Vaccination against influenza viruses reduces infection, not hospitalization or death from respiratory COVID-19: a systematic review and meta-analysis

Abstract

Severe acute respiratory syndrome coronavirus 2 (SARS‑CoV‑2) causes COVID-19 and has brought a huge burden in terms of human lives. Strict social distance and influenza vaccination have been recommended to avoid co-infections between influenza viruses and SARS-CoV-2. Scattered reports suggested a protective effect of influenza vaccine on COVID-19 development and severity.

We analyzed 51 studies on the capacity of influenza vaccination to affect infection with SARS-CoV-2, hospitalization, admission to Intensive Care Units (ICU), and mortality. All subjects taken into consideration did not receive any anti-SARS-CoV-2 vaccine, although their status with respect to previous infection with SARS-CoV-2 is not known. Comparison between vaccinated and not-vaccinated subjects for each of the four endpoints was expressed as odds ratio (OR), with 95% confidence intervals (CIs); all analyses were performed by DerSimonian and Laird model, and Hartung-Knapp model when studies were less than ten.

In a total of 61,029,936 subjects from 33 studies influenza vaccination reduced frequency of SARS-CoV-2 infection [OR plus 95% Confidence Intervals (C.I.) = 0.70 (0.65–0.77)]. The effect was significant in all studies together, in healthcare workers and in the general population; distance from influenza vaccination and the type of vaccine were also of importance. In 98,174 subjects from 11 studies frequency of ICU admission was reduced with influenza vaccination [OR (95% C.I.) = 0.71 (0.54–0.94)]; the effect was significant in all studies together, in pregnant women and in hospitalized subjects. In contrast, in 4,737,328 subjects from 14 studies hospitalization was not modified [OR (95% C.I.) = 1.05 (0.82–1.35)], and in 4,139,660 subjects from 19 studies, mortality was not modified [OR (95% C.I.) = 0.76 (0.26–2.20)]. This study emphasizes the importance of influenza vaccination in the protection against SARS-CoV-2 infection.

Introduction

Since early reports, severe COVID-19 has been found to be more frequent in the elderly, in patients with diabetes mellitus, obesity, or pre-existing cardiovascular diseases, and it has been associated with development of cardiovascular diseases [1–4].

Many studies focused on the similarities, differences, and relationship between SARS-CoV-2 and the influenza viruses (in particular influenza A virus) that cause influenza. SARS-CoV-2 and influenza share some aspects of pulmonary symptomatology [5–7] and, in the case of co-infection with these two viruses, the prognosis of COVID-19 has been shown to be worse [8–16].

During the COVID-19 pandemic, a significant reduction of influenza cases has been observed worldwide [17–23], likely as a result of the social distance measures adopted to contain the spread of SARS-CoV-2 [24, 25]; in addition, influenza vaccination has been strongly recommended [26], mainly to avoid SARS-CoV-2 and influenza virus co-infections. It is of particular interest whether influenza vaccination directly affects the severity of COVID-19. Previous epidemiological studies performed in 2020 in Italy, a country highly affected by the first wave of COVID-19, have shown that influenza vaccination was associated with a lower rate of SARS-CoV-2 infection and with lower mortality [27–30]. A systematic review of October 2020 focused on the possible association between influenza vaccination and SARS-CoV-2 infection confirmed that, in the majority of the 12 studies considered, influenza vaccination was associated with a lower rate of infection and lower severity of COVID-19 [31]. Nevertheless, these findings were partially confirmed in two more recent meta-analyses, performed in 13 and 23 studies, respectively [32, 33]. These latter meta-analyses showed that influenza vaccination was associated with a reduced risk of infection and hospitalization, but that influenza vaccination did not significantly affect the admission to intensive care units (ICU) or the death rate. These somewhat contrasting data give a potent mandate to extend the analysis on the association between influenza vaccination, SARS-CoV-2 infection, and COVID-19 severity.

Beside influenza vaccination, other vaccines, such as the BCG [34, 35] or pneumococcal [36] vaccine, have been evaluated for their efficacy in reducing the SARS-CoV-2 infection and/or disease severity, yielding contrasting results. Overall, as of today, no consensus has been reached about the effects of influenza vaccination on the infection with SARs-CoV-2.

The aim of this study is to understand how influenza vaccination affects COVID-19 development and severity. To achieve this goal, we performed a meta-analysis of all available studies that took into consideration the association between influenza vaccination and SARS-CoV-2 infection, as well as the association between influenza vaccination and hospitalization, admission of infected patients to ICU, and mortality. Notably, at sensitivity analysis, we also took into consideration five different groups of individuals: i) healthcare workers; ii) the general population; and special populations such as iii) elderly individuals; iv) poor health individuals (subjects at high-risk of influenza complications, kidney transplant patients, advanced-cancer patients); v) pregnant woman. Our analyses demonstrate that influenza vaccination reduces the frequency of SARS-CoV-2 infection and the admission in ICU, but it has only a small effect on the severity of COVID-19, as shown by frequency of hospital admission and, above all, mortality.

Materials and Methods

Search Strategy and Inclusion Criteria

This meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline [37]. Eligible studies were prospective and retrospective studies, both cohort and case-control studies reporting infection in influenza vaccinated and not-vaccinated subjects; in addition, eligible studies were those reporting hospitalization, admission to intensive care units (ICU) and mortality in influenza vaccinated and not-vaccinated subjects. Three authors (AEP, FS, LC) independently searched relevant literature in databases including PubMed, Embase, and Cochrane Library from inception until January 31, 2023. The following keywords were used for disease and intervention: Influenza vaccine or Flu Vaccine or Influenza vaccination or Flu vaccination, and COVID-19 or SARS-CoV-2. The title and abstracts written in English language were reviewed to recognize eligible studies. Additional studies were also manually searched through the references cited in reviews. If the results of one study were reported in more publications, only the most recent and complete data were included in analysis. When required, authors of the studies were also contacted by mail to obtain more details. The following studies were excluded: descriptive studies, editorials, review articles, systematic reviews and meta-analyses, case reports, and studies that did not provide risk ratios or effect sizes. Decisions on trials to include were taken by the authors (AEP, FS, LC), and disagreements were resolved by discussion. The reason for exclusion of other trials was specified (lack of details, no controls, Figure 1). In total: 51 studies, 33 studies for infection (with 37 comparison arms) [44–76], 14 studies (with 14 comparison arms) for hospitalization [49, 52, 59, 60, 63, 75, 77–84], 11 studies (with 11 comparison arms) for admission to ICU [48, 59, 60, 79, 83–85], and 19 studies (with 20 comparison arms) for mortality [49, 59, 63, 67, 75, 81–84, 86–94] fulfilled the inclusion criteria. Table 1A, 1B, 1C, and 1D show details of studies included in this meta-analysis. The protocol of the meta-analysis has been registered (Prospero, CRD42023400802). The following data were extracted: authors, year of publication, country, type of study, mean age of participants. For each group additional items were extracted: population (kind of subjects), method of diagnosis, season of vaccination, date of event, distance in months between vaccination and event, kind of vaccine employed.

Figure 1.

Flow-chart of the analysis performed according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (PRISMA). After screening of literature according to search terms, most papers were excluded for reasons indicated in the squares.

Table 1.

A. Studies evaluating the association between influenza vaccination and SARS-CoV-2 infection. Authors (and references), year of publication, country, kind of study, sample size, mean age, population studied, method for diagnosis, kind of vaccine, vaccination season, and Newcastle Ottawa Scale (NOS) are reported.

Study	Year	country	Kind of study	Sample size	Mean age	Population	Method for diagnosis	Vaccine	Vaccination season	NOS
Jehi [44]	2020	US	PC	11,672	Nd	General	RT-PCR	Nd	2019–2020	8
Vila-Corcoles [45]	2020	Spain	RC	1,547	65.8	age > 50	RT-PCR	Nd	2019–2020	7
Caban-Martinez [46]	2020	US	CS	203	Nd	Fireworkers and nurses	IgG and IgM	Nd	2019–2020	5
Noale [47]	2020	Italy	CS	6,680	44.2 and 70.8 (2 cohorts)	Age < and > 65 (2 cohorts)	RT-PCR	Nd	2019–2020	6
Caratozzolo [48]	2020	Italy	RC	848	79.7	Dementia	RT-PCR and antigenic	Nd	2019–2020	5
Zein [49]	2020	US	RC	13,220	55.4	General	Nd	Inactivated 4v	2019–2020	7
Martinez-Baz [50]	2020	Spain	RC	9,745	Nd	Healthworkers	RT-PCR and antigenic	Inactivated 3v	2019–2020	7
Bersanelli [51]	2020	Italy	PC	955	69.5	Cancer	RT-PCR	Nd	2019–2020	6
Ragni [52]	2020	Italy	CC	17,608	Nd	General	RT-PCR	Inactivated 4v, 3v with adjuvant	2019–2020	7
Belingheri [53]	2020	Italy	CS	3,520	Nd	Healthworkers	RT-PCR	Inactivated 4v	2019–2020	7
Green [54]	2020	Israel	CS	22,563	39.2	General	RT-PCR	Inactivated 4v	2019–2020	8
Massoudi [55]	2021	Iran	CC	261	39.5	Healthworkers	RT-PCR	Inactivated 4v	2019–2020	6
Rivas [56]	2021	US	RC	6,087	41.4	Healthworkers	IgG	Nd	2019–2020	6
Kissling [57]	2021	Europe	CC	1,887	Nd	General	RT-PCR	Nd	2019–2020	6
Erismis [58]	2021	Turkey	RC	203	Nd	Healthworkers	Nd	Nd	2019–2020	5
Conlon [59]	2021	US	RC	27,201	47.2	General	RT-PCR	Nd	2019–2020	8
Pawlowski [60]	2021	US	RC	25,582	74.6	General	RT-PCR	HD, IN, 4v	2019–2020	8
Fernandez-Prada [61]	2021	Spain	CC	188	64.6	General	RT-PCR	Nd	2019–2020	7
Kowalska [62]	2021	Poland	CS	5,376	43.9	General	IgG and IgM	Inactivated 4v	2019–2020	6
Bozek [63]	2021	Poland	RC	2,303	53.7	Age > 40 and < 60	RT-PCR	Inactivated 4v	2019–2020	8
Huang [64]	2021	US	CS	55,667,977	Nd	Age > 65	Nd	HD, 3v with adjuvant	2019–2020	7
King [65]	2021	US	CC	1,736	38	General	RT-PCR	Nd	2019–2020	8
Alkathlan [66]	2021	SA	CS	424	Nd	Healthworkers	Nd	Nd	2019–2020	4
Xiang [67]	2021	UK	PC	30,835	Nd	General	RT-PCR	Nd	2019–2020	8
Pépin [68]	2021	Canada	CC	2,985	54.5	Healthworkers and General (2 cohorts)	RT-PCR	Inactivated	2019–2020	6
Debisarun [69]	2021	Netherl	RC	17,755	Nd	Healthworkers	RT-PCR	Inactivated 4v	2019–2020 and 2020–2021	6
Shosha [70]	2022	Bahrain	RC	3,563	40,6	Healthworkers	RT-PCR	Live attenuated 4v	2019–2020	7
Domnich [71]	2022	Italy	RC	2,561	46.8	Healthworkers	RT-PCR	Inactivated 4v	2020–2021	6
Satir [72]	2022	Turkey	PC	232	44.5	Kidney transplant	RT-PCR	Inactivated 4v	2019–2020	7
Alòs [73]	2022	Spain	RC	429,537	Nd	High risk people	RT-PCR and antigenic	Inactivated 3v/4v	2020–2021	7
Van Laak [74]	2022	Ned	PC	223,580	63.8	High risk people	GP	Nd	2019–2020	7
Hosseini-Mogaddham [75]	2022	Canada	PC	4,471,348	Nd	Age > 66	RT-PCR	Nd	2019–2020 and 2020–2021 (2 cohorts)	7
Tayar [76]	2023	Qatar	CC	2,576	Nd	Healthworkers	RT-PCR	Inactivated 4v	2020–2021	7

CC = Case Control; CS = Cross Sectional; PC = Prospective Cohort; RC = Retrospective Cohort; Ned = Netherlands; SA = Saudi Arabia; US = United States; UK = United Kingdom; H = high dosage; IN = intranasal; 4v = tetravalent; 3v = trivalent; nd = not determined

Table 1. B.

Studies evaluating the association between influenza vaccination and hospitalization due to COVID-19. Authors (and references), year of publication, country, kind of study, sample size, mean age, population studied, method for diagnosis, kind of vaccine, vaccination season, and Newcastle Ottawa Scale (NOS) are reported.

study	year	country	Kind of study	Sample size	Mean age	Population	Method for diagnosis	Vaccine	Vaccination season	NOS
Ilic [47]	2020	Serbia	RC	107	39.1	Healthworkers	RT-PCR	3v, 4v	2019–2020	7
Zein [49]	2020	US	RC	13,220	55.4	General	Nd	Inactivated 4v	2019–2020	7
Ragni [52]	2020	Italy	CC	17,608	Nd	General	RT-PCR	4v, 3v with adjuvant	2019–2020	7
Conlon [59]	2021	US	RC	27,201	47.2	General	RT-PCR	Nd	2019–2020	8
Pawlowski [60]	2021	US	RC	25,582	74.6	General	RT-PCR	HD, recombinant 4v, IN	2019–2020	8
Bozek [63]	2021	Poland	RC	2,303	53.7	General	RT-PCR	Inactivated 4v	2020–2021	8
Hosseini-Mogaddham [75]	2022	Canada	PC	4,471,348	Nd	Age > 66	RT-PCR	Nd	2019–2020 and 2020–2021 (2 cohorts)	7
Gobbato [78]	2020	Italy	RC	3,010	60.0	General	Nd	Nd	2019–2020	7
Yang [79]	2021	US	RC	2,005	43.6	General	Nd	Nd	2019–2020	8
Wilcox [80]	2021	UK	RC	6,921	52.4	General	RT-PCR	Nd	2019–2020	8
Greco [81]	2021	Italy	RC	952	71.5	General	RT-PCR	Nd	2019–2020	8
Massari [82]	2021	Italy	PC	115,945	Nd	General	RT-PCR	Nd	2019–2020	8
Taghioff [83]	2021	US	RC	75,754	52.6	General	Nd	3v, IN, Inactivated	2019–2020 and 2020–2021	8
Giner-Soriano [84]	2022	Spain	RC	309,039	49.3	General	RT-PCR or suspected cases	Nd	2019–2020	8

CC = Case Control; CS = Cross Sectional; PC = Prospective Cohort; RC = Retrospective Cohort; US = United States; UK = United Kingdom; H = high dosage; IN = intranasal; 4v = tetravalent; 3v = trivalent; nd = not determined

Table 1. C.

Studies evaluating the association between influenza vaccination and admission to intensive care unit due to COVID-19. Authors (and references), year of publication, country, kind of study, sample size, mean age, population studied, method for diagnosis, kind of vaccine, vaccination season, and Newcastle Ottawa Scale (NOS) are reported.

Study	Year	Country	Kind of study	Sample size	Mean age	Population	Method for diagnosis	Vaccine	Vaccination season	NOS
Zein [49]	2020	US	RC	13,220	55.4	General	Nd	4v	2019–2020	7
Conlon [59]	2021	US	RC	27,201	47.2	General	RT-PCR	Nd	2019–2020	8
Pawlowski [60]	2021	US	RC	25,582	74.6	General	RT-PCR	HD, 4v, IN	2019–2020	8
Yang [79]	2021	US	RC	2,005	43.6	General	Nd	Nd	2019–2020	8
Taghioff [83]	2021	US	RC	75,754	52.6	General	Nd	3v, IN, inactivated	2019–2020 and 2020–2021	8
De La Cruz Conty [85]	2021	Spain	PC	1,150	33.0	Pregnant women	RT-PCR	Nd	2019–2020	7
Candelli [86]	2021	Italy	RC	602	60.6	General	RT-PCR	Nd	2019–2020	8
Umasabor-Bubu [87]	2021	US	RC	588	68.4	General	RT-PCR	Nd	2019–2020	7
Fernandez Ibanez [88]	2021	Spain	RC	410	70.7	General	RT-PCR	Nd	2019–2020	8
Kline [89]	2021	US	RC	149	58.1	General	RT-PCR	Nd	2019–2020	8
Paganoti [90]	2022	Brasil	RC	3,119	30,1	Pregnant women	RT-PCR or antigenic	Nd	2020–2021	7

CC = Case Control; CS = Cross Sectional; PC = Prospective Cohort; RC = Retrospective Cohort; US = United States; UK = United Kingdom; H = high dosage; IN = intranasal; 4v = tetravalent; 3v = trivalent; nd = not determined

Table 1. D.

Studies evaluating the association between influenza vaccination and mortality due to COVID-19. Authors (and references), year of publication, country, kind of study, sample size, mean age, population studied, method for diagnosis, kind of vaccine, vaccination season, and Newcastle Ottawa Scale (NOS) are reported.

Study	year	country	Kind of study	Sample size	Mean age	Population	Method for diagnosis	Vaccine	Vaccination season	NOS
Zein [49]	2020	US	RC	13,220	55.4	General	Nd	Inactivated 4v	2019–2020	7
Conlon [59]	2021	US	RC	27,201	47.2	General	RT-PCR	Nd	2019–2020	8
Bozek [63]	2021	Poland	RC	2,303	53.7	General	RT-PCR	Inactivated 4v	2020–2021	8
Xiang [67]	2021	UK	PC	27,147	Nd	General	RT-PCR	Nd	2019–2020	8
Hosseini-Mogaddham [75]	2022	Canada	PC	4,471,348	Nd	Age > 66	RT-PCR	Nd	2019–2020 and 2020–2021	7
Wilcox [80]	2021	UK	RC	6,921	52.4	General	RT-PCR	Nd	2019–2020	8
Greco [81]	2021	Italy	RC	952	71.5	Age > 66	RT-PCR	Nd	2019–2020	8
Massari [82]	2021	Italy	PC	115,945	Nd	General	RT-PCR	Nd	2019–2020	8
Taghioff [83]	2021	US	RC	75,754	52.6	General	Nd	3v, IN, inactivated	2019–2020 and 2020–2021	8
Giner-Soriano [84]	2022	Spain	RC	309,039	49.3	General	RT-PCR or suspected cases	Nd	2019–2020	8
Candelli [86]	2021	Italy	RC	602	60.6	General	RT-PCR	Nd	2019–2020	8
Umasabor-Bubu [87]	2021	US	RC	588	68.4	Age > 66	RT-PCR	Nd	2019–2020	7
Fernandez Ibanez [88]	2021	Spagna	RC	410	70.7	Age > 66	RT-PCR	Nd	2019–2020	8
Kline [89]	2021	US	RC	149	58.1	General	RT-PCR	Nd	2019–2020	8
Paganoti [90]	2022	Brasil	RC	3,119	30.1	Pregnant women	RT-PCR or antigenic	Nd	2019–2020 and 2020–2021	7
Azzi [91]	2020	US	PC	229	59.0	Kidney transplant	RT-PCR and IgG	Nd	2019–2020	5
Ortiz-Prado [92]	2020	Ecuador	CC	9,468	44.9	General	RT-PCR	Nd	2019–2020	7
Angulo-Zamudio [93]	2021	Mexico	RC	1,737	46.8	General	RT-PCR	Nd	2019–2020	8
El-Qutob [94]	2021	Spain	RC	255	68.4	Age > 66	RT-PCR	Nd	2019–2020	7

CC = Case Control; CS = Cross Sectional; PC = Prospective Cohort; RC = Retrospective Cohort; US = United States; UK = United Kingdom; H = high dosage; IN = intranasal; 4v = tetravalent; 3v = trivalent; nd = not determined

Quality Assessment

Quality of reports was assessed independently by reviewers according to Newcastle-Ottawa Quality Assessment Scale (NOS) for Cohort Studies [38]. The NOS scale is based on a “star system” in which a study is judged from three broad perspectives: study group selection, group comparability, and ascertaining the outcome of interest. The variables considered are: risk of bias linked to the selection of participants, confounding variables, performance, detection and measurement of exposure, attrition and reporting biases. Since there is not yet a validated version of the NOS for cross-sectional studies, a specially designed scale was used to evaluate them. Disagreement for the quality assessment was resolved by discussion. A score was eventually built, classifying the research articles as poor, intermediate, or good quality, based on the number of the above criteria available for each publication. The NOS score of each study is reported in Tables 1A to 1D.

Statistical Analysis

Comparison between Flu-vaccinated and not-vaccinated subjects for each of the four endpoints (frequency of infection, hospitalization, admission to ICU, and mortality) was expressed as odds ratio (OR), with 95% confidence intervals (CIs); all analyses were performed by a random-effects model according to DerSimonian and Laird [39]. When studies were ≤ 10, the Hartung-Knapp model was used to confirm analysis [40]. Heterogeneity was assessed through Q and I2 statistics for each comparison, and potential sources of heterogeneity were discussed where appropriate [41]. Heterogeneity was considered statistically significant for a p value <0.05. Sensitivity analyses were performed to evaluate sub-group effects, as well as to evaluate distance of vaccination to the event of interest (be it frequency of infection, hospitalization, admission to ICU, and mortality), and to evaluate the effect of different kinds of vaccine, when available. Through meta-regression, we evaluated the possible role of several patients’ and study characteristics on the frequency of endpoints. This was done independently of statistically significant heterogeneity. The dependent variable was the frequency of the event of interest (be it frequency of infection, hospitalization, admission to ICU, and mortality). The role of each covariate in heterogeneity was expressed by Wald test estimated by the meta-regression. The following covariates were included in the meta-regression analysis: number of subjects enrolled, age, kind of study (prospective or retrospective), distance of vaccination to the event of interest, method for diagnosis, kinds of vaccine, quality of studies (NOS). Meta-regression was performed considering all studies together. In a secondary analysis, we also evaluated the existence of a potential publication bias, that means the tendency of authors and editors to publish studies in which the experimental results achieved statistical significance, more favorably than in studies in which the results were not significant, which would ultimately introduce bias into the overall published literature [42]. Funnel-plot asymmetry was evaluated by using the Egger’s test for small study effects through the meta-bias routine [43]. All statistical analyses were performed by Stata 17 (Stata Corporation, College Station, TX, USA) for MacIntosh.

Results

Based on 33 papers analyzed, total subjects evaluated for the effect of influenza vaccination on SARS-CoV-2 infection were 61,029,936, of which 15,950,169 were vaccinated, and 45,079,687 were not vaccinated. Vaccinated subjects received one of the following influenza vaccines: i) tetravalent (n = 152,924); ii) trivalent (n = 12,900,000); iii) trivalent/tetravalent (n = 198,499). For 2,532,262 subjects it was not possible to determine the type of influenza vaccine that was administered. All the subjects taken into consideration did not receive any anti-SARS-CoV-2 vaccine, although we cannot exclude previous infections with SARS-CoV-2 in the populations analyzed.

Influenza vaccination was associated with a reduced frequency of SARS-CoV-2 infection, as shown in Table 2A, Figure 2. When sub-group analyses were performed, we found that the effect of influenza vaccination was significant in healthcare workers, in the general population, in elderly subjects, and in poor health individuals, when all the studies were considered together (Table 2A, Supplemental Figure 1). Nevertheless, more restrictive models (Hartung & Knapp) show a more conservative effect (Table 2A). Vaccination took place October to December, with no difference between different populations, and the distance was around 5 months (1 to 11 months) between vaccination and infection (5.1±2.35 months, hospitalization (5.0±1.41 months), admission to ICU (5.0±1.33 months), and death (5.5±1.37 months), with no differences between various populations; in sensitivity analysis, the time distance between influenza vaccination and the positivity to SARS-CoV-2 was also of relevance in all studies, as well in healthcare workers and in the general population (Table 2A, Supplemental Figure 2). Moreover, the protection against SARS-CoV-2 infection in individuals that received the tetravalent influenza vaccine (table 2A, Supplemental Figure 3, Supplemental Figure 4), but not other types of influenza vaccines (Supplemental Figure 5), was significant in all studies, both in healthcare workers and in the general population. In all studies, a short distance from influenza vaccination and use of tetravalent vaccine appeared to exert a synergistic effect against infection with SARS-CoV-2 (Table 2A, Supplemental Figure 6).

Table 2A.

Effect of influenza vaccination on SARS-CoV-2 infection in all studies and in different populations. Sensitivity analysis for all studies and for different populations are shown (distance from vaccination to infection, type of vaccine employed). When studies are less than 10, Dersimonian and Hartung & Knapp models were used.

	Dersimonian	Hartung & Knapp	Studi	Cochran Q, p
All studies	0.70 (0.65–0.77)		37	1486, 0.001
Distance <7 months	0.71 (0.64–0.78)		29	1217, 0.001
Any vaccine	0.71 (0.62–0.80)		19	893, 0.001
4v Vaccine	0.60 (0.48–0.74)		15	264, 0.001
4v.1 Vaccine	0.57 (0.49–0.67)		13	69, 0.001
4v Vaccine and distance <7 months	0.58 (0.48–0.70)	0.51 (0.29–0.87)	10	65, 0.001
4v.1 Vaccine and distance <7 months	0.56 (0.45–0.68)	0.47 (0.26–0.86)	9	62, 0.001
Healthcare workers	0.59 (0.43–0.81)		13	115, 0.001
Distance <7 months	0.57 (0.40–0.82)		11	101, 0.001
Any vaccine	0.55 (0.37–0.81)	0.50 (0.23–1.06)	8	105, 0.001
4v Vaccine	0.50 (0.36–0.70)	0.43 (0.20–0.95)	7	42, 0.001
4v.1 Vaccine	0.46 (0.32–0.65)	0.38 (0.15–0.93)	6	37, 0.001
4v Vaccine and distance <7 months	0.51 (0.35–0.75)	0.43 (0.17–1.09)	6	39, 0.001
4v.1 Vaccine and distance <7 months	0.45 (0.30–0.69)	0.36 (0.12–1.11)	5	35, 0.001
General population	0.77 (0.62–0.95)		13	297, 0.001
Distance <9 months	0.78 (0.63– 0.98)		12	291, 0.001
Any vaccine	0.69 (0.47–1.01)	0.69 (0.48–0.98)	6	195, 0.001
4v Vaccine	0.60 (0.54–0.66)	0.60 (0.51–0.71)	5	5.4, NS
4v.1 Vaccine	0.60 (0.54–0.65)	0.60 (0.51–0.71)	5	5.4, NS
4v Vaccine and distance <7 months	0.56 (0.51–0.62)	0.56 (0.50–0.64)	3	0.29 NS
4v.1 Vaccine and distance <7 months	0.53 (0.51–0.62)	0.56 (0.50–0.64)	3	0.29, NS
Elderly people	0.79 (0.71–0.88)	0.85 (0.60 −1.21)	5	179, 0.001
Distance <9 months	0.79 (0.71–0.88)	0.85 (0.60–1.21)	5	179, 0.001
Any vaccine	0.72 (0.71–0.72)	0.72 (0.71–0.72)	1	nd
Sick people	0.44 (0.30–0.65)	0.42 (0.09–1.89)	6	383, 0.001
Distance <7 months	0.30 (0.13–0.72)	0.31 (0.03–3.16)	5	341, 0.001
Any vaccine	0.94 (0.81–1.10)	0.89 (0.65–1.22)	4	39, 0.001
4v Vaccine	0.89 (0.66–1.20)	0.84 (0.49–1.42)	3	23, 0.001
4v.1 Vaccine	0.83 (0.49–1.42)	0.70 (0.39–1.27)	2	0.12, NS
4v Vaccine and distance <7 months	0.99 (0.73–1.34)	0.99 (0.71–1.38)	2	17, 0.001
4v.1 Vaccine and distance <7 months	0.84 (0.74–0.96)	0.84 (0.74–0.96)	1	nd

Significant effects at either Dersimonian or Hartung & Knapp models are in bold. 4v = tetravalent vaccine; 4v.1 = inactivated or attenuated tetravalent vaccine

Figure 2.

Frequency of SARS-CoV-2 infection and influenza vaccination. Left panel: forest plot of pooled hazard ratios of SARS-CoV-2 infection in vaccinated and not vaccinated subjects. Right panel: funnel plot with Egger detection of small bias effect.

Based on 14 papers analyzed that reported the information, total subjects evaluated for effect of influenza vaccination on hospitalization were 4,737,328, of which 2,602,803 vaccinated and 2,134,525 not vaccinated. Influenza vaccination was not associated with effect on hospitalization of infected subjects, as shown in Table 2B, Figure 3. Only two studies, one of which was performed in elderly, showed a significant effect.

Table 2B.

Effect of influenza vaccination on hospitalization in all studies and in different populations. Sensitivity analysis for all studies and for different populations are shown (distance from vaccination to hospitalization). When studies are less than 10, Dersimonian and Hartung & Knapp models were used.

	Dersimonian	Hartung & Knapp	Studi	Cochran Q, p
All studies	1.05 (0.82–1.35)		14	1249, 0.001
distance <4.5 months	0.54 (0.52–0.57)	0.53 (0.37–0.75)	2	0.2, NS
General population	1.15 (0.92–1.45)	1.14 (0.74–1.73)	9	340, 0.001
distance <7 months	1.18 (0.89–1.57)	1.17 (0.73–1.87)	9	313, 0.001
Healthworkers	0.71 (0.27–1.87)	0.71 (0.27–1.87)	1	nd
Elderly	0.88 (0.49–1.58)	0.86 (0.40–1.87)	4	86, 0.001
distance <4.5 months	0.54 (0.52–0.57)	0.53 (0.37–0.75)	1	nd

Significant effects at either Dersimonian or Hartung & Knapp models are in bold.

Figure 3.

Frequency of hospitalization in SARS-CoV-2 infected subjects and influenza vaccination. Left panel: forest plot of pooled hazard ratios of hospitalization due to COVID-19 in vaccinated and not vaccinated subjects. Right panel: funnel plot with Egger detection of small bias effect.

Based on 11 studies analyzed that reported the information, total subjects evaluated for effect of influenza vaccination on ICU admission were 98,174, of which 44,553 were vaccinated and 53,621 were not vaccinated. Influenza vaccination was associated with a significant reduction of ICU admission, as shown in Table 2C, Figure 4. The effect was significant in all studies together, in pregnant women and in hospitalized subjects, when the two latter groups were considered together, and in the general population when age was < 50 years; the effect was even greater when pregnant women, hospitalized subjects, and general population with age < 50 years were combined Table 2C, Supplemental Figure 7).

Table 2C.

Effect of influenza vaccination on admission to Intensive Care Units in all studies and in different populations. Sensitivity analysis for all studies and for different populations are shown (age. Combined populations). When studies are less than 10, Dersimonian and Hartung & Knapp models were used.

	Dersimonian	Hartung & Knapp	Studi	Cochran Q, p
All studies	0.71 (0.54–0.94)		11	47, 0.001
Age < 60 years	0.69 (0.49–0.96)	0.65 (0.37–1.14)	7	34, 0.001
Age < 50 years	0.44 (0.22–0.87)	0.43 (0.18–1.02)	4	19, 0.001
General population	0.77 (0.51–1.19)	0.75 (0.41–1.37)	7	37, 0.001
Age < 60 years	0.63 (0.36–1.10)	0.63 (0.36–1.10)		33, 0.001
Age < 50 years	0.27 (0.17–0.41)	0.25 (0.12–0.52)	2	0.55 NS
Pregnant women	0.74 (0.61–0.90)	0.74 (0.61–0.90)	2	0.0, NS
Hospitalized subjects	0.46 (0.30–0.70)	0.42 (0.20–0.89)	2	0.58, NS
Pregnant women + Hospitalized subjects	0.61 (0.44–0.85)	0.57 (0.34–0.97)	4	4, NS
Pregnant women + Hospitalized subjects + General population aged< 50 years	0.44 (0.28–0.70)	0.43 (0.23–0.78)	6	21, 0.001

Significant effects at either Dersimonian or Hartung & Knapp models are in bold.

Figure 4.

Frequency of admission to intensive care units in SARS-CoV-2 infected subjects and influenza vaccination. Left panel: forest plot of pooled hazard ratios of admission to Intensive Care Units due to COVID-19 in vaccinated and not vaccinated subjects. Right panel: funnel plot with Egger detection of small bias effect.

Based on 19 studies that reported the information, total subjects evaluated for effect of influenza vaccination on mortality were 4,139,660, of which 2,703,073 were vaccinated and 1,436,587 were not vaccinated. Influenza vaccination was not associated with a significant effect on mortality, as shown in Table 2D, Figure 5; an effect was only observed in one study in pregnant women, and in one study in which the distance from influenza vaccination and SARS-CoV-2 infection was < 4 months; however, combination of the two studies did not show any effect. Newcastle Ottawa Scales (NOSs) were generally high, indicating a good quality of the studies. No significant meta-regression appeared for any of the endpoints (frequency of infection, hospitalization, admission to ICU, and mortality) and patients’ and study characteristics. No publication bias appeared for any of the comparisons. However, heterogeneity was virtually always very high, as reported in Tables 2A to 2D.

Table 2D.

Effect of influenza vaccination on mortality in all studies and in different populations. Sensitivity analysis for all studies and for different populations are shown (distance from vaccination to death, age). When studies are less than 10, Dersimonian and Hartung & Knapp models were used.

	Dersimonian	Hartung & Knapp	Studi	Cochran Q, p
All studies	0.76 (0.26–2.20)		20	24920, 0.001
Distance < 5 months	0.75 (0.06–9.28)	0.75 (0.17–3.23)	7	17978, 0.001
Distance < 4 months	0.01 (0.01–0.01)	0.01 (0.01–0.01)	1	nd
Pregnant women	0.39 (0.28–0.53)	0.39 (0.28–0.53)	1	nd
General population	0.98 (0.51–1.89)		12	3473, 0.001
Distance < 5 months	1.36 (0.63–2.96)	1.34 (0.56–3.24)	4	205, 0.001
Age > 65 years	0.56 (0.04–7.83)	0.56 (0.0–14.32)	5	3820, 0.001
Distance < 5 monthss	0.14 (0.01–19.20)	0.14 (0.01–131.68)	2	219, 0.001
Distance < 4 months	0.01 (0.01–0.01)	0.01 (0.01–0.01)	1	nd
Hospitalized subjects	0.79 (0.08–7.82)	0.79 (0.06–9.77)	2	28, 0.001
Pregnant women + age > 65 years with distance < 4 months	0.57 (0.27–1.21)	0.57 (0.23–1.42)	2	421, 0.001

Significant effects at either Dersimonian or Hartung & Knapp models are in bold.

Figure 5.

Mortality in SARS-CoV-2 infected subjects and influenza vaccination. Left panel: forest plot of pooled hazard ratios of mortality due to COVID-19 in vaccinated and not vaccinated subjects. Right panel: funnel plot with Egger detection of small bias effect.

Discussion

In 2020, when vaccines against SARS-CoV-2 were not yet available, preventive measures such as face masks and social distancing were the only remedies to combat COVID-19. A non-specific protection against SARS-CoV-2 was initially hypothesized for the BCG vaccine, whose favorable effects on child mortality went beyond prevention of child TBC [95]. However, the efficacy of BCG vaccine has not been confirmed in spite of early reports [34, 35]. The influenza vaccination was strongly recommended since the beginning of the pandemics to avoid co-infections [26]) and early studies highlighted a possible association between influenza vaccination and a reduced SARS-CoV-2 infection [27–30]. Since influenza almost disappeared during the 2020 season [17–23] and thus the prevention of co-infection could not explain this possible protective effect of the influenza vaccination on COVID-19, it was hypothesized that influenza vaccination directly played protective roles against SARS-CoV-2 infection. The ensuing studies on the association between influenza vaccination and SARS-CoV-2 infection and COVID-19 severity were the basis for this meta-analysis [44–94].

The main findings of our meta-analysis are that influenza vaccination reduces the frequency of SARS-CoV-2 infection but has very small effects on severity of COVID-19 in infected subjects, as shown by frequency of hospital admission and mortality. Nevertheless, isolation in ICU appeared to be reduced for several populations analyzed. In particular:

1. Influenza vaccination reduces the frequency of SARS-CoV-2 infection

Coming from the analysis of 33 studies (for a total of 37 comparison arms and over 60-million individuals), this result supports and extends previous observations and reviews [31–33]. The novelty of our findings is represented by the different populations examined (healthcare workers, general population, elderly individuals, poor health individuals, and pregnant women), by the analysis of the vaccines used (tetravalent vs others), and by the possible interaction of type of vaccine and distance from vaccination. When the information about the type of influenza vaccine employed was described, we could not identify any difference in the effect of tetravalent vaccine and of inactivated tetravalent vaccine. In all studies, a short distance from influenza vaccination and use of tetravalent vaccine appeared to exert a synergistic effect against infection with SARS-CoV-2. These data support the hypothesis of the existence of a direct effect of the influenza vaccination on SARS-CoV-2 infection. This effect can be either specific or non-specific, as discussed below.

2. Influenza vaccination has negligible effects on severity of COVID-19 in infected subjects

2.1. Hospitalization

Only two studies out of 14, in which subjects were elderly people vaccinated less than 4.5 months earlier, showed a significant effect of influenza vaccination against hospitalization.

2.2. ICU admission

In contrast to hospitalization, influenza vaccination had a significant effect on prevention of ICU admission; this applied to all studies, to the general population below the age of 50 years, to pregnant women, to hospitalized patients, and the effect was even more significant when the three latter categories were combined.

2.3. Mortality

Only two out of 20 studies showed some effect of influenza vaccination on mortality caused by SARS-CoV-2, one in pregnant women, the other in elderly people vaccinated by less than four months; interestingly, combination of the two studies did not significantly change the results of our analysis.

The possible mechanisms of action

At present, there is no explanation for the preventive effect of influenza vaccination against SARS-CoV-2 infection. Influenza viruses (mainly represented by influenza A virus) and SARS-CoV-2 are different viruses, although they share routes of transmission and part of pulmonary symptomatology [5–7].

Therefore, a direct effect on SARS-CoV-2 of the adaptive immune response (antibody production and/or T cell responses) elicited by influenza vaccination seems unlikely.

An alternative mechanism to explain the protection of influenza vaccination against the infection with SARS-CoV-2 might be represented by trained immunity [96]. Trained immunity is a phenomenon that potentiate the responses of innate immune cells such as macrophages, neutrophils, and dendritic cells via the induction of “memory-like” features that are induced in response to infections or vaccines [96–98]. These memory-like responses lead to enhanced protection upon (re)encounter with the same, or a non-related, microorganism. The existence of trained immunity was initially proposed for the BCG vaccine [96]. Children, who are not in the general indications for influenza vaccines, were very little infected during the first year of the SARS-CoV-2 pandemic; the increased interferon responses in children [99, 100] would protect against SARS-CoV-2 even in the absence of trained immunity, not necessary because of a strong primary response of the host.

Vaccines usually affect adaptive immunity, namely B and T lymphocytes. In the case of trained immunity, infections and vaccines can re-program cells of innate immunity, with a possible improvement of the non-specific response to other infections. This protective effect appears to be long-lasting (between 3 months and 1 year, although, for live vaccines, the protection against heterologous infections has been shown up to 5 years) and reminds adaptive immune memory, although without antigen-specificity [101$]$. Influenza vaccination can stimulate the activity of NK cells, which are innate immune lymphocytes that rapidly respond to an infection [98, 102] involved in the elimination of virus-infected cells by shaping the adaptive response mediated by T-cells, and that have been shown to bear a certain level of “memory” [103, 104]. Activated NK cells are relevant source of INF-g which has a crucial role in instructing CD8⁺ T cell expansion and contraction, as adaptive immune responses to several pathogens, including viruses [105, 106]. In severe SARS-CoV2 infection, NK cells are hypo-responsive, [107] and the anti-SARS-CoV2 vaccination (based on mRNA technology) transiently stimulates high production of NK cells and B cell [107].

In one study [69], influenza vaccination was associated with changes of monocytes and CD4+ lymphocytes, in spite of no change of total circulating leukocytes, and with the downregulation of mediators of the systemic inflammatory response and an increased production of anti-inflammatory cytokines. Intriguingly, a similar mechanism, based on regulation of inflammation, has been suggested for a totally different drug, metformin, that was able to reduce the risk of severe COVID-19 and of long COVID [108].

Interferons might be a key factor in the interaction between influenza vaccination and SARS-CoV-2 infection; different types of interferons are produced in the lungs of COVID-19 patients based on disease severity and location along the respiratory tract [109, 110]. It is thus possible to speculate that the transient induction of interferons that follows influenza viruses encounter, or the influenza vaccination may affect the capacity of the host to control the infection with SARS-CoV-2 [111–113].

Our data demonstrate that the protective effect of influenza vaccination fades over time. Influenza vaccines have been shown to induce interferons to different extents [111–113], suggesting that the broad capacity of interferons, produced in response to influenza vaccine, to “interfere” with the spread and/or replication of viruses may explain the cross-protection against SARS-CoV-2 observed in individuals that were recently vaccinated against influenza.

An additional factor may be represented by the “healthy vaccinee” (or “healthy user”) effect, that can act as a confounding factor. Motivated subjects, together with voluntary vaccination, might also observe in a more proper and strict way other sanitary measures, that can by themselves prevent infection, leading to an over-evaluation of the activity of the influenza vaccines [114]. In particular, one study [76] was aimed at reducing differences between general population and healthcare workers in the approach to vaccination. In agreement with this observation, the effect of influenza vaccines was somewhat more evident in healthcare workers than in the general population.

Limitations

There are several limitations to this meta-analysis. First, no controlled trial was available, and only prospective cohort studies, retrospective cohort studies, case control studies, and cross-sectional studies were available. Second, we should consider that the incidence and severity of SARS-CoV-2 infections have changed from one year to the next during the pandemic, and that the severity has particularly decreased in 2022; although the extent and complexity (in terms of geographic location and period of sampling) of the cohorts utilized for this study cannot take into account the changes in the pathogenicity associated with the emergence of new variants of concern [115], the associations we identified seem to hold true regardless of these differences.

A further limitation is that studies showing a relationship between frequency of vaccination and diffusion of infection could not be considered because they did not show crude data [27–30]. Third, the type of vaccine used was known only for a limited number of individuals, and therefore an analysis of this aspect was not possible across all our analyses. Fourth, we applied two models of analysis, and some of the comparisons were statistically significant only with one of the models. On the other hand, the quality of studies, as assessed through the Newcastle Ottawa Scale was elevated, with a few exceptions, and no publication bias was evident. Finally, even though all subjects in the studies did not receive any anti-SARS-CoV-2 vaccine, we had no information about previous infection with SARS-CoV-2, and, with few exceptions [50, 51, 57, 63, 65, 71, 88, 89], about recent infections with the influenza virus, and we had no data on the possible involvement of the cardiovascular system in the studies evaluated [1–4].

Conclusion

Influenza vaccination was associated with reduced rate of SARS-CoV-2 infection in the majority of studies examined. In contrast, influenza vaccination seems to exert minimal effects on the severity of COVID-19, as assessed through rates of hospitalization and death rates. Of note, influenza vaccination reduced the admission to ICU for several populations analyzed. Our data, thus, show that influenza vaccination prevents the infection with SARS-CoV-2, but that, upon infection, it does not alter the response of the host to this new coronavirus. This preventive effect might be of relevance for individuals with co-morbidities, in particular patients with or at risk of cardiovascular diseases [1–4]. Overall, our analysis reveals the importance, together with other preventive measures, of influenza vaccination against COVID-19, that reduces the impact of SARS-CoV-2 and of similar viruses on the general population and of more exposed subjects [116].

DATA AVAILABILITY STATEMENT:

Data are available from the corresponding authors on the basis of reasonable request.