A field intervention study of the effects of window and door opening on bedroom IAQ, sleep quality, and next-day cognitive performance

Abstract

Indoor Air Quality (IAQ) and sleep quality measurements over a period of two weeks were performed all night in 40 bedrooms in Denmark during the heating season. In the first week, the bedroom conditions were typical of what participants would normally experience during sleep. In the second week, the participants were asked to open the doors or windows if they had been closed or the opposite. A change in the 95th percentile of the measured CO2 concentration by more than 200 ppm in the expected direction on the same weekdays of the two-week measurement period was taken to indicate that an effective intervention had taken place. The measurements in the 29 bedrooms that met this criterion were grouped depending on how the windows or doors had been manipulated. Objectively measured and subjectively rated bedroom IAQ improved when the windows were open except that the NO2 concentration was slightly higher. Sleep was longer under this condition and sleep quality was subjectively assessed to be better. Similar effects were not observed when the doors were open although the 95th percentile of CO2 concentration decreased by as much as when the windows were open. No effects were seen in the 11 bedrooms in which the change to the bedroom conditions made by the participants did not change the CO2 concentration by at least 200 ppm, as would be expected. The present study provides evidence that sufficient dilution and/or removal of pollutants is necessary to ensure good bedroom IAQ and good sleep quality.

Full text

Building and Environment 225 (2022) 109630
Available online 28 September 2022
0360-1323/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
A eld intervention study of the effects of window and door opening on
bedroom IAQ, sleep quality, and next-day cognitive performance
Xiaojun Fan
a
,
*
, Chenxi Liao
b
, Mariya P. Bivolarova
a
, Chandra Sekhar
c
, Jelle Laverge
b
, Li Lan
d
,
Anna Mainka
e
, Mizuho Akimoto
f
, Pawel Wargocki
a
a
International Centre for Indoor Environment and Energy, Department of Environmental and Resource Engineering (DTU Sustain), Technical University of Denmark,
Denmark
b
Research Group Building Physics, Construction, and Climate Control, Department of Architecture and Urban Planning, Ghent University, Belgium
c
Department of the Built Environment, National University of Singapore, Singapore
d
Department of Architecture, School of Design, Shanghai Jiao Tong University, China
e
Department of Air Protection, Faculty of Energy and Environmental Engineering, Silesian University of Technology, Poland
f
Department of Architecture, Waseda University, Japan
ARTICLE INFO
Keywords:
Bedroom ventilation
Intervention
Air quality
Sleep quality
Next-day cognitive performance
ABSTRACT
Indoor Air Quality (IAQ) and sleep quality measurements over a period of two weeks were performed all night in
40 bedrooms in Denmark during the heating season. In the rst week, the bedroom conditions were typical of
what participants would normally experience during sleep. In the second week, the participants were asked to
open the doors or windows if they had been closed or the opposite. A change in the 95
th
percentile of the
measured CO
2
concentration by more than 200 ppm in the expected direction on the same weekdays of the two-
week measurement period was taken to indicate that an effective intervention had taken place. The measure-
ments in the 29 bedrooms that met this criterion were grouped depending on how the windows or doors had been
manipulated. Objectively measured and subjectively rated bedroom IAQ improved when the windows were open
except that the NO
2
concentration was slightly higher. Sleep was longer under this condition and sleep quality
was subjectively assessed to be better. Similar effects were not observed when the doors were open although the
95
th
percentile of CO
2
concentration decreased by as much as when the windows were open. No effects were seen
in the 11 bedrooms in which the change to the bedroom conditions made by the participants did not change the
CO
2
concentration by at least 200 ppm, as would be expected. The present study provides evidence that sufcient
dilution and/or removal of pollutants is necessary to ensure good bedroom IAQ and good sleep quality.
1. Introduction
Sleep plays a central role in human health and well-being [1]. Good
sleep quality enhances our immune system [2] and reduces the risk of
obesity [3,4] and chronic diseases [3]. It also improves next-day
cognitive performance (e.g. concentration, reaction time, and compre-
hension [5,6]) and reduces the risk of occupational injuries [7–10], all of
which have economic implications.
Although limited in number, existing studies show that poor indoor
air quality (IAQ) in bedrooms negatively affects sleep quality; these
studies mainly focused on how changing bedroom ventilation will affect
sleep quality. The IAQ in bedrooms is often characterized by measuring
carbon dioxide (CO
2
) which is a marker of ventilation effectiveness in
the presence of building occupants [11]. Recently Sekhar et al. [12] and
Akimoto et al. [13] summarized these studies and found that CO
2
levels
in many bedrooms are high indicating inadequate ventilation and
implying poor bedroom IAQ. They also proposed a tentative relationship
between bedroom CO
2
concentration during sleep and sleep quality.
This relationship suggests that CO
2
concentration should be <800 ppm
to avoid negative effects on sleep quality, that between 800 ppm and
1100 ppm sleep quality may be negatively affected, that levels above
1100 ppm have consistently been shown to have negative effects on
sleep quality and that sleeping at levels >2600 ppm is likely to reduce
next-day cognitive performance. Recent studies by Fan et al. [6] and Lan
et al. [14] support these conclusions.
A few studies measured IAQ and pollutants in bedrooms. Canha et al.
summarized these studies in a review and found that most studies
* Corresponding author.
E-mail address: xiafan@dtu.dk (X. Fan).
Contents lists available at ScienceDirect
Building and Environment
journal homepage: www.elsevier.com/locate/buildenv
https://doi.org/10.1016/j.buildenv.2022.109630
Received 27 July 2022; Received in revised form 12 September 2022; Accepted 20 September 2022

Building and Environment 225 (2022) 109630
2
measured mainly CO
2
; in some studies particles, carbon monoxide, total
volatile organic compounds (VOCs), and formaldehyde were also
monitored [15]. Some of these pollutants exceeded limit values pre-
scribed by standards and guidelines [16,17], although their impact on
sleep quality has not yet been elucidated [18].
Ventilation is typically used to improve bedroom IAQ. Mechanical
ventilation is not a common method for ventilating bedrooms [19–21].
In the Swedish housing stock, 59% of 3696 houses did not have any form
of mechanical ventilation system [20]. 75% of 304 surveyed homes in
Finland were naturally ventilated [21]. A recent survey in Denmark
found that 40% of 475 bedrooms were naturally ventilated [19]. Natural
ventilation is a method of increasing ventilation by specially designed
systems that use natural forces, such as wind-driven [22] and
buoyancy-driven ventilation [23], but we could not nd any informa-
tion on their use in the above-mentioned studies in Finland [21] or
Denmark [19]. We therefore assume that natural ventilation was ach-
ieved by the opening of windows by building occupants, sometimes
enhanced by opening doors. Sekhar et al. [12] reviewed international
and national standards and guidelines that prescribe bedroom ventila-
tion and concluded that many bedrooms do not meet existing ventilation
requirements. This was especially the case during the heating season for
bedrooms classied as having natural ventilation; in such bedrooms,
window opening is the only way to increase bedroom ventilation and
improve IAQ [24–30].
Although opening a window by building occupants, usually referred
to as natural ventilation, is a common mean of improving IAQ in bed-
rooms, people may not always do so. In a recent survey in Denmark, 70%
of 510 respondents preferred to keep the bedroom window closed during
sleep [19]. The doors to a bedroom can also be kept open to improve
IAQ, although in the above-mentioned survey in Denmark as many as
48% of respondents slept with the doors closed [19]. Similar results
were observed for bedrooms in China [28,29,31]. A eld study con-
ducted in China during spring and autumn observed that both the
window and door in 41% of 104 bedrooms were closed during sleep, and
only 52% kept either the window or door open [32]. A Norwegian
survey observed that only 39% of 1001 respondents opened bedroom
windows at night [33]. A study conducted in 500 bedrooms in Denmark
showed that a window was open in only 20% of the bedrooms at night,
although in most of them the bedroom door was kept open [26]. This
was probably because the measurements were made in children’s
bedrooms.
Only a few eld intervention studies have been conducted to inves-
tigate the effects of window and door opening on bedroom ventilation,
IAQ, and sleep quality [5,30,34–36]. Canha et al. [30] explored the
effects on IAQ during sleep of four different window and door congu-
rations in one naturally ventilated bedroom. They found that ventilation
rates were increased signicantly by opening the window, door, or both,
resulting in a lower level of some pollutants, particularly CO
2
. However,
they also observed the presence of some pollutants that originated
outdoors or from other parts of a dwelling, especially particulate matter.
Strøm-Tejsen et al. [5] showed that increasing the ventilation rate by
opening the window or turning on inaudible outdoor air inlet fans in
dormitory rooms reduced sleep onset latency, increased sleep efciency,
improved subjectively rated sleep quality, improved next-day cognitive
performance and reduced CO
2
concentration during sleep. Laverge and
Janssens [36] reported that window opening in eight naturally venti-
lated dormitory rooms caused participants to report being more rested
and reduced the measured duration of light sleep. The measured CO
2
concentration was also reduced. Liao et al. [35] found in a study with 27
subjects that window opening reduced snoring and the number of
awakenings at night. Mishra et al. [34] observed that window or door
opening in 17 bedrooms resulted in deeper sleep as reported by the
participants; lower CO
2
concentrations were found, and the objectively
measured number of awakenings at night decreased, sleep efciency
increased.
The studies mentioned above either did not monitor sleep quality,
were cross-sectional or generally carried out in dormitories with stu-
dents [5,30,34–36]. Not all of them focused on the effects of window and
door opening on the levels of different pollutants in bedrooms and sleep
quality. Concentration of CO
2
was a major metric used to characterize
bedroom IAQ, however the effect on IAQ could be different, depending
on whether the doors or windows were open even though both actions
can have similar effects on CO
2
levels in bedrooms. Taking these limi-
tations into account, our study was designed to supplement the evidence
on the types of benets that can be expected when sleeping with
bedroom doors or windows open.
2. Method
2.1. Approach
The present eld intervention study was carried out between
September and December 2020 in the Capital Region of Denmark, where
the climate is typical of a temperate zone. It is part of a large cross-
sectional eld measurement conducted in 84 bedrooms and focusing
on bedroom ventilation and sleep quality [37]. A subset of these mea-
surements is analysed in this study, focusing on how an intervention in
bedroom ventilation affected IAQ and sleep quality. Participants slept in
their own bedrooms for two consecutive weeks. In the rst week, they
slept under their normal bedroom conditions. In the second week, they
were asked to change the bedroom ventilation conditions by opening
bedroom windows or doors during the night if they had been closed in
the rst week, or closing them during the night if they had been open in
the rst week. Whether this intervention had substantively altered
bedroom ventilation was determined empirically in a subsequent anal-
ysis. An online questionnaire was used to collect information on the
participants, their bedrooms and sleep quality in the previous month
prior to taking part in this study. The bedroom environment and each
participant’s sleep quality were monitored continuously using in-
struments and were subjectively rated by each participant in online
sleep diaries that were completed on selected evenings and mornings.
The intervention that was made at the end of the rst week was also
recorded in the sleep diary. In addition, wrist skin temperature was
measured. The instruments were installed in a box that was supplied
together with an envelope containing all the necessary information
about the study. A short grammatical reasoning test was presented at the
end of each sleep diary to objectively quantify cognitive performance
before and after each night’s sleep.
Abbreviations
IAQ Indoor Air Quality
CO
2
Carbon dioxide
VOCs Total volatile organic compounds
SI Supplementary information
SD Standard deviation
PSQI Pittsburgh sleep quality index
RH Relative humidity
NO
2
Nitrogen dioxide
PM
1
Particulate matter with the aerodynamic diameter ≤1
μ
m
PM
2.5
Particulate matter with the aerodynamic diameter ≤2.5
μ
m
PM
10
Particulate matter with the aerodynamic diameter ≤10
μ
m
REM Rapid eye movement
GSQS Groningen sleep quality scale
X. Fan et al.

Building and Environment 225 (2022) 109630
3
2.2. Participants
The participants in the present study were recruited primarily from
among the respondents to an online survey that was conducted in early
2020 [19]. As too few replied to the invitation, we recruited more
participants by posting the invitation on social media. A total of 84
participants were recruited. Each participant received a DKK 30 voucher
for the coffee shop and among all participants, we randomly selected six
who received an actigraphy watch. As a general rule, we did not recruit
volunteers if they had reported a chronic disease or a sleep disorder
when completing the online questionnaire that had been used in an
online survey in 2020 [19].
Of the 84 recruited, 64 participated in the two-week measurements
during which an intervention was made in the second week but a
complete set of measured data was obtained from only 40 participants.
The measurements performed in their bedrooms were used for the an-
alyses. Table 1 summarises the information about all 40 participants and
after grouping them according to the intervention type. The average age
of participants in each subgroup was similar, although the standard
deviation (SD) was large; any potential effect on sleep quality caused by
age or other external factors was eliminated by the within-subjects
design in which only responses from the same participant under
different conditions were compared. Detailed information on all 64
participants who participated in this study is presented in Table S1 in the
Supplementary Information (SI).
2.3. Measurements
2.3.1. Physical measurements of the bedroom environment
Sixteen boxes, in which instruments for measuring CO
2
, air tem-
perature, relative humidity (RH) and a data logger, a light sensor and a
tablet with a 4-G internet connection provided by either a SIM card or a
router had been installed contained also an actigraphy watch, a skin
temperature sensor attached to a wristband made of Velcro and a multi-
divider socket, were prepared and used in the study (Fig. S1 in the SI). In
eight of the boxes a portable IAQ monitor had also been installed.
Air temperature, RH, and CO
2
concentration were continuously
monitored at intervals of 5 min using a Vaisala GMW90R (Vaisala
Corporation, Finland) connected to a HOBO UX120-006 M 4-channel
analogue data logger (Onset computer corporation, USA). The HOBO
U12-012 data logger had a built-in light sensor (Onset computer cor-
poration, USA), which measured and recorded the illuminance every 5
min. Selected pollutants (VOCs, nitrogen dioxide (NO
2
), particulate
matter with the aerodynamic diameter ≤1
μ
m (PM
1
), 2.5
μ
m (PM
2.5
),
and 10
μ
m (PM
10
)) were measured by the IAQ monitor Flow 2 (Plume
Labs, France) at intervals of 1 min, as in two of our previous studies [6,
39]. The accuracy and specications of all the instruments are listed in
Table S2 in the SI. The CO
2
sensors were calibrated just before the study
began, while the other sensors were factory calibrated.
The participants were instructed to plug in the box, to keep it plug-
ged in throughout the entire measuring period, and to place it at bed
height about 1 m away from the pillow, preferably on a night table as
recommended in the Instruction shown in Fig. S2 in the SI [30].
2.3.2. Objective measurement of sleep quality
Wrist-worn actigraphy watches were used to measure objective sleep
quality. These were either Fitbit Charge 2 or Fitbit Alta HR models. Both
of them have a sensitivity comparable with polysomnography, as was
documented in a study with subjects suffering from obstructive sleep
apnea [40]. Total sleep duration, time in bed, number of awakenings,
the duration of any periods awake after sleep onset, and the duration of
any periods of deep sleep, light sleep, and Rapid Eye Movement (REM)
sleep were all derived by proprietary software analysis of the continuous
records of heart rate and wrist movement that were subsequently
uploaded from the wrist-worn units, which the participants wore on the
non-dominant hand.
2.3.3. Subjective responses and questionnaires
During recruitment, participants completed the online questionnaire
that had been used in our previous survey [19]. The questionnaire ob-
tained information about each participant, including the characteristics
of their dwelling, bedroom and surroundings, information on bedroom
airing behaviour and the ventilation system, and information about
sleep habits including the questions whose answers could be used to
derive the PSQI.
Ten minutes before sleep, participants completed an evening sleep
diary. It consisted of questions concerning the number of nap times and
their length throughout the day, perceived sleepiness at the time of
answering the question and earlier during the daytime, activities (ex-
ercise and screen time before sleep), diet, smoking, any measures taken
to facilitate sleep, health status, estimated time of going to sleep, and the
perceived quality of the bedroom environment.
Ten minutes after waking up, the participants completed a morning
sleep diary. This obtained information on the time the participants woke
up, the number of awakenings during sleep and the reasons for any
awakenings, the number of adults and children in the bedroom during
that night, the perceived quality of the bedroom environment during
sleep and when answering the questions, whether any bedroom
Table 1
Anthropometric information about the 40 participants (Mean ±SD).
Items In total
a
Bedrooms in which the 95
th
percentile of CO
2
concentration between two weeks differed by
>200 ppm ≤200 ppm
Doors Open vs Closed Windows Open vs Closed
Nights No. where the data was obtained 58 20 23 15
Participants No 40 13
c
16 11
Sex 40 Females Males Females Males Females Males
7 6 7 9 5 6
Age (y) 32 ±12 28 ±4 28 ±3 30 ±10 35 ±18 28 ±10 27 ±5
Body Mass Index (kg/m
2
) 22.8 ±3.6 22.2 ±3.2 26.6 ±3.5 20.0 ±1.4 24.4 ±3.8 20.8 ±1.8 21.8 ±2.6
Living in Denmark ≥1 year 36 6 6 6 8 4 6
<1 year 4 1 0 1 1 1 0
Smoker 2 0 0 1 1 0 0
Chronic diseases Yes 3 0 1 1 0 1 0
No 36 7 5 6 9 4 5
Shift worker Yes 5 0 0 1 0 2 2
No 34 7 6 6 9 3 3
PSQI
b
6 ±3 5 ±2 6 ±5 6 ±2 5 ±3 6 ±2 5 ±1
a
Two participants did not answer all the questions.
b
Pittsburgh Sleep Quality Index [38].
c
Two participants shared bedroom.
X. Fan et al.

Building and Environment 225 (2022) 109630
4
windows or doors had been open that night, current sleepiness level and
nally, subjective sleep quality as indicated by the answers to the
questions in the Groningen Sleep Quality Scale (GSQS) [41]. A question
regarding deep sleep was reported separately as it did not count toward
the total GSQS score.
Subjective assessments of the bedroom environment included ther-
mal sensation, odour/noise/light intensity, air freshness/dryness, and
the acceptability of the thermal/IAQ/acoustic/visual environments.
These assessments and scales had been used in previous studies [6]. The
scales marked by the participants are shown in the SI together with the
scoring of the scales. Sleepiness was assessed on a six-point Likert scale
as follows: very sleepy (0), sleepy (1), somewhat sleepy (2), somewhat
awake (3), awake (4), and wide awake (5); mean values were calculated
to represent the level of sleepiness: the lower the mean, the sleepier the
participant.
Evening and morning sleep diaries were available both in Danish and
English and accessible online through links or QR codes. The partici-
pants were asked to complete these diaries at least twice a week and on
the same two days in each week from Monday night to Friday morning to
reduce potential bias caused by any systematic differences in activities
on different weekdays.
2.3.4. Cognitive performance
Once a sleep diary had been completed, a 3-min version of Badde-
ley’s test was presented to the participant. Baddeley’s test is a gram-
matical reasoning test measuring how well a participant understands the
relationship between objects as described in words and is used
frequently to measure cognitive performance [42]. It has been used
previously to measure cognitive performance after sleeping in poor IAQ
conditions [5,6].
2.3.5. Physiological measurements
Skin temperature on the dominant hand’s wrist was measured
continuously during sleep and recorded at intervals of 5 min. An iButton
DS1922L (Maxim integrated; USA) sensor was used and attached to a
wristband made of Velcro. Skin temperature has been shown to be a
good marker of thermal sensation so we used it in our study [43–45]. No
other physiological measurements were performed, although they might
have been useful, to maintain the realism and ensure that sleeping
conditions were not disturbed, for example by additional sensors.
2.3.6. Estimation of ventilation rates
Metabolically generated CO
2
was used to estimate ventilation rates
from the rate of decay of the measured CO
2
concentration each morning
[39,46,47]. The participants were asked to avoid re-entering their
bedroom for at least 30 min and to leave the bedroom conditions as they
had been during the previous night during this period. The 95
th
percentile of the CO
2
concentration during sleep was also calculated and
used to estimate the ventilation rate. This was assumed to be close to the
steady-state CO
2
concentration.
The above methods can provide an estimate of the total ventilation
rates in bedrooms but they cannot provide an estimate of the proportion
of outdoor air supplied to each bedroom.
2.4. Experimental procedure
The participants received detailed instructions via email and in the
envelope supplied together with the instrumentation box; videos with
all the necessary instructions were posted on YouTube (https://www.yo
utube.com/channel/UC8luzg7Uifjcd-217HuOmMA/videos).
Fig. 1 shows the measurement procedure for each night during the
two-week-long experimental period; evening and morning sleep diaries
were not taken every night.
Participants were asked to maintain their daily sleep patterns and
lifestyle routines. They were also asked to provide a photograph or a
sketch showing the dimensions of their bedroom, the location of the bed,
windows, doors, instrument box, and any outdoor air inlets. We checked
either the sketches or photos from the participants and found that they
were placed as instructed. Two researchers could be contacted at any
time throughout the period if there were any queries but this did not
occur.
The present study conformed to the guidelines in the Helsinki
Declaration. Written informed consent was obtained from each partici-
pant. The data collected via online questionnaires were pseudonymized
and stored on the Technical University of Denmark (DTU) server to
comply with the General Data Protection Regulation (GDPR) re-
quirements. Our proposal was approved by DTU and archived under
DOCX 19/1002413. Participants were informed that they were free to
discontinue their participation at any time but this did not occur.
2.5. Screening and classication of measured data
Although the measurements were made over a period of two weeks,
only data from the nights for which the sleep diary data were available
to identify the status of the windows and doors clearly are reported here;
all other data will be reported separately.
Measurements from the two-week measurements were available for
64 bedrooms. We screened raw data and excluded the measurements
from 24 bedrooms because they were incomplete (Fig. S3 in the SI). The
reasons for exclusion were as follows: for two participants only a single
week of objectively measured sleep data was available; three partici-
pants had no historical CO
2
data; two participants completed their sleep
diaries in only one week; four participants did not have objectively
measured sleep data and a completed sleep diary for the same weekday
nights in both weeks; four participants made an intervention that was
not comparable between weeks (for example they opened the doors with
the windows closed in the rst week, while in the second week they
opened the windows with the doors closed); one participant did not
make any interventions; the rest either did not report making the
intervention or the changes in CO
2
concentration did not match the
intervention they claimed to have made, e.g., the CO
2
concentration was
higher when they had reported sleeping with open windows or doors.
Fig. 1. An example of the experimental procedure with the sleep diaries for one night. Participants adopted these procedures without completing the sleep diaries on
the other nights each week.
X. Fan et al.

Building and Environment 225 (2022) 109630
5
We thus used data from only 40 participants in the analyses. We
compared the 95
th
percentile of CO
2
concentration for the same week-
day nights in each of the two weeks. The bedrooms in which the dif-
ference in the 95
th
percentile of CO
2
concentration was >200 ppm were
used in the analysis of the effects of an intervention; we chose 200 ppm
considering the accuracy of the CO
2
sensors and to account for the
possibility of incomplete mixing, and to ensure that there was a
measurable change in bedroom ventilation. Twenty-nine bedrooms met
this criterion.
As our focus was on the effects of open bedroom doors and windows,
the data from the 29 bedrooms in which the 95
th
percentile CO
2
differed
by >200 ppm between weeks were compared between two conditions
depending on the window and door status in each of the two weeks. In
one group, consisting of 13 bedrooms, the doors were either open or
closed each week, while the windows were always closed. In another
group consisting of 16 bedrooms, the windows were either open or
closed during each week while the doors in most of the bedrooms were
closed. In one bedroom the door was always open independently of
whether the windows were open or closed; in four bedrooms the win-
dows were open together with open doors while the doors were closed
when the windows were closed.
The data from the 11 bedrooms in which the 95
th
percentile CO
2
differed by ≤200 ppm between weeks were analysed for any difference
in effect between two cases: Closed condition (the windows and doors
were closed except for one bedroom where the doors were open for two
weeks); Open condition (windows or doors were open). These analyses
were treated as a form of control as no effects were expected considering
that the CO
2
concentration differed so little between the two weeks,
indicating that the intervention had not affected bedroom ventilation to
any meaningful extent.
Information about the participants in each of these three groups is
shown in Table 1. The detailed characteristics of window and door
opening are presented in Table S3 in the SI.
2.6. Analysis
A Shapiro-Wilk test was used to examine whether the data were
normally distributed. For normally distributed data, a paired-samples t-
test was used, but we otherwise used the non-parametric Wilcoxon
Matched-Pairs Signed-Ranks test. The data that had been measured
more than once each night were subjected to analysis of variance with a
repeated-measures design; the Greenhouse-Geisser method was used to
adjust the violation of sphericity. Post-hoc analysis was performed using
the Bonferroni test. The statistical analyses were performed using IBM
SPSS Statistics 22 (SPSS Inc, Chicago, IL, USA). The signicance level
was set at P =0.05 (2-tail). In the analyses, the status of window and
door opening were independent variables while all other parameters
measured or reported by the participants were dependent variables.
The effect size was calculated using Cohen’s method [48]. Cohen’s f
examining the practical importance of outcomes based on their variance
denes the small (0.1), medium (0.25), and large (0.4) effect sizes [49].
Cohen’s d examining the practical importance of outcomes by
comparing the mean values also denes small (0.2), medium (0.5), and
large (0.8) effect sizes [49]. Small, medium, and large effect sizes imply
that 58%, 69%, and 79% of the results were higher than the mean value,
respectively [49].
Considering the limited number of participants within each subgroup
after data screening in the present study, Bonferroni post-hoc analysis
was also performed when the effect size (Cohen’s f) of intervention ef-
fects, time effects, or the interaction of intervention and time was large.
3. Results
The characteristics of the 40 bedrooms from which data were ana-
lysed are summarized in Table 2. They were primarily non-smoking
spaces located in multi-story apartment buildings in suburban areas
but a few were in a student dormitory building. Many of the buildings
had been constructed before the 1980s and after the 2010s. The average
volume and oor area of the bedrooms were approximately 36 m
3
and
13 m
2
, respectively. According to the incidence of air terminals reported
by the participants, most of the bedrooms had exhaust ventilation. We
did not verify this information or whether the ventilation systems were
in operation at night to avoid the need to enter the bedrooms. More than
half of the bedrooms were singly occupied during the measurement
period. The information for all 64 surveyed bedrooms is presented in
Table S4 in the SI.
Table 3 shows the objectively measured bedroom environmental
quality during sleep. The air change rate was on average lower than the
statutory minimum of 0.5 h
−1
when the doors and windows were closed
but when they were open it increased considerably. The mean 95
th
percentile of CO
2
concentration (and the mean CO
2
concentration)
during sleep was reduced from 2916 ppm to 1415 ppm (mean concen-
tration was reduced from 2362 ppm to 1293 ppm) when the doors were
open and from 2310 ppm–904 ppm (mean from 1820 ppm–761 ppm)
Table 2
Information about surveyed bedrooms (Mean ±SD).
Items In
total
a
Bedrooms in which the 95
th
percentile of CO
2
concentration
between two weeks differed by
>200 ppm ≤200
ppm
Doors
Open vs
Closed
Windows
Open vs
Closed
Building types Detached house 9 1 5 3
Row-house 3 0 3 0
Multi-story
apartment
building
24 11 8 5
Others, (i.e.
cottages)
2 0 0 2
Non-smoking
dwellings
Yes 38 12 15 11
No 2 1 1 0
Built year of
the
dwellings
Before 1960 10 6 3 1
1961–1981 4 2 1 1
1982–1995 1 0 1 0
1996–2009 2 0 2 0
After 2010 7 1 4 2
Don’t know 15 4 5 6
Dwellings’
location
b
Urban 8 0 4 4
Suburban 31 12 12 7
Rural 1 1 0 0
Living in
dormitory
Yes 9 1 4 4
No 30 12 12 6
Bedroom
located oor
<0 (basement) 0 0 0 0
Ground oor (0) 16 9 6 1
≥1st oor 23 4 10 9
Bedroom size (m
3
) 36 ±
12
36 ±12 32 ±11 35 ±
14
Bedroom
ventilation
c
Mechanical
ventilation
8 2 1 5
Exhaust
ventilation
28 8 15 5
Natural
ventilation
3 3 0 0
Occupants in
bedrooms
during sleep
1 25 4 13 8
≥2 15 9 3 3
a
Two participants did not answer all the questions.
b
Residential location was deduced from the zip codes. Urban regions refer to
the areas with the rst two numbers of zip codes 25 or below; suburban regions
refer to the areas with the rst two numbers of postcodes 26–31, 34–36, 40,
50–52, 70, 80–82, and 90–92, and the other areas in the capital region of
Denmark are rural.
c
Bedrooms with air terminals were considered to have a fully balanced me-
chanical ventilation system, bedrooms with trickle vents and air terminals in the
bathroom were considered to have mechanical ventilation with exhaust only;
others were naturally ventilated bedrooms.
X. Fan et al.

Building and Environment 225 (2022) 109630
6
when the windows were open.
Mean RH was signicantly higher when the doors and windows were
closed, and the mean RH was in the range of 40–60%. The mean tem-
perature decreased signicantly when the windows were open but by
only 0.8 ◦C.
The mean NO
2
concentration during sleep increased when either the
doors or windows were open; when a window was open, the NO
2
con-
centration was higher, though not signicantly (P <0.10). The con-
centrations of VOCs and PM
10
were signicantly lower when the
windows were open, indicating that they originated indoors. Boor et al.
summarized in a review that particles could be suspended by body
movements in bed [50], which is likely to explain the observed higher
PM
10
concentration with the windows closed. Door opening also tended
to reduce the level of VOCs (P <0.10). The concentrations of PM
2.5
and
PM
1,
and the illuminance level were unaffected by either intervention.
In the bedrooms in which the 95
th
percentile CO
2
concentration
differed by ≤200 ppm between weeks it still decreased when the win-
dows and/or doors were opened although it remained in the range of
900-1100 ppm. No other measurements differed between weeks in this
failed intervention group.
Subjective ratings of the bedroom environment are shown in
Table S5 in the SI. Signicant differences in the ratings following post-
hoc analysis are summarized in Figs. 2 and 3. The acceptability of
bedroom IAQ increased (Fig. 2A), and the odour intensity decreased
Table 3
Objectively measured bedroom environment during sleep (Mean ±SD). Cohen’s d. **P <0.01; *P <0.05.
Parameters Bedrooms in which the 95
th
percentile of CO
2
concentration between two weeks differed by
>200 ppm ≤200 ppm
Doors Position Windows Position Closed
b
Open
c
P-
value
d
Closed Open P-value d Closed Open P-value d
Air change rate (h
−1
) 0.29 ±
0.24
0.78 ±
1.14
0.007** 0.62 0.34 ±
0.31
1.21 ±
1.14
<0.001** 1.08 0.77 ±
0.58
1.24 ±
1.4
0.875 0.44
95
th
percentile of CO
2
concentration (ppm)
d
2916 ±
960
1415 ±
495
<0.001** 2.02 2310 ±
944
904 ±
406
<0.001** 1.98 1067 ±
516
1017 ±
530
0.083 0.10
Mean CO
2
concentration
(ppm)
e
2362 ±
728
1293 ±
465
<0.001** 1.80 1820 ±
706
761 ±
273
<0.001** 2.02 975±
±462
893 ±
425
0.015* 0.19
NO
2
(ppb)
a
1.2 ± 1.3 4.7 ± 4.5 0.043* 1.14 3.5 ±3.8 10.4 ±
12.5
0.056 0.77 4.0 ±4.3 4.0 ±4.8 0.695 0.01
VOCs (ppb)
a
198.6 ±
66.1
164.1 ±
49.9
0.086 0.62 205.6 ±
60.7
156.0 ±
59.3
0.001** 0.86 170.9 ±
26.8
169.1 ±
46.6
0.867 0.05
PM
10
(
μ
g/m
3
)
a
24.1 ±
24.5
23.1 ±
23.6
0.594 0.04 46.8 ±
37.1
26.3 ±
22.1
0.026* 0.70 23.1 ±
24.9
20.0 ±
18.8
0.701 0.14
PM
2.5
(
μ
g/m
3
)
a
4.7 ±3.1 4.5 ±2.9 0.953 0.06 5.4 ±3.9 3.9 ±2.5 0.124 0.47 4.1 ±3.1 3.8 ±2.9 0.701 0.13
PM
1
(
μ
g/m
3
)
a
1.9 ±1.6 1.9 ±1.5 0.575 0.01 1.8 ±1.5 1.7 ±1.2 0.875 0.08 1.7 ±1.5 1.6 ±1.4 0.223 0.04
Temperature (◦C) 24.7 ±
3.8
24.7 ±
2.9
0.774 0.03 23.2 ±
1.3
22.4 ±
1.9
0.008** 0.51 24.3 ±
1.3
23.7 ±
1.8
0.175 0.36
Relative humidity (%) 54 ± 7 51 ± 7 <0.001** 0.54 53 ± 6 48 ± 6 0.001** 0.78 44 ±8 42 ±10 0.278 0.20
Illuminance (Lux) 11.3 ±
7.1
12.5 ±
9.3
0.420 0.14 11.4 ±
9.0
10.2 ±
6.2
0.794 0.16 8.9 ±6.3 10.0 ±
8.6
0.638 0.14
a
Data from six participants with nine nights of measurements was available when the doors were closed or open; Data from 10 participants with 14 nights of
measurements was available when the windows were closed or open; Data from 9 participants with 13 nights of measurements was available when the difference in the
95
th
of CO
2
concentration was ≤200 ppm between two weeks.
b
The windows and doors were both closed except for one bedroom with the doors open.
c
The windows or doors were open.
d
The number of bedrooms with the CO
2
concentration <1100 ppm, which according to Refs. [12,13] indicates the level below which it is less probable that there are
effects on sleep quality: >200 ppm group: doors closed - 0 bedroom, doors open −3 bedrooms of 13, windows closed – 1 bedroom, windows open 12 bedrooms of 16
bedrooms; <200 ppm group: closed - 7 bedrooms, open - 7 bedrooms of 11 bedrooms.
e
The number of bedrooms with the CO
2
concentration <1100 ppm, which according to Refs. [12,13] indicates the level below which it is less probable that there are
effects on sleep quality: >200 ppm group: doors closed - 0 bedroom, doors open – 6 bedrooms of 13, windows closed – 2 bedrooms, windows open 13 bedrooms of 16
bedrooms; <200 ppm group: closed - 8 bedrooms, open - 9 bedrooms of 11 bedrooms.
Fig. 2. (A) The acceptability of IAQ; (B) odour intensity; and (C) air freshness before, during (recalled), and after sleep under the different conditions. Cohen’s d. **P
<0.01; *P <0.05.
X. Fan et al.

Building and Environment 225 (2022) 109630
7
(Fig. 2B) when the windows were open compared with the condition
when they were closed; door opening did not produce similar effects.
There was a small but signicant improvement in the ratings of air
freshness when the doors were open, while in the case of window
opening the improvement was greater and showed that the air was rated
as much fresher (Fig. 2C). Thermal sensation decreased when the win-
dows were open from around neutral to slightly cool; no such effect was
observed when the doors were open (Fig. 3A). There were no signicant
differences between weeks in the measured wrist skin temperature
(Fig. S4 in the SI). However, the skin temperature was from the sleep
period, and the acceptability of the thermal environment, which was
rated while awake, improved when the windows were open, again with
no similar effects when the doors were open (Fig. 3B). No other signif-
icant differences were observed.
The acceptability of the IAQ decreased, and the odour intensity
increased in the morning compared with the evening, independent of
the status of the doors. Similar changes were observed when the win-
dows were closed. The air was rated to be stufer, and the thermal
sensation was warmer when the doors and windows were closed. The
perceived light intensity increased after sleep compared with before
sleep independently of the window status (Fig. S5A in the SI). The
subjective responses made by participants did not differ between weeks
in the failed intervention group (Table S5 and Fig. S6 in the SI).
Objectively measured sleep quality was compared with what is
currently recommended by the U.S. National Sleep Foundation [51]. The
results are tabulated in Table S6 in the SI and show that the sleep quality
of the participants in the present study would not be regarded as poor.
There were no signicant differences in objectively measured
Fig. 3. (A) Thermal sensation; and (B) acceptability of the thermal environment before, during (recalled), and after sleep under the different conditions. Cohen’s d.
**P <0.01; *P <0.05.
Fig. 4. Objectively measured sleep length under the different conditions. Cohen’s d. *P <0.05.
X. Fan et al.

Building and Environment 225 (2022) 109630
8
parameters dening sleep quality (Table S6 in the SI) except for sleep
duration, which was signicantly longer when the windows were open
(Fig. 4). There were no differences in objectively measured sleep quality
between weeks in the failed intervention group.
Subjectively rated sleep quality tended to improve when the win-
dows were open (P <0.10); no similar effect was seen when the doors
were open (Fig. 5). Self-reported sleepiness is summarized in Table S7 in
the SI. Signicant post-hoc results are shown in Fig. 6. The self-reported
sleepiness level was lower when the windows were open; similar effects
were observed in the morning compared with the evening in this con-
dition; no other differences were seen. The percentage of participants
reporting having a deep sleep increased when the windows were open;
no such effect was seen when the doors were open (Fig. 7). There was no
difference in self-reported number of awakenings during sleep and the
reasons for waking up were random, which supports the realism of this
study (Table S8 in the SI). There were no signicant effects on subjec-
tively reported sleep quality, sleepiness level or depth of sleep in the
failed intervention group (Table S7 and Figs. S7–8 in the SI).
The performance of Baddeley’s test before and after sleep is shown in
Fig. 8 and Table S9 in the SI. There was a signicant decrease in the
percentage of errors after sleeping with windows open. We observed
that the performance before sleep differed when the doors were open
compared with when they were closed but this difference cannot be
attributed to the sleeping conditions. There were no signicant differ-
ences in performance before and after sleep in the failed intervention
group.
4. Discussion
Our results show that opening windows or doors signicantly
reduced CO
2
concentration during sleep. This is conventionally inter-
preted to indicate that in both cases bedroom IAQ improved. However,
this was not the case as shown by other measurements. Only when the
windows were open did the participants rate bedroom IAQ better. Sleep
quality also improved in this condition, which is consistent with the
published studies reviewed in the Introduction section [34–36]. No such
improvements were seen when the doors were open. We believe that
door opening did not provide adequate removal and dilution of pollut-
ants in bedrooms even though we were not able to conrm this hy-
pothesis with the limited measurements that were made. The reduced
levels of CO
2
when the doors were open suggest that air from other parts
of the dwelling was either drawn or diffused into the bedrooms. The CO
2
concentration of this air was low during the sleep period, as other spaces
in the dwelling were not occupied, and consequently the bedroom CO
2
concentration was reduced, though not by as much as when the windows
were open (Table 3).
As shown in the review by Canha et al., bedroom air during sleep
contains numerous pollutants whose levels are higher than the limit
values prescribed by the standards and guidelines [15]. These pollutants
can enter bedrooms from other parts of the dwelling [30]. For example,
cooking oil fumes originating from the kitchen were associated with
overall poor sleep quality [52]. Additionally, exposure to increased
PM
10
concentration was signicantly associated with increased
obstructive sleep apnea [53], which may disturb sleep. A recent
cross-sectional study showed that sleep stages were affected during ex-
posures to NO
2
, PM
2.5
, and O
3
; as a result some decreases in cognitive
capacity were observed [54]. Finally, Chen et al. concluded that
long-term exposures to PM
2.5
, PM
10
, and NO
2
were associated with poor
sleep quality in rural China [55]. If proper removal or dilution of these
and other pollutants is not achieved by the air that enters bedrooms, no
improved bedroom IAQ and sleep quality should be expected. This may
have been the case when the internal doors were open in the present
study, even though the total concentration of VOCs was lower. On the
other hand, window opening was able to provide sufcient dilution and
removal of some of these pollutants. For example, the present study
showed that the total concentration of VOCs and PM
10
levels were lower
and perceived IAQ was improved when the windows were open. How-
ever, window opening increased NO
2
concentration, which thus pre-
sumably originated outdoors. This could counteract the positive effect of
reduced exposure to other pollutants because exposure to NO
2
can in-
crease the risk of sleep apnea [56]. Future studies should closely look at
the impact of outdoor air pollution on sleep quality and consequently
Fig. 5. Subjective measurements of sleep quality under the different conditions. Cohen’s d.
X. Fan et al.

Building and Environment 225 (2022) 109630
9
examine measures that should be taken to reduce their levels.
It is worth noting that in an earlier study involving multiple CO
2
measurements in a bedroom having windows with fully open trickle
vents, Sekhar et al. showed that the bedroom was better ventilated by
the incoming outdoor air through the trickle vents, which was further
enhanced with the assistance of a kitchen hood or bathroom extraction
fan if an internal bedroom door was open [25]. We are unable to verify
whether this occurred in the present study because the operation of
ventilation systems during the measurements period was not checked, as
mentioned earlier. The CO
2
measurements just outside the bedroom
door would help characterize the airow direction in future studies.
It is reasonable to assume that the expectations of participants could
bias the results of the present study particularly their subjective
responses (a Hawthorne effect) [34]. As opening a window or a door is
usually assumed to improve bedroom IAQ, the participants might expect
that they have positive effects but no such effects on subjective responses
were found. Additionally, in the failed intervention group in which
bedroom IAQ remained unchanged even though doors and windows
were opened, no positive or negative effects of expectation on subjective
responses were found. Another reasonable hypothesis is that becoming
familiar with the routines of the experiment during the rst week might
have helped the participants to sleep better in the second week. If so, as
more participants went from open to closed windows in the second week
than in the reverse direction (10 versus 6, respectively, as shown in
Table 4; details summarized in Table S3 in the SI), this might have
spuriously appeared to favour closed windows. That open windows were
Fig. 6. Self-reported sleepiness before and after sleep under the different conditions (Mean ±SD). Cohen’s d.*P <0.05.
Fig. 7. of the participants who reported (recalling after sleep) having a “deep sleep” under the different conditions. Cohen’s d. *P <0.05.
X. Fan et al.

Building and Environment 225 (2022) 109630
10
found to result in better sleep is thus a conservative conclusion.
The present results support the tentative relationship between
bedroom ventilation (as indicated by mean CO
2
level) and sleep quality
that has been recently suggested [12,13]. According to this relationship,
no effects on sleep quality would be predicted in bedrooms with doors
open, as observed in the present study. The present results support the
recommendation that the outdoor air supply rate in bedrooms should be
>10 L/s per person [39] to ensure that sleep quality is undisturbed and
that the resulting CO
2
concentration is below 750 ppm.
A change in the 95
th
percentile of the CO
2
concentration that was less
than 200 ppm was insufcient to produce any measurable changes in the
outcomes in the present study, lending support to the analysis. There
could be many reasons why the change in the 95
th
percentile of the CO
2
concentration after the intervention was lower than 200 ppm in 11
bedrooms. The most plausible reason is that ve of them had mechanical
ventilation and ve had extract ventilation, so opening windows or
doors would have had a very limited effect on CO
2
levels which were
already low. The recently published paper analysing all results from the
present study collected in the rst week showed that the CO
2
concen-
tration measured in bedrooms with mechanical ventilation was lower
compared with bedrooms with other types of ventilation [37], which
further supports this interpretation.
An apparent effect on next-day cognitive performance was seen in
the present study at a mean CO
2
level of 1800 ppm, lower than the 2600
ppm suggested by the above authors as the threshold for such effects.
However, although next-day cognitive performance improved signi-
cantly after sleeping in bedrooms with windows open and did not
improve signicantly after sleeping with windows closed, which is
compatible with a benecial effect of sleeping with open windows, no
signicant difference in performance could be shown between the
conditions with windows open and closed. The present results therefore
do not constitute an extension of the ndings of Strøm-Tejsen et al. [5] to
lower CO
2
concentrations.
In the present experiment, cognitive performance was measured in
the bedroom, i.e., with the same IAQ as during the sleep period, so the
observed effects of poor bedroom IAQ could be due either to the
resulting poor sleep quality or to a direct effect of poor IAQ at the time
the test was performed; the latter effect has been documented in many
studies [57,58]. These two effects cannot be separated in the present
experiment.
The CO
2
level is generally used to estimate the ventilation rate and
IAQ [12,13,15]. Reduced CO
2
concentration in the present study did
indicate increased dilution, but it was not a good predictor of bedroom
IAQ (Fig. 2 and Table 3). CO
2
may not always be a good marker of IAQ as
indicated in the recently published ASHRAE’s Position Document [59].
Future studies should consider measuring other contaminants as well as
CO
2
to obtain a better estimate of bedroom IAQ.
The present results show that ventilation with outdoor air results in
improved bedroom IAQ and sleep quality. This suggests that effective
measures must be taken to ensure sufcient delivery of clean outdoor air
to bedrooms. These measures include mechanical ventilation, natural
Fig. 8. The accuracy of Baddeley’s test under the different conditions (Mean ±SD). Cohen’s d. *P <0.05.
Table 4
The number of participants (nights of measurements each week) who changed
the doors and windows from open to closed and vice versa between weeks.
First week to second week Bedrooms in which the 95
th
percentile of CO
2
concentration between two weeks differed by
>200 ppm ≤200 ppm
Doors Open
vs Closed
Window Open
vs Closed
Open vs
Closed
a
Changing from open doors to
closed doors
5 (8) - - - -
Changing from closed doors to
open doors
8 (12) - - - -
Changing from open windows to
closed windows
- - 10 (15) - -
Changing from closed windows to
open windows
- - 6 (8) - -
Changing from open windows or
doors to closed windows and
doors
- - - - 5 (7)
Changing from closed windows
and doors to open windows or
doors
- - - - 6 (8)
a
Closed: the windows and doors were both closed except for one bedroom
with the doors open; Open: the windows or doors were open.
X. Fan et al.

Building and Environment 225 (2022) 109630
11
ventilation in which windows are operated automatically as a function
of different parameters, or simply providing the possibility of opening
windows to occupants for airing the bedrooms. As described in the
Introduction section and as shown by Sekhar et al. [12], most bedrooms
are ventilated only by voluntary window opening, which cannot take
place while asleep even if bedroom IAQ becomes worse during the night.
The effectiveness of ventilation by window opening depends on many
factors, among which outdoor conditions and wind pressure differences
play the most important role. Window opening can be discouraged by
conditions outside the building, including perceived security and
ambient noise as well as privacy considerations. In areas with high
outdoor air pollution, window opening can even make the IAQ in bed-
rooms worse by allowing outdoor pollutants to enter the bedroom.
Finally, window opening can allow rain to enter and may increase
bedroom temperature during hot weather, or increase energy use for
heating during cold weather. On the other hand, retrotting bedrooms
with mechanical ventilation may involve technical difculties and can
be costly. In the absence of mechanical ventilation or specially designed
natural ventilation systems, other means of improving bedroom IAQ
could be considered [60]. One way would be the use of air cleaners.
However, information on their application for improving bedroom IAQ
is limited [61,62] and no data are available on whether their use would
improve sleep quality. Future studies should closely address this matter
and focus on other methods and retrots that would result in improved
bedroom IAQ.
5. Limitations
To keep the realism, the study was performed in actual bedrooms on
weekdays. This has signicant implications because not all potentially
disturbing factors could be controlled. Yet, the present study was an
intervention performed in two subsequent weeks, and we compared the
measurements obtained in one week against the measurements in the
other week and performed analyses as within-subject comparisons. We
believe that this approach to some extent controlled for external factors.
A similar approach was used successfully in previous studies examining
the effects on work performance in the laboratory [63] and in the eld
[64], as well as in the study examining the effect of bedroom ventilation
on sleep quality and next-day cognitive performance [5].
With the window open, the relative humidity and temperature in
bedrooms during sleep were lower. This could potentially affect sleep
quality [6,44] and bedroom IAQ [65], but we were unable to separate
these effects in the subsequent analyses. Yet, participants did not report
thermal discomfort in any of the conditions examined so the inuence of
these small differences in thermal conditions on sleep quality is expected
to be minor.
We did not measure noise levels. However, most of the surveyed
bedrooms were located in suburban areas where the ambient noise
levels during sleep can generally be assumed to be low. Consequently,
window opening should not have caused much sleep disturbance due to
external noise. In support of this conclusion, participants rated the noise
intensity as low and the acoustic environment in bedrooms as highly
acceptable.
Window opening can increase air speed and cause draft problems.
We did not assess the effects of these two parameters. If anything, they
would be expected to reduce the positive effects that were observed
when the windows were open in the season when the present study was
performed.
The estimation of ventilation rates from the decay rate of metaboli-
cally generated CO
2
requires several assumptions including the level of
CO
2
in the air supplied to bedrooms and good mixing within the bed-
rooms. The estimated ventilation rate reects the local conditions where
the CO
2
concentration was measured and should not be assumed to
apply to the entire volume of the bedroom. A discussion of air distri-
bution and exposure is important in this context, as depending on the
sleep position and air distribution, which can be inuenced by the
position of the bed in relation to the window and the door, the actual
exposure of participants could be different from what was estimated
from the measured CO
2
concentration [25,50,66,67]. We were not able
to examine the air distribution in the bedrooms. However, as we
compared the effects of high and low levels of CO
2
in the same bed-
rooms, our results are valid even without a knowledge of air mixing. In
future studies it would be useful to measure the actual exposures of
sleeping occupants and the air distribution in bedrooms.
Our results are based on a smaller number of observations than was
intended. One reason is that we used stringent inclusion criteria,
including that the sleeping diaries should have been completed on the
same weekday night in each week of the two-week measurements and
that the intervention must have changed the 95
th
percentile of CO
2
by
more than 200 ppm. Conrmation in studies with a larger sample size
and longer duration would be useful.
We did not collect full information about the buildings, bedrooms
and participants. One of the reasons was that the participants already
had to perform many measurements and provide many responses and
we did not want to further inconvenience them. Also, some information
could be considered as sensitive and the participants might have been
unwilling to share it with us or might even have withdrawn from
participating in this study if they were requested to provide it. In future
studies it would be useful to collect more information on climatic data
and building location, the economic status of the participants, their
occupation and the type of bedding used, including an estimation of its
thermal insulation, and the use of curtains, etc, so that this information
can be included as additional variables in the statistical models. Inter-
vention studies like the one presented in this paper can control for many
of these factors as the responses from the same participants were
compared. In bedrooms with more than one participant it would be
useful to collect data from both occupants, if both agree to participate in
the monitoring program.
The present study was conducted during the heating season in a
temperate climate zone. Consequently the ndings require verications
in other climate zones and periods of the year before they can be widely
applied.
6. Conclusions
A eld intervention study was conducted in bedrooms during the
heating season. The effects of bedroom door and window opening on
IAQ, sleep quality, and next-day cognitive performance were deter-
mined. Sleeping with the window open improved bedroom IAQ and
provided benets for sleep quality. No such effects were observed when
sleeping with the door open. The present study provides evidence that
bedroom ventilation with outdoor clean air resulting in improved
bedroom IAQ is important for sleep quality.
CRediT authorship contribution statement
Xiaojun Fan: Writing – original draft, Visualization, Validation,
Methodology, Investigation, Formal analysis, Data curation, Conceptu-
alization. Chenxi Liao: Methodology, Investigation, Conceptualization.
Mariya P. Bivolarova: Investigation. Chandra Sekhar: Writing – re-
view & editing. Jelle Laverge: Writing – review & editing. Li Lan:
Writing – review & editing. Anna Mainka: Writing – review & editing.
Mizuho Akimoto: Writing – review & editing. Pawel Wargocki:
Writing – review & editing, Supervision, Resources, Project adminis-
tration, Methodology, Investigation, Funding acquisition, Data curation,
Conceptualization.
Declaration of competing interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
X. Fan et al.

Building and Environment 225 (2022) 109630
12
Data availability
Data will be made available on request.
Acknowledgments
This work is part of ASHRAE research project on bedroom ventilation
and sleep quality (grant number: 1837-RP). Xiaojun Fan was jointly
supported by the China Scholarship Council (CSC) (grant number:
201906370016), DTU Department of Civil Engineering (now Depart-
ment of Environmental and Resource Engineering), Otto Mønsteds
Foundation (grant number: 20-70-1043), and S.C. Van Fonden (grant
number: 2021-0100). Chenxi Liao was supported by VLAIO research
project on smart ventilation (grant number: HBC.2020.2520) and
Research Foundation – Flanders (FWO), Belgium (grant number:
V409120 N). Chandra Sekhar received nancial support from the Na-
tional University of Singapore for his sabbatical visit to DTU. Anna
Mainka was supported by the National Agency for Academic Exchange
of Poland (grant number: PPI/APM/2018/1/00004). Pawel Wargocki
was nancially supported by Fonden af 20. december (grant number:
155808-15). The authors thank Prof. Jørn Toftum for his guidance in
constructing the online questionnaires’ webpage and the 84 households
participating in this study. Prof. David P. Wyon is acknowledged for
proofreading the nal version of the manuscript.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.buildenv.2022.109630.
References
[1] P. Wargocki, L. Lan, Z. Lian, D.P. Wyon, Thermal environment, IAQ and sleep,
ASHRAE J. 60 (2018) 60–63.
[2] L. Besedovsky, T. Lange, M. Haack, The sleep-immune crosstalk in health and
disease, Physiol. Rev. 99 (2019) 1325–1380, https://doi.org/10.1152/
physrev.00010.2018.
[3] M. Miller, F. Cappuccio, Inammation, sleep, obesity and cardiovascular disease,
Curr. Vasc. Pharmacol. 5 (2007) 93–102, https://doi.org/10.2174/
157016107780368280.
[4] B. Morrissey, E. Taveras, S. Allender, C. Strugnell, Sleep and obesity among
children: a systematic review of multiple sleep dimensions, Pediatr. Obes. 15
(2020) 1–21, https://doi.org/10.1111/ijpo.12619.
[5] P. Strøm-Tejsen, D. Zukowska, P. Wargocki, D.P. Wyon, The effects of bedroom air
quality on sleep and next-day performance, Indoor Air 26 (2016) 679–686, https://
doi.org/10.1111/ina.12254.
[6] X. Fan, H. Shao, M. Sakamoto, K. Kuga, L. Lan, D.P. Wyon, K. Ito, M.P. Bivolarova,
C. Liao, P. Wargocki, The effects of ventilation and temperature on sleep quality
and next-day work performance : pilot measurements in a climate chamber, Build.
Environ. 209 (2022), 108666, https://doi.org/10.1016/j.buildenv.2021.108666.
[7] K.M. Lawson, S. Lee, Better previous night sleep is associated with less next day
work-to-family conict mediated by higher work performance among female
nursing home workers, Sleep Heal 4 (2018) 485–491, https://doi.org/10.1016/j.
sleh.2018.07.005.
[8] S. Lee, C. Mu, B.D. Gonzalez, C.E. Vinci, B.J. Small, Sleep health is associated with
next-day mindful attention in healthcare workers, Sleep Heal 7 (2021) 105–112,
https://doi.org/10.1016/j.sleh.2020.07.005.
[9] L.M. Swanson, J.T. Arnedt, M.R. Rosekind, G. Belenky, T.J. Balkin, C. Drake, Sleep
disorders and work performance: ndings from the 2008 national sleep foundation
sleep in America poll, J. Sleep Res. 20 (2011) 487–494, https://doi.org/10.1111/
j.1365-2869.2010.00890.x.
[10] M. Alhainen, M. H¨
arm¨
a, J. Pentti, J.M. Ervasti, M. Kivim¨
aki, J. Vahtera,
S. Stenholm, Sleep duration and sleep difculties as predictors of occupational
injuries: a cohort study, Occup. Environ. Med. (2021), https://doi.org/10.1136/
oemed-2021-107516 oemed-2021-107516.
[11] A. Persily, Development and application of an indoor carbon dioxide metric, Indoor
Air 32 (2022) 1–13, https://doi.org/10.1111/ina.13059.
[12] C. Sekhar, M. Akimoto, X. Fan, M. Bivolarova, C. Liao, L. Lan, P. Wargocki,
Bedroom ventilation: review of existing evidence and current standards, Build.
Environ. 184 (2020), 107229, https://doi.org/10.1016/j.buildenv.2020.107229.
[13] M. Akimoto, C. Sekhar, M.P. Bivolarova, C. Liao, X. Fan, J. Laverge, L. Lan,
P. Wargocki, Reviewing how bedroom ventilation affects IAQ and sleep quality,
ASHRAE J. 63 (2021) 56–60.
[14] L. Lan, Y. Sun, D.P. Wyon, P. Wargocki, Pilot study of the effects of ventilation and
ventilation noise on sleep quality in the young and elderly, Indoor Air 31 (2021)
2226–2238, https://doi.org/10.1111/ina.12861.
[15] N. Canha, C. Teixeira, M. Figueira, C. Correia, How is indoor air quality during
sleep? Rev. Field Stud. Atmosphere (Basel) 12 (2021) 110, https://doi.org/
10.3390/atmos12010110.
[16] World Health Organization, WHO Guidelines for Indoor Air Quality: Selected
Pollutants, 2010.
[17] EN 16798–1:2019, Energy Performance of Buildings – Ventilation for Buildings –
Part 1 : Indoor Environmental Input Parameters for Design and Assessment of
Energy Performance, 2019.
[18] L. Lan, Z. Lian, Sleep and indoor air quality, in: Y. Zhang, P.K. Hopke, C. Mandin
(Eds.), Handb. Indoor Air Qual., Springer, Singapore, 2021, pp. 1–16, https://doi.
org/10.1007/978-981-10-5155-5_66-1.
[19] C. Liao, M. Akimoto, M.P. Bivolarova, C. Sekhar, J. Laverge, X. Fan, L. Lan,
P. Wargocki, A survey of bedroom ventilation types and the subjective sleep quality
associated with them in Danish housing, Sci. Total Environ. 798 (2021), 149209,
https://doi.org/10.1016/j.scitotenv.2021.149209.
[20] H. Stymne, C. Axel Boman, J. Kronvall, Measuring ventilation rates in the Swedish
housing stock, Build. Environ. 29 (1994) 373–379, https://doi.org/10.1016/0360-
1323(94)90037-X.
[21] L. Øie, H. Stymne, C.-A. Boman, V. Hellstrand, The ventilation rate of 344 Oslo
residences, Indoor Air 8 (1998) 190–196, https://doi.org/10.1111/j.1600-
0668.1998.t01-1-00006.x.
[22] J. Seifert, Y. Li, J. Axley, M. R¨
osler, Calculation of wind-driven cross ventilation in
buildings with large openings, J. Wind Eng. Ind. Aerod. 94 (2006) 925–947,
https://doi.org/10.1016/j.jweia.2006.04.002.
[23] X.H. Ren, R.Z. Liu, Y.H. Wang, L. Wang, F.Y. Zhao, Thermal driven natural
convective ows inside the solar chimney ush-mounted with discrete heating
sources: reversal and cooperative ow dynamics, Renew. Energy 138 (2019)
354–367, https://doi.org/10.1016/j.renene.2019.01.090.
[24] J. Few, C. Elwell, Open or closed ? use of windows and doors at home : ventilation
rates in occupied dwellings, in: ASHRAE Top. Conf. Proceedings. Am. Soc. Heating,
Refrig. Air Cond. Eng. Inc., 2021, pp. 1–8.
[25] C. Sekhar, M. Bivolarova, M. Akimoto, P. Wargocki, Detailed characterization of
bedroom ventilation during heating season in a naturally ventilated semi-detached
house and a mechanically ventilated apartment, Sci. Technol. Built Environ. 27
(2020) 158–180, https://doi.org/10.1080/23744731.2020.1845019.
[26] G. Bek¨
o, T. Lund, F. Nors, J. Toftum, G. Clausen, Ventilation rates in the bedrooms
of 500 Danish children, Build. Environ. 45 (2010) 2289–2295, https://doi.org/
10.1016/j.buildenv.2010.04.014.
[27] J. Hou, Y. Sun, Q. Chen, R. Cheng, J. Liu, X. Shen, H. Tan, H. Yin, K. Huang, Y. Gao,
X. Dai, L. Zhang, B. Liu, J. Sundell, Air change rates in urban Chinese bedrooms,
Indoor Air 29 (2019) 828–839, https://doi.org/10.1111/ina.12582.
[28] J. Hou, Y. Zhang, Y. Sun, P. Wang, Q. Zhang, X. Kong, J. Sundell, Air change rates
at night in northeast Chinese homes, Build. Environ. 132 (2018) 273–281, https://
doi.org/10.1016/j.buildenv.2018.01.030.
[29] P.L. Cheng, X. Li, Air inltration rates in the bedrooms of 202 residences and
estimated parametric inltration rate distribution in Guangzhou, China, Energy
Build. 164 (2018) 219–225, https://doi.org/10.1016/j.enbuild.2017.12.062.
[30] N. Canha, J. Lage, S. Candeias, C. Alves, S.M. Almeida, Indoor air quality during
sleep under different ventilation patterns, Atmos. Pollut. Res. 8 (2017) 1132–1142,
https://doi.org/10.1016/j.apr.2017.05.004.
[31] X. Xu, L. Lan, J. Shen, Y. Sun, Z. Lian, Five hypotheses concerned with bedroom
environment and sleep quality: a questionnaire survey in Shanghai city, China,
Build, Environ. Times 205 (2021), 108252, https://doi.org/10.1016/j.
buildenv.2021.108252.
[32] X. Zhang, G. Luo, J. Xie, J. Liu, Associations of bedroom air temperature and CO2
concentration with subjective perceptions and sleep quality during transition
seasons, Indoor Air 31 (2021) 1004–1017, https://doi.org/10.1111/ina.12809.
[33] B. Bjorvatn, J. Mrdalj, I.W. Saxvig, T. Aasnæs, S. Pallesen, S. Waage, Age and sex
differences in bedroom habits and bedroom preferences, Sleep Med. 32 (2017)
157–161, https://doi.org/10.1016/j.sleep.2017.01.003.
[34] A.K. Mishra, A.M. van Ruitenbeek, M.G.L.C. Loomans, H.S.M. Kort, Window/door
opening-mediated bedroom ventilation and its impact on sleep quality of healthy,
young adults, Indoor Air 28 (2018) 339–351, https://doi.org/10.1111/ina.12435.
[35] C. Liao, M. Delghust, P. Wargocki, J. Laverge, Effects of window opening on the
bedroom environment and resulting sleep quality, Sci. Technol. Built Environ. 27
(2021) 995–1015, https://doi.org/10.1080/23744731.2021.1927177.
[36] J. Laverge, A. Janssens, Analysis of the inuence of ventilation rate on sleep
pattern, Indoor Air (2011) 2011.
[37] C. Liao, X. Fan, M. Bivolarova, J. Laverge, C. Sekhar, M. Akimoto, A. Mainka,
L. Lan, P. Wargocki, A cross-sectional eld study of bedroom ventilation and sleep
quality in Denmark during the heating season, Build. Environ. (2022), 109557,
https://doi.org/10.1016/j.buildenv.2022.109557.
[38] D.J. Buysse, C.F. Reynolds, T.H. Monk, S.R. Berman, D.J. Kupfer, The Pittsburgh
sleep quality index: a new instrument for psychiatric practice and research,
Psychiatr. Res. 28 (1989) 193–213, https://doi.org/10.1016/0165-1781(89)
90047-4.
[39] X. Fan, M. Sakamoto, H. Shao, K. Kuga, K. Ito, L. Lan, P. Wargocki, Emission rate of
carbon dioxide while sleeping, Indoor Air 31 (2021) 2142–2157, https://doi.org/
10.1111/ina.12911.
[40] F. Moreno-Pino, A. Porras-Segovia, P. L´
opez-Esteban, A. Art´
es, E. Baca-García,
Validation of tbit charge 2 and tbit alta hr against polysomnography for
assessing sleep in adults with obstructive sleep apnea, J. Clin. Sleep Med. 15 (2019)
1645–1653, https://doi.org/10.5664/jcsm.8032.
[41] W. Mulder-Hajonides van der Meulen, J.R. Wijnberg, J.J. Hollander, I.P.F. De
Diana, R.H. Van den Hoofdakker, Measurement of subjective sleep quality, Eur.
Sleep Res. Soc. 98 (1980).
X. Fan et al.

Building and Environment 225 (2022) 109630
13
[42] A.D. Baddeley, A 3 min reasoning test based on grammatical transformation,
Psychonomic Sci. 10 (1968) 341–342, https://doi.org/10.3758/BF03331551.
[43] X. Fan, W. Liu, P. Wargocki, Physiological and psychological reactions of sub-
tropically acclimatized subjects exposed to different indoor temperatures at a
relative humidity of 70, Indoor Air 29 (2019) 215–230, https://doi.org/10.1111/
ina.12523.
[44] L. Lan, L. Pan, Z. Lian, H. Huang, Y. Lin, Experimental study on thermal comfort of
sleeping people at different air temperatures, Build. Environ. 73 (2014) 24–31,
https://doi.org/10.1016/j.buildenv.2013.11.024.
[45] P. Strøm-Tejsen, D.P. Wyon, D. Zukowska, A. Jama, Finger temperature as a
predictor of thermal comfort for sedentary passengers in a simulated aircraft cabin,
in: 9th Int. Heal. Build. Conf. Exhib., 2009, p. 382.
[46] American Society for Testing, Materials International, Standard Test Method for
Determining Air Change in a Single Zone by Means of a Tracer Gas Dilution, 2017,
https://doi.org/10.1520/E0741-11R17.ization. ASTM Standard E741-11.
[47] D. Laussmann, D. Helm, Air change measurements: using tracer gases, IntechOpen,
London, United Kingdom, in: N. Mazzeo (Ed.), Chem. Emiss. Control. Radioact.
Pollut. Indoor Air Qual., 2012, pp. 365–406, https://doi.org/10.5772/1030.
[48] L. Lan, Z. Lian, Application of statistical power analysis - how to determine the
right sample size in human health, comfort and productivity research, Build.
Environ. 45 (2010) 1202–1213, https://doi.org/10.1016/j.buildenv.2009.11.002.
[49] J. Cohen, Statistical Power Analysis for the Behavioral Sciences, second ed.,
Lawrence Erlbaum Associates, Mahwah, New Jersey, 1988.
[50] B.E. Boor, M.P. Spilak, J. Laverge, A. Novoselac, Y. Xu, Human exposure to indoor
air pollutants in sleep microenvironments: a literature review, Build. Environ. 125
(2017) 528–555, https://doi.org/10.1016/j.buildenv.2017.08.050.
[51] M. Ohayon, E.M. Wickwire, M. Hirshkowitz, S.M. Albert, A. Avidan, F.J. Daly,
Y. Dauvilliers, R. Ferri, C. Fung, D. Gozal, N. Hazen, A. Krystal, K. Lichstein,
M. Mallampalli, G. Plazzi, R. Rawding, F.A. Scheer, V. Somers, M. V Vitiello,
National Sleep Foundation’s sleep quality recommendations: rst report, Sleep
Heal 3 (2017) 6–19, https://doi.org/10.1016/j.sleh.2016.11.006.
[52] F. Wei, G. Nie, B. Zhou, L. Wang, Y. Ma, S. Peng, S. Ou, J. Qin, L. Zhang, S. Li,
R. Zou, X. Zeng, Z. Zhang, Y. Zou, Association between Chinese cooking oil fumes
and sleep quality among a middle-aged Chinese population, Environ. Pollut. 227
(2017) 543–551, https://doi.org/10.1016/j.envpol.2017.05.018.
[53] S. Lappharat, N. Taneepanichskul, S. Reutrakul, N. Chirakalwasan, Effects of
bedroom environmental conditions on the severity of obstructive sleep apnea,
J. Clin. Sleep Med. 14 (2018) 565–573, https://doi.org/10.5664/jcsm.7046.
[54] C.-C. Lo, W.-T. Liu, Y.-H. Lu, D. Wu, C.-D. Wu, T.-C. Chen, Y.-T. Fang, Y.-C. Lo, Y.-
Y. Chen, L. Kang, C.-Y. Tsai, Y.-L. Lee, K.-J. Chuang, K.-F. Ho, T.-Y. Chang, H.-
C. Chuang, Air pollution associated with cognitive decline by the mediating effects
of sleep cycle disruption and changes in brain structure in adults, Environ. Sci.
Pollut. Res. (2022), https://doi.org/10.1007/s11356-022-19482-7.
[55] G. Chen, H. Xiang, Z. Mao, W. Huo, Y. Guo, C. Wang, S. Li, Is long-term exposure to
air pollution associated with poor sleep quality in rural China? Environ. Int. 133
(2019), 105205 https://doi.org/10.1016/j.envint.2019.105205.
[56] M.E. Billings, D. Gold, A. Szpiro, C.P. Aaron, N. Jorgensen, A. Gassett, P.J. Leary, J.
D. Kaufman, S.R. Redline, The association of ambient air pollution with sleep
apnea: the multi-ethnic study of atherosclerosis, Ann. Am. Thorac. Soc. 16 (2019)
363–370, https://doi.org/10.1513/AnnalsATS.201804-248OC.
[57] X. Zhang, P. Wargocki, Z. Lian, C. Thyregod, Effects of exposure to carbon dioxide
and bioefuents on perceived air quality, self-assessed acute health symptoms, and
cognitive performance, Indoor Air 27 (2017) 47–64, https://doi.org/10.1111/
ina.12284.
[58] P. Wargocki, J.A. Porras-Salazar, S. Contreras-Espinoza, W. Bahneth, The
relationships between classroom air quality and children’s performance in school,
Build. Environ. 173 (2020), 106749, https://doi.org/10.1016/j.
buildenv.2020.106749.
[59] ASHRAE, ASHRAE Position Document on Indoor Carbon Dioxide, 2022, p. 17.
[60] W. Ye, X. Zhang, J. Gao, G. Cao, X. Zhou, X. Su, Indoor air pollutants, ventilation
rate determinants and potential control strategies in Chinese dwellings: a literature
review, Sci. Total Environ. 586 (2017) 696–729, https://doi.org/10.1016/j.
scitotenv.2017.02.047.
[61] E. Cooper, Y. Wang, S. Stamp, E. Burman, D. Mumovic, Use of portable air puriers
in homes: operating behaviour, effect on indoor PM2.5 and perceived indoor air
quality, Build. Environ. 191 (2021), 107621, https://doi.org/10.1016/j.
buildenv.2021.107621.
[62] S. Batterman, L. Du, G. Mentz, B. Mukherjee, E. Parker, C. Godwin, J.Y. Chin,
A. O’Toole, T. Robins, Z. Rowe, T. Lewis, Particulate matter concentrations in
residences: an intervention study evaluating stand-alone lters and air
conditioners, Indoor Air 22 (2012) 235–252, https://doi.org/10.1111/j.1600-
0668.2011.00761.x.
[63] P. Wargocki, Perceived air quality, sick building syndrome (SBS) symptoms and
productivity in an ofce with two different pollution loads, Indoor Air 9 (1999)
165–179, https://doi.org/10.1111/j.1600-0668.1999.t01-1-00003.x.
[64] P. Wargocki, D.P. Wyon, P.O. Fanger, The performance and subjective responses of
call-center operators with new and used supply air lters at two outdoor air supply
rates, Indoor Air 14 (2004) 7–16, https://doi.org/10.1111/j.1600-
0668.2004.00304.x.
[65] L. Fang, Impact of temperature and humidity on perception of indoor air quality
during immediate and longer whole-body exposures, Indoor Air 8 (1998) 276–284,
https://doi.org/10.1111/j.1600-0668.1998.00008.x.
[66] J. Laverge, M. Spilak, A. Novoselac, Experimental assessment of the inhalation
zone of standing, sitting and sleeping persons, Build. Environ. 82 (2014) 258–266,
https://doi.org/10.1016/j.buildenv.2014.08.014.
[67] J. Laverge, A. Novoselac, R. Corsi, A. Janssens, Experimental assessment of
exposure to gaseous pollutants from mattresses and pillows while asleep, Build.
Environ. 59 (2013) 203–210, https://doi.org/10.1016/j.buildenv.2012.08.020.
X. Fan et al.