Cross-sector decision landscape in response to COVID-19: A qualitative network mapping analysis of North Carolina decision-makers

Introduction

The COVID-19 pandemic response has demonstrated the interconnectedness of individuals, organizations, and other entities jointly contributing to the production of community health. This response has involved stakeholders from numerous sectors who have been faced with new decisions, objectives, and constraints. We examined the cross-sector organizational decision landscape that formed in response to the COVID-19 pandemic in North Carolina.

Methods

We conducted virtual semi-structured interviews with 44 organizational decision-makers representing nine sectors in North Carolina between October 2020 and January 2021 to understand the decision-making landscape within the first year of the COVID-19 pandemic. In line with a complexity/systems thinking lens, we defined the decision landscape as including decision-maker roles, key decisions, and interrelationships involved in producing community health. We used network mapping and conventional content analysis to analyze transcribed interviews, identifying relationships between stakeholders and synthesizing key themes.

Results

Decision-maker roles were characterized by underlying tensions between balancing organizational mission with employee/community health and navigating organizational vs. individual responsibility for reducing transmission. Decision-makers' roles informed their perspectives and goals, which influenced decision outcomes. Key decisions fell into several broad categories, including how to translate public health guidance into practice; when to institute, and subsequently loosen, public health restrictions; and how to address downstream social and economic impacts of public health restrictions. Lastly, given limited and changing information, as well as limited resources and expertise, the COVID-19 response required cross-sector collaboration, which was commonly coordinated by local health departments who had the most connections of all organization types in the resulting network map.

Conclusions

By documenting the local, cross-sector decision landscape that formed in response to COVID-19, we illuminate the impacts different organizations may have on information/misinformation, prevention behaviors, and, ultimately, health. Public health researchers and practitioners must understand, and work within, this complex decision landscape when responding to COVID-19 and future community health challenges.

Association of Simulated COVID-19 Vaccination and Nonpharmaceutical Interventions With Infections, Hospitalizations, and Mortality

Importance

Vaccination against SARS-CoV-2 has the potential to significantly reduce transmission and COVID-19 morbidity and mortality. The relative importance of vaccination strategies and nonpharmaceutical interventions (NPIs) is not well understood.

Objective

To assess the association of simulated COVID-19 vaccine efficacy and coverage scenarios with and without NPIs with infections, hospitalizations, and deaths.

Design, Setting, and Participants

An established agent-based decision analytical model was used to simulate COVID-19 transmission and progression from March 24, 2020, to September 23, 2021. The model simulated COVID-19 spread in North Carolina, a US state of 10.5 million people. A network of 1 017 720 agents was constructed from US Census data to represent the statewide population.

Exposures

Scenarios of vaccine efficacy (50% and 90%), vaccine coverage (25%, 50%, and 75% at the end of a 6-month distribution period), and NPIs (reduced mobility, school closings, and use of face masks) maintained and removed during vaccine distribution.

Main Outcomes and Measures

Risks of infection from the start of vaccine distribution and risk differences comparing scenarios. Outcome means and SDs were calculated across replications.

Results

In the worst-case vaccination scenario (50% efficacy, 25% coverage), a mean (SD) of 2 231 134 (117 867) new infections occurred after vaccination began with NPIs removed, and a mean (SD) of 799 949 (60 279) new infections occurred with NPIs maintained during 11 months. In contrast, in the best-case scenario (90% efficacy, 75% coverage), a mean (SD) of 527 409 (40 637) new infections occurred with NPIs removed and a mean (SD) of 450 575 (32 716) new infections occurred with NPIs maintained. With NPIs removed, lower efficacy (50%) and higher coverage (75%) reduced infection risk by a greater magnitude than higher efficacy (90%) and lower coverage (25%) compared with the worst-case scenario (mean [SD] absolute risk reduction, 13% [1%] and 8% [1%], respectively).

Conclusions and Relevance

Simulation outcomes suggest that removing NPIs while vaccines are distributed may result in substantial increases in infections, hospitalizations, and deaths. Furthermore, as NPIs are removed, higher vaccination coverage with less efficacious vaccines can contribute to a larger reduction in risk of SARS-CoV-2 infection compared with more efficacious vaccines at lower coverage. These findings highlight the need for well-resourced and coordinated efforts to achieve high vaccine coverage and continued adherence to NPIs before many prepandemic activities can be resumed.

Examining the bidirectional relationship between food insecurity and healthcare spending

OBJECTIVE

To improve food insecurity interventions, we sought to better understand the hypothesized bidirectional relationship between food insecurity and health care expenditures.

DATA SOURCE

Nationally representative sample of the civilian noninstitutionalized population of the United States (2016-2017 Medical Expenditure Panel Survey [MEPS]).

STUDY DESIGN

In a retrospective longitudinal cohort, we conducted two sets of analyses: (a) two-part models to examine the association between food insecurity in 2016 and health care expenditures in 2017; and (b) logistic regression models to examine the association between health care expenditures in 2016 and food insecurity in 2017. We adjusted for demographic and socioeconomic variables as well as 2016 health care expenditures and food insecurity.

DATA COLLECTION

Health care expenditures, food insecurity, and medical condition data from 10 886 adults who were included in 2016-2017 MEPS.

PRINCIPAL FINDINGS

Food insecurity in 2016, compared with being food secure, was associated with both a higher odds of having any health care expenditures in 2017 (OR 1.29, 95% CI: 1.04 to 1.60) and greater total expenditures ($1738.88 greater, 95% CI: $354.10 to $3123.57), which represents approximately 25% greater expenditures. Greater 2016 health care expenditures were associated with slightly higher odds of being food insecure in 2017 (OR 1.007 per $1000 in expenditures, 95% CI: 1.002 to 1.012, P =0.01). Exploratory analyses suggested that poor health status may underlie the relationship between food insecurity and health care expenditures.

CONCLUSIONS

A bidirectional relationship exists between food insecurity and health care expenditures, but the strength of either direction appears unequal. Higher health care expenditures are associated with a slightly greater risk of being food insecure (adjusted for baseline food insecurity status) but being food insecure is associated with substantially greater subsequent health care expenditures (adjusted for baseline health care expenditures). Interventions to address food insecurity and poor health may be helpful to break this cycle.

The Joint Impact of COVID-19 Vaccination and Non-Pharmaceutical Interventions on Infections, Hospitalizations, and Mortality: An Agent-Based Simulation

Background

Vaccination against SARS-CoV-2 has the potential to significantly reduce transmission and morbidity and mortality due to COVID-19. This modeling study simulated the comparative and joint impact of COVID-19 vaccine efficacy and coverage with and without non-pharmaceutical interventions (NPIs) on total infections, hospitalizations, and deaths.

Methods

An agent-based simulation model was employed to estimate incident SARS-CoV-2 infections and COVID-19-associated hospitalizations and deaths over 18 months for the State of North Carolina, a population of roughly 10.5 million. Vaccine efficacy of 50% and 90% and vaccine coverage of 25%, 50%, and 75% (at the end of a 6-month distribution period) were evaluated. Six vaccination scenarios were simulated with NPIs (i.e., reduced mobility, school closings, face mask usage) maintained and removed during the period of vaccine distribution.

Results

In the worst-case vaccination scenario (50% efficacy and 25% coverage), 2,231,134 new SARS-CoV-2 infections occurred with NPIs removed and 799,949 infections with NPIs maintained. In contrast, in the best-case scenario (90% efficacy and 75% coverage), there were 450,575 new infections with NPIs maintained and 527,409 with NPIs removed. When NPIs were removed, lower efficacy (50%) and higher coverage (75%) reduced infection risk by a greater magnitude than higher efficacy (90%) and lower coverage (25%) compared to the worst-case scenario (absolute risk reduction 13% and 8%, respectively).

Conclusion

Simulation results suggest that premature lifting of NPIs while vaccines are distributed may result in substantial increases in infections, hospitalizations, and deaths. Furthermore, as NPIs are removed, higher vaccination coverage with less efficacious vaccines can contribute to a larger reduction in risk of SARS-CoV-2 infection compared to more efficacious vaccines at lower coverage. Our findings highlight the need for well-resourced and coordinated efforts to achieve high vaccine coverage and continued adherence to NPIs before many pre-pandemic activities can be resumed.

Use of Machine Learning Approaches in Clinical Epidemiological Research of Diabetes

PURPOSE OF REVIEW

Machine learning approaches-which seek to predict outcomes or classify patient features by recognizing patterns in large datasets-are increasingly applied to clinical epidemiology research on diabetes. Given its novelty and emergence in fields outside of biomedical research, machine learning terminology, techniques, and research findings may be unfamiliar to diabetes researchers. Our aim was to present the use of machine learning approaches in an approachable way, drawing from clinical epidemiological research in diabetes published from 1 Jan 2017 to 1 June 2020.

RECENT FINDINGS

Machine learning approaches using tree-based learners-which produce decision trees to help guide clinical interventions-frequently have higher sensitivity and specificity than traditional regression models for risk prediction. Machine learning approaches using neural networking and "deep learning" can be applied to medical image data, particularly for the identification and staging of diabetic retinopathy and skin ulcers. Among the machine learning approaches reviewed, researchers identified new strategies to develop standard datasets for rigorous comparisons across older and newer approaches, methods to illustrate how a machine learner was treating underlying data, and approaches to improve the transparency of the machine learning process. Machine learning approaches have the potential to improve risk stratification and outcome prediction for clinical epidemiology applications. Achieving this potential would be facilitated by use of universal open-source datasets for fair comparisons. More work remains in the application of strategies to communicate how the machine learners are generating their predictions.

Platelet-endothelial associations may promote cytomegalovirus replication in the salivary gland in mice

Platelet decline is a feature of many acute viral infections, including cytomegalovirus (CMV) infection in humans and mice. Platelet sequestration in association with other cells, including endothelium and circulating leukocytes, can contribute to this decline and influence the immune response to and pathogenesis of viral infection. We sought to determine if platelet-endothelial associations (PEAs) contribute to platelet decline during acute murine CMV (mCMV) infection, and if these associations affect viral load and production. Male BALB/c mice were infected with mCMV (Smith strain), euthanized at timepoints throughout acute infection and compared to uninfected controls. An increase in PEA formation was confirmed in the salivary gland at all post-inoculation timepoints using immunohistochemistry for CD41+ platelets co-localizing with CD34+ vessels. Platelet depletion did not change amount of viral DNA or timecourse of infection, as measured by qPCR. However, platelet depletion reduced viral titer of mCMV in the salivary glands while undepleted controls demonstrated robust replication in the tissue by plaque assay. Thus, platelet associations with endothelium may enhance the ability of mCMV to replicate within the salivary gland. Further work is needed to determine the mechanisms behind this effect and if pharmacologic inhibition of PEAs may reduce CMV production in acutely infected patients.

Cost-effectiveness of Preventive Therapy for Tuberculosis With Isoniazid and Rifapentine Versus Isoniazid Alone in High-Burden Settings

Background

A short-course regimen of 3 months of weekly rifapentine and isoniazid (3HP) has recently been recommended by the World Health Organization as an alternative to at least 6 months of daily isoniazid (isoniazid preventive therapy [IPT]) for prevention of tuberculosis (TB). The contexts in which 3HP may be cost-effective compared to IPT among people living with human immunodeficiency virus are unknown.

Methods

We used a Markov state transition model to estimate the incremental cost-effectiveness of 3HP relative to IPT in high-burden settings, using a cohort of 1000 patients in a Ugandan HIV clinic as an emblematic scenario. Cost-effectiveness was expressed as 2017 US dollars per disability-adjusted life year (DALY) averted from a healthcare perspective over a 20-year time horizon. We explored the conditions under which 3HP would be considered cost-effective relative to IPT.

Results

Per 1000 individuals on antiretroviral therapy in the reference scenario, treatment with 3HP rather than IPT was estimated to avert 9 cases of TB and 1 death, costing $9402 per DALY averted relative to IPT. Cost-effectiveness depended strongly on the price of rifapentine, completion of 3HP, and prevalence of latent TB. At a willingness to pay of $1000 per DALY averted, 3HP is likely to be cost-effective relative to IPT only if the price of rifapentine can be greatly reduced (to approximately $20 per course) and high treatment completion (85%) can be achieved.

Conclusions

3HP may be a cost-effective alternative to IPT in high-burden settings, but cost-effectiveness depends on the price of rifapentine, achievable completion rates, and local willingness to pay.

CT colonography: determination of optimal CT technique using a novel colon phantom

The aim of this study was to determine the thickest slice at the lowest radiation dose for detection of colon polyps larger than 5mm in diameter at computed tomographic (CT) colonography. A colon phantom containing haustral folds, flexures, and straight segments was constructed of borosilicate. One hundred forty simulated polyps (5, 7, 10, and 12 mm) of various shapes (sessile, flat, and pedunculated) were attached at different colon locations (wall, base of fold, on the fold and fold tip). Polyps were positioned parallel, perpendicular, and oblique to the CT gantry. The air-filled phantom was scanned at different slice thicknesses (1.25-5 mm) and x-ray tube currents (5-308 mA). All polyps were identified in all data sets except one (1.25 mm slice thickness, 5 mA). In this acquisition, image noise reduced polyp visibility, and five of 140 (3%) polyps could not be identified. Unidentified polyps were 5 mm, flat or sessile in shape, located on the colon wall or base of the fold, and oblique or parallel to CT gantry. All tested CT techniques provided optimal polyp detection except settings at 1.25 mm and 5 mAs. Thin collimation (<5 mm) scans may not be necessary to detect clinically significant polyps.

Accuracy of models for indirect posterior restorations

Effects of materials and techniques on the accuracy of models used to make indirect restorations was measured using a 4-unit posterior model containing a MOD and full crown preparation. Improved stone and fast setting epoxy dies backed with either improved stone or a thermoplastic hot melt stone were made from single-viscosity addition silicone impressions. Technique variations included heating or not heating the impression, cooling and pouring dies and placement of the hot melt stone on set or unset epoxy. The dimensions of the MOD (L, W, H) and of the crown (W, H) dies were measured at 1 and 24 h. No clinically significant changes occurred between 1 and 24 h. The stone control reproduced the dimensions of the master die best, and models made by pouring epoxy into the impression followed by immediate pouring of the hot melt stone gave the poorest reproduction. Other variations in technique using epoxy for the anatomical portion gave no practical differences in accuracy. Of the epoxy dies, those prepared from a previously heated impression with hot melt poured after the epoxy set had the best values; however, epoxy dies were smaller than stone dies. The fast set epoxy was noteworthy for rapid processing and sharp detail, however, negative changes for W and H of the crown and H and L of the MOD showed that a die spacer would be essential in the preparation of indirect restorations.