Association of school infrastructure on health and achievement among children with asthma

Abstract

Objective:

To determine whether school infrastructure is associated with health and academic outcomes among elementary school children with asthma.

Methods:

We conducted a retrospective cohort study of linked medical, academic, and facilities data from a large mid-Atlantic school district of the United States. All K—5 students with asthma who were enrolled under the state’s Children’s Health Insurance Program were included. We estimated associations of the infrastructure quality of the student’s school, as assessed by an engineering firm in Summer 2011 and represented by the Facility Condition Index (FCI), with asthma health outcomes, absenteeism, and standardized test scores in math and reading in the two academic years thereafter.

Results:

6,558 students were identified, the majority non-Hispanic Black, across 130 schools. Most schools (97/130, 75%) were in very poor or worse condition. In cluster-adjusted models accounting for demographics, grade, school-specific area deprivation, and inhaled corticosteroid use, a one standard deviation increase in FCI, corresponding to greater infrastructure deficiency, was associated with higher rates of asthma-related hospitalizations (incidence rate ratio [IRR] 1.16; 95% confidence interval [CI] 1.03, 1.32), more absenteeism (IRR 1.05; 95% CI 1.01, 1.08), and lower scores in math (mean difference [MD] −3.3; 95% CI −5.5, −1.2) and reading (MD −3.0; 95% CI −5.1, −0.9). There were no differences in rates of asthma-related emergency visits or steroid prescriptions.

Conclusions:

Children with asthma attending schools with poorer infrastructure had worse health and academic outcomes. Public policy emphasizing reinvestment in school infrastructure may be a potential means of addressing asthma disparities.

INTRODUCTION

Despite recognition that school conditions are important, America’s schools are inadequately maintained. Over half of schools in the United States reported needing to spend money on infrastructure repairs in 2012, and these schools were more likely to be in cities and have a higher prevalence of students living in poverty.¹ Deficiencies in school infrastructure are in turn linked to worse indoor air quality, including higher concentrations of nitrogen dioxide and carbon monoxide.² Because children spend six to eight hours per day in school, these deficiencies may have an important influence on health and academic outcomes, especially among elementary school-aged children, when lung and brain development is most vulnerable.³

Asthma, the most common chronic childhood illness, is a condition especially sensitive to environmental quality.^4,5 Exposure to indoor air pollution, allergens, and temperature extremes are known to provoke asthma exacerbations, and environmental remediation has been studied as a tool to improve asthma outcomes.^6,7 Almost all such studies, however, have focused on the home, and there is a need to further define the impact of the school on children with asthma.^8,9 Inadequate ventilation and poor thermal conditions have been linked to worse academic performance in some studies.^10,11 Yet, few have investigated effects on health, and none to our knowledge studied children with chronic respiratory illness.

Therefore, we sought to comprehensively examine the impact of school infrastructure on health outcomes, absenteeism, and academic achievement in young children with asthma. In 2011, a large mid-Atlantic urban school district initiated a decade-long project to renovate and replace the district’s aging educational buildings. As part of this program, a complete assessment of the district’s facilities was performed, providing a unique opportunity to address our objective across an entire district.¹² We hypothesized that infrastructure deficiencies would be associated with a higher risk of asthma exacerbation, greater absenteeism, and lower academic achievement among elementary school-aged children with asthma.

METHODS

Study design

We performed a retrospective study of students with asthma enrolled in the district’s elementary grades (K-5) from September 2011 to June 2013. These years were chosen as they were immediately after the district-wide facility condition assessment, before major renovations undertaken by the district in response to the assessment, and sufficiently distant to address concerns related to privacy.

Data linkage was performed at the individual level and is described in Figure 1. Identifiers of all students enrolled during the time period of interest were provided by the school district to the Hilltop Institute at the University of Maryland, the data steward of the Maryland Children’s Health Insurance Program (MCHP), the state's Medicaid expansion program.¹³ Healthcare claims data were extracted for all matched students who were continuously enrolled in MCHP/Medicaid, with dates aggregated to the month level due to the sensitive nature of the information. The matched dataset was sent to the Baltimore Education Research Consortium (BERC), a research-practice partnership between the school district and Maryland-based universities, which linked each student with their individual attendance and achievement information and with facility-level information based on enrollment records. All individual identifiers were subsequently removed from the dataset before it was made available to the investigators.

Figure 1.

Linkage process. Identifiers sent by the school district to MCHP were removed after linkage to protect student confidentiality; MCHP: Maryland Children’s Health Program; BERC: Baltimore Education Research Consortium

We identified children with asthma from this dataset, defined by a prescription fill for an inhaled corticosteroid or at least one health encounter with an International Classifications of Disease (ICD) code for asthma during the two-year span. No exclusion criteria were applied. We also considered alternative definitions of (1) at least one ICD code for asthma and (2) at least one inpatient or two outpatient codes for asthma, and these did not meaningfully impact interpretation of outcomes.

This study was conducted with regulatory and ethics approval from the Institutional Review Boards or equivalents of the school district, State of Maryland Department of Health, Johns Hopkins University School of Medicine, and the Hilltop Institute at the University of Maryland.

Variable definitions

The primary exposure was the facility condition index (FCI) of the student’s school as measured in Summer 2011.¹² The FCI is an industry standard measure and represents the ratio of the cost of addressing a school’s infrastructure deficiencies over the cost of rebuilding the school altogether, allowing comparisons of deficiencies across different campuses. It was objectively determined through on-site inspection conducted by a professional engineering firm.¹² Factors related to indoor air quality are key to the FCI, including heating, ventilation, and air conditioning (HVAC), building envelope (the integrity of the barrier between conditioned indoor air and outside), and plumbing. We examined FCI as either a continuous measure or by the deficiency classification in the report: ≤0.1 (good), 0.11-0.30 (average), 0.31-0.50 (poor), 0.51-0.74 (very poor), and ≥0.75 (replace). As a student could transfer schools between academic years, the student’s corresponding FCI was updated between years.

Outcomes were ascertained for the two academic years following facilities assessment. We examined health outcomes related to asthma exacerbation. We identified asthma-related steroid fills (an oral corticosteroid fill within fourteen days of a healthcare visit for asthma), asthma-related emergency department (ED) visits (an ED visit with asthma as the first or second diagnostic code), and asthma-related hospitalizations (an inpatient admission with asthma as the principal diagnosis).^14,15

Academic achievement in math and reading was assessed by the Maryland State Assessment (MSA) test. The MSA was administered to 3-5^th graders. The scores range from 240-650, with higher scores representing better academic performance. Absenteeism was defined by the count of days absent and enrolled.

Additional covariates were extracted for purposes of regression adjustment. Demographic information on race and ethnicity was provided by MCHP and BERC, and inhaled corticosteroid use was determined based on prescription refill data. Race and ethnicity were included due to known racial and ethnic differences in risk of asthma exacerbation. We also extracted the area deprivation index (ADI) of each school over the survey period. The ADI is a summative representation of neighborhood-level socioeconomic status and incorporates measures such as median family income and housing value.¹⁶ The ADI is nationally scaled to the 1—99^th percentile, with higher values denoting greater degree of deprivation. The ADI was unavailable for two schools due to data suppression and were singly imputed at the mean value; exclusion of these schools did not meaningfully alter results.

Statistical analysis

We summarized student- and school-level characteristics. The association of FCI with monthly rates of asthma-related ED visits, hospitalizations, and corticosteroid prescriptions was evaluated using multivariable negative binomial regression. The association of FCI with annual number of days absent was evaluated in a similar fashion with the additional inclusion of an offset term for the number of days enrolled. The association of FCI to MSA scores was estimated by multivariable linear regression.

Because academic achievement is measured once per year, the primary analysis required students be enrolled in the same school through an academic year (those who transferred schools between years but stayed at the same school each year were included). For other outcomes, we performed a sensitivity analysis which allowed students to transfer schools, enroll, or disenroll during the academic year.

All models were adjusted for grade, sex, race, ethnicity, inhaled corticosteroid use, school-based area deprivation index, and calendar time (month-year for health outcomes and year for attendance and achievement outcomes) and accounted for the complex multi-way clustering among the student’s repeated measures and the student/school observations as students were permitted to transfer schools within the study period¹⁷. Results of negative binomial models are presented as incidence rate ratios (IRR) with corresponding 95% confidence intervals (CI) and linear regression models as mean differences (MD) with 95% CIs corresponding to a 0.25 (one standard deviation) increase in the FCI. We repeated the previously described analyses categorizing FCI by the report classification (0.0-0.30, 0.31-0.50, 0.51-0.74, and ≥0.75). A two-sided p-value of <0.05 was used to define statistical significance. Analyses were performed in Stata 15 (StataCorp; College Station, TX).

RESULTS

Student and school characteristics

Approximately two-thirds of students in this district were continuously covered by MCHP/Medicaid, reflecting a large proportion of children attending from low-income households. Of the 49,796 unique elementary school students enrolled at any time during academic years 2011—2012 or 2012—2013, 33,201 (67%) were successfully matched to health utilization, achievement and attendance, and facility condition data. Within the linked dataset, 7,869 students had an ICD code for asthma or filled an inhaled corticosteroid prescription, resulting in an overall asthma prevalence rate of 23.7%. Of these students with asthma, 6,558 (83%) were continuously enrolled in a given elementary school for at least one academic year and comprised the final study population.

Most students with asthma were male, Black, and non-Hispanic (Table 1). Kindergarteners were over-represented due to first-time enrollees across the two academic years. FCIs of the 130 elementary schools that students attended ranged from 0.03 to 1.23, with 75% of schools considered as very poor or candidates for complete replacement with an FCI ≥0.51 (eFigure 1 at the school level, eFigure 2 at the student level). At the time of facility assessment, the average school was 52 years old, and the oldest school was 101 years old.

TABLE 1.

Participant, school, and outcome characteristics

STUDENT-LEVEL CHARACTERISTICS (N=6,558)
Grade Level	n	%
K	2,010	30.7
1	1,060	16.2
2	967	14.8
3	845	12.9
4	853	13.0
5	823	12.6
Female Sex	2,682	40.9
Race
Black	5,909	90.1
White	546	8.3
Othera	103	1.6
Hispanic Ethnicity	246	3.8
SCHOOL-LEVEL CHARACTERISTICS (N=130)
Facility Condition Index
Mean (SD)	0.63	0.25
Average to Great	17	13.1
Poor	16	12.3
Very Poor	57	43.9
Replace	40	30.8
Area Deprivation Index, mean (SD)	65	24
OUTCOME CHARACTERISTICS
Asthma-Related Outcomes
ED Visits	4,319
Hospitalizations	367
Steroid Prescription Fills	4,684
Attendance
Days Absent	133,604
Days Enrolled	1,904,574
Math Grades, mean (SD)a	397	33.01
Reading Grades, mean (SD)a	401	37.98

^aAmerican Indian/Alaskan Native, Asian, Native Hawaiian/Other Pacific Islander, multiple races

^bAmong 3-5^th graders (3,201 students with scores over two years)

Worse facility condition was associated with higher risk of asthma-related hospitalizations

Of the 6,558 students, 2,273 (35%) had at least one asthma-related ED visit, 300 (5%) had at least one asthma-related hospitalization, and 2,382 (36%) had at least one asthma-related corticosteroid fill (Figure 2). Over 105,900 person-months of observation, there were 4,319 asthma-related ED visits, 367 asthma-related hospitalizations, and 4,684 asthma-related corticosteroid prescriptions.

Figure 2.

Prevalence of asthma-related health events during the study period.

Figure 3 summarizes the associations of FCI to study outcomes, and model coefficients are included in the supplement. In adjusted analysis, an increase in FCI of 0.25, approximately one standard deviation, was associated with a 16% higher rate of asthma-related hospitalizations (incidence rate ratio [IRR] 1.16; 95% confidence interval [CI] 1.03, 1.32) but not with differences in asthma-related ED visits (IRR 1.02; 95% CI 0.97, 1.07) or corticosteroid prescriptions (IRR 1.01; 95% CI 0.95, 1.08) (eTable 1). Because the number of asthma-related ED visits was unexpectedly high, we explored an alternate definition, an ICD code in the first diagnostic position rather than first or second positions and identified 1,940 visits (a 55% reduction). However, FCI was not associated with a difference in ED visit rates (IRR 1.02; 95% CI 0.94, 1.10) with this more restrictive definition. The results remained consistent when FCI was modeled as a categorical variable (eTable 2).

Figure 3.

Association of facility condition index (FCI) with study outcomes. Achievement scores were only collected for 3-5^th graders. aIRR: adjusted incidence rate ratio; aMD: adjusted mean difference; CI: confidence interval; ED: emergency department

We performed a sensitivity analysis including an additional 1,311 students with asthma who had enrolled, disenrolled, or transferred schools during the academic year. In this larger population of 7,869 students, FCI was associated with higher risk of asthma-related hospitalizations (IRR 1.14; 95% CI 1.01, 1.29) but not with other health outcomes.

Worse facility condition was associated with higher absenteeism and lower student achievement

Over 1,904,217 days of enrollment, there were 133,553 absences, corresponding to an overall absenteeism rate of 7%. A 0.25-point increase in FCI was associated with a 5% higher absenteeism rate (aIRR 1.05; 95% CI 1.01, 1.08) (eTable 3). Including students who enrolled, disenrolled, or transferred mid-year did not meaningfully alter the association (aIRR 1.07; 95% CI 1.02, 1.11).

MSA scores were available for 3,201 students (including second graders who completed the exam during the second year of observation). A 0.25-point increase in FCI was associated with both lower math scores (adjusted mean difference [aMD] −3.31; 95% CI −5.47, −1.15) and reading scores (aMD −3.02; 95% CI −5.12, −0.91), corresponding to a standardized effect size of 0.1 and 0.08, respectively (eTable 3). Because absenteeism is plausibly in the causal pathway between infrastructure and grades, we reran the model with additional adjustment for days absent from school and found minimal attenuation of the relationship between FCI and math (aMD −3.0; 95% CI −5.09, −0.91) and reading (aMD −2.9; 95% CI −4.96, −0.84) scores.

DISCUSSION

In this retrospective cohort study of 6,558 elementary school students with asthma, we report that children enrolled in schools with worse facility condition had higher rates of hospitalization due to asthma, higher rates of absenteeism, and worse academic performance in math and reading. To our knowledge, this is the first study to describe an association between poorer school infrastructure, defined by the FCI, and worse individual-level health and academic outcomes for students. Our investigation suggests that this industry-standard measure captures components of school condition that may have tangible effects on young children with asthma.

Proper and timely facilities maintenance is central to Environmental Protection Agency recommendations for maintaining adequate school environmental quality.¹⁸ Yet, three-fourths of schools in this study were considered in very poor or replacement condition when they were first assessed. School districts serving predominantly people of color and those sited in cities, such as the school district represented in this study, receive significantly less funding per child compared to districts serving White populations and those in suburban and rural areas, a serious inequity that is magnified by the chronic under-investment in school infrastructure that occurs at a national level.^19,20 This study reinforces that the consequences of such policy decisions impact not only academic achievement but also public health, and it provides important rationale for a renewed emphasis on public policy that prioritizes school infrastructure. These data further support broad research relating deficiencies in school infrastructure and environment to worse health and academic outcomes among all students.^21,22

The impact of chronic illnesses such as asthma disproportionately affect children who reside in low-income communities and who identify as racial or ethnic minorities.^23-25 This is reflected in our data, where both the prevalence of asthma and the rates of healthcare utilization among those with asthma were higher than national averages.²⁶ These results, when taken together with reports identifying worse school infrastructure among districts serving low-income communities and minorities, also suggest that some socioeconomic and racial disparities in asthma outcomes may be explainable by disparities in school infrastructure.

Our results bolster current evidence relating school infrastructure and environmental quality with worse academic performance and absenteeism in children.^{11,27,28,10,29} We note that the effect size, while modest when traditionally interpreted, is considered significant within the context of educational policy research.^30,31 Interestingly when these outcomes are considered together we found that the relationship between FCI and lower academic performance was not substantially explained by absenteeism, reinforcing the potential pleiotropic impacts of poor school infrastructure.

More importantly, this investigation addresses the comparatively limited understanding of the association between school infrastructure and asthma outcomes.^32,33 The association between FCI and asthma-related hospitalizations is notable, as hospitalizations reflect more severe asthma exacerbations and have health and academic repercussions well beyond the index event, and they represent a substantial proportion of total healthcare spending and societal cost attributable to asthma.³⁴ Therefore, investments in improving school infrastructure may have significant benefit for the health and academic growth of the children and may, in turn, result in healthcare cost savings.

Conversely, we found no association between FCI and asthma-related ED visits or corticosteroid prescriptions. In contrast to hospitalizations, the decision to present for emergency care is often influenced by factors other than disease severity, especially structural barriers related to healthcare access.³⁵ This is corroborated by one-third of our population registering an ED visit for asthma. Alternative use of a more restrictive case definition produced an implausible estimate of asthma ED visit rates that was lower than national estimates among all children, suggesting that either approach may be imprecise in this population.³⁶ These considerations are also relevant for corticosteroid prescriptions. While we conditioned the corticosteroid fill to be within a certain number of days of an asthma-related clinical visit, in line with other studies, this definition has not yet been validated in this population.¹³ Due to privacy considerations limiting data availability, we were unable to examine clinical charting to determine the accuracy of these measures, and this is an area for further investigation.

This study should also be contextualized within the overall literature base highlighting home environmental remediation as a method to improve asthma outcomes³⁷. Addressing infrastructure deficiencies in the school may be a valuable component of a more comprehensive strategy to improve environmental exposures in the built environment.

While we have identified the FCI as a valuable predictor of student outcomes, the integrated nature of the measure does not permit inference on which specific aspects of facility infrastructure are most important. Based on prior literature, we anticipate that these would be most strongly related to elements which plausibly impact environmental quality, including HVAC (temperature, humidity, carbon dioxide), structural envelope (entrainment of ambient air pollution, rodents, and pests), and plumbing (dampness and mold).^9,33 Because the FCI is somewhat standardized to campus size, it also does not capture all aspects of school quality which may plausibly affect asthma outcomes, such as crowding. As the district’s comprehensive renovation and replacement program finishes, future opportunities to measure the longitudinal impact of school infrastructure will become available. Preliminary data has already indicated that renovations have been effective in improving indoor air quality among schools in this district, establishing rationale for investigating the causal effect of school renovation on academic and health outcomes.³⁸

This study has two notable strengths. First, the study population comprised Medicaid-eligible children, most of whom were racial minorities, a population known to be at high risk for adverse asthma outcomes and therefore more vulnerable to infrastructure deficiencies and inequitable public policy. Second, the ability to link data sources from many stakeholders provides a unique level of insight into the potential health and academic impacts of school infrastructure. However, some limitations are acknowledged. First, our results are obtained from a single urban school district, and not all students in the district received health coverage through MCHP; thus, caution should be used when applying our conclusions to students and districts not represented by this study. Second, because of school zoning, there may be confounding through geospatial factors, such as socially advantaged students attending better schools and differences in ambient air quality, teacher retention, and classroom size. However, because of district-wide school choice policies in effect during this period, students were not required to attend their zoned school, accounting for approximately one-third of students; additionally, the study design also allowed FCI to vary across time to account for biases introduced by such transfers, and we adjusted for local school-area deprivation. Residual geospatial confounding around the student’s home remains possible, but we did not have access to this information due to data limitations. Third, although all students were enrolled in MCHP, allowing some inference of socioeconomic status and healthcare access, we did not have access to more granular data to predict these factors more accurately. Finally, students may be absent from school for reasons not attributable to infrastructure deficiencies. However, this would bias relationships to the null, as the degree of misclassification would not be expected to be different among strata of FCI.

In a retrospective cohort study of 6,558 elementary school-aged children with asthma, we found that students enrolled in schools with worse infrastructure experienced greater rates of hospitalization due to asthma, were more likely to be absent from school, and had lower academic achievement in math and reading. These results underscore the possibility that school infrastructure deficiencies may harm the learning and health of children with asthma. Further study on the changes in these outcomes associated with infrastructure remediation are warranted to strengthen causal inference and to estimate potential magnitudes of effect.

WHAT’S NEW.

In a large, district-wide study of the effects of school infrastructure on asthma, poor facility condition was found to predict asthma-related hospitalizations, absenteeism, and lower math and reading achievement. Chronic underinvestment in schools may exacerbate disparities in asthma.