Laboratory-Confirmed COVID-19 Among Adults Hospitalized With COVID-19–Like Illness with Infection-Induced or mRNA Vaccine-Induced SARS-CoV-2 Immunity

Nine States, January-September 2021

Catherine H. Bozio, PhD; Shaun J. Grannis, MD; Allison L. Naleway, PhD; Toan C. Ong, PhD; Kristen A. Butterfield, MPH; Malini B. DeSilva, MD; Karthik Natarajan, PhD; Duck-Hye Yang, PhD; Suchitra Rao, MBBS; Nicola P. Klein, MD, PhD; Stephanie A. Irving, MHS; Brian E. Dixon, PhD; Kristin Dascomb, MD, PhD; I-Chia Liao MPH; Sue Reynolds, PhD; Charlene McEvoy, MD; Jungmi Han; Sarah E. Reese, PhD; Ned Lewis, MPH; William F. Fadel, PhD; Nancy Grisel, MPP; Kempapura Murthy MBBS; Jill Ferdinands, PhD; Anupam B. Kharbanda, MD; Patrick K. Mitchell, ScD; Kristin Goddard, MPH; Peter J. Embi, MD; Julie Arndorfer, MPH; Chandni Raiyani, MPH; Palak Patel, MBBS; Elizabeth A. Rowley, DrPH; Bruce Fireman, MA; Nimish R. Valvi, DrPH, MBBS; Eric P. Griggs, MPH; Matthew E. Levy, PhD; Ousseny Zerbo, PhD; Rachael M. Porter, MPH; Rebecca J. Birch, MPH; Lenee Blanton, MPH; Sarah W. Ball, ScD; Andrea Steffens, MPH; Natalie Olson, MPH; Jeremiah Williams, MPH; Monica Dickerson, MPH; Meredith McMorrow, MD; Stephanie J. Schrag, DPhil; Jennifer R. Verani, MD; Alicia M. Fry, MD; Eduardo Azziz-Baumgartner, MD; Michelle Barron, MD; Manjusha Gaglani, MBBS; Mark G. Thompson, PhD; Edward Stenehjem, MD


Morbidity and Mortality Weekly Report. 2021;70(44):1539-1544. 

In This Article


In this multistate analysis of hospitalizations for COVID-19–like illness among adults aged ≥18 years during January–September 2021 whose previous infection or vaccination occurred 90–179 days earlier, the adjusted odds of laboratory-confirmed COVID-19 were higher among unvaccinated and previously infected patients than among those who were fully vaccinated with 2 doses of an mRNA COVID-19 vaccine without previous documentation of a SARS-CoV-2 infection. Secondary analyses that did not adjust for time since infection or vaccination or adjusted time since infection or vaccination differently as well as before and during Delta variant predominance produced similar results. These findings are consistent with evidence that neutralizing antibody titers after receipt of 2 doses of mRNA COVID-19 vaccine are high;[5,6] however, these findings differ from those of a retrospective records-based cohort study in Israel,†† which did not find higher protection for vaccinated adults compared with those with previous infection during a period of Delta variant circulation. This variation is possibly related to differences in the outcome of interest and restrictions on the timing of vaccination. The Israeli cohort study assessed any positive SARS-CoV-2 test result, whereas this study examined laboratory-confirmed COVID-19 among hospitalized patients. The Israeli cohort study also only examined vaccinations that had occurred 6 months earlier, so the benefit of more recent vaccination was not examined. This report focused on the early protection from infection-induced and vaccine-induced immunity, though it is possible that estimates could be affected by time. Understanding infection-induced and vaccine-induced immunity over time is important, particularly for future studies to consider.

In this study, the benefit of vaccination compared with infection without vaccination appeared to be higher for recipients of Moderna than Pfizer-BioNTech vaccine, which is consistent with a recent study that found higher vaccine effectiveness against COVID-19 hospitalizations for Moderna vaccine recipients than for Pfizer-BioNTech vaccine recipients.[7] In this study, the protective effect of vaccination also trended higher for adults aged ≥65 years than for those aged 18–64 years. However, considering the limited data by both product type and age, additional research is needed on the relative protection of vaccination versus infection without vaccination across demographic groups and vaccine products, as well as vaccination in previously infected persons.

The findings in this report are subject to at least seven limitations. First, although this analysis was designed to compare two groups with different sources of immunity, patients might have been misclassified. If SARS-CoV-2 testing occurred outside of network partners' medical facilities or if vaccinated persons are less likely to seek testing, some positive SARS-CoV-2 test results might have been missed and thus some patients classified as vaccinated and previously uninfected might also have been infected. In addition, despite the high specificity of COVID-19 vaccination status from these data sources, misclassification is possible. Second, the aOR could not be further stratified by time since infection or vaccination because of sparse data and limited ability to control for residual confounding that could be magnified within shorter intervals. The aOR that did not adjust for time might also be subject to residual confounding, particularly related to waning of both types of immunity. Third, selection bias might be possible if vaccination status influences likelihood of testing and if previous infection influences the likelihood of vaccination. Previous work from the VISION network did not identify systematic bias in testing by vaccination status, based on data through May 2021.[1] Fourth, residual confounding might exist because the study did not measure or adjust for behavioral differences between the comparison groups that could modify the risk of the outcome. Fifth, these results might not be generalizable to nonhospitalized patients who have different access to medical care or different health care–seeking behaviors, particularly outside of the nine states covered. Sixth, the statistical model incorporated the use of a weighted propensity score method which is subject to biases in estimates or standard errors if the propensity score model is misspecified. Numerous techniques were used to reduce potential suboptimal specification of the model, including but not limited to including a large set of covariates for machine learning estimation of propensity scores, including covariates in both regression and propensity models, ensuring large sample sizes and checking stability of weights, and conducting secondary analyses to assess robustness of results. Finally, the study assessed COVID-19 mRNA vaccines only; findings should not be generalized to the Janssen vaccine.

In this U.S.-based epidemiologic analysis of patients hospitalized with COVID-19–like illness whose previous infection or vaccination occurred 90–179 days earlier, vaccine-induced immunity was more protective than infection-induced immunity against laboratory-confirmed COVID-19, including during a period of Delta variant predominance. All eligible persons should be vaccinated against COVID-19 as soon as possible, including unvaccinated persons previously infected with SARS-CoV-2.