Topic A8: IAQ & perceived air quality
THE EFFECT OF AIR QUALITY ON SLEEP
Peter STRØM-TEJSEN1,*, Pawel WARGOCKI1, David P. WYON1, and Anna
1 International Centre for Indoor Environment and Energy,
Department of Civil Engineering, Technical University of Denmark
2 Department of Heating, Ventilation and Dust Removal Technologies, Faculty of
Environmental Engineering and Energy, Silesian University of Technology, Poland
*Corresponding email: email@example.com
Keywords: Indoor air quality, Sleep quality, Actigraph, Ventilation rate
The effect of air quality on sleep was examined for occupants of 14 identical single-
occupancy dormitory rooms. The subjects, half women, were exposed to two conditions
(open/closed window), each for one week, resulting in night-time average CO2 levels of 660
and 2585 ppm, and air temperatures of 24.7 and 23.9°C, respectively. Sleep was assessed
from movement data recorded on wristwatch-type actigraphs and from online morning
questionnaires, including the Groningen Sleep Quality scale, questions about the sleep
environment, next-day well-being, SBS symptoms, and two tests of mental performance.
Although no significant effects on the sleep quality scale or on next-day performance could be
shown, there were significant and positive effects of a higher ventilation rate (open window)
on the actigraph measured sleep latency and on the subjects’ assessment of the freshness of
the air, their ability to fall asleep and nasal dryness. There was a negative effect on reported
Sleep is essential for a person’s health and well-being. In studies conducted among Finnish
(Tynjälä et al., 1999) and Dutch children (Meijer et al., 2000), a strong correlation between
sleep quality and their concentration the next day was found. Both studies show that good and
refreshing sleep is one of the constituents for general well-being among adolescents. Many
factors are thought to influence sleep quality, among those the indoor environmental quality
(IEQ) parameters which include: air temperature and relative humidity, air velocity,
particulate matter concentration, illumination level, sound level and ventilation rate. The
effect of air quality was investigated in the present study.
Only a few field studies of sleep as a function of indoor air quality have been carried out. A
study of typical Belgian houses by Laverge and Janssens (2011a) estimated that exposure to
poor air quality is up to 16 times higher in the bedroom. Laverge and Janssens (2011b) then
asked 10 students to sleep in their dormitories for one month, in periods with open window
(high ventilation rate) and with closed window (low ventilation rate). CO2 concentration, air
temperature and relative humidity were measured throughout. The sleep pattern of the
subjects was measured using actigraphy, and the subjects completed a questionnaire every
morning to report their sleep quality. As usable data was obtained from only 6 subjects there
were no significant effects of the intervention, although sleep efficiency (proportion of time in
bed spent asleep) tended to be less when windows were open. The objective of the present
study was to investigate the effect of open windows on indoor air quality, sleep quality and
Facilities and subjects
The experiment took place in the Campus Village at the Technical University of Denmark
(DTU), 10 km north of Copenhagen. This housing complex for international students consists
of twenty identical units, each housing up to ten students. All rooms are identical (3.6 m in
length, 3.0 m in width and 2.4 m in height) with one double-framed window located opposite
the door. An air vent placed in the outer wall was sealed during the experiment. Each housing
complex unit includes common toilet, bath and kitchen facilities with mechanical exhaust,
creating a negative pressure in the corridor. The rooms are furnished with a sofa/bed, a
wardrobe, a desk, and sometimes additional private furniture.
Twenty of the occupants participated in the study, but data from 6 subjects was omitted from
the analysis after it was determined that a leaking air seal had affected the physical conditions
to which some of the subjects were exposed and because insufficient data had been obtained
from some other subjects.
The remaining 14 subjects originated from 10 nations, with an equal number of males and
females. Each subject was exposed to two experimental conditions, open and closed window,
in a balanced order of exposure, each condition lasting one week. In the “open window
condition” the one window sash was held open by a 10 cm long plastic window stay. In the
daytime the subjects were allowed to close or open the window according to their preferences.
They were asked to adjust the electric heater below the window to achieve their preferred
thermal condition for sleeping in both conditions, and to maintain their normal life style,
although with restricted alcohol and caffeine consumption. The experiment was performed
from Saturday the 22nd of September to Saturday the 8th of December 2012 with outdoor
temperatures between -7°C and 11°C at night. The experimental condition in the rooms was
changed on Saturdays. Only data from the four nights between Monday and Friday were used
in the subsequent data analysis.
Physical measurements and questionnaires
During the experimental period the air temperature, relative
humidity and CO2 concentration were measured at 5-
minute intervals. Two measuring stations were used, one in
the centre of each side wall. A HOBO U12-012 data logger
was used to measure air temperature and relative humidity
with an accuracy of ± 0.35 K and ± 2.5%, respectively. A
Vaisala GM20 CO2 sensor, calibrated for the range 0-5000
ppm, was used to measure the CO2 levels with an accuracy
of ± (2% of range + 2% of reading).
The activity of the subjects was recorded each minute for
two weeks using an actigraph. The Philips Actiwatch 2
Figure 1. Philips Actiwatch 2.
(Figure 1) is a small, actigraphy data logger designed for clinical and scientific use. It records
a measure of gross motor activity that can be used to visualise rest activity patterns and
quantify physical activity. Actigraphy is a well-established method for field studies of sleep
(Kushida et al., 2001).
As part of a screening process, the subjects were asked to fill in a recruitment questionnaire
and a background questionnaire, which made it possible to exclude people suffering from
asthma, allergy, sensitive skin or sleeping disorders, and people smoking or using medication.
The background questionnaire was based on the Pittsburgh Sleep Quality Index (PSQI)
(Buysse et al., 1989), which contains questions about sleeping habits during the past month.
Every morning the subjects were asked to fill in an on-line questionnaire no later than 10
minutes after waking up. It consisted of questions about sleep quality, sleep environment,
SBS symptoms, sleep symptoms and well-being the previous day. Two on-line performance
tests were applied, a Grammatical Reasoning Test (Baddeley, 1968) and the Tsai-Partington
Numbers Test (Ammons, 1955), the latter with the modifications introduced by Wyon (1969).
The questionnaire included fifteen questions from the Groningen Sleep Quality (GSQ) Scale
(Mulder Hajonides et al., 1980) which were to be marked true or false. It also included visual-
analogue scales rating 7 aspects of the sleep environment, 13 SBS symptoms, 4 aspects of
sleep quality and 2 next-day symptoms. Additional questions were about clothing worn
during sleep, reasons for any awakenings, how many times the subjects woke up or got out of
bed, and what time they went to bed and woke up.
Data processing and statistical analysis
All of the physical measurements of air temperature, relative humidity and CO2 concentration
can be assumed to be Normally distributed, so they were registered as average, minimum,
maximum and standard deviation values. The data from the online morning questionnaire
were tested for Normality using the Shapiro-Wilks Test. Data from the Tsai-Partington
Numbers Test, Baddeley’s Reasoning Test, and the actiwatches were not Normally distributed.
The pair-wise differences between the two conditions were tested using the non-parametric
Wilcoxon Matched-Pair Signed-Ranks Test. The P-values reported in the Results section are
for a 2-tailed test of the difference between conditions of the 4-day mean values.
RESULTS AND DISCUSSION
Physical measurements of the indoor environment
The average night-time air temperature in the 14 rooms was 23.9°C in the closed window
condition and 24.7°C in the open window condition. There were considerable differences
between individual subjects, with a minimum average value of 16.3°C and a maximum of
27.8°C. The temperature was on average 1.6 K higher in the open window condition for 10 of
the 14 subjects, and 1.4 K lower for the remaining 4 subjects. The female subjects were on
average exposed to a 3 K higher temperature than the male subjects.
The night-time average values of relative humidity for each subject were in the range
between 40% to 72% for the closed window condition and 23% to 64% for the open window
condition, all subjects experiencing lower humidity with an open window. The average with
closed window was 54%, and 40% with open window.
Average night-time values of the CO2 concentration in the 14 rooms during the experiment
are shown in Figure 2, documenting the large effect of the intervention. The average values of
CO2 concentration were between 1730 ppm to 3900 ppm for the closed window condition and
525 ppm to 840 ppm when the window was
open. The average CO2 concentration was
2585 ppm in the closed window condition
and 660 ppm in the open window condition,
corresponding to a difference of 8-9 times in
air exchange rate. The CO2 concentration in
an occupied room is a good indicator of the
air change rate, and thus of changes in the
concentration of pollutants originating from
other sources, e.g. building materials. In an
11 m2 single-occupied bedroom, classed in
Europe as a bedroom with low occupant
density, the measured values for the closed
window condition are considered
unacceptable, although they are not unusual.
Studies including 500 Danish children
(Bekö et al., 2010) showed that 57% of the
bedrooms did not fulfil the minimum
ventilation requirements stipulated in
EN 15251 (2007).
Table 1 shows the results from the Wilcoxon Signed-Rank Test providing P-values for two-
tailed tests with statistically significant differences (P<0.05) shown in bold.
Table 1. Results from the statistical analysis of the morning questionnaire with P-values from
the Wilcoxon Signed-Rank Test.
Variable P-value Comments Variable P-value Comments
GSQ SCALE SBS SYMPTOMS
Score 0.1080 Better when open Nasal dryness 0.0480 Less dry when open
Mouth dryness 0.2859
Air humidity 0.7299 Skin dryness 0.0869 More dry when open
Freshness of air
More fresh when open
Eye dryness 0.3003
Air movement 0.0555 More movement when open Eye clearness 0.2634
More dry when open
Illumination 0.7776 Thirst 0.9750
Isolation of cover 0.2455
SLEEP SYMPTOMS Mental state 0.7776
Quality of sleep 0.8753 Alertness 0.9250
Duration of sleep
Lightness of sleep 0.5936 Wellbeing 0.9165
Ability to fall asleep
Better ability when open
Sleepiness 0.0516 Less sleepy when open TSAI-PARTINGTON TEST
Ability to concentrate
Better ability when open
All P-values are 2-tailed. Bold: P<0.05.
Results for some of the variables are presented graphically as quartiles in Figures 3-11. Each
box represents the interquartile range, and the horizontal line dividing the box is the median.
Figure 2. Comparison of the CO
concentration during night-
time, each bar
representing a 4-day average.
The difference between conditions on the Groningen Sleep Quality (GSQ) Scale did not
reach significance (P<0.1078), although there was a tendency for subjects to report sleeping
better, not worse, with the window open (Figure 3). It should be recalled that the (non-
significant) tendency noted by Laverge and Janssens (2011b, op. cit.) was in the opposite
In the Sleep environment section, the results show a significant difference (P<0.0010) for
Freshness of air (Figure 4). The subjects reported the air to be fresher in the open window
condition, as expected, since the experiment took place in a residential area away from
possible sources of outdoor pollution. Air movement was not reported significantly
differently (P<0.0555), although there was an almost significant tendency for the subjects to
report more air movement with the window open (Figure 5). A difference might have been
expected for Noise, but the subjects did not find the environment to be any noisier in the open
window condition, probably because internal noise from other students dominated in a quiet
Figure 3. Score from the
Figure 4. Ratings of
freshness of air.
Figure 5. Ratings of
In the Sleep symptoms section a significant difference was found only for Ability to fall
asleep (P<0.0303). The subjects reported that it was easier to fall asleep in the open window
condition (Figure 6).
The statistical analysis of Next-day symptoms did not show significant differences, but there
was an almost significant tendency for the subjects to feel less sleepy (P<0.0516) and to have
a greater ability to concentrate (P<0.0806) the day after sleeping with open window (Figures
7 and 8).
Figure 6. Ratings of
ability to fall asleep.
Figure 7. Ratings of
Figure 8. Ratings of
ability to concentrate.
From the statistical analysis of the assessments of the 13 SBS symptoms, there was found a
statistically significant difference between the two conditions for Nasal dryness (P<0.0480)
and Lip dryness (P<0.0413), see Figures 9 and 10. The subjects felt their nose to be less dry
but their lips to be drier with the window open. Although not statistically significant
(P<0.0869), there was a tendency for more Skin dryness in the open window condition
(Figure 11). Drier lips and skin in the open window condition may be due to the lower relative
humidity in this condition. The more pronounced nasal dryness in the condition with closed
window could be caused by the higher concentration of pollutants due to the low air exchange
rate. Field investigations by Sundell and Lindvall (1993) concluded that the indoor air
humidity is not an important factor for the sensation of dryness that might be caused by
pollutants in the air. Results from laboratory experiments performed by Fang et al. (2004)
appear to agree that indoor air pollutants may contribute to symptoms that are similar to the
sensation of dryness.
Figure 9. Ratings of nasal
Figure 10. Ratings of lip
Figure 11. Ratings of skin
There were no significant effects of the intervention on Baddeley’s Grammatical Reasoning
Test or the Tsai-Partington Test, and no trends were apparent in the quartile graphs (not
The results from the statistical analysis of
the actigraph data are shown in Table 2,
where sleep duration is the time spent
sleeping, excluding intervening periods
spent awake; sleep latency is the time
required to fall asleep; snooze time is the
time required to become active after
finally awakening; sleep efficiency is the
percentage of time in bed spent asleep.
The results show a significant improvement in Sleep latency (P<0.0480) in the open window
condition, which leads to a positive tendency for Sleep efficiency (P<0.0736) since more time
is left for sleeping. The result for sleep latency is in agreement with the subjects’ own
assessment of their ability to fall asleep. The process of falling asleep occurs as body
metabolism and central body temperatures are both reduced. For the latter to happen, the body
must lose stored heat. Increased air movement and lower RH make this easier to achieve, by
Table 2. P-values obtained from the Wilcoxon
Signed-Rank Test for the actigraph data.
Shorter when open
Efficient when open
No. of awakenings
All P-values are 2-tailed. Bold: P<0.05.
increasing evaporation from exposed skin, especially facial skin. This is a possible
mechanism for the observed and reported decrease in sleep latency.
Having established both subjectively and objectively that sleep latency was reduced by having
the window open, it is reasonable to expect less sleepiness and better ability to concentrate
after a night with better sleep. It is then appropriate to use a 1-tail P-value as an estimate of
the probability of the observed change, in which case a significant improvement was found
both for sleepiness (P<0.0258) and ability to concentrate (P<0.0403) the day after sleeping
with the window open.
The study was performed in the homes of the subjects, in their normal sleeping environment.
This was made possible by using actigraphy and online questionnaires. Although this added to
the realism of the study, the physical parameters of the sleep environment were not controlled
and the subjects were not blinded to the intervention. Confounding occurred because opening
a window affects the relative humidity and the air movement as well as the air quality. This
study was of the effects of a simple intervention, one that can be made in most bedrooms,
namely opening a window.
The subjects were half men, half women, coming from 10 different countries and 3 different
continents, which makes the results more generally applicable. The probability of obtaining
statistically significant results would probably have been increased by using a less diverse
group of subjects and by increasing the number of subjects.
The use of identical student dormitory rooms reduced the number of other confounding
variables such as the outdoor noise level, and indoor and outdoor pollution sources. The
rooms are designed for single occupancy so there was no disturbance from other people in the
room. However, the student lifestyle, with no regular schedule during the week will have
increased individual variation and thus reduced the probability of obtaining significant results.
The physical measurements of the indoor environment showed a marked difference between
the two conditions. The average CO2 concentrations were 2585 ppm in the closed window
condition and 660 ppm in the open window condition, with only minor differences for the
measured temperature, so the objective of the present study was achieved.
The intervention of opening the window had a significant positive effect on the assessed
freshness of air. The lower relative humidity in the open window condition resulted in more
lip and skin dryness, but there was less nasal dryness, possibly because the concentration of
all airborne pollutants was reduced.
The subjectively assessed ability to fall asleep was significantly greater in the open window
condition, a result supported by the actigraphy data. The Groningen Sleep Quality Scale
showed a tendency for the subjects to sleep better with the window open. Subjects reported
feeling less sleepy and being better able to concentrate the day after sleeping with the window
The experiment reported in this paper is part of an on-going study entitled: “Energy-efficient
bedroom ventilation that may improve sleep and next-day well-being” at the International
Centre for Indoor Environment and Energy (ICIEE) of the Technical University of Denmark
(DTU). Financial support for the project is provided by the Danish Agency for Science,
Technology and Innovation (Ministry of Science, Innovation and Higher Education).
Ammons CH (1955) Task for the study of perceptual learning and performance variables.
Perceptual & Motor Skills, 5, 11-1.
Baddeley AD (1968) A 3-min reasoning test based on grammatical transformation.
Psychonomic Science, 10(10), 341-342.
Bekö G, Lund T, Nors F, Toftum J, and Clausen G (2010) Ventilation rates in the bedrooms
of 500 Danish children. Building and Environment, 45(10), 2289-2295.
Buysse DJ, Reynolds III CF, Monk TH, Berman SR, and Kupfer DJ (1989) The Pittsburgh
Sleep Quality Index: A new instrument for psychiatric practice and research. Journal of
Psychiatric Research, 28(2), 193-213.
EN 15251:2007(E). Indoor environmental input parameters for design and assessment of
energy performance of buildings addressing indoor air quality, thermal environment,
lighting and acoustics. CEN, Brussels.
Fang L, Wyon DP, Clausen G, and Fanger PO (2004) Impact of indoor air temperatures and
humidity in an office on perceived air quality, SBS symptoms and performance. Indoor
Air, 14(Suppl. 7), 74-81.
Kushida CA, Chang A, Gadkary C, Guilleminault C, Carrillo O, and Dement WC (2001)
Comparison of actigraphic, polysomnographic and subjective assessment of sleep
parameters in sleep-disordered patients. Sleep Medicine, 2(5), 389-396.
Laverge J and Janssens A (2011a) IAQ exposure of sleeping occupants under different
residential ventilation configurations. In: Proceedings of the 12th International Conference
on Air Distribution in Rooms - Roomvent 2011, Trondheim, Paper 125.
Laverge J and Janssens A (2011b) Analysis of the influence of ventilation rate on sleep
pattern. In: Proceedings of the 12th International Conference on Indoor Air Quality and
Climate - Indoor Air 2011, Austin, Texas, Paper a51-3.
Meijer AM, Habekothé HT, and van den Wittenboer GL (2000) Time in bed, quality of sleep
and school functioning of children. Journal of Sleep Research, 9(2), 145-153.
Mulder Hajonides van der Meulen WREH, Wijnberg JR, Hollander JJ, De Diana IPF, and van
den Hoofdakker RH (1980) Measurement of subjective sleep quality. European Sleep
Research Society Abstracts, 5, 98.
Sundell J and Lindvall T (1993) Indoor air humidity and sensation of dryness as risk
indicators of SBS. Indoor Air, 3(4), 382-290.
Tynjälä J, Kannas L, Levälahti E, and Välimaa R (1999) Perceived sleep quality and its
precursors in adolescents. Health Promotion International, 14(2), 155-166.
Wyon DP (1969) The effects of moderate heat stress on the mental performance of children.
National Swedish Institute for Building Research, Document D8/69, 83p.