Available via license: Public Domain Mark 1.0
Content may be subject to copyright.
Available via license: Public Domain Mark 1.0
Content may be subject to copyright.
Environmental Health Perspectives
•
v o l u m e 120
|
n u m b e r 12
|
December 2012
1671
Research
Because humans produce and exhale carbon
dioxide (CO
2
), concentrations of CO
2
in
occupied indoor spaces are higher than con-
centrations outdoors. As the ventilation rate
(i.e., rate of outdoor air supply to the indoors)
per person decreases, the magnitude of the
indoor–outdoor difference in CO
2
concen-
tration increases. Consequently, peak indoor
CO
2
concentrations, or the peak elevations
of the indoor concentrations above those in
outdoor air, have often been used as rough
indicators for outdoor-air ventilation rate per
occupant (Persily and Dols 1990). e need
to reduce energy consumption provides an
incentive for low rates of ventilation, leading
to higher indoor CO
2
concentrations.
Although typical outdoor CO
2
concentra-
tions are approximately 380 ppm, outdoor
levels in urban areas as high as 500 ppm have
been reported (Persily 1997). Concentrations
of CO
2
inside buildings range from outdoor
levels up to several thousand parts per million
(Persily and Gorfain 2008). Prior research
has documented direct health effects of CO
2
on humans, but only at concentrations much
higher than those found in normal indoor
settings. CO
2
concentrations > 20,000 ppm
cause deepened breathing; 40,000 ppm
increases respiration markedly; 100,000 ppm
causes visual disturbances and tremors and has
been associated with loss of consciousness; and
250,000 ppm CO
2
(a 25% concentration) can
cause death (Lipsett et al. 1994). Maximum
recommended occupational exposure limits
for an 8-hr workday are 5,000 ppm as a time-
weighted average, for the Occupational Safety
and Health Administration (OSHA 2012)
and the American Conference of Government
Industrial Hygienists (ACGIH 2011).
Epidemiologic and intervention research
has shown that higher levels of CO
2
within
the range found in normal indoor settings
are associated with perceptions of poor air
quality, increased prevalence of acute health
symptoms (e.g., headache, mucosal irrita-
tion), slower work performance, and increased
absence (Erdmann and Apte 2004; Federspiel
et al. 2004; Milton et al. 2000; Seppanen
et al. 1999; Shendell et al. 2004; Wargocki
et al. 2000). It is widely believed that these
associations exist only because the higher
indoor CO
2
concentrations at lower outdoor
air ventilation rates are correlated with higher
levels of other indoor-generated pollutants
that directly cause the adverse effects (Mudarri
1997; Persily 1997). us CO
2
in the range
of concentrations found in buildings (i.e.,
up to 5,000 ppm) has been assumed to have
no direct impacts on occupants’ perceptions,
health, or work performance.
Researchers in Hungary have questioned
this assumption (Kajtar et al. 2003, 2006).
e authors reported that controlled human
exposures to CO
2
between 2,000 ppm and
5,000 ppm, with ventilation rates unchanged,
had subtle adverse impacts on proofreading of
text in some trials, but the brief reports in con-
ference proceedings provided limited details.
is stimulated our group to test effects
of variation in CO
2
alone, in a controlled
environment, on potentially more sensitive
high-level cognitive functioning. We investi-
gated a hypothesis that higher concentrations
of CO
2
, within the range found in buildings
and without changes in ventilation rate, have
detrimental effects on occupants’ decision-
making performance.
Methods
is study addresses responses among human
participants under three different condi-
tions in a controlled environmental chamber
out fitted like an office, with CO
2
concen-
trations of approximately 600, 1,000, and
2,500 ppm. Six groups of four participants
were scheduled for exposure to each of the
three conditions for 2.5 hr per condition.
The experimental sessions for each group
took place on a single day, at 0900–1130,
1230–1500, and 1600–1830 hours, with 1-hr
breaks outside the exposure chamber between
sessions. During the first break, participants
ate a self-provided lunch. e order in which
participants were exposed to the different
CO
2
concentrations was balanced across
groups, including all possible orders of low-,
Address correspondence to M.J. Mendell, Lawrence
Berkeley National Laboratory, 1 Cyclotron Rd.,
90R3058, Berkeley, CA 94720 USA. Telephone:
(510) 486-5762. Fax: (510) 486-6658. E-mail:
mjmendell@lbl.gov
Funding for this research was provided by
Collaborative Activities for Research and Technology
Innovation (CARTI), which supports research in the
areas of air quality and water resource management.
CARTI, part of the Syracuse Center of Excellence
located in Syracuse, New York, is supported by the
U.S. Environmental Protection Agency under award
EM-83340401-0. Information about CARTI is
available at http://www.syracusecoe.org/coe/sub1.
html?skuvar=68.
e authors declare they have no actual or potential
competing financial interests.
Received 28 November 2011; accepted 20 September
2012.
Is CO
2
an Indoor Pollutant? Direct Effects of Low-to-Moderate CO
2
Concentrations on Human Decision-Making Performance
Usha Satish,
1
Mark J. Mendell,
2
Krishnamurthy Shekhar,
1
Toshifumi Hotchi,
2
Douglas Sullivan,
2
Siegfried Streufert,
1
and William J. Fisk
2
1
Department of Psychiatry and Behavioral Science, Upstate Medical University, State University of New York, Syracuse, New York, USA;
2
Indoor Environment Department, Lawrence Berkeley National Laboratory, Berkeley, California, USA
Ba c k g r o u n d : Associations of higher indoor carbon dioxide (CO
2
) concentrations with impaired
work performance, increased health symptoms, and poorer perceived air quality have been attrib-
uted to correlation of indoor CO
2
with concentrations of other indoor air pollutants that are also
influenced by rates of outdoor-air ventilation.
oB j e c t i v e s : We assessed direct effects of increased CO
2
, within the range of indoor concentrations,
on decision making.
M
e t h o d s : Twenty-two participants were exposed to CO
2
at 600, 1,000, and 2,500 ppm in an
office-like chamber, in six groups. Each group was exposed to these conditions in three 2.5-hr ses-
sions, all on 1 day, with exposure order balanced across groups. At 600 ppm, CO
2
came from out-
door air and participants’ respiration. Higher concentrations were achieved by injecting ultrapure
CO
2
. Ventilation rate and temperature were constant. Under each condition, participants com-
pleted a computer-based test of decision-making performance as well as questionnaires on health
symptoms and perceived air quality. Participants and the person administering the decision-making
test were blinded to CO
2
level. Data were analyzed with analysis of variance models.
re s u l t s : Relative to 600 ppm, at 1,000 ppm CO
2
, moderate and statistically significant decrements
occurred in six of nine scales of decision-making performance. At 2,500 ppm, large and statistically
significant reductions occurred in seven scales of decision-making performance (raw score ratios,
0.06–0.56), but performance on the focused activity scale increased.
co n c l u s i o n s : Direct adverse effects of CO
2
on human performance may be economically impor-
tant and may limit energy-saving reductions in outdoor air ventilation per person in buildings.
Confirmation of these findings is needed.
ke y w o r d s : carbon dioxide, cognition, decision making, human performance, indoor environ-
mental quality, ventilation. Environ Health Perspect 120:1671–1677 (2012). http://dx.doi.
org/10.1289/ehp.1104789 [Online 20 September 2012]
Satish et al.
1672
v o l u m e 120
|
n u m b e r 12
|
December 2012
•
Environmental Health Perspectives
medium-, and high-concentration sessions.
Participants and the person administering the
tests of decision-making performance were
not informed about specific CO
2
conditions
in each session. During each exposure condi-
tion, participants completed a computer-based
test of decision-making performance in which
they were presented with scenarios and asked
to make decisions based on a standardized
protocol (Krishnamurthy et al. 2009; Satish
et al. 2009; Streufert and Satish 1997). Before
and after each test of decision-making perfor-
mance, participants also completed computer-
based questionnaires on perceived indoor air
quality and health symptoms.
We received approval for the study proto-
col and the informed consent procedures from
the Human Subjects Committee at Lawrence
Berkeley National Laboratory (LBNL). We
recruited primarily from among a local popula-
tion of university students, all at least 18 years
old. We scheduled 24 participants, with extras
in case of no-shows, for participation. All
partici pants provided written informed consent
before participation. Scheduled partici pants
were provided a small amount of financial
compensation for their time.
Exposure protocol. Experimental ses-
sions were conducted in a chamber facility
at LBNL. e chamber has a 4.6 m × 4.6 m
floor plan, 2.4 m high ceiling, standard gyp-
sum board walls, and vinyl flooring, and is
equipped with four small desks, each with an
Internet-connected computer. The chamber
is located inside a heated and cooled build-
ing, with all external surfaces of the cham-
ber surrounded by room-temperature air. e
chamber has one window (~ 1 m × 1 m) that
views the interior of the surrounding indoor
space; hence, changes in daylight or the view
to outdoors were not factors in the research.
e chamber has a relatively airtight envelope,
including a door with a refrigerator-style seal.
e chamber was positively pressurized rela-
tive to the surrounding space. A small heating,
ventilating, and air-conditioning system served
the chamber with thermally conditioned air
filtered with an efficient particle filter. e out-
door air supply rate was maintained constant
at approximately 3.5 times the 7.1 L/sec per
person minimum requirement in California
(California Energy Commission 2008); the
flow rate was monitored continuously with
a venturi flow meter (model VWF 555 - 4”;
Gerand Engineering Co, Minneapolis, MN).
CO
2
was recorded in real time at 1-min
intervals. During the baseline sessions, with
participants and outdoor air as the only indoor
source of CO
2
, measured CO
2
concentrations
were approximately 600 ppm. In sessions with
CO
2
added, CO
2
from a cylinder of ultra-
pure CO
2
(at least 99.9999% pure) was added
to the chamber supply air, upstream of the
supply-air fan to assure mixing of the CO
2
in
the air, at the rate needed to increase the CO
2
concentration to either 1,000 or 2,500 ppm.
A mass flow controller monitored and regu-
lated injection rates in real time. All other con-
ditions (e.g., ventilation rate, temperature)
remained unchanged.
e outdoor air exchange rate of the cham-
ber was about 7/hr; and in sessions with CO
2
injected into the chamber, injection started
before the participants entered the chamber. In
sessions with no CO
2
injection, CO
2
concen-
trations were close to equilibrium levels 25 min
after the start of occupancy, and in sessions
with CO
2
injection (because CO
2
injection
started before participants entered the cham-
ber), 10–15 min after the start of occupancy.
Before participants entered the chamber,
the desired chamber temperature and ventila-
tion rate were established at target values of
23
o
C (73
o
F) and 100 L/sec (210 ft
3
/min).
Indoor chamber temperature during the experi-
mental sessions was maintained at approxi-
mately 23
o
C (73.4
o
F) by proportionally
controlled electric resistance heating in the
supply airstream. Relative humidity (RH) was
approximately 50% ± 15%. We continuously
monitored temperature and RH in real time.
Temperature was averaged for each session
for comparisons.
Calibrations of all instruments were
checked at the start of the study. Calibration
of the CO
2
monitors was checked at least
every week during experiments using primary
standard calibration gases. Given the instru-
ments used and calibration procedures, we
anticipated measurement accuracies of ± 5%
at the lowest CO
2
concentrations and as high
as ± 3% at the highest concentrations. Real-
time logged environmental data (CO
2
, tem-
perature, RH, outdoor air supply rate) were
downloaded from environmental monitors to
Excel and imported into SAS statistical analy-
sis software (version 9.1; SAS Institute Inc.,
Cary, NC).
The design of the CO
2
injection system
included features to prevent unsafe CO
2
con-
centrations from developing in the event of a
failure in the CO
2
injection system or human
error. e CO
2
cylinder was outdoors so that
any leaks would be to outdoors. A pressure
relief valve located downstream of the pressure
regulator was also located outdoors and set to
prevent pressures from exceeding our target
pressure at the inlet of the mass flow controller
by > 50%. Valves would automatically stop
CO
2
injection if the outdoor air ventilation
to the chamber or the ventilation fan failed.
A flow limiter prevented CO
2
concentrations
from exceeding 5,000 ppm if the mass flow
controller failed in the fully open position,
and a second CO
2
analyzer with control sys-
tem would automatically stop CO
2
injection
if the concentration exceeded 5,000 ppm.
Also, a research associate monitored CO
2
concentrations in the chamber using a real-
time instrument. Given the purity level of the
carbon dioxide in the gas cylinder (99.9999%)
and the rate of outdoor air supply to the cham-
ber, the maximum possible chamber air con-
centration of impurities originating from the
cylinder of CO
2
was only 2 ppb. e impurity
of highest concentration was likely to be water
vapor, and at a concentration ≤ 2 ppb, short-
term health risks from exposures to impurities
would have been far less than risks associated
with exposures to many normal indoor or out-
door pollutants. Finally, before participants
entered the chamber we added CO
2
from the
cylinder to the chamber air, and collected an
air sample on a sorbent tube for analysis by
thermal desorption gas chromatography mass
spectrometry. ere was no evidence that the
CO
2
injection process increased indoor con-
centrations of volatile organic compounds
(VOCs). VOCs at low concentrations, typi-
cal of indoor and outdoor air concentrations,
were detected.
On the morning of each of 6 experi-
mental days, groups of participants came to
LBNL for a full day of three experimental ses-
sions. To ensure a full set of four participants
for each scheduled day (after one unantici-
pated no-show on each of the first 2 days),
we scheduled five participants each day and
selected four at random to participate. On
each experimental day, as soon as all partici-
pants had arrived, the selected participants
were seated in the environmental chamber
facility. Before they entered the chamber, a
research associate distributed to participants
a handout describing the session plans and
answered any questions.
During the first 45 min of each session,
participants were free to perform school work,
read, or engage in any quiet, nondisruptive
activity. Participants were then asked by the
LBNL research associate to complete the
computer-based questionnaire on perceived
air quality and symptoms, available via web
connection on the laptop computers on their
desks. Participants then had a 10-min break,
to stretch or exit the chamber to use the bath-
room, but no participant elected to exit the
chamber during a session.
A 20-min protocol was then used to train
participants in the decision-making task. A
technician trained in administering this test
was present to answer questions before the
test, and could enter the chamber to answer
questions during the test. We estimated that
CO
2
emissions of the technician, who was in
the chamber for about 10 min during each
session, would increase chamber CO
2
con-
centrations by no more than 17 ppm. (e
technician was not required to give informed
consent for this because the study conditions
are commonly experienced in indoor environ-
ments and are not associated with adverse
Effects of CO
2
on decision-making performance
Environmental Health Perspectives
•
v o l u m e 120
|
n u m b e r 12
|
December 2012
1673
health effects.) Over the next 1.5 hr, partici-
pants took the computerized test of decision-
making performance, which involved reading
text displayed on a laptop computer and
selecting among possible responses to indicate
their decisions.
When the performance test was com-
pleted, participants repeated the computer-
based questionnaire on perceived air quality
and symptoms and then left the chamber until
the next session. At any time during each ses-
sion, participants were free to exit the facility
to use a nearby bathroom, but were asked to
return within 10 min. Participants were also
free to terminate their participation and leave
the facility at any time during the day, but no
participants exercised these options.
Testing of decision-making performance.
We used a testing method designed to assess
complex cognitive functioning in ways more
relevant to the tasks of workers in build-
ings than the tests of simulated office work
generally used (e.g., proofreading text, add-
ing numbers) (Wargocki et al. 2000). A
computer-based program called the Strategic
Management Simulation (SMS) test col-
lects data on performance in decision making
under different conditions. e SMS test has
been used to study the impact on people’s
decision-making abilities of different drugs,
VOCs from house painting, stress overload,
head trauma, and the like (Breuer and Satish
2003; Cleckner 2006; Satish et al. 2004, 2008;
Swezey et al. 1998). (SMS testing is available
for research by contract with State University
of New York Upstate Medical University,
and for commercial applications via Streufert
Consulting, LLC. See http://www.upstate.
edu/psych/research/sms.php.)
e SMS measures complex human behav-
iors required for effectiveness in many work-
place settings. e system assesses both basic
cognitive and behavioral responses to task
demands, as well as cognitive and behavioral
components commonly considered executive
functions. The system and its performance
have been described in prior publications (e.g.,
Breuer and Satish 2003; Satish et al. 2004;
Swezey et al. 1998). Participants are exposed to
diverse computer-generated situations present-
ing real-world equivalent simulation scenarios
that are proven to match real-world day-to-day
challenges. Several paral lel scenarios are avail-
able, allowing retesting individuals without
bias due to experience and learning effects.
Participants are given instructions via text mes-
sages on a user-friendly computer interface, and
respond to the messages using a drop-down
menu of possible decisions. All participants
receive the same quantity of information at
fixed time points in simulated time, but partici-
pants have flexibility to take actions and make
decisions at any time during the simulation, as
in the real world. e absence of requirements
to engage in specific actions or to make deci-
sions at specific points in time, the absence of
stated demands to respond to specific informa-
tion, the freedom to develop initiative, and the
freedom for strategy develop ment and deci-
sion implementation allow each participant to
use his or her own preferred or typical action,
planning, and strategic style. The SMS sys-
tem generates measure ment profiles that reflect
the underlying decision-making capacities of
the individual.
e computer calculates SMS performance
measures as raw scores, based on the actions
taken by the participants, their stated future
plans, their responses to incoming information,
and their use of prior actions and outcomes.
The validated measures of task performance
vary from relatively simple competencies such
as speed of response, activity, and task orienta-
tion, through intermediate level capabilities
such as initiative, emergency responsiveness,
and use of information, to highly complex
thought and action processes such as breadth
of approach to problems, planning capac-
ity, and strategy. The nine primary factors
and factor combinations that have predicted
real-world success are basic activity level (num-
ber of actions taken), applied activity (oppor-
tunistic actions), focused activity (strategic
actions in a narrow endeavor), task orientation
(focus on concurrent task demands), initia-
tive (development of new/creative activities),
information search (openness to and search for
information), information usage (ability to use
information effectively), breadth of approach
(flexibility in approach to the task), and basic
strategy (number of strategic actions).
e raw scores assigned for each measure
are linearly related to performance, with a
higher score indicating superior performance.
Interpretation is based on the relationship to
established standards of performance excel-
lence among thousands of previous SMS par-
ticipants (Breuer and Streufert 1995; Satish
et al. 2004, 2008; Streufert and Streufert
1978; Streufert et al. 1988; Streufert and
Swezey 1986). Percentile ranks are calculated
through a comparison of raw scores to the
overall distribution of raw scores from a ref-
erence population of > 20,000 U.S. adults,
16–83 years of age, who had previously com-
pleted the SMS. e reference population was
constructed nonrandomly to be generally rep-
resentative of the job distribution among the
adult U.S. population, including, for example,
college students, teachers, pilots, medical resi-
dents, corporate executives, homemakers, and
the unemployed. e percentile calculations
for individual participants are not further
adjusted for age, sex, or education level.
Data management and analysis. e main
predictor variable of interest was CO
2
, included
in analyses as a categorical variable with three
values: 600, 1,000, and 2,500 ppm. Real-time
CO
2
concentrations and temperature were
averaged for each session for comparison.
Nine measures from the SMS, representing
validated independent assessments of
performance in complex task settings, were
compared across CO
2
conditions. Raw scores
on the different SMS measures were computer-
calculated based on procedures (software
formulas) that are discussed by Streufert and
Swezey (1986). The formulas are based on
numerically and graphically scored decision
actions, on the interrelationships among
decisions over time, the interrelationships
among decisions with incoming information,
as well as decision planning and other
components of participant activity. Each of
the activity event components that are used
in the formulas are collected by the SMS
computer software program (Streufert and
Swezey 1986). A separate SMS software system
is subsequently used to calculate the value for
each measure. Where appropriate—where
maximum performance levels have limits
(cannot be exceeded)—the obtained scores
are expressed by the program as percentages of
maximally obtainable values. A higher score on
a measure indicates better performance in that
area of performance. For each measure, ratios
of scores across conditions were calculated to
show the magnitude of changes.
Initial data analysis used multivariate
analy sis of variance (MANOVA) to assess
overall significance across all conditions,
to assure that subsequent (post hoc) analy-
sis across the nine different simulation mea-
sures would be legitimate. With high levels
of significance established, post hoc analysis
for each simulation measure using analysis
of variance (ANOVA) techniques becomes
possible. Separate ANOVA procedures across
CO
2
conditions were used for each of the
nine SMS measures (within participants, with
participants as their own controls). Percentile
ranks were calculated from the raw scores
and normative data, without adjustments for
demographic or other variables. Percentile
levels are divided into categories with descrip-
tive labels based on prior test findings from
different populations, normal and impaired.
Results
Because 2 of the 24 originally scheduled
partici pants cancelled at a time when they
could not be replaced, 22 participants pro-
vided complete SMS data. Of these, 10 were
male; 18 were 18–29 years of age, and 4 were
30–39 years of age. One participant had com-
pleted high school only, 8 had completed
some college, and 13 had a college degree.
None were current smokers, 1 reported cur-
rent asthma, and 5 reported eczema, hay
fever, or allergy to dust or mold.
Median CO
2
values for the low, medium,
and high CO
2
conditions were 600, 1,006,
Satish et al.
1674
v o l u m e 120
|
n u m b e r 12
|
December 2012
•
Environmental Health Perspectives
and 2,496 ppm (which we refer to as 600,
1,000, and 2,500 ppm), and ranges were 132,
92, and 125 ppm, respectively (Table 1).
Temperatures in the study chamber were
controlled effectively, varying overall within
about 0.2
o
C (from 22.9 to 23.1
o
C in each
condition), and with median values across the
three CO
2
conditions varying < 0.1
o
C.
e raw scores for each of the SMS perfor-
mance measures were plotted for each partici-
pant according to CO
2
level (Figure 1). e
plots indicate clear relationships between raw
scores and CO
2
level for all performance mea-
sures other than focused activity and informa-
tion search, with dramatic reductions in raw
scores at 2,500 ppm CO
2
for some measures
of decision-making performance.
For seven of nine scales of decision-making
performance (basic activity, applied activity,
task orientation, initiative, information usage,
breadth of approach, and basic strategy), mean
raw scores showed a consistently monotonic
decrease with increasing CO
2
concentrations,
with all overall p
-values < 0.001 (Table 2). In
post hoc pairwise comparisons by CO
2
con-
centration, performance on these seven scales
differed between concentrations with p
< 0.01
for all comparisons, except for performance on
the task orientation, initiative, and basic strat-
egy scales between 600 and 1,000 ppm CO
2
(p
< 0.05, p < 0.10, and p < 0.05, respectively)
(Table 3). For these seven scales, compared with
mean raw scores at 600 ppm CO
2
, mean raw
scores at 1,000 ppm CO
2
were 11–23% lower,
and at 2,500 ppm CO
2
were 44–94% lower.
Relative to raw scores at 1,000 ppm CO
2
, raw
scores at 2,500 ppm were 35–93% lower.
For information search, mean raw scores
were similar at all three CO
2
conditions.
Neither the overall analysis across the three
conditions (Table 2) nor the post hoc pair-
wise analyses (Table 3) indicated significant
differences. For focused activity, raw scores
at 600 ppm CO
2
and 1,000 ppm CO
2
were
nearly identical (16.27 and 16.09), but the
mean raw score at 2,500 ppm was higher
(19.55), resulting in an overall p-value
≤ 0.001 (Table 2). Post hoc tests indicated
no difference between mean raw scores at 600
and 1,000 ppm CO
2
, but significant differ-
ences (p
≤ 0.01) between the mean raw score
at 2,500 ppm CO
2
and scores at both 600
and 1,000 ppm (Table 3).
Figure 2 shows the percentile scores on
the nine scales at the three CO
2
conditions
(based on the raw scores shown in Table 2),
with the percentile boundaries for five nor-
mative levels of performance: superior, very
good, average, marginal, and dysfunctional.
At 1,000 ppm CO
2
relative to 600 ppm, per-
centile ranks were moderately diminished at
most. However, at 2,500 ppm CO
2
, percen-
tile ranks for five performance scales decreased
to levels associated with marginal or dysfunc-
tional performance.
Discussion
Synthesis and interpretation of findings.
Performance for six of nine decision-making
measures decreased moderately but signifi-
cantly at 1,000 ppm relative to the baseline of
Table 1. CO
2
concentrations during study
conditions.
CO
2
condition
CO
2
concentration (ppm)
Minimum Median Maximum Range
Low 542 600 675 132
Medium 969 1,006 1,061 92
High 2,418 2,496 2,543 125
Overall 542 1,006 2,543 —
Figure 1. Plots of individual scores, by condition, for each of the SMS measures of decision-making performance (n= 22 subjects).
100
80
60
40
20
0
200
150
100
50
0
200
150
100
50
0
25
20
15
10
5
0
30
20
10
0
30
20
10
0
30
20
10
0
30
20
10
0
15
10
5
0
5102015 25 5102015 25 5102015 25
5102015 25 5102015 25 5102015 25
5102015 25 5102015 25 5102015 25
Basic activity Applied activity Focused activity
Task orientation Initiative Information search
Information usage Breadth of approach Basic strategy
ScoreScoreScore
ScoreScoreScore
ScoreScoreScore
600 ppm 1,000 ppm 2,500 ppm
Subject Subject Subject
Subject Subject Subject
Subject Subject Subject
Effects of CO
2
on decision-making performance
Environmental Health Perspectives
•
v o l u m e 120
|
n u m b e r 12
|
December 2012
1675
600 ppm, and seven decreased substantially at
2,500 ppm. For an eighth scale, “information
search,” no significant differences were seen
across conditions. In contrast to other scales,
an inverse pattern was seen for “focused activ-
ity,” with the highest level of focus obtained
at 2,500 ppm and the lowest at 600 ppm.
Thus, most decision-making variables
showed a decline with higher concentrations
of CO
2
, but measures of focused activity
improved. Focused activity is important for
overall productivity, but high levels of focus
under nonemergency conditions may indicate
“overconcentration.” Prior research with the
SMS has shown repeatedly that individuals
who experience difficulty in functioning [e.g.,
persons with mild-to-moderate head injuries
(Satish et al. 2008), persons under the influ-
ence of alcohol (Streufert et al. 1993), and
persons suffering from allergic rhinitis (Satish
et al. 2004)] tend to become highly focused on
smaller details at the expense of the big picture.
High levels of predictive validity for
the SMS (r
> 0.60 with real-world success
as judged by peers and as demonstrated by
income, job level, promotions, and level in
organizations), as well as high levels of test–
retest reliability across the four simulation sce-
narios (r
= 0.72–0.94) have repeatedly been
demonstrated (Breuer and Streufert 1995;
Streufert et al. 1988). Additional validity is
demonstrated by the deterioration of various
performance indicators with 0.05% blood
alcohol intoxication and seriously diminished
functioning with intoxication at the 0.10 level
(Satish and Streufert 2002). Baseline scores
at 600 ppm CO
2
for the participants in this
study, mostly current science and engineering
students from a top U.S. university, were all
average or above.
Although the modest reductions in
multiple aspects of decision making seen at
1,000 ppm may not be critical to individuals,
at a societal level or for employers an exposure
that reduces performance even slightly could
be economically significant. e substantial
reductions in decision-making performance
with 2.5-hr exposures to 2,500 ppm CO
2
indicate, per the available norms for the SMS
test, impairment that is of importance even
for individuals. ese findings provide initial
evidence for considering CO
2
as an indoor
pollutant, not just a proxy for other pollut-
ants that directly affect people.
CO
2
concentrations in practice. The
real-world significance of our findings, if
confirmed, would depend on the extent to
which CO
2
concentrations are ≥ 1,000 and
≥ 2,500 ppm in current or future buildings.
ere is strong evidence that in schools, CO
2
concentrations are frequently near or above
the levels associated in this study with sig-
nificant reductions in decision-making per-
formance. In surveys of elementary school
classrooms in California and Texas, average
CO
2
concentrations were > 1,000 ppm, a sub-
stantial proportion exceeded 2,000 ppm, and
in 21% of Texas classrooms peak CO
2
con-
centration exceeded 3,000 ppm (Corsi et al.
2002; Whitmore et al. 2003). Given these
concentrations, we must consider the possibil-
ity that some students in high-CO
2
classrooms
are disadvantaged in learning or test taking.
We do not know whether exposures that cause
decrements in decision making in the SMS
test will inhibit learning by students; how-
ever, we cannot rule out impacts on learning.
We were not able to identify CO
2
measure-
ments for spaces in which students take tests
related to admission to universities or graduate
schools, or from tests related to professional
accreditations, but these testing environments
often have a high occupant density, and thus
might have elevated CO
2
levels.
Table 2. Mean raw scores for nine outcome variables at three conditions of CO
2
concentration among
22 participants, and comparison using MANOVA.
Outcome variables
Conditions (ppm of CO
2
) (mean ± SD)
Overall
F-statistic
(df = 2,42) p-Value600 ppm 1,000 ppm 2,500 ppm
Basic activity 69.59 ± 7.04 59.23 ± 7.12 38.77 ± 7.57 172.77 < 0.001
Applied activity 117.86 ± 39.28 97.55 ± 35.51 62.68 ± 31.86 72.13 < 0.001
Focused activity 16.27 ± 3.20 16.09 ± 3.70 19.55 ± 3.40 17.26 < 0.001
Task orientation 140.82 ± 28.66 125.41 ± 28.62 50.45 ± 31.66 115.08 < 0.001
Initiative 20.09 ± 6.96 16.45 ± 6.70 1.41 ± 1.26 81.45 < 0.001
Information search 20.36 ± 3.06 21.5 ± 3.20 20.91 ± 3.08 2.51 > 0.10
Information usage 10.32 ± 3.21 7.95 ± 2.24 3.18 ± 1.71 129.20 < 0.001
Breadth of approach 9.36 ± 1.36 7.82 ± 1.56 2.32 ± 1.17 679.88 < 0.001
Basic strategy 27.23 ± 5.48 23.95 ± 5.65 1.68 ± 1.32 414.51 < 0.001
df, degrees of freedom.
Table 3. Comparison of mean raw scores for nine decision-making measures between three different CO
2
concentrations among 22 participants.
Variables
Ratios of condition scores
a
Score at 1,000 ppm/
score at 600 ppm
Score at 2,500 ppm/
score at 1,000 ppm
Score at 2,500 ppm/
score at 600 ppm
Basic activity 0.85
#
0.65
#
0.56
#
Applied activity 0.83
#
0.64
#
0.53
#
Focused activity 0.99 1.22
#
1.20
#
Task orientation 0.89** 0.40
#
0.36
#
Initiative 0.82* 0.09
#
0.07
#
Information search 1.06 0.97 1.03
Information usage 0.77
#
0.40
#
0.31
#
Breadth of approach 0.84
#
0.30
#
0.25
#
Basic strategy 0.88** 0.07
#
0.06
#
df, degrees of freedom.
a
p-Values based on F-test, df = 1,21, calculated for difference between score in numerator and score in denominator.
*p < 0.10. **p < 0.05.
#
p < 0.01.
Figure 2. Impact of CO
2
on human decision-making performance. Error bars indicate 1 SD.
Superior
Average
Marginal
Dysfunctional
Very good
95th percentile
75th percentile
50th percentile
25th percentile
Basic
activity
Applied
activity
Focused
activity
Information
search
Initiative
600 ppm CO
2
1,000 ppm CO
2
2,500 ppm CO
2
Task
orientation
Information
usage
Breadth of
approach
Basic
strategy
Satish et al.
1676
v o l u m e 120
|
n u m b e r 12
|
December 2012
•
Environmental Health Perspectives
In general office spaces within the United
States, CO
2
concentrations tend to be much
lower than in schools. In a representative survey
of 100 U.S. offices (Persily and Gorfain 2008),
only 5% of the measured peak indoor CO
2
concentrations exceeded 1,000 ppm, assuming
an outdoor concentration of 400 ppm. One
very small study suggests that meeting rooms
in offices, where important decisions are some-
times made, can have elevated CO
2
concentra-
tions—for example, up to 1,900 ppm during
30- to 90-min meetings (Fisk et al. 2010).
In some vehicles (aircraft, ships, sub-
marines, cars, buses, and trucks), because of
their airtight construction or high occupant
density, high CO
2
concentrations may be
expected. In eight studies within commercial
aircraft, mean CO
2
concentrations in the pas-
senger cabins were generally > 1,000 ppm and
ranged as high as 1,756 ppm, and maximum
concentrations were as high as 4,200 ppm
(Committee on Air Quality in Passenger
Cabins of Commercial Aircraft 2002). We
did not identify data on CO
2
concentrations
in automobiles and trucks. One small study
(Knibbs et al. 2008) reported low ventilation
rates in vehicles with ventilation systems in
the closed or recirculated-air positions. From
those results, and using an assumption of one
occupant and a 0.0052 L/sec CO
2
emission
rate per occupant (Persily and Gorfain 2008),
we estimated steady-state CO
2
concentra-
tions in an automobile and pickup truck of
3,700 ppm and 1,250 ppm, respectively,
above outdoor concentrations. ese numbers
would increase in proportion to the num-
ber of occupants. It is not known whether
the findings of the present study apply to the
decision making of vehicle drivers, although
such effects are conceivable.
There is evidence that people wearing
masks for respiratory protection may inhale
air with highly elevated CO
2
concentrations.
In a recent study, dead-space CO
2
concentra-
tions within a respirator (i.e., N95 mask) were
approximately 30,000 ppm (Roberge et al.
2010), suggesting potentially high CO
2
con-
centration in inhaled air. e inhaled concen-
tration would be lower than that within the
mask, diluted by approximately 500 mL per
breath inhaled through the mask. Although
the study did not report the actual inhaled-air
CO
2
concentrations, partial pressures of CO
2
in blood did not differ with wearing the mask.
Caretti (1999) reported that respirator wear
with low-level activity did not adversely alter
cognitive performance or mood.
Findings by others. e Hungarian studies
briefly reported by Kajtar et al. (2003, 2006)
were the only prior studies on cognitive effects
of moderate CO
2
elevations that we identi-
fied. In these studies, the ventilation rate in an
experimental chamber was kept constant at a
level producing a chamber CO
2
concentration
of 600 ppm from the occupant-generated
CO
2
; in some experiments, however, the
chamber CO
2
concentration was increased
above 600 ppm, to as high as 5,000 ppm,
by injecting 99.995% pure CO
2
from a gas
cylinder into the chamber. In two series of
studies, participants blinded to CO
2
concen-
trations performed proofreading significantly
more poorly in some but not all sessions with
CO
2
concentrations of 4,000 ppm relative to
600 ppm. Similar, marginally significant dif-
ferences were seen at 3,000 versus 600 ppm.
(Differences were seen only in proportion of
errors found, not in speed of reading.) The
studies by Kajtar et al. (2003, 2006) were
small (e.g., 10 participants) and found only a
few significant associations out of many trials;
these results may have been attributable to
chance, but they did suggest that CO
2
con-
centrations found in buildings may directly
influence human performance. Our research,
which was motivated by the Hungarian stud-
ies, involved lower concentrations of CO
2
, a
larger study population, and different methods
to assess human performance.
Prior studies on CO
2
exposures, mostly
at higher levels, have focused on physiologic
effects. CO
2
is the key regulator of respiration
and arousal of behavioral states in humans
(Kaye et al. 2004). e initial effects of inhal-
ing CO
2
at higher concentrations are increased
partial pressure of CO
2
in arterial blood
(PaCO
2
) and decreased blood pH. However,
PaCO
2
is tightly regulated in healthy humans
through reflex control of breathing, despite
normal variation within and between indi-
viduals (Bloch-Salisbury et al. 2000). Inhaled
CO
2
at concentrations of tens of thousands
of parts per million has been associated with
changes in respiration, cerebral blood flow,
cardiac output, and anxiety (Brian 1998; Kaye
et al. 2004; Lipsett et al. 1994; Roberge et al.
2010; Woods et al. 1988). Little research has
documented physiological impacts of moder-
ately elevated CO
2
concentrations, except one
small study that reported changes in respira-
tion, circulation, and cerebral electrical activity
at 1,000 ppm CO
2
(Goromosov 1968).
We do not have hypotheses to explain why
inhaling moderately elevated CO
2
, with the
expected resulting increases in respiration, heart
rate, and cardiac output to stabilize PaCO
2
,
would affect decision-making performance.
Bloch-Salisbury et al. (2000) have summarized
prior knowledge on effects of elevated PaCO
2
.
PaCO
2
has a direct linear relationship with
cerebral blood flow in a broad range above and
below normal levels, through dilation and con-
striction of arterioles. Moderately elevated (or
reduced) PaCO
2
has dramatic effects on central
nervous system and cortical function. Bloch-
Salisbury et al. (2000) reported that experi-
mental changes in PaCO
2
in humans within
the normal range (in 2-hr sessions involving
special procedures to hold respiration constant
and thus eliminate the normal reflex control of
PaCO
2
through altered breathing), showed no
effects on cognitive function or alertness but
caused significant changes in electroencephalo-
gram power spectra.
Limitations. This study successfully con-
trolled the known environmental confounding
factors of temperature and ventilation rate.
Although exposures to CO
2
in prior sessions
may theoretically have affected performance
in subsequent sessions, such carryover effects
should not invalidate study results because of
the balanced order of exposures. Suggestion
effects were unlikely, because participants and
the researcher explaining the SMS to them
were blinded to specific conditions of each ses-
sion. Although we conclude that the causality
of the observed effects is clear, the ability to
generalize from this group of college/university
students to others is uncertain. Effects of CO
2
between 600 and 1,000 ppm and between
1,000 and 2,500 ppm, and effects for longer
and shorter periods of time are also uncertain.
e strength of the effects seen at 2,500 ppm
CO
2
is so large for some metrics as to almost
defy credibility, although it is possible that
such effects occur without recognition in daily
life. Replication of these study findings, includ-
ing use of other measures of complex cogni-
tive functioning and measures of physiologic
response such as respiration and heart rate, is
needed before definitive conclusions are drawn.
Implications for minimum ventilation
standards. e findings of this study, if rep-
licated, would have implications for the stan-
dards that specify minimum ventilation rates in
buildings, and would also indicate the need to
adhere more consistently to the existing stan-
dards. Many of the elevated CO
2
concentra-
tions observed in practice are a consequence of
a failure to supply the amount of outdoor air
specified in current standards; however, even
the minimum ventilation rates in the leading
professional standard [American Society of
Heating, Refrigerating and Air-Conditioning
Engineers (ASHRAE) 2010] correspond to
CO
2
concentrations > 1,000 ppm in densely
occupied spaces. There is current interest in
reducing ventilation rates and the rates required
by standards, to save energy and reduce energy-
related costs. Yet large reductions in ventilation
rates could lead to increased CO
2
concentra-
tions that may adversely affect decision-making
performance, even if air-cleaning systems or
low-emission materials were used to control
other indoor pollutants. It seems unlikely that
recommended minimum ventilation rates
in future standards would be low enough to
cause CO
2
levels > 2,500 ppm, a level at which
decre ments in decision-making performance
in our findings were large, but standards with
rates that result in 1,500 ppm of indoor CO
2
are conceivable.
Effects of CO
2
on decision-making performance
Environmental Health Perspectives
•
v o l u m e 120
|
n u m b e r 12
|
December 2012
1677
Conclusions
Increases in indoor CO
2
concentrations result-
ing from the injection of ultrapure CO
2
, with
all other factors held constant, were associated
with statistically significant and meaningful
reductions in decision-making performance.
At 1,000 ppm CO
2
, compared with 600 ppm,
performance was significantly diminished on
six of nine metrics of decision-making per-
formance. At 2,500 ppm CO
2
, compared
with 600 ppm, performance was significantly
reduced in seven of nine metrics of perfor-
mance, with percentile ranks for some perfor-
mance metrics decreasing to levels associated
with marginal or dysfunctional performance.
The direct impacts of CO
2
on performance
indicated by our findings may be economically
important, may disadvantage some individuals,
and may limit the extent to which outdoor air
supply per person can be reduced in buildings
to save energy. Confirmation of these findings
is needed.
Re f e R e n c e s
ACGIH (American Conference of Governmental Industrial
Hygienists). 2011. TLVs and BEIs. Cincinnati, OH:American
Conference of Governmental Industrial Hygienists.
ASHRAE (American Society of Heating, Refrigerating and Air-
Conditioning Engineers). 2010. ANSI/ASHRAE Standard
62.1-2010: Ventilation for Acceptable Indoor Air Quality.
Atlanta, GA:ASHRAE.
Bloch-Salisbury E, Lansing R, Shea SA. 2000. Acute changes
in carbon dioxide levels alter the electroencephalogram
without affecting cognitive function. Psychophysiology
37(4):418–426.
Breuer K, Satish U. 2003. Emergency management simulations: an
approach to the assessment of decision-making processes
in complex dynamic crisis environments. In: From Modeling
To Managing Security: A System Dynamics Approach
(González JJ, ed). Kristiansand, Norway:Norwegian
Academic Press, 145–156.
Breuer K, Streufert S. 1995. The strategic management simula-
tion (SMS): a case comparison analysis of the German SMS
version. In: Corporate Training for Effective Performance
(Mulder M, Brinkerhoff RO, eds). Boston:Kluwer, 1–17.
Brian JE Jr. 1998. Carbon dioxide and the cerebral circulation.
Anesthesiology 88(5):1365–1386.
California Energy Commission. 2008. 2008 Building Energy
Efficiency Standards for Residential and Nonresidential
Buildings. CEC-400-2008-001-CMF. Sacramento, CA:California
Energy Commission.
Caretti DM. 1999. Cognitive performance and mood during respi-
rator wear and exercise. Am Ind Hyg Assoc J 60(2):213–218.
Cleckner LB. 2006. Pilot study of using strategic management
simulation to assess human productivity. In: Proceedings
of Indoor Environmental Quality: Problems, Research, and
Solutions Conference 2006, 17–19 July 2006, Durham, NC.
Red Hook, NY:Air and Waste Management Association;
Curran Associates, Inc., 973–981.
Committee on Air Quality in Passenger Cabins of Commercial
Aircraft. 2002. The Airliner Cabin Environment and the
Health of Passengers and Crew. Washington, DC:National
Research Council.
Corsi RL, Torres VM, Sanders M, Kinney KL. 2002. Carbon
dioxide levels and dynamics in elementary schools: results
of the TESIAS Study. In: Indoor Air 2002, 9th International
Conference on Indoor Air Quality and Climate (Levin H,
ed). Monterey, CA:Indoor Air, 74–79.
Erdmann CA, Apte MG. 2004. Mucous membrane and lower
respiratory building related symptoms in relation to indoor
carbon dioxide concentrations in the 100-building BASE
dataset. Indoor Air 14(s8):127–134.
Federspiel CC, Fisk WJ, Price PN, Liu G, Faulkner D, Dibartolomeo
DL, et al. 2004. Worker performance and ventilation in a call
center: analyses of work performance data for registered
nurses. Indoor Air 14(s8):41–50.
Fisk WJ, Sullivan DP, Faulkner D, Eliseeva E. 2010. CO
2
Monitoring for Demand Controlled Ventilation in Commercial
Buildings. LBNL-3279E. Berkeley, CA:Lawrence Berkeley
National Laboratory.
Goromosov MS. 1968. The Physiological Basis of Health
Standards for Dwellings. Geneva:World Health Organization.
Kajtar L, Herczeg L, Lang E. 2003. Examination of influence of
CO
2
concentration by scientific methods in the laboratory.
In: Proceedings of Healthy Buildings 2003, 7–11 December
2003. Singapore:Stallion Press, 176–181.
Kajtar L, Herczeg L, Lang E, Hrustinszky T, Banhidi L. 2006.
Influence of carbon-dioxide pollutant on human well-being
and work intensity. In: Proceedings of Healthy Buildings
2006, 4–8 June 2006, Lisbon, Portugal:Universidade do
Porto, 85–90.
Kaye J, Buchanan F, Kendrick A, Johnson P, Lowry C, Bailey J,
et al. 2004. Acute carbon dioxide exposure in healthy
adults: evaluation of a novel means of investigating the
stress response. J Neuroendocrinol 16(3):256–264.
Knibbs LD, De Dear RJ, Morawska L, Atkinson SE. 2008. On-road
quantification of the key characteristics of automobile HVAC
systems in relation to in-cabin sub micrometer particle
pollution [Abstract]. In: 11th International Conference on
Indoor Air Quality and Climate: Indoor Air 2008. Copenhagen,
Denmark. Available: http://eprints.qut.edu.au/13822/
[accessed 23 October 2012].
Krishnamurthy S, Satish U, Foster T, Streufert S, Dewan M,
Krummel T. 2009. Components of critical decision making
and ABSITE assessment: toward a more comprehensive
evaluation. J Grad Med Educ 1(2):273–277.
Lipsett MJ, Shusterman DJ, Beard RR. 1994. Inorganic com-
pounds of carbon, nitrogen, and oxygen. In: Patty’s
Industrial Hygiene and Toxicology (Clayton GD, Clayton FD,
eds). New York:John Wiley & Sons, 4523–4554.
Milton DK, Glencross PM, Walters MD. 2000. Risk of sick leave
associated with outdoor air supply rate, humidification,
and occupant complaints. Indoor Air 10(4):212–221.
Mudarri DH. 1997. Potential correction factors for interpreting
CO
2
measurements in buildings. ASHRAE Transactions
103(2):244–255.
OSHA (Occupational Safety and Health Administration). 2012.
Sampling and Analytical Methods: Carbon Dioxide in
Workplace Atmospheres. Available: http://www.osha.gov/
dts/sltc/methods/inorganic/id172/id172.html [accessed
7 May 2012].
Persily AK. 1997. Evaluating building IAQ and ventilation with
carbon dioxide. ASHRAE Transactions 103(2):193–204.
Persily A, Dols WS. 1990. The relation of CO
2
concentra-
tion to office building ventilation. In: Air Change Rate
and Airtightness in Buildings (Sherman MH, ed). West
Conshohocken, PA:ASTM, 77–92.
Persily AK, Gorfain J. 2008. Analysis of Ventilation Data from
the U.S. Environmental Protection Agency Building
Assessment Survey and Evaluation (BASE) Study. NISTIR-
7145-Revised. Gaithersburg, MD:National Institute for
Standards and Technology.
Roberge RJ, Coca A, Williams WJ, Powell JB, Palmiero AJ. 2010.
Physiological impact of the N95 filtering facepiece respira-
tor on healthcare workers. Respir Care 55(5):569–577.
Satish U, Manring J, Gregory R, Krishnamurthy S, Streufert S,
Dewan M. 2009. Novel assessment of psychiatry residents:
SMS simulations. Accreditation Council for Graduate Medical
Education (ACGME) Bulletin, 18–23 January. Available: http://
www.lj.se/newsdoc_files/upl_6120257683244412738.pdf
[accessed 23 October 2012].
Satish U, Streufert S. 2002. Value of a cognitive simulation in
medicine: towards optimizing decision making performance
of healthcare personnel. Qual Saf Health Care 11(2):163–167.
Satish U, Streufert S, Dewan M, Voort SV. 2004. Improvements
in simulated real-world relevant performance for patients
with seasonal allergic rhinitis: impact of desloratadine.
Allergy 59(4):415–420.
Satish U, Streufert S, Eslinger PJ. 2008. Simulation-based
executive cognitive assessment and rehabilitation after
traumatic frontal lobe injury: a case report. Disabil Rehabil
30(6):468–478.
Seppanen O, Fisk WJ, Mendell MJ. 1999. Association of ven-
tilation rates and CO
2
concentrations with health and
other responses in commercial and institutional buildings.
Indoor Air 9(4):226–252.
Shendell DG, Prill R, Fisk WJ, Apte MG, Blake D, Faulkner D.
2004. Associations between classroom CO
2
concentra-
tions and student attendance in Washington and Idaho.
Indoor Air 14(5):333–341.
Streufert S, Pogash RM, Gingrich D, Kantner A, Lonardi L,
Severs W, et al. 1993. Alcohol and complex functioning.
J Appl Soc Psychol 23(11):847–866.
Streufert S, Pogash R, Piasecki M. 1988. Simulation based
assessment of managerial competence: reliability and
validity. Personnel Psychology 41(3):537–557.
Streufert S, Satish U. 1997. Graphic representations of process-
ing structure: the time event matrix. J Appl Soc Psychol
27(23):2122–2148.
Streufert S, Streufert SC. 1978. Behavior in the Complex
Environment. New York:John Wiley and Sons.
Streufert S, Swezey RW. 1986. Complexity, Managers, and
Organizations. Orlando, FL:Academic Press.
Swezey RW, Streufert S, Satish U, Siem FM. 1998. Preliminary
development of a computer-based team performance
assessment simulation. Int J Cogn Ergonomics 2:163–179.
Wargocki P, Wyon DP, Sundell J, Clausen G, Fanger PO. 2000.
The effects of outdoor air supply rate in an office on per-
ceived air quality, sick building syndrome (SBS) symptoms
and productivity. Indoor Air 10(4):222–236.
Whitmore CA, Clayton A, Akland A. 2003. California Portable
Classrooms Study, Phase II: Main Study, Final Report,
Vol II. Research Triangle Park, NC:RTI International.
Woods SW, Charney DS, Goodman WK, Heninger GR. 1988.
Carbon dioxide–induced anxiety: behavioral, physiologic,
and biochemical effects of carbon dioxide in patients with
panic disorders and healthy subjects. Arch Gen Psychiatry
45(1):43–52.
