In this brief, we describe the characteristics of fully vaccinated hospitalized patients who have COVID-19 breakthrough infections, in comparison to people who are not fully vaccinated and hospitalized with COVID-19. This analysis is based on data from Cosmos, a HIPAA-defined Limited Data Set of more than 120 million patients from over 140 Epic organizations, including 250 hospitals across all 50 states and data on nearly 10 million admissions. We assessed over 120,000 COVID-19 hospitalizations in the Cosmos dataset from June 2021 through September 2021. (Adults in the U.S. generally could have been fully vaccinated and protected from most cases of severe illness or death due to COVID-19 by June 2021).

We categorize patients as being “fully vaccinated” after they have received the required dose(s) of a COVID-19 vaccine and were at least 14 days from their single-dose vaccine or their second dose in an mRNA series; we did not require a booster dose to be considered fully vaccinated. We consider a fully vaccinated person to have a “breakthrough” COVID-19 hospitalization if they were admitted to the hospital and had a COVID-19 diagnosis. We group everyone else hospitalized with a COVID-19 diagnosis, including partially vaccinated people, into the “not fully vaccinated” group. (This is consistent with other methodologies). The unit of our analysis is a hospitalization; in uncommon cases when people have multiple admissions for COVID-19, they are counted multiple times.  

Overall, we find that age is highly correlated with breakthrough hospitalizations. The majority of COVID-19 breakthrough hospitalizations were of elderly people (ages 65+), whereas the majority of people not fully vaccinated and hospitalized with COVID-19 were non-elderly adults (ages 18-64). In addition, a greater share of people hospitalized with a breakthrough COVID-19 infection had a comorbidity than people hospitalized with COVID-19 who were not fully vaccinated.

We also find that breakthrough COVID-19 hospitalizations included fewer COVID-related respiratory complications or treatments, suggesting fully vaccinated patients hospitalized with breakthrough COVID-19 may have been more likely to be hospitalized for unrelated reasons. If this is the case, it would mean that the gap in risk of COVID-19 hospitalization between vaccinated and unvaccinated people is even greater than previously known.

Findings

An overwhelming majority of COVID-19 hospital admissions were among unvaccinated or partially vaccinated people

From June to September 2021, 85% of COVID-19 hospitalizations were among those not fully vaccinated against COVID-19 and 15% were among fully vaccinated people. The share of COVID-19 hospitalizations among people not fully vaccinated was similar in June, July, August, and September 2021 (87%, 87%, 86% and 83%, respectively).

The share of COVID-19 hospitalizations by vaccination status in Cosmos data is consistent with a recent CDC MMWR study of 13 jurisdictions, which similarly found fully vaccinated people were 14% of COVID-19 hospitalizations in the June-July period.  

In the U.S., as of September 30, 2021, 56% of the population, 67% of the adults, and 83% of the elderly were fully vaccinated against COVID-19. People unvaccinated against COVID-19 make up a smaller share of the U.S. population (44%) but the vast majority (85%) of the hospitalizations with COVID-19.

Most people hospitalized with breakthrough COVID-19 infections were elderly

In June through September 2021, the majority of breakthrough COVID-19 hospitalizations (69%) were of people ages 65 or older. Only 10% of breakthrough COVID-19 hospitalizations were of people under age 50.

Unvaccinated and partially vaccinated people hospitalized with COVID-19, however, were much more likely to be younger adults. People ages 65 and older represent 3 in 10 hospitalizations of people not fully vaccinated (29%). People ages 49 and younger represented 4 in 10 (41%) of COVID-19 hospitalizations among not fully vaccinated patients.

There are several possible reasons for the differences in age distribution. A higher share of older adults have been fully vaccinated than young adults, and therefore may be more likely to be represented in breakthrough cases. Nonetheless, compared to the age distribution of fully vaccinated people in the U.S., people ages 65 and over make up a disproportionately large share of breakthrough COVID-19 hospitalizations. In the U.S., 25% of those fully vaccinated as of September 30, 2021 were 65 and older, whereas, in the Epic sample, 69% of the fully vaccinated hospitalized with breakthrough COVID-19 were 65 and older.  

Elderly adults make up a disproportionally large share of breakthrough COVID-19 hospitalizations relative to the vaccinated population

Older adults may be disproportionately represented in breakthrough hospitalizations because older adults are more at risk of severe illness from a COVID-19 infection and are more likely to have comorbidities, which also increase their risk of COVID-19 hospitalization. Similarly, older adults and people with certain conditions may be more likely to have weakened immune systems, and it may be that the COVID-19 vaccines offer less protection for this group. Another possible explanation could be that older adults were among the first to receive the COVID-19 vaccines and may have experienced a decline in immunity over time. However, these patterns may also be explained in part by the fact that older adults are more likely to be hospitalized for any cause and a COVID-19 infection may be incidental in some of these hospitalizations, as we discuss more below.

More people hospitalized with a breakthrough COVID-19 infection had a comorbidity than those not fully vaccinated and hospitalized with COVID-19

In June through September 2021, more fully vaccinated adults hospitalized with breakthrough COVID-19 had chronic conditions including, hypertension, diabetes, heart failure, or chronic obstructive pulmonary disease (COPD) than those hospitalized with COVID-19 who were not fully vaccinated (see methods). More adults hospitalized with COVID-19 who were not fully vaccinated were obese than those hospitalized with a breakthrough COVID-19 infection. These patterns hold true for both elderly and non-elderly adults, but it is worth noting that the vast majority of non-elderly and elderly adults hospitalized with COVID-19, regardless of their vaccination status, had at least one of these conditions. The CDC has recommended vaccination especially for older people and younger adults with certain medical conditions, such as heart disease, diabetes, lung disease, asthma and obesity have a greater risk of becoming severely ill if they get infected with COVID-19. 

Fully vaccinated people hospitalized with breakthrough COVID-19 infections may be more likely to be hospitalized for other reasons than those not fully vaccinated

Hospitalizations typically include multiple diagnoses for each patient. In this analysis, consistent with other methodologies, hospitalizations were included so long as the patient had a COVID-19 diagnosis (test confirmed COVID-19 diagnosis during or within 14-days of the hospital admission). However, some people may have been hospitalized primarily for a reason other than COVID-19. For example, among the breakthrough hospitalizations reported to the CDC, 15% of patients were not experiencing COVID-19 symptoms or were admitted for a reason other than COVID-19. However, the CDC has not published similar data for unvaccinated people hospitalized with a COVID-19 diagnosis.

To see whether there are differences in the apparent primary reason for hospitalization between vaccinated and unvaccinated people hospitalized with a COVID-19 diagnosis, we identified four key complications and treatments based on clinical recommendations: presence of viral pneumonia or respiratory failure diagnosis, ventilator support, or treatment with dexamethasone. Although this is an incomplete list of indications, hospitalizations with a COVID-19 diagnosis primarily because of COVID-19, regardless of the patient’s vaccination status, may include at least one of these treatments or complications. These indicators were selected because they were clearly identifiable in Cosmos data.

We find that fewer people hospitalized with a breakthrough COVID-19 diagnosis had one of these complications or treatments associated with COVID-19 than patients who were not fully vaccinated and were hospitalized with COVID-19. While 72% of COVID-19 hospitalizations of people who were not fully vaccinated included a diagnosis of viral pneumonia, only 57% of breakthrough hospitalizations did. Similarly, while 69% of COVID-19 hospitalizations of people who were not fully vaccinated included dexamethasone treatment, only 55% of breakthrough hospitalizations did. Fewer fully vaccinated people hospitalized with COVID-19 required ventilator support or experienced respiratory failure than were patients who were not fully vaccinated.

While most people, regardless of vaccination status, who are hospitalized with a COVID-19 diagnosis appear to be undergoing treatment or experiencing complications consistent with COVID-19, these findings suggest that people hospitalized with breakthrough COVID-19 infections may be more likely to be primarily hospitalized for a reason unrelated to COVID-19 than are patients who are not fully vaccinated.   

Among people hospitalized with a COVID-19 diagnosis who had at least one of these additional COVID-19 indications during the admission (i.e., among those who had viral pneumonia or respiratory failure or were treated with ventilator support or dexamethasone), 13% were fully vaccinated and 87% were not fully vaccinated.

Adults hospitalized with COVID-19 who were fully vaccinated had shorter hospital stays than those not fully vaccinated

Adults hospitalized with COVID-19 who were fully vaccinated had shorter average hospital stays than those who were not fully vaccinated within the same age group. For example, among patients over age 65, the median COVID-19 hospital stay was 1.1 days shorter for those who were fully vaccinated (5.6 days) than for those not fully vaccinated (6.7 days).

Discussion

Despite the widespread availability of COVID-19 vaccines, in September, COVID-19 was the second-leading cause of death in the U.S. overall and the first leading cause of death among people age 35 to 54.

In June through September 2021, fully vaccinated people with breakthrough COVID-19 represented 15% of patients hospitalized with COVID-19. Although people not fully vaccinated continue to represent the vast majority of COVID-19 hospitalizations, understanding the characteristics of people with breakthrough hospitalizations can help inform ongoing policymaking around the pandemic response and can also inform individuals about their potential risks. Our findings show that people hospitalized with COVID-19 who were fully vaccinated are more likely to have other risk factors for serious illness such as older age or a chronic condition like hypertension or diabetes.

This analysis has several limitations. First, it is based on a convenience sample and it is not weighted to match the U.S. population. However, our results closely mirror those of the CDC. Another limitation of this analysis, discussed further in the Methods section, is that COVID-19 vaccination status is not always reported. Additionally, although we are able to identify patients with a COVID-19 diagnosis, some patients in both the fully vaccinated group and the not fully vaccinated group may have been hospitalized primarily for a reason other than COVID-19.

While we do not always know the primary reason for a hospitalization, it appears that people hospitalized with breakthrough COVID-19 infections may be more likely than unvaccinated people to be hospitalized for a reason other than their COVID-19 infection. Hospitalizations among fully vaccinated people with a breakthrough COVID-19 infection were less likely to include treatments related to COVID-19 or other respiratory diagnoses than hospitalizations with COVID-19 among people who were not fully vaccinated. Fully vaccinated patients with breakthrough COVID-19 infections also had shorter average hospital stays than patients who were not fully vaccinated. Though there remains considerable attention on breakthrough COVID-19 cases, the vast majority of serious COVID-19 hospitalizations and deaths remain among unvaccinated patients. An overwhelming majority of the COVID-19 hospitalizations and deaths since June 2021 have been among people not yet vaccinated and more than 163,000 deaths could have been avoided with vaccination.

METHODS

This analysis was done using Cosmos, a HIPAA-defined limited data set of electronic health records from over 140 Epic organizations. Estimates of COVID-19 hospitalizations by vaccination status in this report are based on this population and not weighted to be nationally representative.

Hospital admissions with COVID-19 were defined as admissions with either a documented COVID-19 diagnosis (ICD-10 diagnosis code U07.01) on the hospitalization record or admissions with other respiratory diagnoses involving a patient who tested positive for COVID-19 or received a COVID-19 diagnosis within 14 days of admission. When people have multiple admissions for COVID-19, each admission is counted as a unique observation.

Patients were categorized as “fully vaccinated” if they had received the required dose(s) of a COVID-19 vaccine and were at least 14 days from their single-dose vaccine or their second dose in an mRNA series; we did not require a booster dose to be considered fully vaccinated. In the Cosmos dataset, there were a total of 8,180, 17,348, 50,359, and 47,266 admissions with a COVID-19 diagnosis in June, July, August, and September 2021, respectively. Of the 123,153 hospitalizations with COVID-19 in the Cosmos dataset in June to September 2021, 18,364 (15%) hospitalizations were among fully vaccinated people with breakthrough COVID-19 infections.

Patients’ vaccination status was determined based on information in Cosmos. Cosmos includes COVID-19 vaccination status if COVID-19 vaccine was: 1) administered or patient-reported using Epic workflows at a Cosmos contributor, 2) recorded in a state immunization registry and queried at a Cosmos contributor, 3) included in a claim received through external payer claims at a Cosmos contributor, or 4) documented by some method above at another organization that is queried by a Cosmos contributor using Epic’s Care Everywhere exchange. The share of hospitalizations with COVID-19 by vaccination status in the Cosmos dataset are consistent with the CDC data on 13 jurisdictions in the U.S.

Comorbidities were identified based on relevant codes during the admission, and therefore, may not have been present before the admission (i.e., may not be pre-existing conditions). Hypertension includes a diagnosis of essential hypertension (I10), whether controlled or uncontrolled. COVID-related complications and treatment during the admission were selected based on clinical recommendations.

We also used CDC data on COVID-19 vaccination rates for the U.S. population and HHS data on COVID-19 hospitalizations among adults to assess overall vaccination and hospitalization rates in the U.S.

Justin Lo, PhD, is with Epic Research and Dustin Cotliar, MD, MPH, is an independent consultant. Krutika Amin, PhD,  Matthew Rae, MPH, and Cynthia Cox, MPH, are with KFF.