A Saliva-Based Serological and Behavioral Analysis of SARS-CoV-2 Antibody Prevalence in Howard County, Maryland.

The objective of the study was to estimate severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) seroprevalence in the Howard County, Maryland, general population and demographic subpopulations attributable to natural infection or coronavirus disease 2019 (COVID-19) vaccination and to identify self-reported social behaviors that may affect the likelihood of recent or past SARS-CoV-2 infection. A cross-sectional, saliva-based serological study of 2,880 residents of Howard County, Maryland, was carried out from July through September 2021. Natural SARS-CoV-2 infection prevalence was estimated by inferring infections among individuals according to anti-nucleocapsid immunoglobin G levels and calculating averages weighted by sample proportions of various demographics. Antibody levels between BNT162b2 (Pfizer-BioNTech) and mRNA-1273 (Moderna) recipients were compared. Antibody decay rate was calculated by fitting exponential decay curves to cross-sectional indirect immunoassay data. Regression analysis was carried out to identify demographic factors, social behaviors, and attitudes that may be linked to an increased likelihood of natural infection. The estimated overall prevalence of natural infection in Howard County, Maryland, was 11.9% (95% confidence interval, 9.2% to 15.1%), compared with 7% reported COVID-19 cases. Antibody prevalence indicating natural infection was highest among Hispanic and non-Hispanic Black participants and lowest among non-Hispanic White and non-Hispanic Asian participants. Participants from census tracts with lower average household income also had higher natural infection rates. After accounting for multiple comparisons and correlations between participants, none of the behavior or attitude factors had significant effects on natural infection. At the same time, recipients of the mRNA-1273 vaccine had higher antibody levels than those of BNT162b2 vaccine recipients. Older study participants had overall lower antibody levels compared with younger study participants. The true prevalence of SARS-CoV-2 infection is higher than the number of reported COVID-19 cases in Howard County, Maryland. A disproportionate impact of infection-induced SARS-CoV-2 positivity was observed across different ethnic/racial subpopulations and incomes, and differences in antibody levels across different demographics were identified. Taken together, this information may inform public health policy to protect vulnerable populations. IMPORTANCE We employed a highly innovative noninvasive multiplex oral fluid SARS-CoV-2 IgG assay to ascertain our seroprevalence estimates. This laboratory-developed test has been applied in NCI's SeroNet consortium, possesses high sensitivity and specificity according to FDA Emergency Use Authorization guidelines, correlates strongly with SARS-CoV-2 neutralizing antibody responses, and is Clinical Laboratory Improvement Amendments-approved by the Johns Hopkins Hospital Department of Pathology. It represents a broadly scalable public health tool to improve understanding of recent and past SARS-CoV-2 exposure and infection without drawing any blood. To our knowledge, this is the first application of a high-performance salivary SARS-CoV-2 IgG assay to estimate population-level seroprevalence, including identifying COVID-19 disparities. We also are the first to report differences in SARS-CoV-2 IgG responses by COVID-19 vaccine manufacturers (BNT162b2 [Pfizer-BioNTech] and mRNA-1273 [Moderna]). Our findings demonstrate remarkable consistency with those of blood-based SARS-CoV-2 IgG assays in terms of differences in the magnitude of SARS-CoV-2 IgG responses between COVID-19 vaccines.

Using machine learning on clinical data to identify unexpected patterns in groups of COVID-19 patients.

As clinicians are faced with a deluge of clinical data, data science can play an important role in highlighting key features driving patient outcomes, aiding in the development of new clinical hypotheses. Insight derived from machine learning can serve as a clinical support tool by connecting care providers with reliable results from big data analysis that identify previously undetected clinical patterns. In this work, we show an example of collaboration between clinicians and data scientists during the COVID-19 pandemic, identifying sub-groups of COVID-19 patients with unanticipated outcomes or who are high-risk for severe disease or death. We apply a random forest classifier model to predict adverse patient outcomes early in the disease course, and we connect our classification results to unsupervised clustering of patient features that may underpin patient risk. The paradigm for using data science for hypothesis generation and clinical decision support, as well as our triaged classification approach and unsupervised clustering methods to determine patient cohorts, are applicable to driving rapid hypothesis generation and iteration in a variety of clinical challenges, including future public health crises.