Randomized Trial of Dynamic Choice HIV Prevention at Antenatal and Postnatal Care Clinics in Rural Uganda and Kenya.

BACKGROUND: Pregnant and postpartum women in Sub-Saharan Africa are at high risk of HIV acquisition. We evaluated a person-centered dynamic choice intervention for HIV prevention (DCP) among women attending antenatal and postnatal care. SETTING: Rural Kenya and Uganda. METHODS: Women (aged 15 years or older) at risk of HIV acquisition seen at antenatal and postnatal care clinics were individually randomized to DCP vs. standard of care (SEARCH; NCT04810650). The DCP intervention included structured client choice of product (daily oral pre-exposure prophylaxis or postexposure prophylaxis), service location (clinic or out of facility), and HIV testing modality (self-test or provider-administered), with option to switch over time and person-centered care (phone access to clinician, structured barrier assessment and counseling, and provider training). The primary outcome was biomedical prevention coverage-proportion of 48-week follow-up with self-reported pre-exposure prophylaxis or postexposure prophylaxis use, compared between arms using targeted maximum likelihood estimation. RESULTS: Between April and July 2021, we enrolled 400 women (203 intervention and 197 control); 38% were pregnant, 52% were aged 15-24 years, and 94% reported no pre-exposure prophylaxis or postexposure prophylaxis use for ≥6 months before baseline. Among 384/400 participants (96%) with outcome ascertained, DCP increased biomedical prevention coverage 40% (95% CI: 34% to 47%; P < 0.001); the coverage was 70% in intervention vs. 29% in control. DCP also increased coverage during months at risk of HIV (81% in intervention, 43% in control; 38% absolute increase; 95% CI: 31% to 45%; P < 0.001). CONCLUSION: A person-centered dynamic choice intervention that provided flexibility in product, testing, and service location more than doubled biomedical HIV prevention coverage in a high-risk population already routinely offered access to biomedical prevention options.

Adaptive selection of the optimal strategy to improve precision and power in randomized trials.

Benkeser et al. demonstrate how adjustment for baseline covariates in randomized trials can meaningfully improve precision for a variety of outcome types. Their findings build on a long history, starting in 1932 with R.A. Fisher and including more recent endorsements by the U.S. Food and Drug Administration and the European Medicines Agency. Here, we address an important practical consideration: how to select the adjustment approach-which variables and in which form-to maximize precision, while maintaining Type-I error control. Balzer et al. previously proposed Adaptive Pre-specification within TMLE to flexibly and automatically select, from a prespecified set, the approach that maximizes empirical efficiency in small trials (N < 40). To avoid overfitting with few randomized units, selection was previously limited to working generalized linear models, adjusting for a single covariate. Now, we tailor Adaptive Pre-specification to trials with many randomized units. Using V-fold cross-validation and the estimated influence curve-squared as the loss function, we select from an expanded set of candidates, including modern machine learning methods adjusting for multiple covariates. As assessed in simulations exploring a variety of data-generating processes, our approach maintains Type-I error control (under the null) and offers substantial gains in precision-equivalent to 20%-43% reductions in sample size for the same statistical power. When applied to real data from ACTG Study 175, we also see meaningful efficiency improvements overall and within subgroups.

Community-Wide Universal Human Immunodeficiency Virus (HIV) Test and Treat Intervention Reduces Tuberculosis Transmission in Rural Uganda: A Cluster-Randomized Trial.

BACKGROUND: Human immunodeficiency virus (HIV) treatment reduces tuberculosis (TB) disease and mortality; however, the population-level impact of universal HIV-test-and-treat interventions on TB infection and transmission remain unclear. METHODS: In a sub-study nested in the SEARCH trial, a community cluster-randomized trial (NCT01864603), we assessed whether a universal HIV-test-and-treat intervention reduced population-level incident TB infection in rural Uganda. Intervention communities received annual, population-level HIV testing and patient-centered linkage. Control communities received population-level HIV testing at baseline and endline. We compared estimated incident TB infection by arms, defined by tuberculin skin test conversion in a cohort of persons aged 5 and older, adjusting for participation and predictors of infection, and accounting for clustering. RESULTS: Of the 32 trial communities, 9 were included, comprising 90 801 participants (43 127 intervention and 47 674 control). One-year cumulative incidence of TB infection was 16% in the intervention and 22% in the control; SEARCH reduced the population-level risk of incident TB infection by 27% (adjusted risk ratio = 0.73; 95% confidence interval [CI]: .57-.92, P = .005). In pre-specified analyses, the effect was largest among children aged 5-11 years and males. CONCLUSIONS: A universal HIV-test-and-treat intervention reduced incident TB infection, a marker of population-level TB transmission. Investments in community-level HIV interventions have broader population-level benefits, including TB reductions.

Dynamic choice HIV prevention intervention at outpatient departments in rural Kenya and Uganda.

OBJECTIVE: HIV prevention service delivery models that offer product choices, and the option to change preferences over time, may increase prevention coverage. Outpatient departments in sub-Saharan Africa diagnose a high proportion of new HIV infections, but are an understudied entry point to biomedical prevention. DESIGN: Individually randomized trial of dynamic choice HIV prevention (DCP) intervention vs. standard-of-care (SOC) among individuals with current/anticipated HIV exposure risk at outpatient departments in rural Kenya and Uganda (SEARCH; NCT04810650). METHODS: Our DCP intervention included 1) product choice (oral preexposure prophylaxis [PrEP] or postexposure prophylaxis [PEP]) with an option to switch over time, 2) HIV provider- or self-testing, 3) service location choice (community vs. clinic-based), and 4) provider training on patient-centered care. Primary outcome was proportion of follow-up covered by PrEP/PEP over 48 weeks assessed via self-report. RESULTS: We enrolled 403 participants (61% women; median 27 years, IQR 22-37). In the DCP arm, 86% ever chose PrEP, 15% ever chose PEP over 48 weeks; selection of HIV self-testing increased from 26 to 51% and of out-of-facility visits from 8 to 52%. Among 376 of 403 (93%) with outcomes ascertained, time covered by PrEP/PEP was higher in DCP (47.5%) vs. SOC (18.3%); difference = 29.2% (95% confidence interval: 22.7-35.7; P  < 0.001). Effects were similar among women and men (28.2 and 31.0% higher coverage in DCP, respectively) and larger during periods of self-reported HIV risk (DCP 64.9% vs. SOC 26.3%; difference = 38.6%; 95% confidence interval: 31.0-46.2; P  < 0.001). CONCLUSION: A dynamic choice HIV prevention intervention resulted in two-fold greater time covered by biomedical prevention products compared to SOC in general outpatient departments in eastern Africa.

Randomized Trial of a "Dynamic Choice" Patient-Centered Care Intervention for Mobile Persons With HIV in East Africa.

BACKGROUND: Persons with HIV (PWH) with high mobility face obstacles to HIV care engagement and viral suppression. We sought to understand whether a patient-centered intervention for mobile PWH would improve viral suppression and retention in care, and if so, which subgroups would benefit most. METHODS: In a randomized trial, we evaluated the effect of an intervention designed to address barriers to care among mobile (≥2 weeks out of community in previous year) PWH with viral nonsuppression or recent missed visits in Kenya and Uganda (NCT04810650). The intervention included dynamic choice of a "travel pack" (emergency antiretroviral therapy [ART] supply, discrete ART packaging, and travel checklist), multimonth and offsite refills, facilitated transfer to out-of-community clinics, and hotline access to a mobility coordinator. The primary outcome was viral suppression (<400 copies/mL) at 48 weeks. Secondary outcomes included retention in care and ART possession. RESULTS: From April 2021 to July 2022, 201 participants were enrolled and randomized (102 intervention, 99 control): 109 (54%) were female participants and 101 (50%) from Kenya; median age was 37 years (interquartile range: 29-43). At 48 weeks, there was no significant difference in viral suppression in intervention (85%) vs. control (86%). The intervention improved retention in care (risk ratio: 1.06[1.02-1.1]; P < 0.001) and ART possession (risk ratio: 1.07[1.03-1.11]; P < 0.001), with larger effect sizes among persons with baseline nonsuppression and high mobility (≥2 weeks out of community in previous 3 months). CONCLUSIONS: Mobile PWH-centered care should be considered for high-risk mobile populations, including nonsuppressed and highly mobile PWH, to improve retention in care and sustain viral suppression over time. TRIAL REGISTRATION: NCT04810650.

A community-based dynamic choice model for HIV prevention improves PrEP and PEP coverage in rural Uganda and Kenya: a cluster randomized trial.

INTRODUCTION: Optimizing HIV prevention may require structured approaches for providing client-centred choices as well as community-based entry points and delivery. We evaluated the effect of a dynamic choice model for HIV prevention, delivered by community health workers (CHWs) with clinician support, on the use of biomedical prevention among persons at risk of HIV in rural East Africa. METHODS: We conducted a cluster randomized trial among persons (≥15 years) with current or anticipated HIV risk in 16 villages in Uganda and Kenya (SEARCH; NCT04810650). The intervention was a client-centred HIV prevention model, including (1) structured client choice of product (pre-exposure prophylaxis [PrEP] or post-exposure prophylaxis [PEP]), service location (clinic or out-of-clinic) and HIV testing modality (self-test or rapid test), with the ability to switch over time; (2) a structured assessment of patient barriers and development of a personalized support plan; and (3) phone access to a clinician 24/7. The intervention was delivered by CHWs and supported by clinicians who oversaw PrEP and PEP initiation and monitoring. Participants in control villages were referred to local health facilities for HIV prevention services, delivered by Ministry of Health staff. The primary outcome was biomedical prevention coverage: a proportion of 48-week follow-up with self-reported PrEP or PEP use. RESULTS: From May to July 2021, we enrolled 429 people (212 intervention; 217 control): 57% women and 35% aged 15-24 years. Among intervention participants, 58% chose PrEP and 58% chose PEP at least once over follow-up; self-testing increased from 52% (baseline) to 71% (week 48); ≥98% chose out-of-facility service delivery. Among 413 (96%) participants with the primary outcome ascertained, average biomedical prevention coverage was 28.0% in the intervention versus 0.5% in the control: a difference of 27.5% (95% CI: 23.0-31.9%, p<0.001). Impact was larger during periods of self-reported HIV risk: 36.6% coverage in intervention versus 0.9% in control, a difference of 35.7% (95% CI: 27.5-43.9, p<0.001). Intervention effects were seen across subgroups defined by sex, age group and alcohol use. CONCLUSIONS: A client-centred dynamic choice HIV prevention intervention, including the option to switch between products and CHW-based delivery in the community, increased biomedical prevention coverage by 27.5%. However, substantial person-time at risk of HIV remained uncovered.

Effect of a brief alcohol counselling intervention on HIV viral suppression and alcohol use among persons with HIV and unhealthy alcohol use in Uganda and Kenya: a randomized controlled trial.

INTRODUCTION: Unhealthy alcohol use significantly contributes to viral non-suppression among persons with HIV (PWH). It is unknown whether brief behavioural interventions to reduce alcohol use can improve viral suppression among PWH with unhealthy alcohol use in sub-Saharan Africa (SSA). METHODS: As part of the SEARCH study (NCT04810650), we conducted an individually randomized trial in Kenya and Uganda of a brief, skills-based alcohol intervention among PWH with self-reported unhealthy alcohol use (Alcohol Use Disorders Identification Test-Consumption [AUDIT-C], prior 3 months, ≥3/female; ≥4/male) and at risk of viral non-suppression, defined as either recent HIV viral non-suppression (≥400 copies/ml), missed visits, out of care or new diagnosis. The intervention included baseline and 3-month in-person counselling sessions with interim booster phone calls every 3 weeks. The primary outcome was HIV viral suppression (<400 copies/ml) at 24 weeks, and the secondary outcome was unhealthy alcohol use, defined by AUDIT-C or phosphatidylethanol (PEth), an alcohol biomarker, ≥50 ng/ml at 24 weeks. RESULTS: Between April and September 2021, 401 persons (198 intervention, 203 control) were enrolled from HIV clinics in Uganda (58%) and Kenya (27%) and alcohol-serving venues in Kenya (15%). At baseline, 60% were virally suppressed. Viral suppression did not differ between arms at 24 weeks: suppression was 83% in intervention and 82% in control arms (RR: 1.01, 95% CI: 0.93-1.1). Among PWH with baseline viral non-suppression, 24-week suppression was 73% in intervention and 64% in control arms (RR 1.15, 95% CI: 0.93-1.43). Unhealthy alcohol use declined from 98% at baseline to 73% in intervention and 84% in control arms at 24 weeks (RR: 0.86, 95% CI: 0.79-0.94). Effects on unhealthy alcohol use were stronger among women (RR 0.70, 95% CI: 0.56-0.88) than men (RR 0.93, 95% CI: 0.85-1.01) and among participants with a baseline PEth⩽200 ng/ml (RR 0.68, 95% CI: 0.53-0.87) versus >200 ng/ml (RR 0.97, 95% CI: 0.92-1.02). CONCLUSIONS: In a randomized trial of 401 PWH with unhealthy alcohol use and risk for viral non-suppression, a brief alcohol intervention reduced unhealthy alcohol use but did not affect viral suppression at 24 weeks. Brief alcohol interventions have the potential to improve the health of PWH in SSA by reducing alcohol use, a significant driver of HIV-associated co-morbidities.

A multilevel health system intervention for virological suppression in adolescents and young adults living with HIV in rural Kenya and Uganda (SEARCH-Youth): a cluster randomised trial.

BACKGROUND: Social and cognitive developmental events can disrupt care and medication adherence among adolescents and young adults living with HIV in sub-Saharan Africa. We hypothesised that a dynamic multilevel health system intervention helping adolescents and young adults and their providers navigate life-stage related events would increase virological suppression compared with standard care. METHODS: We did a cluster randomised, open-label trial of young individuals aged 15-24 years with HIV and receiving care in eligible clinics (operated by the government and with ≥25 young people receiving care) in rural Kenya and Uganda. After clinic randomisation stratified by region, patient population, and previous participation in the SEARCH trial, participants in intervention clinics received life-stage-based assessment at routine visits, flexible clinic access, and rapid viral load feedback. Providers had a secure mobile platform for interprovider consultation. The control clinics followed standard practice. The primary, prespecified endpoint was virological suppression (HIV RNA <400 copies per mL) at 2 years of follow-up among participants who enrolled before Dec 1, 2019, and received care at the study clinics. This trial is registered with ClinicalTrials.gov, NCT03848728, and is closed to recruitment. FINDINGS: 28 clinics were enrolled and randomly assigned (14 control, 14 intervention) in January, 2019. Between March 14, 2019, and Nov 26, 2020, we recruited 1988 participants at the clinics, of whom 1549 were included in the analysis (785 at intervention clinics and 764 at control clinics). The median participant age was 21 years (IQR 19-23) and 1248 (80·6%) of 1549 participants were female. The mean proportion of participants with virological suppression at 2 years was 88% (95% CI 85-92) for participants in intervention clinics and 80% (77-84) for participants in control clinics, equivalent to a 10% beneficial effect of the intervention (risk ratio [RR] 1·10, 95% CI 1·03-1·16; p=0·0019). The intervention resulted in increased virological suppression within all subgroups of sex, age, and care status at baseline, with greatest improvement among those re-engaging in care (RR 1·60, 95% CI 1·00-2·55; p=0·025). INTERPRETATION: Routine and systematic life-stage-based assessment, prompt adherence support with rapid viral load testing, and patient-centred, flexible clinic access could help bring adolescents and young adults living with HIV closer towards a goal of universal virological suppression. FUNDING: Eunice Kennedy Shriver National Institute of Child Health and Human Development, US National Institutes of Health.

Blurring cluster randomized trials and observational studies: Two-Stage TMLE for subsampling, missingness, and few independent units.

Cluster randomized trials (CRTs) often enroll large numbers of participants; yet due to resource constraints, only a subset of participants may be selected for outcome assessment, and those sampled may not be representative of all cluster members. Missing data also present a challenge: if sampled individuals with measured outcomes are dissimilar from those with missing outcomes, unadjusted estimates of arm-specific endpoints and the intervention effect may be biased. Further, CRTs often enroll and randomize few clusters, limiting statistical power and raising concerns about finite sample performance. Motivated by SEARCH-TB, a CRT aimed at reducing incident tuberculosis infection, we demonstrate interlocking methods to handle these challenges. First, we extend Two-Stage targeted minimum loss-based estimation to account for three sources of missingness: (i) subsampling; (ii) measurement of baseline status among those sampled; and (iii) measurement of final status among those in the incidence cohort (persons known to be at risk at baseline). Second, we critically evaluate the assumptions under which subunits of the cluster can be considered the conditionally independent unit, improving precision and statistical power but also causing the CRT to behave like an observational study. Our application to SEARCH-TB highlights the real-world impact of different assumptions on measurement and dependence; estimates relying on unrealistic assumptions suggested the intervention increased the incidence of TB infection by 18% (risk ratio [RR]=1.18, 95% confidence interval [CI]: 0.85-1.63), while estimates accounting for the sampling scheme, missingness, and within community dependence found the intervention decreased the incident TB by 27% (RR=0.73, 95% CI: 0.57-0.92).

Uptake of a patient-centred dynamic choice model for HIV prevention in rural Kenya and Uganda: SEARCH SAPPHIRE study.

INTRODUCTION: Person-centred HIV prevention delivery models that offer structured choices in product, testing and visit location may increase coverage. However, data are lacking on the actual uptake of choices among persons at risk of HIV in southern Africa. In an ongoing randomized study (SEARCH; NCT04810650) in rural East Africa, we evaluated the uptake of choices made when offered in a person-centred, dynamic choice model for HIV prevention. METHODS: Using the PRECEDE framework, we developed a persont-centred, Dynamic Choice HIV Prevention (DCP) intervention for persons at risk of HIV in three settings in rural Kenya and Uganda: antenatal clinic (ANC), outpatient department (OPD) and in the community. Components include: provider training on product choice (predisposing); flexibility and responsiveness to client desires and choices (pre-exposure prophylaxis [PrEP]/post-exposure prophylaxis [PEP], clinic vs. off-site visits and self- or clinician-based HIV testing) (enabling); and client and staff feedback (reinforcing). All clients received a structured assessment of barriers with personalized plans to address them, mobile phone access to clinicians (24 hours/7 days/week) and integrated reproductive health services. In this interim analysis, we describe the uptake of choices of product, location and testing during the first 24 weeks of follow-up (April 2021-March 2022). RESULTS: A total of 612 (203 ANC, 197 OPD and 212 community) participants were randomized to the person-centred DCP intervention. We delivered the DCP intervention in all three settings with diverse populations: ANC: 39% pregnant; median age: 24 years; OPD: 39% male, median age 27 years; and community: 42% male, median age: 29 years. Baseline choice of PrEP was highest in ANC (98%) vs. OPD (84%) and community (40%); whereas the proportion of adults selecting PEP was higher in the community (46%) vs. OPD (8%) and ANC (1%). Personal preference for off-site visits increased over time (65% at week 24 vs. 35% at baseline). Interest in alternative HIV testing modalities grew over time (38% baseline self-testing vs. 58% at week 24). CONCLUSIONS: A person-centred model incorporating structured choice in biomedical prevention and care delivery options in settings with demographically diverse groups, in rural Kenya and Uganda, was responsive to varying personal preferences over time in HIV prevention programmes.

Population-level viremia predicts HIV incidence at the community level across the Universal Testing and Treatment Trials in eastern and southern Africa.

Universal HIV testing and treatment (UTT) strategies aim to optimize population-level benefits of antiretroviral treatment. Between 2012 and 2018, four large community randomized trials were conducted in eastern and southern Africa. While their results were broadly consistent showing decreased population-level viremia reduces HIV incidence, it remains unclear how much HIV incidence can be reduced by increasing suppression among people living with HIV (PLHIV). We conducted a pooled analysis across the four UTT trials. Leveraging data from 105 communities in five countries, we evaluated the linear relationship between i) population-level viremia (prevalence of non-suppression-defined as plasma HIV RNA >500 or >400 copies/mL-among all adults, irrespective of HIV status) and HIV incidence; and ii) prevalence of non-suppression among PLHIV and HIV incidence, using parametric g-computation. HIV prevalence, measured in 257 929 persons, varied from 2 to 41% across the communities; prevalence of non-suppression among PLHIV, measured in 31 377 persons, from 3 to 70%; population-level viremia, derived from HIV prevalence and non-suppression, from < 1% to 25%; and HIV incidence, measured over 345 844 person-years (PY), from 0.03/100PY to 3.46/100PY. Decreases in population-level viremia were strongly associated with decreased HIV incidence in all trials (between 0.45/100PY and 1.88/100PY decline in HIV incidence per 10 percentage points decline in viremia). Decreases in non-suppression among PLHIV were also associated with decreased HIV incidence in all trials (between 0.06/100PY and 0.17/100PY decline in HIV incidence per 10 percentage points decline in non-suppression). Our results support both the utility of population-level viremia as a predictor of incidence, and thus a tool for targeting prevention interventions, and the ability of UTT approaches to reduce HIV incidence by increasing viral suppression. Implementation of universal HIV testing approaches, coupled with interventions to leverage linkage to treatment, adapted to local contexts, can reduce HIV acquisition at population level.

Defining and estimating effects in cluster randomized trials: A methods comparison.

Across research disciplines, cluster randomized trials (CRTs) are commonly implemented to evaluate interventions delivered to groups of participants, such as communities and clinics. Despite advances in the design and analysis of CRTs, several challenges remain. First, there are many possible ways to specify the causal effect of interest (eg, at the individual-level or at the cluster-level). Second, the theoretical and practical performance of common methods for CRT analysis remain poorly understood. Here, we present a general framework to formally define an array of causal effects in terms of summary measures of counterfactual outcomes. Next, we provide a comprehensive overview of CRT estimators, including the t-test, generalized estimating equations (GEE), augmented-GEE, and targeted maximum likelihood estimation (TMLE). Using finite sample simulations, we illustrate the practical performance of these estimators for different causal effects and when, as commonly occurs, there are limited numbers of clusters of different sizes. Finally, our application to data from the Preterm Birth Initiative (PTBi) study demonstrates the real-world impact of varying cluster sizes and targeting effects at the cluster-level or at the individual-level. Specifically, the relative effect of the PTBi intervention was 0.81 at the cluster-level, corresponding to a 19% reduction in outcome incidence, and was 0.66 at the individual-level, corresponding to a 34% reduction in outcome risk. Given its flexibility to estimate a variety of user-specified effects and ability to adaptively adjust for covariates for precision gains while maintaining Type-I error control, we conclude TMLE is a promising tool for CRT analysis.

Universal HIV Testing and Treatment With Patient-Centered Care Improves ART Uptake and Viral Suppression Among Adults Reporting Hazardous Alcohol Use in Uganda and Kenya.

OBJECTIVES: Determine whether patient-centered, streamlined HIV care achieves higher antiretroviral therapy (ART) uptake and viral suppression than the standard treatment model for people with HIV (PWH) reporting hazardous alcohol use. DESIGN: Community cluster-randomized trial. METHODS: The Sustainable East Africa Research in Community Health trial (NCT01864603) compared an intervention of annual population HIV testing, universal ART, and patient-centered care with a control of baseline population testing with ART by country standard in 32 Kenyan and Ugandan communities. Adults (15 years or older) completed a baseline Alcohol Use Disorders Identification Test-Consumption (AUDIT-C) and were classified as no/nonhazardous (AUDIT-C 0-2 women/0-3 men) or hazardous alcohol use (≥3 women/≥4 men). We compared year 3 ART uptake and viral suppression of PWH reporting hazardous use between intervention and control arms. We compared alcohol use as a predictor of year 3 ART uptake and viral suppression among PWH, by arm. RESULTS: Of 11,070 PWH with AUDIT-C measured, 1723 (16%) reported any alcohol use and 893 (8%) reported hazardous use. Among PWH reporting hazardous use, the intervention arm had higher ART uptake (96%) and suppression (87%) compared with control (74%, adjusted risk ratio [aRR] = 1.28, 95% CI: 1.19 to 1.38; and 72%, aRR = 1.20, 95% CI: 1.10 to 1.31, respectively). Within arm, hazardous alcohol use predicted lower ART uptake in control (aRR = 0.86, 95% CI: 0.78 to 0.96), but not intervention (aRR = 1.02, 95% CI: 1.00 to 1.04); use was not predictive of suppression in either arm. CONCLUSIONS: The Sustainable East Africa Research in Community Health intervention improved ART uptake and viral suppression among PWH reporting hazardous alcohol use and eliminated gaps in ART uptake between PWH with hazardous and no/nonhazardous use. Patient-centered HIV care may decrease barriers to HIV care for PWH with hazardous alcohol use.

SpiderLearner: An ensemble approach to Gaussian graphical model estimation.

Gaussian graphical models (GGMs) are a popular form of network model in which nodes represent features in multivariate normal data and edges reflect conditional dependencies between these features. GGM estimation is an active area of research. Currently available tools for GGM estimation require investigators to make several choices regarding algorithms, scoring criteria, and tuning parameters. An estimated GGM may be highly sensitive to these choices, and the accuracy of each method can vary based on structural characteristics of the network such as topology, degree distribution, and density. Because these characteristics are a priori unknown, it is not straightforward to establish universal guidelines for choosing a GGM estimation method. We address this problem by introducing SpiderLearner, an ensemble method that constructs a consensus network from multiple estimated GGMs. Given a set of candidate methods, SpiderLearner estimates the optimal convex combination of results from each method using a likelihood-based loss function. K $$ K $$ -fold cross-validation is applied in this process, reducing the risk of overfitting. In simulations, SpiderLearner performs better than or comparably to the best candidate methods according to a variety of metrics, including relative Frobenius norm and out-of-sample likelihood. We apply SpiderLearner to publicly available ovarian cancer gene expression data including 2013 participants from 13 diverse studies, demonstrating our tool's potential to identify biomarkers of complex disease. SpiderLearner is implemented as flexible, extensible, open-source code in the R package ensembleGGM at https://github.com/katehoffshutta/ensembleGGM.

Weight Change Following Switch to Dolutegravir for HIV Treatment in Rural Kenya During Country Roll-Out.

INTRODUCTION: Switch to dolutegravir (DTG) in treatment-experienced people living with HIV (PLH) is associated with excess weight gain in some settings; data are limited from rural low-income settings with low obesity prevalence. METHODS: In rural Kenya, we conducted a retrospective cohort study at 8 HIV clinics and a single-site prospective cohort study including adults switching to DTG during countrywide transition to DTG/tenofovir DF(TDF)/emtricitabine as first-line HIV treatment. In the retrospective analysis, we used preswitch data to model postswitch weight trajectory had each participant not switched to DTG and contrasted observed vs. predicted postswitch weight. In the prospective analysis, we measured weight post-DTG switch and evaluated predictors of 6-month weight change. RESULTS: Our retrospective cohort included 4445 PLH who switched to DTG between 2018 and 2020. Mean 12-month weight change was 0.6 kg preswitch and 0.8 kg postswitch. Among those on TDF throughout (n = 3374; 83% on efavirenz preswitch), 12-month postswitch weight was 0.7 kg more than predicted for women (95% CI: 0.4, 1.0) and similar among men (0.04 kg; 95% CI -0.3, 0.4). In our prospective cohort (n = 135, 100% female), mean 6-month weight change was +0.4 kg (IQR -1.1, 2.0 kg). Predicted gain varied by baseline food insecurity: +1.1 kg (95% CI: 0.34, 1.87) among food secure, -0.09 kg (95% CI -0.71, 0.54) among moderate insecure, and +0.27 kg (95% CI -0.82, 1.36) among severe insecurity. CONCLUSION: In contrast to some reports of large weight gain following switch to DTG, we observed small weight increases in women and no weight change in men following DTG switch when on TDF throughout. Weight gain may be attenuated by food insecurity, though was modest even among food secure.

A Demonstration of Modified Treatment Policies to Evaluate Shifts in Mobility and COVID-19 Case Rates in US Counties.

Mixed evidence exists of associations between mobility data and coronavirus disease 2019 (COVID-19) case rates. We aimed to evaluate the county-level impact of reducing mobility on new COVID-19 cases in summer/fall of 2020 in the United States and to demonstrate modified treatment policies to define causal effects with continuous exposures. Specifically, we investigated the impact of shifting the distribution of 10 mobility indexes on the number of newly reported cases per 100,000 residents 2 weeks ahead. Primary analyses used targeted minimum loss-based estimation with Super Learner to avoid parametric modeling assumptions during statistical estimation and flexibly adjust for a wide range of confounders, including recent case rates. We also implemented unadjusted analyses. For most weeks, unadjusted analyses suggested strong associations between mobility indexes and subsequent new case rates. However, after confounder adjustment, none of the indexes showed consistent associations under mobility reduction. Our analysis demonstrates the utility of this novel distribution-shift approach to defining and estimating causal effects with continuous exposures in epidemiology and public health.

Preconception sleep duration, sleep timing, and shift work in association with fecundability and live birth among women with a history of pregnancy loss.

OBJECTIVE: To evaluate the associations between preconception sleep characteristics and shift work with fecundability and live birth. DESIGN: Secondary analysis of the Effects of Aspirin in Gestation and Reproduction study, a preconception cohort. SETTING: Four US academic medical centers. PATIENT(S): Women aged 18-40 with a history of 1-2 pregnancy losses who were attempting to conceive again. INTERVENTION(S): Not applicable. MAIN OUTCOME MEASURES(S): We evaluated baseline, self-reported sleep duration, sleep midpoint, social jetlag, and shift work among 1,228 women who were observed for ≤6 cycles of pregnancy attempts to ascertain fecundability. We ascertained live birth at the end of follow up via chart abstraction. We estimated fecundability odds ratios (FORs) using discrete, Cox proportional hazards models and risk ratios (RRs) for live birth using log-Poisson models. RESULT(S): Sleep duration ≥9 vs. 7 to <8 hours (FOR: 0.81, 95% confidence interval [CI], 0.61; 1.08), later sleep midpoints (3rd tertile vs. 2nd tertile: FOR: 0.85; 95% CI, 0.69, 1.04) and social jetlag (continuous per hour; FOR: 0.93, 95% CI: 0.86, 1.00) were not associated with reduced fecundability. In sensitivity analyses, excluding shift workers, sleep duration ≥9 vs. 7 to <8 hours (FOR: 0.62; 95% CI, 0.42; 0.93) was associated with low fecundability. Night shift work was not associated with fecundability (vs. non-night shift work FOR: 1.17, 95% CI, 0.96; 1.42). Preconception sleep was not associated with live birth. CONCLUSION(S): Overall, there does not appear to be a strong association between sleep characteristics, fecundability, and live birth. Although these findings may suggest weak and imprecise associations with some sleep characteristics, our findings should be evaluated in larger cohorts of women with extremes of sleep characteristics. CLINICAL TRIAL REGISTRATION NUMBER: Clinicaltrials.gov NCT00467363.

Two-Stage TMLE to reduce bias and improve efficiency in cluster randomized trials.

Cluster randomized trials (CRTs) randomly assign an intervention to groups of individuals (e.g., clinics or communities) and measure outcomes on individuals in those groups. While offering many advantages, this experimental design introduces challenges that are only partially addressed by existing analytic approaches. First, outcomes are often missing for some individuals within clusters. Failing to appropriately adjust for differential outcome measurement can result in biased estimates and inference. Second, CRTs often randomize limited numbers of clusters, resulting in chance imbalances on baseline outcome predictors between arms. Failing to adaptively adjust for these imbalances and other predictive covariates can result in efficiency losses. To address these methodological gaps, we propose and evaluate a novel two-stage targeted minimum loss-based estimator to adjust for baseline covariates in a manner that optimizes precision, after controlling for baseline and postbaseline causes of missing outcomes. Finite sample simulations illustrate that our approach can nearly eliminate bias due to differential outcome measurement, while existing CRT estimators yield misleading results and inferences. Application to real data from the SEARCH community randomized trial demonstrates the gains in efficiency afforded through adaptive adjustment for baseline covariates, after controlling for missingness on individual-level outcomes.

Two or more significant life-events in 6-months are associated with lower rates of HIV treatment and virologic suppression among youth with HIV in Uganda and Kenya.

Youth living with HIV in sub-Saharan Africa have poor HIV care outcomes. We determined the association of recent significant life-events with HIV antiretroviral treatment (ART) initiation and HIV viral suppression in youth aged 15-24 years living with HIV in rural Kenya and Uganda. This was a cross-sectional analysis of 995 youth enrolled in the SEARCH Youth study. At baseline, providers assessed recent (within 6 months) life-events, defined as changes in schooling/employment, residence, partnerships, sickness, incarceration status, family strife or death, and birth/pregnancy, self-reported alcohol use, being a parent, and HIV-status disclosure. We examined the frequencies of events and their association with ART status and HIV viral suppression (<400 copies/ul). Recent significant life-events were prevalent (57.7%). Having >2 significant life-events (aOR = 0.61, 95% CI:0.45-0.85) and consuming alcohol (aOR = 0.61, 95% CI:0.43-0.87) were associated with a lower odds of HIV viral suppression, while disclosure of HIV-status to partner (aOR = 2.39, 95% CI:1.6-3.5) or to family (aOR = 1.86, 95% CI:1.3-2.7), being a parent (aOR = 1.8, 95% CI:1.2-2.5), and being single (aOR = 1.6, 95% CI:1.3-2.1) had a higher odds. This suggest that two or more recent life-events and alcohol use are key barriers to ART initiation and achievement of viral suppression among youth living with HIV in rural East Africa.Trial registration: ClinicalTrials.gov identifier: NCT03848728..