Strength and durability of indirect protection against SARS-CoV-2 infection through vaccine and infection-acquired immunity =========================================================================================================================== * Sophia T. Tan * Isabel Rodríguez-Barraquer * Ada T. Kwan * Seth Blumberg * Hailey J. Park * Justine Hutchinson * David Leidner * Joseph A. Lewnard * David Sears * Nathan C. Lo ## Abstract Early investigation revealed that COVID-19 vaccines confer indirect protection to fully susceptible and unvaccinated persons, defined as a reduced risk of SARS-CoV-2 infection among social contacts of vaccinated individuals. However, indirect protection from infection-acquired immunity and its comparative strength and durability to vaccine-derived indirect protection in the current epidemiologic context of high levels of vaccination, prior infection, and novel variants are not well characterized. Here, we show that both infection-acquired and vaccine-derived immunity independently yield indirect protection to close social contacts with key differences in their strength and waning. Analyzing anonymized data from a system-wide SARS-CoV-2 surveillance program of 177,319 residents across 35 California state prisons from December 2021 to December 2022 in a case-control design, we find that vaccine-derived indirect protection against Omicron SARS-CoV-2 infection is strongest within three months post-vaccination [30% (95% confidence interval: 20-38%)] with subsequent modest protection, whereas infection-acquired immunity provides 38% (24-50%) indirect protection to roommates for 6 months after SARS-CoV-2 infection, with moderate indirect protection persisting for over one year. Variant-targeted vaccines (bivalent formulation including Omicron subvariants BA.4/BA.5) confer strong indirect protection for at least three months [40% (3-63%)]. These results have important implications for understanding the long-term transmission dynamics of SARS-CoV-2 and can guide vaccine policy and public health measures, especially in high-risk environments such as prisons. Keywords * COVID-19 * SARS-CoV-2 * vaccination * indirect protection * transmission * epidemiology * correctional health ## Main Text Transmission dynamics of SARS-CoV-2 are driven in part by population immunity generated from vaccine-derived and infection-acquired immunity, which confer both direct and indirect protection1,2. Direct protection is the benefit to an individual with vaccine-derived and/or infection-acquired immunity against developing an infection or disease after exposure2. In contrast, indirect protection refers to the reduced risk of infection among social contacts of individuals with vaccine-derived and/or infection-acquired immunity due to infection prevention (sterilizing immunity) and reduced infectiousness2,3. Understanding the dynamics of vaccine-derived and infection-acquired indirect protection is needed to understand population-level transmission of SARS-CoV-2 (e.g., epidemic waves due to waning population immunity), especially in settings with high population immunity, and to guide public health control measures (e.g., value of booster doses)1. Indirect protection against SARS-CoV-2 infection has primarily been shown through studies that found a decrease in infectiousness of SARS-CoV-2 infections among vaccinated individuals3–10, with some evidence that suggests reinfections are less infectious than primary infections4. However, few studies have been able to quantify overall indirect protection, which includes indirect protection due to infection prevention in addition to reduced infectiousness3,11. These studies that estimate overall indirect protection have focused on unvaccinated individuals in immunologically naïve populations, so there is limited evidence on the role of indirect protection in mitigating transmission in contemporary populations with high levels of vaccination and natural infection with emerging novel variants. The temporal dynamics (strength and durability over time) of both vaccine-derived and infection-acquired indirect protection and the benefits of additional booster doses and variant-targeting vaccines (e.g., bivalent vaccine) also remain unclear. Investigation on these topics has been limited, in part given its study requires intensive testing of a large study population over the entire pandemic. Studying the impact of an individual’s vaccine-derived and infection-acquired protection on the risk of SARS-CoV-2 infection in their close social contacts is of particular relevance to public health control measures in high-transmission environments such as prisons4. At the start of the COVID-19 pandemic, prison populations experienced rates of infection more than five times higher than the general population in the United States12–14. The incarcerated population has since experienced sustained SARS-CoV-2 transmission due to dense congregate living and suboptimal ventilation conditions14–17. Characterizing the dynamics of indirect protection can improve understanding of population susceptibility over time and inform public health control measures, such as optimizing the interval timing of additional vaccine doses to slow transmission and informing reactive vaccination efforts during outbreaks. In this study, we use a modified test-negative case-control design to quantify overall indirect protection from both vaccine-derived and infection-acquired immunity, including their strength and durability, and considering different vaccine formulations. We perform our study retrospectively within a SARS-CoV-2 surveillance program in the California prison system aimed at widespread testing and isolation of cases to reduce transmission. The study aims to understand the complex dynamics of indirect protection for SARS-CoV-2 infection and is directly relevant to infection control and vaccine measures in the incarcerated population and other high-risk environments. ## Results We conducted a retrospective test-negative case-control study using anonymized data from the California Correctional Health Care Services (CCHCS) and their system-wide SARS-CoV-2 surveillance program of 177,319 residents across 35 California state prisons to measure vaccine-derived and infection-acquired indirect protection. We defined indirect protection as the difference in SARS-CoV-2 infection risk between individuals living with roommates with and without vaccine-derived and/or infection-acquired immunity. We used a test-negative design to ensure similar testing practices between cases and controls and a clearly defined period of exposure18,19. The study period was from December 15, 2021, to December 15, 2022, to study Omicron variants/sub-variants over a period with consistent, high-volume testing within the surveillance program (Figure 1 and Supplementary Table 1). The dominant circulating variants during this study period were Omicron BA.1, BA.2, BA.4, and BA.5 based on genomic surveillance in a subset of isolates, which also reflect variants circulating in California and the United States20. Descriptions of the system-wide testing, quarantine, and isolation practices implemented to reduce transmission are included in the Supplementary Notes. ![Figure 1.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/10/09/2024.07.23.24310889/F1.medium.gif) [Figure 1.](http://medrxiv.org/content/early/2024/10/09/2024.07.23.24310889/F1) Figure 1. SARS-CoV-2 infections, testing, and vaccination in California state prisons. We analyzed anonymized retrospective data from a SARS-CoV-2 surveillance program of residents incarcerated across 35 California state prisons. We plotted the total number of weekly SARS-CoV-2 infections and weekly SARS-CoV-2 tests (top) and cumulative vaccine coverage (bottom) in the entire population. Data are shown from March 2020 to December 2022, although the study period was December 2021 to December 2022 (shaded in gray) during circulation of Omicron variant/sub-variants. SARS-CoV-2 testing rates were consistent over the study period. Vaccine administration switched from ancestral monovalent vaccines to bivalent vaccines in September 2022. ### Study population Over the study period, we identified 36,754 confirmed SARS-CoV-2 infections, 11,331 of which were reinfections (Figure 1 and Figure 2). In December 2021, 62% of residents had received at least one vaccine dose, and 17% of residents had received at least one booster dose (Figure 1). Most residents received mRNA vaccines for their primary series (83%). We identified 6,472 COVID-19 cases meeting our study criteria, defined as a resident with a positive SARS-CoV-2 test, residence in a two-person room, and without a positive test within the preceding 90 days. Subsequently, we identified 246,444 potential controls, defined as a resident with a negative SARS-CoV-2 test, residence in a two-person room, and without a positive test within the preceding 90 days or subsequent 14 days. Residents could have multiple positive and/or negative SARS-CoV-2 tests that met study criteria (Supplementary Notes). We defined their roommate based on the housing arrangements 3-6 days prior to test collection in the case or control to reflect the biological latent period between exposure and detectable infection and movement of residents for quarantine after SARS-CoV-2 exposure. We tested alternative timings of the roommate definition in sensitivity analyses. We required that cases, controls, and their roommates were incarcerated before April 2020 to ensure complete record of prior SARS-CoV-2 infection. A complete description of inclusion and exclusion criteria can be found in the Methods and Figure 2. ![Figure 2.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/10/09/2024.07.23.24310889/F2.medium.gif) [Figure 2.](http://medrxiv.org/content/early/2024/10/09/2024.07.23.24310889/F2) Figure 2. Study population flow chart. We designed a test-negative case-control study to measure the indirect protection provided by COVID-19 vaccination and/or infection-acquired immunity. We analyzed anonymized retrospective data from a SARS-CoV-2 surveillance program of residents incarcerated in the California state prison system from December 15, 2021, to December 15, 2022. We identified individuals with a new SARS-CoV-2 infection (cases) and individuals with a negative SARS-CoV-2 test (controls) at the same time while residing in the same building. Cases and controls were required to co-reside in rooms with a single other resident in the 3-6 days leading up to test collection, to account for the latent period from exposure to detectable infection. We required cases and controls and their roommates to be incarcerated since March 2020 to ensure a complete record of prior infection over the pandemic. Cases and controls were matched in a 1:2 ratio based on multiple characteristics (although a subset was only matched 1:1 based on available controls meeting study criteria), including vaccination and prior infection status. We then evaluated differences in SARS-CoV-2 infection outcome in cases/controls based on the vaccine and prior infection history of their roommate. The sample size of the study population is shown at various stages of applying the study criteria and matching. Cases and controls were matched in a variable 1:2 ratio by time (tests within two days), COVID-19 vaccine status (by dose), prior SARS-CoV-2 infection status, time since their last vaccine dose and/or infection, and building of residence and security level, which largely determined activities and number of possible social contacts, and demographic factors. Cases and controls were matched exactly by their vaccine and prior infection status to isolate differential risk due to differences in the indirect protection from their roommates’ vaccine- and infection-acquired immunity. The final sample size included 4,640 cases and 7,824 controls; 3,184 cases were matched to two controls and 1,456 cases were matched to a single control. Match quality of cases and controls is shown in Supplementary Figure 1. Test acceptance in the 14 days prior to study inclusion was similar between cases and controls (61% of cases and 63% of controls had a SARS-CoV-2 test) and between roommates of cases and controls (61% of case roommates and 63% of control roommates had a SARS-CoV-2 test). Roommates of cases were more likely to have a new SARS-CoV-2 infection within the preceding 4 days (44% of case roommates compared to 15% of control roommates with testing data in this period) (Supplementary Table 1). Characteristics of the cases, matched controls, and their roommates are shown in Table 1. View this table: [Table 1.](http://medrxiv.org/content/early/2024/10/09/2024.07.23.24310889/T1) Table 1. Characteristics of the study population including COVID-19 cases, matched controls, and their roommates in California prisons. ### Indirect protection from COVID-19 vaccine and infection-acquired immunity Cases were more likely to co-reside with unvaccinated individuals (14% of cases had an unvaccinated roommate compared to 12% of matched controls) and less likely to co-reside with individuals with prior SARS-CoV-2 infection (47% of cases had roommates with documented prior infection compared to 51% of controls). On average, roommates were more recently vaccinated (mean of 149 days prior to study inclusion) than recently infected (mean of 460 days prior to study inclusion) (Table 1). With an adjusted model, co-residing with a vaccinated resident (received any vaccine dose at least 14 days prior to study inclusion) was associated with 22% indirect protection (95% CI: 13-31%) against SARS-CoV-2 infection (Figure 3 and Supplementary Table 2). When defining vaccine status by the number of doses received by roommates, we found that each additional dose of COVID-19 vaccine was associated with 7% indirect protection (95% CI: 4-11%), with up to 27% indirect protection (95% CI: 24-29%) from a resident with two or more booster doses (Figure 3 and Supplementary Table 3). Co-residing with a resident with prior SARS-CoV-2 infection (at least 14 days prior to study inclusion) was associated with 16% indirect protection (95% CI: 8-23%) (Figure 3 and Supplementary Table 2). We found co-residing with a resident with hybrid immunity (both vaccine-derived and infection-acquired immunity) was associated with 36% indirect protection (95% CI: 25-46%) (Figure 3 and Supplementary Table 4). ![Figure 3.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/10/09/2024.07.23.24310889/F3.medium.gif) [Figure 3.](http://medrxiv.org/content/early/2024/10/09/2024.07.23.24310889/F3) Figure 3. Overall vaccine-derived and infection-acquired indirect protection to close social contacts against SARS-CoV-2 infection. We estimated the indirect protection that COVID-19 vaccination and infection-acquired immunity provided to their roommate. We defined indirect protection as change in risk of SARS-CoV-2 infection in an individual based on their roommate’s vaccine and prior infection status. We adjusted for age and risk of severe COVID-19 of both the case and control and their roommates. Residents in California state prisons were less likely to be infected by the Omicron SARS-CoV-2 variant if they co-resided with an individual with vaccine-derived and/or infection-acquired protection. The mechanism of protection is likely that individuals with vaccination and/or prior infection are less likely to become infected (and then transmit infection) or are less infectious upon infection or reinfection. Residents with hybrid immunity were more likely to be recently vaccinated than recently infected. We plotted the mean (point estimate) and associated 95% confidence intervals (bars) for indirect protection. We defined vaccination both as a binary variable and by dose and infection-acquired immunity as binary. Separate regression models were fit for any vaccination and any infection, vaccination by dose, and hybrid immunity. ### Strength and waning of indirect protection from COVID-19 vaccines and infection-acquired immunity We assessed the strength and durability of indirect protection from vaccine-derived and infection-acquired immunity over time since their most recent vaccine, infection, or vaccine/infection. COVID-19 vaccination among residents was associated with 30% indirect protection (95% CI: 20-38%) within the first three months of vaccination (Figure 4 and Supplementary Table 5), with the strongest protection within 2 months of vaccination (Supplementary Figure 2 and Supplementary Table 6). Indirect protection from vaccine-derived immunity demonstrated waning after 3 months, with some subsequent modest protection (indirect protection ranged from 13-18%) (Figure 4 and Supplementary Table 5). ![Figure 4.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/10/09/2024.07.23.24310889/F4.medium.gif) [Figure 4.](http://medrxiv.org/content/early/2024/10/09/2024.07.23.24310889/F4) Figure 4. Comparative strength and durability of vaccine-derived and infection-acquired indirect protection to close social contacts against SARS-CoV-2 infection. We estimated the strength and durability of indirect protection that COVID-19 vaccine-derived and SARS-CoV-2 infection-acquired immunity provided to their close social contact (roommate). Residents in California state prisons were found to have benefit (e.g., less likely to be infected by the Omicron SARS-CoV-2 variant) when residing with an individual with infection-acquired or vaccine-derived protection; both sources of indirect protection waned over time, but infection-acquired protection yielded stronger and more durable indirect protection. We estimated indirect protection based on time since last vaccine dose (left), time since last SARS-CoV-2 infection (middle), and time since either most recent vaccine or infection (right). We plotted the mean (point estimate) and associated 95% confidence intervals (bars) for indirect protection. We fit separate models for vaccine-derived immunity, infection-acquired immunity, and most recent vaccination or infection. Indirect protection from infection-acquired immunity in the study population was stronger and more durable over time than indirect protection from COVID-19 vaccination. Co-residing with an individual with a SARS-CoV-2 infection within six months was associated with 38% indirect protection (95% CI: 24-50%). Prior SARS-CoV-2 infection more than one year ago prior to testing was associated with 12% indirect protection (95% CI: 3-19%) (Figure 4 and Supplementary Table 5). When measuring time since either the most recent vaccination or infection, indirect protection demonstrated waning over time (Figure 4 and Supplementary Table 5). We did not detect a multiplicative interaction between vaccine-derived indirect protection and infection-acquired indirect protection, which suggests the strength of vaccine-derived indirect protection does not differ between roommates with evidence of prior infection and those without prior infection (Supplementary Table 9). Additionally, both vaccine-derived and infection-acquired indirect protection from roommates were broadly similar among cases and controls with different levels of direct protection (no prior immunity, vaccine-derived immunity only, infection-acquired immunity only, and hybrid immunity), although this analysis was underpowered in some groups (Supplementary Table 10). ### Indirect protection from variant-targeted COVID-19 vaccines Starting September 2022, residents received bivalent vaccine doses, which targeted both the ancestral SARS-CoV-2 strain and Omicron variants BA.4 and BA.5 to match circulating Omicron subvariants21,22. By December 2022, 20.5% of residents received a bivalent vaccine dose. We found that a bivalent COVID-19 vaccine dose provided 40% indirect protection (95% CI: 3-63%) within the first three months of vaccine receipt (September – December 2022); we estimated that ancestral monovalent vaccines conferred 30% indirect protection (95% CI: 21-39%) for three months in the period before bivalent vaccine introduction (December 2021 – August 2022; Figure 5 and Supplementary Table 11). ![Figure 5.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/10/09/2024.07.23.24310889/F5.medium.gif) [Figure 5.](http://medrxiv.org/content/early/2024/10/09/2024.07.23.24310889/F5) Figure 5. Indirect protection from variant-targeting bivalent COVID-19 vaccine and comparison to ancestral monovalent vaccine during Omicron era. We estimated the indirect protection from a variant-targeting vaccine (bivalent vaccine formulated with ancestral strain and Omicron subvariants BA.4/5 to target circulating variants) in California state prisons. Our goal was to determine how concordance of vaccine formulation and circulating variants would affect vaccine-derived indirect protection. Administration of bivalent vaccines began in September 2022, which yielded approximately 3 months of follow up; therefore, estimates were only available for 3 months. We include estimates of indirect protection from ancestral monovalent vaccines during the study period prior to bivalent vaccine introduction. We estimated indirect protection and plotted the mean (point estimate) and associated 95% confidence intervals (bars) for indirect protection. ### Negative control analysis As a negative control analysis, we tested a negative control exposure of influenza vaccination in roommates. We found no evidence of indirect protection against SARS-CoV-2 infection from influenza vaccination within the past year (0.6% [95% CI: (-8-8%)]) (Supplementary Table 12). ### Sensitivity analyses Our findings were also similar with different definitions for timing of co-residence for cases and controls. When alternatively defining co-residence on the third day prior to test collection in cases and controls, we found having a roommate with vaccine-derived immunity or infection-acquired immunity was associated with 23% (95% CI: 14-31%) and 16% (95% CI: 9-22%) indirect protection, respectively. When defining co-residence as the week leading up to test collection, we estimated indirect protection was 22% (95% CI: 13-31%) from vaccine-derived immunity and 12% (4-19%) from infection-acquired immunity (Supplementary Table 13). Study findings were robust to alternative specifications in matching and statistical model, including 1:1 matching, alternative matching criteria, and alternative model covariates (Supplementary Table 14). We found similar results when we used unconditional logistic regression, controlling for all matched factors, and when we accounted for repeated observations of residents in the study population over time (Supplementary Table 15). ## Discussion In this study, we found that both vaccine-derived and infection-acquired immunity yield measurable and meaningful indirect protection against Omicron SARS-CoV-2 infection, indicating their role in reducing transmission and governing transmission dynamics. Our results suggest that infection-acquired immunity provides stronger and more durable indirect protection compared to vaccine-derived immunity, and indirect protection from both wanes over time. We also evaluated indirect protection from variant-targeting (bivalent) vaccines and found they provide strong indirect protection within the first three months of vaccination. Our findings have implications for understanding long-term viral transmission dynamics of SARS-CoV-2, which are likely governed by these indirect effects (i.e. rise in transmission and epidemic waves due in part to waning indirect protection). Our results also highlight that indirect protection from vaccines exist in a population with high cumulative vaccination and prior infection during a period of intense transmission, which is most relevant to this stage of the COVID-19 pandemic. This work has relevance to vaccine policy and public health measures in high-risk environments, such as reactive vaccination during periods of outbreaks and/or consideration of higher frequency of routine vaccination to maximize indirect protection. We observe infection-acquired immunity may generate stronger and more durable indirect protection than vaccine-derived indirect protection. This finding becomes most clear when accounting for time since vaccine and/or infection because the study population is more recently vaccinated than infected. The strength of infection-acquired indirect protection could be explained by many mechanisms, including that natural infection generates a more robust immune response. Another explanation is that recent infection occurs with more contemporary variants compared to vaccination with an ancestral strain formulation, meaning infection generates an immune response more tailored to circulating variants. We also find that indirect protection from both vaccine-derived and infection-acquired immunity wanes over time. This could be explained by many immunologic mechanisms (e.g., waning antibody titers), but also by viral evolution and increasing mismatch between the generated host immunity and circulating variants over time. Additional considerations strengthen the study findings. We identify strong and consistent indirect protection generated by vaccine-derived and infection-acquired immunity across multiple primary and sensitivity analyses, accounting for both mechanisms of indirect protection: infection prevention and reduced infectiousness upon infection. We find the indirect protection from vaccination is dose-dependent, and indirect protection from both vaccine-derived and infection-acquired immunity are time-dependent. The dose- and time-dependence are not only scientifically and policy relevant, but also strengthen the validity of our findings and reduce the likelihood of confounding. The time-dependence in indirect protection found in our study also matches literature on the comparative strength and waning of direct protection from vaccines and infection-acquired immunity23–25. We find that roommates of controls, who were more likely to have a history of prior vaccination and SARS-CoV-2 infection, were also less likely to have a positive SARS-CoV-2 test in the 4 days prior to study inclusion than roommates of cases (Supplementary Table 1). This evidence supports the mechanism of indirect protection, e.g., reduction of infection risk in the roommate due to vaccination and/or infection-acquired immunity. Finally, we employ a negative control exposure (influenza vaccination). In this analysis, we found no evidence for indirect protection against SARS-CoV-2 infection from influenza vaccination, which strengthens the causal implications of indirect protection from COVID-19 vaccination in our study population. This study is important for improving health and addressing inequities in the incarcerated population due to COVID-19. Over the pandemic, this population has experienced high rates of SARS-CoV-2 transmission largely due to structural and environmental risk factors, including dense congregate living and suboptimal ventilation conditions12–17. Although severe COVID-19 (defined as hospitalization or death) outcomes were rare during the Omicron era, high SARS-CoV-2 infection incidence, despite high levels of vaccine coverage and prior infection, and risk for complications (such as long COVID) highlight the need for continued optimization of health policy for this vulnerable population. We find evidence that additional vaccine doses can provide additive indirect protection to close social contacts, even in individuals with infection-acquired immunity, which suggests that residents who co-reside together can provide and receive indirect benefits that reduce transmission risk with additional vaccination. Housing arrangements based on vaccine and/or prior infection status may be particularly relevant for high-risk residents (e.g., 65+ years, immunocompromised, unvaccinated) and may reduce their risk of infection by nearly 40%. Furthermore, we find transmission often occurs outside of their rooms (most cases were not exposed to a SARS-CoV-2 positive roommate leading up to test collection) (Supplementary Table 1) and benefits of indirect protection could extend to a broader spatial scale (e.g., building level). Therefore, increasing uptake of booster doses in residents and especially among staff, where even vaccination with primary series alone has lagged behind residents (65%)26, would likely reduce transmission. Our findings also suggest that reactive vaccination campaigns could be used to mitigate transmission (via vaccine-derived indirect effects on transmission) during future outbreaks, especially if future variants are more virulent, and residents may benefit from considering more frequent routine COVID-19 vaccination. This study has limitations. This study design does not distinguish between indirect protection from infection prevention and indirect protection from reduced infectiousness; the relative contribution of these mechanisms may vary between individuals (e.g., immunity status). This test-negative case-control study design is observational and thus subject to potential confounding27. For example, residents with vaccine-derived and/or infection-acquired immunity may have unobserved differences in behavior (e.g., fewer social interactions, a more risk-averse social cohort, masking) that lower their risk of infection and/or transmission to social contacts, and we do not have data on behavior such as masking. However, our analysis demonstrates time-dependent indirect protection (waning) from both sources, which argues against the presence of such confounding factors. The negative control analysis (influenza vaccination) further reduces the likelihood of residual confounding. We also match cases and controls by all observed characteristics (including vaccine and prior infection status), and roommate assignments are expected to be random with respect to vaccine and prior infection status. While testing is frequent in this population throughout the pandemic, there may be some misclassification of prior infection status or timing due to imperfect case ascertainment27, though any misclassification is more likely nondifferential (Supplementary Table 1). While cases are required to have a new positive SARS-CoV-2 infection, timing of exposure and infection is unknown. However, we observed robust results when we varied our definition for timing of co-residence and roommate definitions. We also did not have access to cycle threshold values for PCR tests or any individual-level data on symptoms, serologic status, or viral genome for cases. Our analysis examines indirect protection at the room level to focus on the benefit of room-level housing arrangement policies that consider vaccine or prior infection status, although a substantial fraction of transmission occurs outside the room through interactions with other residents and staff. A higher proportion of outside-room transmission would reduce the indirect protection conferred by a roommate, so our analysis may underestimate the population-level indirect protection generated by vaccines and infection-acquired immunity. The absolute magnitude of indirect protection against SARS-CoV-2 infection will depend on the transmission environment and social context. This study focuses on indirect protection from Omicron SARS-CoV-2 infection, though these findings may generalize to other variants and vaccine formulations. Our study population is a subset of the overall California incarcerated population during the pandemic that co-resides in close contact with a single person, and our specific estimates on indirect protection are most applicable to similar high-risk transmission environments. This study finds that infection-acquired and vaccine-derived immunity confer indirect protection to close social contacts, with stronger and more durable indirect protection from immunity acquired from prior SARS-CoV-2 infection. These findings have implications for understanding transmission dynamics of SARS-CoV-2 and can inform vaccine and public health control measures for high-risk environments. ## Methods ### Study design We used a test-negative case-control design to evaluate indirect protection of COVID-19 vaccination and prior infection conferred to residents in California prisons18,19. We defined indirect protection as the difference in risk of SARS-CoV-2 infection between individuals with an unvaccinated and previously uninfected roommate and individuals with a vaccinated and/or previously infected roommate. While the test-negative case-control design is most often applied to measure direct vaccine effectiveness by comparing vaccine status in cases and controls19, we match cases and controls by vaccine status and prior infection history to identify differential risk between cases and controls solely from their roommates’ vaccine status and prior infection history. ### Data We used anonymized person-level data from CCHCS on demographics, SARS-CoV-2 testing, COVID-19 vaccination, and nightly housing information for incarcerated persons in the California state prison system from March 1, 2020, to December 15, 2022. We defined a study period from December 15, 2021, to December 15, 2022, based on a pre-specified objective of studying contemporary variants (e.g., Omicron subvariants BA.1, BA.2, BA.4, and BA.5) while ensuring consistent system-wide testing practices. The majority of tests were polymerase chain reaction (78%). Residents were isolated if they tested positive for SARS-CoV-2. Isolation, quarantine, and testing practices and vaccine administration during the study period are further described in the Supplementary Notes. ### COVID-19 cases and controls The inclusion and exclusion criteria for cases and controls are shown in Figure 2. Cases were defined as residents with a positive SARS-CoV-2 test (first positive test in at least 90 days). Controls were defined as residents with a negative SARS-CoV-2 test with no positive test in the preceding 90 days or subsequent 14 days. Both cases and controls were required to reside in rooms of only two residents in the 3-6 days leading up to their test. The timing of this housing requirement was chosen to represent the latent period between exposure and detectable infection and to account for movement of residents in response to SARS-CoV-2 exposure and was varied in sensitivity analyses. Cases, controls, and their roommates were required to have been incarcerated before April 1, 2020, to ensure a more complete history of documented SARS-CoV-2 infection. ### Statistical analysis We performed matching of cases and controls by person-level characteristics and time to improve precision and control for unobserved factors (Supplementary Notes). Cases and controls were first matched exactly by time (tests within two days), building and security level (which largely determines a resident’s social contacts), COVID-19 vaccine status (unvaccinated, partially vaccinated, primary series alone, one booster dose, two or more booster doses), and prior SARS-CoV-2 infection (binary). We matched cases and controls by vaccine and prior infection status to limit confounding from direct protection. Cases and controls were then further matched to minimize differences in time since their most recent SARS-CoV-2 infection and/or COVID-19 vaccination and age (in years) and risk score for severe COVID-19 (weighted score of risk factors for severe COVID-19 used by CCHCS) between cases and controls and between their roommates. Cases and controls were ineligible for matching if they resided together. Cases and controls were matched in a 1:2 ratio (although a subset was matched 1:1 based on available controls) though this was varied in a sensitivity analysis. We did not allow residents to be repeated within a matched group. Descriptive data on the quality of matches are in Supplementary Figure 1. To estimate the indirect protection from vaccine-derived and infection-acquired immunity, we fit conditional logistic regression models, defining strata for matched sets of cases and controls27. The model outcome was SARS-CoV-2 infection (case or control), and the primary exposures of interest were the COVID-19 vaccine status (binary) and prior infection history (binary) of the roommate. Indirect protection (protection generated by roommate against SARS-CoV-2 infection) was estimated from one minus the adjusted odds ratio (OR)18,19. We adjusted for age and severe COVID-19 risk in cases, controls, and their roommates. Since the model was stratified by matched case and control pairs, we did not adjust for covariates that were exactly matched (e.g., building and security level, COVID-19 vaccine status, and prior infection history of cases and controls). We treated repeated observations of a single resident in different matched groups independently, although we tested this assumption in sensitivity analyses. We defined onset of vaccine protection as 14 days after receipt of each dose, which is consistent with literature29,30. We defined infection-acquired immunity as prior infection more than 14 days prior to testing to remove individuals who had an active SARS-CoV-2 infection. We fit additional models to explore the dose- and time-dependence of indirect protection. We defined COVID-19 vaccine status in roommates numerically by doses (unvaccinated, partially vaccinated, primary series alone, one booster dose, two or more booster doses). We assessed durability of indirect protection from vaccine-derived and infection-acquired immunity in separate models by defining exposures by time. We defined key time categories based on vaccine literature25,31, including <3 months, 3-6 months, 6-12 months, and 12+ months. We also measured indirect protection over time since most recent immunizing event (most recent vaccination or infection). Each model was adjusted for age and risk of severe COVID-19 of cases, controls, and their roommates and, if applicable, binary vaccine status or prior infection status of the case and control. We tested interactions between vaccine and prior infection exposures and measured indirect protection from hybrid immunity (both COVID-19 vaccine and prior SARS-CoV-2 infection). We stratified indirect protection by immune status in cases and controls to explore the relationship between indirect and direct protection. Bivalent COVID-19 vaccines targeting both the ancestral SARS-CoV-2 strain and Omicron subvariants BA.4 and BA.5 became the only available vaccine in the study population in September 2022. We performed an analysis where we estimated the indirect protection from bivalent vaccines (received on or after September 1, 2022). We compared these estimates to the indirect protection from monovalent vaccine (<3 months, 3+ months), although these estimates were from the period prior to bivalent introduction (Supplementary Notes). We performed a negative control analysis by testing a negative control exposure of influenza vaccination in roommates (instead of COVID-19 vaccination)32. Influenza vaccination was chosen since it should have no causal impact on SARS-CoV-2 infection or transmission. Analysis was conducted in R (version 4.3.1). All code is publicly available33. ### Sensitivity analyses We conducted various sensitivity analyses of study design and analytical decisions. We tested different housing requirements for roommates of cases and controls (3 day, 0-6 days, 6-9 days). We assessed differences in results when matching cases and controls in a 1:1 ratio. We assessed robustness of results to adjustment for time since last COVID-19 vaccine and time since last SARS-CoV-2 infection. Since matching may introduce bias28, we also tested an unconditional logistic regression model with the same study population of matched cases and controls that adjusted for all factors that had been matched exactly, including building and security level, vaccine status of cases and controls, and prior infection status of cases and controls. To assess the sensitivity of our results to repeated measures of the same residents, we fit a conditional logistic regression model without repeated measures and an unconditional logistic regression model with person-level cluster robust errors. ### Ethics This study was approved by the IRB at Stanford University and UCSF. The IRB included a waiver of consent given use of retrospective secondary data without direct identifiers that were collected for public health surveillance. Additionally, this study was deemed to be minimal risk and has direct relevance to improving the health of the population (Supplementary Notes). Isolation, quarantine, and testing practices are further described in the Supplementary Notes. ### Data availability Data requests may be made to California Correctional Health Care Services and are subject to controlled access. ### Code availability All analytic code is publicly available33. ## Supporting information Supplementary Information [[supplements/310889_file02.pdf]](pending:yes) ## Data Availability Data requests may be made to California Correctional Health Care Services and are subject to controlled access. All analytic code is publicly available at [https://github.com/sophttan/covid-indirects](https://github.com/sophttan/covid-indirects). ## Disclaimer The study content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. ## Funding NCL is supported by the National Institutes of Health, NIAID New Innovator Award (DP2AI170485). ## Author contributions Ms. Sophia Tan and Dr. Nathan Lo had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: STT, DS, NCL Statistical analysis: STT, NCL Analytic coding: STT, HJP Acquisition, analysis, or interpretation of data: All authors First draft of the manuscript: STT, NCL Critical revision of the manuscript: All authors Contributed intellectual material and approved final draft: All authors ## Competing interests JAL has received grants, honoraria, and speaker fees from Pfizer; grants and honoraria from Merck, Sharp, & Dohme; honoraria from Valneva; and honoraria from VaxCyte; all unrelated to the subject of this work. ATK and DS received funding from California Prison Health Care Receivership. NCL reports consulting fees from the World Health Organization related to guidelines on neglected tropical diseases, which are outside the scope of the present work. The remaining authors have no disclosures. ## Acknowledgments We thank Dr. Heidi Bauer and Dr. Richard Sun for valuable input and acknowledge individuals at the Office of the California Prison Health Care Receivership and California Correctional Health Care Services. We acknowledge the individuals who provided the data underlying these analyses. ## Footnotes * Authorship and affiliations updated; Manuscript and supplementary information updated to include additional analyses * Received July 23, 2024. * Revision received October 8, 2024. * Accepted October 9, 2024. * © 2024, Posted by Cold Spring Harbor Laboratory The copyright holder for this pre-print is the author. All rights reserved. The material may not be redistributed, re-used or adapted without the author's permission. ## References 1. 1.Anderson RM, May RM. Infectious Diseases of Humans: Dynamics and Control. Oxford University Press; 1992. 2. 2.Halloran ME, Haber M, Longini IM Jr., Struchiner CJ. Direct and Indirect Effects in Vaccine Efficacy and Effectiveness. Am J Epidemiol. 1991;133(4):323–331. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/oxfordjournals.aje.a115884&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=1899778&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F10%2F09%2F2024.07.23.24310889.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=A1991EZ15700001&link_type=ISI) 3. 3.Hayek S, Shaham G, Ben-Shlomo Y, et al. Indirect protection of children from SARS-CoV-2 infection through parental vaccination. Science. 2022;375:1155–1159. 4. 4.Tan ST, Kwan AT, Rodríguez-Barraquer I, et al. Infectiousness of SARS-CoV-2 breakthrough infections and reinfections during the Omicron wave. Nat Med. 2023;29(2):358–365. 5. 5.Prunas O, Warren JL, Crawford FW, et al. Vaccination with BNT162b2 reduces transmission of SARS-CoV-2 to household contacts in Israel. Science. 2022;375(6585):1151–1154. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1126/science.abl4292&link_type=DOI) 6. 6.Lyngse FP, Mortensen LH, Denwood MJ, et al. Household transmission of the SARS-CoV-2 Omicron variant in Denmark. Nat Commun. 2022;13(1):5573. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41467-022-33328-3&link_type=DOI) 7. 7.Baker JM. SARS-CoV-2 B.1.1.529 (Omicron) Variant Transmission Within Households — Four U.S. Jurisdictions, November 2021–February 2022. MMWR Morb Mortal Wkly Rep. 2022;71. 8. 8.Madewell ZJ, Yang Y, Longini IM Jr., Halloran ME, Dean NE. Household Transmission of SARS-CoV-2: A Systematic Review and Meta-analysis. JAMA Netw Open. 2020;3(12):e2031756. 9. 9.Jung J, Kim JY, Park H, et al. Transmission and Infectious SARS-CoV-2 Shedding Kinetics in Vaccinated and Unvaccinated Individuals. JAMA Netw Open. 2022;5(5):e2213606. 10. 10.Richterman A, Meyerowitz EA, Cevik M. Indirect Protection by Reducing Transmission: Ending the Pandemic With Severe Acute Respiratory Syndrome Coronavirus 2 Vaccination. Open Forum Infect Dis. 2022;9(2):ofab259. 11. 11.Salo J, Hägg M, Kortelainen M, et al. The indirect effect of mRNA-based COVID-19 vaccination on healthcare workers’ unvaccinated household members. Nat Commun. 2022;13(1):1162. [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F10%2F09%2F2024.07.23.24310889.atom) 12. 12.Burkhalter E, Colón I, Derr B, et al. Incarcerated and Infected: How the Virus Tore Through the U.S. Prison System. The New York Times. [https://www.nytimes.com/interactive/2021/04/10/us/covid-prison-outbreak.html](https://www.nytimes.com/interactive/2021/04/10/us/covid-prison-outbreak.html). April 10, 2021. Accessed April 11, 2024. 13. 13.Data Visualizations. COVID Prison Project. Accessed April 11, 2024. [https://covidprisonproject.com/data-visualizations/](https://covidprisonproject.com/data-visualizations/) 14. 14.Saloner B, Parish K, Ward JA, DiLaura G, Dolovich S. COVID-19 Cases and Deaths in Federal and State Prisons. JAMA. 2020;324(6):602–603. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jama.2020.12528&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F10%2F09%2F2024.07.23.24310889.atom) 15. 15.Chin ET, Ryckman T, Prince L, et al. COVID-19 in the California State Prison System: an Observational Study of Decarceration, Ongoing Risks, and Risk Factors. J Gen Intern Med. 2021;36(10):3096–3102. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/s11606-021-07022-x&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=34291377&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F10%2F09%2F2024.07.23.24310889.atom) 16. 16.Kwan A, Sklar R, Cameron DB, et al. Respiratory pandemic preparedness learnings from the June 2020 COVID-19 outbreak at San Quentin California State Prison. Int J Prison Health. 2022;19(3):306–321. 17. 17.Sklar R, Noth E, Kwan A, Sear D, Bertozzi S. Ventilation conditions during COVID-19 outbreaks in six California state carceral institutions. PLOS ONE. 2023;18(11):e0293533. 18. 18.Lewnard JA, Patel MM, Jewell NP, et al. Theoretical Framework for Retrospective Studies of the Effectiveness of SARS-CoV-2 Vaccines. Epidemiology. 2021;32(4):508. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1097/EDE.0000000000001366&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=34001753&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F10%2F09%2F2024.07.23.24310889.atom) 19. 19.Lewnard JA, Tedijanto C, Cowling BJ, Lipsitch M. Measurement of Vaccine Direct Effects Under the Test-Negative Design. Am J Epidemiol. 2018;187(12):2686–2697. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/aje/kwy163&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30099505&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F10%2F09%2F2024.07.23.24310889.atom) 20. 20.CDC. COVID Data Tracker. Centers for Disease Control and Prevention. March 28, 2020. Accessed June 9, 2022. [https://covid.cdc.gov/covid-data-tracker](https://covid.cdc.gov/covid-data-tracker) 21. 21.Pfizer and BioNTech Granted FDA Emergency Use Authorization of Omicron BA.4/BA.5-Adapted Bivalent COVID-19 Vaccine Booster for Ages 12 Years and Older | Pfizer. Accessed April 12, 2024. [https://www.pfizer.com/news/press-release/press-release-detail/pfizer-and-biontech-granted-fda-emergency-use-authorization](https://www.pfizer.com/news/press-release/press-release-detail/pfizer-and-biontech-granted-fda-emergency-use-authorization) 22. 22.COVID-19 Vaccine Booster Questions & Answers. Accessed April 12, 2024. [https://www.cdph.ca.gov/Programs/CID/DCDC/Pages/COVID-19/Vaccine-Booster-QA.aspx](https://www.cdph.ca.gov/Programs/CID/DCDC/Pages/COVID-19/Vaccine-Booster-QA.aspx) 23. 23.Altarawneh HN, Chemaitelly H, Hasan MR, et al. Protection against the Omicron Variant from Previous SARS-CoV-2 Infection. N Engl J Med. 2022;386(13):1288–1290. [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F10%2F09%2F2024.07.23.24310889.atom) 24. 24.Chin ET, Leidner D, Lamson L, et al. Protection against Omicron from Vaccination and Previous Infection in a Prison System. N Engl J Med. 2022;387(19):1770–1782. 25. 25.Rennert L, Ma Z, McMahan CS, Dean D. Effectiveness and protection duration of Covid-19 vaccines and previous infection against any SARS-CoV-2 infection in young adults. Nat Commun. 2022;13(1):3946. [PubMed](http://medrxiv.org/lookup/external-ref?access_num=35803915&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F10%2F09%2F2024.07.23.24310889.atom) 26. 26.Population COVID-19 Tracking. COVID-19 Information. Accessed April 15, 2024. [https://www.cdcr.ca.gov/covid19/population-status-tracking/](https://www.cdcr.ca.gov/covid19/population-status-tracking/) 27. 27.Sullivan SG, Tchetgen Tchetgen EJ, Cowling BJ. Theoretical Basis of the Test-Negative Study Design for Assessment of Influenza Vaccine Effectiveness. Am J Epidemiol. 2016;184(5):345–353. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/aje/kww064&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=27587721&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F10%2F09%2F2024.07.23.24310889.atom) 28. 28.Pearce N. Analysis of matched case-control studies. BMJ. 2016;352:i969. [FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiRlVMTCI7czoxMToiam91cm5hbENvZGUiO3M6MzoiYm1qIjtzOjU6InJlc2lkIjtzOjE2OiIzNTIvZmViMjVfNi9pOTY5IjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjQvMTAvMDkvMjAyNC4wNy4yMy4yNDMxMDg4OS5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 29. 29.Baden LR, El Sahly HM, Essink B, et al. Efficacy and Safety of the mRNA-1273 SARS-CoV-2 Vaccine. N Engl J Med. 2021;384(5):403–416. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1056/NEJMoa2035389&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F10%2F09%2F2024.07.23.24310889.atom) 30. 30.Polack FP, Thomas SJ, Kitchin N, et al. Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine. N Engl J Med. 2020;383(27):2603–2615. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1056/NEJMoa2034577&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F10%2F09%2F2024.07.23.24310889.atom) 31. 31.Menegale F, Manica M, Zardini A, et al. Evaluation of Waning of SARS-CoV-2 Vaccine– Induced Immunity: A Systematic Review and Meta-analysis. JAMA Netw Open. 2023;6(5):e2310650. 32. 32.Lipsitch M, Tchetgen Tchetgen E, Cohen T. Negative controls: a tool for detecting confounding and bias in observational studies. Epidemiol Camb Mass. 2010;21(3):383–388. 33. 33.GitHub Repository. [https://github.com/sophttan/covid-indirects](https://github.com/sophttan/covid-indirects)