CC BY
0Accepted Manuscript
The impact of working in a green certified building on cognitive function and health
Piers MacNaughton, Usha Satish, Jose Guillermo Cedeno Laurent, Skye Flanigan, Jose Vallarino, Brent Coull, John D. Spengler, Joseph G. Allen
PII: S0360-1323(16)30472-3
DOI: 10.1016/j.buildenv.2016.11.041
Reference: BAE 4723
To appear in: Building and Environment
Received Date: 27 June 2016
Revised Date: 23 November 2016
Accepted Date: 24 November 2016
Please cite this article as: MacNaughton P, Satish U, Laurent JGC, Flanigan S, Vallarino J, Coull B, Spengler JD, Allen JG, The impact of working in a green certified building on cognitive function and health, Building and Environment (2016), doi: 10.1016/j.buildenv.2016.11.041.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
The Impact of Working in a Green Certified Building on Cognitive Function and Health Authors
Piers MacNaughton1, Usha Satish2, Jose Guillermo Cedeno Laurent1, Skye Flanigan1, Jose Vallarino1, Brent Coull3, John D. Spengler1, Joseph G. Allen1
Affiliations
1 Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, MA, USA
2 Psychiatry and Behavioral Sciences, SUNY-Upstate Medical School, Syracuse, NY, USA Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA, USA
Corresponding Author:
Joseph G. Allen, DSc, MPH Assistant Professor
Harvard T.H. Chan School of Public Health
401 Park Drive
Landmark Center, 404-L
Boston, MA 02215
e: JGAllen@hsph.harvard.edu
o: 617-384-8475
Short running title: Green Certified Buildings and Cognitive Function
Abstract
Thirty years of public health research have demonstrated that improved indoor environmental quality is associated with better health outcomes. Recent research has demonstrated an impact of the indoor environment on cognitive function. We recruited 109 participants from 10 high-performing buildings (i.e. buildings surpassing the ASHRAE Standard 62.1-2010 ventilation requirement and with low total volatile organic compound concentrations) in five U.S. cities. In each city, buildings were matched by week of assessment, tenant, type of worker and work functions. A key distinction between the matched buildings was whether they had achieved green certification. Workers were administered a cognitive function test of higher order decision-making performance twice during the same week while indoor environmental quality parameters were monitored. Workers in green certified buildings scored 26.4% (95% CI: [12.8%, 39.7%]) higher on cognitive function tests, controlling for annual earnings, job category and level of schooling, and had 30% fewer sick building symptoms than those in non-certified buildings. These outcomes may be partially explained by IEQ factors, including thermal conditions and lighting, but the findings suggest that the benefits of green certification standards go beyond measureable IEQ factors. We describe a holistic "buildingomics" approach for examining the complexity of factors in a building that influence human health.
Keywords: Green Certification, Office Buildings, Cognitive Function, Indoor Environmental Quality, Buildingomics
1 1.0 Introduction
2 Thirty years of public health science and building science have demonstrated that buildings play
3 a key role in shaping our health [1-5]. Buildings have the capacity to create conditions that are
4 harmful to health or conducive to health: they determine our exposure to outdoor pollutants, by
5 either facilitating entry of particles of outdoor origin indoors, or acting as a barrier and removing
6 them through enhanced filtration [6]; they govern exposure to chemicals of concern, such as
7 volatile organic compounds (VOCs), flame retardants and polyfluorinated compounds, which
8 can be ubiquitous or nonexistent, depending on the decisions we make regarding building
9 materials and products [7, 8]; buildings either protect us from noise or contribute to the problem
10 through the introduction of indoor sources, poor noise insulation, or poor acoustical design [9,
11 10]; they can induce eye strain or improve alertness through impacts on circadian rhythm,
12 depending on the lighting system [11, 12]; buildings can protect us during heat events, or create
13 environments that magnify the problem through solar heat gain [13, 14]; and buildings can either
14 wall us off from nature or connect us to it [15, 16].
15 The scientific literature around buildings and health has identified the foundations of a
16 healthy building including factors such as ventilation, air quality, thermal comfort, noise and
17 lighting, and this body of research has served as the basis for green certification standards to
18 define their indoor environmental quality (IEQ) guidelines. A review of leading, global green-
19 building standards - LEED New Construction 2009, Green Star Office v3, BREEAM New
20 Construction 2012, BCA Green mark for new non-residential buildings v4.1 2013, and DGNB
21 New Office v2012 - demonstrates the approach of these certification standards toward IEQ. All
22 of the rating systems offer credits for thermal comfort, indoor air quality (IAQ) and lighting; all
23 but LEED NC 2009 have credits for acoustics; and Green STAR v3 and LEED NC 2009 have
24 credits specifically for ventilation. However, building owners and developers can opt for certain
25 credits, and IEQ represents only 4-20% of the total score a building can obtain. Of the reviewed
26 rating systems, only LEED NC 2009 has mandatory IEQ credits, for minimum IAQ performance
27 and environmental tobacco smoke control [17].
28 The adoption rates of the optional IEQ credits in LEED NC 2009 give an indication of
29 how building owners are prioritizing certain aspects of IEQ [17]. We extracted the data and
30 found that the vast majority of projects obtain credits for low-emitting adhesives, paints and
31 flooring systems (Table 1). Increased ventilation is much less widely adopted, despite strong
32 evidence for health and performance benefits of higher ventilation rates [18, 19]. While some
33 credits are preferentially adopted and others not, buildings that seek LEED NC 2009 obtain on
34 average 9 of the 15 possible IEQ credits, not including the required fundamental commissioning
35 credit under the energy and atmosphere credit category.
36 Table 1. Credit adoption rates for select optional IEQ credits in 5,490 LEED New Construction
37 2009 certified buildings (USGBC, 2016)._
Credit_% Adoption
EQc2: Increased ventilation 40.9%
EQc4.1: Low-emitting materials - adhesives and sealants 86.5%
EQc4.2: Low-emitting materials - paints and coatings 94.4%
EQc4.3: Low-emitting materials - flooring systems 79.1%
EQc4.4: Low-emitting materials - composite wood and agrifiber products 58.6% EQc5: Indoor chemical and pollutant source control 40.7%
EQc6.1 EQc6.2 EQc7.1 EQc7.2 EQc8.1 EQc8.2
Controllability of systems - lighting 66.4%
Controllability of systems - thermal comfort 39.1%
Thermal comfort - design 79.4%
Thermal comfort - verification 59.2%
Daylight and views - daylight 19.5%
Daylight and views - views 38.3%
39 The literature suggests that these credits translate into improved IEQ. Our previous
40 review of green buildings and health identified 17 studies and found that, overall, occupants
41 report better IEQ and fewer health problems in these buildings compared to non-certified
42 buildings. These studies found lower levels of VOCs, formaldehyde, allergens, nitrogen dioxide,
43 and particulate matter in green buildings, which have been separately shown to impact health.
44 Six of the reviewed studies tracked the health of occupants in addition to IEQ, and all six found
45 improvements in the green buildings [20]. These include reduced asthma and allergy symptoms
46 in offices [21]; reduced respiratory symptoms, fewer sick building symptoms, and better self-
47 reported well-being in public housing [22-24]; and fewer medical errors and decreased mortality
48 in hospitals [25]. Of these studies, Newsham et al. used an approach similar to this study by
49 recruiting green and conventional office building pairs and measuring IEQ. They found an
50 improvement in IEQ, a reduction in symptoms, and better reported sleep quality in the green
51 buildings [26]. A follow up paper by Colton et al. published since the time of our review found
52 that in addition to fewer asthma symptoms, hospital visits and school absences were reduced in
53 the green certified public housing development [27]. Comparisons of buildings in poor condition
54 to green buildings provide an opportunity to see the biggest potential effect, but may falsely
55 attribute benefits to certification.
56 As part of our efforts to determine the factors that drive better human health in buildings,
57 we previously conducted a study in a controlled setting to investigate several IEQ factors -
58 ventilation, CO2, and VOCs - and their impact on cognitive function scores. We found
59 significant impacts on human decision-making performance related to all three of these factors
60 (Allen et al., 2015). Others have also found independent effects of ventilation, CO2 and VOCs on
61 cognitive function and other physiological responses at levels commonly found in indoor
62 environments [19, 28-31]. In this current study, we looked at buildings that are high-performing
63 across these indicators of IEQ and investigated the potential for additional benefits of green
64 certification on cognitive function, environmental perceptions, and health.
66 2.0 Methods
67 Study Design - Overview
68 Workers from 10 office buildings in five U.S. cities (two buildings per city) were recruited to
69 participate in a week-long assessment. 12 participants were initially recruited from each building.
70 Participants completed surveys about their health and environmental perceptions and took a
71 cognitive test on the Tuesday and Thursday of the assessment. All buildings are high-performing
72 buildings, defined as buildings surpassing the ASHRAE Standard 62.1-2010 minimum
73 acceptable per person ventilation requirement and with low (<250 (ig/m3) TVOC concentrations;
74 however, six of the buildings were renovated to green via the LEED certification framework
75 while the remaining four did not seek green certification during renovation [32].
77 Participant and Building Recruitment
78 The building assessments took place in urban areas of the following cities: Boston,
79 Massachusetts (9/29/2015-10/2/2015); Washington DC (10/26/2015-10/30/2015); Denver,
80 Colorado (11/9/2015-11/13/2015); San Jose, California (11/30/2015-12/4/2015); and Los
81 Angeles, California (12/14/2015-12/18/2015 and 2/1/2016-2/5/2016). In each city, the buildings
82 were matched strictly by tenant and loosely by age and size (Table 3). In the first four cities, the
83 buildings were also matched by the dates of assessment, and the buildings were recruited such
84 that one building was LEED-certified and the other not. The goal of matching was to select two
85 high-performing buildings in each city that were as similar to each other as possible with the key
86 distinction being that one pursued LEED certification. In the last city, Los Angeles, two green
87 certified buildings were recruited and the assessments occurred on different dates due to an
88 earlier enrolled building dropping out of the study prior to the assessment; a second building was
89 subsequently recruited. The study team visited each building prior to the assessment to: 1)
90 perform a an initial assessment of the heating, ventilation and air conditioning (HVAC) systems,
91 2) ensure that the building classification as high-performing was valid, and 3) recruit
92 participants.
93 After obtaining permission from the building owner, building management and tenant, 12
94 participants were recruited to participate in a five day health assessment in each building. Final
95 participant numbers by building are presented in Table 3. As mentioned previously, the same
96 tenant was used in each city to ensure similar work functions, and all of the companies employ
97 primarily knowledge workers (i.e. administrative, professional, technical and managerial
98 positions). Asthmatics were excluded during recruitment. We did not restrict recruitment to
99 select areas of each building to limit potential selection bias, but we are unable to demonstrate
100 that our participants are representative of the building population. The study protocol was
101 reviewed and approved by the Harvard T.H. Chan School of Public Health Institutional Review
102 Board. All participants signed informed consent documents and were compensated $100.
103 Table 2. Demographic breakdown of participants in each building classification
High-Performing High-Performing Green Certified Non-Certified
Number of Participants Gender Male Female
55% 45%
54% 46%
20-30 31-40 41-50 51-60 61-70
39% 21% 21% 18% 1%
28% 33% 15% 15% 8%
Ethnicity
White/Caucasian 70% 56%
Black or African American 6% 10%
Asian 7% 8%
Latino 7% 13%
Other 9% 13% Highest level of Schooling
High School Graduate 0% 10%
Some College 12% 26%
College Degree 63% 49%
Graduate Degree 25% 15% Job Category
Managerial 22% 10%
Professional 45% 54%
Technical 6% 18%
Secretarial or Clerical 18% 15%
Other 9% 3% Total Annual Earnings
<$50,000 34% 13%
$50,000-$75,000 21% 41%
$75,000-$100,000 10% 21%
$100,000-$150,000 27% 18%
>$150,000 7% 8%
1 Includes 2 participants in green certified buildings and 1 in non-certified buildings who did not complete the baseline survey
105 Building Assessment
106 The building assessment consisted of three parts. First, the study team conducted an inspection of
107 the building systems along with the building engineers from each facility. The study team
108 recorded the type and condition of the systems, how they are typically operated, and the
109 frequency of building commissioning tasks such as changing the filters. Second, the study team
110 characterized each test space. The test spaces were defined by the unique ventilation zones in
111 which the participants were located. The baseline assessment of the test spaces characterized the
112 building, office and cleaning materials in the space; the air supply and exhaust strategies; and the
113 environmental controls such as operable windows and thermostat set points. On each cognitive
114 testing day, a separate assessment was conducted of the ventilation rates, noises, odors and
115 occupancy in each test space. Lastly, the building manager was provided a survey asking about
116 general building information, building policies, and utility costs. All elements of the building
117 assessment were adapted from the EPA BASE study [33]. These elements were designed to
118 assess the building as a whole rather than just the IEQ of the participant's workstations. The
119 building assessments did not intend to validate the certification of building; therefore, we cannot
120 say whether the green certified buildings still meet the criteria for certification nor whether the
121 non-certified buildings would classify as a green certified building had they gone through the
122 certification process at the time of the study. We anticipate that the organizations responsible for
123 the non-certified buildings would seek certification if it was possible since the same
124 organizations did obtain certification for the green certified buildings in our study.
126 Table 3. Building characteristics of the 10 high-performing buildings included in the study
City Type Size (sq. ft) Year of Type/Year of Ventilation Number of
Construction Certification1 Strategy2 Participants
Boston Non-Certified <50,000 1929 NA CV, RC 12
Boston Certified <50,000 1929 LEED EB v3 Platinum in 2012 VAV, SP 12
DC Non-Certified >500,000 1935 NA VAV, RC 11
DC Certified >500,000 1917 Pending CV, SP 12
Denver Non-Certified 50,000-100,000 1938 NA CV, RC 8
Denver Certified 50,000-100,000 1938 LEED CI v3 Silver in 2011 CV, RC 12
San Jose Non-Certified 50,000-100,000 1971 NA CV, RC 9
San Jose Certified >500,000 1934 LEED EB v3 Gold in 2015 VAV, RC 12
Los Angeles Certified <50,000 1953 LEED EB v3 Platinum in 2013 VAV, RC 11
Los Angeles Certified <50,000 1929 Pending VAV, RC 10
1 EB = Existing Buildings, CI = Commercial Interiors
2 CV = Constant Volume, VAV = Variable Air Volume, SP = Single pass with energy recovery ventilator, RC = Partial recirculation with reheat
128 Environmental Assessment
129 A complete characterization of the IEQ in each test space was conducted on each cognitive
130 testing day. Each participant was outfitted with a Netatmo Weather Station (Netatmo, Boulogne-
131 Bellancourt) in their cubicle to measure temperature, humidity, carbon dioxide concentrations in
132 parts per million (ppm), and sound levels (in decibels) every 5 minutes for each participant. The
133 units were tested with 400 and 1000 ppm CO2 calibration gas before and after the field
134 campaign. If the sensor had drifted, the CO2 data was adjusted first by the offset from the 400
135 ppm reading and second by a scaling factor to match the 1000 ppm reading of the instrument to
136 1000 ppm. This process corrected both the intercept and slope of the collected data to match
137 experimentally derived values. The CO2 data was then used to produce ventilation (cfm of
138 outdoor air per person) and air exchange rates (ACH) for each participant-day of the study. For
139 ventilation rate, the 90th percentile CO2 concentration during occupied hours was taken as the
140 steady-state concentration of CO2 using the method described by Ludwig et al., and for air
141 exchange rate, the decays curves of CO2 were analyzed using the tracer gas method described in
142 ASTM Standard E741-11 [34, 35]. Briefly, when test spaces changed from fully occupied to
143 unoccupied, the rate of decay of occupant generated CO2 can be used to estimate air exchange
144 rates using the validated methodology set forth by ASTM. These approaches have some
145 limitations; for example, air from other zones with elevated CO2 levels can bias air exchange rate
146 calculations and assumptions about occupant CO2 generation rates may be inaccurate.
147 Air sampling was performed for 62 common VOCs and 14 common aldehydes in each
148 building in the test space with the most participants present during each cognitive testing day.
149 VOCs were collected using summa canisters according to EPA method TO-15. Aldehydes were
150 collected on an 8-hour integrated active air sample (0.4 L/min flow rate) according to EPA
151 method TO-11. ALS Analytical Laboratories conducted the analyses of these samples
152 (Cincinnati, OH). 25 VOCs and four aldehydes were not detected in any of the samples. Each
153 test space was also equipped with at least one commercial sensor package (FengSensor,
154 Tsinghua University, Beijing) to measure the same parameters as the Netatmo as well as light
155 levels in lux and particulate matter less than 2.5 (im in diameter (PM2.5) in (ig/m3. These sensors
156 were installed on the first day of the assessment (Monday) and collected on the final day of the
157 assessment (Friday).
159 Health Assessment
160 Participants were provided a Basis Peak Watch (Basis an Intel Company, San Francisco) for the
161 duration of the assessment, which tracked the participants' heart rate, skin temperature, galvanic
162 skin response, physical activity (i.e. steps and calorie expenditure) and sleep patterns (i.e. sleep
163 duration, tossing and turning, number of interruptions). The participants also completed a series
164 of questionnaires over the course of the study. The first was a baseline survey about their
165 perceptions of their work environment and health. The second survey was completed each study
166 day at the end of the workday, a total of five times for each participant, which asked about their
167 environment and whether they experienced any of 19 sick building syndrome (SBS) symptoms
168 on that day. A follow-up survey was completed on the final day of the study asking questions
169 about the previous week, such as satisfaction with noise, lighting, thermal comfort and odors in
170 their cubicle. These surveys were adapted from the EPA BASE study as well and used in our
171 previous research on green buildings [30, 33].
172 Cognitive function was assessed using the Strategic Management Simulation (SMS)
173 software on Tuesday and Thursday at approximately 15:00. The participants completed two
174 different scenarios to avoid potential learning effects, and the frequency of each scenario was
175 balanced between green certified and non-certified buildings. The SMS tool is a validated,
176 computer-based test that measures higher-order decision making ability across nine domains of
177 cognitive function, ranging from basic activity levels to strategy. The SMS tool, and how to
178 interpret scores in each cognitive domain, has been extensively described in the literature [36179 38]. Briefly, the SMS tool immerses the participant in a 1.5 hour long real-life scenario, where
180 they have to respond to several plot lines that emerge over the course of the simulation. These
181 plot lines are validated for content and designed to capture cognitive functions representative of
182 productivity in the real world. As a result, validations of the SMS testing have found a high
183 degree of correlation between performance on the SMS test and other indicators of productivity
184 such as salary at age and number of employees supervised at age [36]. Participants are given the
185 flexibility to approach the simulation in their own thinking style, with no stated demands and a
186 wide breadth of available responses. The types of decisions and plans the participant makes and
187 the events to which they link these actions are processed by the software through a series of
188 algorithms that compute scores for each domain. The SMS study team is blinded to the building
189 status (green certified vs. non-certified). Participants' cognitive function scores on Tuesday and
190 Thursday were, on average, highly consistent. More detailed methodology about the cognitive
191 testing is described elsewhere [19, 29, 39].
193 Statistical Methods
194 The IEQ data collected in this study experienced building-level clustering, which was accounted
195 for with hierarchical statistical tests. Two-sample t-tests with clustered data were used to test for
196 significant differences in IEQ between green certified and non-certified buildings. For analyses
197 of participant outcomes, such as cognitive function and sleep, the data was additionally clustered
198 by the repeated measurements on each participant. Generalized linear mixed effect models were
199 used to model the associations between building classification and these outcomes, treating
200 participant ID and building ID as a random effect:
201 Cog.Scoreij k = /?i + /?2 * (Green Certified) -I- blt + b2i:k + e^^ (1)
202 where Cog.Score,jj: is the average cognitive score for subject i on day j in building k, normalized
203 to the non-certified, high-performing buildings; Pi is the fixed intercept; P2 is the fixed effect of
204 high-performing, certified buildings compared to high-performing, non-certified buildings; bn is
205 the random effect of intercept for subject i; and b2i.kis the random effect of intercept for building
206 k. Additional models were run with the following variables: job category, annual earnings, level
207 of schooling and thermal comfort as indicator variables and previous night's sleep as a
208 continuous variable. The residuals were normally distributed and homoscedastic for all models.
209 We used penalized splines to graphically assess linearity in the associations between continuous
210 variables and outcome measures.
211 The SMS tool provides raw scores for nine domains of cognitive function. To allow
212 comparisons between domains, the cognitive function scores were normalized to scores in the
213 non-certified building by dividing each score by the average score in the non-certified buildings
214 in that domain, as has been done in previous studies using the SMS test [39]. The average
215 cognitive score is an average score across the nine domains. Thermal comfort is a binary variable
216 that reflects whether or not a participant was within the thermal comfort zone specified by
217 ASHRAE Standard 55-2004 on any particular day of the assessment [40] (Figure SI). Relative
218 humidity and temperature from the Netatmo were entered in the Fanger thermal comfort
219 equations to estimate whether the percent of people dissatisfied with the thermal conditions
220 would exceed 10% [41]. We assume constant radiant temperatures (same as dry bulb
221 temperature), air velocities (0.15 m/s), metabolic rates (1 met), and clothing (1 clo) between
222 participants.
223 To assess sleep, we developed an index to characterize each night of sleep across three
224 well-known indicators of sleep quality: sleep duration, tossing and turning, and number of
225 interruptions. It was calculated using data from the Basis Watch for each night of sleep the
226 participants had during the assessment according to equation (2):
„ Sleep.Duration „ _„, Toss.Tivrn „„„, Num.Int
227 Sleep Score = 100%-----10%---10%--(2)
r 420 85 4 v '
228 where Sleep. Duration is the number of minutes the participant spent sleeping between 9PM and
229 9AM the following day, Toss.Turn is the number of minutes during which the watch registered
230 motion via the accelerometer (the maximum Toss.Turn in this study was 85), and Num.Int is the
231 number of times during a night of sleep that the sleep activity changed from asleep to awake and
232 then back to asleep (the maximum Num.Int in this study was 4). If the participant slept for longer
233 than 420 minutes, or 7 hours, the first term was capped at 100%. Nights when the watch was not
234 worn or worn improperly were removed from the analysis, resulting in a total sample size of 260
235 nights, 100 of which preceded a cognitive testing day. The average Sleep Score was 83.1% with
236 a standard deviation of 19.7%. Sleep Scores and thermal comfort were added to the model in
237 Equation 1 to test their effect on cognitive function. Analyses were performed using the open-
238 source statistical package R version 3.2.0 (R Project for Statistical Computing, Vienna, Austria).
240 3.0 Results
241 The non-certified buildings and green certified buildings had similar air quality; the low CO2,
242 low TVOC and high ventilation rates indicate that the buildings were high-performing at the time
243 of the assessment (Figure 1). The ventilation rates exceeded the ASHRAE 62.1 -2010 standard
244 for 84% of participants, which could mitigate the buildup of airborne contaminants. The green
245 certified buildings were on average brighter (374 lux vs. 163 lux), louder (51.8 dB vs. 48.9 dB),
246 and drier (38.4% vs. 45.9%) than the non-certified buildings; however, only the difference in
247 relative humidity was statistically significant (Figure 1). Differences in humidity may be driven
248 by the ventilation strategies in the green certified buildings, which more frequently had variable
249 air volume ventilation systems and energy recovery ventilators (ERVs). In the cases when
250 outdoor humidity was high, buildings with ERVs had lower indoor humidity levels.
Vent (cfm/person)
AER (ACH)
C02 (ppm)
High-performing
TVOCs (jig/m3) PM2 J (ng/mä)
High-performing Green Certified
Temp (°C)
RH (%) =!=
Noise (dB) Light (lux)
Figure 1. Boxplots of indoor environmental quality (IEQ) parameters in high-performing, non-certified buildings and high-performing, green certified buildings. Vent, AER, CO2, Temp, RH and Noise are measured by the Netatmo in every workstation each day, TVOCs are measured with summa canisters in every test space each cognitive testing day, and PM25 and Light are measured by the Feng Sensor in every test space each day. An asterisk (*) denotes that the
257 building classifications are statistically significantly different from each other for that IEQ
258 parameter after adjusting for clustering by building.
260 Between-subject analyses were necessary to compare participants in different building
261 classifications. Table 2 shows the demographic information for the participants in each building
262 classification: the matching criteria resulted in the two groups having similar job classifications,
263 gender and ages. The green certified buildings had a slightly larger percentage of
264 white/Caucasian participants and participants with a college or graduate degree. These buildings
265 also had more participants at both the lower and higher end of the range of annual earnings. We
266 added these variables as predictors to the cognitive function models to test if they influenced
267 baseline cognitive abilities. While some of these variables had non-significant associations with
268 cognitive test scores, the effect estimate of building classification did not change when these
269 parameters were added to the model, indicating that the findings are not a result of residual
270 confounding.
271 The impact of building classification on each domain of cognitive function is summarized
272 in Figure 2. On average, participants in the high-performing, green certified buildings scored
273 26.4% (95% CI: [12.8%, 39.7%]) higher on the SMS cognitive test than those in the high-
274 performing, non-certified buildings (p-value < 0.001). Cognitive scores were statistically
275 significantly higher for 7 of the 9 domains with the largest impacts on crisis response, applied
276 and focused activity level and strategy. No differences in scores were seen for basic activity level
277 or information seeking. For the average scores, the model's R was 0.28, indicating that 28% of
278 the variability in cognitive function scores is explained by the building classification alone.
High-performing Green^CertifiscP
Basic Applied Focused Task
Activity Level Activity Level Activity Level
Crisis Information Information Breadth Response Seeking Usage of Approach
Strategy Average
280 281 282
03 Cl I
Cognitive Domain
Figure 2. Cognitive scores and 95% confidence intervals for each domain of the SMS tool after controlling for participant, normalized to high-performing buildings, for participants in high-performing and high-performing, green certified buildings
284 Of the IEQ parameters assessed in the buildings, the largest differences were seen for
285 relative humidity. The non-certified buildings were more frequently outside the ASHRAE
286 Standard 55 thermal comfort zone than the green certified buildings due to their higher
287 humidities (Figure S1). Both building classifications had participant-days where the building was
288 too cold to comply with ASHRAE Standard 55. After controlling for building classification,
289 participants scored 5.4% higher on the cognitive tests, averaged across the nine domains of
290 cognitive function, on days when they took the SMS test within the thermal comfort zone than
291 when they took it without (Figure 3). This finding is not statistically significant at the 95%
292 confidence level.
293 Previous night's sleep was also associated with cognitive function scores. A 25%
294 increase in Sleep Scores was associated with a 2.8% increase in cognitive function scores. Sleep
295 quality was influenced by day-time exposures in the office: participants in the green certified
296 buildings had 6.4% higher Sleep Scores than those in the non-certified buildings. This may be in
297 part a result of higher light levels in the green buildings; a 300 lux increase in illuminance during
298 the day was associated with a 2.9% increase in Sleep Scores that night. However, these findings
299 are not statistically significant (Figure 3).
304 Figure 3. Effect of a) thermal comfort on cognitive function scores, b) yesterday's sleep on
305 cognitive function scores, c) building classification on Sleep Scores, and d) light levels on Sleep
306 Scores, using generalized linear mixed effect models with 95% confidence intervals, treating
307 building and participant as random effects. The effect size for thermal comfort is comparing
308 cognitive scores from tests taken by participants within the ASHRAE Standard 55-2013 comfort
309 zone to those without. The effect sizes for yesterday's sleep and light correspond to a 25%
310 change in Sleep Score and 300 lux change in illuminance respectively.
312 In addition to improved cognitive function scores, participants in green certified buildings
313 reported better environmental perceptions and fewer symptoms than those in non-certified
314 buildings. Participants in green certified buildings were generally more satisfied with daylighting
315 and electrical lighting in their workspace, and less frequently reported the temperature being too
316 hot or too cold, the air movement being too much or too little, the air being too dry or too humid,
317 and the presence of chemical, tobacco and other odors (Figure S2). These perceptions are linked
318 to varying degrees to the monitored IEQ in the spaces. For example, relative humidities were
319 15.9% higher when participants reported the air was too humid and 9.3% lower when they
320 reported the air was too dry. Importantly, for the same change in monitored IEQ conditions,
321 participants in the green certified buildings report a larger improvement based on environmental
322 perceptions. Lastly, participants in the non-certified buildings reported 0.5 (30%) more
323 symptoms each day than those in the green certified buildings. Symptom counts are higher when
324 participants report an issue with environmental conditions. Environmental perceptions and total
325 symptom counts were not associated with cognitive function scores when introduced into the
326 mixed effect models.
328 4.0 Discussion
329 Previous research by our team, and others, has identified IAQ as a key driver of cognitive
330 function. In particular, CO2, TVOCs, and ventilation all have independent impacts on cognitive
331 function, even at levels deemed to be acceptable by the relevant codes and standards [19, 28, 29,
332 39]. Many office buildings on the market now fit the classification as high-performing by
333 surpassing the ASHRAE Standard 62.1 ventilation requirement and having low TVOC
334 concentrations (<250 (ig/m3). The findings of this study indicate that even among high-
335 performing buildings that meet these IEQ criteria, additional benefits to cognitive function and
336 health may be achieved by seeking green building certification. Participants in high-performing,
337 green certified buildings had better environmental perceptions, 30% fewer sick building
338 symptoms, 26.4% higher cognitive function scores and 6.4% higher Sleep Scores than
339 participants in the high-performing, non-certified buildings even after controlling for annual
340 earnings, job categories, and level of schooling. The reduction in self-reported symptoms and
341 improvements in environmental perceptions support previous research in green buildings [23, 24,
342 27, 30, 42]. Participant's environmental perceptions do track actual IEQ conditions, but
343 participants in green certified buildings are more likely to have a positive response even when
344 IEQ conditions are the same. This observation, along with participants reporting more symptoms
345 when they report problems with environmental conditions, highlights the limitations of using
346 subjective metrics when assessing building performance or occupant wellbeing. For the cognitive
347 function results, some of the domains that had the largest differences in scores (crisis response,
348 information usage, and strategy) are the most highly correlated with other measures of
349 productivity such as salary at age [36]. This aligns with Allen et al. that found these same
350 domains to be the most impacted by CO2, TVOCs and ventilation. By comparison, lowering
351 TVOC concentrations from -580 (ig/m3 to -40 (ig/m3 caused a 61% increase in cognitive
352 function scores in that study compared to 26.4% increase from working in a green certified
353 building in this study.
354 While much of the effect of green certification on cognitive test scores is unexplained,
355 the effect may be partly attributed to several IEQ parameters. The green certified buildings were
356 generally less humid than the non-certified buildings, and as a result a larger proportion of
357 participants in these buildings were in the thermal comfort zone defined by ASHRAE 55 (Figure
358 S1). Participants outside this thermal comfort zone scored 5.4% lower on the cognitive
359 simulations, but the finding was not statistically significant. The detriments to cognitive function
360 align with previous research on thermal conditions and performance. In a review of 24 papers,
361 Seppanen et al. found that work performance was optimized at temperatures within the ASHRAE
362 Standard 55 zone, and that the benefits were seen using various different indicators of cognitive
363 function ranging from simple cognitive tests to objectively reported work performance [43]. The
364 impacts on the SMS tool indicate that high order decision-making may also be affected by these
365 exposures.
366 Not surprisingly, our study suggests that previous night's sleep is a driver of cognitive
367 function scores. More interesting is that better Sleep Scores were associated with better lighting
368 conditions in the building. This is biologically plausible, considering previous research linking
369 exposure to daylighting or blue-enriched lighting before sleep to sleep repression. Warmer light
370 colors, such as those at dusk, trigger the body to release melatonin, which has a fatiguing effect,
371 and late-night screen use can delay or suppress the release of melatonin [44]. Similarly, a larger
372 contrast between daytime light exposures and nighttime light exposures leads to a larger
373 amplitude in daily melatonin secretion cycles [45]. Daylighting and blue-enriched lighting during
374 the day helps align the body's circadian rhythm and improve sleep quality at night [12]. This
375 effect was observed in our study: brighter lighting in the office during the day was associated
376 with higher Sleep Scores at night, and participants in the green certified buildings, which were
377 generally brighter, had 6.4% higher Sleep Scores than those in the non-certified buildings. This
378 finding supports previous research by Newsham et al. on sleep quality in green buildings [26].
379 Investigating real-world office buildings, as opposed to a simulated environment, posed
380 several limitations on the study. First, the case-control study design required between-subject
381 comparisons. To minimize baseline differences in cognitive function, we matched the buildings
382 by tenant and job categories. Adding annual earnings, level of education, and job category to our
383 models did not influence the effect size of building classification on cognitive function scores,
384 nor were these factors statistically significantly associated with cognitive scores. Second, the
385 environmental conditions were variable between buildings and could not be modified by the
386 study team. The variability in exposures also limits the ability for the factors we did measure to
387 produce a quantifiable effect. Third, missing data for some outcomes, such as sleep, reduced the
388 power of those analyses. Fourth, while the sample size of participants was sufficiently powered,
389 factors that vary on building level, such as ventilation system type, have a sample size of 10 and
390 were underpowered. With this sample size we were not able to identify which individual green
391 credits were drivers of better performance, nor were we able to obtain the same level of building-
392 related design data from the non-certified buildings (precisely because they did not go through
393 the certification process). As such, it is possible that green certification in our study may simply
394 be a proxy for more relevant indicators of building performance. Fifth, we assessed the IEQ of
395 the workstations of our participants, which may not be representative of the building as a whole.
396 During our building assessment, we did not observe major differences in building systems,
397 operation or maintenance for areas of the building in which we did not have participants. As the
398 buildings were all high-performing, the results of the study may not be representative of
399 conventional or problem buildings. In addition, the study population is representative of the
400 general population of knowledge workers and may not be generalizable to other worker
401 populations.
402 The findings in this study hint at the complexity of understanding all of the building
403 related factors that can influence human health and performance. The measured IEQ variables
404 only accounted for part of the impact of green certification on productivity and health. Other
405 aspects of the green certification process - such as commissioning of building systems, 3rd party
406 reviews of IEQ performance, and the commitment to sustainability and health of owners and
407 building managers - may play a role in how occupants perceive and perform in a building. Here,
408 we advocate for a holistic, "buildingomics" approach. Omics research describes efforts to
409 understand the totality of a given research field, currently best exemplified by genomics research
410 and the ambitious undertaking of the Human Genome Project. This has spurred a set of related -
411 omics research areas: transcriptomics, proteomics, metabolomics, epigenomics. And, in the field
412 of exposure science, the relatively new and equally challenging efforts to characterize human
413 exposures over the course of a person's lifetime - the exposome [46]. We now propose
414 "buildingomics" to capture the complexity of the research of health in buildings.
415 "Buildingomics" is the totality of factors in indoor environments that influence human health,
416 well-being and productivity of people who work in those spaces. The primary challenge is that
417 buildings serve a variety of purposes and the potential exposures span several fields of study;
418 thus multi-disciplinary teams that include building scientists, exposure scientists,
419 epidemiologists, toxicologists, materials scientists, architects, designers, and social/behavioral
420 scientists are necessary to characterize all the building-related factors that influence health in
421 buildings.
423 5.0 Conclusions
424 Our findings show that in high-performing buildings additional benefits to health and
425 productivity may be obtained through green certification. In a sample of 10 high-performing
426 buildings, participants in green certified buildings had 26.4% higher cognitive function scores,
427 better environmental perceptions and fewer symptoms than those in high-performing, non-
428 certified buildings. This outcome may be partially explained by IEQ factors, including thermal
429 conditions and lighting, but the findings suggest that the benefits of green certification standards
430 go beyond measureable IEQ factors. Building-level factors may play an important role in
431 occupant health and cognitive function yet have been largely overlooked. We describe the need
432 for a holistic, "buildingomics" approach to studying the drivers of human health and
433 performance in buildings.
435 Acknowledgements
436 We would like to acknowledge our partners who allowed our team to investigate the impact of
437 buildings on their employees and tenants. We thank both the study participants and field staff for
438 volunteering their time. We also thank the reviewers of this manuscript for their thoughtful
439 comments that have strengthened the paper. This research was supported by a gift from United
440 Technologies to the Center for Health and the Global Environment at the Harvard T.H. Chan
441 School of Public Health. United Technologies was not involved in the recruitment of buildings
442 or participants, data collection, data analysis, data presentation, or drafting of the manuscript.
443 Dr. MacNaughton's time was supported by NIEHS environmental epidemiology training grant
444 5T32ES007069-35.
445 References
446 [1] Wargocki, P., D. Wyon, J. Sundell, G. Clausen, and P.O. Fanger. 2000. The effects of
447 outdoor air supply rate in an office on perceived air quality, Sick Building Syndrome
448 (SBS) symptoms and productivity. Indoor Air-International Journal Of Indoor Air
449 Quality And Climate, 10(4): p. 222-236.
450 [2] Spengler, J.D. and K. Sexton. 1983. Indoor Air Pollution: A Public Health Perspective.
451 Science, 221(4605): p. 9-17.
452 [3] Weschler, C.J. 2009. Changes in indoor pollutants since the 1950s. Atmospheric
453 Environment, 43(1): p. 153-169.
454 [4] Mendell, M.J., W.J. Fisk, K. Kreiss, H. Levin, D. Alexander, W.S. Cain, J.R. Girman, C.J.
455 Hines, P.A. Jensen, D.K. Milton, L P. Rexroat, and K.M. Wallingford. 2002. Improving
456 the Health of Workers in Indoor Environments: Priority Research Needs for a National
457 Occupational Research Agenda. American Journal of Public Health, 92(9): p. 1430458 1440.
459 [5] Spengler, J. Indoor air quality handbook, ed. J.F. McCarthy, J.M. Samet, and J.D. Spengler.
460 2001, New York: McGraw-Hill.
461 [6] Rudd, A. Ventilation system effectiveness and tested indoor air quality impacts, ed. D.
462 Bergey, et al. 2014: U.S. Department of Energy, Energy Efficiency & Renewable Energy,
463 Building Technologies Office.
464 [7] Allen, J.G., M.D. McClean, H.M. Stapleton, and T.F. Webster. 2008. Linking PBDEs in
465 House Dust to Consumer Products using X-ray Fluorescence. Environmental Science &
466 Technology, 42(11): p. 4222-4228.
467 [8] Dodson, R.E., L.J. Perovich, A. Covaci, N. Van Den Eede, A.C. Ionas, A.C. Dirtu, J.G.
468 Brody, and R.A. Rudel. 2012. After the PBDE phase-out: a broad suite of flame
469 retardants in repeat house dust samples from California. Environmental science &
470 technology, 46(24): p. 13056.
471 [9] Heakyung, C.Y. 2000. Differences in performance with different background sound and
472 ambient noise in three open office plans. The Journal of the Acoustical Society of
473 America, 108: p. 2632.
474 [10] Jahncke, H. 2012. Open-plan office noise: the susceptibility and suitability of different
475 cognitive tasks for work in the presence of irrelevant speech. Noise &Amp, 14(61): p.
476 315-320.
477 [11] Stone, P.T. 2009. A model for the explanation of discomfort and pain in the eye caused by
478 light. Lighting Research & Technology, 41(2): p. 109-121.
479 [12] Viola, A.U., L.J.M. James Lm Fau - Schlangen, D.-J. Schlangen Lj Fau - Dijk, and D.J.
480 Dijk. 2008. Blue-enriched white light in the workplace improves self-reported alertness,
481 performance and sleep quality. (0355-3140 (Print)).
482 [13] Hatvani-Kovacs, G., M. Belusko, N. Skinner, J. Pockett, and J. Boland. 2016. Drivers and
483 barriers to heat stress resilience. Science of the Total Environment.
484 [14] Park, J. Heat Stress and Human Capital Production (Job Market Paper), in JOB MARKET
485 PAPER - Unpublished Manuscript, Harvard University Economics Department, H.
486 University, Editor. 2016.
487 [15] Ulrich, R.S. 1984. View through a Window May Influence Recovery from Surgery. Science,
488 224(4647): p. 420-421.
489 [16] Wilson, E.O. Biophilia. 1986: President and Fellows of Harvard College.
490 [17] Da Silva, N.A.F., P. Wargocki, and K.W. Tham. Building certification schemes and the
491 quality of indoor environment, in Technical University of Denmark, Department of Civil
492 Engineering. 2015, DTU Civil Engineering Report R335.
493 [18] MacNaughton, P., J. Pegues, U. Satish, S. Santanam, J.D. Spengler, and J. Allen. 2015.
494 Economic, Environmental and Health Implications of Enhanced Ventilation in Office
495 Buildings. International Journal of Environmental Research and Public Health, 12.
496 [19] Maddalena, R., M.J. Mendell, K. Eliseeva, W.R. Chan, D.P. Sullivan, M. Russell, U. Satish,
497 and W.J. Fisk. 2015. Effects of ventilation rate per person and per floor area on perceived
498 air quality, sick building syndrome symptoms, and decision-making. Indoor Air, 25(4):
499 p. 362-370.
500 [20] Allen, J.G., P. MacNaughton, J.G.C. Laurent, S.S. Flanigan, E.S. Eitland, and J.D. Spengler.
501 2015. Green Buildings and Health. Current Environmental Health Reports, 2(3): p. 250502 258.
503 [21] Singh, A., M. Syal, S.C. Grady, and S. Korkmaz. 2010. Effects of green buildings on
504 employee health and productivity. American journal ofpublic health, 100(9): p. 1665.
505 [22] Colton, M., P. MacNaughton, J. Vallarino, J. Kane, M. Bennett-Fripp, J. Spengler, and G.
506 Adamkiewicz. 2014. Indoor Air Quality in Green Vs Conventional Multifamily Low-
507 Income Housing. Environmental Science & Technology, 48(14): p. 7833-7833.
508 [23] Breysse, J., D.E. Jacobs, W. Weber, S. Dixon, C. Kawecki, S. Aceti, and J. Lopez. 2011.
509 Health Outcomes and Green Renovation of Affordable Housing. Public Health Reports,
510 126(Suppl 1): p. 64-75.
511 [24] Jacobs, D.E., E. Ahonen, S.L. Dixon, S. Dorevitch, J. Breysse, J. Smith, A. Evens, D.
512 Dobrez, M. Isaacson, C. Murphy, L. Conroy, and P. Levavi. 2014. Moving Into Green
513 Healthy Housing. Journal of Public Health Management and Practice: p. 1.
514 [25] Thiel, C.L., K.L. Needy, R. Ries, D. Hupp, and M.M. Bilec. 2014. Building design and
515 performance: A comparative longitudinal assessment of a Children's hospital. Building
516 and Environment, 78: p. 130-136.
517 [26] Newsham, G., B. Birt, C. Arsenault, L. Thompson, J. Veitch, S. Mancini, A. Galasiu, B.
518 Gover, I. Macdonald, and G. Burns. Do green buildings outperform conventional
519 buildings? Indoor environment and energy performance in North American offices, in
520 National Research Council Canada. Research Report; no. RR-329. 2012, National
521 Research Council Canada.
522 [27] Colton, M., J.G. Laurent, P. MacNaughton, J. Kane, M. Bennett-Fripp, J. Spengler, and G.
523 Adamkiewicz. 2015. Health Benefits of Green Public Housing: Associations With
524 Asthma Morbidity and Building-Related Symptoms. (1541-0048 (Electronic)).
525 [28] Satish, U., L. Cleckner, and J. Vasselli. 2013. Impact of VOCs on decision making and
526 productivity. Intelligent Buildings International, 5(4): p. 213-220.
527 [29] Satish, U., M.J. Mendell, K. Shekhar, T. Hotchi, D.P. Sullivan, S. Streufert, and W.J. Fisk.
528 2012. Is CO2 an indoor pollutant? Direct effects of Low-to-moderate CO2 concentrations
529 on human decision-making performance. Environmental Health Perspectives, 120(12):
530 p. 1671.
531 [30] MacNaughton, P., J. Spengler, J. Vallarino, S. Santanam, U. Satish, and J. Allen. 2016.
532 Environmental perceptions and health before and after relocation to a green building.
533 Building and Environment, 104: p. 138-144.
534 [31] Vehvilainen, T., H. Lindholm H Fau - Rintamaki, R. Rintamaki H Fau - Paakkonen, A.
535 Paakkonen R Fau - Hirvonen, O. Hirvonen A Fau - Niemi, J. Niemi O Fau - Vinha, and J.
536 Vinha. 2015. High indoor CO concentrations in an Office Environment Increases the
537 Transcutaneous CO Level and Sleepiness during Cognitive Work. (1545-9632
538 (Electronic)).
539 [32] ASHRAE. 2016. Standard 62.1-2016 -- Ventilation for Acceptable Indoor Air Quality.
540 American Society for Heating, Refrigeration, and Air-Conditioning Engineers, Inc.
541 [33] EPA. Building Assessment Survey and Evaluation. 1998 [cited 2015 1/22]; Available from:
542 http://www.epa. gov/iaq/base/study overview.html.
543 [34] Ludwig, J., J. McCarthy, B. Baker, R. Caron, and D. Hanson. 2000. A Review of Selected
544 Methodologies to Determine Outdoor Air Ventilation Rates in BASE Study Buildings.
545 Air and Waste Management Assoc Conference.
546 [35] ASTM. Standard Test Method for Determining Air Change in a Single Zone by Means of a
547 Tracer Gas Dilution. 2011, ASTM International.
548 [36] Streufert, S., R. Pogash, and M. Piasecki. 1988. Simulation-based assessment of managerial
549 competence: Reliability and validity. Personnel Psychology, 41(3): p. 537-557.
550 [37] Satish, U., S. Streufert, M. Dewan, and S. Voort. 2004. Improvements in Simulated Real-
551 world Relevant Performance for Patients with Seasonal Allergic Rhinitis: Impact of
552 Desloratadine. Allergy, 59(4): p. 415-420.
553 [38] Breuer, K. and U. Satish. Emergency Management Simulations: An Approach to the
554 Assessment of Decision-making Processes in Complex Dynamic Crisis Environments, in
555 From Modeling to Managing Security: A Systems Dynamics Approach (J.J. G, ed). , N.A.
556 Press, Editor. 2003: Norway. p. 145-156.
557 [39] Allen, J.G., P. MacNaughton, S. Santanam, U. Satish, and J. Spengler. 2015. Associations
558 of Cognitive Function Scores with Carbon Dioxide, Ventilation, and Volatile Organic
559 Compound Exposures in Office Workers: A Controlled Exposure Study of Green and
560 Conventional Office Environments. Environmental Health Perspectives, 123(10).
561 [40] ASHRAE. 2010. Standard 55-2010 -- Thermal Environmental Conditions for Human
562 Occupancy. American Society for Heating, Refrigeration, and Air-Conditioning
563 Engineers, Inc.
564 [41] Fanger, P.O. Analysis and Applications in Environmental Engineering. 1970, New York:
565 McGraw-Hill Book Company.
566 [42] Garland, E., S.H. Steenburgh Et Fau - Sanchez, A. Sanchez Sh Fau - Geevarughese, L.
567 Geevarughese A Fau - Bluestone, L. Bluestone L Fau - Rothenberg, A. Rothenberg L Fau
568 - Rialdi, M. Rialdi A Fau - Foley, and M. Foley. 2013. Impact of LEED-certified
569 affordable housing on asthma in the South Bronx. Prog Community Health Partnership,
570 7(1): p. 29-37.
571 [43] Seppänen, O., J.F. William, and L.-G. Quanhong. Effect of temperature on task performance
572 in office environment, in 5th International Conference on Cold Climate Heating,
573 Ventilating and Air Conditioning. 2006: Moscow, Russia.
574 [44] Lockley, S.W., C.A. Brainard Gc Fau - Czeisler, and C.A. Czeisler. 2003. High sensitivity
575 of the human circadian melatonin rhythm to resetting by short wavelength light. (0021576 972X (Print)).
577 [45] Takasu, N.N., S. Hashimoto, Y. Yamanaka, Y. Tanahashi, A. Yamazaki, S. Honma, and K-
578 i. Honma. 2006. Repeated exposures to daytime bright light increase nocturnal melatonin
579 rise and maintain circadian phase in young subjects under fixed sleep schedule. American
580 Journal of Physiology - Regulatory, Integrative and Comparative Physiology, 291(6): p.
581 R1799-R1807.
582 [46] Rappaport, S.M. 2010. Implications of the exposome for exposure science. (1559-064X
583 (Electronic)).
Highlights:
1. 26.4% higher cognitive test scores in high-performing, green certified buildings
2. 6.4% higher Sleep Quality scores in high-performing, green certified buildings
3. 30% fewer symptoms in high-performing, green certified buildings
4. Thermal comfort and sleep quality associated with higher cognitive scores
5. "Buildingomics": the totality of factors in buildings that influence health
