Faculty Bibliography

BACKGROUND:Mathematical models guide tuberculosis (TB) target-setting, yet most assume homogeneous "all-to-all" mixing. We compared projected intervention impacts between an all-to-all compartmental model and a Barabási-Albert (BA) scale‑free social network model under otherwise identical disease assumptions. METHODS:We calibrated transmission parameters so both models produced similar baseline trends, then introduced vaccination (coverage 30-70%; efficacy 80-95%) and treatment (20-50% increases in recovery) after a 400‑day burn‑in. Outcomes were assessed 300 days post‑intervention. RESULTS:Under 60% coverage, increasing vaccine efficacy from 80% to 95% yielded smaller projected reductions in active TB with the network model than with all‑to‑all mixing. Treatment improvements showed the same pattern: lower reductions under the network than the all‑to‑all model at modest efficacy, converging at high efficacy/coverage. Findings were robust across baseline prevalence scenarios. CONCLUSIONS:Accounting for social networks can attenuate projected impacts for sub‑optimal TB interventions. Forecasts and target‑setting should include sensitivity to social network structure.

BACKGROUND:Co-occurring alcohol, substance, and mood-related (CASM) conditions are prevalent, mutually reinforcing, and under-diagnosed contributors to morbidity, mortality, and health disparities. OBJECTIVE:To evaluate screening strategies leveraging the predictive value arising from patterns of CASM co-occurrence in populations with high CASM prevalence. DESIGN/METHODS:Individual-based health risks model validated to predict US life expectancy and causes of death by sex and age decile, including CASM conditions of depression, anxiety, chronic pain, and unhealthy alcohol, tobacco, opioid and stimulant use. The model includes CASM co-occurrence patterns, mutual reinforcement across CASM conditions, and reduced engagement in other preventative care due to CASM. PARTICIPANTS/METHODS:Veterans Aging Cohort Study (VACS), a large longitudinal cohort of in-care US veterans. INTERVENTIONS/METHODS:(1) Screening alcohol, tobacco, and/or depression symptoms; (2) adding further screening of CASM conditions likely to co-occur with those screened positive, with variation in the minimum co-occurrence rate; (3) screening all CASM conditions (hypothetical maximum). MAIN MEASURES/METHODS:Estimated life expectancy (LE) and quality-adjusted life-years (QALYs). KEY RESULTS/RESULTS:The maximum strategy added 0.52 years to estimated LE (95% CI: 0.51 - 0.54) and 0.68 QALYs/person (95% CI: 0.67 - 0.69). Screening individual CASM conditions added a small fraction of this benefit, the largest LE gain from tobacco screening: 0.08 years (95% CI: 0.07 - 0.09). Screening for depression, alcohol, and tobacco provided 34.6% of the maximum strategy's LE gain (0.19 years, 95% CI: 0.17 - 0.20). Additionally screening conditions with moderate (≥ 20%) probability of co-occurring with those already screened positive provided 84.8% of the maximum strategy's LE gain. Screening all CASM conditions if depression, alcohol, and/or tobacco screened positive provided 86.6% of the maximum strategy's LE gain. CONCLUSIONS:Compared to common practice of screening one or few CASM conditions, large health benefits are possible by further assessing CASM conditions most likely to co-occur with those already screening positive, improving health without increasing up-front screening burden in populations with high CASM prevalence.

OBJECTIVES/OBJECTIVE:HIV viral load (VL) testing using polymerase chain reaction (PCR) is a mainstay of treatment monitoring, but is technically demanding, time-consuming, and costly. We investigated whether antibody measurements can detect VL rebound, leveraging a recent analytic treatment interruption (ATI) study, NCT03225118. METHODS:We tested longitudinal specimens in N=22 ATI participants (91% male, 38-61 year age range, 2.1-15.9 years VL suppression pre-ATI) using Limiting Antigen (LAg) antibody assays, measured as normalized optical density (ODn). We used Bayesian inference to fit linear mixed effects models, including time between first detectable VL and ODn rise, time between peak VL and peak ODn, and ODn rate-of-change after VL re-suppression. RESULTS:LAg ODn increased 4.36 weeks (95% CI: 3.58, 5.79) after treatment interruption and 2.78 weeks (95% CI: 1.94, 4.15) after VL became detectable. ODn peaked 1.03 weeks (95% CI: 0.25, 1.93) after VL peaked, increasing 1.70 fold (95% CI: 1.39, 2.08) compared to pre-interruption. After VL re-suppression, ODn declined by -0.015 units per week (95% CI: -0.017, -0.013) over one year of follow-up. CONCLUSIONS:LAg antibody levels rose in all participants 2-4 weeks after VL detection in treatment interruption. Longitudinal antibody measurements could support facile, rapid, low-cost HIV treatment monitoring.

BACKGROUND:Characterising disparities in HIV infection across populations by gender, age, and HIV risk is key information to guide intervention priorities. We aimed to assess how indicators measuring HIV acquisitions, transmissions, or potential long-term infections influence estimates of the contribution of different populations to new infections, including key populations (including female sex workers, their clients, men who have sex with men). METHODS:) measured the proportion of new infections averted if transmission involving a specific population was blocked over a specific time period. We compared estimates of the four indicators across seven populations and 15 settings and assessed if the contribution of specific populations ranked differently across indicators for ten settings. FINDINGS/RESULTS:), whereas more infections were transmitted than acquired in non-key population men aged 25 years and older (median 1·4 times more) and clients of female sex workers (1·6 times more) in all but one model. Estimates of the 10-year tPAFs accounting for transmission in the long-term were substantially larger than the direct transmission indicator for all populations, especially for female sex workers (2·0 times higher). INTERPRETATION/CONCLUSIONS:Indicators that reflect HIV acquisitions and transmissions in the short and long term can be used to capture the complexity of HIV epidemics across different populations and timeframes. The added nuance would improve the effectiveness of the HIV prevention response across all populations at risk. FUNDING/BACKGROUND:US National Institutes of Health and UK Medical Research Council. TRANSLATION/UNASSIGNED:For the French translation of the abstract see Supplementary Materials section.

Extreme heat events are a growing health threat, but their impact is heterogeneous because different settings have different levels of heat adaptation. Previous reviews have assessed morbidity and mortality as a function of meteorological conditions such as air temperature. We aimed to conduct the first systematic review and meta-analysis of the association between meeting local definitions of extreme heat events and risks of hospitalization and mortality overall and by population segment. We searched PubMed, Web of Science, Scopus, ScienceDirect, and Google Scholar on February 05, 2025, with no restrictions on language or publication date. Data extraction and risk of bias assessment were conducted by multiple reviewers. We estimated the risk ratio (RR) for hospitalization and mortality when meeting local definitions of extreme heat events using a random effects model to account for heterogeneity. In subgroup analysis, we examined variations by date, global region, country income group, extreme heat event definition type (absolute thresholds, percentile-based, composite indices), and population characteristics including age, socioeconomic context, and urban versus rural settings. In sensitivity analysis, we assessed the robustness of results by excluding lower-quality studies and applying alternative regression models. From 6,015 initial records, 21 studies (n = 126,930,288 individuals) met inclusion criteria. The pooled RR for mortality associated with extreme heat events was 1.24 (95% CI: 1.06-1.46) with substantial heterogeneity across studies (I² = 99.8%), which was explored through subgroup, meta-regression, and sensitivity analysis. The meta-regression showed a significant increase in RR over time (p < 0.05). No significant publication bias was detected (Begg's test, p = 0.458). In subgroup analysis, higher RRs were observed in studies from Europe (RR range: 1.16-4.24) and low- and middle-income countries, in older adults (RR range: 1.16-2.24), in urban populations, in older populations, and in studies using absolute temperature thresholds or composite indices. Findings were similar when excluding lower-quality studies and applying alternative regression models. Extreme heat events were significantly associated with mortality, with risks increasing over time and elevated among older adults and urban populations. Municipal and health authorities should prioritize extreme heat adaptation, as mortality is likely to increase as the climate warms.

INTRODUCTION/BACKGROUND:Depression is the most common mental disorder in sub-Saharan Africa (SSA). Group interpersonal therapy (IPT-G) is a recommended first-line treatment for depression, shown to be safe and effective in clinical trials. However, less is known about its real-world retention and effectiveness when delivered at scale in SSA. We describe retention patterns and associated factors in a large IPT-G programme in Uganda and Zambia. METHODS:test. RESULTS:Among 45 349 clients, overall average attendance was 82%. Four classes emerged from attendance patterns: high attendance (63%), moderate attendance (27%), early dropout (6%) and late dropout (4%). Relative to the high attendance class, the early drop-out class had higher odds of being age <25 (adjusted OR (aOR) 1.19, 95% CI 0.99 to 1.44), in teletherapy (aOR 3.46, 95% CI 2.89 to 4.13) and presenting with moderate than moderately severe to severe depression (aOR 1.27, 95% CI 1.17 to 1.40). The overall mean reduction in PHQ-9 scores was 13.0 (SD 4.2), but the early drop-out group showed smaller improvements (10.32, SD 8.78) compared with the high attendance group (13.3, SD 6.15, p=0.001). CONCLUSION/CONCLUSIONS:Retention in a scaled IPT-G programme in Uganda and Zambia was high. Still, early dropout-linked to younger age, teletherapy and moderate depression-was associated with less reduction in depressive symptoms. These findings highlight areas for innovation in IPT-G implementation to improve engagement and outcomes.

BACKGROUND:The USA has traditionally been the largest donor to health programmes in low-income and middle-income countries (LMICs). In January 2025, almost all such funding was stopped and prospects for its resumption are uncertain. The suddenness of the funding cuts makes it difficult for national health programmes in LMICs to adapt. We aimed to estimate the impact of these cuts on deaths and other outcomes (new infections, number of family planning users, and unplanned pregnancies) for four health areas that have been a focus of a substantial amount of US foreign assistance: HIV, tuberculosis, family planning, and maternal and child health. METHODS:We applied established mathematical models to the countries receiving US foreign assistance in each domain to estimate health impacts over the period 2025 to 2030. We used six models of HIV, three different approaches to estimate family planning impact, and one model each for tuberculosis and maternal and child health, applying these models to as many as 80 countries. We compared model projections assuming constant funding (status quo) with projections assuming complete elimination of US funding in each country. Some models also considered partial cuts or restoration of funding over time. FINDINGS/RESULTS:A complete cessation of US funding without replacement by other sources would lead to drastic increases in deaths from 2025 to 2030: 4·1 million (range 1·6-6·6) additional AIDS-related deaths across 55 countries, 606 900 (95% uncertainty interval [UI] 466 000-768 800) additional tuberculosis deaths across 79 countries, 40-55 million additional unplanned pregnancies and 12-16 million unsafe abortions across 51 countries, and 2·5 million (1·3-4·5) additional child deaths from causes other than HIV and tuberculosis across 24 countries. Restoration of funding for HIV treatment but not prevention would avoid most of the increase in deaths but still result in nearly 1 million more new HIV infections from 2025 to 2030. INTERPRETATION/CONCLUSIONS:Substantial progress has been made in improving global health in the past few decades. This progress has strengthened hope in reaching global development goals. However, the recent funding cuts threaten to change these trajectories and could lead to sharp increases in avoidable mortality for the poorest countries. Even a partial restoration of US funding would combat the most severe effects and provide time for countries that have received substantial US foreign assistance to adjust to the new funding landscape. FUNDING/BACKGROUND:Economic and Social Research Council; Engineering and Physical Sciences Research Council; European and Developing Countries Clinical Trials Partnership; Gates Foundation; Global Fund to Fight AIDS, Tuberculosis, and Malaria; Open Philanthropy; UK Foreign, Commonwealth & Development Office; UK Medical Research Council; UN Population Fund; UNAIDS; US National Institute of Allergy and Infectious Diseases; University of Edinburgh; US National Institutes of Health; US President's Emergency Plan for AIDS Relief; Wellcome Trust; World Bank; WHO.

BACKGROUND:Communities hardest-hit by early SARS-CoV-2 outbreaks accrued more immunity, but prioritizing these communities for vaccination could reduce health disparities. Optimal vaccine allocation depends on inequality aversion, i.e., willingness to trade off aggregate health benefits to increase distributional equity. We evaluated the impact of vaccine prioritization strategies on COVID-19 infections and mortality in New York City (NYC). METHODS:We used a susceptible-exposed-infected-recovered COVID-19 transmission model calibrated to NYC neighborhood-level data to compare three vaccine distribution strategies: 1) uniform across neighborhoods (no prioritization); 2) prioritizing hardest-hit neighborhoods (exposure-based prioritization); and 3) prioritizing hardest-hit neighborhoods while maintaining mitigation measures in other neighborhoods (exposure-based prioritization plus mitigation). The model accounted for vaccine efficacy, rollout pace, pre-vaccine immunity, and heterogeneous neighborhood exposure risk. We categorized 42 NYC neighborhoods into quintiles of cumulative COVID-19 mortality rates from March 1, 2020, until first vaccine availability (December 14, 2020). We modeled total deaths and equally-distributed-equivalent (EDE) deaths (i.e., the equally preferred number of deaths, considering equity and efficiency) across a range of inequality aversion (Atkinson's index, ε=0-20). RESULTS:Exposure-based prioritization plus mitigation was estimated to avert the most citywide COVID-19 deaths (32.5 %) relative to no vaccination, regardless of adjustment for inequality aversion. Relative to no prioritization, exposure-based prioritization was estimated to avert 45 % fewer citywide deaths but generated 2.5 % more EDE-adjusted deaths at an Atkinson index of 10. Exposure-based prioritization outperformed no prioritization at an Atkinson index of ≥ 6. CONCLUSIONS:Prioritizing vaccination within the hardest-hit communities, paired with sustained mitigation efforts in communities with the greatest advantage, resulted in the greatest overall reduction in mortality and inequities. Emergency response teams should consider a community's ability to continue non-pharmaceutical mitigation efforts when allocating limited pharmaceutical supplies.