You are on page 1of 11

Int Arch Occup Environ Health (2008) 82:21–30

DOI 10.1007/s00420-008-0301-9


Sick building syndrome in relation to air exchange rate,
CO2, room temperature and relative air humidity in university
computer classrooms: an experimental study
Dan Norbäck · Klas Nordström

Received: 10 July 2007 / Accepted: 15 January 2008 / Published online: 2 February 2008
© Springer-Verlag 2008

Objective To study the eVects of ventilation and temperature changes in computer classrooms on symptoms in students.
Methods Technical university students participated in a
blinded study. Two classrooms had higher air exchange
(4.1–5.2 ac/h); two others had lower (2.3–2.6 ac/h) air
exchange. After 1 week, ventilation conditions were interchanged between the rooms. The students reported symptoms during the last hour, on a seven-step rating scale.
Room temperature, relative air humidity (RH) carbon dioxide (CO2), PM10 and ultra-Wne particles (UFP) were measured simultaneously (1 h). Illumination, air velocity,
operative temperature, supply air temperature, formaldehyde, NO2 and O3 were measured. Multiple logistic regression was applied in cross-sectional analysis of the Wrst
answer (N = 355). Those participating twice (N = 121)
were analysed longitudinally.
Results Totally 31% were females, 2.9% smokers and
3.8% had asthma. Mean CO2 was 993 ppm (674–
1,450 ppm), temperature 22.7°C (20–25°C) and RH 24%
(19–35%). Lower and higher air exchange rates corresponded to a personal outdoor airXow of 7 l/s*p and 10–13
L/s*P, respectively. Mean PM10 was 20 g/m3 at lower
and 15 g/m3 at higher ventilation Xow. Ocular, nasal and
throat symptoms, breathlessness, headache and tiredness
were signiWcantly more common at higher CO2 and temperature. After mutual adjustment, ocular (OR = 1.52 per 1°C),

D. Norbäck (&) · K. Nordström
Department of Medical Sciences,
Occupational and Environmental Medicine,
Uppsala University, University Hospital,
751 85 Uppsala, Sweden

nasal (OR = 1.62 per 1°C) and throat symptoms (OR = 1.53
per 1°C), headache (OR = 1.51 per 1°C) and tiredness
(OR = 1.54 per 1°C) were signiWcantly associated with
temperature; headache was associated only with CO2
(OR = 1.19 per 100 ppm CO2). Longitudinal analysis demonstrated that increased room temperature was related to
tiredness (P < 0.05).
Conclusion Computer classrooms may have CO2 above
1,000 ppm and temperatures above 22°C. Increased temperature and CO2 may aVect mucosal membrane symptoms, headaches and tiredness. Room temperature was
most important. CO2 associations may partly be temperature eVects.
Keywords Indoor air quality · Ventilation · Room
temperature · Sick building syndrome (SBS) · University

There are a large number of cross-sectional studies demonstrating associations between various personal factors and
indoor environmental exposures and the prevalence of
mucosal, dermal and general symptoms. These symptoms
are sometimes called the sick building syndrome (SBS)
(Apter et al. 1994; Hodgson 1995). There are considerably
fewer experimental Weld studies on medical symptoms in
relation to changes of indoor environmental parameters
such as temperature (Wyon 1992; Reinikainen and Jaakkola 2001), relative air humidity (RH) (Nordström et al.
1994; Norbäck et al. 2000, 2006; Reinikainen et al. 1991,
1992; Reinikainen and Jaakkola 2001) and ventilation
(Godish and Spengler 1996; Seppänen and Fisk 2004;
Wargocki et al. 2002).


Materials and methods Study design A single blinded cross-over study manipulating the air exchange rate was performed in four computer classrooms at a department at Lund Technical University. but the study was limited to morning and afternoon lectures with full classes. but higher ventilation rates may not be so (Godish and Spengler 1996). questionnaire data were collected twice. There are. A second aim was to study eVects of variation of CO2 levels.4 m above the Xoor. . Classrooms are densely populated. Kamijima et al. but studies on associations between indoor air quality and health in schools are rare (Daisey et al. Another review (EUROVEN) concluded that a low ventilation rate is associated with bad health eVects and decreased performance in oYces (Wargocki et al. increased RH and reduced CO2 concentration can reduce 123 Int Arch Occup Environ Health (2008) 82:21–30 mucosal. Each classroom had 6–8 supply air tackles placed 2. we found no previous study on indoor environment in computer classrooms. Righi et al. and after the lunch break with new students during afternoon classes. The number of students Xuctuated during the lectures. 2003). 2004). 2007. Wrst during morning classes.5 l/s. In the ventilation standard in Sweden it is recommended that indoor CO2 levels should be below 1. Lindivent. it was concluded that beneWcial health eVects could be achieved by reducing Carbon dioxide (CO2) down to 800 ppm (Seppänen and Fisk 2004). Lund. but this regulation was not active during the experiment since the CO2 levels and room temperature were above the critical levels (800 ppm. Sweden. Besides a study on chemical emissions in a newly built computer classroom in an elementary school (Yura et al. CO2 could be used as an indicator of human emissions. one referred to as ‘lower air-exchange rate’ corresponding to a personal supply ventilation Xow of about 7 l/s. The ventilation Xow could be electronically regulated by changing the frequency of the fan. In another review. one below and one above current ventilation standards.0°C) during normal daytime occupancy of the classrooms.3 to 11. 2002). and Monday and Tuesday the next week (week 2). Neuner and Seidler 2006 ). All computers were turned on during the lectures.000 ppm and that outdoor airXow should be at least 7 l/s per person and additionally 0.35 l/s and m2 Xoor surface (National Swedish Board of Occupational Safety and Health 2000). Data was collected during 2 weeks in February 2004. When the ventilation Xow was increased. One longitudinal school study showed that an increase of the personal outdoor air Xow rate from 1. 2005). using two levels of air exchange. The following hypothesis was tested: Increased air exchange rate. Two levels of air-exchange rate were selected.22 The eVects of building ventilation on humans have been reviewed. The rooms were used by some students outside the lecture time. the inlet area was increased. The ventilation system was programmed to be controlled by sensors measuring CO2 levels and room temperature in the exhaust air. reduced room temperature. there are a few studies on indoor air quality in university buildings (Branis et al. reduced the risk for asthmatic symptoms in pupils (Smedje and Norbäck 2000). The main aim was to study the eVects of an experimental change of the air exchange rate in computer classrooms on medical symptoms among university students. and thus the air velocity of the supply air was kept constant. and a ‘higher airexchange rate’ which was the maximum Xow of the ventilation system (corresponding to 10–13 l/s). usually with 2–3 students per computer. All classrooms had supply–exhaust ventilation with frequency-controlled fans and individual regulation of the supply airXow. Monday and Tuesday 1 week (week 1).000 ppm CO2 and a minimum personal outdoor air Xow of 8 l/s (ASHRAE 1999). The questionnaire were distributed in the end of the lectures. but only a few have dealt with medical symptoms compatible with SBS (Bakke et al. The recall period for symptoms was 1 h. large glass windows can increase the thermal problems during warmer parts of the year. dermal and general symptoms. Godish and Spengler reported that there is limited evidence to suggest that ventilation rate increases up to 10 l/s per person may be eVective in reducing symptoms. A recent review on the school environment concluded that classroom ventilation is typically inadequate. which can lead to both thermal discomfort and perception of poor indoor air quality. room temperature and RH. Each day. with no exams. 2005. when the students had been at least 1 h in the classroom. 2005. The increased use of computers may increase the thermal load. however. but data analysis was restricted to average values for the same period as the recall period for the symptoms. 22. in combination with a change of the inlet areas of the supply air tackle (VAV system. Climate measurements were performed during the whole day. There was one exhaust air tackle in each room situated in one corner of the ceiling. 2002). To our knowledge. through installation of new ventilation systems with displacement ventilation. and computers have been identiWed as an important source of sensory pollution load (olf) (Bako-Biro et al. very few epidemiological studies on health eVects of ventilation in schools or university buildings. Sweden). The study was a single-blinded intervention study. Moreover. The ASHRAE ventilation standard has similar requirements of maximum 1. There were normal and similar types of lectures both weeks.

The total air Xow in the main supply air ducts in each room was measured by tracer gas technique on Tuesday afternoon both weeks.02–1 m.5 l/min.000 to 8.900 W). and Dust. 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 Tuesdays.080 W from illumination and 2. very strong symptom(s) was coded ‘5’ and unbearable symptom(s) was coded ‘6’. in southern Sweden. sinusitis.400–3. In addition. was determined by the CAMNEA method (Palmgren et al.Int Arch Occup Environ Health (2008) 82:21–30 All students (age 20–25 years) who were in the morning or afternoon classes on Monday and Tuesday the Wrst week were invited to participate.200 W from the students. throat symptoms. USA). very slight symptom(s) was coded ‘1’. In addition. leading to a total heat load ranging from 6. The concentration of N2O in the supply air was measured by SensAir 2000 1 l (Sense Air. Illumination at the students’ work sites were measured using a Hagner Ec1x lux meter (Hagner AB. respectively. Each classroom had 16–24 old-model com- 23 puters with VDU display with vacuum tubes. measuring the Xow rate of N2O using the mass Xow meter Bronkhorst High-Tech BV (Bronkhorst High Tech BV. The air exchange rate was calculated by dividing the total supply air Xow in the room with the room volume. large windows. During this week. dermal symptoms. The same individual instrument was used in the same classroom. An absence of symptoms was coded ‘0’. measuring particles from »1 to 10 m (PM10).4 m (1. Sweden). irrespective of the week. USA).500 W (mean 6. at diVerent points of the supply air duct. nausea. there were questions on gender.400–3. Particles were measured using both P-Trak™ (Model 8525 UltraWne Particle Counter). Supply air temperature was measured by Digital Thermometer TES 1316 (Instrumental Sales and Rental. contact lens wearing. TSI Incorporated. During the second week. The detection limit for viable organisms was 30 colony-forming units (cfu) per m3 of air. Room temperature. asking for the degree of symptoms and requiring answers on a six-graded scale. 4-h sampling time). The total concentration of airborne moulds and bacteria. Viable moulds and bacteria were determined by incubation on two diVerent media. ocular symptoms. adapted from a previous study on cabin environment in aircraft (Norbäck et al. Canada). headache and tiredness. asking them to rate diVerent medical symptoms during the last hour. The questionnaire contained 11 symptomrelated questions. The instruments were placed on separate desks at the same heights as the breathing zone of the students. C) had lower air exchange rate and two (A. This was done by adding a known Xow of N2O in the supply air system. situated away from major streets.Trackt™ (Model 8520. was used in a cross-sectional analysis. in the city of Lund. The estimated daytime heat eVect per room was 2. Airborne micro-organisms were sampled on 25 mm nucleopore Wlters with a pore size of 0. smoking. the B and C classrooms had higher air exchange rate and the A and D had lower air exchange rate. One question on ear problems and another on musculoskeletal symptoms were not analysed here. 2006). furry pet allergy and food allergy/intolerance.600 W from computers (mean eVect 150 W/computer). but some did not follow this instruction. All had a linoleum Xoor. equivalent temperature and air velocity were measured by a Brule & Kjaer 1213 thermal comfort instrument at computer work stations near the windows in each room. The students participating during the Wrst week were asked to sit in the same classrooms the week after. The Wrst answered questionnaire from each participant. which was normal since they could not be opened. ON. Delsbo. but only data from the same time period as the recall period of the questionnaire (1-h mean values) were used in the calculations. during two consecutive weeks. doctor-diagnosed asthma. strong symptom(s) was coded ‘4’. slight symptom(s) was coded ‘2’. painted walls and painted wood Wbre acoustic absorbents in the ceiling. Solna. The Netherlands). RH and CO2 were measured using a Q-Trackt™ IAQ Monitor (TSI Incorporated. Sweden). Classroom energy and ventilation characterisation The classrooms were on the third Xoor of a Wve-Xoor brick building in the 1960s. and measuring the N2O concentration in the end of the supply ducts. hay fever. Measurements were taken during the whole day. All windows were closed during the experiment. The supply air Xow was measured by a tracer gas dilution method. two of the classrooms (B. 720–1. Data from a subpopulation of students participating during both weeks were used in a longitudinal analysis. supplying a constant Xow of N2O in each supply air duct. Formaldehyde was sampled during 1 week by another diVusion sampler (Levin 123 . measuring particles in the size range 0. room temperature. and all students in the rooms were invited to participate during the second week. The questions included nasal symptoms. moderate symptom(s) was coded ‘3’. both weeks. Ruurlo. dyspnea. 1986). D) had higher air exchange rate The study was repeated on Monday and Tuesday the next week. Assessment of personal factors and perceived indoor air quality The students received a questionnaire at the end of the lecture. sampling one minute average intervals. The instruments were calibrated by the Swedish service laboratory for TSI equipment.

Last the group diVerences in change of each score were Table 1 Prevalence of personal factors among the participants in diVerent data sets Cross-sectional analysis supply ventilation Xow a Female Current smoking Ex-smoker Longitudinal data analysis (N = 121) (%)b Total material (N = 355) (%) Lower (N = 186) (%) Higher (N = 169) (%) 33. RH and air exchange rate was calculated using Kendall’s tau beta.5 21. correlation between symptom scores and CO2. Only 121 participated twice.0 7.7 5. controlling for sex.8 29. followed by mutual adjustment.5 6. mean CO2 was 263 ppm lower and mean air exchange was 1. A total of 355 students participated during either week one or week two. otherwise they did not diVer from the total material.4 13.1 1. one-third were females. room temperature.3 19. irrespectively of week Longitudinal analysis is based on subjects participating twice Lower = 6.3 4.8 21. 230 answered the Wrst questionnaire at week1 and 115 answered the Wrst questionnaire at week two. very slight or slight symptoms as 0.8 8. very strong. Multiple logistic regression was applied to study associations between symptoms and measured indoor environment. and no symptom. a cross-sectional analysis was performed. A longitudinal study was performed in those students participating twice. current smoking. As a next step the symptom scales were dichotomised. 33 had increased ventilation. Results Among those present in the classrooms during the study period.7 14. including the Wrst questionnaire answer by any student during week 1 or week 2.9 5.0 19. except a higher prevalence of current smokers among those with a higher Xow. Two-tailed tests and a Wve percent level of signiWcance were used. and active movement of air and other artefacts is minimised.9 l/s and person a b 123 . more than 90% agreed to answer the questionnaire.0 Food allergy/intolerance 9. During higher air-exchange conditions. 1988). coding moderate.5–7. and 38 had unchanged conditions because they did not sit in the same room both weeks. Among the 355 participants. Those participating twice had somewhat more hay fever and less furry pet allergy.9 30.1 2.1°C lower. Female students rated ocular symptoms and headache higher. RH and air exchange rate in the models. as compared to lower Statistical analysis Initially.7 19. Mann–Whitney U-test was used to compare diVerences in symptom scores between males and females.1–12. the theoretical sampling rate can be used to calculate the pollutant concentrations.0 l/s and person Higher = 10.1 Contac lens wearer 20. room temperature.24 Int Arch Occup Environ Health (2008) 82:21–30 tested by Mann–Whitney U-test. There were no obvious diVerences in prevalence of personal factors between those answering the questionnaire during lower and higher air-exchange rate. Otherwise there were no gender diVerences in symptom rating. We used a ‘badge type’ sampler fully based on the theory for diVusion sampling (Ferm and Svanberg 1998. For each perception the diVerence in the symptom score was calculated by subtracting the Wrst week score from that of the second.6 8. keeping CO2.3 2. with additional control for contact lens wearing for ocular symptoms only. the mean temperature was 1.8 31.1 6. a few had asthma or were smokers and one-Wfth (23%) had either pollen or furry pet allergy (atopy) (Table 1). et al. With this type of sampler. Ferm 2001).4 g NO2 m3 and 4 g O3 m3. Initially each exposure variable was introduced separately. 50 of them had decreased ventilation from week one to week two.38 ac/h higher. strong. The main symptoms were headache and tiredness. asthma.4 3. In a crude analysis.8 Cross-sectional analysis is based on the Wrst answered questionnaire.6 3. mean RH (24%) was the same at both conditions. one-Wfth wore contact lenses.7 18.1 Fury pet allergy 14. while male students rated breathing diYculties and sinusitis symptoms higher (Table 2).5 7. The lower detection limits for 1-week sampling are 0.6 4. Nitrogen dioxide and ozone were sampled by diVusion sampling for 7 days.9 Hay fever 16. and unbearable symptoms as 1. and was analysed by high-pressure liquid chromatography.3 Doctor’s diagnosed asthma 4. hay fever and furry pet allergy.

0°C lower in room B and 1.2(1. corresponding to a change of the personal air Xow by 3 l/s. 19. There were no associations between RH and any type of symptoms (Table 5). No symptom. very strongly. 1. Alternaria sp.5°C in room C. As a next step.470 pt/cm3 (range 450–2.5 Tiredness 3. When introducing each exposure factor separately. the supply air temperature increased to 16.8) 42.5 M(SD) Total prevalence (%) 16.5°C in room B and 18.16.890) during higher air exchange and 1.6(1.04 1.4(0. 95% CI 1. very strongly = 5. the symptom scales were dichotomised and multiple logistic regression analysis was applied.5(1. current smoking. RH.0(1.8) 24.7 Nausea 1.8) 1.3(0.2(0.4(1.2) 8.1(1. There were no signiWcant correlations between RH and any symptom score (Table 4). very little or slightly symptom coded as 0 P-values by Mann–Whitney U-test air-exchange conditions (Table 3).5) 2. slightly = 2.2(1.7(1. The mean concentration of viable bacteria (77–108 cfu/m3) and viable moulds (74–84 cfu/m3) were similar during both conditions.0) NS 1.5(2. Airexchange rate was negatively correlated with breathing diYculties and dermal symptoms.0) NS 1.08 m/s in room B and 0.510) during lower rate.9(1. asthma. 0. Since contact-lens wearing was related to eye symptoms (adj OR 2.1(1. and dermal symptoms were associated with CO2 levels only.2) 2.000 per m3).1(1.5°C in room A.4(1. 654 lux (range 485–777) in room B. additional controlling for this factor was done when analysing eye symptoms.4) 1. breathlessness.3) 1. strongly = 4. P = 0.4) 17.0°C lower in room D.Int Arch Occup Environ Health (2008) 82:21–30 25 Table 2 Mean rating of diVerent symptom ratings and prevalence of symptoms (cross-sectional analysis) Type of symptom Males (N = 240) Females (N = 115) P-value Total (N = 355) M(SD) M(SD) Eye symptoms 1. 1.3(0.30.3) 2.5 Dermal symptoms 1. moderately = 3. The operative temperature near the windows was 0. The mean illumination level was 673 lux (range 414–825) in room A. unbearable = 6) Moderately.7 Sinusitis 1.9) 1. Detected species included Cladosporium sp. The air velocity near the windows was 0.7 Symptoms ratings was judged from 0 to 6 (None = 0.0(1. For 123 . while the concentration of total moulds was below the detection limit (<10. The 1-h mean concentration of ultra-Wne particles (UFP) was 1. each of the four exposure variables (temperature. nasal and throat symptoms.4) NS 2.10 m/s in room D.4) 16. The average change for signiWcant associations was 0. Supply air temperature was measured at diVerent points of the supply air ducts. 511 lux (range 71–700) in room C and 503 lux (range 150–816) in room D. as compared to the air temperature.7) 0. The mean number of students in the classrooms was 19 during higher-ventilation conditions and 19 during lower-ventilation conditions.8) 3.03). the magnitude of the associations with room temperature was unchanged. strongly.9(1.4(0.1°C lower in room D. there was a signiWcant positive correlation between higher room temperature and lower CO2 levels and all nine types of symptom scores. After mutual adjustment (Model II). The mean NO2 concentration measured during 1 week was 11 g/m3 indoors and 10 g/m3 outdoors. 17.8) 3. 0.04 2.8°C lower in room A.4(1.6(1.4(3. hay fever and furry pet allergy (Model I). headache and tiredness.5(1. Operative temperature and air velocity was measured the Wrst week only.6 Headache 2. and unbearable symptoms coded as 1. Penicillium spp and yeast.13 m/s in room A.000 per m3.8) 0.4 points) for a change of air exchange by 1 ac/h.5°C after the aircooling units.5) 0. air exchange) was introduced separately into the models.1(1. but due to a heat exchange eVect between supply and exhaust air ducts.7) NS 3. There was a large day-to-day variation of outdoor UFP concentrations (data not shown). both CO2 level and room temperature were positively associated with ocular.9) 4. and the concentration of O3 was 22 g/m3 indoors and 68 g/m3 outdoors. The average change on the scales was similar (0. very slight = 1.2–0. The supply air temperature was 12. CO2.5°C in room D. In the crude analysis.510 pt/cm3 (range 440–2.8 Throat symptoms 2.03 1. The indoor concentration of formaldehyde measured during 1 week was low (2–4 g/m3) in all rooms.4 points on the six-point scale for a change of 1°C room temperature.1) 1. The 1-h mean PM10 was 15 g/m3 (range 6–24) at higher air exchange and 20 g/m3 (range 16–23) at lower air-exchange rate.0) 0. The total concentration of total bacteria was 15.3(1.4 Breathing diYculties 1.4) Nasal symptoms 2. Breathing diYculties were negatively associated with air exchange rate. Initially. but with large Xuctuations from 10 to 40 students during the lectures.10 m/s in room C.09–4.0) 6.3(1. Sinusitis was associated with room temperature only. controlling for sex.4) NS 2. but the associations were no longer statistically signiWcant for breathing diYculties and sinusitis.04 2.

350) Week 2 1. using Pearson correlation coeYcient. the negative association RH and tiredness became statistically signiWcant.10* 0.03 ¡0. 123 P = 0.13*** 0.01 ¡0.11*** 0. There was no signiWcant correlation between RH and either room temperature or air exchange rate.0 6.0(22–24) Week 1 23(22–24) 25(22–28) 23(22–24) 21(19–25) Week 2 24(23–25) 32(31–35) 24(23–25) 27(25–30) Week 1 910(750–1.07 5.56 2.36 3. except for headache. and negatively correlated with air exchange rate (R = ¡0.0(23–25) Week 2 21. similar associations were found as in Model II when adjusting for all four exposure factors (data not shown).64.04 ¡0.13** 0.001).040(900–1. P < 0.06 ¡0.21*** ¡0.11** 0.3(23–25) 24. CO2 was positively associated with room temperature (R = 0. After mutual adjustment. P < 0. air exchange rate increased from week 1 to week 2 Carbon dioxide (ppm) Table 4 Crude analysis of associations between indoor climate and symptom rating scores (N = 355) (cross-sectional analysis) Type of perception CO2 (ppm) Temperature (°C) RH (%) Air exchange (ac/h) Eye symptoms 0. after mutual adjustment.190(940–1.9 6.6 10.001) and RH (R = 0.16 2.460) 830(800–900) 930(840–1.13.001 CO2.0(21–21) 23.26 Week 2 3.15*** 0.01).04 Nausea 0.0(21–23) 22.8 12.05 **P < 0. air exchange rate decreased from week 1 to week 2 In room B and C.8(20–23) 22.18*** 0.09* Dermal symptoms 0. Correlation between the four exposure factors was calculated on individual level. . When analysing the logistic regression models with mutual adjustment for only CO2 and room temperature (excluding RH and air exchange rate).57.02 ¡0.15*** 0.5 11.001).180(950–1.210(1.09* 0.300) 1.11* 0.17*** ¡0. Finally.11** 0.08 Nasal symptoms 0.02 ¡0.14*** 0.03 Symptom scores were coded from 0 to 6 The values are correlation coeYcients (Kendals-Tau beta) *P < 0.8(22–24) 21.19*** 0.02 Headache 0.3(22–23) 24.26 Table 3 Indoor climate and ventilation in the four computer classrooms during week 1 and week 2 Int Arch Occup Environ Health (2008) 82:21–30 Exposure factor Decreased air exchange Increased air exchange A D B C M(min–max) M(min–max) M(min–max) M(min–max) 162 171 207 207 Week 1 17(10–30) 19(11–29) 21(11–30) 19(11–29) Week 2 22(13–31) 22(15–40) 19(11–25) 20(14–25) Room volume (m3) Number of students Air exchange rate(ac/h) Week 1 4.8 Week 2 6.250) 1.030) 1.07 Breathing diYculties 0.19*** Sinusitis 0.9 7.06 ¡0.030) Temperature (°C) Relative humidity (%) In room A and D.000–1. there were signiWcant positive associations between air exchange rate and both headache and tiredness (Table 5).03 Throat symptoms 0.53 Supply ventilation Xow (l/s*p) Week 1 10.11** 0.83 3.01 ¡0.12** 0.01 ¡0.1 Week 1 22. P < 0.07 Tiredness 0.01 ***P < 0. Room temperature was negatively associated with air exchange rate (R = ¡0. the magnitude of the associations was reduced and.03 3.12** ¡0. was no longer statistically signiWcant.61.300) 850(680–1.

19(1. the exposure to low or high air-exchange could be considered as randomised.19–1.57(0.200 to 920 ppm. while those who received increased air Xow had an average decrease of CO2 from 1. for headache.72–1.49–2.91) 0.49–2.16–1.26) 1. irrespective of ventilation conditions.64(0.18(1.19(0.78(0.04–1.99–2.48(1.32–0.69(0.37) 1.51(0.03–3.06–2.67) 0.51(1. When calculating the association between change of symptoms score and change of ventilation. and 1 ac/h Finally.61) 1.03) Model I 1.53(1.14(0.05) Model II 1.14(1.72(0. with respect to change of airexchange rate.88)** 1.55(1.19* 0.79–1.38(0. Lastly we analysed changes of medical symptom scores in relation to change of CO2.23)* 0.24–1.97–2.24–1.52(0.29–1.76) 0.16.55–1.02) 0.77)*** 0.15(0. and one exposure variable (additional control of contact lens wearing for eye symptoms only). furry pet allergy.11–2. strongly.33)* 1.18–1.55–2.93(1.18(0.17) 1.16) 0.06–1.17)* 1.03–4.05–1.37) 1.35(0.99)* 1.84–1.21–2.52) 1. Numerically.39) 1. room temperature and RH among the 121 students who participated twice.94(0.12–2.00) 0.31)** 1. 10% RH.34) 1.g.12–2.45(0.00(0.35(0.Int Arch Occup Environ Health (2008) 82:21–30 27 Table 5 Associations between indoor climate and symptom rating (N = 355).56) 1.16–3.56(1. Adjusted OR with 95% CI calculated per change of 100 ppm CO2. Increase of room temperature was associated with an increased tiredness score (Kendall’s tau beta 0.87) Model I 1.13–2. e.79)*** 0.83)** Model II 1.15(0.46) 0.09(0.02)*** 0.59–2.09) 0.86) 1. irrespective of which room they were sitting in.00) Moderately.47(1.49(0.01–1.12) Model I 1.00–2.33–1.07–1.19–1.35–2.00)** 0.08–2. Students who received reduced air Xow during the study period had an average increase of CO2 from 1. There was a decrease of symptoms in all three groups.85) 0.75(0.61) 1.88–2.86) 0.11–1.05)** 0.27(0.20–0.90–1.00–1.003–1.43(0.69) 1.51(1. Moreover.170 ppm. very strongly. CO2.71(0.29(0.49((0.05(0.94)*** 1.16) Model II 1. 1°C temperature.17–1.83(0. hay fever.030 to 1.33(1.32–1. since similar types of students were sitting in 123 .77)*** 0. and variation of the other environmental factors was not controlled experimentally.02).44) 0.52(1.07) Model I 1.73) 1.76(0.85–4. the study had good statistical power.75–2.76(0.98–1.77) 1.81–1.95(0.72)* 1. current. P = 0.03) Model II 1. The strength of the study is that it was experimental.83(1.47(0.60–3.71–1. Selection bias is less likely since the participation rate was high (more than 90%).41)* 1.19(1.68) Model II 1.13) Model II 1.53) 1.55(0.91–1.00–1.83)*** 1.43(1. and unbearable symptoms coded as 1.62(1.67–4.60–2.90(0.31–1.50(0.20) 1.55) 1.43–1.18–2.39)** 1. none of the diVerence reached statistical signiWcance.09) 0.62(1.60–1.08–2.13) 1.49(1. Associations between CO2 and symptoms can partly be explained as a temperature eVect.25–6.30(0.45(0.43) Model I 1.52(0.81) Model I 1.21(1. there was a trend to a higher proportion of students with decreased symptom score in relation to increased air exchange for some symptom.30–1. RH and air exchange rate. Due to the large number of participants in the cross-sectional part. and the prevalence of subjects with decreased and increased symptom score was calculated (Table 6).15)** Model I 1. very little or slightly symptom coded as 0 Model I: control for gender.54(1. The association between air exchange and medical symptoms was less consistent.91–2.77(0.08) Model II 1. while room temperature seemed to be the most important climate factor. Model II (control for same confounders as in model I but with mutual adjustment for CO2.33)** Model I 1.21(1.64–1.10(0. comparing those who received decreased and increased ventilation.61–1.44(1. change symptom score during the study period was calculated for those participating twice.82(0.22(0.04–3. The disadvantage is that the experiment only included manipulation of one factor (air exchange rate).14–3. temperature.09–2.34)** 1. blinded to the participants and included both cross-sectional and a longitudinal analyses. The students were divided in three groups. and there were no obvious diVerence in prevalence of personal factors between participants in the cross-sectional and the longitudinal analyses.12(0.19(0.05–1. Discussion We were able to demonstrate that there were fewer medical symptoms among university students in computer classrooms at lower room temperature and lower CO2 levels.26–8. smoking.21(1.40)* 1. No symptom.23) Model II 1.34) Model II 1.32(1. by multiple logistic regression (mode II: mutual adjustment) Type of symptoms Eye symptoms Nasal symptoms Throat symptoms Breathing diYculties Sinusitis symptoms Dermal symptoms Headache Tiredness Nausea CO2 (ppm) Temperature (°C) RH (%) Air exchange rate (ac/h) OR(95%CI) OR(95%CI) OR(95%CI) OR(95%CI) Model I 1.40)** 1.22(0.47) 0.73) 1.74) 0. asthma.30(0.16(1.48–3.88–1.16–1.

PM10-values in classrooms were in the range of 50–150 g/m3 (Simoni et al. which is in contrast to most other studies on SBS. possibly because contact lenses may disturb the ocular tear Wlm. NO2 and O3 were relatively low in the computer classrooms. As a number of statistical tests were made. and less than previous PM10 levels in schools. In another study on indoor environment in European primary schools. mucous membrane symptoms were less prevalent. The indoor levels of formaldehyde. This could be explained as recall bias related to expectations of a general environmental improvement in the building. especially for temperature and CO2. some Wndings could be due to mass signiWcance. The 123 prevalence of contact-lens wearing was relatively high (20%). where females usually report more medical symptoms than men (Stenberg and Wall 1995). Moreover. One study in a university lecture room measured a 12-h mean value of 42 g/m3 (Branis et al. compared to WHO air quality guidelines for formaldehyde (100 g/m3. CO2 concentration in particular. where no signiWcant association was found between contact lenses and ocular symptoms. but in addition. There was no general trend of more symptoms in females. and thus had unchanged ventilation conditions. The study was an experimental study with respect to change of air-exchange rate. The mean PM10 level was 15 g/m3 during increased airexchange conditions and 20 g/m3 during normal conditions. and we found an association between contact lenses and eye symptoms. but since the study was blinded. This is well below the WHO air quality guideline of 50 g/m3 (24-h mean). we evaluated associations between CO2. such bias is less likely to have inXuenced the observed associations between symptoms and measured environmental factors.01 for many Wndings. Many students gave low ratings to the intensity of symptoms on the six-step rating scales. but also by crowdedness (number of students). As the students and the teachers had no previous knowledge of ventilation conditions (high or normal air exchange rate). There were . 30 min value) (WHO 1987). we do not believe that our overall conclusions are seriously biased by selection or response errors or by chance Wndings. Our data are in contrast to a previous study of civil aviation pilots. One explanation could be that the male and female students had a similar type of education and similar types of work tasks. but there was a consistent pattern in the Wndings using diVerent statistical models. in the follow up study (N = 121) Int Arch Occup Environ Health (2008) 82:21–30 Type of symptom Eye symptoms Nasal symptoms Change of ventilation conditions from week 1 to week 2 Decreased (N = 50) (%) Unchanged (N = 38) (%) Increased (N = 33) (%) Decreased 39 54 35 Increased 13 13 22 Decreased 42 40 35 Increased 10 11 22 Decreased 19 46 41 Increased 20 11 16 Breathing diYculties Decreased 29 37 33 Increased 16 10 12 Sinusitis Decreased 19 20 22 Increased 7 6 17 Dermal symptoms Decreased 10 20 22 Increased 16 14 13 Throat symptoms Headache Totally 38 did not sit in the same room twice. NO2 (40 g/m3. statistical signiWcance levels were below 0. No diVerences were statistically signiWcant Symptom change Tiredness Nausea Decreased 35 36 47 Increased 26 15 16 Decreased 48 49 59 Increased 16 20 14 Decreased 33 20 27 Increased 10 11 10 all four classrooms. annual mean) (WHO 2005) and O3 (100 g/m3. 2006). Among these factors. room temperature and RH and symptoms. Headache and tiredness were most prevalent.28 Table 6 Percentage of subjects with decreased and increased symptom rating after intervention. recall bias in relation to the exposure is also less likely. Thus. irrespective of change of ventilation conditions. 2006). is not only partly inXuenced by air-exchange rate. 8-h value) (WHO 2005). 2005). There was a general trend of reduced symptom-reporting during the study period. despite the very low RH in aircraft (Lindgren et al. DiVerent symptoms were tested statistically.

PM(2. Nakajima T (2005) Indoor air pollution due to 2-ethyl-1-hexanol airborne concentrations. (1999) Ventilation for Acceptable Indoor Air Quality. Sidman J. Indoor Built Environ 9:28–34 Nordström K. Lindahl R. Takeuchi Y. mainly related to access to a free computer. After mutual adjustment. Bracker A. Seppänen and Fisk 2004. Aviat Space Environ Med 77:832–837 Levin J-O. we could not demonstrate any signiWcant eVect from change of air exchange rate. Indoor Air 14:178–187 Branis M. Ohno H. Wargocki et al. Wålinder R. Wieslander G (2006) Changes of ocular and nasal signs and symptoms in air crew. pp 298–303 Godish T. Wieslander G. emission sources and subjective symptoms in classroom users. GA. not controlled by the researchers. it is important to keep CO2-levels below current standard (1. while the associations between CO2 levels and symptoms were reduced and mostly no longer statistically signiWcant. Nippon Koshu Eisei Zasshi 52:1021–1031 (in Japanese with English abstract) Lindgren T.3 to 11. Wargocki P. Our results are in agreement with conclusions from previous review articles on human response to building ventilation (Godish and Spengler 1996. independent eVects of room temperature could be demonstrated. A six week longitudinal study. Norbäck D. due to the combined heat eVect of both students and electric equipment. In conclusion. Health (2000) The design of the workplace AFS 2000. Shibata E. Int J Hyg Environ Health 209:367–375 Norbäck D. Norbäck D (2007) Gender and the physical and psychosocial work environments are related to indoor air symptoms. Blomqvist G. In the longitudinal analysis. Sakai K. J Appl Bacteriol 61:401–406 Reinikainen LM. sun shields or suYciently high ventilation Xow. Our study indicates that room temperature might be the most important indoor climate factor in computer classrooms. Heinonen OP (1991) The eVect of air humidiWcation on diVerent symptoms in oYce workers—an epidemiological study. Apte MG (2003) Indoor air quality. Ström G. through installation in schools of new ventilation systems with displacement ventilation. American Society for Heating. Norbäck D. related to air humidiWcation on intercontinental Xights. France. Wieslander G. and variations in sun radiation. Atlanta. decreased the risk for asthmatic symptoms in pupils (Smedje and Norbäck 2000). Acknowledgments This study was partly supported by grants from the Swedish Council for Worklife Research and the Swedish Foundation for Health Care Sciences and Allergy Research. The number of subjects in a room aVects CO2. Room temperature was most important factor. 29 References Apter A. Yamada T. Nordström K. SBS symptoms and productivity in oYces. Angell WJ. 26–28 September 2001.5) and PM(1) in a classroom. associations between air-exchange rate and symptoms were less consistent and depended on the type of model used. Seidel HJ (2006) Adaptation of oYce workers to a new building-impaired well-being as part of the sick-building-syndrome. Malmberg P (1986) Collection of airborne micro-organisms on Nucleopore Wlters. Spengler JD (1996) Relationships between ventilation and indoor air quality: a review. The possible loss of learning ability due to medical symptoms caused by poor indoor air quality in computer classrooms deserves more attention. In the longitudinal analysis. Thermal problems can be severe in computer classrooms. Domasova M (2005) The eVect of outdoor air and indoor human activity on mass concentrations of PM(10). Occup Environ Med 51:683–688 Palmgren U. In the cross-sectional analysis. which can be achieved either by reducing the number of students or increasing the ventilation Xow. Indoor Air 13:53–64 Ferm M. Standard 62-1999. room temperature and RH. Environ Int 17:243–250 123 . There are a few other experimental studies on eVects of ventilation Xow in university buildings or computer classrooms. Indoor Air 6:135–145 Hodgson M (1995) The sick-building syndrome. one previous experimental study has identiWed computers as an important source of sensory pollution load (olf) (Bako-Biro et al. We found that room temperature and CO2 levels were positively associated with diVerent types of symptoms. Refrigerating and Air conditioning Engineers Bakke JV. In: Proceedings of international conference measuring air pollutants by diVusive sampling. tear Wlm stability. Andersson K (1988) High performance liquid chromatographic determination of formaldehyde in indoor air in the ppb to ppm range using diVusive sampling and hydrazone formation.42 (in Swedish) Neuner R. Leung WY (1994) Epidemiology of the sick building syndrome. Rezacova P. Lindgren T. and biomarkers in nasal lavage. variation of outdoor temperature during the day. Atmos Environ 32:1377–1381 Ferm M (2001) Validation of a diVusive sampler for ozone in workplace atmospheres according to EN838. Jaakkola JJK. Weschler CJ. Environ Technol Lett 9:1423–1430 National Swedish Board of Occupational Safety. Hodgson M. Svanberg P-A (1998) Cost-eYcient techniques for urban.000 ppm). Occupational Medicine: State Art Rev 10:167–175 Kamijima M. J Occup Environ Med 49:641–650 Bako-Biro Z.and background measurements of SO2 and NO2. A four months longitudinal study. Scand J Work Environ Health 32:138–144 Norbäck D. Ishihara S. might inXuence room temperature. Andersson K (2006) Perception of the cockpit environment among pilots on commercial aircraft. Fanger PO (2004) EVects of pollution from personal computers on perceived air quality. 2002).5 l/s. suggesting that personal ventilation Xow should be at least 10 l/s and that beneWcial health eVects could be achieved when reducing CO2 levels down to 800 ppm. only change of room temperature was signiWcantly related to any symptom. Venge P (2000) The eVect of air humidiWcation on symptoms and nasal patency. Moen BE. 2004). Environ Res 99:143–149 Daisey JM. ventilation and health symptoms in schools: an analysis of existing information. J Allergy Clin Immunol 94:277–288 ASHRAE. Akselsson R (1994) The eVect of humidiWcation on the Sick Building Syndrome and perceived indoor air quality in hospitals. Increasing the personal outdoor air Xow rate from 1. It is important to control room temperature in computer classrooms through air conditioning. estimation and analysis-CAMNEA method. computer classrooms may have CO2 levels above 1. Moreover. Montpellier. Moreover. Moreover.Int Arch Occup Environ Health (2008) 82:21–30 large Xuctuations of number of students.000 ppm and classroom temperatures above 22°C.

Seppänen O. Wall S (1995) Why do women report “sick building symptoms” more often than men? Soc Sci Med 40:491–502 Wargocki P. Sci Total Environ 286:41–50 Seppänen OA.who. Jaakkola JJK (2001) EVects of temperature and humidiWcation in the oYce environment. Canciani M. Shimizu T (2005) Indoor air pollution in newly built or renovated elementary schools and its eVect on health in children. Norbäck D (2000) New ventilation systems at selected schools in Sweden—EVects on asthma and exposure. Pickering A. Ciccarese V. Eur Respir J 28(suppl 50):837 Smedje G. Sigsgaard T. Seppänen O (1992) The eVect of air humidiWcation on symptoms and perception of indoor air quality in oYce workers: A 6-period cross-over trial. 2005 (http://www. WHO. Geneva. Copenhagen. Fisk WJ (2004) Summary of human responses to ventilation. Arch Environ Health 55:18–25 123 Int Arch Occup Environ Health (2008) 82:21–30 Stenberg B. Indoor Air 14(Suppl 7):102–118 Simoni M. Norback D. Arch Environ Health 56:365–368 Righi E. Wieslander G. Gyntelberg F. Harrison P. Sundell J. 1987 World Health Organization (WHO) WHO air quality guidelines global update 2005. Fantuzzi G. Viegi G. Jaakola JJK. European Series No 23. Nystad W.30 Reinikainen LM. Environ Technol 3:313–322 Yura A. Annesi-Maesano I. Hanssen SO. Iki M. Bischof W. Wouters P (2002) Ventilation and health in non-industrial indoor environments: report from a European multidisciplinary scientiWc consensus meeting (EUROVENT). Fanger PO. Air quality guidelines for Europe. Indoor Air 2:113–128 World Health Organization (WHO) Regional oYce for Europe. Proceedings from 16th ERS Annual Congress. Brundrett G. Predieri G (2002) Air quality and well-being perception in subjects attending university libraries in Modena (Italy). Sestini P (2006) Relationships between school indoor environment and respiratory health in children of Wve European Countries (HESE study). Arch Environ Health 47:8–15 Reinikainen LM. Aggazzotti Wyon D (1992) Sick buildings and the experimental approach. WHO Regional Publications. Nippon Koshu Eisei Zhasshi 52:715–726 (in Japanese with English abstract) .

.Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.