Carbon dioxide (CO2) demand-controlled ventilation in university computer classrooms and possible effects on headache, fatigue and perceived indoor environment: an intervention study

Original Article

DOI: 10.1007/s00420-012-0756-6

Cite this article as:
Norbäck, D., Nordström, K. & Zhao, Z. Int Arch Occup Environ Health (2013) 86: 199. doi:10.1007/s00420-012-0756-6

Abstract

Purpose

To study the effects of a CO2 demand-controlled ventilation system (variable flow) in computer classrooms on perceived air quality and sick building syndrome.

Methods

University students (27 % women) participated in a blinded study. Two classrooms had variable flow (mean 5.56 ac/h); two others had constant ventilation flow (mean 5.07 ac/h). After one week, ventilation conditions were shifted. The students reported symptoms/perceptions during the last hour on rating scales. Temperature, air humidity, CO2, PM10 and number concentration of particles were measured simultaneously. Cat (Fel d 1), dog (Can f 1), horse (Equ cx) and house dust mites (Der f 1 and Der p 1) allergens were measured in dust. Those participating twice in the same classroom (N = 61) were analysed longitudinally.

Results

Mean CO2 was 784 ppm (9 % of time >1,000 ppm) with variable flow and 809 ppm with constant flow conditions (25 % of time >1,000 ppm). Mean temperature (22.6 °C), PM10 (18 μg/m3) and number concentration (1,860 pt/cm3) were unchanged. The median levels of cat, dog, horse and Der f 1 allergens were 10,400 ng/g, 4,900 ng/g, 13,700 U/ng and 260 ng/g dust, respectively. There were slightly less headache (p = 0.003), tiredness (p = 0.007) and improved perceived air quality (p = 0.02) with variable flow.

Conclusions

Use of a CO2-controlled ventilation system, reducing elevated levels of CO2, may slightly reduce headache and tiredness and improve perceived air quality. The high levels of pet allergens, due to track in of allergens from the home and possible accumulation due to electrostatic forces, illustrate a need for improved cleaning.

Keywords

Furry pet allergens Indoor air quality Ventilation Room temperature Sick building syndrome (SBS) University students 

Introduction

The effects of building ventilation on humans have been reviewed, suggesting that ventilation rate increases up to 10 L/s per person (Godish and Spengler 1996), or reducing CO2 down to 800 ppm (Seppänen et al. 1999) would be beneficial. Another review (EUROVEN) concluded that a low ventilation rate is associated with health effects and decreased performance in offices (Wargocki et al. 2002). Carbon dioxide has been used as an indicator of ventilation. In the ventilation standard in Sweden, it is recommended that indoor CO2 levels should be below 1,000 ppm and that outdoor air flow should be at least 7 L/s per m2 and additionally 0.35 L/s per m2 floor surface (National Swedish Board of Occupational Safety and Health 2000). The need for sufficient ventilation has been pointed out in a previous review on health and student performance in relation to the school environment (Daisey et al. 2003; Mendell and Heath 2005).

There is an increasing demand on energy saving in buildings. Demand-controlled ventilation is one way to save energy in buildings (Fisk and De Almeida 1998; Haghighat and Donnini 1992; Mysen et al. 2003; Roth et al. 2003; Schell and Inthout 2001). In this case, the ventilation flow is reduced when the building is not occupied, by sensor-controlled ventilation. Commonly CO2 sensors, infra red (IR) occupancy sensors, temperature sensors, or relative air humidity sensor can be used to control the ventilation flow. It has been estimated that typically the investment in such systems would pay off within a few years (Fisk and De Almeida 1998). The effect of demand-controlled ventilation on indoor air quality is unclear. One publication points out the potential problem with improper CO2 sensor installation or sensor failure, leading to unacceptable levels of indoor pollutants (Roth et al. 2003). Most technical studies on demand-controlled ventilation have been dealing with energy saving in dwellings (Pavlovas 2004; Nielsen and Drivsholm 2010). Two Norwegian studies on demand-controlled ventilation in schools suggested considerable potential for energy saving (Mysen et al. 2005; Wachenfeldt et al. 2007). We have not found any studies on effects of demand-controlled ventilation on symptoms or environmental perceptions among pupils or staff in schools or university buildings.

Classrooms are densely populated, which can lead to both thermal discomfort and perception of poor indoor air quality. Moreover, large glass windows can increase the thermal problems during warmer parts of the year. The increased use of computers may increase the thermal load, and computers have been identified as an important source of sensory pollution load (olf) (Bako-Biro et al. 2004). There are few studies on health effects of ventilation flow or thermal conditions in university buildings (Bakke et al. 2008). In 2004, we performed an experimental blinded study in four computer classrooms at a technical university in South Sweden, using two different constant flow conditions (Norbäck and Nordström 2008a, b). Room temperature and CO2 levels were often above accepted comfort standards, and there were associations between both thermal conditions and ventilation flow and sick building syndrome (SBS) (Norbäck and Nordström 2008a) and perceived indoor environment (Norbäck and Nordström 2008b). The results initiated environmental improvements, such as reduced number of computers in the classrooms and increased ventilation flow. The environmental improvements enable us to perform a new study in 2005, comparing a CO2-controlled ventilation system with constant flow conditions, in the same computer classrooms.

The main aim was to study the effects of a variable flow ventilation system, controlled by CO2 levels and temperature, as compared to a constant flow system in computer classrooms. The following hypothesis was tested: Use of the variable ventilation system may reduce mucosal, dermal and general symptoms and influence perceptions of the indoor environment and air quality. The hypothesis was tested in a single-blinded intervention study with cross-over design, using two ventilation conditions, either with variable ventilation conditions or with a constant ventilation flow near the recommended ventilation standards. Moreover, since we found no data on furry pet allergen levels in university buildings and a relatively large proportion (14 %) of the university students at Lund University had previously reported furry pet allergy (Norbäck and Nordström 2008a), we measured pet allergens in settled dust. Allergens in educational buildings may have implications for the sensitised person. One Swedish study showed that children with cat allergy had an increased risk of exacerbation of asthma at school, measured as asthma symptoms, peak expiratory flow and use of asthma medication if they were in classes with many cat owners (Almqvist et al. 2001). Another Swedish school study found an increased incidence of asthma diagnosis in schools with higher levels of cat allergen in dust (Smedje and Norbäck 2001).

The protocol of the study was approved by the Ethical Committee of the Medical Faculty of Uppsala University.

Materials and methods

Study design

A single-blinded cross-over study manipulating the mode of ventilation (demand-controlled variable flow or constant flow) was performed in four computer classrooms (A, B, C and D) at a department at Lund Technical University, Sweden. Data were collected during 2 weeks in April 2005, Monday and Tuesday one week (week 1), and Monday and Tuesday the next week (week 2). The basic flow with no occupancy was set to 4 L/s per person, calculated for 24 persons in smaller rooms (A, D) and 32 persons in larger rooms (B, C). This would correspond to an air exchange rate of 2.0–2.2 ac/h when the classrooms are empty. The calculated personal air flow during occupancy was designed to be around 10 L/s, in accordance with recommended ventilation standards.

All classrooms had supply-exhaust ventilation with frequency-controlled fans and individual regulation of the supply air flow. The ventilation flow could be electronically regulated by changing the frequency of the fan, in combination with a change in the inlet areas of the supply air tackle (VAV system, Lindivent, Lund, Sweden). When the ventilation flow was increased, the inlet area was increased, and thus the air velocity of the supply air was kept constant. Each classroom had 6–8 supply air tackles placed 2.4 m above the floor. There was one exhaust air tackle in each room situated in one corner of the ceiling. The ventilation system was programmed to be controlled by sensors measuring CO2 levels and temperature in the exhaust air, increasing the ventilation flow to maximum when the CO2 levels or temperature were above the critical levels (800 ppm, 22.0 °C).

The rooms were used by some students outside the lecture time, in mornings, evenings and in the weekend. However, the study was limited to morning and afternoon lectures with full classes. Each day, questionnaire data were collected twice, first during morning classes and after the lunch break with new students during afternoon classes. The questionnaire was distributed at the end of the lectures, when the students had been at least 1 h in the classroom. The recall period for symptoms was 1 h. Climate measurements were performed during the whole day (9 a.m.–3 p.m.), but each subject was addressed the average values for the same period as the recall period for the symptoms (1-h). All computers were turned on during the lectures. The number of students fluctuated during the lectures, usually with 2–3 students per computer. There were normal and similar types of lectures both weeks, with no exams.

All the students (age, 20–25 years) who were in the morning or afternoon classes on Monday and Tuesday the first week were invited to participate. During this week, two of the classrooms (B, C) had constant flow conditions (VAV-system off) and two (A, D) had variable flow (VAV-system on). The study was repeated on Monday and Tuesday the next week, and all students in the rooms were invited to participate the second week. During the second week, the B and C classrooms had variable flow conditions and the A and D had constant flow. The students participating during the first week were asked to sit in the same classrooms the week after, but some did not follow this instruction. The total air flow in the main supply air ducts in each room, during constant flow conditions, was measured by tracer gas technique on Tuesday afternoon both weeks. This was done by adding a known flow of N2O in the supply air system and measuring the N2O concentration in the end of the supply ducts. The ventilation flow during variable flow conditions were measured by in-built sensors in each supply air duct, using wireless connection to a hand-held computer. The first answered questionnaire from each participant, irrespectively of week, was used to describe the population and to study gender differences. Data from a sub-population of students participating both weeks were used in a longitudinal analysis to study changes in symptoms and perceptions in relation to type of ventilation system.

Assessment of personal factors, perceived indoor air quality and medical symptoms

The students received a questionnaire at the end of the lecture. It included questions on personal factors, such as gender, smoking, contact lens wearing, doctor-diagnosed asthma, hay fever, furry pet allergy and food allergy/intolerance. In addition, there were eight questions on perception of the indoor environment, used in previous studies (Lindgren et al. 2006; Norbäck and Nordström 2008b), requiring answers on six-category scales, coded from 0 to 6. One question asked about the immediate perception of the air quality when entering the classroom (the first 15 min), and the other asked about perceptions the last hour. For temperature, air humidity and air movement, which can be either too low or too high, 3 indicates the neutral point (not too low, not too high). For air quality (current or the first 15 min), 6 is the best option (extremely good). For noise in general, noise from the ventilation system, illumination and odour, 0 indicates the best option.

Finally, the questionnaire contained 11 symptom questions, asking for the degree of symptoms and requiring answers on a 6-graded scale, adapted from previous studies (Norbäck et al. 2006; Norbäck and Nordström 2008a,). One question on ear problems and another on musculoskeletal symptoms were not analysed here. An absence of symptoms was coded ‘0’, very slight symptom(s) was coded ‘1’, slight symptom(s) was coded ‘2’, moderate symptom(s) was coded ‘3’, strong symptom(s) was coded ‘4’, very strong symptom(s) was coded ‘5’ and unbearable symptom(s) was coded ‘6’. The questions included nasal symptoms, ocular symptoms, throat symptoms, dermal symptoms, sinusitis symptoms, dyspnoea, nausea, headache and tiredness.

Classroom energy and ventilation characterisation

The classrooms were on the third floor of a five-floor brick building from the 1960s, situated away from major streets in the city of Lund in southern Sweden. All had a linoleum floor, large windows, painted walls and painted wood-fibre acoustic absorbents in the ceiling. All windows were closed during the experiment, which was normal since they could not be opened. Doors were kept closed, except when students entered or left the rooms. Each classroom had 12–16 computers of an old type, with VDU display using vacuum tubes. All computers were turned on during the lectures. The estimated daytime heat effect per room was 1,800–2,400 W from computers (mean effect 150 W/computer), 720–1,080 W from illumination and 2,400–3,200 W from the students, resulting in a total heat load ranging from 4,900–6,700 W (mean 5,800 W). The supply air flow was measured by a tracer gas dilution method, supplying a constant flow of N2O in each supply air duct and measuring the flow rate of N2O by the mass flow meter Bronkhorst High-Tech BV (Bronkhorst High Tech BV, Ruurlo, The Netherlands). The concentration of N2O in the supply air was measured by SensAir 2000 1L (Sense Air, Delsbo, Sweden). The air exchange rate was calculated by dividing the total supply air flow in the room with the room volume.

Methods to measure climate and air pollution in the classrooms

Indoor and outdoor measurements were performed in parallel with the questionnaire study on Mondays and Tuesday two consecutive weeks, from 9 a.m. to 3 p.m. each day. Room temperature, relative air humidity and CO2 were measured with a Q-Trak™ IAQ Monitor (TSI Incorporated, USA), sampling 1 min average intervals. Measurements were performed during the whole day, but only data from the same time period as the recall period of the questionnaire (1-h mean values) were used in the calculations. Particles were measured with both P-Trak™ (Model 8525 Ultrafine Particle Counter; TSI Incorporated, USA), measuring particles in the size range 0.02–1 μm (number concentration), and DustTrak™ (Model 8520; TSI Incorporated, USA), measuring particles from approximately 0.3–10 μm (PM10). The instruments were calibrated by the Swedish service laboratory for TSI equipment. The same individual instrument was used in each respective classroom, both weeks. The instruments were placed on separate desks at the same heights as the breathing zone of the students. Supply air temperature was measured by a Digital Thermometer TES 1316 (Instrumental Sales and Rental, Ontario, Canada), at different points of the supply air duct.

Airborne micro-organisms were sampled on 25 mm Nuclepore filters with a pore size of 0.4 μm (1.5 L/min; 4-h sampling time). The total concentration of airborne moulds and bacteria was determined by the CAMNEA method (Palmgren et al. 1986). Viable moulds and bacteria were determined by incubation on two different media. The detection limit for viable organisms was 30 colony-forming units (cfu) per m3 of air. Formaldehyde was sampled during one week using another diffusion sampler (Levin et al. 1988) and then analysed by high-pressure liquid chromatography.

Allergen measurements

Settled dust was collected from desks, chairs and the floor by a 1,000-W-vacuum cleaner (Philips Vital 371) provided with a special dust collector (ALK Abello, Copenhagen, Denmark) equipped with a Millipore filter (pore size 6 μm). Vacuum cleaning was performed for totally 4 min per sample, 2 min on the floor and 2 min on other surfaces (desks, chairs), as in previous studies (Kim et al. 2005; Zhao et al. 2006). Each classroom was divided into two halves, one near the corridor and the other near the windows, and one sample was collected from each half. The filters were sealed in plastic bags and stored at −20 °C until extraction. The total amount of dust was weighed in each sample. Samples of settled dust (100 mg) were extracted in 2 ml of phosphate buffered saline containing 0.05 % Tween 20 (1/20 W/v) by rotating mixing for 2 h at room temperature. Samples were then centrifuged at 4,500 rpm for 10 min followed by another centrifugation of the supernatant at 10,000 rpm for 10 min. The final supernatant was transferred to Micro tubes (Sarstedt, Germany) and stored frozen at −20 °C until analysed for the content of allergen. Enzyme-Linked Immunosorbent Assay (ELISA) was applied to determine the allergen levels of cat (Fel d1), dog (Can f1) (Indoor Biotechnologies Ltd, Manchester, UK), and horse (Equ c x) (Mabtech, Stockholm, Sweden) as previously described (Kim et al. 2005; Zhao et al. 2006).

Statistical analysis

Mann–Whitney U test was used to compare differences in environmental perceptions between men and women. The study on symptoms/perceptions was restricted to those students participating and sitting in the same classroom both weeks. For each perception or symptom rating, the difference in the score was calculated for each subject by subtracting the first week’s score from that of the second. Finally, the differences in change in each group’s score were tested by Mann–Whitney U test. Two-tailed tests and a five-per cent level of significance were used. The prevalence of SBS symptoms was calculated, coding moderately, strongly, very strongly and unbearable symptom as ‘1’ and no symptoms, very little or slightly symptoms as ‘0’. The symptom prevalence was used in a descriptive table only; all statistical calculations were performed with the original scores.

Results

Among those present in the classrooms during the study period, more than 90 % agreed to answer the questionnaire at least once. A total of 232 students participated during either week 1 or week 2, 150 answered the first questionnaire week 1 and 82 answered the first questionnaire week 2. Participation in the computer classes was not compulsory, so the students could choose to be away if they felt they know the topic enough. Only 71 participated twice, 37 of them had variable flow week 1 and constant flow week 2, 24 had constant flow week 1 and variable flow week 2. Totally 10 had unchanged conditions because they did not sit in the same room both weeks and were excluded from the longitudinal analysis that comprised 61 subjects.

Among the 232 participants, one-third was women, one-fifth had contact lenses, a few had asthma or were smokers and one-fifth (25 %) had either pollen or furry pet allergy (atopy) (Table 1). There was no obvious difference in prevalence of personal factors between those from the two different ventilation sequence groups. Those participating twice had somewhat more hay fever and furry pet allergy; otherwise they did not differ from the total material (Table 1). Female students rated room temperature as colder (Table 2) and male students reported more sinusitis symptoms (Table 3). Otherwise there were no gender differences in symptom rating or environmental perceptions.
Table 1

Prevalence of personal factors among the participants in different data sets

 

All participants witha

Longitudinal analysisb (N = 61) (%)

Total materiala (N = 232) (%)

Constant flow (N = 1,112) (%)

Variable flow (N = 120) (%)

  

Female

29.0

26.0

31.1

27.2

Current smoker

4.5

1.8

3.3

3.1

Ex-smoker

9.1

7.1

6.7

8.1

Contact lens wearer

19.1

17.7

18.0

18.4

Hay fever

19.4

20.7

25.0

20.1

Fury pet allergy

18.7

14.4

20.0

16.5

Food allergy/intolerance

11.1

7.1

9.8

9.1

Doctor’s diagnosed asthma

4.7

7.3

6.7

6.0

aBased on the first answered questionnaire, irrespectively of week

bLongitudinal analysis is based on subjects participating twice and sitting in the same classroom

Table 2

Mean rating of different aspects of indoor environment

Type of perception

Males (N = 169)

M(SD)

Females (N = 63)

M(SD)

p valuea

Total (N = 232) M(SD)

Room temperature

 Too cold (0)–too hot (6)

3.5 (1.0)

3.2 (0.8)

0.003

3.4 (0.9)

Air humidity

 Extremely dry (0)–extremely humid (6)

2.7 (0.9)

2.5 (0.9)

0.17

2.6 (0.9)

Air movement (draught)

 No movement (0)–extremely draughty (6)

2.2 (1.0)

2.4 (1.1)

0.34

2.2 (1.0)

Air quality

 Extremely poor (0)–extremely good (6)

2.1 (1.0)

2.1 (0.9)

0.63

2.1 (1.0)

Odour

 No odour (0)–extremely strong odour (6)

2.0 (1.4)

1.9 (1.4)

0.71

2.0 (1.4)

Illumination

 Very good (0)–very poor (6)

2.4 (1.3)

2.6 (1.4)

0.23

2.5 (1.3)

Noise in general

 No disturbing noise (0)–very disturbing (6)

3.3 (1.5)

3.5 (1.2)

0.58

3.3 (1.4)

Noise from ventilation system

 No disturbing noise (0)–very disturbing (6)

2.0 (1.5)

1.9 (1.5)

0.67

1.9 (1.5)

Immediate perception of air quality

  Extremely poor (0)–extremely good (6)

2.5 (1.2)

2.8 (1.0)

0.07

2.6 (1.2)

Environmental ratings was judged from from 0 to 6. All questions asked for perceptions during the last hour, except immediate perception which covered the first 15-min in the classroom

aDifference in score (0–6) calculated by Mann–Whitney U test

Table 3

Mean rating of different symptom ratings and prevalence of symptoms

Type of symptom

Males (N = 169)

M (SD)

Females (N = 63)

M (SD)

p valuea

Total (N = 232)

M (SD)

Total prevalence (%)

Eye symptoms

0.9 (1.3)

0.9 (1.3)

0.99

0.9 (1.3)

13.9

Nasal symptoms

0.8 (1.2)

0.6 (1.1)

0.35

0.8 (1.2)

10.8

Throat symptoms

1.1 (1.4)

0.8 (1.2)

0.26

1.0 (1.4)

16.6

Breathing difficulties

0.4 (1.1)

0.2 (0.6)

0.08

0.4 (1.0)

5.4

Sinusitis symptoms

0.4 (1.0)

0.1 (0.4)

0.008

0.3 (0.9)

4.5

Dermal symptoms

0.4 (0.9)

0.4 (0.7)

0.99

0.3 (0.8)

3.6

Headache

1.3 (1.5)

1.3 (1.5)

0.84

1.3 (1.5)

21.5

Tiredness

2.1 (1.7)

1.7 (1.4)

0.09

2.0 (1.6)

35.6

Nausea

0.4 (1.0)

0.3 (0.7)

0.46

0.4 (0.9)

4.9

Symptoms ratings was judged from 0 to 6 (none = 0, very slight = 1, slightly = 2, moderately = 3, strongly = 4, very strongly = 5, unbearable = 6)

All questions asked for symptoms during the last hour

When calculating prevalence (%), moderately, strongly, very strongly, and unbearable symptoms were coded as 1; No symptom, very little or slightly symptom were coded as 0

ap values by Mann–Whitney U test comparing the scores (0–6) between men and women

During variable ventilation conditions, the mean value for all four classrooms (week 1 and combined) was 22.6 °C, 28 % RH and 784 ppm CO2 and mean air exchange rate during lectures was 5.56 ac/h. The ventilation fluctuated during lectures, the air exchange rate ranged from 2.89 to 6.95 ac/h (mean 5.56 ac/h) and personal supply ventilation flow varied from 8.9 to 37.5 L/s (mean 16.1 L/s). The highest personal air flow rate occurred when there were few students in the classrooms. During constant flow conditions, the mean value for all four classrooms (week 1 and 2 combined) was 22.5 °C, 30 % RH, 809 ppm CO2. The mean air exchange rate during constant flow conditions was 5.07 ac/h, and the mean personal supply ventilation flow was 13.3 L/s (Table 4). The proportion of time with CO2 levels above 1,000 ppm was 8.7 % during variable flow conditions and 25.4 % during constant flow (p < 0.001, tested for 1-min mean values). The maximum 1-min mean value was 1,224 ppm, and minimum was 600 ppm. The proportion of time with temperatures above 23.0 °C was similar during both conditions. The mean number of students in the four classrooms was 19 during variable flow conditions and 20 during constant flow conditions, but with large fluctuations from 4 to 33 students during the lectures. The supply air temperature was 13.0 °C in the supply air system, measured after the air cooling units situated near the air intake, but due to a heat exchange effect between supply and exhaust air ducts, supply air temperature increased to 16.5 °C in room A, 19.0 °C in room D, 17.0 °C in room B and 18.0 °C in room C.
Table 4

Indoor climate and ventilation in the four computer classrooms (A, B, C and D) during week 1 and week 2

Exposure factor

Variable flow week 1 and constant flow week 2

Constant flow week 1 and variable flow week 2

A M (min–max)

D M (min–max)

B M (min–max)

C M (min–max)

Room volume (m3)

162

171

207

207

Number of students

 Week 1

19 (10–26)

17 (4–31)

23 (6–30)

15 (2–23)

 Week 2

21 (4–31)

22 (8–33)

25 (13–32)

15 (7–21)

Air exchange rate (ac/h)

 Week 1

4.89

5.99

5.67

4.45

 Week 2

4.78

5.37

5.89

5.47

Supply ventilation flow (L/s*p)

 Week 1

12.8

19.4

14.2

17.1

 Week 2

10.2

11.6

13.3

19.2

Temperature (°C)

 Week 1

22.5 (22–24)

22.0 (22–23)

22.7 (22–23)

22.6 (21–24)

 Week 2

22.1 (21–24)

22.5 (22–24)

23.6 (22–24)

22.7 (22–23)

Relative air humidity (%)

 Week 1

35 (33–37)

36 (34–39)

35(32–38)

35 (32–37)

 Week 2

23 (21–33)

25 (23–39)

21 (20–22)

21 (20–21)

Carbon dioxide (ppm)

 Week 1

810 (600–960)

680 (560–840)

970 (920–990)

770 (470–990)

 Week 2

720 (600–880)

670 (600–840)

990 (940–1,020)

620 (610–630)

Min–max temperature, RH and CO2 are given as range of mean values for the 1-h recall period for the questionnaire

Arithmetic mean values (M) are total means of all 1-h means during the week

Air exchange and supply ventilation flow during variable flow conditions are spot measurements daytime during two weekdays

The mean PM10 in all four classrooms was 17 μg/m3 (range of classroom mean, 10–24) at variable flow conditions and 18 μg/m3 (range of classroom mean, 8–23) at constant flow conditions. The mean number concentration of particles measured by the P-trak was 1,880 pt/cm3 (range of classroom mean, 1,210–2,480) during variable flow conditions and 1,850 pt/cm3 (range of classroom mean, 640–3,760) at constant flow. There was a 2–3 times day-to-day variation of indoor number concentrations. None of these differences were statistically significant, if testing 1-min mean values. The mean concentration of viable bacteria in the four classrooms (<80–100 cfu/m3) and viable moulds (<80–100 cfu/m3) was similar during both conditions. No particular species could be identified. The mean concentration of total bacteria in the four classrooms was <10,000–41,000 per m3, and the concentration of total moulds was <10,000–50,000 per m3, similar during both conditions. The indoor concentration of formaldehyde measured during 1 week was low (10–16 μg/m3) in all rooms, and 2 μg/m3 outdoors. The levels of both cat, dog and horsed allergens were high in vacuumed dust from all classrooms, and considerably higher than in dust samples collected in Swedish primary school, using the same method (Kim et al. 2005) (Table 5).
Table 5

Allergen concentration in vacuumed settled dust in computer classrooms, as compared to Swedish primary schools

 

N

Median (IQR)

Min–max value

Median (QR) in schoolsa

Cat allergen (Fel d 1) (ng/g dust)

8

10,400 (8,970–11,380)

7,300–14,100

860 (450–1,425)

Dog allergen/Can f 1) (ng/g dust)

8

4,930 (4,090–5,560)

4,000–6,700

750 (443–3,000)

Horse allergen (Equ c x) (U/g dust)

8

12,940 (9,490–17,370)

10,800–22,100

945 (535–2,200)

House dust mite (Der f 1) (ng/g dust)

8

270 (210–330)

<40–460

<200 (<200–<200)

Median (IQR) = median with inter-quartile range. Note: Totally two dust samples per classroom. The house dust mite allergen Der p 1 was below the detection limit (<40 ng/g dust) in all samples

aData from 24 classrooms in eight primary schools in mid-Sweden (Kim et al. 2005)

Finally, we studied change in environmental ratings and change in symptom score during the study period for those participating twice sitting in the same classroom (N = 61), excluding those who did not sit in the same classroom twice (N = 10) (Table 6). There was no difference in prevalence of women between the two groups with different test sequence. In general, both environmental perceptions and symptom rating were better during variable flow conditions. The changes were mainly occurring in the group who had constant flow week 1 and variable flow week 2. Statistically significant difference, in favour of the variable flow conditions, was observed for immediate perception of air quality (p = 0.02), headache (p = 0.003), and tiredness (p = 0.007). There was no significant effect on mucosal symptoms (eye, nose, and throat). Effects on other medical symptom were not analysed, since they were too rare.
Table 6

Change in ratings of environmental perceptions, headache and tiredness in relation to system activation sequence (N = 61)

Type of perception

Variable flow week 1

Constant flow week 2 (N = 37)

Constant flow week 1

Variable flow week 2 (N = 24)

Two-tailed p valuea

Week 1 M (SD)

Week 2 M (SD)

Change in mean

Week 1 M (SD)

Week 2 M (SD)

Change in mean

Room temperature

 Too cold (0)–too hot (6)

3.7 (0.9)

3.6 (1.0)

−0.1

3.2 (0.8)

3.2 (0.6)

0

0.79

Air humidity

 Extremely dry (0)–extremely humid (6)

2.8 (0.8)

2.6 (0.8)

−0.2

2.6 (0.8)

2.5 (0.7)

−0.1

0.66

Air movement (draught)

 No movement (0)–extremely draughty (6)

2.4 (1.1)

2.2 (1.1)

−0.2

2.6 (1.2)

2.6 (0.7)

0

0.25

Air quality

 Extremely poor (0)–extremely good (6)

2.0 (1.0)

1.8 (2.3)

−0.2

2.3 (0.9)

3.2 (1.2)

0.9

0.09

Odour

 No odour (0)–extremely strong odour (6)

2.4 (1.2)

2.2 (1.3)

−0.2

2.9 (1.4)

1.3 (1.2)

−1.6

0.35

Noise from ventilation system

 No disturbing noise (0)–very disturbing (6)

2.4 (1.7)

1.8 (1.4)

−0.6

1.8 (1.3)

1.3 (1.2)

0.5

0.67

Immediate perception of air quality

 Extremely poor (0)–extremely good (6)

2.8 (1.1)

2.8 (1.5)

0

2.6 (1.0)

3.5 (1.1)

1.1

0.02

Headache

 None (0)–unbearable (6)

1.2 (1.7)

1.3 (1.6)

0.1

1.7 (1.4)

0.7 (1.0)

−1.0

0.003

Tiredness

 None (0)–unbearable (6)

1.9 (1.5)

2.0 (1.6)

0.1

1.9 (1.3)

1.1 (1.4)

−0.8

0.007

All questions asked for perceptions/symptoms during the last hour, except immediate perception which covered the first 15 min

aCalculated by Mann–Whitney U test comparing difference in changes in scores (0–6), calculated for each subject, between the two groups

Discussion

Our study indicates that use of a CO2 demand-controlled ventilation system may improve the perception of indoor air quality among university students in computer classrooms and slightly reduce headache and tiredness. The average air exchange was about 10 % higher during lectures when the system was active, and the proportion of time with CO2 levels above 1,000 ppm was reduced, while mean room temperature, relative air humidity, PM10 or number concentration by P-trak were unchanged.

The strength of the study is that it was experimental and blinded to the participants. Moreover, the cross-over design has better power as compared to a study design with one intervention group and one control group. A large number of students participated in the study once, but relatively few participated twice sitting in the same classroom. We did not ask non-participants why they did not come the second week. The lectures were not compulsory, and the main reason for not attending was most probably because they felt they already know the topic enough. One limitation of our cross-over study design was that we only changed exposure conditions once, and the follow-up time was only one week. Selection bias is less likely since most students participated at least once (more than 90 %), and there was no obvious difference in prevalence of personal factors between total participants and participant in the longitudinal analysis. Moreover, the exposure to variable or constant flow conditions could be considered to be randomised, since similar types of students were sitting in all four classrooms. As the students and the teachers had no knowledge of ventilation conditions, recall bias in relation to the exposure is also less likely. As a number of statistical tests were made, some findings could be due to mass significance, but there was a reasonable pattern in the findings, and significance levels were high (p = 0.003–0.007) for the CNS-symptoms. Thus, we do not believe that our overall conclusions are seriously biased by selection or response errors or by-chance findings.

Many students gave low ratings to the intensity of symptoms on the 6-step rating scales. Headache and tiredness were most prevalent; mucous membrane symptoms and other symptoms were less prevalent. Women had no excess of symptoms, the only gender difference was for sinusitis, where men reported a higher prevalence. The lack of gender differences is in agreement with the previous study on technical university students in 2004 (Norbäck and Nordström 2008b). This is in contrast to most other studies on SBS, where women usually report more medical symptoms than men (Stenberg and Wall 1995). One explanation could be that the male and female students had a similar type of education and similar types of work tasks.

The mean PM10 level was 17 μg/m3 during variable flow conditions and 18 μg/m3 at constant flow. This is well below the WHO air quality guideline of 50 μg/m3 (24-h mean) (WHO 2005) and similar as in our previous study in the same computer classrooms (Norbäck and Nordström 2008a) but lower than previous PM10 levels in schools. One study in a university lecture room measured a 12-h mean value of 42 μg/m3 (Branis et al. 2005). In another study on indoor environment in European primary schools, PM10 values in classrooms were in the range of 50–150 μg/m3 (Simoni et al. 2010).

We found high levels of cat (Fel d 1), dog (Can f 1) and horse (Eq c x) allergens in settled dust in all classrooms, but low levels of house dust mite allergens (de f 1). The pet allergen levels were 7–14 times higher than levels previously measured in primary schools in Sweden by the same method (Kim et al. 2005). We found no previous publications on pet allergen levels in university buildings. Since pets are not allowed, neither in Swedish schools nor in Swedish University buildings, the most probably source is track in of allergens from the home environment, public transportation, or other indoor environments, by cloths or hair. Normally, pets are not allowed in the student dormitories, but some students live in own apartments or with their parents. The allergen levels were surprisingly high; one explanation could be that electrostatic forces from the vide display units (VDU) may accumulate allergens in the indoor environment. The VDU units were of the old type (not LCD). Since 16.5 % of the students reported furry pet allergy, the allergen contamination in computer rooms as well as in university buildings in general deserves further attention.

The demand-controlled ventilation with variable ventilation flow slightly reduced headache and tiredness and improved the perception of indoor air quality, despite only small changes in the mean ventilation flow and CO2 levels during lecture time. There was no difference in complaints on draught or noise from the ventilation system between constant and variable flow conditions, and the mean number of students or proportion of women was similar during both conditions. Moreover, the average PM10 or number concentration was the same during constant flow and variable flow conditions. The mean CO2 level was 809 ppm at constant flow and 784 ppm at variable flow conditions. It is surprising that such small changes in mean CO2 levels can have any impact on SBS or environmental perceptions. However, there was a greater difference in the proportion of the time with CO2 levels above 1,000 ppm, from 25 % at constant flow to 9 %, with variable flow. The highest CO2 level was 1,224 ppm. One interpretation of the results is that the proportion of time with CO2 levels above 1,000 ppm, which was reduced during variable flow conditions, may have a significant but numerically small influence on headache, tiredness and the perceived air quality. One previous blinded intervention study in dwellings demonstrated that when the ventilation flow was slightly reduced in winter and mean CO2 levels were slightly increased from 920 to 980 ppm in bedrooms, the air quality was perceived as poorer while headache, tiredness and other medical symptoms were not influenced (Engvall et al. 2005). We have previously demonstrated that changes in CO2 levels from 1,185 ppm to 922 ppm in the same classrooms, using two different constant flow conditions, reduced many SBS symptoms including headache and tiredness (Norbäck and Nordström 2008a) and improved environmental perceptions (Norbäck and Nordström 2008b). Ventilation demands are influenced by the pollution sources in the particular indoor environment. We have no data chemical emissions from indoor sources in the classrooms, except formaldehyde. Computer classrooms may have specific sources of VOC and other chemicals, and computers have been identified as an important source of sensory pollution load (olf) (Bako-Biro et al. 2004).

Room temperature and humidity are important factors that may influence SBS symptoms and perceived air quality (Fang et al. 2004), and compared with the direct impact of temperature and humidity on perception of air quality, sensory emissions from the building materials may have a secondary influence (Fang et al. 1999). The mean room temperature for the four rooms did not differ between week 1 and week 2, but the relative air humidity was on average 12 % lower the second week as compared to the first week, because of lower outdoor temperature. A similar difference in relative air humidity in Sweden in winter may influence the perception of air dryness and SBS symptoms (Nordström et al. 1994). However, we calculated changes in symptoms or perception from week 1 to week 2 for each subject and compared changes between the two groups who had different test sequence. This study design enabled us to control for trends in symptoms reporting or general changes in the indoor climates, such as the relative air humidity, influencing all rooms in a similar way. In addition, since the participants were sitting in the same classroom twice, we adjusted for constant effects of indoor factors in a particular classroom.

The current ventilation standard in Sweden requires a ventilation flow of 7–8 L/s per person and a CO2 level below 1,000 ppm (National Swedish Board of Occupational Safety and Health 2000). The ventilation flow during reference condition (constant flow) was relatively high (mean 13.3 L/s per person), which is higher than the current Swedish ventilation standard. The associations between ventilation flow and medical symptoms or air quality perceptions have been previously reviewed. One early review concluded that about half of the CO2 studies suggest that the risk of SBS continued to decrease with decreasing CO2 below 800 ppm (Seppänen et al. 1999). A recent multidisciplinary review on ventilation and health concluded that higher ventilation rates in offices, up to 25 L/s per person, are associated with reduced prevalence of SBS (Sundell et al. 2011).

There are a few other experimental studies on health effects in pupils or students of changing the ventilation flow in schools. Increasing the personal outdoor air flow rate from 1.3 to 11.5 L/s, through installation in schools of new ventilation systems with displacement ventilation, decreased the risk for asthmatic symptoms in pupils (Smedje and Norbäck 2000). Our previous experimental study in the same computer classrooms, using two constant flow conditions (2.3–2.6 vs. 4.1–5.2 ac/h), had beneficial effects on both medical symptoms (Norbäck and Nordström 2008a) and perceived indoor environment (Norbäck and Nordström 2008b). There are a number of technical scientific articles on energy-saving aspects of demand-controlled ventilation (Fisk and De Almeida 1998; Haghighat and Donnini 1992; Mysen et al. 2003, 2005; Nielsen and Drivsholm 2010; Pavlovas 2004; Roth et al. 2003; Schell and Inthout 2001; Wachenfeldt et al. 2007), mostly based on simulations or theoretical considerations. However, we found no other field study studying effects of demand-controlled ventilation on medical symptoms or environmental perceptions among users of the buildings.

There are two previous technical studies on energy saving in schools by demand-controlled ventilation systems. One study evaluated two demand-controlled systems in Norwegian primary schools. One was based on a CO2 sensor, and the other was an IR occupancy sensor–based system. The CO2 sensor system and the IR sensor system were estimated to reduce the energy use due to ventilation by 38 and 51 %, respectively. (Mysen et al. 2005). Another study from Norway concluded that a CO2-controlled demand control displacement ventilation reduced the total heating energy demand by 21 %, the amount of unrecovered heat in the exhaust ventilation air by 54 % and the average air flow by 50 %. Conventional constant air volume mixing ventilation was used as reference system (Wachenfeldt et al. 2007). Unfortunately, we did not study the energy-saving effects in our study, but it is likely that use of a demand-controlled ventilation system can lead to considerable energy saving in computer classrooms. Since demand-controlled ventilation can reduce energy consumption in buildings and may be more common in the future, more evaluations are needed on the effects of demand control ventilation on human health and indoor perceptions.

In conclusion, use of a CO2 demand-controlled ventilation system in computer classrooms may slightly reduce headache and tiredness and improve perceived air quality, even if the mean levels of CO2 is below current ventilation standards. This indicates that proportion of time with CO2 levels above 1,000 ppm can be important in certain indoor environments. The high levels of pet allergens, possible due to track in of allergens from the home in combination with electrostatics effects from the VDU, are a new finding and illustrate a need for improved cleaning in computer classrooms.

Acknowledgments

This study was partly supported by grants from the Swedish Research Council for Environment, Agriculture and Spatial Planning (FORMAS), the Swedish Foundation for Health Care Sciences and Allergy Research, and the County Council of Uppsala, Sweden.

Conflict of interest

The authors declare that they have no conflict of interest.

Copyright information

© Springer-Verlag 2012

Authors and Affiliations

  1. 1.Department of Medical Science, Occupational and Environmental Medicine, University HospitalUppsala UniversityUppsalaSweden
  2. 2.Department of Environmental Health, School of Public HealthFudan UniversityShanghaiPeople’s Republic of China

Personalised recommendations