Beth Ann Griffin

A Tutorial on Propensity Score Estimation for Multiple Treatments Using Generalized Boosted Models

The use of propensity scores to control for pretreatment imbalances on observed variables in non-randomized or observational studies examining the causal effects of treatments or interventions has become widespread over the past decade. For settings with two conditions of interest such as a treatment and a control, inverse probability of treatment weighted estimation with propensity scores estimated via boosted models has been shown in simulation studies to yield causal effect estimates with desirable properties. There are tools (e.g., the twang package in R) and guidance for implementing this method with two treatments. However, there is not such guidance for analyses of three or more treatments. The goals of this paper are twofold: (1) to provide step-by-step guidance for researchers who want to implement propensity score weighting for multiple treatments and (2) to propose the use of generalized boosted models (GBM) for estimation of the necessary propensity score weights. We define the causal quantities that may be of interest to studies of multiple treatments and derive weighted estimators of those quantities. We present a detailed plan for using GBM to estimate propensity scores and using those scores to estimate weights and causal effects. We also provide tools for assessing balance and overlap of pretreatment variables among treatment groups in the context of multiple treatments. A case study examining the effects of three treatment programs for adolescent substance abuse demonstrates the methods.

Housing affordability and health among homeowners and renters.

BACKGROUND Although lack of affordable housing is common in the U.S., few studies have examined the association between housing affordability and health. PURPOSE Using quasi-experimental methods, the aim of this study was to examine whether housing affordability is linked to a number of important health outcomes, controlling for perceptions of neighborhood quality, and determining whether this association differs by housing tenure (renting versus owning). METHODS Data from the 2008 Southeastern Pennsylvania Household Health Survey, a telephone-based survey of 10,004 residents of Philadelphia and its four surrounding counties, were analyzed. The association between housing affordability and health outcomes was assessed using propensity score methods to compare individuals who reported living in unaffordable housing situations to similar individuals living in affordable ones. RESULTS Overall, 48.4% reported difficulty paying housing costs. People living in unaffordable housing had increased odds of poor self-rated health (AOR=1.75, 95% CI=1.33, 2.29); hypertension (AOR=1.34, 95% CI=1.07, 1.69); arthritis (AOR=1.92, 95% CI=1.56, 2.35); cost-related healthcare nonadherence (AOR=2.94, 95% CI=2.04, 4.25); and cost-related prescription nonadherence (AOR=2.68, 95% CI=1.95, 3.70). There were no significant associations between housing affordability and heart disease, diabetes, asthma, psychiatric conditions, being uninsured, emergency department visits in the past year, obesity, and being a current smoker. Renting rather than owning a home heightened the association between unaffordable housing and self-rated health (AOR=2.55, 95% CI=1.93, 3.37 for renters and not significant among homeowners) and cost-related healthcare nonadherence (AOR=4.74, 95% CI=3.05, 7.35 for renters and AOR=1.99, 95% CI=1.15, 3.46 for homeowners). CONCLUSIONS The financial strain of unaffordable housing is associated with trade-offs that may harm health. Programs that target housing affordability for both renters and homeowners may be an important means for improving health.

Effectiveness of Cognitive Tutor Algebra I at Scale

This article examines the effectiveness of a technology-based algebra curriculum in a wide variety of middle schools and high schools in seven states. Participating schools were matched into similar pairs and randomly assigned to either continue with the current algebra curriculum for 2 years or to adopt Cognitive Tutor Algebra I (CTAI), which uses a personalized, mastery-learning, blended-learning approach. Schools assigned to implement CTAI did so under conditions similar to schools that independently adopt it. Analysis of posttest outcomes on an algebra proficiency exam finds no effects in the first year of implementation, but finds evidence in support of positive effects in the second year. The estimated effect is statistically significant for high schools but not for middle schools; in both cases, the magnitude is sufficient to improve the median student’s performance by approximately eight percentile points.

Ambient Particulate Matter Air Pollution and Venous Thromboembolism in the Women's Health Initiative Hormone Therapy Trials

Background The putative effects of postmenopausal hormone therapy on the association between particulate matter (PM) air pollution and venous thromboembolism (VTE) have not been assessed in a randomized trial of hormone therapy, despite its widespread use among postmenopausal women. Objective In this study, we examined whether hormone therapy modifies the association of PM with VTE risk. Methods Postmenopausal women 50–79 years of age (n = 26,450) who did not have a history of VTE and who were not taking anticoagulants were enrolled in the Women’s Health Initiative Hormone Therapy trials at 40 geographically diverse U.S. clinical centers. The women were randomized to treatment with estrogen versus placebo (E trial) or to estrogen plus progestin versus placebo (E + P trial). We used age-stratified Cox proportional hazard models to examine the association between time to incident, centrally adjudicated VTE, and daily mean PM concentrations spatially interpolated at geocoded addresses of the participants and averaged over 1, 7, 30, and 365 days. Results During the follow-up period (mean, 7.7 years), 508 participants (2.0%) had VTEs at a rate of 2.6 events per 1,000 person-years. Unadjusted and covariate-adjusted VTE risk was not associated with concentrations of PM < 2.5 μm (PM2.5) or < 10 μm (PM10)] in aerodynamic diameter and PM × active treatment interactions were not statistically significant (p > 0.05) regardless of PM averaging period, either before or after combining data from both trials [e.g., combined trial-adjusted hazard ratios (95% confidence intervals) per 10 μg/m3 increase in annual mean PM2.5 and PM10, were 0.93 (0.54–1.60) and 1.05 (0.72–1.53), respectively]. Findings were insensitive to alternative exposure metrics, outcome definitions, time scales, analytic methods, and censoring dates. Conclusions In contrast to prior research, our findings provide little evidence of an association between short-term or long-term PM exposure and VTE, or clinically important modification by randomized exposure to exogenous estrogens among postmenopausal women.

The relationship between urban sprawl and coronary heart disease in women

Studies have reported relationships between urban sprawl, physical activity, and obesity, but - to date - no studies have considered the relationship between sprawl and coronary heart disease (CHD) endpoints. In this analysis, we use longitudinal data on post-menopausal women from the Womens Health Initiative (WHI) Clinical Trial to analyze the relationship between metropolitan statistical area (MSA)-level urban compactness (the opposite of sprawl) and CHD endpoints including death, any CHD event, and myocardial infarction. Models control for individual and neighborhood socio-demographic characteristics. Women who lived in more compact communities at baseline had a lower probability of experiencing a CHD event and CHD death or MI during the study follow-up period. One component of compactness, high residential density, had a particularly noteworthy effect on outcomes. Finally, exploratory analyses showed evidence that the effects of compactness were moderated by race and region.

Individual and Neighborhood Socioeconomic Status and the Association between Air Pollution and Cardiovascular Disease.

Background: Long-term fine particulate matter (PM2.5) exposure is linked with cardiovascular disease, and disadvantaged status may increase susceptibility to air pollution-related health effects. In addition, there are concerns that this association may be partially explained by confounding by socioeconomic status (SES). Objectives: We examined the roles that individual- and neighborhood-level SES (NSES) play in the association between PM2.5 exposure and cardiovascular disease. Methods: The study population comprised 51,754 postmenopausal women from the Women’s Health Initiative Observational Study. PM2.5 concentrations were predicted at participant residences using fine-scale regionalized universal kriging models. We assessed individual-level SES and NSES (Census-tract level) across several SES domains including education, occupation, and income/wealth, as well as through an NSES score, which captures several important dimensions of SES. Cox proportional-hazards regression adjusted for SES factors and other covariates to determine the risk of a first cardiovascular event. Results: A 5 μg/m3 higher exposure to PM2.5 was associated with a 13% increased risk of cardiovascular event [hazard ratio (HR) 1.13; 95% confidence interval (CI): 1.02, 1.26]. Adjustment for SES factors did not meaningfully affect the risk estimate. Higher risk estimates were observed among participants living in low-SES neighborhoods. The most and least disadvantaged quartiles of the NSES score had HRs of 1.39 (95% CI: 1.21, 1.61) and 0.90 (95% CI: 0.72, 1.07), respectively. Conclusions: Women with lower NSES may be more susceptible to air pollution-related health effects. The association between air pollution and cardiovascular disease was not explained by confounding from individual-level SES or NSES. Citation: Chi GC, Hajat A, Bird CE, Cullen MR, Griffin BA, Miller KA, Shih RA, Stefanick ML, Vedal S, Whitsel EA, Kaufman JD. 2016. Individual and neighborhood socioeconomic status and the association between air pollution and cardiovascular disease. Environ Health Perspect 124:1840–1847; http://dx.doi.org/10.1289/EHP199

Disparities in Detection and Treatment History Among Mothers With Major Depression in Los Angeles

OBJECTIVE We sought to determine disparities in detection and treatment histories among a group of racial and ethnically diverse mothers with major depression. METHOD Our sample included 276 racially and ethnically diverse mothers who participated in the Los Angeles Family and Neighborhood Survey and who were classified with major depression based on the Comprehensive International Diagnostic Interview-Short Form. We used logistic regression to assess the association between demographic factors and previous detection with major depression, mental health specialty use, and the use of a primary care physician among these women. The demographic factors examined included race and ethnicity, immigration status, marital status, education, income, body mass index (BMI), maternal age, number of children, childrens ages, history of emotional problems, and history of diabetes. RESULTS Results indicated that 69% of mothers had not been previously detected with major depression nor had they sought mental health treatment in the 12 months before the interview. The odds of having been previously diagnosed with major depression were significantly higher among White and single mothers, as well as among mothers with higher BMIs and those with a history of emotional problems. Nonimmigrant mothers without emotional problems had a higher odds of having seen a mental health specialist in the 12 months before the interview compared with immigrant mothers without emotional problems; no differences in mental health treatment were found between nonimmigrant and immigrant mothers with emotional problems. Finally, African-American mothers and those with a history of diabetes had significantly higher odds of seeing a primary care physician compared with Hispanic mothers and those with no history of diabetes, respectively. CONCLUSION Our analyses of a population of depressed mothers living in Los Angeles highlight the need for identification and treatment of racial minority and immigrant mothers.

The effectiveness of community-based delivery of an evidence-based treatment for adolescent substance use

This study evaluates the effectiveness of motivational enhancement therapy/cognitive behavioral therapy-5 (MET/CBT-5) when delivered in community practice settings relative to standard community-based adolescent treatment. A quasi-experimental strategy was used to adjust for pretreatment differences between the MET/CBT-5 sample (n = 2,293) and those who received standard care (n = 458). Results suggest that youth who received MET/CBT-5 fared better than comparable youth in the control group on five of six 12-month outcomes. A low follow-up rate (54%) in the MET/CBT-5 sample raised concerns about nonresponse bias in the treatment effect estimates. Sensitivity analyses suggest that although modest differences in outcomes between the MET/CBT-5 nonrespondents and respondents would yield no significant differences between the two groups on two of the six outcomes, very large differences in outcomes between responders and nonresponders would be required for youth receiving MET/CBT-5 to have fared better had they received standard outpatient care.

Propensity Score Analysis with Survey Weighted Data

Abstract Propensity score analysis (PSA) is a common method for estimating treatment effects, but researchers dealing with data from survey designs are generally not properly accounting for the sampling weights in their analyses. Moreover, recommendations given in the few existing methodological articles on this subject are susceptible to bias. We show in this article through derivation, simulation, and a real data example that using sampling weights in the propensity score estimation stage and the outcome model stage results in an estimator that is robust to a variety of conditions that lead to bias for estimators currently recommended in the statistical literature. We highly recommend researchers use the more robust approach described here. This article provides much needed rigorous statistical guidance for researchers working with survey designs involving sampling weights and using PSAs.

Associations Between Abstinence in Adolescence and Economic and Educational Outcomes Seven Years Later Among High-Risk Youth

AIM In this study, we investigated the relationship between abstinence and long-term educational and economic outcomes among a sample of high-risk youth. METHODS Multivariable regression models were used to estimate associations between abstinence and outcomes among a sample of 13-17 year-olds referred to group homes in Los Angeles in 1999-2000 and followed for 87 months afterwards. Abstinence was measured during the first year of the study. We considered differential effects based on the duration of abstinence (12 vs. 6 months) and type of abstinence (all substances vs. use of alcohol and/or marijuana) on three 87-month outcomes: having received a high-school diploma or equivalent by age 20, institutionalization in the past 90 days, and total legitimate income for the past 90 days. RESULTS Abstinence from all substances for 12 months was associated with positive long-term educational and economic outcomes relative to using any drug during the same time interval. Abstaining from all substances for 12 months was also associated with an increased likelihood of being a legitimate wage earner and decreased likelihood of being institutionalized relative to using only alcohol and/or marijuana during that time interval. No effect on long-term outcomes was seen among youth who abstained for only 6 months relative to those who used drugs during this interval, or for youth who used only alcohol and/or marijuana over 12 months vs. those who used other drugs during this interval. CONCLUSIONS The results presented here justify continued and expanded efforts to promote long periods of abstinence from all drugs for high-risk youth.