Veridical causal inference using propensity score methods for comparative effectiveness research with medical claims

Health Services and Outcomes Research Methodology - Medical insurance claims are becoming increasingly common data sources to answer a variety of questions in biomedical research. Although...     

Improving estimation and prediction in linear regression incorporating external information from an established reduced model

We consider a situation where there is rich historical data available for the coefficients and their standard errors in a linear regression model describing the association between a continuous outcome variable Y and a set of predicting factors X , from a large study. We would like to use this summary information for improving inference in an expanded model of interest, Y given X , B . The additional variable B is a new biomarker, measured on a small number of subjects in a new dataset. We formulate the problem in an inferential framework where the historical information is translated in terms of nonlinear constraints on the parameter space and propose both frequentist and Bayes solutions to this problem. We show that a Bayesian transformation approach proposed by Gunn and Dunson is a simple and effective computational method to conduct approximate Bayesian inference for this constrained parameter problem. The simulation results comparing these methods indicate that historical information on E( Y | X ) can improve the efficiency of estimation and enhance the predictive power in the regression model of interest E( Y | X , B ). We illustrate our methodology by enhancing a published prediction model for bone lead levels in terms of blood lead and other covariates, with a new biomarker defined through a genetic risk score.     

Associations between Extreme Precipitation and Gastrointestinal-Related Hospital Admissions in Chennai,India

Background: Understanding the potential links between extreme weather events and human health in India is important in the context of vulnerability and adaptation to climate change. Research exploring such linkages in India is sparse.Objectives: We evaluated the association between extreme precipitation and gastrointestinal (GI) illness-related hospital admissions in Chennai, India, from 2004 to 2007.Methods: Daily hospital admissions were extracted from two government hospitals in Chennai, India, and meteorological data were retrieved from the Chennai International Airport. We evaluated the association between extreme precipitation (≥ 90th percentile) and hospital admissions using generalized additive models. Both single-day and distributed lag models were explored over a 15-day period, controlling for apparent temperature, day of week, and long-term time trends. We used a stratified analysis to explore the association across age and season.Results: Extreme precipitation was consistently associated with GI-related hospital admissions. The cumulative summary of risk ratios estimated for a 15-day period corresponding to an extreme event (relative to no precipitation) was 1.60 (95% CI: 1.29, 1.98) among all ages, 2.72 (95% CI: 1.25, 5.92) among the young (≤ 5 years of age), and 1.62 (95% CI: 0.97, 2.70) among the old (≥ 65 years of age). The association was stronger during the pre-monsoon season (March–May), with a cumulative risk ratio of 6.50 (95% CI: 2.22, 19.04) for all ages combined compared with other seasons.Conclusions: Hospital admissions related to GI illness were positively associated with extreme precipitation in Chennai, India, with positive cumulative risk ratios for a 15-day period following an extreme event in all age groups. Projected changes in precipitation and extreme weather events suggest that climate change will have important implications for human health in India, where health disparities already exist.Citation: Bush KF, O’Neill MS, Li S, Mukherjee B, Hu H, Ghosh S, Balakrishnan K. 2014. Associations between extreme precipitation and gastrointestinal-related hospital admissions in Chennai, India. Environ Health Perspect 122:249–254; http://dx.doi.org/10.1289/ehp.1306807     

Hemodynamic,Autonomic, and Vascular Effects of Exposure to Coarse Particulate Matter Air Pollution from a Rural Location

Background: Fine particulate matter (PM) air pollution is associated with numerous adverse health effects, including increased blood pressure (BP) and vascular dysfunction. Coarse PM substantially contributes to global air pollution, yet differs in characteristics from fine particles and is currently not regulated. However, the cardiovascular (CV) impacts of coarse PM exposure remain largely unknown.Objectives: Our goal was to elucidate whether coarse PM, like fine PM, is itself capable of eliciting adverse CV responses.Methods: We performed a randomized double-blind crossover study in which 32 healthy adults (25.9 ± 6.6 years of age) were exposed to concentrated ambient coarse particles (CAP; 76.2 ± 51.5 μg/m³) in a rural location and filtered air (FA) for 2 hr. We measured CV outcomes during, immediately after, and 2 hr postexposures.Results: Both systolic (mean difference = 0.32 mmHg; 95% CI: 0.05, 0.58; p = 0.021) and diastolic BP (0.27 mmHg; 95% CI: 0.003, 0.53; p = 0.05) linearly increased per 10 min of exposure during the inhalation of coarse CAP when compared with changes during FA exposure. Heart rate was on average higher (4.1 bpm; 95% CI: 3.06, 5.12; p < 0.0001) and the ratio of low-to-high frequency heart rate variability increased (0.24; 95% CI: 0.07, 0.41; p = 0.007) during coarse particle versus FA exposure. Other outcomes (brachial flow-mediated dilatation, microvascular reactive hyperemia index, aortic hemodynamics, pulse wave velocity) were not differentially altered by the exposures.Conclusions: Inhalation of coarse PM from a rural location is associated with a rapid elevation in BP and heart rate during exposure, likely due to the triggering of autonomic imbalance. These findings add mechanistic evidence supporting the biological plausibility that coarse particles could contribute to the triggering of acute CV events.Citation: Brook RD, Bard RL, Morishita M, Dvonch JT, Wang L, Yang HY, Spino C, Mukherjee B, Kaplan MJ, Yalavarthi S, Oral EA, Ajluni N, Sun Q, Brook JR, Harkema J, Rajagopalan S. 2014. Hemodynamic, autonomic, and vascular effects of exposure to coarse particulate matter air pollution from a rural location. Environ Health Perspect 122:624–630; http://dx.doi.org/10.1289/ehp.1306595     

Propensity score‐based diagnostics for categorical response regression models

For binary or categorical response models, most goodness‐of‐fit statistics are based on the notion of partitioning the subjects into groups or regions and comparing the observed and predicted responses in these regions by a suitable chi‐squared distribution. Existing strategies create this partition based on the predicted response probabilities, or propensity scores, from the fitted model. In this paper, we follow a retrospective approach, borrowing the notion of balancing scores used in causal inference to inspect the conditional distribution of the predictors, given the propensity scores, in each category of the response to assess model adequacy. We can use this diagnostic under both prospective and retrospective sampling designs, and it may ascertain general forms of misspecification. We first present simple graphical and numerical summaries that can be used in a binary logistic model. We then generalize the tools to propose model diagnostics for the proportional odds model. We illustrate the methods with simulation studies and two data examples: (i) a case‐control study of the association between cumulative lead exposure and Parkinson's disease in the Boston, Massachusetts, area and (ii) and a cohort study of biomarkers possibly associated with diabetes, from the VA Normative Aging Study. Copyright © 2013 John Wiley & Sons, Ltd.     

Selection of nonlinear interactions by a forward stepwise algorithm: Application to identifying environmental chemical mixtures affecting health outcomes

In this paper, we propose a stepwise forward selection algorithm for detecting the effects of a set of correlated exposures and their interactions on a health outcome of interest when the underlying relationship could potentially be nonlinear. Though the proposed method is very general, our application in this paper remains to be on analysis of multiple pollutants and their interactions. Simultaneous exposure to multiple environmental pollutants could affect human health in a multitude of complex ways. For understanding the health effects of multiple environmental exposures, it is often important to identify and estimate complex interactions among exposures. However, this issue becomes analytically challenging in the presence of potential nonlinearity in the outcome-exposure response surface and a set of correlated exposures. Through simulation studies and analyses of test datasets that were simulated as a part of a data challenge in multipollutant modeling organized by the National Institute of Environmental Health Sciences ( http://www.niehs.nih.gov/about/events/pastmtg/2015/statistical/ ), we illustrate the advantages of our proposed method in comparison with existing alternative approaches. A particular strength of our method is that it demonstrates very low false positives across empirical studies. Our method is also used to analyze a dataset that was released from the Health Outcomes and Measurement of the Environment Study as a benchmark beta-tester dataset as a part of the same workshop.     

PM2.5-induced changes in cardiac function of hypertensive rats depend on wind direction and specific sources in Steubenville, Ohio

BACKGROUND: Increases in particulate matter less than 2.5 μm (PM(2.5)) in ambient air is linked to acute cardiovascular morbidity and mortality. Specific components and potential emission sources of PM(2.5) responsible for adverse health effects of cardiovascular function are unclear. Methods: Spontaneously hypertensive rats were implemented with radiotelemeters to record ECG responses during inhalation exposure to concentrated ambient particles (CAPs) for 13 consecutive days in Steubenville, OH. Changes in heart rate (HR) and its variability (HRV) were compared to PM(2.5) trace elements in 30-min time frames to capture acute physiological responses with real-time fluctuations in PM(2.5) composition. Using positive matrix factorization, six major source factors were identified: (i) coal/secondary, (ii) mobile sources, (iii) metal coating/processing, (iv) iron/steel manufacturing, (v) lead and (vi) incineration. Results: Exposure-related changes in HR and HRV were dependant on winds predominately from either the northeast (NE) or southwest (SW). During SW winds, the metal processing factor was associated with increased HR, whereas factors of incineration, lead and iron/steel with NE winds were associated with decreased HR. Decreased SDNN was dominated during NE winds by the incinerator factor, and with SW winds by the metal factor. Metals and mobile source factors also had minor impacts on decreased SDNN with NE winds. Individual elemental components loaded onto these factors generally showed significant associations, although there were some discrepancies. Conclusions: Acute cardiovascular changes in response to ambient PM(2.5) exposure can be attributed to specific PM constituents and sources linked with incineration, metal processing, and iron/steel production.     

Efficient designs of gene-environment interaction studies: implications of Hardy-Weinberg equilibrium and gene-environment independence

It is important to investigate whether genetic susceptibility variants exercise the same effects in populations that are differentially exposed to environmental risk factors. Here, we assess the power of four two-stage case-control design strategies for assessing multiplicative gene-environment (G-E) interactions or for assessing genetic or environmental effects in the presence of G-E interactions. We considered a di-allelic single nucleotide polymorphism G and a binary environmental variable E under the constraints of G-E independence and Hardy-Weinberg equilibrium and used the Wald statistic for all tests. We concluded that (i) for testing G-E interactions or genetic effects in the presence of G-E interactions when data for E are fully available, it is preferable to ascertain data for G in a subsample of cases with similar numbers of exposed and unexposed and a random subsample of controls; and (ii) for testing G-E interactions or environmental effects in the presence of G-E interactions when data for G are fully available, it is preferable to ascertain data for E in a subsample of cases that has similar numbers for each genotype and a random subsample of controls. In addition, supplementing external control data to an existing case-control sample leads to improved power for assessing effects of G or E in the presence of G-E interactions. Copyright ? 2012 John Wiley & Sons, Ltd.