Investigation of the Environmental and Socio-Economic Characteristics of Counties with a High Asthma Burden to Focus Asthma Action in Utah

Rising adult asthma prevalence (AAP) rates and asthma emergency room (AER) visits constitute a large burden on public health in Utah (UT), a high-altitude state in the Great Basin Desert, USA. This warrants an investigation of the characteristics of the counties with the highest asthma burden within UT to improve allocation of health resources and for planning. The relations between several predictor environmental, health behavior and socio-economic variables and two health outcome variables, AAP and AER visits, were investigated for UT’s 29 counties. Non-parametric statistical comparison tests, correlation and linear regression analysis were used to determine the factors significantly associated with AER visits and AAP. Regression kriging with Utah small area data (USAD) as well as socio-economic and pollution data enabled local Moran’s I cluster analysis and the investigation of moving correlations between health outcomes and risk factors. Results showed the importance of desert/mining dust and socio-economic status as AAP and AER visits were greatest in the south of the state, highlighting a marked north–south divide in terms of these factors within the state. USAD investigations also showed marked differences in pollution and socio-economic status associated with AAP within the most populous northern counties. Policies and interventions need to address socio-economic inequalities within counties and between the north and south of the state. Fine (PM2.5) and coarse (PM10) particulate matter monitors should be installed in towns in central and southern UT to monitor air quality as these are sparse, but in the summer, air quality can be worse here. Further research into spatiotemporal variation in air quality within UT is needed to inform public health interventions such as expanding clean fuel programs and targeted land-use policies. Efforts are also needed to examine barriers to routine asthma care.


USGS UT WHO
United States Geological Survey Utah World Health Organization 2. Methods 2.1. Questions used from BRFSS to determine AAP, smoking and obesity AAP, smoking and obesity rates were determined from the BRFSS using the following questions: 1)"Have you ever been told by a doctor {nurse or other health professional} that you have asthma?" Current asthma is defined as an affirmative response to that question followed by an affirmative response to the subsequent question "Do you still have asthma?" [65] and 2) "Do you now smoke cigarettes every day, some days, or not at all?" [66], with individuals reporting any smoking being considered as smokers. Obesity was determined from BRFSS questions about height and weight and calculated based on a BMI of 30-99.8 [66].

CL PM2.5 Data limitations
The following statement of data limitations comes with the CL PM2.5 data: "measures estimate average annual concentration of fine PM pollution in the county, and can miss "important short-term fluctuations in air quality (such as stagnation events), local patterns (high concentrations near roads and other major sources), and other pollutants (such as ozone, etc.). Further, these estimates are based on seasonal averages. Even within counties with low average fine PM concentrations, locations can experience days of dangerously elevated levels. It should be noted that these data are derived from only one air quality model among several. Like all models, this air quality model has errors. There is also a large time lag (up to 5 years) between when these data are collected and when the modeled results become available." https://www.countyhealthrankings.org/app/UT/2019/measure/factors/125/description

Investigating USAD and census tract data through regression kriging
As data at the USAD level (n=99) and census tract level (n=588) were available for the HII [55] ( Figure 5b) and CT PM2.5, respectively, it was decided to investigate whether these data could be used to gain further insight into AAP and AER patterns. The HII and CT PM2.5 data were aggregated to the county-level by averaging the values for the five nearest neighbor USAD/census tracts (Figure 5a-b not shown for CT PM2.5) to allow correlation analysis with AAP and AER data at the county-level.
The correlations between HII and AAP and AER visits at the county-level were 0.49 and 0.13, respectively. Given the stronger correlation with AAP than AER visits, the HII data were used to regression krige the AAP data to the USAD level (n=99). The AER visit data correlated better with the CT PM2.5 data (Summer 2011 max. r=0.323 and Winter 2011-2014 mean r=-0.347) so the modelled PM2.5 data were used to regression krige AER visits. The county-level AAP was used as the dependent variable in regression with the county-level HII data as a single independent variable, then a variogram of the regression residuals was computed and modelled. The regression residuals were ordinary kriged to the USAD level. The regression equation from the county-level was then used to predict AAP ( 0.040558 4.884658 where corresponds to HII county-level data) at the USAD level and the kriged residuals were added to these regressed values (Figure 2c). To determine the success of the regression kriging, the RK values were aggregated to the county-level using averaging of the five nearest neighbor values (Figure 2d). These values were then correlated and compared with the original county-level AAP values. The correlation coefficient was r=0.93 and the mean RMSE ( Figure 2e) was 0.35%. This and the result that follows for AER visits suggests that the combination of using regression association and spatial association in regression kriging is successful at predicting AAP and AER visits at the USAD level and census tract level. The same procedure was followed with the AER data and the CT PM2.5 data. The regression equation used was =0.3965 0 -1.4886 1 + 27.915 where 0 and 1 correspond to Summer2011Max and Winter2011to2014Mean, respectively.
Once the RK AER data (Figure 2f) were aggregated to the county-level (Figure 2g), the correlation with the original county-level AER data was r=0.876 and the mean RMSE (Figure 2h) was 2.197.