Efficient and robust propensity‐score‐based methods for population inference using epidemiologic cohorts

Dynamics of SARS-CoV-2 Seroprevalence in a Large US population Over a Period of 12 Months

Due to a combination of asymptomatic or undiagnosed infections, the proportion of the United States population infected with SARS-CoV-2 was unclear from the beginning of the pandemic. We previously established a platform to screen for SARS-CoV-2 positivity across a representative proportion of the US population, from which we reported that almost 17 million Americans were estimated to have had undocumented infections in the Spring of 2020. Since then, vaccine rollout and prevalence of different SARS-CoV-2 variants have further altered seropositivity trends within the United States population. To explore the longitudinal impacts of the pandemic and vaccine responses on seropositivity, we re-enrolled participants from our baseline study in a 6- and 12- month follow-up study to develop a longitudinal antibody profile capable of representing seropositivity within the United States during a critical period just prior to and during the initiation of vaccine rollout. Initial measurements showed that, since July 2020, seropositivity elevated within this population from 4.8% at baseline to 36.2% and 89.3% at 6 and 12 months, respectively. We also evaluated nucleocapsid seropositivity and compared to spike seropositivity to identify trends in infection versus vaccination relative to baseline. These data serve as a window into a critical timeframe within the COVID-19 pandemic response and serve as a resource that could be used in subsequent respiratory illness outbreaks.

Variable inclusion strategies through directed acyclic graphs to adjust health surveys subject to selection bias for producing national estimates

Along with the rapid emergence of web surveys to address time-sensitive priority topics, various propensity score (PS)-based adjustment methods have been developed to improve population representativeness for nonprobability- or probability-sampled web surveys subject to selection bias. Conventional PS-based methods construct pseudo-weights for web samples using a higher-quality reference probability sample. The bias reduction, however, depends on the outcome and variables collected in both web and reference samples. A central issue is identifying variables for inclusion in PS-adjustment. In this paper, directed acyclic graph (DAG), a common graphical tool for causal studies but largely under-utilized in survey research, is used to examine and elucidate how different types of variables in the causal pathways impact the performance of PS-adjustment. While past literature generally recommends including all variables, our research demonstrates that only certain types of variables are needed in PS-adjustment. Our research is illustrated by NCHS' Research and Development Survey, a probability-sampled web survey with potential selection bias, PS-adjusted to the National Health Interview Survey, to estimate U.S. asthma prevalence. Findings in this paper can be used by National Statistics Offices to design questionnaires with variables that improve web-samples' population representativeness and to release more timely and accurate estimates for priority topics.

Undiagnosed SARS-CoV-2 seropositivity during the first 6 months of the COVID-19 pandemic in the United States

Asymptomatic SARS-CoV-2 infection and delayed implementation of diagnostics have led to poorly defined viral prevalence rates in the United States and elsewhere. To address this, we analyzed seropositivity in 9089 adults in the United States who had not been diagnosed previously with COVID-19. Individuals with characteristics that reflected the U.S. population (n = 27,716) were selected by quota sampling from 462,949 volunteers. Enrolled participants (n = 11,382) provided medical, geographic, demographic, and socioeconomic information and dried blood samples. Survey questions coincident with the Behavioral Risk Factor Surveillance System survey, a large probability-based national survey, were used to adjust for selection bias. Most blood samples (88.7%) were collected between 10 May and 31 July 2020 and were processed using ELISA to measure seropositivity (IgG and IgM antibodies against SARS-CoV-2 spike protein and the spike protein receptor binding domain). The overall weighted undiagnosed seropositivity estimate was 4.6% (95% CI, 2.6 to 6.5%), with race, age, sex, ethnicity, and urban/rural subgroup estimates ranging from 1.1% to 14.2%. The highest seropositivity estimates were in African American participants; younger, female, and Hispanic participants; and residents of urban centers. These data indicate that there were 4.8 undiagnosed SARS-CoV-2 infections for every diagnosed case of COVID-19, and an estimated 16.8 million infections were undiagnosed by mid-July 2020 in the United States.

Adjusted logistic propensity weighting methods for population inference using nonprobability volunteer-based epidemiologic cohorts

Many epidemiologic studies forgo probability sampling and turn to nonprobability volunteer-based samples because of cost, response burden, and invasiveness of biological samples. However, finite population (FP) inference is difficult to make from the nonprobability sample due to the lack of population representativeness. Aiming for making inferences at the population level using nonprobability samples, various inverse propensity score weighting methods have been studied with the propensity defined by the participation rate of population units in the nonprobability sample. In this article, we propose an adjusted logistic propensity weighting (ALP) method to estimate the participation rates for nonprobability sample units. The proposed ALP method is easy to implement by ready-to-use software while producing approximately unbiased estimators for population quantities regardless of the nonprobability sample rate. The efficiency of the ALP estimator can be further improved by scaling the survey sample weights in propensity estimation. Taylor linearization variance estimators are proposed for ALP estimators of FP means that account for all sources of variability. The proposed ALP methods are evaluated numerically via simulation studies and empirically using the naïve unweighted National Health and Nutrition Examination Survey III sample, while taking the 1997 National Health Interview Survey as the reference, to estimate the 15-year mortality rates.