Study selection

Of 1661 query results, 1262 were eligible for title and abstract screening, and 113 were eligible for full-text review, yielding 17 studies reporting single dose COVID-19 mRNA VE in populations with a history of prior infection (Supplementary Table 1). One study known to be eligible for inclusion but not identified in the search or included in reference lists of relevant systematic reviews was also included, for a total of 18 studies (Fig. 1). Two studies appeared to fit the inclusion criteria but single dose VE was reported only in figures without corresponding values for point estimates and 95% CIs and thus were excluded^16,17.

Study characteristics

The 18 included studies encompassed over 5.8 million total patients (Supplementary Table 2). Studies were conducted in the United States (6), Canada (3), Brazil (2), Israel (2), United Kingdom (2), China (Hong Kong [1]), France (1), and Singapore (1). Study designs included eight test-negative case-control, one matched case-control, eight retrospective cohort, and one prospective cohort. All included studies reported single dose VE for original mRNA vaccines, and no studies reported single dose VE for BA.1 or BA.4/5 bivalent or monovalent XBB.1.5-adapted formulations. VE endpoints reported in studies included infection, symptomatic infection, hospitalization, and hospitalization or death.

Ten studies included adult populations, of which five were conducted in high-risk populations including health system workers (3), nursing home residents (1), and adults ≥60 years (1). Four studies included both adolescent and adult populations (aged ≥12 years [3]; ≥16 years [1]). Three studies reported VE for adolescents aged 12–16 or 12–17 years, and three studies reported VE for children aged 5–11 years. No studies were identified reporting single dose VE for children under 5 years of age with prior infection. Ten studies reported VE during periods predominated by the original strain, or Alpha or Delta variants, while 11 studies reported VE during Omicron predominance. Ten studies reported single dose VE for the BNT162b2 vaccine alone while the remaining eight studies reported VE estimates for any mRNA vaccine (BNT162b2 or mRNA-1273).

VE during pre-Omicron predominance

Ten studies conducted in the United States, United Kingdom, France, Israel, and Brazil reported single dose mRNA VE for individuals with history of prior infection during periods of wild-type, Alpha, or Delta strain predominance (Supplementary Table 2). These studies used three different referent groups: immunologically naïve (unvaccinated individuals with no history of prior infection [5 studies])^{18,19,20,21,22}, unvaccinated individuals with a history of prior infection (5 studies)^{20,23,24,25,26}, and individuals vaccinated with two doses with no history of prior infection (1 study, Supplementary Tables 3–5)²⁷.

Compared to SARS-CoV-2 naïve individuals (i.e., unvaccinated individuals with no history of prior infection), one dose VE among individuals with history of prior infection ranged from 78 to 92% against infection^18,19,20, and 85–98% against symptomatic infection during wild-type, Alpha, or Delta strain predominance^21,22. In all five studies, single dose VE among those with prior infection was similar (i.e., had overlapping 95% CIs) to two dose VE among those with prior infection (Table 1); two dose VE with prior infection ranged from 78 to 85% against infection^18,19,20, and 96 to 99% against symptomatic infection (Supplementary Table 3)^21,22. Three studies found that effectiveness against infection with a single dose in individuals previously infected with SARS-CoV-2 was comparable to two vaccine doses without prior infection^18,19,20. Two studies found that one dose plus prior infection afforded significantly higher protection against symptomatic infection than two vaccine doses without prior infection (Table 1)^21,22.

Five studies reported one dose VE among individuals with history of prior infection compared to unvaccinated individuals with history of prior infection during wild-type, Alpha, and Delta predominance. One dose VE against infection ranged from 52 to 78% in study populations that included adolescents and adults^23,24, and was lower (11%, 95% CI: −98 to 60) in a study of nursing home residents (Fig. 2)²⁰. One dose VE was 45–77% against symptomatic infection^23,25, 59% (95% CI: 41 to 71) against hospitalization²⁶, and 62% (95% CI: 41 to 75) against hospitalization or death²⁵. One dose without prior SARS-CoV-2 infection had lower VE than one dose with prior infection^19,20,21,22, and two doses plus prior infection conferred similar^20,24,25,26 or higher²⁵ protection as one dose plus prior infection (Table 1, Supplementary Table 4).

In the only study that directly compared receipt of one dose with prior SARS-CoV-2 infection to two doses with prior infection as the referent group, there was no significant difference in VE for one or two doses with prior infection (Supplementary Table 5)²⁷.

VE during Omicron predominance

Eleven studies conducted in Brazil, Canada, China (Hong Kong), Israel, Singapore, the United States, and the United Kingdom reported single dose mRNA VE for individuals with history of prior infection during Omicron predominance. Six studies reported VE compared to immunologically naïve individuals^{21,28,29,30,31,32}, and six studies reported VE compared to unvaccinated individuals with history prior infection^{23,26,29,31,33,34}. One study used single dose plus prior infection as a referent group for estimating effectiveness of two or three doses plus prior infection³⁴, and one study used two doses plus no prior infection as the singular referent group³⁵.

Compared to immunologically naïve individuals, VE ranged from 32 to 89% against infection^28,29,30, 57–97% against symptomatic infection^21,30,31, 80–94% against hospitalization^29,32, and was 89% (95% CI: 82 to 93) for the only estimate against hospitalization or death (Fig. 2)³¹. Effectiveness of a single dose with prior infection was comparable^28,29,32 or higher than for one or two doses without prior infection against Omicron infection^29,30, symptomatic infection^21,30, and hospitalization (Table 2, Supplementary Table 3)^29,32. A single dose with prior infection conferred similar protection as two doses with prior infection against all endpoints in 10 estimates^{21,28,29,30,31,32}, with one estimate reporting higher effectiveness against symptomatic infection for two doses plus prior infection than one dose plus prior infection³¹.

Compared to unvaccinated individuals with prior infection, VE of one dose plus prior infection ranged from 8 to 71% against infection^23,29,33,34, 39–67% against symptomatic infection^23,31, and 25–60% against hospitalization^26,29 or hospitalization or death³¹ (Fig. 2). VE of one dose plus prior infection was higher than the VE of one dose without prior infection^21,29,30,32, and was comparable to two doses plus prior infection in 7 of 9 estimates across multiple endpoints (Table 2, Supplementary Table 4)^26,29,31,33.

VE by time since dose

Time since receipt of a single mRNA dose was inconsistently reported in eligible studies. Five studies provided no information on time since a single dose in their analyses^{23,24,30,33,34}, while eight studies noted only that VE was measured at least 7 or 14 days after receipt of the dose (Fig. 2)^{18,22,25,26,27,31,32,35}. Five studies reported the range (i.e., minimum and maximum) or median time since dose for single dose VE estimates^{19,20,21,28,29}; follow-up ranged from a median of 14 days²⁰, to up to a range of 14 months²⁹. Only 4 of 18 studies reported single dose VE estimates stratified by time since dose^19,21,23,29.

VE in children and adolescents

During both pre-Omicron and Omicron periods, one dose of original mRNA vaccine plus prior infection provided effective protection against infection in children aged 5–11 years (5 of 6 estimates)^23,28,33, and against infection (4 of 5 estimates)^23,33 and symptomatic infection (4 of 5 estimates)^21,23 in adolescents aged 12–16 or 12–17 years (Fig. 2).

VE in the elderly

Two studies focused on elderly populations^20,32. A study of adults aged ≥60 years during Omicron reported strong protection against hospitalization for a single dose plus prior infection during periods of BA.1, BA.2, and BA.4/5 variant predominance³². A pre-Omicron study of nursing home residents reported high VE against infection when compared to unvaccinated individuals without prior infection, but found no evidence of added protection with vaccination when compared to unvaccinated residents with prior infection²⁰.

Heterogeneity in studies

Studies had significant heterogeneity in study populations and endpoints. After grouping estimates from the 18 included studies by the five pre-defined groups for synthesis, the largest group size achieved was two studies, for two groups (Supplementary Fig. 1). The high degree of heterogeneity prevented the calculation of median VE by group, meta-analysis, and synthesis of the impact of other factors on VE, including time since dose, order of infection and vaccination, and SARS-CoV-2 variant.

Risk of bias in studies

In the Newcastle-Ottawa Quality Assessment, 15 studies scored eight or nine stars, and three studies scored seven stars (Supplementary Tables 6–7). No studies scored below the seven-star threshold for exclusion. The two case-control studies that scored seven stars had increased risk of bias primarily due to self-reported ascertainment of vaccination status^22,28. While four cohort studies were conducted in high-risk populations including health system workers and nursing home residents^18,19,20,34, the single cohort study that scored seven stars lacked details describing ascertainment of vaccination status in a health system worker population³⁴. Five studies reported ≥1 VE estimate with 95% CIs spanning ≥50 percentage points (Supplementary Tables 3–5)^{20,23,27,29,31}, which suggests low certainty of evidence for these estimates.