Identification of complier and noncomplier average causal effects in the presence of latent missing-at-random (LMAR) outcomes: a unifying view and choices of assumptions.

The study of treatment effects is often complicated by noncompliance and missing data. In the one-sided noncompliance setting where of interest are the complier and noncomplier average causal effects, we address outcome missingness of the latent missing at random type (LMAR, also known as latent ignorability). That is, conditional on covariates and treatment assigned, the missingness may depend on compliance type. Within the instrumental variable (IV) approach to noncompliance, methods have been proposed for handling LMAR outcome that additionally invoke an exclusion restriction-type assumption on missingness, but no solution has been proposed for when a non-IV approach is used. This article focuses on effect identification in the presence of LMAR outcomes, with a view to flexibly accommodate different principal identification approaches. We show that under treatment assignment ignorability and LMAR only, effect nonidentifiability boils down to a set of two connected mixture equations involving unidentified stratum-specific response probabilities and outcome means. This clarifies that (except for a special case) effect identification generally requires two additional assumptions: a specific missingness mechanism assumption and a principal identification assumption. This provides a template for identifying effects based on separate choices of these assumptions. We consider a range of specific missingness assumptions, including those that have appeared in the literature and some new ones. Incidentally, we find an issue in the existing assumptions, and propose a modification of the assumptions to avoid the issue. Results under different assumptions are illustrated using data from the Baltimore Experience Corps Trial.

Multiple imputation for propensity score analysis with covariates missing at random: some clarity on within and across methods.

In epidemiology and social sciences, propensity score methods are popular for estimating treatment effects using observational data, and multiple imputation is popular for handling covariate missingness. However, how to appropriately use multiple imputation for propensity score analysis is not completely clear. This paper aims to bring clarity on the consistency (or lack thereof) of methods that have been proposed, focusing on the within approach (where the effect is estimated separately in each imputed dataset and then the multiple estimates are combined) and the across approach (where typically propensity scores are averaged across imputed datasets before being used for effect estimation). We show that the within method is valid and can be used with any causal effect estimator that is consistent in the full-data setting. Existing across methods are inconsistent, but a different across method that averages the inverse probability weights across imputed datasets is consistent for propensity score weighting. We also comment on methods that rely on imputing a function of the missing covariate rather than the covariate itself, including imputation of the propensity score and of the probability weight. Based on consistency results and practical flexibility, we recommend generally using the standard within method. Throughout, we provide intuition to make the results meaningful to the broad audience of applied researchers.

The Importance of Equity Value Judgments and Estimator-Estimand Alignment in Measuring Disparity and Identifying Targets to Reduce Disparity.

The choice of which covariates to adjust for (so-called allowability designation (AD)) in health disparity measurements reflects value judgments about inequitable versus equitable sources of health differences, which is paramount for making inferences about disparity. Yet, many off-the-shelf estimators used in health disparity research are not designed with equity considerations in mind, and they imply different ADs. We demonstrated the practical importance of incorporating equity concerns in disparity measurements through simulations, motivated by the example of reducing racial disparities in hypertension control via interventions on disparities in treatment intensification. Seven causal decomposition estimators, each with a particular AD (with respect to disparities in hypertension control and treatment intensification), were considered to estimate the observed outcome disparity and the reduced/residual disparity under the intervention. We explored the implications for bias of the mismatch between equity concerns and the AD in the estimator under various causal structures (through altering racial differences in covariates or the confounding mechanism). The estimator that correctly reflects equity concerns performed well under all scenarios considered, whereas the other estimators were shown to have the risk of yielding large biases in certain scenarios, depending on the interaction between their ADs and the specific causal structure.

Sensitivity analysis for principal ignorability violation in estimating complier and noncomplier average causal effects.

An important strategy for identifying principal causal effects (popular estimands in settings with noncompliance) is to invoke the principal ignorability (PI) assumption. As PI is untestable, it is important to gauge how sensitive effect estimates are to its violation. We focus on this task for the common one-sided noncompliance setting where there are two principal strata, compliers and noncompliers. Under PI, compliers and noncompliers share the same outcome-mean-given-covariates function under the control condition. For sensitivity analysis, we allow this function to differ between compliers and noncompliers in several ways, indexed by an odds ratio, a generalized odds ratio, a mean ratio, or a standardized mean difference sensitivity parameter. We tailor sensitivity analysis techniques (with any sensitivity parameter choice) to several types of PI-based main analysis methods, including outcome regression, influence function (IF) based and weighting methods. We discuss range selection for the sensitivity parameter. We illustrate the sensitivity analyses with several outcome types from the JOBS II study. This application estimates nuisance functions parametrically - for simplicity and accessibility. In addition, we establish rate conditions on nonparametric nuisance estimation for IF-based estimators to be asymptotically normal - with a view to inform nonparametric inference.

Comparison of methods that combine multiple randomized trials to estimate heterogeneous treatment effects.

Individualized treatment decisions can improve health outcomes, but using data to make these decisions in a reliable, precise, and generalizable way is challenging with a single dataset. Leveraging multiple randomized controlled trials allows for the combination of datasets with unconfounded treatment assignment to better estimate heterogeneous treatment effects. This article discusses several nonparametric approaches for estimating heterogeneous treatment effects using data from multiple trials. We extend single-study methods to a scenario with multiple trials and explore their performance through a simulation study, with data generation scenarios that have differing levels of cross-trial heterogeneity. The simulations demonstrate that methods that directly allow for heterogeneity of the treatment effect across trials perform better than methods that do not, and that the choice of single-study method matters based on the functional form of the treatment effect. Finally, we discuss which methods perform well in each setting and then apply them to four randomized controlled trials to examine effect heterogeneity of treatments for major depressive disorder.

Estimation of place-based vulnerability scores for HIV viral non-suppression: an application leveraging data from a cohort of people with histories of using drugs.

The relationships between place (e.g., neighborhood) and HIV are commonly investigated. As measurements of place are multivariate, most studies apply some dimension reduction, resulting in one variable (or a small number of variables), which is then used to characterize place. Typical dimension reduction methods seek to capture the most variance of the raw items, resulting in a type of summary variable we call "disadvantage score". We propose to add a different type of summary variable, the "vulnerability score," to the toolbox of the researchers doing place and HIV research. The vulnerability score measures how place, as known through the raw measurements, is predictive of an outcome. It captures variation in place characteristics that matters most for the particular outcome. We demonstrate the estimation and utility of place-based vulnerability scores for HIV viral non-suppression, using data with complicated clustering from a cohort of people with histories of injecting drugs.

Methods for Integrating Trials and Non-experimental Data to Examine Treatment Effect Heterogeneity.

Estimating treatment effects conditional on observed covariates can improve the ability to tailor treatments to particular individuals. Doing so effectively requires dealing with potential confounding, and also enough data to adequately estimate effect moderation. A recent influx of work has looked into estimating treatment effect heterogeneity using data from multiple randomized controlled trials and/or observational datasets. With many new methods available for assessing treatment effect heterogeneity using multiple studies, it is important to understand which methods are best used in which setting, how the methods compare to one another, and what needs to be done to continue progress in this field. This paper reviews these methods broken down by data setting: aggregate-level data, federated learning, and individual participant-level data. We define the conditional average treatment effect and discuss differences between parametric and nonparametric estimators, and we list key assumptions, both those that are required within a single study and those that are necessary for data combination. After describing existing approaches, we compare and contrast them and reveal open areas for future research. This review demonstrates that there are many possible approaches for estimating treatment effect heterogeneity through the combination of datasets, but that there is substantial work to be done to compare these methods through case studies and simulations, extend them to different settings, and refine them to account for various challenges present in real data.

Flexible propensity score estimation strategies for clustered data in observational studies.

Existing studies have suggested superior performance of nonparametric machine learning over logistic regression for propensity score estimation. However, it is unclear whether the advantages of nonparametric propensity score modeling are carried to settings where there is clustering of individuals, especially when there is unmeasured cluster-level confounding. In this work we examined the performance of logistic regression (all main effects), Bayesian additive regression trees and generalized boosted modeling for propensity score weighting in clustered settings, with the clustering being accounted for by including either cluster indicators or random intercepts. We simulated data for three hypothetical observational studies of varying sample and cluster sizes. Confounders were generated at both levels, including a cluster-level confounder that is unobserved in the analyses. A binary treatment and a continuous outcome were generated based on seven scenarios with varying relationships between the treatment and confounders (linear and additive, nonlinear/nonadditive, nonadditive with the unobserved cluster-level confounder). Results suggest that when the sample and cluster sizes are large, nonparametric propensity score estimation may provide better covariate balance, bias reduction, and 95% confidence interval coverage, regardless of the degree of nonlinearity or nonadditivity in the true propensity score model. When the sample or cluster sizes are small, however, nonparametric approaches may become more vulnerable to unmeasured cluster-level confounding and thus may not be a better alternative to multilevel logistic regression. We applied the methods to the National Longitudinal Study of Adolescent to Adult Health data, estimating the effect of team sports participation during adolescence on adulthood depressive symptoms.

It's all about balance: propensity score matching in the context of complex survey data.

Many research studies aim to draw causal inferences using data from large, nationally representative survey samples, and many of these studies use propensity score matching to make those causal inferences as rigorous as possible given the non-experimental nature of the data. However, very few applied studies are careful about incorporating the survey design with the propensity score analysis, which may mean that the results do not generate population inferences. This may be because few methodological studies examine how to best combine these methods. Furthermore, even fewer of them investigate different non-response mechanisms. This study examines methods for handling survey weights in propensity score matching analyses of survey data under different non-response mechanisms. Our main conclusions are: (i) whether the survey weights are incorporated in the estimation of the propensity score does not impact estimation of the population treatment effect, as long as good population treated-comparison balance is achieved on confounders, (ii) survey weights must be used in the outcome analysis, and (iii) the transferring of survey weights (i.e., assigning the weights of the treated units to the comparison units matched to them) can be beneficial under certain non-response mechanisms.

Does Marijuana Use at Ages 16-18 Predict Initiation of Daily Cigarette Smoking in Late Adolescence and Early Adulthood? A Propensity Score Analysis of Add Health Data.

Given the declining trend in adolescent cigarette smoking and increase in general access to marijuana, it is important to examine whether marijuana use in adolescence is a risk factor for subsequent cigarette smoking in late adolescence and early adulthood. Preliminary evidence from a very small number of studies suggests that marijuana use during adolescence is associated with later smoking; however, to control confounding, previously published studies used regression adjustment, which is susceptible to extrapolation when the confounder distributions differ between adolescent marijuana users and non-users. The current study uses propensity score weighting, a causal inference method not previously used in this area of research, to weight participants based on their estimated probability of exposure given confounders (the propensity score) to balance observed confounders between marijuana users and non-users. The sample consists of participants of Add Health (a nationally representative dataset of youth followed into adulthood) who were 16-18, with no history of daily cigarette smoking at baseline (n = 2928 for female and 2731 for male sub-samples). We assessed the effect of adolescent marijuana use (exposure, ascertained at wave 1) on any daily cigarette smoking during the subsequent 13 years (outcome, ascertained at wave 4). Analyses suggest that for females (but not males) who used marijuana in adolescence, marijuana use increased the risk for subsequent daily smoking: OR = 1.71, 95% CI = (1.13, 2.59). We recommend that adolescent marijuana use be viewed as a possible risk factor for subsequent initiation of daily cigarette smoking in women.

Low Uptake of Voluntary Medical Male Circumcision Among High Risk Men in Malawi.

Malawi is one of 14 priority countries for voluntary medical male circumcision (VMMC) initiatives with the lowest VMMC uptake. Using data from a study of 269 men accessing VMMC in southern Malawi and latent class analysis, men were classified based on four risk factors: ever tested for HIV, condom use at last sex, having casual/concurrent sexual partners, and using alcohol before sex. Two distinct classes were identified: 8% of men were classified as high risk, while 92% were classified as low/medium risk. Poisson regression modeling indicated that men who had lower education (risk ratio [RR] 1.07, p < 0.05) and were ages 19-26 (RR 1.07, p < 0.05) were more likely to be in the high risk group. The low numbers of men in the high risk category seeking services suggests the need to implement targeted strategies to increase VMMC uptake among such high risk men.

The Internalized Homophobia Scale for Vietnamese Sexual Minority Women: Conceptualization, Factor Structure, Reliability, and Associations With Hypothesized Correlates.

We developed the first Vietnamese Internalized Homophobia (IH) scale for use with Vietnamese sexual minority women (SMW). Drawing from existing IH scales in the international literature and based on prior qualitative research about SMW in the Viet Nam context, the scale covers two domains: self-stigma (negative attitudes toward oneself as a sexual minority person) and sexual prejudice (negative attitudes toward homosexuality/same-sex relations in general). Scale items, including items borrowed from existing scales and items based on local expressions, were reviewed and confirmed by members of the target population. Quantitative evaluation used data from an anonymous web-based survey of Vietnamese SMW, including those who identified as lesbian (n = 1187), or as bisexual (n = 641) and those who were unsure about their sexual identity (n = 353). The scale was found to consist of two highly correlated factors reflecting self-stigma (not normal/wholesome and self-reproach and wishing away same-sex sexuality) and one factor reflecting sexual prejudice, and to have excellent internal consistency. Construct validity was evidenced by subscale associations with a wide range of hypothesized correlates, including perceived sexual stigma, outness, social support, connection to other SMW, relationship quality, psychological well-being, anticipation of heterosexual marriage, and endorsement of same-sex marriage legalization. Self-stigma was more strongly associated with psychosocial correlates, and sexual prejudice was more associated with endorsement of legal same-sex marriage. The variations in these associations across the hypothesized correlates and across sexual identity groups were consistent with the minority stress model and the IH literature, and exhibited context-specific features, which are discussed.

Condomless Anal Intercourse Among Males and Females at High Risk for Heterosexual HIV Infection.

BACKGROUND: Understanding and addressing heterosexual HIV transmission requires attention to the range and context of heterosexual sexual behaviors. We sought to determine population-based prevalence of condomless anal intercourse (CAI) among individuals at increased heterosexual HIV risk in Baltimore and to identify demographic, behavioral, and health-related correlates. METHODS: Data were from a cross-sectional study of 185 males and 198 females at increased heterosexual risk for HIV recruited using respondent-driven sampling as part of Centers for Disease Control and Prevention's National HIV Behavioral Surveillance Project in Baltimore, August-December 2010. Bivariate and multivariate logistic regression examined factors associated with heterosexual CAI. RESULTS: The sample was majority African American, with a mean age of 38 years among men and 34 years among women. Forty-two percent of men (95% confidence interval, 30.9%-52.0%) and 38% of women (95% confidence interval, 29.4%-47.2%) reported any CAI in the past year, with variance by partner type and sex. Among men, CAI was significantly associated with homelessness, casual and exchange partners, same-sex partner in past year, and substance use. Among women, CAI was significantly associated with lower education, casual and exchange partners, same-sex partner in past year, multiple partners, and substance use. In adjusted sex-specific models, males and females with increasing numbers of partners were more likely to engage in CAI. CONCLUSIONS: It is important to recognize the efficiency of transmission of HIV and other sexually transmitted infections through CAI. There is a need to broaden heterosexual sexual health promotion and HIV/sexually transmitted infection prevention to adequately and appropriately address risks and prevention strategies for anal intercourse.

Syringe exchange in the United States: a national level economic evaluation of hypothetical increases in investment.

To examine whether increasing investment in needle/syringe exchange programs (NSPs) in the US would be cost-effective for HIV prevention, we modeled HIV incidence in hypothetical cases with higher NSP syringe supply than current levels, and estimated number of infections averted, cost per infection averted, treatment costs saved, and financial return on investment. We modified Pinkerton's model, which was an adaptation of Kaplan's simplified needle circulation theory model, to compare different syringe supply levels, account for syringes from non-NSP sources, and reflect reduction in syringe sharing and contamination. With an annual $10 to $50 million funding increase, 194-816 HIV infections would be averted (cost per infection averted $51,601-$61,302). Contrasted with HIV treatment cost savings alone, the rate of financial return on investment would be 7.58-6.38. Main and sensitivity analyses strongly suggest that it would be cost-saving for the US to invest in syringe exchange expansion.

Trajectories of craving in the course of pharmacotherapy trials for methamphetamine use disorder.

AIMS: The aim of this study was to measure trajectories of craving for methamphetamine during the course of pharmacotherapy trials for methamphetamine use disorder. DESIGN, SETTING AND PARTICIPANTS: Craving trajectories were identified using Group-Based Trajectory Modeling. The association of craving trajectories with drug use trajectories was examined using a dual trajectory model. Association of craving trajectories with other health and social outcomes was also examined. The study used pooled data from five randomized controlled pharmacotherapy trials for methamphetamine use disorder. A total of 866 adults with methamphetamine use disorder participated in randomized controlled pharmacotherapy trials. MEASUREMENT: Craving was assessed weekly using the Brief Substance Craving Scale. Drug use was assessed using urine toxicology. Alcohol- and drug-related problems, as well as psychiatric, medical, legal, employment and relationship problems, were measured using the Addiction Severity Index. FINDINGS: A three-trajectory model with high, medium and low craving trajectories was selected as the most parsimonious model. Craving trajectories were associated with methamphetamine use trajectories in the course of trial; 88.4% of those in the high craving trajectory group had a consistently high frequency of methamphetamine use compared with 18.7% of those in the low craving group. High craving was also associated with less improvement in most other outcomes and higher rate of dropout from treatment. In turn, low craving was associated with a rapidly decreasing frequency of methamphetamine use, greater improvement in most other outcomes and a lower rate of dropout. Participants on modafinil daily and ondansetron 1 mg twice daily were less likely to be in the high craving group compared with those on placebo. CONCLUSIONS: Trajectories of methamphetamine craving in the course of clinical trials for methamphetamine use disorder appear to be both highly variable and strongly associated with greater frequency of drug use, other drug-related outcomes and dropout from trials. Two medications, modafinil daily and ondansetron at a dose of 1 mg two times daily, appear to be associated with greater reduction in craving in the course of treatment compared with placebo. A decrease in methamphetamine craving shows promise as an early indicator of recovery from methamphetamine use disorder.

Causal mediation analysis: From simple to more robust strategies for estimation of marginal natural (in)direct effects.

This paper aims to provide practitioners of causal mediation analysis with a better understanding of estimation options. We take as inputs two familiar strategies (weighting and model-based prediction) and a simple way of combining them (weighted models), and show how a range of estimators can be generated, with different modeling requirements and robustness properties. The primary goal is to help build intuitive appreciation for robust estimation that is conducive to sound practice. We do this by visualizing the target estimand and the estimation strategies. A second goal is to provide a "menu" of estimators that practitioners can choose from for the estimation of marginal natural (in)direct effects. The estimators generated from this exercise include some that coincide or are similar to existing estimators and others that have not previously appeared in the literature. We note several different ways to estimate the weights for cross-world weighting based on three expressions of the weighting function, including one that is novel; and show how to check the resulting covariate and mediator balance. We use a random continuous weights bootstrap to obtain confidence intervals, and also derive general asymptotic variance formulas for the estimators. The estimators are illustrated using data from an adolescent alcohol use prevention study. R-code is provided.

Clarifying causal mediation analysis: Effect identification via three assumptions and five potential outcomes.

Causal mediation analysis is complicated with multiple effect definitions that require different sets of assumptions for identification. This article provides a systematic explanation of such assumptions. We define five potential outcome types whose means are involved in various effect definitions. We tackle their mean/distribution's identification, starting with the one that requires the weakest assumptions and gradually building up to the one that requires the strongest assumptions. This presentation shows clearly why an assumption is required for one estimand and not another, and provides a succinct table from which an applied researcher could pick out the assumptions required for identifying the causal effects they target. Using a running example, the article illustrates the assembling and consideration of identifying assumptions for a range of causal contrasts. For several that are commonly encountered in the literature, this exercise clarifies that identification requires weaker assumptions than those often stated in the literature. This attention to the details also draws attention to the differences in the positivity assumption for different estimands, with practical implications. Clarity on the identifying assumptions of these various estimands will help researchers conduct appropriate mediation analyses and interpret the results with appropriate caution given the plausibility of the assumptions.

Comparing the performance of statistical methods that generalize effect estimates from randomized controlled trials to much larger target populations.

Policymakers use results from randomized controlled trials to inform decisions about whether to implement treatments in target populations. Various methods - including inverse probability weighting, outcome modeling, and Targeted Maximum Likelihood Estimation - that use baseline data available in both the trial and target population have been proposed to generalize the trial treatment effect estimate to the target population. Often the target population is significantly larger than the trial sample, which can cause estimation challenges. We conduct simulations to compare the performance of these methods in this setting. We vary the size of the target population, the proportion of the target population selected into the trial, and the complexity of the true selection and outcome models. All methods performed poorly when the trial size was only 2% of the target population size or the target population included only 1,000 units. When the target population or the proportion of units selected into the trial was larger, some methods, such as outcome modeling using Bayesian Additive Regression Trees, performed well. We caution against generalizing using these existing approaches when the target population is much larger than the trial sample and advocate future research strives to improve methods for generalizing to large target populations.

Clarifying causal mediation analysis for the applied researcher: Defining effects based on what we want to learn.

The incorporation of causal inference in mediation analysis has led to theoretical and methodological advancements-effect definitions with causal interpretation, clarification of assumptions required for effect identification, and an expanding array of options for effect estimation. However, the literature on these results is fast-growing and complex, which may be confusing to researchers unfamiliar with causal inference or unfamiliar with mediation. The goal of this article is to help ease the understanding and adoption of causal mediation analysis. It starts by highlighting a key difference between the causal inference and traditional approaches to mediation analysis and making a case for the need for explicit causal thinking and the causal inference approach in mediation analysis. It then explains in as-plain-as-possible language existing effect types, paying special attention to motivating these effects with different types of research questions, and using concrete examples for illustration. This presentation differentiates 2 perspectives (or purposes of analysis): the explanatory perspective (aiming to explain the total effect) and the interventional perspective (asking questions about hypothetical interventions on the exposure and mediator, or hypothetically modified exposures). For the latter perspective, the article proposes tapping into a general class of interventional effects that contains as special cases most of the usual effect types-interventional direct and indirect effects, controlled direct effects and also a generalized interventional direct effect type, as well as the total effect and overall effect. This general class allows flexible effect definitions which better match many research questions than the standard interventional direct and indirect effects. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

Outcome evaluation of a "common factors" approach to develop culturally tailored HIV prevention interventions for people who inject drugs.

BACKGROUND: Current models of HIV prevention intervention dissemination involve packaging interventions developed in one context and training providers to implement that specific intervention with fidelity. Providers rarely implement these programs with fidelity due to perceived incompatibility, resource constraints, and preference for locally-generated solutions. Moreover, such interventions may not reflect local drug markets and drug use practices that contribute to HIV risk. PURPOSE: This paper examines whether provider-developed interventions based on common factors of effective, evidence-based behavioral interventions led to reduction in drug-related HIV risk behaviors at four study sites in Ukraine. METHODS: We trained staff from eight nongovernmental organizations (NGOs) to develop HIV prevention interventions based on a common factors approach. We then selected four NGOs to participate in an outcome evaluation. Each NGO conducted its intervention for at least N = 130 participants, with baseline and 3-month follow-up assessments. RESULTS: At three sites, we observed reductions in the prevalence of both any risk in drug acquisition and any risk in drug injection. At the fourth site, prevalence of any risk in drug injection decreased substantially, but the prevalence of any risk in drug acquisition essentially stayed unchanged. CONCLUSIONS: The common factors approach has some evidence of efficacy in implementation, but further research is needed to assess its effectiveness in reducing HIV risk behaviors and transmission. Behavioral interventions to reduce HIV risk developed using the common factors approach could become an important part of the HIV response in low resource settings where capacity building remains a high priority.