Before 2020, primary objectives for the SUPERNOVA network included surveillance to estimate the incidence of norovirus and other causes of acute gastroenteritis in inpatient and outpatient settings, and active and passive surveillance to determine the inpatient burden of adult respiratory syncytial virus, influenza viruses, and other viruses that cause acute respiratory illness (Figure 2). Acute gastroenteritis surveillance using stool specimens ordered for clinician-directed diagnostic testing began at four sites in 2011, and expanded in 2015 to include medical chart review and collection of saliva and serum specimens, with a fifth site (Palo Alto) added in 2017 (16, 17). Acute respiratory illness surveillance with laboratory confirmation began as a pilot at one site (Houston) in 2016, with a second site (Los Angeles) added in 2017. Acute respiratory illness surveillance included: (1) identification of patients who had been admitted with a diagnosis consistent with acute respiratory illness and/or had received clinician-ordered viral respiratory diagnostic testing; (2) medical record reviews and chart abstractions to obtain demographic and clinical information including clinical course; and (3) collection of residual respiratory and serum specimens (18). By 2019–2020, surveillance for acute gastroenteritis and acute respiratory illness was including ~900 and ~1,600 patients, respectively.

Beginning in March 2020, SUPERNOVA shifted focus to expand acute respiratory illness surveillance to prospective inpatient surveillance with a focus on COVID-19 at all five participating VAMC sites, while pausing surveillance for acute gastroenteritis. Objectives included understanding characteristics of patients with COVID-19 and the spectrum of disease severity, COVID-19 hospitalization rates, underlying medical conditions that increase risk of severe outcomes, virus variants, immune responses, multisystem inflammatory syndrome in adults, and health care use among patients with COVID-19. Additionally, given inclusion of case and control groups, this surveillance platform was positioned to assess COVID-19 vaccine effectiveness among hospitalized Veterans.

Beginning in February 2020, clinician-directed testing and then universal inpatient screening at hospital admission for SARS-CoV-2 were implemented at the five VAMC sites. Patients with COVID-19 (i.e., cases) were eligible for inclusion in SUPERNOVA if they had a positive molecular SARS-CoV-2 test result within the first 72 hours of hospitalization (Figure 3). Cases were considered symptomatic if they had ≥1 COVID-19-like signs or symptoms documented in their medical records (e.g., admission history and physical examination, or emergency department note)¹.

To allow for case-control analyses, for each case, 1–2 inpatients without COVID-19 (i.e., controls) were targeted for inclusion in SUPERNOVA (Figure 3). In keeping with existing surveillance procedures for acute respiratory illness within SUPERNOVA, controls were identified from among other patients with respiratory symptoms. Patients were eligible for inclusion as a non-COVID-19 control if they had both a negative SARS-CoV-2 molecular test result (reverse transcription-polymerase chain reaction [RT-PCR] or isothermal nucleic acid amplification test [NAAT]) within 14 days prior to admission or within the first 72 hours of hospitalization, and, if they presented with ≥1 respiratory symptom² (primary controls) at admission. An additional set of controls without respiratory symptoms was included beginning in early 2021. Secondary controls, defined as persons admitted for a non-respiratory acute illness (e.g., chest pain, hypotension) with a negative test result for SARS-CoV-2 and without respiratory symptoms, were also included. Primary or secondary controls who later had a positive SARS-CoV-2 antigen or molecular test result during hospitalization would become ineligible as controls; however, controls could have a positive test result for a different pathogen (e.g., influenza). To ensure similar participation dates, controls were identified daily according to hospital admission dates falling within 1–14 days of COVID-19 case admission dates.

VAMC electronic health records (EHR) of participants with and without COVID-19 were reviewed. Using standardized case report forms, information was collected on patient demographic and clinical characteristics, including SARS-CoV-2 and other viral pathogen test results; history of vaccination against COVID-19 (including state immunization registry data, where available), influenza, and pneumococcal disease; history of present illness and clinical presentation with vital signs and laboratory values upon admission; past medical history including underlying medical conditions; imaging and interventions including intensive care unit admission and mechanical ventilation; treatments provided during hospitalization (e.g., anti-inflammatory agents, anticoagulants, antivirals, vasopressors, hemodialysis, and other therapies); hospital complications and discharge diagnoses using ICD-10 codes; advance directive code status, and disposition (including deaths). Medical records of both cases and controls were reviewed again 90 days after hospital discharge to assess outcomes including clinical and functional status, and additional health care use.

Residual clinical respiratory swab specimens and serum or plasma specimens were collected from both COVID-19 case-patients and controls where possible. Residual respiratory specimens were stored at −70°C before shipment to CDC for testing using CDC's 2019-Novel Coronavirus Real-Time RT-PCR Diagnostic Panel (19). Residual serum or plasma specimens (1–2 mL) were retrieved after clinically indicated testing was completed, divided into aliquots (minimum 250 microliters each) and frozen at < −20°C for further testing. Available laboratory testing at CDC included testing for SARS-CoV-2 antibodies, molecular diagnostics for select respiratory pathogens, or SARS-CoV-2 whole genome sequencing of respiratory specimens with qualifying cycle threshold values. Specimens with a Ct value <33 were submitted for sequencing, and those with genome coverage >90% had a confirmed lineage reported. Results of research-use only laboratory tests were not expected to be provided to VAMC patients or their clinicians.

Modified protocols were reviewed and approved by the VAMC Research and Development Committee and institutional review boards for each of the five sites; in addition, the activity was reviewed by CDC and conducted consistent with applicable federal law and policy³. Sites tailored a generic consent form to provide their site-specific information and include necessary authorizations for release of medical records, for clinical specimen collection, and for potential future testing.

Standardized case report forms and data dictionaries were developed and used across all participating sites. Data were entered into a standardized surveillance database in REDCap, a secure web application for building and managing online surveys and databases designed to comply with Health Insurance Portability and Accountability Act (HIPAA) regulations (20, 21). Original participant medical records located at the participating sites were linked to surveillance data solely through sequentially assigned participant identification numbers. Local data were stored behind secure institutional firewalls. Following federal data handling requirements, data were transferred to CDC using participant identification numbers; personally identifiable information, such as patient name or medical record number, was not shared with CDC or other sites.

In addition to simple descriptive analyses of data from COVID-19 cases, including median and interquartile range (IQR), formal statistical comparisons of data from cases and controls may be performed to estimate pathogen-specific relative risk of disease progression and mortality. For example, to estimate the effects of underlying medical conditions on adverse clinical outcomes including mortality, respiratory failure, mechanical ventilation, intensive care unit (ICU) admission, and length of hospital stay, we calculated adjusted risk ratios (aRR) and 95% confidence intervals (CI) for these outcomes among participants with and without COVID-19 hospitalized from February 23–October 31, 2020.

Vaccine effectiveness (VE) against COVID-19 hospitalization among Veterans may be estimated using data obtained from this surveillance system by comparing vaccination status between hospitalized adults with and without laboratory-confirmed SARS-CoV-2 infection. Using a modified, prospective, test-negative case-control design (23, 24) to identify COVID-19 cases and controls with COVID-like signs or symptoms identified through passive and active surveillance, VE can be calculated using multivariable logistic regression with case status as the outcome and vaccination status as the main predictor, adjusting for potential confounders (e.g., site, age, sex, race/ethnicity, and underlying health status), where estimated VE = (1–adjusted odds ratio⁴) ×100 (25). To address potential misclassification of controls, VE estimates will be calculated using both primary and secondary controls. SUPERNOVA is one of several surveillance platforms CDC is using to assess effectiveness of COVID-19 vaccines in various populations (26). SUPERNOVA data can be used to evaluate effectiveness of full or partial vaccination against COVID-19 as well as specific vaccine products, reflecting real-world use. Data from SUPERNOVA showed that, during February 1–August 6, 2021, VE of mRNA COVID-19 vaccines against COVID-19 hospitalization was 80% among adults aged ≥65 years compared with 95% among adults aged 18–64 years (27).

Genomic sequencing of specimens collected in SUPERNOVA may be used to identify emergence of novel respiratory pathogens such as SARS-CoV-2 variants, which are a cause of concern due to their potential to exacerbate the trajectory of the COVID-19 pandemic (28). Monitoring for variants, particularly among VA patients with multiple hospital admissions for COVID-19, may provide evidence regarding disease severity, potential reinfections, and variant influence on countermeasures including diagnostics, therapeutics, and vaccines. Among SUPERNOVA participants, Delta became the predominant SARS-CoV-2 variant in July 2021 (27).

SUPERNOVA data may be able to address whether antibodies to SARS-CoV-2 or other common coronaviruses might modulate COVID-19 illness severity, using serum representing a spectrum of disease from adults hospitalized with COVID-19.

Case series have been reported of Multisystem inflammatory syndrome in adults (MIS-A) associated with SARS-CoV-2 infection (29). Data on laboratory values and imaging studies from SUPERNOVA may be used to identify possible cases of MIS-A to target these cases for additional chart review and follow-up.

COVID-19 can result in late sequelae and prolonged illness (“long COVID,” or post-acute sequelae of SARS-CoV-2 infection), but relatively little is known about typical clinical course and health care use following hospital discharge for COVID-19 (30). SUPERNOVA data, which include information on disposition and health care use through at least 90 days following initial hospital discharge for both COVID-19 cases and controls, might help fill this knowledge gap.

Limitations to this surveillance approach include the reliance on current policy of universal screening for SARS-CoV-2 infection of inpatients at the participating sites to identify cases and controls; if these policies were to change, misclassification of COVID-19 cases and controls might be more likely. In addition, timeliness of reporting varies by pandemic phase and site; during pandemic surges in COVID-19 cases at respective sites, adjustments included prioritizing identification of cases rather than controls, and implementing flexible or extended deadlines for data entry and submission to CDC. Furthermore, although a test-negative design for assessing VE minimizes differences by healthcare seeking behavior, other COVID-19 risk mitigation measures could be relevant. Addition of an interview component could help identify differential behaviors associated with COVID-19 acquisition or transmission between cases and controls, such as compliance with non-pharmaceutical interventions (e.g., mask-wearing, physical distancing) or known close contact with a person with COVID-19 in the 14 days preceding symptom onset.