Mixed methods approach to examining the implementation experience of a phone-based health research survey investigating risk factors for SARS-CoV-2 infection in California (preprint)

Objective: To describe the implementation of a test-negative design case-control study in California during the Coronavirus Disease 2019 (COVID-19) pandemic. Methods: Between February 24, 2021 - February 24, 2022, 34 interviewers called 38,470 SARS-CoV-2-tested Californians to enroll 1,885 cases and 1,871 controls in a 20-minute telephone survey. We estimated adjusted odds ratios for answering the phone and consenting to participate using mixed effects logistic regression. We used a web-based anonymous survey to compile interviewer experiences. Results: Cases had 1.29-fold (95% CI: 1.24-1.35) higher adjusted odds of answering the phone and 1.69-fold (1.56-1.83) higher adjusted odds of consenting to participate compared to controls. Calls placed from 4pm to 6pm had the highest adjusted odds of being answered. Interviewers who faced participants with dire need for social services or harassment experienced poor mental health. Conclusions: We suggest calling during afternoons and allocating more effort towards enrolling controls when designing a case-control study. Remaining adaptive to the dynamic needs of the team is critical to a successful study, especially in a pandemic setting.

Receipt of COVID-19 and seasonal influenza vaccines in California (USA) during the 2021-2022 influenza season (preprint)

Background Despite lower circulation of influenza virus throughout 2020-2022 during the COVID-19 pandemic, seasonal influenza vaccination has remained a primary tool to reduce influenza-associated illness and death. The relationship between the decision to receive a COVID-19 vaccine and/or an influenza vaccine is not well understood. Methods We assessed predictors of receipt of 2021-2022 influenza vaccine in a secondary analysis of data from a case-control study enrolling individuals who received SARS-CoV-2 testing. We used mixed effects logistic regression to estimate factors associated with receipt of seasonal influenza vaccine. We also constructed multinomial adjusted marginal probability models of being vaccinated for COVID-19 only, seasonal influenza only, or both as compared with receipt of neither vaccination. Results Among 1261 eligible participants recruited between 22 October 2021 - 22 June 2022, 43% (545) were vaccinated with both seasonal influenza vaccine and >1 dose of a COVID-19 vaccine, 34% (426) received >1 dose of a COVID-19 vaccine only, 4% (49) received seasonal influenza vaccine only, and 19% (241) received neither vaccine. Receipt of >1 COVID-19 vaccine dose was associated with seasonal influenza vaccination (adjusted odds ratio [aOR]: 3.72; 95% confidence interval [CI]: 2.15-6.43); this association was stronger among participants receiving >1 COVID-19 booster dose (aOR=16.50 [10.10-26.97]). Compared with participants testing negative for SARS-CoV-2 infection, participants testing positive had lower odds of receipt of 2021-2022 seasonal influenza vaccine (aOR=0.64 [0.50-0.82]). Conclusions Recipients of a COVID-19 vaccine were more likely to receive seasonal influenza vaccine during the 2021-2022 season. Factors associated with individuals' likelihood of receiving COVID-19 and seasonal influenza vaccines will be important to account for in future studies of vaccine effectiveness against both conditions. Participants who tested positive for SARS-CoV-2 in our sample were less likely to have received seasonal influenza vaccine, suggesting an opportunity to offer influenza vaccination before or after a COVID-19 diagnosis.

Effectiveness of nirmatrelvir-ritonavir against hospital admission: a matched cohort study in a large US healthcare system (preprint)

ABSTRACT Background In the United States, oral nirmatrelvir-ritonavir (Paxlovid™) is authorized for use among patients aged ≥12 years with mild-to-moderate SARS-CoV-2 infection who are at risk for progression to severe COVID-19, including hospitalization. However, effectiveness under real-world conditions has not been well established. Methods We undertook a matched, observational cohort study of non-hospitalized individuals with SARS-CoV-2 infection to compare outcomes between those who received or did not receive nirmatrelvir-ritonavir within the Kaiser Permanente Southern California healthcare system. Individuals were matched on testing date, age, sex, treatment/care setting, symptoms status (including presence or absence of acute COVID-19 symptoms at testing, and time from symptom onset to testing), history of vaccination and SARS-CoV-2 infection, Charlson comorbidity index, and prior-year healthcare utilization. Time to hospital admission was compared between matched COVID-19 cases who received or did not receive nirmatrelvir-ritonavir. Primary analyses evaluated treatment effectiveness against any hospital admission and acute respiratory infection (ARI)-associated hospital admission, with dispense occurring 0–5 days symptom onset. Secondary analyses evaluated effectiveness against the same endpoints for all treatment dispenses. We measured treatment effectiveness as (1–adjusted hazards ratio [aHR])×100%, estimating the aHR via Cox proportional hazards models accounting for match strata and additional patient characteristics. Results Analyses included 4,329 nirmatrelvir-ritonavir recipients and 20,980 matched non-recipients who were followed ≥30 days after a positive SARS-CoV-2 outpatient test. Overall, 23,603 (93.3%) and 19,564 (78.1%) of 25,039 participants had received ≥2 and ≥3 COVID-19 vaccine doses, respectively. A total of 23,858 (94.2% of 25,039) patients were symptomatic at the point of testing, with a 2.1 day mean time from symptom onset to testing. For patients dispensed nirmatrelvir-ritonavir 0–5 days after symptom onset, effectiveness in preventing all hospital admissions was 88.1% (95% confidence interval: 49.0–97.5%) over 15 days and 71.9% (25.3–90.0%) over 30 days, respectively. Effectiveness in preventing ARI-associated hospital admissions was 88.3% (12.9–98.8%) and 87.3% (18.3–98.5%) over 15 and 30 days, respectively. In expanded analyses that included patients receiving treatment at any point during their clinical course, effectiveness was 86.6% (54.9–96.3%) and 78.0% (46.2–91.4%) in preventing all hospital admissions over 15 and 30 days, respectively, and 93.7% (52.5–99.4%) and 92.8% (53.9–99.1%) in preventing ARI-associated hospital admissions over 15 and 30 days. Subgroup analyses identified similar effectiveness estimates among patients who had received ≥2 COVID-19 vaccine doses. Implications In a real-world setting with high levels of COVID-19 vaccine and booster uptake, receipt of nirmatrelvirritonavir 0–5 days after symptom onset was associated substantial reductions in risk of hospital admission among individuals testing positive for SARS-CoV-2 infection in outpatient settings. Funding US Centers for Disease Control and Prevention, US National Institutes of Health

Impact of individual-level characteristics and transmission mitigation behaviors on SARS-CoV-2 infection and seroprevalence in a large Northern California Bay Area cohort (preprint)

Comprehensive data on transmission mitigation behaviors and SARS-CoV-2 infection and serostatus are needed from large, community-based cohorts to identify SARS-CoV-2 risk factors and impact of public health measures. From July 2020 to March 2021, {approx}5,500 adults from the East Bay Area, California were followed over three data collection rounds. We estimated the prevalence of antibodies from SARS-CoV-2 infection and COVID-19 vaccination, and self-reported COVID-19 test positivity. Population-adjusted SARS-CoV-2 seroprevalence was low, increasing from 1.03% (95% CI: 0.50-1.96) in Round 1 (July-September 2020), to 1.37% (95% CI: 0.75-2.39) in Round 2 (October-December 2020), to 2.18% (95% CI: 1.48-3.17) in Round 3 (February-March 2021). Population-adjusted seroprevalence of COVID-19 vaccination was 21.64% (95% CI: 19.20-24.34) in Round 3. Despite >99% of participants reporting wearing masks, non-Whites, lower-income, and lower-educated individuals had the highest SARS-CoV-2 seroprevalence and lowest vaccination seroprevalence. Our results demonstrate that more effective policies are needed to address these disparities and inequities.

Predictors of SARS-CoV-2 infection following high-risk exposure (preprint)

ABSTRACT Background Non-pharmaceutical interventions (NPIs) are recommended for COVID-19 mitigation. However, the effectiveness of NPIs in preventing SARS-CoV-2 transmission remains poorly quantified. Methods We conducted a test-negative design case-control study enrolling cases (testing positive for SARS-CoV-2) and controls (testing negative) with molecular SARS-CoV-2 diagnostic test results reported to California Department of Public Health between 24 February-26 September, 2021. We used conditional logistic regression to assess predictors of case status among participants who reported contact with an individual known or suspected to have been infected with SARS-CoV-2 (“high-risk exposure”) within ≤14 days of testing. Results 643 of 1280 cases (50.2%) and 204 of 1263 controls (16.2%) reported high-risk exposures ≤14 days before testing. Adjusted odds of case status were 2.94-fold (95% confidence interval: 1.66-5.25) higher when high-risk exposures occurred with household members (vs. other contacts), 2.06-fold (1.03-4.21) higher when exposures occurred indoors (vs. not indoors), and 2.58-fold (1.50-4.49) higher when exposures lasted ≥3 hours (vs. shorter durations) among unvaccinated and partially-vaccinated individuals; excess risk associated with such exposures was mitigated among fully-vaccinated individuals. Mask usage by participants or their contacts during high-risk exposures reduced adjusted odds of case status by 48% (8-72%). Adjusted odds of case status were 68% (32-84%) and 77% (59-87%) lower for partially- and fully-vaccinated participants, respectively, than for unvaccinated participants. Benefits of mask usage were greatest when exposures lasted ≥3 hours, occurred indoors, or involved non-household contacts. Conclusions NPIs reduced the likelihood of SARS-CoV-2 infection following high-risk exposure. Vaccine effectiveness was substantial for partially and fully vaccinated persons. KEY POINTS SARS-CoV-2 infection risk was greatest for unvaccinated participants when exposures to known or suspected cases occurred indoors or lasted ≥3 hours. Face mask usage when participants were exposed to a known or suspect case reduced odds of infection by 48%.

Use of recently vaccinated individuals to detect bias in test-negative case-control studies of COVID-19 vaccine effectiveness (preprint)

Post-authorization observational studies play a key role in understanding COVID-19 vaccine effectiveness following the demonstration of efficacy in clinical trials. While bias due to confounding, selection bias, and misclassification can be mitigated through careful study design, unmeasured confounding is likely to remain in these observational studies. Phase III trials of COVID-19 vaccines have shown that protection from vaccination does not occur immediately, meaning that COVID-19 risk should be similar in recently vaccinated and unvaccinated individuals, in the absence of confounding or other bias. Several studies have used the estimated effectiveness among recently vaccinated individuals as a negative control exposure to detect bias in vaccine effectiveness estimates. In this paper we introduce a theoretical framework to describe the interpretation of such a bias-indicator in test-negative studies, and outline assumptions that would allow the use of recently vaccinated individuals to correct bias due to unmeasured confounding.

Statistical deconvolution for inference of infection time series (preprint)

Accurate measurement of daily infection incidence is crucial to epidemic response. However, delays in symptom onset, testing, and reporting obscure the dynamics of transmission, necessitating methods to remove the effects of stochastic delays from observed data. Existing estimators can be sensitive to model misspecification and censored observations; many analysts have instead used methods that exhibit strong bias or do not account for delays. We develop an estimator with a regularization scheme to cope with these sources of noise, which we term the Robust Incidence Deconvolution Estimator (RIDE). We validate RIDE on synthetic data, comparing accuracy and stability to existing approaches. We then use RIDE to study COVID-19 records in the United States, and find evidence that infection estimates from reported cases can be more informative than estimates from mortality data. To implement these methods, we release incidental, a ready-to-use R implementation of our estimator that can aid ongoing efforts to monitor the COVID-19 pandemic.

Epidemiology and transmission dynamics of COVID-19 in two Indian states (preprint)

Although most COVID-19 cases have occurred in low-resource countries, there is scarce information on the epidemiology of the disease in such settings. Comprehensive SARS-CoV-2 testing and contact-tracing data from the Indian states of Tamil Nadu and Andhra Pradesh reveal stark contrasts from epidemics affecting high-income countries, with 92.1% of cases and 59.7% of deaths occurring among individuals <65 years old. The per-contact risk of infection is 9.0% (95% confidence interval: 7.5-10.5%) in the household and 2.6% (1.6-3.9%) in the community. Superspreading plays a prominent role in transmission, with 5.4% of cases accounting for 80% of infected contacts. The case-fatality ratio is 1.3% (1.0-1.6%), and median time-to-death is 5 days from testing. Primary data are urgently needed from low- and middle-income countries to guide locally-appropriate control measures.