Association of Influenza Vaccination and Dementia Risk: A Meta-Analysis of Cohort Studies

Abstract

Background:

Dementia is a critical global public health problem. Previous cohort studies have found that influenza vaccination can decrease the risk of dementia.

Objective:

This meta-analysis aimed to systematically examine the relationship between influenza vaccination and dementia risk.

Methods:

We searched PubMed, Embase, Web of Science, ScienceDirect, medRxiv, and bioRxiv for studies investigating dementia risk based on influenza vaccination status, up to September 14, 2022. Relative risks (RRs) and 95% confidence intervals (95% CIs) were pooled in the meta-analysis. Subgroup analyses and sensitivity analyses were conducted as well.

Results:

Of the 4,087 articles initially reviewed, 6 cohort studies were included in the final meta-analysis, and all eligible studies were at low risk of bias. There were 2,087,195 participants without dementia at baseline (mean age: 61.8–75.5 years, 57.05% males), and 149,804 (7.18%) cases of dementia occurred during 4.00–13.00 years of follow-up. Pooled analysis of adjusted RRs found that influenza vaccination could reduce dementia risk by 31% (RR = 0.69, 95% CI: 0.57–0.83). Subgroup analyses showed that in the study with a mean age of 75–80 years or 75%–100% males, the association was generally weakened compared with studies with a mean age of 60–75 years or 25%–50% males. The results were stable in the sensitivity analyses, and no publication bias was observed.

Conclusion:

Influenza vaccination in older adults was markedly associated with a decreased risk of dementia. More mechanistic studies and epidemiological studies are needed to clarify the association between influenza vaccination and decreased dementia risk.

Keywords

Alzheimer’s disease dementia influenza meta-analysis vaccination

INTRODUCTION

Dementia is now the 7th leading cause of mortality globally and one of those with the highest cost to society [1]. The World Alzheimer Report, released in 2018, estimates that at least 55 million people are living with dementia globally, and its prevalence is nearly doubled every 20 years, with the total number rising to 139 million by 2050 [2]. For individuals, dementia can influence the daily activities, living quality, as well as physical and psychological health of the elderly patients to a great extent. Besides, people who diagnosed with severe dementia can easily become completely dependent on their caregivers, which poses mental health risks to caregivers around them [3]. For society, dementia imposes a huge burden on the world economy, with an estimated cost of US$818 billion currently and is expected to increase to US$2 trillion by 2030 [4]. Nowadays, most approved treatments for dementia such as aducanumab provide only limited symptom relief [5], and dementia still cannot be eradicated completely [6]. Therefore, it is of great importance to discover a new method for preventing or delaying the onset of dementia.

The influenza vaccine, which is recommended for all people aged 6 months and older with no contraindications [7], can prevent not only influenza, but also serious complications such as pneumonia and myocarditis [8]. In addition, Davtyan et al. designed a dual vaccine against influenza and Alzheimer’s disease (AD) based on conventional influenza vaccines, which was proved to be safe and effective for AD in both cell and animal experiments [9, 10]. In recent years, influenza vaccination has been found to protect against dementia in older adults, with epidemiological evidence in a large veterans cohort during 7.5 years of follow-up [11], in a cohort of older adults during 4 years in the United States [12], and in several cohorts in Taiwan during 8–13 years [13 –15]. Veronese et al. have conducted a meta-analysis of 5 articles from 2021 and before without subgroup analyses and found that influenza vaccination could reduce the risk of dementia by 29% after adjusting for confounders [16]. But this year, an article published by Bukhbinder et al. added the latest evidence to this line of research [12]. In this meta-analysis, we aimed to systematically analyze all cohort studies to date on the association between influenza vaccination and dementia risk. We also performed subgroup analyses to examine the subgroup differences, which was missing from previous meta-analysis [16]. Hopefully, our study may provide new evidence for the early prevention of dementia.

MATERIALS AND METHODS

Search strategy

In accordance with the updated PRISMA guidelines and MOOSE guidelines [17, 18], we conducted a systematic review and meta-analysis. We searched PubMed, ScienceDirect, Embase, medRxiv, Web of Science, and bioRxiv from inception to 14 September 2022 for potentially relevant studies without language restrictions. The search terms were as follows: (flu OR influenza) AND (vaccin^*) AND (Alzheimer^* OR dement^* OR Lewy OR “Posterior cortical atrophy” OR “Binswanger” OR “Progressive supranuclear palsy” OR “Frontotemporal disorder” OR “Corticobasal degeneration”) [16]. References of review papers were examined as well to identify other eligible studies in the database that were not covered by the original search. The study was registered with PROSPERO (CRD42022345577).

Selection criteria

The research selection was divided into two stages: the preliminary topic selection and the abstract, and then the full-text review of the papers that may meet the requirements. Two review authors (HS and JL) independently assessed the eligibility, and a third researcher (ML) eliminated the differences. Studies that met the following criteria were included: 1) observational studies with a prospective or retrospective cohort design, 2) evaluated associations between influenza vaccination and dementia risk, 3) dementia was diagnosed based on the international diagnostic criteria, and 4) contained sufficient data to calculate hazard ratio (HR), relative risk (RR), or odds ratio (OR) with 95% confidence interval (CI) on the basis of multivariate adjustment [19]. We also used some exclusion criteria: 1) case reports, cross-sectional studies, editorials, commentaries, reviews, conference abstracts, and non-human studies, 2) not relevant to the subject of the meta-analysis, such as studies that did not use influenza vaccination as exposure, 3) studies that did not report any RR (or OR, HR) and 95% CI, 4) duplicate reports or duplicate participants, and 5) studies that did not indicate diagnostic methods for dementia. For redundant publications, only study with the largest sample size or most comprehensive information was chosen. Eligible non-English studies were translated by translators or translation software if necessary.

Data extraction

Data were independently extracted from eligible studies by two authors (HS and JL) using the data extraction sheets. Possible disagreements were resolved by consensus and a third author (ML) if necessary. The information extracted from eligible studies were as follows: 1) title, 2) first author, 3) year of publication, 4) study design, 5) WHO Region, 6) the number of all participants and dementia cases, 7) percentage of males and age range (mean±SD), 8) follow-up length, 9) participant’s background disease, 10) methods of ascertaining influenza vaccination, 11) diagnosis criteria for dementia, 12) effect estimates (HR, RR or OR) with 95% CI, 13) statistical methods, and 14) covariates for adjustment.

Quality assessment

Two authors (HS and JL) evaluated the quality of each eligible study based on the Newcastle-Ottawa Scale (NOS) [20], and disagreements were resolved through consensus. The following items were considered: ascertainment of outcomes of interest (maximum = 3 stars), comparability of study groups (maximum = 2 stars), and selection of participants (maximum = 4 stars). Thus, the highest quality was represented by 9 stars. Those studies scored≤4, 5-6, and 7–9 stars were at high, medium, and low risk of bias, respectively.

Definitions

In this study, dementia was diagnosed based on the International Statistical Classification of Diseases and Related Health Problems 9th Revision (ICD-9), International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10), and Diagnostic and Statistical Manual of Mental Disorders 4th edition (DSM-IV).

Data synthesis and analysis

Statistical analyses were conducted with STATA 17 software. All statistical tests were two-sided, and a p-value of < 0.05 was deemed statistically significant. We selected risk estimates from the multivariate models after full adjustment for covariates. The incidence of dementia was relatively low (less than 10% [21, 22]), so OR, RR, and HR were treated equally and represented by RRs in this meta-analysis [19, 23]. The RRs and 95% CIs were regarded as the effect size for each included study. The heterogeneity across studies was quantified using the I² statistic (I² = 0–60% none to moderate heterogeneity, I² > 60% substantial to high heterogeneity). If the heterogeneity was observed to be low, a fixed-effect model was employed, otherwise the random-effect model (DerSimonian and Laird method [24]) was chosen to bolster the results.

To examine the subgroup differences and potential heterogeneity sources, subgroup analyses were conducted based on the publication year, study design, WHO Region, sample size, follow-up length, population, mean age, sex ratio, exposure measurement, diagnostic criteria for dementia, and adjustments. Funnel plots and Egger’s tests were applied to evaluate potential publication bias. To assess the robustness of the pooled data, sensitivity analyses were performed in two ways, by excluding studies one by one and by excluding articles from a particular country, such as the United States. The impact of each study and a certain type on the overall results were both evaluated.

RESULTS

Characteristics of eligible studies

As shown in Fig. 1, we initially screened 4,087 articles. After the removal of duplicates, there left 3,547 articles, of which 3,541 were irrelevant studies, conference abstracts, commentaries, case reports, editorials, reviews, non-human studies, and cross-sectional studies based on screening of titles and abstracts. Among the 33 examined through the full-text check, 6 eligible studies were ultimately included in this meta-analysis [11–15 , 25].

Fig. 1

Flow chart of the identification of eligible studies.

The characteristics of the 6 studies published between 2001 and 2022 are summarized in Table 1. Of these studies, three were carried out in the Western Pacific Region [13 –15], and the other three in the Region of the Americas [11 , 25]. All studies comprised a retrospective cohort design except for one, which was a prospective cohort study [25]. Two studies were conducted with general patient population samples [11 , 25], while three included specific patient subgroups with chronic conditions (chronic obstructive pulmonary disease [13], chronic kidney disease [15], and periodontitis [14]) and one included veterans patient population [11]. Of 2,087,195 participants (57.05% males, mean age: 61.8–75.5 years) that had no dementia at baseline, a total of 149,804 (7.18%) were diagnosed with dementia during 4.00–13.00 years of follow-up. Methods of ascertaining influenza vaccination included CPT/HCPCS codes [11, 12], ICD-9-CM [13 –15], and questionnaire [25]. All studies reported the diagnosis of dementia in accordance with the ICD-9 [11 –15], ICD-10 [11, 12], or DSM-IV codes [25]. Two recent studies mentioned that dementia was defined by the presence of two or more ICD-9/ICD-10 codes on separate days within the same 12-month period [11, 12]. Four studies used Cox proportional-hazards model for statistical analysis [11 , 13–15] and one used multivariate logistic regression models [25]. In addition, Bukhbinder et al. used propensity score matching (PSM) and Wiemken et al. used inverse probability of treatment weighting (IPTW) to control the confounding factors [11, 12]. Based on the NOS values, all 6 studies received at least 8 stars, implying a low risk of bias for these studies.

Table 1

Main characteristics of studies included in the meta-analysis

Study	Study design	WHO Region	Sample size	Follow-up length (y)	Population	Age range (y) mean±SD	Males (%)	Participants with dementia	Exposure measurement	Definition of dementia	Statistical methods	Adjustments	NOS
Bukhbinder, 2022 [12]	Retrospective cohort study	Region of the Americas	1871774	4	General population	73.7±6.3	43.1	127519	CPT/HCPCS codes	ICD-9, ICD-10^a	PSM	Age, sex, race, geographic region, No. of healthcare visits during look-back, No. of routine “well visit” examinations, comorbidities, medications	8
Wiemken, 2021 [11]	Retrospective cohort study	Region of the Americas	123747	7.5	Veterans	75.5±7.3	96.2	15933	CPT/HCPCS codes	ICD-9, ICD-10^a	IPTW, Cox proportional-hazards model	Age, sex, marital status, race, region, access to private health insurance in addition to VHA coverage, healthcare utilization, comorbidities, medications	9
Luo, 2020 [13]	Retrospective cohort study	Western Pacific Region	19848	11	COPD	71.9±7.9	57.8	2452	ICD-9-CM code V048	ICD-9	Cox proportional-hazards model	COPD-related hospitalization, age, sex, comorbidities, level of urbanization, monthly income	9
Lee, 2020 [14]	Retrospective cohort study	Western Pacific Region	56018	13	Periodontitis	≥50 (mean 61.8)	48.4	2246	ICD-9-CM	ICD-9	Cox proportional-hazards model	Age, sex, comorbidities, medications	9
Liu, 2016 [15]	Retrospective cohort study	Western Pacific Region	11943	8	Chronic kidney disease	73.0±8.0	56.8	1471	ICD-9-CM code V048	ICD-9	Cox proportional-hazards model	Age, sex, comorbidities, level of urbanization, monthly income	9
Verreault, 2001 [25]	Prospective cohort study	Region of the Americas	3865	5	General population	≥65 (mean 74.8)	40	183	questionnaire	DSM-IV	Multivariate logistic regression models	Age, sex, education	9

Main characteristics of 6 eligible studies included in the meta-analysis. WHO, World Health Organization; SD, standard deviation; NOS, Newcastle-Ottawa Scale; PSM, propensity score matching; IPTW, inverse probability of treatment weighting; COPD, chronic obstructive pulmonary disease; CPT, Current Procedural Terminology; HCPCS, Health Care Product Code for Supplies; ICD, the International Statistical Classification of Diseases and Related Health Problems; DSM, the Diagnostic and Statistical Manual of Mental Disorders; VHA, Veterans Health Administration. ^aDementia was defined by the presence of two or more ICD codes on separate days within the same 12-month period.

Influenza vaccination and dementia risk

As shown in Fig. 2, this meta-analysis revealed that compared to those who had not received any influenza vaccination, individuals who had at least one influenza vaccination had a 0.69-fold lower risk of developing dementia (pooled random-effect RR = 0.69, 95% CI: 0.57–0.83). Heterogeneity across all studies was substantial and high (I² = 98.9%, p = 0.000). Since the RR estimates of all eligible studies were adjusted for age and sex (Table 1), it can be assumed that the pooled random-effect RR was not affected by age and sex, which were two confounding factors in the analysis.

Fig. 2

Forest plot of the effect of influenza vaccination on the risk of dementia (The results are expressed as RRs and 95% CIs). RR, relative risk; CI, confidence interval; DL, DerSimonian and Laird method.

Subgroup analyses were performed according to the characteristics of eligible studies. There were significant differences between subgroups in follow-up length (p = 0.001), sample size (p = 0.000), average age (p = 0.000), and percentage of males (p = 0.000) (Table 2). In the study with an average age of 75–80 years [11], the association was weakened compared with studies with an average age of 70–75 years. The correlation was also weakened in the Wiemken’s study with 75%–100% males compared to studies with 25%–50% males. To determine whether these associations were influenced by confounding factors, subgroup analyses were performed according to adjustments such as race, region, comorbidities, medications, healthcare utilization, income, level of urbanization, marital status, and education. The subgroup difference test showed that the subgroup effect of marital status (p = 0.000) was statistically significant (Table 3). This association was weaker in the Wiemken’s study that adjusted for marital status (RR = 0.86) [11] than in studies that did not adjust for marital status (RR = 0.65).

Table 2

Subgroup analyses of the association between influenza vaccination and dementia

Study characteristics	No. Studies	No. participants	No. cases	RR	95% CI		p within groups	p between groups	I² (%)
Overall	6	2087195	149804	0.69	0.57	0.83	0.000	–	98.9
Study design
Retrospective cohort study	5	2083330	149621	0.69	0.56	0.84	0.000	0.645	99.1
Prospective cohort study	1	3865	183	0.75	0.54	1.04	–		0.0
WHO Region
Region of the Americas	3	1999386	143635	0.73	0.54	0.98	0.000	0.581	99.5
Western Pacific Region	3	87809	6169	0.67	0.63	0.70	0.693		0.0
Publication year
Before 2018	2	15808	1654	0.65	0.59	0.72	0.368	0.591	0.0
After 2018	4	2071387	148150	0.70	0.55	0.88	0.000		99.3
Follow-up length (y)
< 5	1	1871774	127519	0.60	0.59	0.61	–	0.001	0.0
5–10	3	139555	17587	0.75	0.59	0.95	0.000		92.4
10–15	2	75866	4698	0.68	0.63	0.72	0.825		0.0
Sample size
> 1500000	1	1871774	127519	0.60	0.59	0.61	–	0.000	0.0
100000–500000	1	123747	15933	0.86	0.83	0.88	–		0.0
50000–100000	1	56018	2246	0.67	0.61	0.74	–		0.0
10000–50000	2	31791	3923	0.66	0.62	0.71	0.399		0.0
< 10000	1	3865	183	0.75	0.54	1.04	–		0.0
Population
General population	2	1875639	127702	0.63	0.53	0.76	0.183	0.347	43.7
Non-general population	4	211556	22102	0.71	0.60	0.84	0.000		95.3
Mean age (y)
60–65	1	56018	2246	0.67	0.61	0.74	–	0.000	0.0
70–75	4	1907430	131625	0.64	0.59	0.70	0.017		70.7
75–80	1	123747	15933	0.86	0.83	0.88	–		0.0
Percentage of males
25%–50%	3	1931657	129948	0.64	0.57	0.71	0.037	0.000	69.7
50%–75%	2	31791	3923	0.66	0.62	0.71	0.399		0.0
75%–100%	1	123747	15933	0.86	0.83	0.88	–		0.0
Exposure measurement
CPT/HCPCS codes	2	1995521	143452	0.72	0.50	1.02	0.000	0.724	99.8
ICD-9 code	3	87809	6169	0.67	0.63	0.70	0.693		0.0
Questionnaire	1	3865	183	0.75	0.54	1.04	–		0.0
Definition of dementia
ICD-9/ICD-10	2	1995521	143452	0.72	0.50	1.02	0.000	0.724	99.8
ICD-9	3	87809	6169	0.67	0.63	0.70	0.693		0.0
DSM-IV	1	3865	183	0.75	0.54	1.04	–		0.0

Subgroup analyses of the association between influenza vaccination and dementia by study design, WHO Region, publication year, follow-up length, sample size, population, average age, sex ratio, exposure measurement, and clinical tools used for dementia diagnosis. RR, relative risk; CI, confidence interval; WHO, World Health Organization; COPD, chronic obstructive pulmonary disease; CPT, Current Procedural Terminology; HCPCS, Health Care Product Code for Supplies; ICD, the International Statistical Classification of Diseases and Related Health Problems; DSM, the Diagnostic and Statistical Manual of Mental Disorders.

Table 3

Subgroup analyses according to adjustments

Adjustments	No. Studies	No. participants	No. cases	RR	95% CI		p within groups	p between groups	I² (%)
Overall	6	2087195	149804	0.69	0.57	0.83	0.000	–	98.9
Race
Yes	2	1995521	143452	0.72	0.50	1.02	0.000	0.692	99.8
No	4	91674	6352	0.67	0.63	0.71	0.748		0.0
Region
Yes	2	1995521	143452	0.72	0.50	1.02	0.000	0.692	99.8
No	4	91674	6352	0.67	0.63	0.71	0.748		0.0
Comorbidities
Yes	5	2083330	149621	0.69	0.56	0.84	0.000	0.645	99.1
No	1	3865	183	0.75	0.54	1.04	–		0.0
Medications
Yes	3	2051539	145698	0.70	0.53	0.93	0.000	0.727	99.5
No	3	35656	4106	0.67	0.62	0.71	0.544		0.0
Income
Yes	2	23713	3923	0.66	0.62	0.71	0.399	0.604	0.0
No	4	2055404	145881	0.71	0.55	0.91	0.000		99.3
Level of urbanization
Yes	2	23713	3923	0.66	0.62	0.71	0.399	0.604	0.0
No	4	2055404	145881	0.71	0.55	0.91	0.000		99.3
Marital status
Yes	1	123747	15933	0.86	0.84	0.89	–	0.000	0.0
No	5	1963448	133871	0.65	0.60	0.70	0.006		72.6
Education
Yes	1	3865	183	0.75	0.54	1.04	–	0.645	0.0
No	5	2083330	149621	0.69	0.56	0.84	0.000		99.1

Subgroup analyses of the association between influenza vaccination and dementia according to adjustments, including race, region, comorbidities, medications, income, level of urbanization, marital status, and education. RR, relative risk; CI, confidence interval.

Publication bias and sensitivity analyses

Funnel plots are shown in Fig. 3. Egger regression tests for the relative risks among the tested population (p = 0.207 > 0.05) indicated that there was no obvious publication bias.

Fig. 3

Funnel plots to estimate publication bias.

To determine the influence of eligible studies on the overall results, sensitivity analyses were carried out. As shown in Fig. 4, the combined results of the remaining studies were not affected significantly after excluding any single study or studies from the United States. Hence, this meta-analysis was considered stable. From the perspective of heterogeneity, when Bukhbinder’s study [12] and Wiemken’s study [11] were removed separately (Fig. 4A), the heterogeneity of the remaining articles decreased (I² = 93.7%, I² = 72.6%); When we removed Bukhbinder’s study and Wiemken’s study [11, 12] at the same time (Fig. 4B), the heterogeneity of the remaining 4 articles disappeared (I² = 0.0%).

Fig. 4

Sensitivity analyses of eligible studies. 4A) Sensitivity analysis of the remaining 5 articles after removing each article one by one. The three long vertical lines in 4A correspond to the pooled RR and 95% CI (RR = 0.69, 95% CI: 0.57–0.83) in the forest plot. The circles and horizontal lines in 4A represent the pooled RRs and 95% CIs of the remaining 5 articles. 4B) Sensitivity analysis of the remaining 4 articles after removing studies from America. RR, relative risk; CI, confidence interval; DL, DerSimonian and Laird method.

DISCUSSION

Principal findings

In this meta-analysis consisting of 6 cohort studies (2,087,195 older adults from different sites with 149,804 cases of dementia during 4.00–13.00 years of follow-up), it was observed that individuals who had at least one influenza vaccination were related to a 31% lower risk for dementia (pooled random-effect RR = 0.69, 95% CI: 0.57–0.83, p = 0.000). Notably, the pooled RR in our meta-analysis was not affected by age or gender, as all qualified studies adjusted their results for these two covariates. Subgroup analyses showed that in the study with a mean age of 75–80 years or 75%–100% males [11], the association was generally weakened compared to studies with a mean age of 70–75 years or 25%–50% males. The results were stable in the sensitivity analyses. In a nutshell, these results suggested that influenza vaccination may help prevent dementia in older adults.

Comparison with other studies

Recently, increasing numbers of observational studies have reported lower risk of dementia in the elderly who had received influenza vaccination compared to those who had never been vaccinated. Previous meta-analysis involving 4 studies indicated that influenza vaccination was related to 29% decreased risk after adjustment for potential confounders [16]. In our meta-analysis, the evidence was further strengthened by the inclusion of 6 eligible studies with adequate subgroup analyses, which was not reported by Veronese et al. [16]. Besides, the sample size was increased approximately 7 times by including 2,087,195 participants at baseline compared to approximately 300,000 in the previous study [16]. Finally, the results of our meta-analysis were robust due to the lack of publication bias.

Potential underlying mechanisms

The first thing to note is that most studies were based on data from medical claims or electronic health records [11 –15]. These data do not contain psychosocial measures such as social support, cognitive stimulation, isolation, and exercise, all of which may partly explain the relationship between influenza vaccination and incident dementia. So far, research about the mechanisms of influenza vaccination on dementia has been mainly carried out in animals. Yang et al. found that early multiple influenza vaccination could exert an immunomodulatory effect in amyloid precursor protein (APP)/presenilin ₁ (PS₁) mice by removing amyloid-β plaques, activating microglia, disrupting Treg-regulated immune system, ultimately enhancing cognitive deficits [26]. Another study demonstrated that influenza vaccination combined with moderate-dose PD1 blockade could reduce amyloid-β deposition and improve cognition in APP/PS₁ mice [27]. Some scholars have proposed that vaccination can ‘train immunity’ [28] and regulate general resistance, which are non-specific effects [29]. These findings suggest that immune dysfunction is related to dementia, and influenza vaccination may exert neuroprotective effects through immune modulation. However, the specific mechanisms and the pathways through which the influenza vaccine works are still unclear, especially in the human population. Besides, a large number of studies indicated that the infectious etiologies are involved in the development of dementia. Pathogenic infections can result in oxidative stress and neurological inflammation, leading to neurodegenerative diseases [30 –32]. This view is supported by evidence from other epidemiological studies. It has been found that not only the flu shot, but numerous types of vaccinations are related to decreased dementia risk, with hepatitis B (HR = 0.82), typhoid (HR = 0.80), hepatitis A (HR = 0.78), influenza (HR = 0.74), herpes zoster (HR = 0.69), tetanus & diphtheria & pertussis (Tdap) (HR = 0.69), and rabies (HR = 0.43) vaccinations being significant [6]. These findings suggest that the protective effect of vaccination is general and not specific to any one type of vaccine. It is worth mentioning that some studies have found it takes at least 2-3 doses of influenza vaccine to show a statistically significant protective effect against dementia [13, 15], and Wiemken et al. proposed at least 6 doses [11]. These indicate that there may be some form of dose-response relationship between influenza vaccination and dementia, and further research is needed to clarify this relationship and whether there exists a minimum effective dose. Furthermore, influenza vaccination might reduce dementia risk via other potential mechanisms. For instance, there is evidence that influenza vaccination may decrease the risk of acute cerebrovascular and cardiovascular events [33, 34], which are risk factors for dementia. Overall, there is indeed a link between influenza vaccination and incident dementia. Easily accessible and low-cost influenza vaccination can serve as a potential intervention for lowering dementiarisk.

Limitations of this study

Nevertheless, this study has several limitations. First, there were only 6 eligible studies and they all had observational design that made demonstration of causality not strong enough. Second, 3 studies were carried out in the Western Pacific Region and the remaining 3 in the Region of the Americas, which imposed geographic limitations on our analysis. Third, 3 studies in Taiwan all used population with certain chronic diseases [13 –15] and 1 study in America used a veterans population [11], which may differ from the general population in aspects that are important for the analysis. Fourth, although all of these studies have adjusted RRs for age, gender and other related covariates, the possibility of unmeasured and residual confounding factors may not be ruled out. In particular, the RR extracted from Verreault et al. was adjusted only for age, sex, and education [25]. Last, although a random-effect model was employed, the results of this meta-analysis should be interpreted in caution due to the significant heterogeneity detected in the pooled analyses. Based on the sensitivity analyses, we believed that the significant heterogeneity was likely to be derived from the articles published by Bukhbinder et al. and Wiemken et al. [11, 12], whose removal separately reduced the heterogeneity of the remaining articles, and simultaneously made the heterogeneity of the remaining articles disappear. It may be because these two studies differed from the other four in the way they measured exposure and diagnosed dementia. Both Bukhbinder’s and Wiemken’s studies used CPT/HCPCS codes for the measurement of influenza vaccination [11, 12], while the other studies used ICD-9 code or questionnaire. These two studies based on electronic health record data were conducted in the United States [11, 12], which has no national influenza vaccine registration compared with Taiwan. It is likely that many patients have received influenza vaccination outside the health care system but were not registered, for example, at pharmacies or health fairs. Therefore, these two U.S. studies were likely to misclassify exposure. Additionally, the two studies of Bukhbinder and Wiemken both used some statistical methods to balance baseline characteristics of different observation groups, such as PSM and IPTW [11, 12], which all belong to the propensity score methods (Table 1). Although the results of other studies have adjusted for confounding factors when analyzing the data, propensity score methods allowed Bukhbinder’s and Wiemken’s observational studies to be designed similar to randomized trials as much as possible, and to separate the design of the studies from the analyses of the effect of exposure on the outcome [35]. So these two studies from the United States may have more complete control over confounding factors. This differentiated these two studies from other studies, which may be a potential source of heterogeneity in this study.

Due to the lack of qualified studies, we recommend that more epidemiological studies on the association between influenza vaccination and dementia should be conducted in the future, especially in different WHO regions, such as the European region. The studies of dose-response relationship are also called for to determine whether there is a threshold dose for the protective effect of influenza vaccination on dementia. Compared to experimental epidemiological study and case-control study, prospective cohort study is a better form of research because it can examine the causal link between flu vaccination and dementia strongly andethically.

Conclusion

The results of this meta-analysis indicate that influenza vaccination in older adults is associated with decreased risk of dementia. Nevertheless, additional mechanistic studies and more epidemiological studies are required to clarify the relationship between influenza vaccination and dementia risk.

Footnotes

ACKNOWLEDGMENTS

We would like to thank Professor Jue Liu, Professor Min Liu, and Miss Huimin Sun from the School of Public Health, Peking University. Jue Liu conceived and designed the study. Huimin Sun did data acquisition, data curation, data analysis, visualization, and writing original draft. Jue Liu and Min Liu reviewed and edited the manuscript.

FUNDING

This work was partly supported by the National Natural Science Foundation of China (72122001, 72211540398, 71934002), National Key Research and Development Project of China (2021ZD0114101, 2021ZD0114104, 2021ZD0114105), National Statistical Science Research Project (2021LY038) and the Fundamental Research Funds for the Central Universities supported by Global Center for Infectious Disease and Policy Research & Global Health and Infectious Diseases Group of Peking University (202204), National Science and Technology Project on Development Assistance for Technology, Developing China-ASEAN Public Health Research and Development Collaborating Center (KY202101004).

CONFLICT OF INTEREST

The authors have no conflict of interest to report.

DATA AVAILABILITY

The data supporting the findings of this study are available on request from the corresponding author.

References

World Health Organization (2021) Global Status Report on the Public Health Response to Dementia. https://www.who.int/publications/i/item/9789240033245 Last updated September 1, 2021, Accessed on September 20, 2022.

Patterson

(2018) World Alzheimer Report 2018. The state of the art of dementia research: New frontiers. Alzheimer’s Disease International, London.

Allen

, Curran

, Duggan

, Cryan

, Chorcoráin

, Dinan

, Molloy

, Kearney

, Clarke

(2017) A systematic review of the psychobiological burden of informal caregiving for patients with dementia: Focus on cognitive and biological markers of chronic stress. Neurosci Biobehav Rev 73, 123–164.

Prince

, Wimo

, Guerchet

, Ali

G-C

, Wu

Y-T

, Prina

(2015) World Alzheimer Report 2015. The global impact of dementia: An analysis of prevalence, incidence, cost and trends. Alzheimer’s Disease International, London.

Dhillon

(2021) Aducanumab: First approval. Drugs 81, 1437–1443.

, Yang

, He

, Xia

, Chen

, Zhou

, Liu

, Sun

(2022) Adult vaccination as a protective factor for dementia: A meta-analysis and systematic review of population-based observational studies. Front Immunol 13, 872542.

Grohskopf

, Alyanak

, Ferdinands

, Broder

, Blanton

, Talbot

, Fry

(2021) Prevention and control of seasonal influenza with vaccines: Recommendations of the Advisory Committee on Immunization Practices, United States, 2021-22 Influenza Season. MMWR Recomm Rep 70, 1–28.

Bellino

, Piovesan

, Bella

, Rizzo

, Pezzotti

, Ramigni

(2020) Determinants of vaccination uptake, and influenza vaccine effectiveness in preventing deaths and hospital admissions in the elderly population; Treviso, Italy, 2014/2015-2016/2017 seasons. Hum Vaccin Immunother 16, 301–312.

Davtyan

, Ghochikyan

, Movsesyan

, Petrushina

, Cadagan

, Zamarin

, Martinez-Sobrido

, Albrecht

, García-Sastre

, Agadjanyan

(2011) A novel strategy for generation of effective and safe Alzheimer’s disease vaccine based on conventional influenza virus vaccine modified to express Ab1-10. Neurodegener Dis 8, https://doi.org/10.1159/000327701.

10.

Davtyan

, Ghochikyan

, Cadagan

, Zamarin

, Petrushina

, Movsesyan

, Martinez-Sobrido

, Albrecht

, García-Sastre

, Agadjanyan

(2011) The immunological potency and therapeutic potential of a prototype dual vaccine against influenza and Alzheimer’s disease. J Transl Med 9, 127.

11.

Wiemken

, Salas

, Hoft

, Jacobs

, Morley

, Scherrer

(2021) Dementia risk following influenza vaccination in a large veteran cohort. Vaccine 39, 5524–5531.

12.

Bukhbinder

, Ling

, Hasan

, Jiang

, Kim

, Phelps

, Schmandt

, Amran

, Coburn

, Ramesh

, Xiao

, Schulz

(2022) Risk of Alzheimer’s disease following influenza vaccination: A claims-based cohort study using propensity score matching. J Alzheimers Dis 88, 1061–1074.

13.

Luo

, Chi

, Fang

, Liu

, Lee

(2020) Influenza vaccination reduces dementia in patients with chronic obstructive pulmonary disease: A nationwide cohort study. J Investig Med 68, 838–845.

14.

Lee

, Chang

, Lin

, Yeh

, Hu

, Wu

, Chen

, Liao

(2020) Risk of dementia in patients with periodontitis and related protective factors: A nationwide retrospective cohort study. J Clin Periodontol 47, 1428–1436.

15.

Liu

, Hsu

, Kao

, Hao

, Liu

, Lin

, Sung

, Wu

(2016) Influenza vaccination reduces dementia risk in chronic kidney disease patients: A population-based cohort study. Medicine (Baltimore) 95, e2868.

16.

Veronese

, Demurtas

, Smith

, Michel

, Barbagallo

, Bolzetta

, Noale

, Maggi

(2022) Influenza vaccination reduces dementia risk: A systematic review and meta-analysis. Ageing Res Rev 73, 101534.

17.

Page

, McKenzie

, Bossuyt

, Boutron

, Hoffmann

, Mulrow

, Shamseer

, Tetzlaff

, Akl

, Brennan

, Chou

, Glanville

, Grimshaw

, Hróbjartsson

, Lalu

, Li

, Loder

, Mayo-Wilson

, McDonald

, McGuinness

, Stewart

, Thomas

, Tricco

, Welch

, Whiting

, Moher

(2021) The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ 372, n71.

18.

Stroup

, Berlin

, Morton

, Olkin

, Williamson

, Rennie

, Moher

, Becker

, Sipe

, Thacker

(2000) Meta-analysis of observational studies in epidemiology: A proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 283, 2008–2012.

19.

, Shi

, Zheng

, Sun

, Huang

, Zhang

, Que

, Sun

, Shi

, Bao

, Deng

, Lu

(2022) Leisure activities and the risk of dementia: A systematic review and meta-analysis. Neurology 99, e1651–e1663.

20.

Luchini

, Stubbs

, Solmi

, Veronese

(2017) Assessing the quality of studies in meta-analyses: Advantages and limitations of the Newcastle Ottawa Scale. World J Metaanal 5, 80–84.

21.

Satizabal

, Beiser

, Chouraki

, Chene

, Dufouil

, Seshadri

(2016) Incidence of dementia over three decades in the Framingham Heart Study. N Engl J Med 374, 523–532.

22.

Gao

, Ogunniyi

, Hall

, Baiyewu

, Unverzagt

, Lane

, Murrell

, Gureje

, Hake

, Hendrie

(2016) Dementia incidence declined in African-Americans but not in Yoruba. Alzheimers Dement 12, 244–251.

23.

Davies

, Crombie

, Tavakoli

(1998) When can odds ratios mislead? BMJ 316, 989–991.

24.

DerSimonian

, Laird

(1986) Meta-analysis in clinical trials. Control Clin Trials 7, 177–188.

25.

Verreault

, Laurin

, Lindsay

, De Serres

(2001) Past exposure to vaccines and subsequent risk of Alzheimer’s disease. CMAJ 165, 1495–1498.

26.

Yang

, He

, Xing

, Zuo

, Yuan

, Wu

, Jiang

, Qi

, Yao

(2020) Influenza vaccination in early Alzheimer’s disease rescues amyloidosis and ameliorates cognitive deficits in APP/PS1 mice by inhibiting regulatory T cells. J Neuroinflammation 17, 65.

27.

Xing

, Zuo

, Hu

, Zheng

, Wang

, Yuan

, Zhou

, Qi

, Yao

(2021) Influenza vaccine combined with moderate-dose PD1 blockade reduces amyloid-β accumulation and improves cognition in APP/PS1 mice. Brain Behav Immun 91, 128–141.

28.

Netea Mihai

, Quintin

, van der Meer Jos

(2011) Trained immunity: A memory for innate host defense. Cell Host Microbe 9, 355–361.

29.

Benn

, Netea

, Selin

, Aaby

(2013) A small jab - a big effect: Nonspecific immunomodulation by vaccines. Trends Immunol 34, 431–439.

30.

Sipilä

, Heikkilä

, Lindbohm

, Hakulinen

, Vahtera

, Elovainio

, Suominen

, Väänänen

, Koskinen

, Nyberg

, Pentti

, Strandberg

, Kivimäki

(2021) Hospital-treated infectious diseases and the risk of dementia: A large, multicohort, observational study with a replication cohort. Lancet Infect Dis 21, 1557–1567.

31.

Sochocka

, Zwolińska

, Leszek

(2017) The infectious etiology of Alzheimer’s disease. Curr Neuropharmacol 15, 996–1009.

32.

Perry

, Cunningham

, Holmes

(2007) Systemic infections and inflammation affect chronic neurodegeneration. Nat Rev Immunol 7, 161–167.

33.

Tsivgoulis

, Katsanos

, Zand

, Ishfaq

, Malik

, Karapanayiotides

, Voumvourakis

, Tsiodras

, Parissis

(2018) The association of adult vaccination with the risk of cerebrovascular ischemia: A systematic review and meta-analysis. J Neurol Sci 386, 12–18.

34.

Ju-Chi Liu

, Sung

, Hao

(2016) Influenza vaccination reduces relative risk of dementia in patients with heart failure. Eur J Heart Failure 18, 344.

35.

Austin

, Stuart

(2015) Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies. Stat Med 34, 3661–3679.