Grantee Research Project Results

2014 Progress Report: Hypertension in Mexican-Americans: Assessing Disparities in Air Pollutant Risks

EPA Grant Number: R834581
Title: Hypertension in Mexican-Americans: Assessing Disparities in Air Pollutant Risks
Investigators: Symanski, Elaine , Piller, Linda B. , Lopez, David S , Chen, Lin-An , Jimenez, Maria , Strom, Sara , Chan, Wenyaw
Current Investigators: Symanski, Elaine , Bondy, Melissa L. , Chen, Lin-An , Jimenez, Maria , Strom, Sara , Chan, Wenyaw
Institution: The University of Texas School of Public Health , The University of Texas MD Anderson Cancer Center , National Chiao-Tung University
Current Institution: The University of Texas School of Public Health , National Chiao-Tung University , The University of Texas MD Anderson Cancer Center
EPA Project Officer: Hahn, Intaek
Project Period: August 1, 2010 through July 31, 2014 (Extended to July 31, 2016)
Project Period Covered by this Report: August 1, 2013 through July 31,2014
Project Amount: $1,250,000
RFA: Understanding the Role of Nonchemical Stressors and Developing Analytic Methods for Cumulative Risk Assessments (2009) RFA Text | Recipients Lists
Research Category: Human Health

Objective:

We are assessing the hypothesis that individual- and neighborhood-level psychosocial stressors exacerbate risks for hypertension associated with fine particulates and other air pollutants among individuals of Mexican origin living in Houston, Texas, who also are participating in the University of Texas M.D. Anderson Cancer Center's Mano a Mano cohort study. Several studies have shown an association between air pollutants and hypertension via oxidative stress and inflammatory pathways. Yet, little is known about the modifying effects of nonchemical stressors on air pollutant risks for hypertension. The incidence of hypertension, a key risk factor for cardiovascular diseases (CVD), has been growing in the United States, and CVD remains the leading cause of death among the U.S. Hispanic population. Novel methods for addressing interactions in a logistic regression context are being developed. We also are applying a community-based approach by involving the community in defining research questions, identifying mechanisms for becoming involved in research activities, and interpreting and disseminating research findings. This program also will help identify culturally specific elements that could be used for an effective intervention in a Latino community to address disparities in air pollutant health risks.

Aim 1: Develop new statistical methods that allow for a rigorous evaluation of interaction between chemical and nonchemical stressors in a logistic regression framework.

Aim 2: Estimate exposure to psychosocial stressors and traffic-related and industrial pollutants.

Aim 3: Examine interactions between air pollution and psychosocial stressors on the prevalence of hypertension, with a focus on quantifying the modifying effects of nonchemical stressors on air pollutant effects.

Summary/Accomplishments (Outputs/Outcomes):

We have completed interviews on 2,481 Mano a Mano cohort members who agreed to participate in our study and completed the air pollution exposure assessment.

Aim 1: Develop new statistical methods that allow for a rigorous evaluation of interaction between chemical and non-chemical stressors in a logistic regression framework.

Under a logistic regression framework, we developed an analytic approach for evaluating interactions between binary and continuous independent variables, which are discretized using a quartile breakdown. We completed an R program for implementing the proposed estimation and testing methods. We conducted a simulation study for examining the interaction effect based on this method. The results are close to our targeted error probability and the power for detecting the interaction effect between two covariates increases with the absolute value of the interaction effect. We also examined the performance of our method when there are more than two covariates with a comparison to the generalized linear model with interaction. However, we found that the target error probabilities are influenced by the discretization of the continuous covariates. We re-worked the simulation to address this issue. We have also completed the application of these methods with a subset of the data that have been collected. We are now in process of writing the manuscript for submission.

In addition to the method that we proposed in the original grant application, we have also developed a novel concept of approaching interaction by borrowing the concept of “isobole” in toxicology (Sorensen et al. (2007)) to define interaction in terms of a “statistical isobole.” We developed the estimation procedure; analytically examined some properties of the estimators; developed a test statistic for testing the interaction effect; and developed procedures to classify an isobole as having a synergistic or an antagonistic effect. Through intensive simulation, we have also calculated the power for interaction test using the statistical isobole. A manuscript describing this work was submitted for publication in the reporting period; revisions were required and the manuscript will be re-submitted by the end of the year.

Aim 2: Estimate exposure to psychosocial stressors and traffic-related and industrial pollutants.

Development of the questionnaire to assess exposures to psychosocial stressors. A paper describing the input received from the community in developing the final questionnaire regarding exposures to psychosocial stressors is in press (Symanski et al., 2015, in press). The manuscript examines the engagement of Mexican-origin neighborhood residents in sharing their perspectives on environmental exposures, psychological stress and health consequences through their participation in focus groups, the establishment of the Neighborhood Council of Advisors (NCA) and interviews for the pilot questionnaire testing. This qualitative multi-level approach allowed for the collection of data that focused questions on the actual community experiences with chemical and non-chemical stressors. In conclusion, the mixed methods approach facilitated community involvement in the development of a culturally appropriate questionnaire that has been administered (interviewing was completed during the reporting period).

Administration of the questionnaire. The questionnaire was administered to 2,481 Mano a Mano cohort members who agreed to participate in our study. Table 1 summarizes demographic and lifestyle characteristics of the study population. A summary of the data on psychosocial factors appears in Table 2.

Air pollution exposure assessment. After excluding participants (n=13) without valid geographic coordinates for residential addresses, the final sample size was N=2,468. We obtained validated air pollution data for Ozone (O₃) and particulate matter of aerodynamic diameter less than 2.5 micrometers (PM_2.5) from the Texas Commission on Environmental Quality (TCEQ). Between 2008 and 2009, there were 48 active O₃ monitoring stations and 13 active PM_2.5 monitoring stations operated by TCEQ in the Houston-Galveston-Brazoria region. We calculated the maximum daily eight-hour running average for O₃ and the mean 24-hour average for PM_2.5 for every day in the study period from January 1, 2008 to December 31, 2013. Monitoring stations with 25% or more missing observations in a calendar year were excluded. Inverse Distance interpolation (IDW) with weight (p=2) was used to calculate exposure estimates for each individual over the baseline year, using the three closest monitors to each participant’s geocoded residential address.

Aim 3: Examine interactions between air pollution and psychosocial stressors on prevalence of hypertension, with a focus on quantifying the modifying effects of nonchemical stressors on air pollutant effects.

Work on this aim did not begin during the reporting period.

Additional Analyses.

We conducted a cross-sectional analysis of polycyclic aromatic hydrocarbons and diesel particulate matter exposures and hypertension among individuals of Mexican origin from the Mano a Mano cohort. Using geographical information systems, we linked modeled annual estimates of PAHs and diesel particulate matter at the census tract level from the 2002 and 2005 U.S. Environmental Protection Agency’s National-Scale Air Toxics Assessment to baseline residential addresses of cohort members who enrolled from 2001 to 2003 or 2004 to 2006, respectively. For each enrollment period, we applied mixed-effects logistic regression models to determine associations between diesel particulate matter and PAHs, separately, and self-reported hypertension while adjusting for confounders and the clustering of observations within census tracts and households. The study population consisted of 11,218 participants of which 77% were women. The mean participant age at baseline was 41 years. Following adjustment for age, there was a dose-dependent, positive association between PAHs and hypertension (medium exposure, adjusted odds ratio (OR)=1.09, 95% CI: 0.88-1.36; high exposure, OR=1.40, 95% CI: 1.01-1.94) for individuals enrolled during 2001-2003; associations were generally similar in magnitude, but less precise, following adjustment for age, gender, smoking, and BMI. No association was detected for the later period. There was no evidence of an association between residential levels of diesel particulate matter and hypertension.

Conclusions:

Using both traditional methods and the methods developed in the project, multiple logistic regression analyses will be applied to evaluate the modifying effects of neighborhood-based and individual-level psychosocial stressors that potentially affect susceptibility to hypertension due to air pollution. To account for pollutant source, we will conduct sensitivity analyses based on our two proximity measures.

The methodology paper for the concept that was proposed in Aim 1 of the project is almost complete and is expected to be submitted within 1 month. The “isobole” paper is in its final stage of data analysis and is expected to be submitted within 2 months.

Progress Summary:

Aim 1: Develop new statistical methods that allow for a rigorous evaluation of interaction between chemical and nonchemical stressors in a logistic regression framework.

Regarding implementing the procedures we proposed in the grant application, we have completed the R program for implementing the proposed method. We assume the logistic regression model as

, where z is a binary variable and x₁ and x₂ are continuous. In the program to develop this method, we discretized each stressor x_i by partitioning its support into four categories using its lower quartile, median and upper quartile as the cutoff points. We then computed the cell probability in each cell using the formula

where

is the probability density function for X₁, X₂,..., and X_p, which follows a joint normal distribution. In this project we choose p = 2 and assume that X₁ and X₂ follow a joint normal probability density function. We further assume Z has a Bernoulli distribution. Note that we have proposed to test for the interaction index between X₁ and Z for each cell, given by the linear combination of the marginal probabilities of P (Y = 1) on each cellthat is,

where

represents the averaged marginal probabilities on the cell when X₁ = i, Z = j ,

represents the averaged marginal probabilities on the cells when X₁ = i,

represents the averaged marginal probabilities on the cells when Z = j, and

represents the averaged marginal probabilities on all cells.

In the past year, we used two approaches for combing the p-values of testing interaction within each cell. One method used the proportion of rejections among cells to determine the p-value for the overall test of interaction. Another method used Hartung's multiple comparison procedure to determine the p-value for the overall test of interaction. In the meantime, we have completed the simulation study for examining the interaction effect based on the aforementioned method. For the simulation, we considered a logistic regression model that includes two continuous covariates, one binary covariate, and possible interaction between the continuous and the binary covariates. We have considered the scenario that no interaction effect exists with various coefficients for the true model. The results for type I error are all about 5 percent, our targeted error probability, when testing for nonexistence of both interaction effects.

As we described in the previous year's report, we have borrowed the concept of isobole in toxicology to define interaction (Sorensen, et al., 2007). Recall that this novel concept defines isobole as

for any constant c. For the scenario that Y, X₁ and X₂ have a joint trivariate normal distribution, we express the isobole in a line segment that is connected by two points

and

. We also continued our endeavor of using this concept to examine the interaction. We have developed the estimation procedure and analytically examined some properties of the estimators. We have also proposed the test statistic for testing the interaction effect for an isobole IB(c). In addition, we have developed procedures to classify an isobole as having a synergistic or antagonistic effect when an interaction effect exists. Through lengthy algebraic manipulations, the asymptotic property for the proposed test statistic has been shown to follow a normal distribution. We have analytically calculated its asymptotic variance. We are developing procedures that can combine the results of each isobole test for use of testing the overall interaction effect of the entire space. Through intensive simulation, we have also calculated the power for interaction test within each isobole. The simulation results indicate that the proposed isobole method is more powerful in detecting interaction effect than the traditional model. While waiting for the project data, we also applied our method to the National Health and Nutrition Examination Survey dataset.

Aim 2: Estimate exposure to psychosocial stressors and traffic-related and industrial pollutants.

The 36-question pilot questionnaire was transformed into the 32-question final questionnaire and submitted to the Institutional Review Board. It was approved on October 8, 2013. As of July 31, 2014, 1,215 of the planned 2,400 participants have completed the final questionnaire.

Aim 3: Examine interactions between air pollution and psychosocial stressors on the prevalence of hypertension, with a focus on quantifying the modifying effects of nonchemical stressors on air pollutant effects.

Work on this aim has not begun.

Future Activities:

We will continue to administer the questionnaire among selected members of the Mano a Mano cohort. A GIS approach will be used to estimate residential ambient air pollutant levels for each participant. For each subject, we will develop metrics for residential proximity to roadways and industrial point sources. Using both traditional methods and the methods developed in the project, multiple logistic regression analyses will be applied to evaluate the modifying effects of neighborhood-based and individual-level psychosocial stressors that potentially affect susceptibility to hypertension due to air pollution. To account for pollutant source, we will conduct sensitivity analyses based on our two proximity measures.

References:

Sorensen H, Cedergreen N, Skovgaard I, Streibig J. An isobole-based statistical model and text for synergism/antagonism in binary mixture toxicity experiments. Environmental and Ecological Statistics 2007;14(4):383-397.

Snipes SA, Thompson B, O'Connor K, Godina R, Ibarra G. Anthropological and psychological merge: design of a stress measure for Mexican farmworkers. Culture, Medicine and Psychiatry 2007;31(3):359-388.

Waller LA, Gotway CA. Applied Spatial Statistics for Public Health Data. Hoboken, NJ: John Wiley & Sons, 2004, 520 pp.

Journal Articles on this Report : 1 Displayed | Download in RIS Format

Publications Views
Other project views:	All 5 publications	3 publications in selected types	All 3 journal articles

Publications
Type	Citation	Project	Document Sources
Journal Article	Symanski E, Karpman M, Jimenez M, Lopez DS, Felknor SA, Upadhyaya M, Strom SS, Bondy ML. Using a community-engaged approach to develop a bilingual survey about psychosocial stressors among individuals of Mexican origin. Journal of Health Care for the Poor and Underserved 2015;26(4):1456-1471.	R834581 (2014) R834581 (2015) R834581 (Final)	Abstract from PubMed Abstract: Johns Hopkins University-Abstract Exit Other: ResearchGate-Abstract Exit

Supplemental Keywords:

air pollution, epidemiology, exposure, sensitive populations , Health, Scientific Discipline, ENVIRONMENTAL MANAGEMENT, Health Risk Assessment, Risk Assessments, Biology, Risk Management, Biochemistry, cumulative exposure, hispanics, cumulative risk, oxidative stress, cardiovascular disease, air pollution, human health risk, latino community, hypertension

Progress and Final Reports:

Original Abstract

The perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.