SARS-CoV-2 breakthrough infections in a retrospective cohort of vaccinated university students and employees

ABSTRACT

Breakthrough SARS-CoV-2 infections, despite widespread vaccination efforts, remain a critical concern in densely populated environments such as universities, where transmission dynamics are influenced by close contact and shared facilities. In this retrospective cohort study conducted at the University of Miami from December 2020 through October 2021, we analyzed infection rates among 38,783 students (61%) and employees (39%) to evaluate vaccine effectiveness and the necessity of ongoing surveillance. Our findings revealed that vaccinated individuals had a lower infection rate (1.8%) compared to their unvaccinated counterparts (2.6%), with students exhibiting fewer breakthrough infections (1.5%) than employees (5.6%). Notably, the median duration from full vaccination to breakthrough infection was shorter in students, suggesting potential differences in exposure risk or immune response. These results highlight the protective impact of vaccines in this population while emphasizing the importance of tailored public health strategies, particularly in institutional settings where behavioral and occupational differences contribute to infection disparities. Continuous monitoring remains essential to refine mitigation efforts and enhance pandemic preparedness.

Introduction

Breakthrough SARS-CoV-2 occurs when fully vaccinated individuals contract the virus, despite vaccination. This phenomenon has been well documented, notably among healthcare workers.¹ As additional doses are now recommended, it is important to assess the optimal timing, vaccine type, and target populations in order to enhance protection and reduce the risk of future outbreaks. Understanding the extent of vaccine-induced protection against emerging SARS-CoV-2 variants remains a critical area of investigation. The interplay between vaccine-induced immunity and emerging variants of the virus underscores the complexity of the pandemic. The efficacy of vaccines, while robust against initial strains, has shown variability in protection against new variants.² This variability highlights the need for ongoing surveillance and adaptation of vaccine strategies to address the evolving viral landscape. Notably, current SARS-CoV-2 vaccines do not confer sterilizing immunity, a limitation common to most respiratory virus vaccines. Neutralizing antibody titers against the SARS-CoV-2 spike protein have been informative to describe waning immunity, but breakthrough infections are another method to determine vaccine durability.^3,4 Studies have looked at breakthrough infections in medical students in China and Iran however no studies have compared students and employees in the same environment.^5,6 Universities represent microcosms of broader communities but with unique challenges and characteristics such as dense living and social environments that can accelerate the transmission of infectious diseases like COVID-19.⁷

This study aims to elucidate the proportion of breakthrough SARS-CoV-2 infections among a retrospective cohort of students and employees at the University of Miami from December 2020 through October 2021. We conducted a retrospective analysis of over 38,000 students and employees, examining infection rates by vaccination status. Our findings reveal important differences in breakthrough infection patterns between these two university populations. This research seeks to provide insights into the effectiveness of current vaccines and the critical role of continuous monitoring in a university setting.

Methods

This retrospective cohort study was conducted at the University of Miami, Miami, FL, covering the period from December 20, 2020, to October 7, 2021. It focused on tracking COVID-19 vaccinations and infections among students and employees.

The study subjects were students who were on campus in class or residence halls anytime during the day. Employees were non-students who came on campus to do their job (teaching, administrative, service). All students and employees had a record in the electronic health record (EHR) Epic, where results of testing and vaccination were recorded.

Vaccination data were sourced from the University and state-wide records in Florida, and were automatically integrated into the EHR. Three FDA-approved vaccines – Pfizer, BioNTech (BNT162b2) referred to as Pfizer, Moderna (mRNA-1273) referred to as Moderna, and Janssen COVID 19 Vaccine (Ad26.COV2.S) referred to as Janssen – were offered to the university community, with Pfizer being the first available. While vaccination was mandatory for all employees and strongly encouraged for students, unvaccinated students or those who did not disclose their vaccination status had to participate in weekly asymptomatic surveillance testing whenever they were on campus during the term.

Fully vaccinated was defined as having received the FDA recommend number of injections for each brand of vaccine: either 2 for Pfizer or 2 for Moderna or 1 for Janssen.

SARS-CoV-2 testing at the university was conducted in specialized testing laboratories. This included diagnostic testing, which required midnasal swab samples for reverse transcriptase polymerase chain reaction (RT-PCR) tests, with results typically available within 24 hours.⁸ The testing protocol was uniformly applied to both asymptomatic individuals undergoing weekly testing per university protocols and to symptomatic individuals or those identified as close contacts of confirmed cases. Individuals in these latter groups were instructed to contact a 24/7 hotline and were tested using the same procedures as the asymptomatic group.

The protocol was fully approved by the University Institutional Review Board, approval number 20221361.

Statistical analyses

Continuous variables with normal distribution were expressed as mean ± standard deviation. Shapiro-Wilk test was used for normality assessment; asymmetrically distributed continuous variables were expressed as median and interquartile range (25th–75th); and categorical variables were expressed as absolute and relative frequencies. For between-group comparisons, Student’s t test or ANOVA were used for symmetrically distributed variables; Mann–Whitney U test or Kruskal Wallis for asymmetrically distributed variables. Fisher’s exact test or Pearson’s chi-square were performed in order to analyze the association among categorical variables. Multivariate analysis was performed using logistic regression model, considering COVID-19 infection as the dependent variable while the covariates included age, vaccine brand, sex, and relation to the university (student or employee). P values of <.05 (two-tailed) were significant and all data were analyzed on SPSS version 28.0 (SPSS Inc., Chicago, IL, USA) or R Statistics.

Results

A total of 38,835 subjects participated in this study, comprising 23,691 (61%) students and 15,147 (39%) employees, with a female representation of 65% and a mean age of 30.6 (±14.6) years. Table 1 provides a comparison of characteristics between students and employees. Out of the total cohort, 75% (n = 29,139) were fully vaccinated. Eight percent (n = 3,108) were partially vaccinated, meaning they had not yet received the second dose when required. A total of 17% of the individuals were unvaccinated (n = 6,588).

Table 1.

Descriptive characteristics comparing students and employees.

	Students	Employees	P-value
Age	23 ± 9	44 ± 13	<.001
Female sex, n (%)	14,593 (61.6)	10,633 (70.2)	<.001
Unvaccinated, n (%)	2,771 (11.7)	3,817 (25.2)	.514
Partially vaccinated, n (%)	2,321 (9.8)	787 (5.2)	<.001
Fully vaccinated, n (%)	18,597 (78.5)	10,542 (69.6)	<.001
Vaccine brand, n (%)			.413
Pfizer BioNTech(BNT162b2)	13,267 (69.1)	5,875 (56.9)
Moderna (mRNA-1273)	4,011 (20.9)	4,009 (38.8)
Janssen (Ad26.COV2.S)	1,299 (6.8)	429 (4.2)
Unknown	617 (2.6)	10 (0.1)

The median interval between the first and second dose was 22 days (IQR: 22–28). The distribution of vaccines administered was as follows: Pfizer (64.8%), Moderna (27.2%), Janssen (5.9%), and unknown (2.1%). Most employees began vaccinating in December 2020 compared to the student population who received their vaccines 5 months later in April 2021 when the FDA and CDC approved vaccination of individuals not in the classic risk groups for COVID 19. It is important to note that while the vaccination program for students officially began in April, some students received their vaccinations earlier from external sources. April 2021 coincided with the 4th wave of COVID in our academic health center with the 5th surge of infections starting in the last week of June and peaking in mid-August 2021 (Figure 1).

Figure 1.

Waves of COVID-19 infection in the academic health center: this figure illustrates the trends of COVID-19 infection waves within the academic health center. The red bars represent the actual number of COVID-19 cases recorded over time, while the green line depicts the predicted number of cases based on epidemiological modeling. Arrows indicate the significant time points at which vaccination campaigns were initiated for employees and students. The x-axis denotes the time in months, and the y-axis represents the number of COVID-19 cases. Analysis of residuals showed independency (Durbin Watson test = 1.905).

Testing and infection rates

Throughout the study period, a total of 111,054 COVID-19 tests were administered, with an average of 2.8 tests per subject (range: 1–22). The mean age of subjects who tested positive for SARS-CoV-2was 30.8 years (±14.7), which was significantly higher than those who tested negative, who had a mean age of 26 years (±11.7) (P < .001). Regarding sex, there was no significant difference in infection rates between females and males (χ² = 1.287, DF = 2, P = .525). Among partially vaccinated individuals, 3% (n = 95) contracted SARS-CoV-2. When considering vaccination status categories (unvaccinated, fully vaccinated, and partially vaccinated), only the unvaccinated status was significantly associated with positive COVID-19 tests (χ² = 22.09, DF = 1, P < .001).

Among the individuals tested, 3.8% of students (n = 903) and 2.3% of employees (n = 351) tested positive for SARS-CoV-2. This difference indicates a significant association between university role and likelihood of infection, with students being at higher risk (χ² = 66.533, DF = 1, P < .001), independent of vaccination status.

When examining the timing of infection in relation to vaccination, 50.5% of students (n = 456) who tested positive had done so before receiving a vaccine, compared to 35.0% (n = 123) of employees (P < .001). Additionally, the mean time from full vaccination to SARS-CoV-2 infection was significantly shorter among students (97 ± 63 days) than among employees (120 ± 82 days; P < .0001) (Figure 2).

Figure 2.

Histogram and Boxplot of time from vaccination to COVID-19 infection: this figure presents two histograms displaying the distribution of days from full vaccination to the first positive COVID-19 test among the university cohort, categorized by their relationship to the university, and colored by their vaccination status. Panel a represents students, while panel B represents employees. Each histogram plots the frequency of individuals against the number of days elapsed since receiving their last vaccine dose until testing positive for COVID-19. The x-axis indicates the time in days, and the y-axis represents the number of individuals. Panel C represents a box plot of the median time from vaccination to a positive COVID-19 test among students and employees.

Subgroup analysis for vaccine impact

For this subgroup analysis, we included students and employees who had no record of SARS-CoV-2 infection prior to vaccination or during the pre-vaccine period. This allowed for a comparison of infection incidence across three groups: fully vaccinated, partially vaccinated, and unvaccinated individuals, without the confounding influence of prior natural immunity. As a result, a total of 32,100 subjects were analyzed, of which 22,852 were vaccinated, 6,234 were unvaccinated, and 3,014 were partially vaccinated.

Among the vaccinated individuals, 412 (1.8%) tested positive for SARS-CoV-2 (breakthrough infection) at any point during the follow-up period, compared to 201 individuals (6.7%) in the partially vaccinated group, and 0.9% (n = 55) in the unvaccinated group, showing an association with full vaccination and lower rates of infection (X2 = 21.877 [DF = 1], P < .0001). Notably, partially vaccinated individuals presented higher rates of SARS-CoV-2 infection than unvaccinated individuals (X2 = 62.41 [DF = 1], P < .0001).

When considering only the initial test result following vaccination, 2.3% of vaccinated subjects tested positive for SARS-CoV-2, whereas 3.6% of unvaccinated subjects tested positive during the same timeframe. The median number of days from full vaccination to a positive COVID-19 test was 112 days (IQR: 37–152). The median number of days from first dose of vaccination (partially vaccination) to a positive COVID-19 test was 98.5 days (IQR: 30–151). Analysis also demonstrated that vaccinated individuals were less likely to test positive compared to unvaccinated individuals within a median time of 109 days (χ² = 31.231 [DF = 1], P < .001).

No significant association was found between the brand of vaccine and the likelihood of breakthrough cases among all subjects fully vaccinated (χ² = 1.705 [DF = 4], P = .790), considering the first test result (median time 109 days). The same was observed when considering COVID-19 positivity at any time after vaccination within the follow-up period of 11 months and 10 days (χ² = 8.550 [DF = 5], P = .128).

Breakthrough infections among students and employees

Our analyses explored the rates of breakthrough SARS-CoV-2infections among students and employees. We found that 1.5% of students presented breakthrough infections within a median of 109 days, which was significantly lower than the 5.6% observed among employees. Notably, students were 75% less likely to experience a breakthrough infection compared to employees, with an odds ratio (OR) of 0.250 (95% CI 0.186–0.337, P < .001) as shown in Tables 2 and 3. This reduced likelihood remained significant even after adjusting for age, sex, and vaccine brand, with an adjusted OR of 0.161 (95% CI 0.086–0.301, P < .001).

Table 2.

Univariate analyses for breakthrough infection post-vaccination considering the relationship to the university.

	Estimated coefficient (S.E)	OR (95% CI)	P-value
Student	−1.828 (0.153)	0.161 (0.086–0.301)	<.001
Sex (Female reference)	−0.121 (0.098)	0.886 (0.605–1.297)	.534
Age	0.019 (0.006)	1.02 (0.999–1.040)	.057
Vaccine brand	0.049 (0.075)	0.876 (0.688–1.116)	.285

Univariate analyses considering COVID-19 infection as the dependent variable. S.E, standard error, OR, Odds ratio; CI, confidence interval.

Table 3.

Multivariate analyses for breakthrough infection post-vaccination considering the relationship to the university. Univariate and multivariate models considering COVID-19 infection as the dependent variable.

	COVID-19 positive (n = 179)	COVID-19 negative (n = 7458)	Univariate OR (95% CI)	Unadjusted P-value	Multivariate OR* (95% CI)	Adjusted* P-value
	N (%)	N (%)
Employees	91 (5.6)	1,534 (94.4)	1		1
Students	88 (1.5)	5,924 (98.5)	0.250 (0.186–0.337)	<.001	0.161 (0.086–0.301)	<.001

*Regression analysis adjusting for sex, age, and vaccine brand. OR, Odds ratio; CI, confidence interval.

Across all vaccine brands, the rate of breakthrough infections was consistently lower among students compared to employees, indicating that the differences in infection rates between these groups persisted regardless of the vaccine brand used (Table 4).

Table 4.

Comparison of breakthrough infection rates by vaccine brand between students and employees.

	Students	Employees
	Breakthrough infection rate (95% CI), %	Breakthrough infection rate (95% CI), %	P-value*
Pfizer (BNT162b2)	0.92 (0.62–1.31)	5.23 (3.55–7.42)	<.0001
Moderna (mRNA-1273)	1.42 (0.85–2.21)	2.9 (1.79–4.44)	.02
Janssen (AD26,COV2.S)	1.52 (0.41–3.9)	9.67 (1.99–28.28)	.005

*Comparison among the frequency of vaccine brands administered between Students and Employees, using Χ² with Bonferroni adjustment.

Furthermore, the median number of days from full vaccination to a positive COVID-19 test was shorter for students, at 111 days (IQR: 35–146), compared to 117 days (IQR: 46–193) for employees, indicating that breakthrough infections occurred sooner in students than in employees (P < .0001, Figure 2).

Discussion

This study presents critical insights into breakthrough SARS-CoV-29 infections among a large cohort of students and employees at the University of Miami. Our analyses revealed a significant disparity in breakthrough infection rates between these groups, with students showing a markedly lower proportion compared to employees, even after adjustments for age. The younger age of the student population (mean age 23 ± 9 years) compared to employees (44 ± 13 years) may have contributed to this difference (P < .001).

This finding is particularly compelling, highlighting the differential impact of COVID-19 on various segments within the same environment, possibly influenced by other factors such as lifestyle, and social behaviors which can differ substantially between students and employees in this population.⁹ In addition to behavioral factors, younger individuals may also mount a more robust immune response, potentially further enhanced by asymptomatic or mild natural exposures to SARS-CoV-2.

Interestingly, while unvaccinated students exhibited a higher proportion of infections, this trend reversed post-vaccination. Vaccinated students experienced notably lower rates of breakthrough infections than their vaccinated employee counterparts, emphasizing the effectiveness of vaccination in reducing infection rates among younger, more socially active segments of the population.

Further analysis delved into additional aspects of breakthrough infections. Notably, the median time from full vaccination to a positive COVID-19 test varied between the groups, with infections occurring sooner among students than employees.

The shorter period of days from vaccination to breakthrough infection in students may be due to differences in the timing of the availability of the vaccine. Most employees began vaccinating in December 2020 compared to the student population who received their vaccines 5 months later in April 2021, when the FDA and CDC approved vaccination of individuals not in the classic risk groups for COVID-19. April 2021 coincided with the 4th wave of COVID-19 in our academic health center with the 5th surge of infections starting in the last week of June and peaking in mid-August 2021 (Figure 1). The 5th surge, due primarily to the delta variant of the SARS-CoV-2 virus (unpublished) may account for the less protection afforded to the students vaccinated later.

In our analysis, we observed a significant difference in vaccination timings between these two cohorts. Employees were vaccinated several months prior to the student population, leading to a notable disparity in their respective antibody titers. It is well-documented that antibody levels, specifically those targeting the SARS-CoV-2 spike protein, wane over time.^10,11 In our study, the temporal variation in vaccination schedules and the possible natural decline in antibody titers over months necessitate a nuanced interpretation of the data, especially when assessing the risk and proportion of breakthrough infections. As antibody levels wane due to the staggered vaccination timing between employees and students, we possibly face a reduction in neutralizing capacity. Previous studies have demonstrated a correlation between declining neutralizing antibody titers and increased risk of breakthrough infections.^10,11 While we lacked antibody data in our cohort, it is plausible that lower titers over time contributed to the observed breakthrough cases, particularly among those vaccinated earlier.This connection highlights the critical need for caution when comparing breakthrough infection rates between these two groups, as varying post-vaccination intervals and declining antibody efficacy can significantly alter the dynamics of infection and immunity.

Additionally, our findings did not suggest a significant association between the brand of vaccine administered and the likelihood of developing a breakthrough infection, which underscores the overall effectiveness of all vaccine brands used in the study across different demographics. However, infection rates were notably lower among vaccinated individuals compared to those unvaccinated, reinforcing the effectiveness of vaccines in reducing the incidence of COVID-19, as evidenced in several studies.^12,13

On the other hand, it is important to account for the role of vaccinations among different SARS-CoV-2 variants. Early in the COVID-19 pandemic, the prevalence of mutant variants was relatively low due to the limited number of infections, which restricted the emergence of escape mutants.¹⁴ However, the subsequent surge in infections, including prolonged cases among immunocompromised individuals, has spurred the evolution of multiple SARS-CoV-2 variants.¹⁵ It is crucial to understand how these variants affect public health strategies and vaccination efforts.¹⁶

This study draws on a large university-based cohort, allowing for meaningful analyses of breakthrough SARS-CoV-2 infections in a real-world setting. As with any large retrospective dataset, it is important to acknowledge potential limitations that may inform the interpretation of our findings.

First, the dataset does not capture prior SARS-CoV-2 infections that may have occurred before the initiation of vaccination, which could influence observed infection rates due to varying levels of naturally acquired immunity. Similarly, antibody titers were not collected as part of routine clinical care during the study period, limiting our ability to assess immunologic correlates of protection or the extent of waning immunity over time.

Infection status in this study was based on PCR-confirmed cases using mid-nasal swabs. While this method is standard and highly specific, it may miss asymptomatic infections or cases arising from variant strains with altered detection profiles, leading to an underestimation of total infections. Data on individual mask use were not available and may have influenced exposure risk. Additionally, the use of serological assays, particularly those targeting antibodies against the SARS-CoV-2 nucleocapsid protein, could have allowed us to distinguish vaccine-induced immunity from natural infection and to identify subclinical infections not captured by PCR testing.^17,18

Our access to test results was limited to those performed within the university system or reported to it. Although compliance with institutional testing requirements was high, cases tested externally and not disclosed may have escaped detection. Other potentially relevant variables, such as comorbidities, behavioral patterns, mask use, or socioeconomic status, were not available in the dataset and could also influence infection susceptibility and vaccine effectiveness.

Despite these considerations, the strengths of this study lie in its large sample size encompassing over 38,000 subjects, structured testing protocols, and the ability to analyze vaccination and infection dynamics across two distinct university populations. Together, these data offer valuable insights into breakthrough infections and can inform public health strategies in similar institutional settings.

It is important to interpret crude infection rates with caution, as they do not reflect key contextual factors such as timing of exposure, vaccination rollout, and surveillance testing. For example, while the unadjusted infection rate appears higher in students (3.8%) than in employees (2.3%), this reflects differential exposure during a peak surge that coincided with the later rollout of vaccines to the student population. In contrast, our multivariate analysis, which adjusts for age, sex, and vaccine brand, shows that students were significantly less likely to experience breakthrough infections than employees (adjusted OR 0.161, 95% CI 0.086–0.301, P < .001).

Similarly, the observed lower infection rate among unvaccinated individuals (0.5%) compared to the fully vaccinated (1.8%) is likely influenced by both the timing of infection relative to vaccination and testing frequency requirements. Unvaccinated individuals may have had fewer opportunities for exposure or less consistent testing, especially if they were not present on campus. In contrast, vaccinated individuals remained on campus during periods of high transmission, and infections occurring months after initial vaccination may reflect waning immunity. These findings underscore the importance of interpreting infection rates within the proper temporal and behavioral context, rather than in isolation.

The comprehensive tracking of vaccination and infection status over a significant period allows for a detailed analysis of the temporal dynamics of breakthrough infections and the sustained effectiveness of the vaccines over time. Previous evidence indicated that maintaining high adherence to regular, voluntary asymptomatic screening with nose and throat swabs is feasible.¹⁹

University-led testing programs have received strong support from students, offering reassurance during a period when the pandemic has significantly impacted student mental health and wellbeing.^20,21

Conclusion

This retrospective cohort underscores the complexity of managing COVID-19 in a university setting, where diverse populations such as students and employees exhibit different risks and outcomes related to breakthrough infections. The lower proportion of breakthrough infections among students compared to employees, despite earlier vaccination timelines in the latter group, suggests that additional factors may influence susceptibility to infection. This suggests that susceptibility to breakthrough infection is likely influenced by a combination of factors, including age, immune response, vaccine timing, behavioral differences (e.g., frequency of exposure or adherence to mitigation strategies), and potentially occupational roles. These findings highlight the importance of tailored public health strategies that consider the unique characteristics of each subgroup within a population. Continued surveillance of breakthrough infections and the adaptation of vaccination strategies are essential for enhancing our understanding and response in future pandemics.

Biography

Joana R. N. Lemos research focuses on the immunological mechanisms driving type 1 diabetes, particularly immune modulation, islet transplantation, and data-driven approaches to translational research. As a member of the Diabetes Research Institute’s (DRI) scientific team, she integrates clinical research, statistical modeling, and scientific strategy to advance innovative diabetes therapies. Her work explores how immune responses contribute to beta cell destruction, identifying biomarkers and immunoregulatory pathways for early intervention.

By analyzing clinical and experimental data, Dr. Lemos helps refine approaches to immune tolerance, predictive disease modeling, and personalized treatments. She collaborates with clinical teams on the long-term follow-up of islet transplant recipients, assessing graft survival, immune rejection, and metabolic outcomes. Her expertise extends to public health and infectious diseases, including efforts to improve COVID-19 diagnostics and vaccination awareness. Dr. Lemos co-authored a publication on the Tera Breathalyzer, a novel diagnostic tool for detecting SARS-CoV-2, contributing to the development of non-invasive testing approaches. With a strong background in biostatistics, clinical trial design, and translational medicine, she helps bridge the gap between scientific discovery and real-world application, supporting research dissemination and strategic advancements in diabetes and immunotherapy.

Funding Statement

This study was supported in part by Grant [MD017347] from the National Institutes of Health to REW.

Disclosure statement

No potential conflict of interest was reported by the author(s).