Advertisement

City-Specific Air Quality Warnings for Improved Asthma Self-Management

Open AccessPublished:June 22, 2019DOI:https://doi.org/10.1016/j.amepre.2019.03.022

      Introduction

      This study presents a framework for identifying “high-risk” days for asthma attacks associated with elevated concentrations of criteria pollutants using local information to warn citizens on days when the concentrations differ from Environmental Protection Agency Air Quality Index (AQI) warnings. Studies that consider the unique mixture of pollutants and the health data specific to a city provide additional information for asthma self-management. This framework is applied to air pollution and asthma data to identify supplemental warning days in Houston, Texas.

      Methods

      A four-step framework was established to identify days with pollutant levels that pose meaningful increased risk for asthma attacks compared with baseline. Historical associations between 18,542 ambulance-treated asthma attacks and air pollutant concentrations in Houston, Texas (2004–2016; analyzed in 2018), were analyzed using a case-crossover study design with conditional logistic regression. Days with historically high associations between pollution and asthma attacks were identified as supplemental warning days.

      Results

      Days with 8-hour maximum ozone >66.6 parts per billion for the 3 previous days and same-day 24-hour nitrogen dioxide >19.3 parts per billion pose an RR of 15% above baseline; concentrations above these levels pose an increased risk of 15% (RR=1.15, 95% CI=1.14, 1.16) and 30% (RR=1.30, 95% CI=1.29, 1.32), respectively. These warnings add an additional 12% days per year over the AQI warnings.

      Conclusions

      Houston uses this framework to identify supplemental air quality warnings to improve asthma self-management. Supplemental days reflect risk lower than the National Ambient Air Quality Standards and consecutive poor air quality days, differing from the AQI.

      INTRODUCTION

      The link between exposure to air pollution and asthma attacks is well established but complex, thus making asthma management difficult.

      U.S. Environmental Protection Agency. Health risk and exposure assessment for ozone. https://nepis.epa.gov/Exe/ZyPDF.cgi/P100KBUF.PDF?Dockey=P100KBUF.PDF. Published August 2014. Accessed October 1, 2018.

      U.S. Environmental Protection Agency. Integrated science assessment for carbon monoxide. https://ofmpub.epa.gov/eims/eimscomm.getfile?p_download_id=494432. Published January 2010. Accessed October 1, 2018.

      U.S. Environmental Protection Agency. Integrated science assessment for oxides of nitrogen – health criteria. http://ofmpub.epa.gov/eims/eimscomm.getfile?p_download_id=526855. Published January 2016. Accessed October 1, 2018.

      U.S. Environmental Protection Agency. Integrated science assessment for sulfur oxides – health criteria. http://ofmpub.epa.gov/eims/eimscomm.getfile?p_download_id=533653. Published December 2017. Accessed October 1, 2018.

      U.S. Environmental Protection Agency. Integrated science assessment for particulate matter. http://ofmpub.epa.gov/eims/eimscomm.getfile?p_download_id=494959. Published December 2009. Accessed October 1, 2018.

      • Alhanti BA
      • Chang HH
      • Winquist A
      • Mulholland JA
      • Darrow LA
      • Sarnat SE
      Ambient air pollution and emergency department visits for asthma: a multi-city assessment of effect modification by age.
      • Gasana J
      • Dillikar D
      • Mendy A
      • Forno E
      • Ramos Vieira E
      Motor vehicle air pollution and asthma in children: a meta-analysis.
      • Li J
      • Ewart G
      • Kraft M
      • Finn PW
      The public health benefits of air pollution control.
      • Penard-Morand C
      • Raherison C
      • Charpin D
      • et al.
      Long-term exposure to close-proximity air pollution and asthma and allergies in urban children.
      • Raun LH
      • Jefferson LS
      • Persse D
      • Ensor KB
      Geospatial analysis for targeting out-of-hospital cardiac arrest intervention.
      • Raun LH
      • Ensor KB
      • Persse D
      Using community level strategies to reduce asthma attacks triggered by outdoor air pollution: a case crossover analysis.
      • Silverman RA
      • Ito K
      • Freese J
      • et al.
      Association of ambient fine particles with out-of-hospital cardiac arrests in New York City.
      • Wendt JK
      • Symanski E
      • Stock TH
      • Chan W
      • Du XL
      Association of short-term increases in ambient air pollution and timing of initial asthma diagnosis among Medicaid-enrolled children in a metropolitan area.
      Triggers can differ by pollutant, mixtures of pollutants, and duration of individual exposure to pollutants.
      • Li J
      • Ewart G
      • Kraft M
      • Finn PW
      The public health benefits of air pollution control.
      • Silverman RA
      • Ito K
      • Freese J
      • et al.
      Association of ambient fine particles with out-of-hospital cardiac arrests in New York City.
      Ambient air quality is unique to each city and influenced by meteorology, geography, vegetation, and activities taking place in the city. Studies have shown the impact of local differences in air quality on asthma.
      • Bell ML
      HEI Health Review Committee
      Assessment of the health impacts of particulate matter characteristics.
      • Domínguez-López D
      • Adame JA
      • Hernández-Ceballos MA
      • Vaca F
      • De la Morena BA
      • Bolívar JP
      Spatial and temporal variation of surface ozone, NO and NO at urban, suburban, rural and industrial sites in the southwest of the Iberian Peninsula.
      • Keeler GJ
      • Morishita M
      • Wagner JG
      • Harkema JR
      Characterization of urban atmospheres during inhalation exposure studies in Detroit and Grand Rapids, Michigan.
      • Lobdell DT
      • Isakov V
      • Baxter L
      • Touma JS
      • Smuts MB
      • Özkaynak H
      Feasibility of assessing public health impacts of air pollution reduction programs on a local scale: New Haven case study.
      • Tang G
      • Wang Y
      • Li X
      • Ji D
      • Hsu S
      • Gao X
      Spatial-temporal variations in surface ozone in Northern China as observed during 2009–2010 and possible implications for future air quality control strategies.
      The most widespread method to warn for potentially hazardous levels of air pollution is the U.S. Environmental Protection Agency (EPA) Air Quality Index (AQI), a tool that provides local information on air quality to the public and recommends how to protect human health.

      AirNow. Air quality index (AQI) basics. https://airnow.gov/index.cfm?action=aqibasics.aqi. Published 2016. Accessed March 24, 2017.

      This system maps concentrations of criteria pollutants to an indexed number and a corresponding air quality category (good, moderate, unhealthy for sensitive groups, unhealthy, very unhealthy, or hazardous). As people with asthma are considered a sensitive group, knowing when air quality is unhealthy for sensitive groups or worse is important information for asthma self-management. The AQI does not address local pollutant mixtures or exposures longer than a day.
      City-specific characterizations of air quality can address both and, therefore, can supplement AQI warnings. Although AQI warnings remain the most common type of alert, the City of Houston developed supplemental warning days to enhance public health protection. This supplemental information aids public officials, healthcare providers, and affected residents in efforts to reduce asthma exacerbation from poor air quality.
      • Keller SA
      • Koonin SE
      • Shipp S
      Big data and city living – what can it do for us?.

      Utah Department of Health. Air quality recess guidance for schools. www.health.utah.gov/asthma/air%20quality/Recess%20Guidance.html. Published 2013. Accessed October 3, 2013.

      • Dimitrova R
      • Lurponglukana N
      • Fernando HJS
      • et al.
      Relationship between particulate matter and childhood asthma – basis of a future warning system for central Phoenix.
      Case-crossover with conditional logistic regression is frequently used to analyze associations between air pollution and acute health effects. This method has been used previously to demonstrate local associations between factors such as ozone (O3), particulate matter ≤2.5 microns, sulfur dioxide (SO2), carbon monoxide (CO), nitrogen dioxide (NO2), and temperature with negative health effects using pollution lagging structures.
      • Raun LH
      • Ensor KB
      • Persse D
      Using community level strategies to reduce asthma attacks triggered by outdoor air pollution: a case crossover analysis.
      • Litchfield IJ
      • Ayres JG
      • Jaakkola JJ
      • Mohammed NI
      Is ambient air pollution associated with onset of sudden infant death syndrome: a case-crossover study in the UK.
      • Raza A
      • Dahlquist M
      • Lind T
      • Ljungman PL
      Susceptibility to short-term ozone exposure and cardiovascular and respiratory mortality by previous hospitalizations.
      • Tsai SS
      • Tsai CY
      • Yang CY
      Fine particulate air pollution associated with increased risk of hospital admissions for hypertension in a tropical city, Kaohsiung, Taiwan.
      • Ensor KB
      • Raun LH
      • Persse D
      A case-crossover analysis of out-of-hospital cardiac arrest and air pollution.
      These methods, however, have not been used previously to inform public health warnings. Case-crossover designs are useful when exposure to the factors is not constant. In case-crossover, a health event (case) serves as its own reference, and individuals act as their own control. Airborne pollution concentrations are used in proxy for personal exposure. Ambient air pollution concentrations when individuals experience an adverse health effect are compared with reference concentrations on different days. Conditional logistic regression is used to approximate the association between pollution and increased RR of the health event while controlling for confounding factors.
      • Ensor KB
      • Raun LH
      • Persse D
      A case-crossover analysis of out-of-hospital cardiac arrest and air pollution.
      These estimated increases in the risk of observing an adverse health event can be used to establish additional pollution concentration–based warnings.
      This work aims to increase the accuracy of how Houston identifies and communicates the link between air quality and asthma risk. Raun et al.
      • Raun LH
      • Ensor KB
      • Persse D
      Using community level strategies to reduce asthma attacks triggered by outdoor air pollution: a case crossover analysis.
      found an association between ambient air pollution and emergency medical service ambulance-treated asthma attacks in Houston using case-crossover analysis with conditional logistic regression. This previous work indicated that people with asthma in Houston were at an increased risk from exposure to consecutive days of O3 and NO2 at levels lower than the National Ambient Air Quality Standards, but it did not identify city-specific concentration warning levels. Houston uses the framework presented in this study to define city-specific concentration levels. Days with high asthma risk were defined by pollution levels with risk 15% above the established historical baseline. These warnings augment the EPA warnings. They are issued on days when the EPA does not signal an air quality warning for sensitive subpopulations. This study provides a methodology to identify supplemental warning days, currently used by the City of Houston to supplement those issued by EPA.

      METHODS

      Study Sample

      The analysis assessed the relationship between ambulance-treated asthma attacks and the air pollutants CO, NO2, particulate matter ≤2.5 microns, O3, and SO2. These pollutants are measured in Houston with a fixed site monitoring network and are known to exacerbate asthma.

      U.S. Environmental Protection Agency. Health risk and exposure assessment for ozone. https://nepis.epa.gov/Exe/ZyPDF.cgi/P100KBUF.PDF?Dockey=P100KBUF.PDF. Published August 2014. Accessed October 1, 2018.

      U.S. Environmental Protection Agency. Integrated science assessment for carbon monoxide. https://ofmpub.epa.gov/eims/eimscomm.getfile?p_download_id=494432. Published January 2010. Accessed October 1, 2018.

      U.S. Environmental Protection Agency. Integrated science assessment for oxides of nitrogen – health criteria. http://ofmpub.epa.gov/eims/eimscomm.getfile?p_download_id=526855. Published January 2016. Accessed October 1, 2018.

      U.S. Environmental Protection Agency. Integrated science assessment for sulfur oxides – health criteria. http://ofmpub.epa.gov/eims/eimscomm.getfile?p_download_id=533653. Published December 2017. Accessed October 1, 2018.

      U.S. Environmental Protection Agency. Integrated science assessment for particulate matter. http://ofmpub.epa.gov/eims/eimscomm.getfile?p_download_id=494959. Published December 2009. Accessed October 1, 2018.

      The statistical metric evaluated was the daily average with the exception of O3, which was the 8-hour maximum, in agreement with National Ambient Air Quality Standards criteria pollutant standards. Hourly pollution data were obtained from the Texas Commission on Environmental Quality's online database.

      Texas Commission on Environmental Quality. Welcome to the Texas Air Monitoring Information System (TAMIS) Web Interface.www17.tceq.texas.gov/tamis/index.cfm?fuseaction=home.welcome. 2014. Accessed March 27, 2018.

      These data were collected using EPA federal reference methods and validated by the Texas Commission on Environmental Quality. Data related to ambulance-treated asthma attacks were obtained from the Houston Fire Department emergency medical service call database where the working assessment was asthma treated with nebulized albuterol. There were 18,587 ambulance-treated asthma attacks in Houston during the study period (January 4, 2004–June 15, 2016). The data were analyzed in 2018.

      Measures

      The analysis was conducted in four steps. First, conditional logistic regression with case-crossover analysis of ambulance-treated asthma attacks was conducted for each pollutant to identify pollutants and lags with a statistically significant association, controlling for apparent temperature. This analysis was conducted in single- and double-pollutant combinations at the following daily pollutant time lags: 0, 1, 2, 3, 0–2, 1–2, and 1–3. The lagging structure was used to identify the effects of previous pollution on asthma outcomes and was informed by previous research.
      • Raun LH
      • Ensor KB
      • Persse D
      Using community level strategies to reduce asthma attacks triggered by outdoor air pollution: a case crossover analysis.
      Each pollution variable's lag value was divided by its IQR to reduce the influence of differently scaled concentrations of pollution. Pollutants and lags with no significant association with ambulance-treated asthma attacks were not evaluated further.
      In the second step, significant pollutants and lags were further evaluated to identify the concentration level associated with a risk of ≥15% above the baseline risk. The baseline was defined as days with low to no risk of pollution-triggered asthma attacks. Because previous research indicated that high levels of 8-hour maximum O3 with a lag of 0–2 and daily average NO2 pose a risk, low levels of these pollutants and their lags were defined as reference baseline days (i.e., below the median).
      • Raun LH
      • Ensor KB
      • Persse D
      Using community level strategies to reduce asthma attacks triggered by outdoor air pollution: a case crossover analysis.
      This analysis was conducted using case-crossover with conditional logistic regression and compared high concentration day bins with reference day bins. The high concentration bin limit was initiated at the median, and the concentration was increased by 5% until the risk above baseline was 15%. If the increase in risk above baseline did not reach 15%, the pollutant and lag were dropped from further analysis.
      In the third step, the high concentration bins for pollutants with an increased risk above baseline of ≥15% were further evaluated to quantify whether the RR associated with the bin was also ≥15% using case-crossover with conditional logistic regression. The pollutant and lag were dropped from the analysis if the risk within the upper bin was <15%.
      The fourth step of the methodology to identify supplemental air pollution warning days was to evaluate the following: (1) if the supplemental warning days differ from EPA AQI warning days and (2) if the percentage of the time warnings that occur is sufficiently low to avoid over-warning. This evaluation is important to strike a balance between providing extra information to Houstonians without desensitizing the public by over-warning.
      In the fourth step, potential warning days were identified between 2010 and 2015. These days were compared with days when EPA AQI warnings were issued as “unhealthy for sensitive groups” or more severe to determine the extent of overlap. Supplemental warning days for the same period were then evaluated by month and year to assess the potential for over-warning. Supplemental warning days are those that do not coincide with EPA AQI warning days and whose frequency of occurrence does not exceed 20%. The AQI may differ from the proposed warnings because it does not address local pollutant mixtures or exposures longer than 1 day.

      Statistical Analysis

      Statistical analysis was conducted in SAS, version 9.4; R, version 3.5.1; and RStudio, version 1.1.456 (2009–2018). Rice University IRB approved all data-collecting procedures for human subjects.

      RESULTS

      There were 18,587 ambulance-treated asthma attacks in Houston during the study period (2004–2016) at an annual average for the years 2004–2015 of 1,491 asthma attacks (SD=174 asthma attacks). Of these calls, 24% of patients were aged ≤18 years, 19% were aged ≥60 years, and 54% were female. The ethnic breakdown was 70% African American, 14% Caucasian, 13% Hispanic, and 3% other. In comparison, the ethnic breakdown for the general population of Houston is 20% African American, 61% Caucasian, 39% Hispanic, and 19% other.

      Social Explorer. American Community Survey Tables: 2012 – 2016. www.socialexplorer.com/data/ACS2016_5yr/metadata/?ds=ACS16_5yr. Accessed March 28, 2019.

      Table 1 shows the apparent temperature and the means, SDs, and percentiles of each pollutant. Correlations between daily pollutant concentrations indicated that NO2 and CO were the most highly correlated of the pollutants (Pearson correlation, r=0.74), followed by NO2 and SO2 (r=0.53) and SO2 and CO (r=0.41), each with p-values <0.0001. Correlation coefficients between other pollutant pairs were <0.25 in magnitude.
      Table 1Statistics of Daily Pollution and Apparent Temperature, January 4, 2004 to June 15, 2016
      Percentile
      VariableMonitors% missing
      % Missing: percentage of time when hourly data were unavailable at any monitor. CO, carbon monoxide; NO2, nitrogen dioxide; O3, ozone; PM2.5, particulate matter ≤2.5 microns; ppb, parts per billion; SO2, sulfur dioxide.
      Mean (SD)5%25%50%75%95%IQR
      CO daily mean (ppb)80.111,017.5 (5.3)1,009.51,014.11,016.91,020.4102.36.2
      NO2 daily average (ppb)100.0912.7 (6.0)4.88.111.716.523.98.4
      PM2.5 daily mean (µg/m3)40.1111.8 (4.9)5.58.410.914.321.45.9
      SO2 daily mean (ppb)70.001.3 (1.5)−0.30.20.82.14.01.9
      O3 8-hour max (ppb)130.0044.6 (18.1)20.631.641.354.379.022.6
      Apparent temperature (°F)150.0973.4 (17.5)42.259.776.088.796.029.0
      a % Missing: percentage of time when hourly data were unavailable at any monitor.CO, carbon monoxide; NO2, nitrogen dioxide; O3, ozone; PM2.5, particulate matter ≤2.5 microns; ppb, parts per billion; SO2, sulfur dioxide.
      The results of the first three steps of the analysis to identify the supplemental warning days are summarized in Table 2. SO2 was not significant and was excluded from further analyses.
      Table 2Identification of Supplemental Warning Days Using EMS-Treated Asthma Attack Data
      Pollutant and lag
      Daily pollutant average for CO, NO2, and PM2.5; 8-hour maximum for O3. CO, carbon monoxide; EMS, emergency medical services; NO2, nitrogen dioxide; O3, ozone; PM2.5, particulate matter ≤2.5 microns; ppb, parts per billion.
      Step 1
      Step 1: identify pollutants /lags posing a significant risk for EMS-treated asthma attacks.
      : Significant RR of increased asthma attack
      Step 2
      Step 2: identify concentration levels of pollutants/lags in Step 1 associated with 15% above baseline.
      : Concentration limit at 15% above baseline,
      Baseline is defined as days with low to no risk of pollution triggered asthma attacks: baseline=below 41.9 ppb for O3 0–2 and below 11.7 ppb for NO2 lag0; pollutants/lags marked as “–” are <15% above baseline.
      concentration (ppb)
      Step 3
      Step 3: identify RR associated above concentration limit in Step 2 posing a risk of 15% or greater.
      : RR above concentration limit
      RR above concentration limit exceeds 15%; days above this limit are potential warning days.
      RR per IQRnRRn
      CO lag 01.03 (1.01, 1.05)18,520504.81.01 (1.01, 1.01)1,047
      CO lag 11.02 (1.00, 1.04)18,511417.31.01 (1.01, 1.01)2,005
      CO lag 0–21.04 (1.01, 1.06)18,520
      CO 1–21.02 (1.00, 1.05)18,517403.41.01 (1.01, 1.01)7,962
      CO 1–31.03 (1.00, 1.05)18,517
      NO2 01.06 (1.03, 1.09)18,52419.31.30 (1.29, 1.32)
      RR above concentration limit exceeds 15%; days above this limit are potential warning days.
      3,167
      NO2 lag 11.04 (1.01, 1.07)18,515
      NO2 lag 0–21.05 (1.02, 1.09)18,524
      NO2 lag 1–21.03 (1.00, 1.06)18,520
      PM2.5 lag 21.02 (1.00, 1.05)18,479
      PM2.5 lag 31.03 (1.01, 1.05)18,479
      PM2.5 lag 1–31.03 (1.00, 1.06)18,491
      O3 lag 01.03 (1.01, 1.06)18,52458.81.08 (1.08, 1.09)3,616
      O3 lag 11.03 (1.01, 1.06)18,524
      O3 lag 21.03 (1.00, 1.05)18,52479.01.12 (1.11, 1.13)3,485
      O3 lag 0–21.05 (1.02, 1.08)18,524
      O3 lag 1–21.04 (1.01, 1.07)18,524
      O3 lag 1–31.04 (1.01, 1.07)18,52466.61.15 (1.14, 1.16)
      RR above concentration limit exceeds 15%; days above this limit are potential warning days.
      1,808
      Note: EMS data: January 4, 2004 through June 15, 2016.
      a Step 1: identify pollutants /lags posing a significant risk for EMS-treated asthma attacks.
      b Step 2: identify concentration levels of pollutants/lags in Step 1 associated with 15% above baseline.
      c Step 3: identify RR associated above concentration limit in Step 2 posing a risk of 15% or greater.
      d Baseline is defined as days with low to no risk of pollution triggered asthma attacks: baseline=below 41.9 ppb for O3 0–2 and below 11.7 ppb for NO2 lag0; pollutants/lags marked as “–” are <15% above baseline.
      e RR above concentration limit exceeds 15%; days above this limit are potential warning days.
      f Daily pollutant average for CO, NO2, and PM2.5; 8-hour maximum for O3.CO, carbon monoxide; EMS, emergency medical services; NO2, nitrogen dioxide; O3, ozone; PM2.5, particulate matter ≤2.5 microns; ppb, parts per billion.
      The significant pollutants and lags were then evaluated to identify the concentration level associated with a risk of ≥15% above a reference group of baseline risk (Step 2). The reference group of baseline risk was defined as days when the concentration is below the median of same-day NO2 (NO2 lag 0) and O3 from same-day through the previous 2 days (O3 lag 0–2). These thresholds demonstrated no association between air pollution and asthma attacks. The median for same-day NO2 was 11.7 parts per billion (ppb) and the median for O3 lag 0–2 was 41.9 ppb. Pollutants and lags that posed a 15% increased significant risk above baseline (Table 2) were CO lag 0 (504.8 ppb), lag 1 (417.3 ppb), and lag 1–2 (403.4 ppb); NO2 lag 0 (19.3 ppb); and O3 lag 0 (58.8 ppb), lag 2 (79.0 ppb), and lag 1–3 (66.6 ppb).
      Only the concentration limits associated with same-day NO2 and O3 from the previous 3 days (O3 lag 1–3) exceeded 15% risk over baseline. At a concentration of 19.3 ppb (85th percentile), same-day NO2 posed an increased risk of asthma attacks of 30% (RR=1.30, 95% CI=1.29, 1.32), and at an average concentration of 66.6 ppb (95th percentile), O3 from the previous 3 days posed an increased risk of asthma attacks of 15% (RR=1.15, 95% CI=1.14, 1.16). These two concentration cut offs serve as the basis for the proposed supplemental warning day concentration limits in Houston.
      The retrospective analysis conducted as part of Step 4 indicated supplemental warning days that would have occurred, based on the two concentration limits described above, resulting in warnings 16% of days between 2010 and 2015. Comparatively, EPA's AQI warning was issued on 9% of days in the same period. There was a 4% overlap with the proposed supplementary warning days (Figure 1) and EPA's AQI warning. In other words, supplementary warnings would have added an additional 12% warning frequency to the current AQI warning system. The retrospective analysis indicated a range of supplemental warning occurrence between 6% and 24% by year during 2010 and 2015 (Figure 2). Monthly analysis of 2010 through 2015 indicated warnings would have occurred between a range of 5% and 29% of the month. Warnings triggered by NO2 occurred most commonly in the cooler months between October and March, whereas warnings triggered by O3 occurred most commonly in the warmer months between April and September. In Figure 2, the black segment indicates the percentage of days when both O3 and NO2 would have triggered a warning on the same day. The maximum occurrence of warning by both pollutants occurred in October at a frequency of 3.2%. Because air quality improved over the entire study period, only the second half was used to evaluate potential warning frequency.
      Figure 1
      Figure 1Venn diagram of retrospective overlap between EPA AQI warning days and proposed supplemental warning days.
      Note: n=2,191 days (2010 to 2015)
      aPublished EPA Air Quality Index warning of Unhealthy or greater.
      bProposed supplemental warning days.
      AQI, Air Quality Index; EPA, Environmental Protection Agency.
      Figure 2
      Figure 2Stacked bar charts of percent of supplemental warnings by pollutant between 2010 and 2015.
      Note: When ozone is >66.6 ppb, NO2 is >19.3 ppb or AQI is >100.
      AQI, Air Quality Index; NO2, nitrogen dioxide; ppb, parts per billion.

      DISCUSSION

      Asthma attacks are a serious concern in Houston. There are approximately 1,500 ambulance-treated attacks a year in this metropolitan area, for a medical emergency that is almost always preventable. The City of Houston was spurred to action by previous research showing an association between air pollution and asthma attacks in Houston. A framework was developed to identify days when air pollution concentrations historically posed high risk for asthma attacks in the city, and the Houston Health Department began warning the community on days when concentrations may exceed these levels. These warnings are intended to supplement the EPA AQI warnings by providing additional information to the community to enhance asthma self-management.
      The warning days identified by this methodology are days when, historically, NO2 or lagged elevated O3 concentrations posed a statistically significant increased risk of an ambulance-treated asthma attack for Houstonians. These pollutants may not be the only trigger of the attacks but may serve as a proxy for the conditions that create the increased risk. The criteria pollutants analyzed in the study are themselves correlated, and therefore, their distinct impact cannot be distinguished (e.g., daily average NO2 and CO are correlated, r=0.73). In addition, Houston's complex mixture of hazardous air pollutants emitted from the large industrial complex, including one of the largest ports in the U.S., likely plays a role in the high-risk days.
      • Keller SA
      • Koonin SE
      • Shipp S
      Big data and city living – what can it do for us?.

      Toxics Release Inventory Program. TRI Explorer Release Chemical Report 2016. Harris County. On and Offsite Emissions. https://iaspub.epa.gov/triexplorer/tri_release.chemical. Accessed March 28, 2019.

      Finally, the role of aeroallergens in the increased risk is unknown because data are not available on a consistent enough basis to be included in the framework models.
      • Raun LH
      • Ensor KB
      • Persse D
      Using community level strategies to reduce asthma attacks triggered by outdoor air pollution: a case crossover analysis.
      However, seasonal and meteorological conditions that cause pollutant mixtures to be elevated may be correlated with days with elevated aeroallergens in Houston, therefore, also represented by proxy.
      • Raun LH
      • Ensor KB
      • Persse D
      Using community level strategies to reduce asthma attacks triggered by outdoor air pollution: a case crossover analysis.
      The intent of the framework is to identify days posing a high risk rather than pinpoint the exact causes of the risk.
      The identified warning days indicate risk of about 30% when the daily average NO2 is expected to exceed 19.3 ppb. The EPA AQI warnings for NO2 are for daily 1-hour maximum NO2 and currently do not accommodate the daily average statistical metric. The daily maximum and the daily average are highly correlated, but prediction of the maximum is much more difficult for use in warnings (r=0.898, p<0.0001). Retrospective comparison indicates the EPA AQI warning for daily maximum occurred 9% of the time in the second half of the study period, whereas a supplemental warning day for Houston would have occurred 12% of the time not covered by the AQI.
      The supplemental warning days identified for Houston based on increased risk of consecutive levels of O3 are also not accommodated in the EPA AQI at this time. The EPA AQI warning is issued based on the daily 8-hour maximum O3 concentration, whereas this research indicates Houstonians also have an increased risk for exposure prolonged over more than 1 day of about 15%. A warning to Houstonians is issued on a day when the previous 3 days’ average 8-hour maximum concentration exceeds 66.6 ppb. The issued warnings remind people with asthma to be self-aware and prepared on these days that have historically posed a problem. The warning, posted on the Houston Health Department webpage at 5:00am on warning days, is, “Asthma Air Aware Day Warning. The Houston Health Department is notifying you that today is a Houston Asthma Air Aware Day. Houston Asthma Air Aware Days occur when outdoor air conditions are similar to days when Houstonians historically experience a high number of asthma problems. Houstonians with asthma should be mindful of today's potential risk.”
      The concentration limits identified in this study show risk on days when AQI warnings were not issued, adding an average of 46 warning days per year between 2010 and 2015 (277 days in 6 years); these additional 12% of days warned would be particularly meaningful for people with asthma.
      Careful attention was paid to the frequency of warning. The frequency was chosen as a balance between increasing the number of warnings to provide extra information compared with providing too many warnings and having the public fatigued by over-warning. The methodology chosen to search for the warning day concentration level in Step 2 represents the best practices identified during in-depth exploratory analysis. Use of classification and regression tree analysis was explored as a method to define natural breaks for the upper concentration risk bins as an alternative to Step 2. This method was abandoned because it yielded low concentration limits resulting in over-warning (e.g., warnings would have occurred 43% of days between 2010 and 2015). Although the 6-year average between 2010 and 2015 sits at 16% warning, the trend from 2010 through 2014 was generally decreasing. With further analysis of the proposed warning concentration limits, it is possible that the warning frequency would decrease even further.

      Limitations

      A limitation of this study is the use of ambulance-treated asthma exacerbations over emergency department visits. The study uses ambulance data because they are maintained by the city, and records are readily available on a real-time basis for ongoing analysis. There is no central location for data pertaining to emergency department visits for asthma attacks. The use of ambulance data likely underestimates the number of asthma exacerbations associated with pollution. Another limitation of this study is that aeroallergen data were not complete enough to include in the analysis. However, the intent is to identify days that historically pose an increased health risk to Houstonians, rather than pinpoint the trigger. Risk posed by aeroallergens may be represented by proxy through correlated pollutants and meteorological conditions. Analysis with aeroallergen data will be included when and if sufficient data are available.
      Another limitation is warning at 15% risk above baseline or on average 46 additional days a year. There is limited research published on the awareness of and compliance with air pollution advisories, and none, to the authors’ knowledge, of how much warning of poor air quality is too much. Future work will include follow-up with the community regarding usefulness of the warnings, at which time the 15% above baseline limit will be adjusted, as needed.

      CONCLUSIONS

      This study provides a city-specific approach to examine exposures that may pose a risk for people with asthma. Houston has developed messaging and a warning system that can be used to make personal health decisions based on local, data-driven risk factors. This study furthers the understanding of how local data on city-specific air quality can help better target air quality warnings for sensitive subpopulations and provides a framework that the Houston Health Department uses to supplement EPA AQI warnings.

      ACKNOWLEDGMENTS

      The authors would like to thank the reviewers for their careful readings of this manuscript. Their attention to detail and thoughtful comments greatly improved the clarity of this work and the strengthening of this method.
      This work was unfunded.
      Author responsibilities were as follows: study concept and design (LR); interpretation of the data (LR, KE); drafting of the manuscript (LR); statistical expertise (KE); data analysis (JP, LC); editing of the manuscript (JP); creating figures and tables (LC); acquisition of the data (DP); and critical revision of the manuscript (DP).
      No financial disclosures were reported by the authors of this paper.

      Appendix. Supplementary materials

      REFERENCES

      1. U.S. Environmental Protection Agency. Health risk and exposure assessment for ozone. https://nepis.epa.gov/Exe/ZyPDF.cgi/P100KBUF.PDF?Dockey=P100KBUF.PDF. Published August 2014. Accessed October 1, 2018.

      2. U.S. Environmental Protection Agency. Integrated science assessment for carbon monoxide. https://ofmpub.epa.gov/eims/eimscomm.getfile?p_download_id=494432. Published January 2010. Accessed October 1, 2018.

      3. U.S. Environmental Protection Agency. Integrated science assessment for oxides of nitrogen – health criteria. http://ofmpub.epa.gov/eims/eimscomm.getfile?p_download_id=526855. Published January 2016. Accessed October 1, 2018.

      4. U.S. Environmental Protection Agency. Integrated science assessment for sulfur oxides – health criteria. http://ofmpub.epa.gov/eims/eimscomm.getfile?p_download_id=533653. Published December 2017. Accessed October 1, 2018.

      5. U.S. Environmental Protection Agency. Integrated science assessment for particulate matter. http://ofmpub.epa.gov/eims/eimscomm.getfile?p_download_id=494959. Published December 2009. Accessed October 1, 2018.

        • Alhanti BA
        • Chang HH
        • Winquist A
        • Mulholland JA
        • Darrow LA
        • Sarnat SE
        Ambient air pollution and emergency department visits for asthma: a multi-city assessment of effect modification by age.
        J Expo Sci Environ Epidemiol. 2016; 26: 180-188
        • Gasana J
        • Dillikar D
        • Mendy A
        • Forno E
        • Ramos Vieira E
        Motor vehicle air pollution and asthma in children: a meta-analysis.
        Environ Res. 2012; 117: 36-45
        • Li J
        • Ewart G
        • Kraft M
        • Finn PW
        The public health benefits of air pollution control.
        J Allergy Clin Immunol. 2012; 130: 22-23
        • Penard-Morand C
        • Raherison C
        • Charpin D
        • et al.
        Long-term exposure to close-proximity air pollution and asthma and allergies in urban children.
        Eur Respir J. 2012; 36: 33-40
        • Raun LH
        • Jefferson LS
        • Persse D
        • Ensor KB
        Geospatial analysis for targeting out-of-hospital cardiac arrest intervention.
        Am J Prev Med. 2013; 45: 137-142
        • Raun LH
        • Ensor KB
        • Persse D
        Using community level strategies to reduce asthma attacks triggered by outdoor air pollution: a case crossover analysis.
        Environ Health. 2014; 13: 58
        • Silverman RA
        • Ito K
        • Freese J
        • et al.
        Association of ambient fine particles with out-of-hospital cardiac arrests in New York City.
        Am J Epidemiol. 2010; 172: 917-923
        • Wendt JK
        • Symanski E
        • Stock TH
        • Chan W
        • Du XL
        Association of short-term increases in ambient air pollution and timing of initial asthma diagnosis among Medicaid-enrolled children in a metropolitan area.
        Environ Res. 2014; 131: 50-58
        • Bell ML
        • HEI Health Review Committee
        Assessment of the health impacts of particulate matter characteristics.
        Res Rep Health Eff Inst. 2012; 161: 5-38
        • Domínguez-López D
        • Adame JA
        • Hernández-Ceballos MA
        • Vaca F
        • De la Morena BA
        • Bolívar JP
        Spatial and temporal variation of surface ozone, NO and NO at urban, suburban, rural and industrial sites in the southwest of the Iberian Peninsula.
        Environ Monit Assess. 2014; 186: 5337-5351
        • Keeler GJ
        • Morishita M
        • Wagner JG
        • Harkema JR
        Characterization of urban atmospheres during inhalation exposure studies in Detroit and Grand Rapids, Michigan.
        Toxicol Pathol. 2007; 35: 15-22
        • Lobdell DT
        • Isakov V
        • Baxter L
        • Touma JS
        • Smuts MB
        • Özkaynak H
        Feasibility of assessing public health impacts of air pollution reduction programs on a local scale: New Haven case study.
        Environ Health Perspect. 2011; 119: 487-493
        • Tang G
        • Wang Y
        • Li X
        • Ji D
        • Hsu S
        • Gao X
        Spatial-temporal variations in surface ozone in Northern China as observed during 2009–2010 and possible implications for future air quality control strategies.
        Atmos Chem Phys. 2012; 12: 2757-2776
      6. AirNow. Air quality index (AQI) basics. https://airnow.gov/index.cfm?action=aqibasics.aqi. Published 2016. Accessed March 24, 2017.

        • Keller SA
        • Koonin SE
        • Shipp S
        Big data and city living – what can it do for us?.
        Signif (Oxf). 2012; 9: 4-7
      7. Utah Department of Health. Air quality recess guidance for schools. www.health.utah.gov/asthma/air%20quality/Recess%20Guidance.html. Published 2013. Accessed October 3, 2013.

        • Dimitrova R
        • Lurponglukana N
        • Fernando HJS
        • et al.
        Relationship between particulate matter and childhood asthma – basis of a future warning system for central Phoenix.
        Atmos Chem Phys. 2012; 12: 2479-2490
        • Litchfield IJ
        • Ayres JG
        • Jaakkola JJ
        • Mohammed NI
        Is ambient air pollution associated with onset of sudden infant death syndrome: a case-crossover study in the UK.
        BMJ Open. 2018; 8e018341
        • Raza A
        • Dahlquist M
        • Lind T
        • Ljungman PL
        Susceptibility to short-term ozone exposure and cardiovascular and respiratory mortality by previous hospitalizations.
        Environ Health. 2018; 17: 37
        • Tsai SS
        • Tsai CY
        • Yang CY
        Fine particulate air pollution associated with increased risk of hospital admissions for hypertension in a tropical city, Kaohsiung, Taiwan.
        J Toxicol Environ Health A. 2018; 81: 567-575
        • Ensor KB
        • Raun LH
        • Persse D
        A case-crossover analysis of out-of-hospital cardiac arrest and air pollution.
        Circulation. 2013; 127: 1192-1199
      8. Texas Commission on Environmental Quality. Welcome to the Texas Air Monitoring Information System (TAMIS) Web Interface.www17.tceq.texas.gov/tamis/index.cfm?fuseaction=home.welcome. 2014. Accessed March 27, 2018.

      9. Social Explorer. American Community Survey Tables: 2012 – 2016. www.socialexplorer.com/data/ACS2016_5yr/metadata/?ds=ACS16_5yr. Accessed March 28, 2019.

      10. Toxics Release Inventory Program. TRI Explorer Release Chemical Report 2016. Harris County. On and Offsite Emissions. https://iaspub.epa.gov/triexplorer/tri_release.chemical. Accessed March 28, 2019.