Faculty Bibliography

"As the lockdowns lift" -- "Here where the valley fans the wind" -- "What if, while working"

Background: Increased drug use has disproportionately impacted rural areas across the U.S. People who use drugs are at risk of overdose and other medical complications, including infectious diseases. Understanding barriers to healthcare access for this often stigmatized population is key to reducing morbidity and mortality, particularly in rural settings where resources may be limited. Methods: We conducted 20 semi-structured interviews with people who use drugs, including 17 who inject drugs, in rural southern Illinois between June 2018 and February 2019. Interviews were analyzed using a modified grounded theory approach where themes are coded and organized as they emerge from the data. Results: Participants reported breaches of trust by healthcare providers, often involving law enforcement and Emergency Medical Services, that dissuaded them from accessing medical care. Participants described experiences of mistreatment in emergency departments, with one account of forced catheterization. They further recounted disclosures of protected health information by healthcare providers, including communicating drug test results to law enforcement and sharing details of counseling sessions with community members without consent. Participants also described a hesitancy common among people who use drugs to call emergency medical services for an overdose due to fear of arrest. Conclusion: Breaches of trust by healthcare providers in rural communities discouraged people who use drugs from accessing medical care until absolutely necessary, if at all. These experiences may worsen healthcare outcomes and further stigmatize this marginalized community. Structural changes including reforming and clarifying law enforcement's role in Emergency Departments as well as instituting diversion policies during arrests may help rebuild trust in these communities. Other possible areas for intervention include stigma training and harm reduction education for emergency medicine providers, as well as developing and implementing referral systems between Emergency Departments and local harm reduction providers and medically assisted drug treatment programs.

SUMMARY:We investigate convolutional neural networks (CNNs) for filtering small genomic variants in short-read DNA sequence data. Errors created during sequencing and library preparation make variant calling a difficult task. Encoding the reference genome and aligned reads covering sites of genetic variation as numeric tensors allows us to leverage CNNs for variant filtration. Convolutions over these tensors learn to detect motifs useful for classifying variants. Variant filtering models are trained to classify variants as artifacts or real variation. Visualizing the learned weights of the CNN confirmed it detects familiar DNA motifs known to correlate with real variation, like homopolymers and short tandem repeats (STR). After confirmation of the biological plausibility of the learned features we compared our model to current state-of-the-art filtration methods like Gaussian Mixture Models, Random Forests and CNNs designed for image classification, like DeepVariant. We demonstrate improvements in both sensitivity and precision. The tensor encoding was carefully tailored for processing genomic data, respecting the qualitative differences in structure between DNA and natural images. Ablation tests quantitatively measured the benefits of our tensor encoding strategy. Bayesian hyper-parameter optimization confirmed our notion that architectures designed with DNA data in mind outperform off-the-shelf image classification models. Our cross-generalization analysis identified idiosyncrasies in truth resources pointing to the need for new methods to construct genomic truth data. Our results show that models trained on heterogenous data types and diverse truth resources generalize well to new datasets, negating the need to train separate models for each data type. AVAILABILITY AND IMPLEMENTATION:This work is available in the Genome Analysis Toolkit (GATK) with the tool name CNNScoreVariants (https://github.com/broadinstitute/gatk). SUPPLEMENTARY INFORMATION:Supplementary data are available at Bioinformatics online.

BACKGROUND:Much remains unknown in rural risk environments, despite a growing crisis in these areas. We adapt a risk environment framework to characterize rural southern Illinois and describe the relations of risk environments, opioid-related overdose, HIV, Hepatitis C, and sexually transmitted infection rates between 2015 and 2017. METHODS:Over two dozen risk environment variables are summarized across zip-code (n = 128) or county levels (n = 16) based on availability and theoretical relevance. We calculate data attribute associations and characterize spatial and temporal dimensions of longitudinal health outcomes and the rural risk environment. We then use a "regional typology analysis" to generate data-driven risk regions and compare health outcomes. RESULTS:Pervasive risk hotspots were identified in more populated locales with higher rates of overdose and HCV incidence, whereas emerging risk areas were isolated to more rural locales that had experienced an increase in analgesic opiate overdoses and generally lacked harm-reduction resources. At-risk areas were characterized with underlying socioeconomic vulnerability but in differing ways, reflecting a nuanced and shifting structural risk landscape. CONCLUSIONS:Rural risk environment vulnerabilities and associated opioid-related health outcomes are multifaceted and spatially heterogeneous. More research is needed to better understand how refining geographies to more precisely define risk can support intervention efforts and further enrich investigations of the opioid epidemic.

OBJECTIVES:Emerging literature shows that racialised police brutality, a form of structural racism, significantly affects health and well-being of racial/ethnic minorities in the USA. While public health research suggests that structural racism is a distal determinant of sexually transmitted infections (STIs) among Black people, no studies have empirically linked police violence to STIs. To address this gap, our study measures associations between police killings and rates of STIs among Black residents of US metropolitan statistical areas (MSAs). METHODS:This cross-sectional ecological analysis assessed associations between the number of Black people killed by police in 2015 and rates of primary and secondary syphilis, gonorrhoea and chlamydia per 100 000 Black residents of all ages in 2016 in 75 large MSAs. Multivariable models controlled for MSA-level demographic and socioeconomic characteristics, police expenditures, violent crime, arrest and incarceration rates, insurance rates and healthcare funding. RESULTS:In 2015, the median number of Black people killed by police per MSA was 1.0. In multivariable models, police killings were positively and significantly associated with syphilis and gonorrhoea rates among Black residents. Each additional police killing in 2015 was associated with syphilis rates that were 7.5% higher and gonorrhoea rates that were 4.0% higher in 2016. CONCLUSIONS:Police killings of Black people may increase MSA-level risk of STI infections among Black residents. If future longitudinal analyses support these findings, efforts to reduce STIs among Black people should include reducing police brutality and addressing mechanisms linking this violence to STIs.

Assessment of the long-term population-level effects of HIV interventions is an ongoing public health challenge. Following the implementation of a Transmission Reduction Intervention Project (TRIP) in Odessa, Ukraine, in 2013-2016, we obtained HIV pol gene sequences and used phylogenetics to identify HIV transmission clusters. We further applied the birth-death skyline model to the sequences from Odessa (n = 275) and Kyiv (n = 92) in order to estimate changes in the epidemic's effective reproductive number (R_e) and rate of becoming uninfectious (δ). We identified 12 transmission clusters in Odessa; phylogenetic clustering was correlated with younger age and higher average viral load at the time of sampling. Estimated R_e were similar in Odessa and Kyiv before the initiation of TRIP; R_e started to decline in 2013 and is now below R_e = 1 in Odessa (R_e = 0.4, 95%HPD 0.06-0.75), but not in Kyiv (R_e = 2.3, 95%HPD 0.2-5.4). Similarly, estimates of δ increased in Odessa after the initiation of TRIP. Given that both cities shared the same HIV prevention programs in 2013-2019, apart from TRIP, the observed changes in transmission parameters are likely attributable to the TRIP intervention. We propose that molecular epidemiology analysis can be used as a post-intervention effectiveness assessment tool.

BACKGROUND:Adequate access to effective treatment and medication assisted therapies for opioid dependence has led to improved antiretroviral therapy adherence and decreases in morbidity among people who inject drugs (PWID), and can also address a broad range of social and public health problems. However, even with the success of syringe service programs and opioid substitution programs in European countries (and others) the US remains historically low in terms of coverage and access with regard to these programs. This manuscript investigates predictors of historical change in drug treatment coverage for PWID in 90 US metropolitan statistical areas (MSAs) during 1993-2007, a period in which, overall coverage did not change. METHODS:Drug treatment coverage was measured as the number of PWID in drug treatment, as calculated by treatment entry and census data, divided by numbers of PWID in each MSA. Variables suggested by the Theory of Community Action (i.e., need, resource availability, institutional opposition, organized support, and service symbiosis) were analyzed using mixed-effects multivariate models within dependent variables lagged in time to study predictors of later change in coverage. RESULTS:Mean coverage was low in 1993 (6.7%; SD 3.7), and did not increase by 2007 (6.4%; SD 4.5). Multivariate results indicate that increases in baseline unemployment rate (β = 0.312; pseudo-p < 0.0002) predict significantly higher treatment coverage; baseline poverty rate (β = - 0.486; pseudo-p < 0.0001), and baseline size of public health and social work workforce (β = 0.425; pseudo-p < 0.0001) were predictors of later mean coverage levels, and baseline HIV prevalence among PWID predicted variation in treatment coverage trajectories over time (baseline HIV * Time: β = 0.039; pseudo-p < 0.001). Finally, increases in black/white poverty disparity from baseline predicted significantly higher treatment coverage in MSAs (β = 1.269; pseudo-p < 0.0001). CONCLUSIONS:While harm reduction programs have historically been contested and difficult to implement in many US communities, and despite efforts to increase treatment coverage for PWID, coverage has not increased. Contrary to our hypothesis, epidemiologic need, seems not to be associated with change in treatment coverage over time. Resource availability and institutional opposition are important predictors of change over time in coverage. These findings suggest that new ways have to be found to increase drug treatment coverage in spite of economic changes and belt-tightening policy changes that will make this difficult.

Over 30 years into the US HIV/AIDS epidemic, Black men who have sex with men (BMSM) continue to carry the highest burden of both HIV and AIDS cases. There is then, an urgent need to expand access to HIV prevention and treatment for all gay and bisexual men, underscoring the importance of the federal initiative 'Ending the Epidemic: A Plan for America'. This research examines structural factors associated with BMSM HIV testing coverage over time (2011-2016) in 85 US Metropolitan Statistical Areas (MSAs). We calculated MSA-specific annual measures of BMSM HIV testing coverage (2011-2016). Variables suggested by the Theory of Community Action (i.e., need, resource availability, institutional opposition and organized support) were analyzed as possible predictors of coverage using multilevel modeling. Relationships between BMSM HIV testing and the following covariates were positive: rates of BMSM living with HIV (b = 0.28), percent of Black residents employed (b = 0.19), Black heterosexual testing rate (b = 0.46), health expenditures per capita (b = 0.16), ACT UP organization presence in 1992 (b = 0.19), and syringe service presence (b = 0.12). Hard drug arrest rates at baseline (b = - 0.21) and change since baseline (b = - 0.10) were inversely associated with the outcome. Need, resources availability, organized support and institutional opposition are important determinants of place associated with BMSM HIV testing coverage. Efforts to reduce HIV incidence and lessen AIDS-related disparities among BMSM in the US require improved and innovative HIV prevention approaches directed toward BMSM including a fuller understanding of structural factors that may influence place variation in BMSM testing patterns and risk behavior in places of high need.

Research Objective: The opioid epidemic in the United States remains a growing public health problem. Understanding opioid prescribing for children, adolescents, and young adults is essential for developing targeted interventions and policies at both national and local levels for this population. In this study, we investigate temporal trends in opioid-prescribing practices in children, adolescents, and young adults in the United States from 2006 to 2018.
Study Design: Cross-sectional analysis of opioid prescriptions from retail pharmacies. Primary outcomes included total number of opioid prescriptions, duration of prescription (overall mean, <=3 days, and >=30 days), amount prescribed in mean morphine milligram equivalents (MME) dispensed per day, and high-dose prescriptions (prescriptions for doses >90 MME/day for those >14 years). Yearly values were reported overall and stratified by age for 2006, 2012, and 2018, with relative percentage change between point estimates for 2006 and 2018. Population Studied: Opioid prescription data from January 1, 2006, to December 31, 2018, for children, adolescents, and young adults <25 years were extracted from the IQVIA XPonent database which accounts for 90% of retail outpatient prescriptions in the United States. Principal Findings: The opioid prescription rate (per 100 persons) decreased by 58.9% overall from 2006 to 2018, declining from a rate of 4.6-1.4 for children 0-5 years, 4.3-1.4 for those 6-9 years, 6.6-2.7 for those 10-14 years, 20.4-10.8 for adolescents 15-19 years, and 35.1-15.3 for young adults 20-24 years. The total MME/day was 36.5 in 2006, 34.4 in 2012, and 31.2 in 2018 (decrease of 14.7% by 2018), with decreased amounts prescribed in all ages, except those <=5 years. The total mean opioid prescription duration remained relatively stable, with duration of 6 days in 2006, 6.8 days in 2012, and 6.2 days in 2018. The number of prescriptions with duration <=3 days was unchanged, but the number of prescriptions with duration >=30 days increased by 31.8%, with the largest increases in children <15 years. High-dose prescriptions for the 15-24 years cohort decreased by >50% between 2006 and 2018.
Conclusion(s): Since 2006, total opioid prescriptions have decreased for children, adolescents, and young adults, and there has been a reduction in the prescription amount and in high-dose opioid prescriptions for adolescents and young adults. However, despite current opioid-prescribing guidelines, as well as policies that limit days' supply, opioids continue to be frequently dispensed and opioid prescriptions with duration >=30 days have increased. Implications for Policy or Practice: Continued policy efforts and the development of clinical practice guidelines for youths are necessary to ensure safe and judicious opioid prescribing for children, adolescents, and young adults. Further study, including investigating geographic patterns and patient-and prescriber-level characteristics, will help inform targeted interventions for appropriate prescribing practices in this population