Maya L. Petersen

Estimation of Direct Causal Effects

Many common problems in epidemiologic and clinical research involve estimating the effect of an exposure on an outcome while blocking the exposures effect on an intermediate variable. Effects of this kind are termed direct effects. Estimation of direct effects is typically the goal of research aimed at understanding mechanistic pathways by which an exposure acts to cause or prevent disease, as well as in many other settings. Although multivariable regression is commonly used to estimate direct effects, this approach requires assumptions beyond those required for the estimation of total causal effects. In addition, when the exposure and intermediate variables interact to cause disease, multivariable regression estimates a particular type of direct effect—the effect of an exposure on an outcome when the intermediate is fixed at a specified level. Using the counterfactual framework, we distinguish this definition of a direct effect (controlled direct effect) from an alternative definition, in which the effect of the exposure on the intermediate is blocked, but the intermediate is otherwise allowed to vary as it would in the absence of exposure (natural direct effect). We illustrate the difference between controlled and natural direct effects using several examples. We present an estimation approach for natural direct effects that can be implemented using standard statistical software, and we review the assumptions underlying our approach (which are less restrictive than those proposed by previous authors).

Diagnosing and Responding to Violations in the Positivity Assumption

The assumption of positivity or experimental treatment assignment requires that observed treatment levels vary within confounder strata. This article discusses the positivity assumption in the context of assessing model and parameter-specific identifiability of causal effects. Positivity violations occur when certain subgroups in a sample rarely or never receive some treatments of interest. The resulting sparsity in the data may increase bias with or without an increase in variance and can threaten valid inference. The parametric bootstrap is presented as a tool to assess the severity of such threats and its utility as a diagnostic is explored using simulated and real data. Several approaches for improving the identifiability of parameters in the presence of positivity violations are reviewed. Potential responses to data sparsity include restriction of the covariate adjustment set, use of an alternative projection function to define the target parameter within a marginal structural working model, restriction of the sample, and modification of the target intervention. All of these approaches can be understood as trading off proximity to the initial target of inference for identifiability; we advocate approaching this tradeoff systematically.

Promoting Transparency in Social Science Research

Social scientists should adopt higher transparency standards to improve the quality and credibility of research. There is growing appreciation for the advantages of experimentation in the social sciences. Policy-relevant claims that in the past were backed by theoretical arguments and inconclusive correlations are now being investigated using more credible methods. Changes have been particularly pronounced in development economics, where hundreds of randomized trials have been carried out over the last decade. When experimentation is difficult or impossible, researchers are using quasi-experimental designs. Governments and advocacy groups display a growing appetite for evidence-based policy-making. In 2005, Mexico established an independent government agency to rigorously evaluate social programs, and in 2012, the U.S. Office of Management and Budget advised federal agencies to present evidence from randomized program evaluations in budget requests (1, 2).

Prevention of Haemophilus influenzae Type b (Hib) Meningitis and Emergence of Serotype Replacement with Type a Strains after Introduction of Hib Immunization in Brazil

Surveillance for Haemophilus influenzae meningitis cases was performed in Salvador, Brazil, before and after introduction of H. influenzae type b (Hib) immunization. The incidence of Hib meningitis decreased 69% during the 1-year period after initiation of Hib immunization (from 2.62 to 0.81 cases/100,000 person-years; P<.001). In contrast, the incidence for H. influenzae type a meningitis increased 8-fold (from 0.02 to 0.16 cases/100,000 person-years; P=.008). Pulsed-field gel electrophoretic analysis demonstrated that H. influenzae type a isolates belonged to 2 clonally related groups, both of which were found before Hib immunization commenced. Therefore, Hib immunization contributed to an increased risk for H. influenzae type a meningitis through selection of circulating H. influenzae type a clones. The risk attributable to serotype replacement is small in comparison to the large reduction in Hib meningitis due to immunization. However, these findings highlight the need to maintain surveillance as the use of conjugate vaccines expands worldwide.

Pillbox Organizers are Associated with Improved Adherence to HIV Antiretroviral Therapy and Viral Suppression: a Marginal Structural Model Analysis

BACKGROUND Pillbox organizers are inexpensive and easily used; however, their effect on adherence to antiretroviral medications is unknown. METHODS Data were obtained from an observational cohort of 245 human immunodeficiency virus (HIV)-infected subjects who were observed from 1996 through 2000 in San Francisco, California. Adherence was the primary outcome and was measured using unannounced monthly pill counts. Plasma HIV RNA level was considered as a secondary outcome. Marginal structural models were used to estimate the effect of pillbox organizer use on adherence and viral suppression, adjusting for confounding by CD4+ T cell count, viral load, prior adherence, recreational drug use, demographic characteristics, and current and past treatment. RESULTS Pillbox organizer use was estimated to improve adherence by 4.1%-4.5% and was associated with a decrease in viral load of 0.34-0.37 log10 copies/mL and a 14.2%-15.7% higher probability of achieving a viral load < or = 400 copies/mL (odds ratio, 1.8-1.9). All effect estimates were statistically significant. CONCLUSION Pillbox organizers appear to significantly improve adherence to antiretroviral therapy and to improve virologic suppression. We estimate that pillbox organizers may be associated with a cost of approximately

Long-term consequences of the delay between virologic failure of highly active antiretroviral therapy and regimen modification

19,000 per quality-adjusted life-year. Pillbox organizers should be a standard intervention to improve adherence to antiretroviral therapy.

A Marginal Structural Model to Estimate the Causal Effect of Antidepressant Medication Treatment on Viral Suppression Among Homeless and Marginally Housed Persons With HIV

OBJECTIVE Studies have found that CD8 T-cell activation, as measured by CD38 expression, in HIV-1-infected individuals on suppressive therapy for longer than 12 months is not predictive of CD4 T-cell recovery. Owing to the fact that reconstitution of memory and naive T-cell populations occurs differentially over time, this study evaluated whether distinct memory/naive CD4 T-cell subsets correlated with CD38 on CD8 T-cells. MATERIALS AND METHODS Whole blood from 13 participants was used to evaluate activation phenotypic markers on CD8 lymphocytes and memory/naive phenotypes on CD4 lymphocytes. These HIV-1-infected individuals had stable CD4 cell counts for more than 1 year while on suppressive combination antiretroviral therapy. RESULTS The results demonstrate that CD4 central memory and naive cell populations contribute to the magnitude of CD4 T-cell reconstitution. CD4 central memory has a significant negative correlation with the percentage of CD38-activated CD8 T-cells. CONCLUSION This suggests that CD8 activation is important in CD4 recovery from a low CD4 T-cell nadir.Objectives:Current treatment guidelines recommend immediate modification of antiretroviral therapy in HIV-infected individuals with incomplete viral suppression. These recommendations have not been tested in observational studies or large randomized trials. We evaluated the consequences of delayed modification following virologic failure. Design/methods:We used prospective data from two clinical cohorts to estimate the effect of time until regimen modification following first regimen failure on all-cause mortality. The impact of regimen type was also assessed. As the effect of delayed switching can be confounded if patients with a poor prognosis modify therapy earlier than those with a good prognosis, we used a statistical methodology – marginal structural models – to control for time-dependent confounding. Results:A total of 982 patients contributed 3414 person-years of follow-up following first regimen failure. Delay until treatment modification was associated with an elevated hazard of all-cause mortality among patients failing a reverse transcriptase inhibitor-based regimen (hazard ratio per additional 3 months delay = 1.23, 95% confidence interval: 1.08, 1.40), but appeared to have a small protective effect among patients failing a protease inhibitor-based regimen (hazard ratio per additional 3 months delay = 0.93, 95% confidence interval: 0.87, 0.99). Conclusion:Delay in modification after failure of regimens that do not contain a protease inhibitor is associated with increased mortality. Protease inhibitor-based regimens are less dependent on early versus delayed switching strategies. Efforts should be made to minimize delay until treatment modification in resource-poor regions, where the majority of patients are starting reverse transcriptase inhibitor-based regimens and HIV RNA monitoring may not be available.

Direct Effect Models

CONTEXT Depression strongly predicts nonadherence to human immunodeficiency virus (HIV) antiretroviral therapy, and adherence is essential to maintaining viral suppression. This suggests that pharmacologic treatment of depression may improve virologic outcomes. However, previous longitudinal observational analyses have inadequately adjusted for time-varying confounding by depression severity, which could yield biased estimates of treatment effect. Application of marginal structural modeling to longitudinal observation data can, under certain assumptions, approximate the findings of a randomized controlled trial. OBJECTIVE To determine whether antidepressant medication treatment increases the probability of HIV viral suppression. DESIGN Community-based prospective cohort study with assessments conducted every 3 months. SETTING Community-based research field site in San Francisco, California. PARTICIPANTS One hundred fifty-eight homeless and marginally housed persons with HIV who met baseline immunologic (CD4+ T-lymphocyte count, <350/μL) and psychiatric (Beck Depression Inventory II score, >13) inclusion criteria, observed from April 2002 through August 2007. MAIN OUTCOME MEASURES Probability of achieving viral suppression to less than 50 copies/mL. Secondary outcomes of interest were probability of being on an antiretroviral therapy regimen, 7-day self-reported percentage adherence to antiretroviral therapy, and probability of reporting complete (100%) adherence. RESULTS Marginal structural models estimated a 2.03 greater odds of achieving viral suppression (95% confidence interval [CI], 1.15-3.58; P = .02) resulting from antidepressant medication treatment. In addition, antidepressant medication use increased the probability of antiretroviral uptake (weighted odds ratio, 3.87; 95% CI, 1.98-7.58; P < .001). Self-reported adherence to antiretroviral therapy increased by 25 percentage points (95% CI, 14-36; P < .001), and the odds of reporting complete adherence nearly doubled (weighted odds ratio, 1.94; 95% CI, 1.20-3.13; P = .006). CONCLUSIONS Antidepressant medication treatment increases viral suppression among persons with HIV. This effect is likely attributable to improved adherence to a continuum of HIV care, including increased uptake and adherence to antiretroviral therapy.

Leveraging rapid community-based hiv testing campaigns for non-communicable diseases in rural uganda

The causal effect of a treatment on an outcome is generally mediated by several intermediate variables. Estimation of the component of the causal effect of a treatment that is not mediated by an intermediate variable (the direct effect of the treatment) is often relevant to mechanistic understanding and to the design of clinical and public health interventions. Robins, Greenland and Pearl develop counterfactual definitions for two types of direct effects, natural and controlled, and discuss assumptions, beyond those of sequential randomization, required for the identifiability of natural direct effects. Building on their earlier work and that of others, this article provides an alternative counterfactual definition of a natural direct effect, the identifiability of which is based only on the assumption of sequential randomization. In addition, a novel approach to direct effect estimation is presented, based on assuming a model directly on the natural direct effect, possibly conditional on a subset of the baseline covariates. Inverse probability of censoring weighted estimators, double robust inverse probability of censoring weighted estimators, likelihood-based estimators, and targeted maximum likelihood-based estimators are proposed for the unknown parameters of this novel causal model.

History-Adjusted Marginal Structural Models and Statically-Optimal Dynamic Treatment Regimens

Background The high burden of undiagnosed HIV in sub-Saharan Africa limits treatment and prevention efforts. Community-based HIV testing campaigns can address this challenge and provide an untapped opportunity to identify non-communicable diseases (NCDs). We tested the feasibility and diagnostic yield of integrating NCD and communicable diseases into a rapid HIV testing and referral campaign for all residents of a rural Ugandan parish. Methods A five-day, multi-disease campaign, offering diagnostic, preventive, treatment and referral services, was performed in May 2011. Services included point-of-care screening for HIV, malaria, TB, hypertension and diabetes. Finger-prick diagnostics eliminated the need for phlebotomy. HIV-infected adults met clinic staff and peer counselors on-site; those with CD4≤100/µL underwent intensive counseling and rapid referral for antiretroviral therapy (ART). Community participation, case-finding yield, and linkage to care three months post-campaign were analyzed. Results Of 6,300 residents, 2,323/3,150 (74%) adults and 2,020/3,150 (69%) children participated. An estimated 95% and 52% of adult female and male residents participated respectively. Adult HIV prevalence was 7.8%, with 46% of HIV-infected adults newly diagnosed. Thirty-nine percent of new HIV diagnoses linked to care. In a pilot subgroup with CD4≤100, 83% linked and started ART within 10 days. Malaria was identified in 10% of children, and hypertension and diabetes in 28% and 3.5% of adults screened, respectively. Sixty-five percent of hypertensives and 23% of diabetics were new diagnoses, of which 43% and 61% linked to care, respectively. Screening identified suspected TB in 87% of HIV-infected and 19% of HIV-uninfected adults; 52% percent of HIV-uninfected TB suspects linked to care. Conclusions In an integrated campaign engaging 74% of adult residents, we identified a high burden of undiagnosed HIV, hypertension and diabetes. Improving male attendance and optimizing linkage to care require new approaches. The campaign demonstrates the feasibility of integrating hypertension, diabetes and communicable diseases into HIV initiatives.