven school buildings were studied in this work. Five out of seven were located in Ciudad Real, a small Spanish city in the center of the Iberian Peninsula (38°59′ N, 03°56′ W) with about 75,000 inhabitants. The other school building was located in Poblete (38°56′ N, 3°58′ 53″ W), a village with 2600 inhabitants located in a rural area (surrounded by farm fields) approximately 8 km South–West of Ciudad Real. The last school was located in a semi-urban area of Miguelturra (39°2′ N, 3°55′ W), 6 km South–East from Ciudad Real. Nineteen different classrooms were surveyed in the seven different schools covering different education levels: six classrooms for preschool children (3–6 years old), seven classrooms for primary education (6–12 years old), and six classrooms used for secondary education (12–18 years old). The criteria considered in selecting the classrooms in each school and the information collected using the study checklist to gather data on the characteristics of the school buildings and of the classrooms surveyed are presented elsewhere [22]. In relation to these previously published preliminary data, this work constitutes a substantial extension of the assessment plan that allowed us to obtain data for investigating the existence of statistically significant associations, as presented in Table S1 (Supplementary Materials). The criteria for selecting schools for this study were based on an affirmative manifestation of intention to install air cleaners in the schools during the upcoming 3 months by the respective school building managers. To ensure that we covered different periods before and after the installation of the air cleaners, a sampling plan covering multiple campaigns was designed. Briefly, the IEQ assessment plan established for this included three sampling campaigns:

• 1st campaign: from 30 September until 27 October 2020;
• 2nd campaign: from 28 October until 24 November 2020;
• 3rd campaign: from 25 November until 22 December 2020.

For each sampling campaign, environmental parameters were continuously monitored during the full operating hours of a school day, under representative conditions of occupancy and use of the classrooms under study. The monitoring work started about 5 to 15 min before the children entered the classroom and lasted until the end of the lectures (or even 1 h after the children left the classroom). Typically, children attended school (preschool and primary school) from 9.00 to 13.00 in September and from 9.00 to 14.00 the rest of the period (October–December). For secondary schools, the teaching hours lasted from 8.30–8.45 until 14.30–14.50 h. For all education levels, there was a 30 min break (period of non-occupancy in the classrooms) in the middle of the teaching period. All monitoring equipment (described in detail in Section 2.2.) was consistently placed at a height representative of the breathing zones of the pupils, also ensuring a distance of at least 1.5 m from walls, doors, or windows and 1 m from occupants [30].

All classrooms were naturally ventilated through the opening of the existing doors and windows. Only one classroom located in a recent educational building was equipped with a mechanical ventilation system able to ensure controlled indoor air exchange rates. Nevertheless, even for this mechanically ventilated classroom, according to the information collected in the field, the teacher preferred to promote natural ventilation. For instance, the practice of opening/closing windows (and the number of windows open) was generally managed by the teachers and was not strictly scheduled. Nevertheless, for the period of this study, most of the changes made in the classrooms were registered by teachers and communicated to the researchers. The heating systems were turned on in November. All classrooms were equipped with radiators (different units depending on the room size) placed under the windows functioning with diesel fuel or natural gas and scheduled from 7.00–8.00 to 11.00 or 13.00 depending on the schools (see

Table 1. Summary characteristics of the studied classrooms.

School	Classroom	Number of Students	Use of Mask	Floor Number	Age of Building (Years)	Area (m2)	Central Heating System			Portable Air Cleaner/Sampling Campaigns
							Type	Number of Radiators	Time on a	1st	2nd	3rd
1	1-a Preschool	22	No	0	25	50	Gas oil	2 under the windows	8.00–13.00	No	Yes-Winix Zero Pro	Yes-Winix Zero Pro
	1-b Primary	22	Yes	1	25	45	Gas oil	1 under the windows	8.00–13.00	No	Yes-Winix Zero Pro	Yes- Winix Zero Pro
	1-c Primary	20	Yes	0	1	50	Gas oil	1 under the windows	8.00–13.00	No	Yes-Winix Zero Pro	Yes- Winix Zero Pro
2	2-a Preschool	25	No	0	40	45	Gas oil	2 under the windows	7.00–11.00	No	No	Yes- Winix Zero
	2-b Primary	26	Yes	1	40	64	Gas oil	1 under the window + 1 near the door	7.00–11.00	No	No	Yes-Winix Zero Pro
	2-c Secondary	28	Yes	2	40	64	Gas oil	1 under the window + 1 near the door	7.00–11.00	No	No	Yes- Winix Zero Pro
	2-d Secondary	23	Yes	2	40	50	Gas oil	3 under the windows	7.00–11.00	No	No	Yes- Winix Zero Pro
3	3-a Preschool	21	No	0	32	58	Gas oil	2 near the doors	7.00–11.00	No	No	No
	3-b Primary	23	Yes	2	32	63	Gas oil	2 near the doors	7.00–11.00	No	No	No
4	4-a Preschool	23	No	1	45	33	Natural gas	2 under the windows	7.30–12.00	No	No	No
	4-b Primary	28	Yes	2	45	38	Natural gas	2 under the windows	7.30–12.00	No	No	No
	4-c Secondary	30	Yes	1	45	51	Natural gas	2 on opposite walls	7.30–12.00	No	No	No
	4-d Secondary	25	Yes	1	45	62	Natural gas	2 under the windows	7.30–12.00	No	No	No
5	5-a Preschool	24	No	0	14	55	Gas oil	3 small radiators under the windows	8.00–13.00	No	No	Yes- Winix Zero Pro
	5-b Primary	24	Yes	1	14	57	Gas oil	2 under the windows	8.00–13.00	No	No	Yes- Winix Zero Pro
6	6-a Preschool	23	No	0	25	49	Gas oil	pipes through the ceiling	7.30–13.00	No	No	No
	6-b Primary	22	Yes	1	25	44	Gas oil	3 under the windows	7.30–13.00	No	No	No
7	7-a Secondary	24	Yes	0	26	57	Gas oil	4 under the windows	7.00–12.00	No	Yes- Winix Zero	Yes- Winix Zero
	7-b Secondary	12	Yes	0	26	57	Gas oil	4 under the windows	7.00–12.00	No	Yes- Winix Zero	Yes- Winix Zero

^a Turned on in November (2nd and 3rd sampling campaign).

).

According to the national recommendations to reduce transmission risk in school, air changes per hour (ACH) of 5 or 6 h⁻¹ should be ensured in a classroom of 100 m² with 25 students aged 5–8 years old [31]. This ACH could be reached through ventilation or with a combination of improved ventilation and air filtration. Schools that effectively have opted to buy portable air cleaners (PACs) during this study typically installed only one unit of PAC per classroom to complement ventilation, especially in the cold season, in order to avoid having windows open for long periods. During the 2nd and 3rd campaigns, some of the schools had PACs with HEPA filters (H13) in classrooms (Table 1).

Considering the period of study, the highest accumulated incidence of COVID-19 cases was registered in the epidemiological week of 2–8 November, with 291.8 cases per 100,000 inhabitants in Spain and 305.2 in Castilla La Mancha [32].

Data collected in this study were analyzed in order to generate evidence to answer the following main research questions:

• What are the IAQ and comfort conditions to which children of different educational levels (pre-, primary, and secondary schools) were exposed during the first 3 months after school reopening due to the COVID-19 pandemic?
• Was the installation of air cleaners with HEPA filters conducted in some surveyed classrooms during this study effective in promoting safe levels of PM?

2.2. Monitoring of Environmental Parameters

Indoor CO₂ concentration, temperature, and relative humidity levels were assessed using a portable data logger (model HD21ABE17, Delta OHM, GHM Group, Baden-Wurttemberg, Germany), presenting the following measurement range and accuracy: 0–5000 ppm, ±50 ppm (+3% of the reading), 1 ppm of resolution for CO₂ (NDIR dual-wavelength sensor); −20 °C to 60 °C, ±0.2 °C for temperature (NTC 10 kΩ sensor); and 0–100%, ±2% for relative humidity (capacitive sensor). PM_2.5 and PM₁₀ mass concentrations were monitored employing a DustTrak DRX aerosol monitor (model 8533, TSI, Inc., Shoreview, MN, USA), able to measure in the concentration range 0.001–150 mg/m³ and an accuracy of ±5%. The smaller size particles were measured by a P-Trak portable condensation particle counter (model 8525, TSI, Inc., Shoreview, MN, USA) that counts airborne particles sizing from 0.02 to 1 μm at a concentration range from 0 to 5 × 10⁵ particles/cm³, with an accuracy of ±5%. Although the measurements conducted by equipment covered quasi-UFP particles lower than 1 μm (that include an important fraction of particles in the ultrafine particles (UFP)-specific size range (<100 nm)), similar to other studies [33], in this document, we will refer to the assessed smaller particulate fraction, simply as UFP.

All data loggers were programmed for a 1-min data logging interval and were operated in the field according to the manufacturer’s recommendations.

2.3. Portable Air Cleaner Characteristics and Operation

A detailed description of the PACs used in some of the classrooms surveyed can be found elsewhere [34]. Briefly, the commercial PACs used in schools were Winix Zero Pro (and Winix Zero (Models AZPU370-IWE and AZBU330-HWE, WINIX España, Gerona, Spain), including a washable pre-filter made of fine mesh for coarse dust, an active charcoal filter for chemical contaminants and a glass micro-fiber HEPA filter H13 for submicron and UFP. Both PACs have some labels for specific certifications, including UK allergy, ECARF (European Centre for Allergy Research Foundation), and AHAM (American Household Appliance Manufacturers) seals.

During the pandemic, existing instructions refereed that portable air cleaners should be used in mode “high” or “turbo” to reduce the risk of aerosol transmission, but due to the noise produced in “turbo” mode, PACs were run most of the time in “high” mode. The clean air delivery rate (CADR) provided at this stage is 330 m³/h for the Winix Zero Pro and 270 m³/h for the Winix Zero, reaching an ACH of about 2.4 h⁻¹ and 1.9 h⁻¹, respectively, for a classroom of 50 m².

2.4. Statistical Analysis

To uniform the period of data availability in the three sampling campaigns, only data collected minute-by-minute for the teaching period from 9.00 to 14.00 for preschool and primary classrooms and 9.00 to 14.30 for secondary classrooms was selected for further analysis. Descriptive statistics for indoor air temperature, relative humidity, CO_2, and airborne PM (PM_2.5, PM₁₀, and UFP) were obtained using Microsoft Excel 2016. Although measurements were not conducted outdoors, for data analysis purposes, air quality and weather platforms were accessed to check the existence of data for the region and period of study. The outdoor air temperature was obtained through the AccuWeather website (www.accuweather.com). In addition, ambient air PM₁₀ levels were acquired from a local air quality monitoring station [35]. Statistical analysis was performed using IBM SPSS Statistics version 27, considering a significance level of p < 0.05. Data normality was checked using the Shapiro–Wilk and Kolmogorov–Smirnov tests. As the distribution of most air parameter levels was skewed, non-parametric tests were used for the analysis. Wilcoxon and Friedman’s tests were employed to investigate the existence of statistically significant differences across the levels of air parameters assessed across the three sampling campaigns. To assess differences in the levels of air parameters according to the dichotomous variables from information collected through a checklist (type of educational microenvironment, location of the classroom within the building (floor), and orientation of classrooms), Mann–Whitney U tests were applied. The Spearman method was used to investigate the existence of significant correlations among the levels of the several air pollutants concomitantly assessed. Samples taken with and without PACs in the same classroom but at different sampling campaigns were considered independent samples. Due to that, the sampling campaigns were compared using the nonparametric test for two independent samples, the Mann–Whitney U test.

3. Results

In general, the levels of the environmental parameters assessed in this study showed a considerable fluctuation across the sample of the surveyed classrooms.

Table 2. Descriptive statistics achieved for the levels of air parameters assessed in pre-, primary, and secondary classrooms (19 classrooms in 7 school buildings) during the 3 sequential sampling campaigns.

		Preschools (6 Classrooms)			Primary Schools (7 Classrooms)			Secondary Schools (6 Classrooms)
Parameter	Campaign	Median	Average (SD)	Min–Max	Median	Average (SD)	Min–Max	Median	Average (SD)	Min–Max
Temperature, °C	1st	21.4	20.9 (1.9)	15.1–23.9	21.3	20.7 (1.3)	17.4–24.2	18.6	18.9 (3.5)	13.5–26.7
	2nd	19.5	19.3 (1.5)	11.3–22.9	19.9	19.8 (1.1)	15.0–22.9	19.5	19.2 (1.0)	11.9–21.7
	3rd	17.2	17.2 (0.9)	12.5–20.6	17.5	18.0 (2.8)	12.7–24.0	17.7	17.8 (1.6)	14.3–22.6
RH, %	1st	43.3	45.4 (10.3)	26.3–66.1	46.6	50.5 (13.2)	26.1–78.5	45.2	42.2 (10.2)	23.5–63.2
	2nd	58.5	54.2 (11.7)	20.0–75.7	52.6	54.6 (4.4)	47.4–69.3	58.0	54.8 (12.6)	31.3–76.5
	3rd	55.8	54.3 (6.6)	40.6–67.7	53.4	49.8 (11.3)	32.2–67.1	48.6	53.0 (11.5)	37.4–86.4
CO2, ppm	1st	548.0	565.3 (57.4)	391.0–1075.0	584.0	625.2 (126.0)	379.0–1341.0	678.0	698.0 (166.6)	393.0–2072.0
	2nd	661.5	661.5 (71.6)	402.0–1147.0	632.0	709.6 (223.1)	396.0–1950.0	745.0	875.1 (240.9)	417.0–3405.0
	3rd	614.5	661.1 (153.8)	419.0–1334.0	688.5	718.3 (123.0)	423.0–2082.0	1041.0	1037.4 (315.2)	426.0–2506.0
PM2.5, µg/m3	1st	44.0	51.2 (25.7)	7.0–364.0	28.0	26.2 (8.8)	1.0–105.0	22.0	27.6 (12.4)	5.0–279.0
	2nd	39.5	52.0 (21.2)	11.0–252.0	51.0	46.3 (17.9)	3.0–480.0	38.5	46.3 (19.6)	12.0–133.0
	3rd	32.0	35.4 (21.1)	8.0–306.0	34.5	39.0 (14.6)	9.0–246.0	14.5	16.8 (7.7)	2.0–64.0
PM10, µg/m3	1st	70.0	86.4 (54.1)	10.0–749.0	41.0	40.4 (12.3)	2.0–141.0	32.5	39.6 (17.8)	6.0–490.0
	2nd	62.5	76.3 (32.6)	13.0–385.0	70.0	68.2 (27.3)	4.0–975.0	46.5	59.9 (30.8)	14.0–176.0
	3rd	44.0	51.7 (34.8)	9.0–497.0	46.0	53.1 (21.0)	10.0–411.0	21.0	22.7 (8.9)	3.0–103.0
UFP, pt/cm3	1st	7115	13,448 (8047)	1714–104,231	6815	6675 (3865)	185–67,053	6399	7011 (3430)	997–29,348
	2nd	10,393	16,103 (9925)	2302–105,571	6393	7791 (3409)	392–37,333	5450	7289 (2674)	1332–55,943
	3rd	6178	11,785 (7106)	1582–61,910	10,075	9928 (5468)	2210–78,246	4252	5025 (1668)	714–39,613

CO₂, carbon dioxide; PM, particulate matter; RH, relative humidity; SD, standard deviation; UFP, ultrafine particles. Min–Max, correspond to the maximum and minimum absolute values that were measured.

shows the summarized and structured results according to the educational environment studied (preschool, primary, and secondary education classrooms) and to the different sampling campaigns (1st, 2nd, and 3rd campaigns). The boxplots showing the distribution of the levels of the air parameters measured during the three campaigns per school building are presented in Figure 1. The main findings of the work will be fully described below.

3.1. Thermal Conditions and Relative Humidity Levels

For the days of sampling, indoor temperatures assessed in the classrooms ranged, on average, from 15.2 to 24.1 °C. No significant differences were observed in the average temperatures obtained for the three educational environments under study (preschool, primary, and secondary classrooms). Nevertheless, as shown in Table 2, the highest absolute temperature (26.7 °C) was obtained during the 1st campaign in a secondary school classroom. In turn, the lowest absolute indoor temperature of 11.3 °C was achieved in a classroom for preschool-age children during the work conducted in the 2nd campaign. In fact, the maximum variation in temperature for the learning period of a school day was obtained in this preschool classroom in the 2nd campaign (difference between maximum and minimum temperature values: 11.6 °C). In this case classroom, temperatures of around 11 °C were measured at the beginning of the teaching day (9 h), reaching a maximum of almost 23 °C at around 13 h.

Statistically significant differences were obtained in the levels of air temperature assessed among the three sampling campaigns (χ²(2) = 4.59, p = 0.003), with the significantly lowest average indoor temperature values being achieved in the 3rd sampling campaign. This corresponds to the period in which the outdoor temperature was also the lowest. In fact, according to AccuWeather records, ambient air temperatures were on average 18.5 °C, 16.1 °C and 9.5 °C for the periods of the study in the 1st, 2nd, and 3rd campaigns, respectively (range of average outdoor temperature values in the days of study: 5 to 26 °C). Statistically significant correlations between indoor and outdoor temperatures were also noticed (r_s = 0.633, p < 0.001). In addition, temperature values assessed indoors were found to be significantly associated with the number of windows and doors that were opened at the time of the monitoring activities (r_s = 0.041, p < 0.001).

In general, higher indoor air temperature values were statistically associated with lower indoor RH levels (

Table 3. Spearman correlation coefficients obtained for the levels of air parameters that were concomitantly assessed indoors.

	1.	2.	3.	4.	5.	6.
1. Temperature	1
2. RH	−0.277 *	1
3. CO2	0.021	−0.016	1
4. PM2.5	0.379 **	−0.088	0.001	1
5. PM10	0.449 ***	−0.111	−0.040	0.962 ***	1
6. UFP	0.306 *	−0.421 **	0.081	0.583 ***	0.547 ***	1

CO₂, carbon dioxide; PM, particulate matter; RH, relative humidity; UFP, ultrafine particles. * Correlation is significant at 0.05 level (2-tailed). ** Correlation is significant at 0.01 level (2-tailed). *** Correlation is significant at 0.001 level (2-tailed).

). No statistically significant differences were obtained between the RH levels measured in the different campaigns or across the different educational environments.

3.2. Indoor CO₂ Concentrations

The average CO₂ concentrations assessed in the 19 classrooms during the 1st, 2nd, and 3rd campaigns ranged from 462 to 904 ppm, from 565 to 1311 ppm, and from 547 to 1334 ppm, respectively. For instance, the assessments conducted in the 2nd and 3rd campaigns exhibited a non-significant trend for presenting higher average CO₂ levels than those observed in the 1st campaign (Table 2). As expected, the CO₂ concentrations measured in this study were significantly associated with the number of children in the classroom (r_s = 0.438, p = 0.001). In addition, indoor CO₂ concentrations were inversely correlated to the number of windows and doors that were opened at the time of the assessments (r_s = −0.386, p < 0.001). Nevertheless, the levels of CO₂ were not found to be significantly associated with the levels of any of the other environmental parameters that were simultaneously assessed in the classrooms (temperature, RH, PM_2.5, PM₁₀, and UFP) (Table 3).

For the 1st campaign, all 19 classrooms surveyed presented indoor CO₂ levels that were, on average, below 1000 ppm. However, the same was only accomplished in 17 (6 preschool, 6 primary, and 5 secondary classrooms) and 14 (6 preschool, 6 primary, and 2 secondary classrooms) out of the 19 classrooms surveyed in the 2nd and 3rd campaigns, respectively. Further, it was found that 6 (3 primary and 3 secondary classrooms), 9 (2 preschool, 2 primary, and 5 secondary classrooms), and 8 (1 preschool, 3 primary, and 4 secondary classrooms) of the studied classrooms presented average CO₂ concentrations that exceeded 700 ppm in the three sequential monitoring campaigns, respectively. For instance, among the educational microenvironments evaluated in this study, preschool classrooms presented, in general, the lowest average CO₂ concentrations (U = 195.00, z = −2.48, p = 0.013), while significantly higher concentrations were obtained in the secondary classrooms (U = 151.00, z = −3.13, p = 0.001).

The investigation of statistical associations of the average CO₂ concentrations with individual characteristics of the school buildings showed that classrooms in older buildings (according to the date of construction) presented a significant trend for displaying increased CO₂ levels (r_s = 0.469, p < 0.001). Classrooms located on the ground floor showed significantly lower mean levels of CO₂ (U = 199.00, z = −2.88, p = 0.003) than classrooms positioned on the 1st or 2nd floors of the school building.

3.3. Airborne Particulate Matter

For the particle size fractions (PM_2.5, PM₁₀, and UFP) assessed in the air of the 19 classrooms studied, it was found that the obtained average values per classroom varied from:

• 7.7 to 96.4 µg PM_2.5/m³, 10.4 to 187.9 µg PM₁₀/m³, and 296 to 24,847 #UFP/cm³ in the 1st campaign (average: PM_2.5, 34.6 µg/m³; PM₁₀, 54.7 µg/m³; and UFP: 8668 pt/cm³);
• 10.9 to 92.8 µg PM_2.5/m³, 16.2 to 138.8 µg PM₁₀/m³, and 2201 to 30,101 #UFP/cm³ in the 2nd campaign (average: PM_2.5, 48.1 µg/m³; PM₁₀, 68.1 µg/m³; and UFP: 9932 pt/cm³);
• 7.1 to 69.5 µg PM_2.5/m³, 10.2 to 110.0 µg PM₁₀/m³, and 2611 to 23,557 #UFP/cm³ in the 3rd campaign (average: PM_2.5, 30.1 µg/m³; PM₁₀, 41.9 µg/m³; and UFP: 8670 pt/cm³).

The obtained values presented great fluctuations in the classrooms of the different school environments (Table 2), with preschool classrooms presenting significantly higher average concentrations of PM_2.5 (U = 209.00, z = −2.08, p = 0.038), PM₁₀ (U = 190.00, z = −2.42, p = 0.015) and UFP (U = 130.00, z = −2.08, p = 0.002) than both primary and secondary classrooms. In turn, it was in classrooms used for secondary education where the significantly lower average concentrations of PM_2.5 (U = 207.00, z = −2.26, p = 0.023), PM₁₀ (U = 178.00, z = −2.78, p = 0.005) and UFP (U = 184.00, z = −2.35, p = 0.018) were obtained. Comparing the levels obtained among the three sampling campaigns, it was found that the PM_2.5 concentrations measured in the 2nd campaign were significantly higher than those levels assessed in the 3rd campaign (z = −2.44, p = 0.013). No statistically significant difference was observed in relation to the levels assessed in the 1st campaign (p > 0.05). However, it is important to disclose that the estimations were conducted just for the characterization of the levels obtained in the three sampling campaigns without taking into consideration the fact that some of the schools installed PACs during the 2nd and 3rd sampling campaigns.

Higher levels of PM_2.5, PM₁₀, and UFP were significantly associated with higher average indoor temperatures (Table 3). In addition, classrooms presenting lower average indoor RH (%) showed significantly higher airborne concentrations of UFP. Levels of the three particle size fractions that were measured were significantly correlated to each other (Table 3). For the fraction PM₁₀, mass concentrations assessed indoors were significantly correlated to the outdoor levels acquired from a local air quality monitoring station (r_s = 0.615, p < 0.001). In addition, a mean indoor-to-outdoor (I/O) PM₁₀ concentration ratio of 2.4 was obtained (1st campaign: 2.2; 2nd campaign: 2.7; 3rd campaign: 2.4).

Levels of all PM size fractions assessed were significantly associated with the number of doors in the classrooms (PM_2.5: r_s = 0.312, p = 0.020; PM₁₀: r_s = 0.375, p = 0.005; UFP: r_s = 0.406, p = 0.003). Interestingly, the number of windows was only significantly but negatively correlated to the number concentrations of UFP (r_s = −0.307, p = 0.027). PM_2.5 and PM₁₀ concentrations were also negatively correlated to the age of the building (PM_2.5: r_s = −0.285, p = 0.035; PM₁₀: r_s = −0.342, p = 0.011). The mean concentrations of PM_2.5, PM₁₀, and UFP were higher in classrooms located on the ground floor than those located on upper floors (PM_2.5: 41.8 vs. 35.1 µg/m³; PM₁₀: 63.7 vs. 49.4 µg/m³; and UFP: 10,953 vs. 7855 pt/cm³), but no statistical significance was found. Classrooms oriented to the outdoor playground presented lower average concentrations of PM₁₀ (U = 182.00, z = −2.23, p = 0.025) than classrooms facing streets. Nevertheless, classrooms from schools having an outdoor playground area with sand or soil (7 out of 19 classrooms) presented significantly higher average indoor levels of PM_2.5 (U = 217.00, z = −2.33, p = 0.019) and PM₁₀ (U = 186.00, z = −2.87, p = 0.004).

3.4. Impact of the Use of Air Cleaners

In order to reduce airborne aerosol concentration in the classrooms making school environments safer for children during the COVID-19 outbreak, the building managers of four schools (school IDs 1, 2, 5, and 7, including 3 preschools, 4 primary, and 4 secondary classrooms) installed air purifiers during the period of study. Figure 2 shows the boxplots of airborne particulate concentration levels (PM_2.5, PM₁₀, and UFP) investigated in the classrooms before installing the PACs and with PACs running in the classrooms. The boxplots represent the statistics of the entire period of teaching where the campaigns without or with PACs have been grouped. Schools 1 and 7 were monitored, with air purifiers operating in the 2nd and 3rd campaigns, while for schools 2 and 5, the air purifiers were run only during the 3rd campaign. Overall, the results obtained from the comparison between campaigns with and without using PACs were very controversial. In the case of preschool classrooms, significantly lower concentrations (p < 0.05) of PM_2.5, PM₁₀, and UFP were observed in the campaigns with PACs running, and even narrower boxplots with respect to the measurements with no air purifiers were observed. In turn, this reduction in airborne particles was not observed in the case of primary classrooms (1-b, 1-c, 2-b, and 5-b); in fact, higher concentrations of PM_2.5, PM₁₀, and UFP were registered in campaigns with the PACs operating. For secondary classrooms, significantly lower concentrations of PM_2.5, PM₁₀, and UFP were observed in the 3rd campaign of school 2 (2-c, 2-d) with the air purifier running, while an increase in the levels of airborne PM were observed in the secondary classrooms of school 7 (7-a, 7-b) with active PACs.

4. Discussion

This work shows that for colder days, the maintenance of the windows opened to comply with the recommendation of health authorities as a strategy for reducing the risk of airborne transmission of SARS-CoV-2 seemed to contribute to substantial thermal losses. This has resulted in observing comfort conditions in classrooms that fail to comply with SINPHONIE guidelines. In accordance with the respective recommendation, temperatures in classrooms should be maintained between 20 and 26 °C [36]. For the colder campaign of this study (3rd campaign), 15 classrooms were found to present average temperature values below 20 °C. The same situation was verified in seven and nine classrooms in the 1st and 2nd campaigns, respectively. The average temperature values obtained in the three sequential campaigns of this work (20.2 °C, 19.5 °C and 17.6 °C, respectively) were lower than those obtained in the SINPHONIE study (21.5 °C) [23]. In addition, very low absolute minimum temperatures reaching values around 11 °C were obtained for the coldest days of the study. For indoor RH, the average levels lower than comfortable range values (<30%) were only obtained in the 1st sampling campaign for two classrooms. Meteorologically, the region is characterized by hot and sunny weather during summer and dry and cold winters. The region is also characterized by an irregular rain regime. Rainfall is low (400–600 L m⁻² per year), being more abundant in autumn and spring. Castilla-La Mancha is included within the so-called “dry Spain” with very high aridity indices [37]. These conditions can play a crucial role in the evolution and transformation of air pollutants. For school buildings with similar characteristics but located in colder regions following the health authorities’ recommendation on the promotion of natural ventilation through opened windows and doors, the possibility of the existence of classrooms with comfort conditions even more compromised than those studied in this work cannot be excluded.

In addition to the associated thermal discomforts that can impact children’s well-being and learning performance, cold temperatures and low RH conditions may also favor the survival and transmission of respiratory viruses, as demonstrated mostly by the influenza virus [38,39]. In fact, it was recently reported that SARS-CoV-2 survives longest at low temperatures and extreme relative humidity, presenting a median estimated virus half-life of over 24 h at 10 °C and 40% RH, but of approximately 1.5 h at 27 °C and 65% [40]. Thus, measures to guarantee thermal and relative humidity levels within the comfortable ranges in indoor environments are fully advised, not only for achieving comfort-related benefits but also to promote airborne virus inactivation and avoid transmission in classrooms.

Further, the coldest indoor temperatures found in the 2nd and 3rd campaigns (and resultant thermal-related discomfort) are likely to increase the resistance of occupants to open windows of the rooms in these periods. This is likely to explain why, although without statistical significance, in general, higher average CO₂ concentrations were found in the coldest campaigns. Only 26% (4 preschool, 1 primary, and 0 secondary) of the surveyed classrooms presented CO₂ concentrations below 1000 ppm during 100% of the teaching hours in these three campaigns (occupancy period). That limit had been recommended in several countries worldwide in the pre-COVID-19 period [41]; also, only 5% (1 primary) presented values <700 ppm, the maximum concentration that was recommended for the outbreak period by several authors [42]. However, it is important to note that the average CO₂ levels assessed in this study were lower than those reported by some of the studies conducted in pre-COVID-19 period, including SINPHONIE [23], being even slightly lower than those works conducted in classrooms also during COVID-19 period [12].

CO₂ has been extensively described as a good indicator of ventilation conditions [43], and recently, the scientific community has pointed out the concentrations of CO₂ as a SARS-CoV-2 infection risk proxy for different indoor environments [9]. Ensuring low CO₂ levels in classrooms seems to reduce the probability that a child breathes the air exhaled by other(s) potentially infected classroom occupant(s). In this regard, there is emerging advice to control CO₂ in enclosed environments, mainly in rooms with a high density of occupants. This fact could ensure that CO₂ limit levels are not exceeded and then that ventilation conditions are sufficient to avoid the accumulation of CO₂ and viruses from the putative-infected occupants indoors. Since it has been recognized that increased CO₂ levels in classrooms can be linked to reduced learning performance [44] and absenteeism, mainly in the heating season [45], guaranteeing low levels of CO₂ will achieve wider benefits to children well-being beyond the COVID-19 situation. In this work, obtained data were extensively explored to investigate the factors that significantly influenced the CO₂ concentrations measured in the classrooms and, thus, identify opportunities for improvement. From this analysis, lower mean levels of CO₂ were reported in classrooms located on the ground floor compared with those located on the upper floors. This can be explained by a lower load of CO₂ to the indoor air because of the lower occupancy registered in the classroom on the ground floor. For instance, in this study, it was found that classrooms located on the ground floor presented significantly higher areas (m²) per occupant (U = 274.50, z = −1.98, p = 0.049) than other classrooms. A possible greater air renovation rate on the ground floor due to the existence of big entrance doors of the building that are frequently opened can also contribute to the dilution of CO₂ levels found in the ground floor rooms. Further, for the sample of school buildings that were studied, the age of the building was significantly associated with a lower area (m²) of the room available per occupant (r_s = −0.277, p = 0.037), and consequently, to the significantly higher CO₂ levels reported. This evidence suggests that strategies for adjusting the occupancy of the classrooms according to existing ventilation conditions of the rooms should be further explored, particularly in older school buildings, to guarantee CO₂ levels below the limits and minimize the risk of airborne disease transmission. As expected, it was also verified that, in general, the concentrations of CO₂ were significantly correlated with the number of occupants in the rooms. However, interestingly, classrooms of secondary schools that presented the lowest occupancy rate (average: 2.5 m²/occupant vs. 2.1 m²/occupant for both pre- and primary classrooms) were the educational environments that exhibited significantly higher CO₂ levels. This can be explained by the fact that students of secondary school age exhibit a higher CO₂ generation rate than younger pupils [46]. In turn, although a similar occupancy rate was verified for pre- and primary classrooms, preschool classrooms presented significantly lower average CO₂ concentrations. According to the evidence presented above, the fact that most of the classrooms for preschool-age children (83% (5 out of 6) vs. primary: 14% (1 out of 7)) were located on the ground floor of the school building can have an influence in this observation. Noteworthy, there were secondary school classrooms located on the ground floor that reached very high CO₂ concentrations (almost 3500 ppm in one of the campaigns). This fact can be explained by the fact that it was observed that the cross-ventilation practice was interrupted because students were able to close windows. This observation was less frequent in classrooms with younger children, where the teachers have more control in keeping the windows open, given the instructions provided by the authorities for the pandemic period.

In terms of PM, the average mass concentrations of PM_2.5 were assessed in 17 (6 pre, 6 primary, and 5 secondary), 18 (6 pre, 6 primary, and 6 secondary), and 14 (5 pre, 6 primary, and 3 secondary) out of the 19 studied classrooms exceeded the recommended limit of 15 μg/m³ for 24 h exposure, in the 1st, 2nd and 3rd campaigns, respectively. In the case of PM₁₀, 10 (4 pre, 4 primary, and 2 secondary), 16 (6 pre, 6 primary, and 4 secondary), and 8 (3 pre and 5 primary) of the studied classrooms exceeded the 45 μg/m³ recommended by the latest WHO guidelines [25], during the 1st, 2nd and 3rd sampling campaign, respectively. In addition, concentrations of PM_2.5 and PM₁₀ exceeding WHO guidelines were consistently found in the three sampling campaigns for twelve and five classrooms, respectively. In relation to outdoor PM₁₀ levels collected from the local air quality monitoring station, values above the WHO guideline were only found for 3 days of the monitoring period (corresponding to the sampling period of one classroom during the 1st campaign and two classrooms during the 2nd campaign). Although no exposure limits are established for UFP, the new WHO guidelines for air quality also include a good practice statement for UFP that refers that low particle number counts (PNCs) can be considered <1000 pt/cm³ (24-h mean), while a high PNC can be considered >10,000 pt/cm³ (24-h mean) or 20,000 particles/cm³ (1-h). In this study, mean UFP number concentrations lower than 1000 pt/cm³ were only obtained for one classroom of a primary school in the 1st campaign. In turn, from the assessments of the 1st sampling campaign, four classrooms presented high levels of UFP (>10,000 pt/cm³), while in the 2nd and 3rd campaigns, the number of classrooms in the same qualitative situation increased to six. According to WHO recommendations, high levels of UFP indicate that good practices for the management of this type of PM should be considered as preventive measures to avoid related health risks.

Noteworthy, in general, the preschool classrooms that were presented above as the educational environment studied that seemed to present the best ventilation conditions (according to the obtained CO₂ concentrations) presented significantly higher mean levels of PM_2.5, PM₁₀, and UFP than those that were measured in primary and secondary classrooms. However, as shown in Table 3, no significant associations were obtained between the levels of the different size fractions of PM and CO₂ concentrations.

From the statistical analysis of these data, the location of the classrooms within the building was found to have an apparent influence on the levels of PM found indoors. Classrooms facing the outdoor playground seemed to be more protected from PM₁₀ originating from traffic than classrooms facing streets. Nevertheless, the existence of a playground area with sand or soil was shown to be a risk factor for increasing PM_2.5 and PM₁₀. Previous studies have shown that indoor particle concentration can be particularly increased in rooms located on the ground floor, namely due to substantial contributions from outdoor pollution sources at the street level (e.g., traffic-related sources) [47]. Nevertheless, in this work, although the average concentrations of the assessed particle size fractions were higher in classrooms located on the ground floor than those on the upper floors of the building, the differences were not statistically significant.

The I/O PM concentration ratio has been widely used by researchers to estimate the contribution of PM of outdoor origin in the PM load detected indoors. Because this study did not include assessments in the outdoor environment, publicly available data from local air quality monitoring stations existing in the area of the surveyed schools’ location were collected. Although PM_2.5 is considered to be of special concern because of the robust evidence showing causal associations between exposure to PM_2.5 air pollution and all-cause mortality [25], the local air quality monitoring station located in the proximity of the surveyed schools does not include the assessment of PM_2.5 in the ambient air. Since above, we reported that most of the classrooms presented average PM_2.5 concentration exceeding the WHO limit value, and the absence of outdoor data constitutes an important limitation to the investigation of the contribution of the ambient air to high PM_2.5 concentration observed in the study classrooms. Since data for ambient air PM₁₀ concentration was available, the referred estimation was conducted only for PM₁₀. The I/O concentration ratio higher than 1 found for PM₁₀ suggests that this particle fraction size is likely to be attributed to both outdoor and indoor sources, with the indoor sources having an apparent greater contribution to concentrations found in the air inside the studied classrooms. However, this value should be interpreted with caution since indoor and outdoor data were acquired with different methodologies, which can introduce substantial bias in the result. In addition, the I/O ratio is probably not sensitive to differential time variations since the I/O concentration ratio obtained for a given pollutant may be higher or lower than the unity on some study days or periods of the day. Consequently, the calculated average value may not be fully representative of the actual origin(s). It is also important to disclose that the study design employed in this work did not consider the implementation of an approach to properly capture the influence on changes in pollution source contributions across the campaigns (e.g., the research team did not register changes on the type of activities, events or duration of windows opening). This compromised the comprehensive understanding of the findings, namely on the reason(s) that contributed to the fact that higher PM2.5 concentrations were found during the 2nd campaign. Thus, it is crucial that further research, including indoor assessment plans covering multiple assessment days/campaigns, consider the integration of procedures and/or tools (logsheets, checklists) to control at a high resolution the temporal changes in important events, behaviors, ventilation in order to identify local and time-dependent factors determining air pollution levels effectively. Even though existing evidence supports that, in addition to outdoor PM concentrations, some classroom characteristics, including the type of furniture (plastic laminate, composite, and plywood furniture), the use of products for waxing and polishing, and the number of counts of airborne fungi in indoor air are among the factors that significantly influence PM concentrations in classrooms [24].

Preschool classrooms are the classrooms with the highest levels of airborne PM but also with the best ventilation conditions in terms of CO₂ levels. In the case of school 1, which has a playground area with sand in front of the preschool classroom, greater reductions in PM_2.5, PM₁₀, and UFP were observed in the 2nd and 3rd campaigns with the installations of the PACs despite the windows being still open. This can be attributed to the fact that during the 1st campaign, the four windows in the classroom and door were totally opened (the temperature outside was pleasant), while during the 2nd and 3rd campaigns, only one or two windows were half open diagonally to the door allowed to have good ventilation but also lower levels of airborne PM due to a reduced air exchange rate (but enough) and the filtration guaranteed by the PACs. For the rest of the preschool classrooms, something similar happened. During the 1st campaign, all windows were completely open, while during the 2nd and 3rd campaigns, the opening of the windows was adjusted in order to reach a compromise between good ventilation (low CO₂ concentrations) and thermal comfort. It is worth mentioning that in preschool classrooms, the opening of windows was controlled exclusively by the teachers.

Noteworthy, in this study, it was found that for primary and secondary classrooms (except school 2), the concentration levels of PM_2.5, PM₁₀, and UFP were higher in the campaigns where air purifiers were running (see Figure 2). The higher levels of airborne PM cannot be attributed generally either to the location of the classroom (all classrooms on the first floor of the building except school 7 that were on the ground floor) or the location of the PAC in the classroom (in the center of the classroom: 1-b, 1-c; in a side of the classroom: 2-b, 2-c, 2-d, 7-a, and 7-b). In the case of classrooms 2-c and 2-d, significantly lower concentrations of airborne PM were observed with PACs, but other factors could influence the levels of indoor PMs, such as the opening of windows (in primary and secondary classrooms, the teachers do not control the opening of the windows as occurred in the preschool classrooms), the concentration outdoors, PACs operating not properly at the level recommended, activities promoting resuspension of particles, etc. In these classrooms, it is necessary to analyze the results on a case-by-case basis. In primary classroom 1-b, during the second campaign, the windows were shut during teaching hours, and only one window was opened diagonally to the door between classes. The CO₂ concentration was between 1000 and 2000 ppm all morning. Therefore, only one air purifier, although placed in the center of the classroom, does not seem to be enough to reduce the level of particles generated inside with respect to the 1st campaign. In secondary classroom 7-a, something similar happened during the second and third campaigns; the windows were shut most of the time, and the CO₂ concentrations were between 1000 and 3000 ppm. In secondary classroom 7-b, during the second campaign, the windows were also closed, and there was only 10 min of ventilation between classes. In classroom 2-b, during the third campaign (with an air purifier), there was an open window, but the CO₂ reached 1000 ppm. Here, the air purifier was placed on one side of the rectangular classroom, which seems to be insufficient to reduce the level of particles generated inside. Something similar happened in classroom 5-b, where the air purifier was placed in a corner of the room.

Therefore, it can be concluded that the classrooms operated with only an air purifier, generally in the center, and under natural ventilation with low levels of CO₂ and adjusting the opening of windows, PM levels were reduced compared with the classrooms with closed or partially closed windows. In these cases, a single air purifier is not capable of reducing the level of particles generated inside, probably due to resuspension, although it is expected that the concentration of PM will be lower than in the absence of an air purifier.

Pacitto et al. studied the effect of ventilation strategies and air purifiers in school gyms on the children’s exposure to airborne particles and concluded that the use of air purifiers (well-sized with respect to the classroom´s size) with windows closed leads to a significant reduction in terms of indoor-to-outdoor concentration ratios [48]. However, in the case of opened windows, the effect of the reduction caused by air purifiers on outdoor-generated submicron particles (10–700 nm) is negligible.

In general, the findings from studies investigating the potential of the use of air purifiers to reduce exposure to airborne particle concentrations indoors are quite positive [48,49]. However, these present data are controversial since the effectiveness of the use of PACs in reducing particle levels in indoor air was only shown for a limited number of classrooms (mainly preschool classrooms). Nevertheless, it is important to disclose that although the present study was conducted in real scenarios during the pandemic period, its study design presents certain limitations. In particular, the comparison of the concentration before and during the use of PACs was carried out through the comparison of data obtained for three different campaigns separated in time from September to December. In addition, the implemented methodology did not include the assessment of the concentration of PM in the surrounding outdoor environment, which would be useful to accurately adjust the models to the relevant influence of the outdoor environment due to the recommendations for maintaining open windows. Although outdoor concentrations of PM₁₀ were acquired from a local monitoring station, the attempt to adjust indoor levels based on these levels resulted in similar controversial results. However, it is important to highlight that the purpose of the PACs during the pandemic was exclusively to try to reduce the virus-laden aerosol, and there is consistent evidence in the literature on this effectiveness. For example, in our previous study [34], the detection of SARS-CoV-2 in the HEPA filters of two of the participating schools in this study was confirmed, suggesting that the virus-laden aerosol can become trapped in the PAC filters. In addition, there is evidence showing that in conditions in which the heating system was operated with one or two PACs, under natural ventilation, the amount of virus-containing aerosol in the air can be decreased by 86–88% compared with settings in which neither natural ventilation nor an air purifier are used [50]. For the cases in this study, only one unit of PAC was used per classroom, and depending on the characteristics and specific needs of the indoor space, this number could be insufficient to improve air quality. In fact, broader data would be needed to draw conclusions that are more consistent on the beneficial impact of the use of PACs in classrooms.

Overall, although the occupancy rate was, on average, very similar in the classrooms of the three studied environments (pre: 2.1; primary: 2.1; secondary: 2.5 m²/occupant), according to the results presented in this work, children of different ages at different education levels can be exposed to significantly different environmental conditions (namely in terms of ventilation and PM pollution). In fact, classrooms for preschool-age children presented better ventilation conditions, which effectively avoided the accumulation of CO₂ in the indoor air, than primary and secondary classrooms, but were significantly more polluted with PM, probably coming from outdoors.

5. Conclusions

In summary, findings from this study showed that the implementation of COVID-19 emergency protocols that included recommendations for improving natural ventilation seemed to be effective in achieving average CO₂ levels in classrooms that were lower than those reported by some of the studies conducted in the pre-COVID-19 period. Nevertheless, this achievement can be compromised on colder days of the heating season, with some of the classrooms presenting average CO₂ levels exceeding 1000 ppm. In addition, evidence was obtained that demonstrated that for cold seasons in climates such as in central Spain, the recommendation for maintaining open windows during the learning period to prevent the risk of airborne infectious diseases could result in important thermal comfort issues. It cannot be excluded that such thermal issues are likely to compromise children’s well-being and learning performance. Further, a substantial number of the classrooms exhibited high levels of airborne PM, and the implementation of one single unit of PAC seems to be a solution that cannot be effective in all classrooms to reduce the levels of PM to safe levels. The outcomes from this study also showed that environmental quality in classrooms could be significantly impacted by several factors, such as building (e.g., age of building, floor), occupant (age), and organizational (occupancy density)-related characteristics. Indeed, although the establishment of common evidence-based guidance is important, based on the findings that demonstrated the existence of significant differences across different educational settings, the corrective measures to employ in each classroom/school should reflect the individual characteristics and needs of the target space/building.