ABSTRACT
Etiologic heterogeneity occurs when distinct sets of events or exposures give rise to different subtypes of disease. Inference about subtype-specific exposure effects from two-phase outcome-dependent sampling data requires adjustment for both confounding and the sampling design. Common approaches to inference for these effects do not necessarily appropriately adjust for these sources of bias, or allow for formal comparisons of effects across different subtypes. Herein, using inverse probability weighting (IPW) to fit a multinomial model is shown to yield valid inference with this sampling design for subtype-specific exposure effects and contrasts thereof. The IPW approach is compared to common regression-based methods for assessing exposure effect heterogeneity using simulations. The methods are applied to estimate subtype-specific effects of various exposures on breast cancer risk in the Carolina Breast Cancer Study.
ABSTRACT
In the analysis of observational studies, inverse probability weighting (IPW) is commonly used to consistently estimate the average treatment effect (ATE) or the average treatment effect in the treated (ATT). The variance of the IPW ATE estimator is often estimated by assuming that the weights are known and then using the so-called "robust" (Huber-White) sandwich estimator, which results in conservative standard errors (SEs). Here we show that using such an approach when estimating the variance of the IPW ATT estimator does not necessarily result in conservative SE estimates. That is, assuming the weights are known, the robust sandwich estimator may be either conservative or anticonservative. Thus, confidence intervals for the ATT using the robust SE estimate will not be valid, in general. Instead, stacked estimating equations which account for the weight estimation can be used to compute a consistent, closed-form variance estimator for the IPW ATT estimator. The 2 variance estimators are compared via simulation studies and in a data analysis of the association between smoking and gene expression.
Subject(s)
Models, Statistical , Computer Simulation , Humans , ProbabilityABSTRACT
In observational genomics data sets, there is often confounding of the effect of an exposure on gene expression. To adjust for confounding when estimating the exposure effect, a common approach involves including potential confounders as covariates with the exposure in a regression model of gene expression. However, when the exposure and confounders interact to influence gene expression, the fitted regression model does not necessarily estimate the overall effect of the exposure. Using inverse probability weighting (IPW) or the parametric g-formula in these instances is straightforward to apply and yields consistent effect estimates. IPW can readily be integrated into a genomics data analysis pipeline with upstream data processing and normalization, while the g-formula can be implemented by making simple alterations to the regression model. The regression, IPW, and g-formula approaches to exposure effect estimation are compared herein using simulations; advantages and disadvantages of each approach are explored. The methods are applied to a case study estimating the effect of current smoking on gene expression in adipose tissue.
Subject(s)
Gene Expression Regulation , Cohort Studies , Computer Simulation , Confidence Intervals , Confounding Factors, Epidemiologic , Humans , Models, Genetic , Probability , Regression Analysis , Smoking/adverse effects , Smoking/geneticsABSTRACT
Accurate power calculations are essential in small studies containing expensive experimental units or high-stakes exposures. Herein, power of the Wilcoxon Mann-Whitney rank-sum test of a continuous outcome is formulated using a Monte Carlo approach and defining [Formula: see text] as a measure of effect size, where [Formula: see text] and [Formula: see text] denote random observations from two distributions hypothesized to be equal under the null. Effect size [Formula: see text] fosters productive communications because researchers understand [Formula: see text] is analogous to a fair coin toss, and [Formula: see text] near 0 or 1 represents a large effect. This approach is feasible even without background data. Simulations were conducted comparing the empirical power approach to existing approaches by Rosner & Glynn, Shieh and colleagues, Noether, and O'Brien-Castelloe. Approximations by Noether and O'Brien-Castelloe are shown to be inaccurate for small sample sizes. The Rosner & Glynn and Shieh, Jan & Randles approaches performed well in many small sample scenarios, though both are restricted to location-shift alternatives and neither approach is theoretically justified for small samples. The empirical method is recommended and available in the R package wmwpow.
Subject(s)
Research Design/statistics & numerical data , Animals , Computer Simulation , Data Interpretation, Statistical , Humans , Models, Statistical , Monte Carlo MethodABSTRACT
BACKGROUND: People who inject drugs (PWID) have a high incidence of HIV, little access to antiretroviral therapy (ART) and medication-assisted treatment (MAT), and high mortality. We aimed to assess the feasibility of a future controlled trial based on the incidence of HIV, enrolment, retention, and uptake of the intervention, and the efficacy of an integrated and flexible intervention on ART use, viral suppression, and MAT use. METHODS: This randomised, controlled vanguard study was run in Kyiv, Ukraine (one community site), Thai Nguyen, Vietnam (two district health centre sites), and Jakarta, Indonesia (one hospital site). PWID who were HIV infected (index participants) and non-infected injection partners were recruited as PWID network units and were eligible for screening if they were aged 18-45 years (updated to 18-60 years 8 months into study), and active injection drug users. Further eligibility criteria for index participants included a viral load of 1000 copies per mL or higher, willingness and ability to recruit at least one injection partner who would be willing to participate. Index participants were randomly assigned via a computer generated sequence accessed through a secure web portal (3:1) to standard of care or intervention, stratified by site. Masking of assignment was not possible due to the nature of intervention. The intervention comprised systems navigation, psychosocial counselling, and ART at any CD4 count. Local ART and MAT services were used. Participants were followed up for 12-24 months. The primary objective was to assess the feasibility of a future randomised controlled trial. To achieve this aim we looked at the following endpoints: HIV incidence among injection partners in the standard of care group, and enrolment and retention of HIV-infected PWID and their injection partners and the uptake of the integrated intervention. The study was also designed to assess the feasibility, barriers, and uptake of the integrated intervention. Endpoints were assessed in a modified intention-to-treat popualtion after exclusion of ineligible participants. This trial is registered on ClinicalTrials.gov, NCT02935296, and is active but not recruiting new participants. FINDINGS: Between Feb 5, 2015, and June 3, 2016, 3343 potential index participants were screened, of whom 502 (15%) were eligible and enrolled. 1171 injection partners were referred, and 806 (69%) were eligible and enrolled. Index participants were randomly assigned to intervention (126 [25%]) and standard of care (376 [75%]) groups. At week 52, most living index participants (389 [86%] of 451) and partners (567 [80%] of 710) were retained, and self-reported ART use was higher among index participants in the intervention group than those in the standard of care group (probability ratio [PR] 1·7, 95% CI 1·4-1·9). Viral suppression was also higher in the intervention group than in the standard of care group (PR 1·7, 95% CI 1·3-2·2). Index participants in the intervention group reported more MAT use at 52 weeks than those in the standard of care group (PR 1·7, 95% CI 1·3-2·2). Seven incident HIV infections occurred, and all in injection partners in the standard of care group (intervention incidence 0·0 per 100 person-years, 95% CI 0·0-1·7; standard of care incidence 1·0 per 100 person-years, 95% CI 0·4-2·1; incidence rate difference -1·0 per 100 person-years, 95% CI -2·1 to 1·1). No severe adverse events due to the intervention were recorded. INTERPRETATION: This vanguard study provides evidence that a flexible, scalable intervention increases ART and MAT use and reduces mortality among PWID. The low incidence of HIV in both groups impedes a future randomised, controlled trial, but given the strength of the effect of the intervention, its implementation among HIV-infected PWID should be considered. FUNDING: US National Institutes of Health.
Subject(s)
Antiretroviral Therapy, Highly Active , HIV Infections/drug therapy , Opiate Substitution Treatment/methods , Substance Abuse, Intravenous/drug therapy , Viral Load/drug effects , Adult , CD4 Lymphocyte Count , Counseling , Feasibility Studies , Female , HIV Infections/complications , HIV Infections/mortality , Humans , Incidence , Indonesia , Male , Methadone/therapeutic use , Proportional Hazards Models , Substance Abuse, Intravenous/complications , Substance Abuse, Intravenous/mortality , Ukraine , Vietnam , Young AdultABSTRACT
BACKGROUND: Human immunodeficiency virus self-testing (HIVST) holds great promise for reaching high-risk key populations who do not access facility-based services. We sought to characterize unsupervised HIVST implementation among men who have sex with men in China. METHODS: We conducted a nationwide online survey in China. Eligible men were at least 16 years, had anal sex with a man, and had recent condomless sex. We assessed benefits (first-time testing, increased testing frequency, confirmatory testing) and potential harms (coercion, violence, suicidality) of HIVST. Among men who have sex with men who reported ever testing for human immunodeficiency virus (HIV), we identified correlates of HIVST as first-time HIV test being a self-test using multivariable logistic regression. RESULTS: Among 1610 men who met the eligibility criteria and started the survey, 1189 (74%) completed it. Three hundred forty-one (29%) of 1189 reported ever self-testing for HIV. Human immunodeficiency virus prevalence was 7% (24/341) among self-testers and 5% (15/306) among non-self-testers. Two hundred (59%) of 341 men who self-tested reported HIVST as a first-time HIV test. Thirty-one (9%) men experienced coercion with HIVST. Thirty-one (78%) of 40 men with positive HIV self-tests sought confirmation. Multivariable analysis revealed that HIVST as first-time HIV test was associated with younger age (odds ratio [OR], 0.95; 95% confidence interval [CI], 0.92-0.99), not being "out" (OR, 2.28; 95% CI, 1.60-3.28), not using the internet to meet sex partners (OR, 0.39; 95% CI, 0.22-0.69), and group sex (OR, 1.74; 95% CI, 1.02-2.9). CONCLUSIONS: Human immunodeficiency virus self-testing reached high-risk individuals that had never received facility-based testing. Further implementation research is needed to better understand HIVST outside of research programs.
Subject(s)
HIV Infections/diagnosis , Mass Screening/statistics & numerical data , Self-Examination/statistics & numerical data , Sexual and Gender Minorities/statistics & numerical data , Adult , China/epidemiology , HIV Infections/epidemiology , Health Plan Implementation , Humans , Logistic Models , Male , Mass Screening/methods , Multivariate Analysis , Prevalence , Self-Examination/methods , Sexual Behavior , Surveys and Questionnaires , Unsafe Sex , Young AdultABSTRACT
OBJECTIVE: Vietnam, Indonesia, and Ukraine have major burdens of IDU and HIV. We estimated the prevalence of depressive symptoms at baseline among people living with HIV who inject drugs, evaluated associations between depression at baseline and 12-month HIV care outcomes and medication-assisted treatment (MAT), and evaluated the study intervention effect by baseline depression subgroups. DESIGN: HPTN 074 was a randomized study. The study intervention included psychosocial counseling, systems navigation, and antiretroviral treatment (ART) at any CD4+ cell count. METHODS: Moderate-to-severe depression was defined as a Patient Health Questionnaire-9 score of 10 or above. ART and MAT were self-reported. Eligibility criteria were: 18-60 years of age, active IDU, and viral load of at least 1000 copies/ml. Adjusted probability differences (aPD) were estimated using inverse-probability weighting. RESULTS: A total of 502 participants enrolled from April 2015 to June 2016. Median age was 35 years; 85% identified as men. Prevalence of baseline moderate-to-severe depression was 14% in Vietnam, 14% in Indonesia, and 56% in Ukraine. No evident associations were detected between baseline depression and ART, viral suppression, or MAT at 12-month follow-up. The study intervention improved the proportions of people who inject drugs achieving 12-month viral suppression in both the depressed [intervention 44%; standard of care 24%; estimated aPDâ=â25% (95% confidence interval: 4.0%, 45%)] and nondepressed subgroups [intervention 38%; standard of care 24%; aPDâ=â13% (95% confidence interval: 2.0%, 25%)]. CONCLUSION: High levels of depressive symptoms were common among people living with HIV who inject drugs in Ukraine but were less common in Vietnam and Indonesia. The study intervention was effective among participants with or without baseline depression symptoms.
Subject(s)
HIV Infections , Pharmaceutical Preparations , Adult , Depression/epidemiology , HIV Infections/complications , HIV Infections/drug therapy , Humans , Indonesia/epidemiology , Male , Ukraine , Vietnam/epidemiology , Viral LoadABSTRACT
BACKGROUND: People who inject drugs (PWID) living with HIV experience inadequate access to antiretroviral treatment (ART) and medication for opioid use disorders (MOUD). HPTN 074 showed that an integrated intervention increased ART use and viral suppression over 52 weeks. To examine durability of ART, MOUD, and HIV viral suppression, participants could re-enroll for an extended follow-up period, during which standard-of-care (SOC) participants in need of support were offered the intervention. METHODS: Participants were recruited from Ukraine, Indonesia and Vietnam and randomly allocated 3:1 to SOC or intervention. Eligibility criteria included: HIV-positive; active injection drug use; 18-60 years of age; ≥1 HIV-uninfected injection partner; and viral load ≥1,000 copies/mL. Re-enrollment was offered to all available intervention and SOC arm participants, and SOC participants in need of support (off-ART or off-MOUD) were offered the intervention. RESULTS: The intervention continuation group re-enrolled 89 participants, and from week 52 to 104, viral suppression (<40 copies/mL) declined from 41% to 29% (estimated 9.4% decrease per year, 95% CI -17.0%; -1.8%). The in need of support group re-enrolled 94 participants and had increased ART (re-enrollment: 55%, week 26: 69%) and MOUD (re-enrollment: 16%, week 26: 25%) use, and viral suppression (re-enrollment: 40%, week 26: 49%). CONCLUSIONS: Viral suppression declined in year 2 for those who initially received the HPTN 074 intervention and improved maintenance support is warranted. Viral suppression and MOUD increased among in need participants who received intervention during the study extension. Continued efforts are needed for widespread implementation of this scalable, integrated intervention.
ABSTRACT
Approaches to deplete persistent HIV infection are needed. We investigated the combined impact of the latency reversing agent vorinostat (VOR) and AGS-004, an autologous dendritic cell immunotherapeutic, on the HIV reservoir. HIV+, stably treated participants in whom resting CD4+ T cell-associated HIV RNA (rca-RNA) increased after VOR exposure ex vivo and in vivo received 4 doses of AGS-004 every 3 weeks, followed by VOR every 72 hours for 30 days, and then the cycle repeated. Change in VOR-responsive host gene expression, HIV-specific T cell responses, low-level HIV viremia, rca-RNA, and the frequency of resting CD4+ T-cell infection (RCI) was measured at baseline and after each cycle. No serious treatment-related adverse events were observed among five participants. As predicted, VOR-responsive host genes responded uniformly to VOR dosing. Following cycles of AGS-004 and VOR, rca-RNA decreased significantly in only two participants, with a significant decrease in SCA observed in one of these participants. However, unlike other cohorts dosed with AGS-004, no uniform increase in HIV-specific immune responses following vaccination was observed. Finally, no reproducible decline of RCI, defined as a decrease of >50%, was observed. AGS-004 and VOR were safe and well-tolerated, but no substantial impact on RCI was measured. In contrast to previous clinical data, AGS-004 did not induce HIV-specific immune responses greater than those measured at baseline. More efficacious antiviral immune interventions, perhaps paired with more effective latency reversal, must be developed to clear persistent HIV infection.
Subject(s)
Dendritic Cells/transplantation , HIV Infections/therapy , HIV-1/drug effects , Histone Deacetylase Inhibitors/therapeutic use , Immunotherapy, Adoptive/methods , Vorinostat/therapeutic use , Adult , CD4-Positive T-Lymphocytes/immunology , Humans , Immunologic Memory/immunology , Male , Middle Aged , T-Lymphocytes, Cytotoxic/immunology , Translational Research, Biomedical , Treatment Outcome , VaccinationABSTRACT
INTRODUCTION: People who inject drugs (PWID) experience high HIV incidence and face significant barriers to engagement in HIV care and substance use treatment. Strategies for HIV treatment as prevention and substance use treatment present unique challenges in PWID that may vary regionally. Understanding differences in the risk structure for HIV transmission and disease progression among PWID is essential in developing and effectively targeting intervention strategies of HIV treatment as prevention. METHODS: We present a baseline analysis of HIV Prevention Trials Network (HPTN) 074, a two-arm, randomized controlled trial among PWID in Indonesia (n = 258), Ukraine (n = 457) and Vietnam (n = 439). HPTN 074 was designed to determine the feasibility, barriers and uptake of an integrated intervention combining health systems navigation and psychosocial counselling for the early engagement of antiretroviral therapy (ART) and substance use treatment for PWID living with HIV. Discordant PWID networks were enrolled, consisting of an HIV-positive index and their HIV-negative network injection partner(s). Among the enrolled cohort of 1154 participants (502 index participants and 652 network partners), we examine regional differences in the baseline risk structure, including sociodemographics, HIV and substance use treatment history, and injection and sexual risk behaviours. RESULTS: The majority of participants were male (87%), with 82% of the enrolled females coming from Ukraine. The overall mean age was 34 (IQR: 30, 38). Most commonly injected substances included illegally manufactured methadone in Ukraine (84.2%), and heroin in Indonesia (81.8%) and Vietnam (99.5%). Injection network sizes varied by region: median number of people with whom participants self-reported injecting drugs was 3 (IQR: 2, 5) in Indonesia, 5 (IQR: 3, 10) in Ukraine and 3 (IQR: 2, 4) in Vietnam. Hazardous alcohol use, assessed using the Alcohol Use Disorders Identification Test - Alcohol Consumption Questions (AUDIT-C), was prominent in Ukraine (54.7%) and Vietnam (26.4%). Reported sexual risk behaviours in the past month, including having two or more sex partners and giving/receiving money or drugs in exchange for sex, were uncommon among all participants and regions. CONCLUSIONS: While regional differences in risk structure exist, PWID particularly in Ukraine need immediate attention for risk reduction strategies. Substantial regional differences in risk structure will require flexible, tailored treatment as prevention interventions for distinct PWID populations.