Introduction

Cardiovascular disease (CVD) is a leading cause of death both in California and the United States.The environment problems and source of pollutions have been shown to increase the risk of CVD rate. PM2.5 and other air pollutants such as ozone, nitrogen dioxide, and sulfur dioxide have been considered in the development of health-based standards. Moreover, people living in communities that were identified as “disadvantage” by California Environmental Protection Agency (CalEPA) are more vulnerable to the effects of pollution than others.

The main purpose of this project is to evaluate the association between pollution burden and CVD rate while accounting for community’s vulnerability to this association in all California counties.

Methods

Data source

The data set was downloaded from (https://data.ca.gov/dataset/calenviroscreen-3-0-results) through API. Once downloaded, the desired information was extracted and formed into a data table. This is a data set including environmental, health, and socioeconomic information of all communities in California State. The key independent variables that were examined in this study were PM2.5, ozone, traffic density, cardiovascular disease rate and community category.

Examine variables

  1. PM2.5 PM2.5 exposure in this data set was an annual mean concentration(ug/m3). After assessing the linearity of PM2.5 and CVD rate, the association was roughly linear.
  2. Ozone Ozone exposure, which is the mean of summer month (May-October) of the daily maximum 8 hours concentration (ppm). The relationship between ozone exposure and CVD rate was linear.
  3. Traffic density Traffic exposure(vehicles-km/hr/km) is represented as the number of vehicles (adjusted by road segment lengths in kilometers) per hour per kilometer of roadways.Traffic variable was highly skewed. In order to easily interpret, the traffic density was cut into four levels, and the likelihood test result suggested that keep traffic density variable categorical was better.
  4. The events of cardio_disease Cardiovascular disease rate is a age-adjusted rate of emergency department(ED) visits for AMI per 10,000 residents. The events of cardio-disease were calculated through multiplying the rate of ED visit per 10,000 residents by the population in that community. Two missing data from population was excluded.
  5. Community The communities were divided into two categories: disadvantaged and non-disadvantaged. California Environmental Protection Agency (CalEPA) identify the communities which are more vulnerable to the effects of pollution than others as disadvantaged communities.

Statistical analysis

An modified Poisson regression model was used to estimate the association of pollutants and cardiovascular disease rate in all communities of California State. Analyses were performed by adjusting for community category confounding factor. Further model fit assessment was tested by Person chi-square test. The overdispersion of the model was also tested to determine if the variance was larger than what would be expected under a Poisson distribution. Negative binomial regression model was used to address overdispersion issue. P-values< 0.05 were considered statistically significant.


Results

The relationship between pollution particles and CVD rate

Figure 1: The correlation between cardiovascular disease (CVD) rate and pollution particle: PM2.5 and ozone, by county.

PM2.5 and ozone, are main particles considered in the study. Pollution score is an averaged pollution burden of environmental effects for each county. It is calculated as the average percentiles,the percentile represents a relative score for the indicator, of the environmental effects indicators. The county with higher score therefore has relatively high pollution burdens. In Figure 1, PM2.5 and ozone are positively related with CVD rate. The pollution score plot is also showing a positive correlation with CVD rate.

Figure 2: State map depict the pollution score and CVD rate data of each county in California.

Counties with high pollution score is consistent with higher CVD rate. For example, Madera, Fresno, Los Angeles, San Joaquin, San Bernadino and Riverside counties have higher pollution score as well as a high CVD rate( Figure 2 ). However, there are some inconsistency. For instance, the northern counties of state shows lower pollution burden but have relatively higher CVD rate. This suggests that other life characteristics and factors may contribute to their high CVD rate.


Pollution burden and community category

County Pollution Score Disadvantaged Community
Fresno 7.483057 0.5979899
Stanislaus 7.286043 0.6489362
Tulare 7.245979 0.4487179
San Bernardino 7.217753 0.4254743
Kern 7.182617 0.4503311
Kings 6.997034 0.4074074
Los Angeles 6.976374 0.4430218
Merced 6.913608 0.7551020
San Joaquin 6.498406 0.5107914
Riverside 6.470632 0.2229581

Figure 3: Figure 3 lists top 10 counties with the higher pollution score. Fresno county has the highest pollution burden and is approximately 60% disadvantaged communities in this county. A high pollution burden is not exactly positive related with high proportion of disadvantaged communities of the county.


Figure 3 : CVD rate by community category.

Among the top 10 cities with high pollution burden, 90% of them are disadvantaged communities. CalEPA classifies communities as “disadvantaged” due to the environmental conditions and vulnerability of people living in those communities. Figure3 and table1 both indicate that living in an disadvantaged community increase the risk of CVD rate.


CVD rate with pollution burden by community category

Figure 4 : Cardiovascular rate vs. pollutant burden by community category.

Figure 4 shows the association between CVD rate and pollution burden by community category. The average pollution burden in disadvantaged communities is obviously higher than non-disadvantage communities. However, the relationship between CVD rate and pollution burden is not as obvious as in Figure 1.


Traffic density level VS PM2.5 and Ozone

Traffic density level (vehicles-km/hr/km) PM2.5 (ug/m3) Ozone (ppm)
[22.4,442] 9.975939 0.0489529
(442,700] 10.304066 0.0469590
(700,1.19e+03] 10.572666 0.0471831
(1.19e+03,4.57e+04] 10.618026 0.0465895

Table 3: Traffic density vs PM2.5 and Ozone

The lowest traffic density level was 22.4-442 (vehicles-km/hr/km).The mean PM2.5 and ozone value associated with the lowest traffic density were 9.98(ug/m3) and 0.049(ppm), respectively. The highest traffic density level was associated the highest PM2.5 value 10.62(ug/m3). Thus, increased PM2.5 was associated with high traffic density level.


Adjusted association between CVD rate and all variables

  Unadjusted Model Negative binomial Model
Predictors Incidence Rate Ratios Conf. Int (95%) P-Value Incidence Rate Ratios Conf. Int (95%) P-Value
(Intercept) 1.53 1.42 – 1.66 <0.001 1.74 1.60 – 1.89 <0.001
community 1.35 1.31 – 1.39 <0.001
ozone_ctr 2.27 2.15 – 2.40 <0.001 2.21 2.08 – 2.35 <0.001
PM2.5 1.01 1.01 – 1.02 <0.001 1.00 0.99 – 1.00 0.463
PM2.5:traffic.q4(1.19e+03,4.57e+04] 0.96 0.94 – 0.97 <0.001 0.93 0.92 – 0.95 <0.001
PM2.5:traffic.q4(442,700] 0.99 0.98 – 1.00 0.173 0.99 0.98 – 1.00 0.043
PM2.5:traffic.q4(700,1.19e+03] 0.98 0.97 – 1.00 0.008 0.97 0.96 – 0.98 <0.001
traffic.q4: [22.4,442] Reference Reference
traffic.q4: (442,700] 1.07 0.96 – 1.19 0.214 1.11 0.99 – 1.24 0.080
traffic.q4:
(700,1.19e+03] 		1.16 1.02 – 1.32 0.025 1.30 1.13 – 1.49 <0.001
traffic.q4:
(1.19e+03,4.57e+04] 		1.47 1.24 – 1.75 <0.001 1.84 1.53 – 2.21 <0.001
Observations 7951 7951
R2 Nagelkerke 0.184 0.239

Table 5: Adjusted association between the pollution and CVD rate.

After checking overdispersion, there was an 1.14 (P<0.001) overdispersion term in the Poisson model. A negative binomial regression model was performed to address overdispersion. In the final negative binomial model, PM2.5 was not statistically associated with CVD rate(p=0.463) compared to unadjusted model(p<0.001), and the exposure had the similar CVD rate as non-exposure(95% CI 0.99,1.00;p=0.463). Every one-unit ozone exposure is associated with 2.21 times the CVD rate(95% CI 2.08,2.35;p<0.001), while 2.27 times before adjusting for community. The highest traffic density exposure was associated with 1.84 times CVD rate (95% CI 1.53,2.21;p<0.001) which increased 25% compared to unadjustment model (95% CI 1.24,1.75;p<0.001). Moreover, there is a significant interaction between PM2.5 and traffic density,and increased PM2.5 was associated with high traffic density level( table3 ).


Conclusion

Pollutants such PM2.5, Ozone and traffic were statistically significantly associated with CVD rate and living in an disadvantaged community increases risk of CVD rate. The baseline of CVD rate after adjustment of community category increased 13%. The CVD rate in highest traffic density level increased 25% compared to unadjusted model. Although the CVD rate associated with PM2.5, and ozone exposure did not change much after adjustment, the people with ozone exposure still had 2.21 times the rate of CVD than the baseline. PM2.5 had a significant interaction with traffic density level. Thus, improvements in air quality would be helpful to reduce overall CVD prevalence across California. In addition, efforts to identify the pollution source that accounting for a community’s vulnerability would beneficial to those disadvantaged communities which suffered most from CVD occurrence.