The evaluation of green school building attributes and their effect on the health and performance of students and teachers in New York State

Transcription

1 FINAL PROJECT REPORT (Proposal Number: # 147 funded by the United States Green Building Council) The evaluation of green school building attributes and their effect on the health and performance of students and teachers in New York State Shao Lin, Christine Kielb, Anne Orsini, Neil Muscatiello Bureau of Environmental and Occupational Epidemiology Center for Environmental Health New York State Department of Health 1

2 Acknowledgement We would like to sincerely thank all of our partners who made this project possible. Collaborators included the following: Staff within the Bureau of Toxic Substances Assessment, Bureau of Occupational Health, and Outreach & Education Unit at the New York State Department of Health; the New York State Education Department s Office of Facilities Planning; New York State United Teachers; and our Technical Review Team and administrators from the United States Green Building Council. In addition we would like to thank United States Congressman Paul Tonko and staffs from the Capital Region, Questar III, Washington-Saratoga-Warren-Hamilton-Essex, and Hamilton-Fulton-Montgomery Board of Cooperative Education Services for their support. 2

3 EXECUTIVE SUMMARY Background The green building movement has in the past focused primarily on the economic benefits or general environmental impacts of buildings rather than the effects on occupants. However, students, teachers and other staff spend a substantial proportion of their time in schools, and children may be particularly susceptible to indoor hazards. The question of whether there are potential health and performance gains associated with green buildings has yet to be answered. Prior studies have focused on adults working in office buildings, adverse health effects and general building conditions or single school factors (e.g., indoor air quality (IAQ)), but few studies have evaluated the impact of the school environment as a whole on occupant performance. Specifically, few examined school cleanliness, various IAQ factors, acoustics and lighting, and their joint effects. To fill some of these gaps, the objective of the current project was to assess if green school building factors individually and jointly affect student performance and health as well as teacher teaching ability and health in New York State (NYS). Methods This study consisted of three components: statewide data linkage analyses, a statewide telephone teacher survey, and a sub-sample of school walkthroughs (the latter of which included objective measurements and occupant surveys). Statewide analyses examining the relationship between school building conditions and student performance and health were conducted by linking NYS Building Condition Survey (BCS) data with student attendance and test performance data, and school district level asthma hospitalization data in NYS public schools (excluding New York City). Sixty-six building-related variables were analyzed separately and also grouped into attributes measuring IAQ, dryness, cleanliness, acoustics and lighting. Additional building attributes reflecting well-maintained building systems were identified including site utilities, building envelope, building interior, plumbing and HVAC system. A summary measure of greenness, the green school index (GSI) was developed, cumulating the variables described above. The second study component consisted of a statewide telephone survey of teachers asking about teacher demographics, teaching experience and hours, health symptoms, perceptions of the school environment, whether certain classroom conditions affected teaching ability and home environmental conditions. The third study component consisted of school walkthroughs and objective measurements conducted in ten schools to evaluate classroom conditions and IAQ as well as measuring temperature, relative humidity, carbon dioxide (CO2), sound and lighting. Surveys of school nurses and teachers were also conducted in these schools. Results This study found that the most commonly reported favorable green school building conditions included absence of visible mold problems (> 96%) and vermin infestations (>96%). On the other hand, the most frequent unfavorable school building problems were visible water damage/leaks (18%-29%) and unsatisfactory rating of components of the building envelope, including, the roof, windows and skylights (17%-20%). Specific indicators affecting IAQ, such as fresh air intakes away from sources of potential diesel exhaust/pollutants (Odds Ratios [ORs]) ranging from ), air intakes free of blockages (OR: 3

4 2.03) and properly functioning dampers (OR: 1.61), as well as a summary measure of IAQ (OR: 1.36), were the most important factors contributing to good student attendance in NYS. The GSI, representing good IAQ, absence of moisture/mold, good lighting, good acoustics, cleanliness and well-maintained building systems was also found to be significantly associated with good attendance in NYS (OR: 1.22). In addition, good lighting fixtures ratings (OR: 1.55) and no active infestations of cockroaches (OR:10.2) were significantly related to good attendance. We also found significantly interactive effects of the geographic region of the school (upstate vs. downstate schools) between the association of some school factors and student outcomes. In other words, we found that many healthy green building conditions were significantly associated with favorable outcomes for the occupants in upstate NY schools, including fresh air intakes away from sources of diesel exhaust/pollutants (ORs ranged from ), air intakes free of blockages (OR: 2.33) or ductwork free of dirt (OR: 2.33), good condition of air filters (OR: 1.40), functioning dampers (OR: 2.08), good overall IAQ/ventilation (OR: 1.39) and good acoustics (OR: 1.47). However, some of these positive relationships were not found in downstate schools except for fresh air intakes away from sources of diesel exhaust/pollutants (ORs ranging from 2.87 to 3.03) and air intakes free of blockages (OR: 1.81). In terms of school performance, good condition of air filters was associated with good 4 th grade academic performance (OR: 1.42) in statewide analyses. In upstate schools, good condition of air filters (OR: 2.27), good humidity/moisture rating (OR: 1.62) and excellent/satisfactory lighting fixtures (OR: 2.97) as well as a good overall IAQ attribute score (OR: 1.68) were found to be significantly associated with good student test scores, but these associations did not hold for downstate schools. The median asthma hospitalization rate from among school-age children by NYS school district (excluding NYC) was per 10,000. A statewide analysis of school district conditions and asthma hospitalization rates found that high cleanliness scores (OR: 1.46) and good building plumbing scores (OR: 1.90) were significantly associated with lower asthma hospitalization rates, after controlling for potential confounders. A statewide teacher survey (N=501) found the most commonly reported of 14 health symptoms to be sinus problems (22.2%) and allergies/congestion (21.4%). For the most common symptoms, large proportions of teachers reported worsening during the school day or week, ranging from 38.3% for sneezing to 82.9% for throat irritation. Similar proportions reported improvement of these symptoms when away from work (62.6% for allergies/congestion to 92.1% for throat irritation). Substantial proportions of teachers (16.7% to 42.3%) also reported that a given symptom affects their work or attendance. Almost half (48.9%) of teachers surveyed reported having one or more symptom, and over 40% reported having at least one allergic-type symptom. Symptoms potentially associated with poor lighting parameters were reported by 29.3% and acoustics-related symptoms such as headaches and throat irritation were reported by 30.9% of teachers. The most favorable school conditions associated with reduced teacher health symptoms included absence of mold or odors from mold, moisture damage, dust/dust reservoirs, vermin, art supplies, scented products, paint, cleaners and excessive noise; adequate storage; classroom climate comfort and ability to control temperature with a thermostat; ability to control glare, and ability to hear students speaking in a normal tone of voice (ORs ranging from 0.15 to 0.60). 4

5 A sub-study consisting of school walkthroughs and measurements found that approximately 64% of classrooms had CO 2 concentrations above 1,000 parts per million (ppm) at some point during the monitoring period, and in nearly 25% of classrooms CO 2 levels were above 1,000 ppm around 40% of the time (although original guidelines from ANSI/ASHRAE stated CO2 levels should be no more than 700 ppm above outdoor measurements-which can allow indoor levels to then be above 1,000-many studies continue use this as the exposure level of interest because indoor air pollutants begin to build up in the space, and occupants may start to feel uncomfortable and begin to complain of various health symptoms). Classroom temperature and humidity were generally within acceptable ranges. This study found that high CO 2 levels measured in classrooms were significantly associated with teachers reporting single or multiple health symptoms. Teachers who felt their symptoms affected teaching ability were more likely to teach in a classroom with higher measured levels of CO 2, although this was not statistically significant. Summary of key findings and conclusions: The most common unfavorable building conditions in NYS public schools were water damage/leaks and unsatisfactory rating of components of the building envelope. The most common favorable building conditions reported in NYS schools were absence of visible mold and vermin infestations. Schools with good IAQ, especially having fresh air intakes located away from sources of diesel exhaust/pollutants and having good ventilation characteristics, were 94%-184% more likely to have a good attendance rate than other schools. Good lighting fixtures and no active infestations of cockroaches were related to good attendance. Schools with high IAQ total scores and a high GSI were more likely to have high student attendance rates (36% and 22 % respectively). Schools with good condition of air filters were also 42% more likely to have good 4 th grade academic performance. The favorable effects of healthy school building conditions on student attendance and performance mainly appeared in upstate schools and less in downstate schools. School cleanliness and good maintenance of school plumbing systems may contribute to lower school district asthma hospitalization rates. Many health symptoms reported by teachers were potentially work-related, and many of these symptoms were perceived to affect their attendance and/or teaching ability. Most favorable school conditions were associated with reduced reports by teachers of health symptoms. Classroom measurements in 10 NYS schools of temperature and humidity were generally within acceptable ranges. High classroom CO 2 levels were significantly associated with teacher reports of three or more health symptoms as well as the reporting of any allergic symptoms. In summary, this study emphasized the potential importance of healthy school building characteristics, especially IAQ and ventilation as well as cleanliness, thermal comfort, lighting and acoustics, on school occupant health and performance, which affects the functioning of the educational system for teachers and students. These results add to the growing body of literature 5

6 addressing the positive impact of healthy and green school characteristics on occupants, and can be used to guide stakeholders in advocating for and implementing changes in the school environment. INTRODUCTION AND BACKGROUND In recent decades, concerns about the adverse effects of building conditions or indoor environment on occupant health has increased, originating from episodes of Sick Building Syndrome during the 1970s in which occupants of the building including residents, office workers or commercial building owners or buyers complained of health symptoms related to their building (Kreiss, 1989). After recognizing the potential health impacts associated with poor indoor environments, the scientific community also began to research the impact of building conditions on occupant performance and productivity (Mendell et al., 2002). Most early studies in this area were focused on office settings and adults. A group of Nordic scientists established the European Multidisciplinary Scientific Network on Indoor Environment and Health to study the association between ventilation and health (EUROVEN) and conducted a systematic review of scientific literature on the effect of ventilation on health, comfort and productivity. A review of 105 papers, showed that many studies were conducted among adults and the majority evaluated office building or home environments except for two studies which studied school buildings (Wargocki et al., 2002). The EUROVEN group reached a consensus that a strong association existed between ventilation and comfort (as indicated by perceived air quality), health symptoms and productivity (indicated by performance of office work). Understanding the public health impact of the school environment is particularly important because of the amount of time spent there by students and staff, and because persistent shortages of funds inhibit the ability to adequately maintain school facilities compared to other commercial buildings. Children, who represent the main population in schools, may be more susceptible to environmental hazards than adults due to the higher volume of air they breathe relative to their body weight, and due to the rapid growth of their organs. Poor school environments may not only adversely impact student learning and performance, but may also have lifelong economic and social consequences for students. According to a U.S. General Accounting Office s report (1995), about 14 million students in the U.S. attended school in buildings below environmental standards and 63% attended schools with one or more building systems having problems necessitating extensive repair or replacement. Without clear documentation regarding the adverse effect of poor school environment on student health and performance, schools have little motivation or incentive to improve the school environment and protect their students and school staff, especially in the face of tight budgets and financial difficulties. Teacher health and performance is also an important issue, both for teachers and for the students who learn from them. A systematic review of the scientific evidence relating school environmental factors to human performance or attendance was conducted by Mendell et al. (2004). The authors categorized overall evidence into four groups such as persuasive (causal relationship) evidence, strongly suggestive evidence, suggestive evidence, and sparse (no evidence or inconsistent evidence of a relationship). Out of 30 relevant studies (including some literature on non-school indoor workplace settings) reviewed, very little persuasive evidence was found to exist. However, the authors identified strongly suggestive evidence linking higher concentrations of indoor nitrogen dioxide to decreased school attendance by children and suggestive evidence of a relationship between low fresh air ventilation in school buildings and reduced occupant performance. This review also described 6

7 indirect evidence supporting several factors such as excess dampness, microbiological/chemical exposures (primarily from home) and pollutants from outdoors in a causal chain relating indoor environment to occupant attendance and performance. A large body of studies has found associations between poor indoor air quality (IAQ), including low ventilation rates, excess moisture or formaldehyde exposure, and asthma or other allergy diseases. Overall, scientific evidence suggests that poor indoor environmental quality adversely influences the attendance and performance of students, primarily through the mechanism of health effects from indoor exposures (USEPA, 2003a) A comprehensive review of the literature was undertaken by the National Research Council s (NRC) in a report entitled Green Attributes for Health and Learning, which reviewed 20 studies that evaluated the impact of overall building conditions or functionality on student achievement. While all buildings were evaluated for both condition and adequacy, Lewis (2000) concluded that building facility measurements could typically explain 10-15% of the differences in test scores across schools while controlling for other factors. Impacts on students suggested that test scores were 5% lower in schools rated in poor vs. fair condition, and 10% lower when comparing poor with excellent condition (NRC, 2006). The United States Environmental Protection Agency (USEPA) has developed a website and guidance documents that provide an overview of the current evidence that involves good IAQ and impacts on occupants. In general, previous studies have found some correlation between overall building condition and student achievement. However, the NRC committee also identified some major limitations of this research such as using building age as a surrogate for building condition, combining scores for different tests rather than studying individual test scores and inadequate control of socio-demographics or prior test scores at the school level. More importantly, they recommended understanding the impact of specific building conditions by measuring the effect of one or more building characteristics upon student and teacher performance. Although numerous studies and reviews have examined the impact of various aspects of the office environment on building-related symptoms, comfort and productivity (Milton et al., 2000; Seppanen et al., 1999; Wargocki et al., 2002; Wyon, 2004; Seppanen and Fisk, 2005; Fanger, 2000; Wang et al., 2005) and a couple have looked at employee health or perceptions of the environment in relation to the school environment (Norback et al., 1990; Smedje et al., 1997), relatively few studies have specifically addressed health and performance issues in teachers in relation to the school environment. Surveillance data from National Institute for Occupational Safety and Health and the Centers for Disease Control and Prevention indicate that asthma in the educational employment sector is an occupational health problem, especially among teachers, and that work-related asthma is associated with the reporting of exposure to indoor air pollutants (Mazurek et al., 2008). A survey of teachers found a link between building-related symptoms in female teachers and mold exposure (Ebbehoj et al., 2001), and Schneider (2003) found that many teachers attributed their health symptoms to poor school conditions. Research on voice problems suggests that teachers are more subject to this problem than other occupations (Titze et al., 1996; Smith et al., 1998) and one study examined teacher voice disorders in relation to school environmental conditions, where teachers who taught younger students, taught in large rooms with more students and had higher background noise levels were the most affected (Preciado et al., 1998). While these studies are informative, no one study has examined a series of teacher health symptoms in relation to a comprehensive array of classroom conditions. 7

8 Until recently, the green building movement has focused more on the economic benefits or general environmental impacts of green schools rather than effects on occupants. However, as more public schools incorporate green building practices, research is needed to understand the potential public health benefits by assessing the extent to which green schools provide a functional learning environment that is healthy and comfortable. Existing guidance documents such as those from Leadership in Energy and Environmental Design (LEED) and New York Collaborative for High Performance Schools (NY-CHPS) primarily outline tools for measuring the quality of the design, renovation and maintenance of green schools. However, there is currently little framework for evaluating whether green building attributes and practices support school occupant health and productivity. Current challenges raised in the NRC report include lack of a clear definition of what constitutes a green school, the difficulty in measuring educational and productivity endpoints, and the existence of confounding factors (NRC, 2006). Studies that have examined the impact of the indoor environment on occupant performance and health have generally only assessed a single school factor, focused on adults in an office setting, and mostly assessed adverse health effects rather than assessing the positive impacts of green school attributes. Some important school factors such as cleanliness, specific IAQ factors, acoustics and lighting were rarely studied. PROJECT GOALS AND OBJECTIVES: The goal of this study was to assess how healthy green school building attributes may positively affect occupant performance and respiratory health. This was achieved through the following objectives: 1) examine the potential impacts of individual healthy green school factors and their joint effect by developing a green school index (GSI) on the performance and health of students in New York State; 2) assess teachers perception of the school environment and its relation to their selfreported health and performance; and 3) validate the reported school building conditions in a subsample of schools using objective measurements. METHODS: IA. Examine the Potential Impacts of Individual Healthy Green School Factors and Their Joint Effects on the Attendance and Test Scores of Students (Methods) Study Design and Population: A cross-sectional study was conducted to link school building conditions with student attendance and performance data, using individual schools as the unit of analysis. Some important school building data such as IAQ were not available in New York City (NYC), therefore, the population of this study was restricted to public schools in the rest of NYS serving students from kindergarten to grade 12. Data Sources: Information on school environmental conditions was obtained from the 2005 Building Condition Survey (BCS), a physical inspection of NYS public school buildings conducted by a team that includes independent certified inspectors trained as engineers or building architects, hired by each school district. The BCS is mandated by Part 155 of the New York State Commissioner of Education Regulations, also known as RESCUE (Rebuild Schools to Uphold Education). The BCS began in 2000 and is conducted every five years as required by the New York State Education 8

9 Department (NYSED). It provides information on building age and size, and rates the overall building condition as well as the condition of 53 individual building systems including the building envelope, plumbing systems, heating ventilation and air conditioning (HVAC) systems, and roofing. Condition ratings are generally categorized as excellent/satisfactory/unsatisfactory and nonfunctioning/critical failure, although some are categorized as good/fair/poor. In 2005, a new section was added to assess environmental parameters related to comfort and health, including IAQ, ventilation, dryness (mold and moisture), appearance, cleanliness, evidence of vermin, acoustics and lighting quality. Most of these new variables are in a yes/no format, and a few have a rating system of good/ fair/poor/unable to determine. Appendix I provides details on the identity and coding of these variables. School attendance and performance data were obtained from the NYSED School Report Card, a yearly assessment of NYS schools. To assess potential outdoor environmental exposures, school locations were geo-coded using Cybersecurity 2008 data and available mapped data such as distance from school to major roads (New York State Department of Transportation), airports (United States Federal Aviation Administration) and industrial facilities (United States Environmental Protection Agency s Toxic Release Inventory) were computed using MapMarker Plus and MapInfo. Exposure Definition: We identified 66 variables related to healthy green school building conditions from the 2005 BCS data. These variables were selected based on review of previous green building research and guidance as being conceptually and biologically related to occupant health and performance. These variables were initially analyzed as distinct exposures. Subsequently, based upon NRC, LEED and NY-CHPS guidelines, these building-related variables were grouped into five attributes which included: 1) IAQ; 2) dryness; 3) cleanliness; 4) acoustics; 5) lighting. Five additional attributes reflecting well-maintained building systems were created including site utilities, building envelope, building interior, plumbing and HVAC system. As described in Appendix II, the IAQ attribute consists of ten variables including five variables related to air intakes and five variables related to ductwork, filters, dampers, outside air adequacy, and overall rating of IAQ/ventilation. The dryness attributes cover moisture and mold. Questions used to define absence of moisture problems included no visible water damage in classrooms or other areas, no active roof leaks in classroom or other areas, no active plumbing leaks in classroom or other areas, no moisture condensation in classroom or other areas, and good humidity/moisture rating. Information about absence of mold problems was gathered from four variables and defined by having no visible mold in classrooms, common areas, return/supply grilles and other areas. The cleanliness attribute was composed of six variables and defined by ratings of good general appearance, good cleanliness rating, and no active infestation of rodents, wood-boring insects, cockroaches or other vermin. Good acoustics and lighting were each defined by single overall rating variables defined as good and, for lighting, an additional variable reflecting the condition of the light fixtures. The well-maintained systems attributes were defined by using the ratings provided for 53 building systems organized under one of six sections in the BCS, including (Appendix II) site utilities (water or fuel), other site utilities (cosmetic features), building envelope, building interior, plumbing, and HVAC system. Finally, a green school index (GSI) was developed, consisting of one cumulative measure for all of the individual variables described above. As detailed in Appendix II, all individual variables for building conditions were coded and responses representing better conditions were used to calculate and assign a score to each attribute category. 9

10 For each school, a score was calculated for each of the 10 attributes by assigning 1 point for having a better condition rating for that variable and subtracting 1 point for having a worse condition rating. System ratings which were coded not applicable/unable to determine/missing were not assigned any points. Depending upon the question, either a yes or no response could be the better condition; for variables rated on a good/fair/poor scale, a point was assigned for good ratings but not for fair/poor ; for variables rated on an excellent/satisfactory/unsatisfactory/non-functioning/ critical failure scale, a point was assigned for excellent/satisfactory ratings but not unsatisfactory/non-functioning/critical failure ratings. Finally, the GSI score was computed by summing up all variables. For analysis, the median score for individual healthy green attributes and for the GSI was calculated for all schools, and a dichotomous variable was created to categorize schools as above or below the median. Schools with six or more missing responses within their BCS were excluded from the analysis. Outcome Definition: Our main outcomes of interest included average daily student attendance and performance on 4 th grade and 8 th grade standardized tests. Average daily attendance rate is provided by NYSED and represents the average proportion of students reported present on each day throughout the school year. The attendance rate was dichotomized to the 90 th percentile vs. the rest of the distribution for the primary analysis. Student performance was measured using standardized tests scores in English, Math and Science (for 4 th and 8 th graders) for the school year. The NYSED reports these performance measures as a proportion of the student population meeting each of four scoring levels as follows: 1) L1-students meeting proficiency in none of the grade level standards, 2) L2-students needing extra help to meet the grade level standards, 3) L3-students meeting their grade level standards (passing), and 4) L4-students exceeding their grade level standards. The L4 measure was used in this study. As there were very small numbers of schools above 90 th percentile of the L4 proportion, we used 75 th percentile or above. To get a summary measure of academic performance, we averaged performance on the three standardized tests by adding the proportion scoring at L4 across the three test types and dividing by three. As the test scores for three subjects were highly correlated within schools, we combined the scores from all subjects together first and will assess the effects of school environment on each subject test respectively in a further analysis. Potential Confounders and Effect Modifiers: As socio-demographic variables are known factors for affecting school attendance and test performance, these variables were treated as potential confounders and controlled at the school level in our multivariate analyses. Covariates controlled in this study included percent of students eligible to receive free lunches, proportion of white students enrolled in the district, district urbanicity (suburban/rural vs. urban), district pupil per teacher ratio, district expenditure per pupil, student density (enrollment per 1,000 square feet), building age (> 50 yrs vs. < 50 years), and location in downstate counties (Rockland, Westchester, Nassau, and Suffolk) vs. upstate counties. The average daily attendance rate median was used dichotomously to control for above median attendance rate vs. below median attendance rate in the analysis of student test performance. In addition, above or below median enrollment of 4 th graders was included in 4 th grade test performance. The presence of high school grades in the school that houses the lower grades was related to decreased attendance rates, a dichotomized variable (presence vs. no presence of high school grades) was included. To address the potential impact of outdoor environmental exposure, school locations were geo-coded and their proximity to traffic, airports, and industrial facilities were 10

11 controlled as well. The potential interactions between school building conditions and sociodemographic variables were also assessed. Statistical Analysis: All analyses were conducted using SAS version 9.1. The BCS data were linked with the school attendance data and to standardized test scores data using a unique 8-digit school code, school name and address. A univariate analysis was first performed to examine the data distribution of both building conditions and student outcomes, respectively. In bivariate analysis examining the crude relationships between building factors and outcomes, we utilized the Cochran- Mantel-Haenszel methods to estimate odds ratios (OR) and 90% confidence intervals (CI). On crude analysis, the OR was the same order and magnitude and direction as the prevalence ratio and was used to estimate the association. In addition, non-parametric one-way analysis was conducted to test the mean differences of the outcome variables by different levels of school conditions. For multivariate analyses, unconditional logistic regression was used to determine the odds ratio (OR) of positive occupant outcomes under the various healthy school conditions or attribute scores while controlling for the covariates as described above. A 90% CI of the point estimate rather than 95% CI was used due to small sample size for some exposure categories and was equivalent to a one-sided test. As significant interactions by geographic regions (downstate vs. upstate counties) were found between some school environmental factors and the outcomes, in addition to a statewide analysis, stratified analyses were conducted by this variable to compute the stratum-specific results. IB. Statewide Analysis of the Relationship between Individual Healthy Green School Factors and Their Joint Effects on District Asthma Hospitalization Rates (Methods) Study Population and Data Sources: A cross-sectional study of the relationship between non-nyc school district building conditions and asthma hospitalization rates among the school-age population was conducted. Data from the 2005 BCS were used to assess building conditions. Data collected through the Statewide Planning and Research Cooperative System (SPARCS) for years were used to calculate district-level asthma hospitalization rates among school-age children. SPARCS is a legislatively mandated population-based database that contains discharge data for at least 95% of all NYS hospitalizations, excluding those at psychiatric and federal hospitals. Census population data were used to calculate denominators and to create population-based sociodemographic variables for use in multivariate analysis. Exposure Definition: Exposure measures derived from the 2005 BCS at the school building level (See Section 1A. Exposure Definition, for details) were aggregated to the school district level. School district level analysis was conducted, rather than school building level, because school building level geographic boundaries were not available. Furthermore, asthma hospitalizations are relatively rare events and many school buildings may not have any students hospitalized for asthma during the study timeframe. The attribute scores and GSI were weighted by school building enrollment and totaled across all schools in a district. In other words, enrollment-weighted districtlevel measures of school building attributes (e.g., IAQ) were developed for each district by multiplying each attribute score in each school by the school s enrollment, and then summing these scores across schools in the district and dividing by the total district enrollment. For instance, if there were two schools in a given district, and one had an enrollment of 100 and the IAQ attribute score was 3, then the enrollment-weighted score for that school was 300. If the IAQ attribute score in the other school was -2, and this school had an enrollment of 200, then the IAQ score for this school 11

12 would be Summing these scores and dividing by the total district enrollment would yield /300= This same process was completed for each attribute category including indoor air quality, dryness, quiet, light and well-maintained systems. The scores for each attribute were totaled across all districts that had at least one building with BCS data so that median values for each enrollment-weighted score could be calculated. Dichotomized variables for each exposure measure were then created to categorize school districts as having weighted scores above or below the median for use in statistical analysis. The same procedure was conducted for the GSI. Outcome definition: The health outcome in this part of the study was asthma hospitalizations defined by the principal diagnosis code of 493 from the International Classification of Diseases, 9 th Edition among school-age children (5-17 years old) admitted from January 2003 December 2006 (2 years before and one year after 2005 to increase power). Hospitalizations that occurred during the summer months (July and August) were excluded due to school not being in session when exposure to conditions at school was not likely. MapInfo was used to geo-code asthma hospitalizations based on residential address information from the SPARCS record, after which hospitalizations were cumulated within each school district. Hospital admission rates were calculated at the school district level using hospitalization counts by school district as the numerator and the 5-18 year old population from the 2000 U.S. Census data as the denominator. Summary statewide measures were calculated from school district hospitalization rates, including a dichotomized measure to categorize school district as having a rate above or below the statewide median. Statistical analysis and confounders: Variables controlled for in the analysis included the percent of enrolled students eligible for a free lunch, district enrollment, proportion of children in the 5-9 age range and white population proportion. The hospitalization data were merged with 2005 BCS data to get a reliable estimate of disease at district level. Bivariate analysis was conducted to calculate the crude relationship between school district asthma hospitalization rates and weighted-attribute and weighted-gsi scores. The Wilcoxon rank test was used to compare schools districts above and below the statewide median hospitalization rate by school district proportion of race, ethnicity, gender, and age, as well as free lunch and GSI. Multivariate logistic regression analysis was conducted to calculate the adjusted relationship between school district asthma hospitalization rates and weighted-attribute and weighted-gsi after controlling for previously mentioned confounders. II. Statewide Telephone Teacher Survey (Methods) Sampling and Population: A telephone survey of NYS public school teachers asking about their health symptoms and classroom environment was conducted in December, Full-time classroom teachers working in non-nyc schools constituted the target population. We randomly selected 425 schools from the 2005 BCS database of 2,751 schools in 650 districts. This sample was checked for similarity to the database as a whole with respect to 1) the number of full-time teachers working in the district and 2) the ability of the district to meet its needs with the resources at hand (district needs index). The sample was sent to New York State United Teachers (NYSUT) for linkage with their database of union members. NYSUT identified all full-time classroom teachers based on their union list and chose one teacher per school to call based on a randomly generated list for each school. We generated and sent NYSUT a supplementary list of schools in the event that one teacher in each school from the primary school list could not be reached. 12

13 Survey procedures: The survey we developed was reviewed by NYSUT Health and Safety specialists, and conducted by Benchmark Polling Group (an in-house Call Center located at NYSUT headquarters). NYSUT Call Center programmers developed a telescript based on the final survey draft for Call Center interviewers to follow. We also verified and approved the telescript. A detailed protocol of survey administration procedures was developed for interviewers to follow. A pilot study of 25 teachers was first conducted to estimate the time it would take to administer the survey and to identify questions that are time-consuming, unclear or awkward to administer. Since no major problems were identified in this phase, pilot study surveys were included in the final database. Calls were made in December of 2009 from Monday through Thursday between 4:30 and 8:30 PM until the quota was reached. For each school, one attempt was made to reach the randomly chosen respondent. If that teacher could not be reached or refused to participate, the reason was noted and the next teacher on the randomly-generated list was called. This process was continued until a completed survey was obtained from each school off the randomly selected 425 school list. Exposure and Outcome Definition: Questions were asked about IAQ (allergens, irritants and odors), housekeeping (storage, clutter, food consumption and flooring), climate (heat, cold, dryness, humidity and fresh air), lighting (adequacy and glare) and acoustics (noise, echo) in the classroom in which teachers spent the most time. The survey also asked teachers about the type and pattern of 14 health symptoms (sinus problems, allergies, headache, sneezing, throat irritation, fatigue, eye irritation, frequent colds, coughing without cold, skin irritation, inability to focus, wheezing, nausea and asthma attacks) and also asked how certain classroom conditions affected their ability to teach. Questions about demographics, teaching experience and hours, grade level and home environmental conditions were also asked. Statistical Analysis: Data analysis was conducted using SAS version 9.1. Descriptive analysis was first performed to generate information about the proportion of respondents having health symptoms, the pattern of these symptoms and the prevalence of various self-reported classroom conditions experienced in the school year. Information about demographics and home exposures were also summarized. Next, bivariate analysis was conducted using summary measures of health symptoms as outcomes. Having at least one symptom and having at least one allergic-type symptom (sinus problems, allergies, sneezing, throat irritation, eye irritation, coughing without cold, skin irritation, wheezing and asthma attacks) were each examined in relation to classroom indoor air quality, housekeeping and climate. Having at least one lighting-related symptom (headache, fatigue, eye irritation and inability to focus) was examined in relation to classroom lighting, and having at least one symptom potentially related to excessive noise and poor acoustics (headache, fatigue, throat irritation, coughing without cold and inability to focus), and having voice problems was examined in relation to classroom acoustics. Classroom variables associated with these respective outcomes at p<0.10 were chosen for logistic regression analyses controlling for age group (age 50+), gender, years worked (above or below median), working more than 40 hours per week, district size (above or below median), district needs index (high need vs. average/low need), elementary vs. MSHS, exposure to tobacco smoke, living within 300 yards of a busy highway) and having a furred or feathered pet at home. One classroom variable was entered into each of these regression equations. III. School Walkthroughs, Objective Measurements, and Staff Surveys (Sub-study, Methods) 13

14 Overview: This sub-study of 10 schools, a small but economically feasible sample, was implemented to complement the statewide analysis of school building conditions and occupant outcomes. The objective of the sub-study was to: 1) gain an understanding of school conditions by conducting school walkthroughs and collecting real-time objective measurements for some key green attributes; 2) validate the school building environment by comparing the objective measurements with the reported data from the 2010 BCS; and 3) link the quantitative measured data with the teacher surveys collected from these schools to explore if the school environment measurements are associated with teacher reported health and performance. School Recruitment: The schools for this sub-study were selected from the Greater Capital District region of NYS (schools were limited to four local Board of Cooperative Educational Services regions; Capital Region, Hamilton-Fulton-Montgomery, Questar III, and Washington-Saratoga- Warren-Hamilton-Essex), given budget considerations and the economic advantage of being able to make day trips to obtain the data for each school. Eligible schools were identified from a list representing different tertiles of socioeconomic (SES) status and different degree of school greenness, the latter a rough estimate from the 2005 BCS. The objective was to enroll at least 10 schools representing a range of SES and school greenness (i.e., more green vs. less green school conditions; high and low SES). A total of approximately 40 letters were sent to district superintendents who had at least one school in their district meeting our recruitment criteria. It was made clear that participation was voluntary. School administrators with interest in participating were asked to contact our project staff. The research team received a number of phone calls from interested administrators, and follow-up phone conferences and/or in-person meetings were scheduled. The NYSDOH staff attended a school faculty meeting at each school to introduce the project to school administrators, building facility managers and teachers, and to address any questions or concerns. Project staff then worked with school administrators who confirmed a willingness to participate in the project to set-up a date for the walkthroughs. Teachers in these schools were given the opportunity to volunteer their classrooms for assessment and monitoring, up to 10 classrooms per school to ensure adequate time to complete all aspects of the walkthroughs and because of limits on equipment availability. School Assessment and Monitoring: The school walkthrough sub-study included three components; school building observation and real-time measurements, teacher and staff surveys, and nurse surveys in these schools. The school walkthroughs were conducted with the building facility managers to better understand general school conditions, discuss potential problems and gather information about existing school policies. Project staff recorded observation data specific to each classroom including general information about the classroom; the status of ventilation units and lighting; identification of mold or moisture problems; and assessment of potential sources of dust and other asthma triggers. Project staff completed an observational form for participating classrooms while the rooms were unoccupied, either before or after school was in session. While in each classroom, background noise measurements were also recorded using a Bruel & Kjaer Hand- Held Analyzer Type Five U HOBO data loggers (Onset Corporation) recorded temperature and relative humidity, and five U HOBO data loggers (Onset Corporation) recorded temperature, relative humidity, and lighting levels. Telaire 7001 (Telaire) CO 2 monitors were connected to the HOBO data loggers to record classroom CO 2 measurements. Telaire 7001 monitors were calibrated prior to data collection using Calibration Kit Data loggers were previously calibrated by the manufacturer. Data loggers were set to collect measurements every five 14

15 minutes for forty-eight hours. Monitoring data were limited to those times during which the teacher indicated that students were in the classroom and measures of tendency and variation were calculated for temperature, relative humidity, CO 2, and lighting. Teachers and staff worked together to identify the best place to position data loggers. Instruments were placed in locations representative of adult breathing zones and out of the reach of children, generally on top of a file cabinet or shelves, and away from supply air. Once data loggers were in position, teachers were instructed not to move them. Teachers in each of the classrooms in which monitoring data were collected were asked to complete occupancy logs for each school day during the forty-eight hour monitoring period. Survey of Teachers: The teacher survey used for the sub-study was the same survey used for the statewide telephone teacher survey. All teachers and staff in each participating school were given the opportunity to complete a paper survey of their perceptions of the school environment and potentially-related health symptoms. The survey collected demographic information about the teacher and their home environment and general information for the following: 14 specific symptoms potentially related to the indoor environment and their perceived work-relatedness, cleanliness, visible problems (e.g., moisture, dust, dust reservoirs, pests, etc.) and odors, flooring, indoor thermal comfort and climate, noise, lighting, and asthma management policies and practices. Teachers were asked to complete and return the survey within two weeks of the walkthrough date. The surveys collected during the sub-study were coded in the same manner as surveys in the statewide teacher survey. A description of these methods can be found in the methods for the statewide teacher survey. Survey of School Nurses: School nurses in each of the 10 schools participating in the sub-study were asked to complete a paper and pencil survey in the spring of This survey asked about asthma prevalence in students and visits to the health office for treatment. Questions were also asked about asthma management in the school setting, including having a comprehensive action plan for each student and medication administration; school practices, policies and education regarding students with asthma; and organizational infrastructure to help manage asthma and associated environmental issues. It also asked questions about the school allergens, irritants and odors, and asked nurses to rate the IAQ of the school. Questions regarding nurse availability were also asked. Statistical Analysis: Descriptive analysis was carried out using SAS version 9.1. Frequency distributions were calculated for each item on the classroom assessment form. Daily average outdoor temperature, relative humidity, precipitation, and wind speed data from the National Weather Service were averaged over the two days and added to the dataset to come up with a single fortyeight hour measure. To determine the extent to which CO 2 levels reached levels indicative of inadequate ventilation (1,000 parts per million (ppm)), the number of monitoring values greater than or equal to 1,000 ppm was calculated and divided by the total number of monitoring values during times when students were present to determine the proportion of CO 2 values above 1,000 ppm during occupied times for each classroom. Similar calculations have been used previously as an indicator of ventilation (Seppanen et al., 1999; Simoni et al., 2010). The median was calculated from these proportions and a dichotomous variable representing classrooms above and below the median was created. Bivariate analyses were conducted to assess the relationship between CO 2 levels and: 1) ventilation units in each classroom and 2) teacher symptom reporting. Both bivariate analyses were limited to data from traditional classrooms to improve comparability. Several outcome variables were assessed for the analysis of teacher symptoms and CO 2 levels, including: 1) median number of symptoms reported, 2) any symptom reported, 3) any allergy-related symptom reported 15