This study aims to examine the long-term outcomes in Veterans infected and uninfected with SARS-CoV-2 using electronic health record information and structured surveys.

Background: Early evidence suggests that the frequency and type of long term complications stemming from SARS-CoV-2 infection may depend on the severity of initial presentation. However, additional unidentified factors may also play a role including those related to the individual, the environment and/or the specific strain of the virus. The long-term health impacts of SARS-CoV-2 will have major implications for health care utilization, care processes and quality and costs of care for Veterans, both within the VA and in the community. Over 200,000 Veterans have had COVID-19. Although the long term consequences of this novel illness are unknown, it is clear they will often be significant. The overarching goal in the Long-Term Outcome Study (LTO) is to advance VA HSR&D research into the consequences and care of COVID-19 infection by developing and releasing integrated multi-modal data and results on long-term outcomes of COVID for use by VA clinical, research and operations communities. The investigators will do so via a coordinated program of data production and analysis using the VA electronic health record (EHR) and other administrative sources (e.g., Medicare), de novo longitudinal survey, and qualitative inquiry. This work will be closely coordinated with the COVID-19 Observational Research Collaboratory (CORC) Coordinating Center (CCC) and the burgeoning community of HSR&D-funded COVID researchers.

This study aims to examine the long-term outcomes in Veterans infected and uninfected with SARS-CoV-2.

Aim 1: Using VINCI-CDW data, patients will be identified who had a SARS-CoV-2 test since February 2020. The investigators will use a combination of traditional epidemiologic studies, and machine learning algorithms to determine the long term, multi-systemic and functional adverse outcomes and/or syndromes related to SARS-CoV-2. Members of the research team are currently developing methods to identify new onset and exacerbations in comorbidities following SARS-CoV-2 illness in Veterans.

Aim 2: Determine Veteran-Reported Outcome Effects Using Structured Survey on domains such as self- reported symptoms, disability, and financial impacts, measured repeatedly over 36 months.

Inclusion Criteria:

• All VA enrollees since February 2020, which includes all patients who were tested for SARS-CoV-2 and are included in the VA COVID-19 Shared Data Resource in CDW.

Exclusion Criteria:

• VA employees

All VA enrollees since February 2020, which includes all patients who were tested for SARS-CoV-2 and are included in the VA COVID-19 Shared Data Resource in CDW.

Beginning six months after the first survey is collected, updated releases of all structured survey items and basic summary scores will be released via VINCI for all VA investigators with appropriate permissions-but specific HIPAA-protected identifiers will not be included. The research information sheet and recruitment letter include the VA's recommended explicit language about this and the option to opt out.

Releases of survey items and scores will occur every six months and will be available indefinitely

Researchers that access this data for future research use may be national or international and affiliated or unaffiliated with the VA, and may only use the data for research purposes. Future studies will have appropriate data use agreements in place as well as receive approval by an IRB and other oversight committees prior to starting any research.

Baseline analysis included all Veterans who tested positive for COVID-19 and their matched comparators who consented to and responded to the first longitudinal survey cycle.

36 months

No adverse events because this was a non interventional study

While COVID-19 status was ascertained early in the pandemic through evidence of PCR test, we were unable to ascertain if uninfected comparators became infected over time due to uptake in home testing. This would have resulted in misclassification of the uninfected and biased our results towards the null.