Influenza Vaccination and Short‐Term Risk of Stroke Among Elderly Patients With Chronic Comorbidities in a Population‐Based Cohort Study

ABSTRACT

The protective effect of influenza vaccination on stroke risk has been inconclusive. In this study, we aimed to investigate the impact of influenza vaccination on the 1‐year risk of stroke in individuals aged 60 years and older with COPD and hypertension or diabetes mellitus. We conducted a retrospective cohort study in four districts of Shanghai, China, from August 2017 to July 2019. Data were collected from various information systems related to chronic disease management, cardiovascular reporting, and immunizations. The incidence of stroke within 1 year was compared between vaccinated and unvaccinated chronic disease patients. Cox proportional hazards regression was used to calculate hazard ratios (HRs). Sensitivity analysis was performed using the Poisson regression model to examine the association between influenza vaccination and stroke incidence, and propensity score matching was employed to address confounding. We found that influenza vaccination was associated with a lower risk of stroke during the two influenza seasons, 2017–2018 (adjusted HR, 0.27; 95% CI, 0.10–0.73) and 2018–2019 (adjusted HR, 0.46; 95% CI, 0.21–1.02). The results from the Poisson regression model (RR, 0.26; 95% CI, 0.10–0.70) were consistent with those obtained from the Cox model analysis. The reduction in stroke risk associated with influenza vaccination ranged from 54% to 73%. Our findings suggest that influenza vaccination is associated with a lower 1‐year risk of stroke in individuals with chronic illnesses, compared to those who are not vaccinated.

Abbreviations

BMI

body mass index

CI

confidence interval

COPD

chronic obstructive pulmonary disease

HR

hazard ratio

ICD

International Classification of Disease

RR

relative risk

SCDC

the Shanghai Municipal Center for Disease Control and Prevention

1. Background

Stroke is a health risk with high prevalence [1]. On a global scale, approximately 9.6 million ischemic strokes and 4.1 million hemorrhagic strokes occur annually, including both intracerebral and subarachnoid hemorrhages. Alarmingly, stroke accounts for an estimated 5.9 million deaths and the loss of 102 million disability‐adjusted life years (DALYs), ranking it as the second leading cause of mortality after ischemic heart disease and the third leading cause of DALYs lost worldwide [2].

In China, the burden of stroke is particularly significant and is on the rise. Nationwide studies and governmental reports revealed that in 2013, there were approximately 11 million prevalent stroke cases, 2.4 million new stroke cases, and 1.1 million stroke‐related deaths [3, 4]. These sobering statistics underscore the urgent need for effective prevention and management strategies.

A range of risk factors for stroke have been identified and validated, including hypertension [5], diabetes, and a variety of lifestyle and environmental factors [6]. In recent years, however, rare but potentially significant risk factors have also emerged, with influenza infection being one such example. Research has demonstrated a link between influenza and an increased risk of stroke [7, 8, 9, 10]. Specifically, studies have reported that influenza virus infection can enlarge the infarct size [7] and is associated with a 39% increase in the odds of ischemic stroke [11]. Trivalent inactivated influenza vaccine (IIV3), the type of influenza vaccine approved by the National Medical Products Administration and widely used in Shanghai during 2017–2019, have demonstrated efficacy variable direct and indirect protection against influenza infection, therefore recommended by the WHO and China CDC for its potential significance in both direct and indirect health benefits [12, 13]. Given that influenza vaccination can effectively reduce the incidence risk of influenza, it is crucial to investigate whether such vaccination can also mitigate the risk of stroke. However, existing studies exploring the association between influenza vaccination and stroke incidence have yielded inconsistent findings. Certain research has demonstrated the protective effect of influenza vaccine and/or influenza infection prevention approaches on stroke among different populations and in multiple settings [14, 15, 16]. However, there are also insignificant or contradictive findings, showing no statistically significant connection between influenza vaccine and stroke risks [17].

Moreover, given that diabetes and hypertension are established and significant risk factors for stroke [18], with hypertension being particularly prominent [19], there is a pressing need for evidence on the protective effects of influenza vaccination in these high‐risk patient populations. Despite the World Health Organization's recommendation to enhance vaccination coverage in individuals with certain chronic illness such as chronic obstructive pulmonary disease (COPD), and aged 65 years and older [20, 21], there is still a significant gap in our understanding of the preventive benefits of influenza vaccination against stroke in patients with diabetes or hypertension.

Therefore, the primary objective of our study is to investigate the protective effect of the seasonal influenza vaccine in reducing stroke among elderly patients with diabetes or hypertension. This study aims to provide valuable insights into the potential merit of influenza vaccination in reducing stroke risk among these vulnerable populations.

2. Methods

2.1. Study Design and Study Population

We conducted a population‐based retrospective cohort study in four districts (Huangpu, Yangpu, Baoshan, and Qingpu) of Shanghai, China, where recommendations have been given on influenza vaccination for chronic disease management populations since 2017. We initially recruited a primary cohort consisting of patients registered in the Shanghai Chronic Disease Management Information System from August 2017 to July 2019, who met the following inclusion criteria: (1) aged 60 years or older, (2) diagnosed with COPD, (3) diagnosed with hypertension or diabetes, (4) no history of stroke before influenza vaccinated, and (5) had a minimum of 1 year of recorded health status to ensure data quality. The exposure variable was influenza vaccination. Individuals were classified as vaccinated if they received the IIV3 between August 1 and March 31 for the concurrent influenza season. Those who did not receive the vaccine were considered unvaccinated. During each cohort year, participants were followed from August 1st until the occurrence of study outcomes, including stroke, death, or the end of the cohort year (July 31). Stroke was defined as I60‐I64 according to the International Classification of Diseases (ICD). A total of 50 842 patients were followed up.

2.2. Data Sources

Data for the study were obtained from three distinct sources: (1) the Shanghai Chronic Disease Management Information System, (2) the Shanghai Immunization Program Information System, and (3) the Shanghai Cardiovascular and Cerebrovascular Event Reporting Information System, which are all housed at the Shanghai CDC. The individual's personal ID was used to link the three information systems.

Throughout Shanghai, patients ≥15 years old diagnosed with hypertension and diabetes are asked if they want to be included in a centralized database–the Chronic Disease Management Information System. Inclusion in the database means that the patients will receive more standardized management of their disease. General practitioners follow up with patients every 3 months at community health centers and input data related to these visits into the Chronic Disease Management Information System. This database contains information on gender, birthdate, occupation, weight, height, smoking status, alcohol consumption, and diagnostic information pertaining to hypertension, diabetes, and COPD. No other individual‐level information was available from the dataset. All patients in the four districts from the Shanghai Chronic Disease Management Information System who met the inclusion criteria were included in this study.

Data in the Immunization Program Information System were captured and entered by vaccination providers at community health care centers. Data were uploaded daily from these health centers’ electronic registries into the Immunization Program Information System. Influenza vaccination information from the 2017–2018 and the 2018–2019 influenza seasons and pneumococcal vaccination information from January 2013 were obtained from the Shanghai Immunization Program Information System. Types and dates of vaccination were extracted from the Immunization Program Information System. The Shanghai CDC and the Shanghai Health Commission implement regular data quality assurances of the Immunization Program Information System.

Data in the Shanghai Cardiovascular and Cerebrovascular Event Reporting Information System were captured automatically by the hospital record system at secondary and tertiary hospitals, then uploaded after being verified and confirmed by clinical doctors. Types and attack time of stroke were extracted from the Shanghai Cardiovascular and Cerebrovascular Event Reporting Information System. The Shanghai CDC and the Shanghai Health Commission implement regular data quality assurances of the Shanghai Cardiovascular and Cerebrovascular Event Reporting Information System.

2.3. Covariates

A series of covariates were incorporated into the study, enlightened by previous research indicating the association between influenza vaccines and stroke [22, 23, 24, 25]. Sociodemographic variables included age (60–69, 70–79, 80–89, and 90–104), gender (male vs. female), and body mass index (BMI, kg/m²). The BMI was further categorized into four groups: underweight (<18.5 kg/m²), normal weight (18.5–23.9 kg/m²), overweight (24–27.9 kg/m²), and obese (≥28 kg/m²). Regarding lifestyle habits, smoking status was dichotomized into current smokers and non‐smokers, while alcohol consumption was similarly classified as current drinkers or abstainers. Pneumonia vaccination status (vaccinated vs. unvaccinated) was also considered.

2.4. Statistical Analysis

Baseline characteristics for each influenza season were stratified based on whether individuals had a history of influenza vaccination. These characteristics were then compared using appropriate statistical tests, including the χ2 test, Fisher's exact test, or the Student t‐test. The Cox proportional hazards regression model served as the primary analytical tool, estimating both unadjusted and adjusted hazard ratios (aHRs) along with their corresponding 95% confidence intervals (CIs). These ratios quantified the association between influenza vaccination and the incidence of stroke during each influenza season. Drawing from previous research [22, 23], the models were adjusted to account for potential confounding variables such as gender, age, hypertension, diabetes, smoking status (current smoker vs. non‐smoker), alcohol consumption, and BMI.

To further delve into the data, the analyses were divided into subgroups by gender, BMI (normal <24 kg/m² vs. abnormal ≥ 24 kg/m²), pneumonia vaccination status (vaccinated vs. unvaccinated), smoking status (current smoker vs. non‐smoker), and alcohol consumption (drinker vs. abstainer). To assess the robustness of our findings, three sensitivity analyses were conducted: (1) to address missing values in some covariates, multiple imputations using chained equations were employed to fill in the gaps. These imputed datasets were then used to rerun the primary analyses, enabling us to test the consistency of our initial results. (2) Propensity score matching was employed to select control subjects who were comparable to cases, minimizing potential biases in treatment selection. The propensity score was derived through multivariable logistic regression, incorporating variables such as age, gender, smoking status, alcohol consumption, and BMI. A greedy nearest neighbor approach with one‐to‐one matching was applied to the logit of the propensity score, using a caliper width of 0.2 in SAS. The covariate balance was evaluated using standardized differences, with differences below 0.10 considered negligible. (3) Given the extremely low incidence rate of stroke, the Poisson regression model was chosen to account for the correlation and calculate relative risks (RRs). This approach allowed us to assess the effectiveness of influenza vaccination in preventing stroke more accurately.

All analyses have been performed using SAS statistical software 9.4 (SAS/STAT user's guide 2012; SAS Institute, Cary, NC). A two‐sided p value <0.05 was considered to be statistically significant.

3. Results

3.1. Patient Characteristics

Table 1 presents the baseline characteristics of the participants in each influenza season, stratified by their influenza vaccination status. In the 2017–2018 influenza season, there were 50 842 participants, while in the 2018–2019 influenza season, there were 47 045 participants. A substantial proportion of the total population had diabetes (38.8%) or hypertension (86.1%). The rates of influenza vaccination varied between seasons, with 1.55% of participants vaccinated in the 2017–2018 influenza season and 1.25% in the 2018–2019 influenza season. Vaccinated participants tended to be younger (<80 years), male, smokers, drinkers, and have hypertension, while unvaccinated participants were more likely to have been diagnosed with diabetes.

TABLE 1.

Characteristics of participants stratified by influenza vaccination status.

Characteristics	2017–2018		2018–2019
	Vaccinated	Unvaccinated	Vaccinated	Unvaccinated
All participants	786 (1.55)	50 056 (98.45)	589 (1.25)	46 456 (98.75)
Age (years)
60–69	223 (1.49) *	14 739 (98.51)	204 (1.41) *	14 217 (98.59)
70–79	367 (2.15) *	16 664 (97.85)	277 (1.72) *	15 868 (98.28)
80–89	184 (1.11) *	16 408 (98.89)	107 (0.73) *	14 603 (99.27)
90–104	12 (0.53) *	2245 (99.47)	1 (0.06) *	1768 (99.94)
Sex
Male	402 (2.03) *	19 424 (97.97)	267 (1.47) *	17 912 (98.53)
Female	384 (1.24) *	30 632 (98.76)	322 (1.12) *	28 544 (98.88)
Smoking status
No	659 (1.69) *	38 280 (98.31)	501 (1.30)	38 027 (98.70)
Yes	53 (2.91) *	1768 (97.09)	23 (1.27)	1790 (98.73)
Alcohol consumption
Nondrinker	664 (1.68) *	38 849 (98.32)	510 (1.30)	38 594 (98.70)
drinker	48 (3.85) *	1198 (96.15)	14 (1.13)	1222 (98.87)
Body mass index (BMI), kg/m2
<18.5	7 (0.98)	708 (99.02)	3 (0.43)	700 (99.57)
18.5–24	364 (1.91)	18 684 (98.09)	248 (1.32)	18 604 (98.68)
24–28	269 (1.64)	16 086 (98.36)	220 (1.36)	15 991 (98.64)
≥28	70 (1.53)	4491 (98.47)	50 (1.11)	4444 (98.89)
Hypertension
No	34 (1.00) *	3376 (99.00)	35 (1.07)	3229 (98.93)
Yes	752 (1.59) *	46 680 (98.41)	554 (1.27)	43 227 (98.73)
Diabetes mellitus
No	516 (1.66) *	30 594 (98.34)	381 (1.33)	28 269 (98.67)
Yes	270 (1.37) *	19 462 (98.63)	208 (1.13)	18 187 (98.87)

^* p < 0.05.

3.2. Influenza Vaccination and 1‐Year Incidence Risk of Stroke

During the 2017–2018 influenza season, there were five cases of stroke among vaccinated participants and 1308 cases among unvaccinated participants. In the 2018–2019 influenza season, there were seven cases of stroke among vaccinated participants and 1344 cases among unvaccinated participants. The incidence rates of stroke were higher among unvaccinated participants (2.61%) compared to vaccinated participants (0.64%) in the 2017–2018 influenza season. Similarly, during the 2018–2019 influenza season, the incidence rates were 2.89% among unvaccinated participants and 1.19% among vaccinated participants (Table 2). Figures S1 and S2 demonstrate the 1‐year cumulative probability of survival for vaccinated and unvaccinated populations in the 2017–2018 and 2018–2019 influenza seasons, respectively.

TABLE 2.

Association between influenza vaccination and risk of stroke during the influenza season.

Cohort	Primary outcome in influenza‐vaccinated subjects, n (%)	Primary outcome in non–influenza‐vaccinated subjects n (%)	Unadjusted HR (95% CI)	p	Adjusted* HR (95% CI)	p
2017–2018	5/786 (0.64)	1308/50 056 (2.61)	0.241 (0.293–0.961)	0.0217	0.272 (0.102–0.726)	0.0093
2018–2019	7/589 (1.19)	1344/46 456 (2.89)	0.351 (0.157–0.781)	0.0103	0.458 (0.205–1.022)	0.0565

Note: Model‐1: crude model; Model‐2: adjusted for age, gender, smoking, alcohol consumption, hypertension and diabetes mellitus.

Abbreviations: CI, confidence interval; HR, hazard ratio.

In the unadjusted analysis, influenza vaccination was associated with a reduced risk of stroke during both influenza seasons (Table 2). Compared to unvaccinated participants, influenza vaccination was associated with a 76% reduction in the incidence rate of stroke during the 2017–2018 influenza season (HR = 0.24; 95% CI, 0.29–0.96; p = 0.0217). Similarly, during the 2018–2019 influenza season, vaccination was associated with a significantly reduced incidence rate of stroke (HR = 0.35; 95% CI, 0.16–0.78; p = 0.010). After adjusting for age, gender, smoking, alcohol consumption, hypertension, and diabetes, the model showed a reduced incidence risk of stroke in the vaccinated group compared to the unvaccinated group during the 2017–2018 influenza season (aHR = 0.27; 95% CI, 0.10–0.73). In the analysis of the 2018–2019 influenza season, the point estimates were similar (aHR = 0.46). However, the association between vaccination and reduced stroke risk did not reach statistical significance (aHR = 0.46; 95% CI, 0.21–1.02, p = 0.057).

3.3. Subgroup Analyses

In the stratified analysis, no statistically significant interaction terms were observed with the various stratification variables, including gender, BMI, pneumonia vaccine status, smoking, and alcohol consumption (all p > 0.05). However, some differences were apparent in the stratified models. During the 2017–2018 influenza season, individuals who did not smoke (aHR = 0.29, 95% CI, 0.11–0.77), did not consume alcohol (aHR = 0.28, 95% CI, 0.11–0.75), or had not been vaccinated against pneumonia (aHR = 0.25, 95% CI, 0.08–0.77) exhibited a similar pattern of association with influenza vaccination and stroke risk as the entire population. Similar trends were observed during the 2018–2019 influenza season (Figures 1 and 2).

FIGURE 1.

Effect of vaccination on the hazard ratio and 95% confidence intervals for the stroke according to subgroups of patients in the 2017–2018 influenza season.

FIGURE 2.

Effect of vaccination on the hazard ratio and 95% confidence intervals for the stroke according to subgroups of patients in the 2018–2019 influenza season.

3.4. Sensitivity Analysis

To further assess the robustness of our findings, we conducted a series of sensitivity analyses.

Firstly, given the low incidence of stroke in the population, we employed the Poisson regression model to estimate the association between influenza vaccination and stroke risk. In the unadjusted Poisson regression model for the 2017–2018 influenza season, vaccination against influenza was associated with a reduction in stroke incidence (HR = 0.24; 95% CI, 0.10–0.59). After adjusting for confounders such as gender, age, and other relevant variables, this association remained statistically significant (aHR = 0.26; 95% CI, 0.10–0.70) (Table S1). Similarly, for the 2018–2019 influenza season, the Poisson regression model also indicated a reduced stroke incidence risk associated with influenza vaccination (HR = 0.41; 95% CI, 0.20–0.86).

Secondly, we addressed missing values in some covariates using multiple imputations with chained equations. This sensitivity analysis did not alter our findings, as influenza vaccination remained associated with stroke incidence in both influenza seasons (p < 0.05 in adjusted models) (Table S2).

Thirdly, we used propensity score matching to select controls similar to cases and minimize treatment selection bias when estimating the effects of vaccination. In the propensity score‐adjusted model, we observed significant differences in stroke incidence risk associated with vaccination. Specifically, the matched HR for vaccination was 0.26 (95% CI, 0.10–0.70, p = 0.008) during the 2017–2018 influenza season and 0.31 (95% CI, 0.13–0.79, p = 0.013) during the 2018–2019 influenza season (Table S3).

Collectively, these sensitivity analyses support the robustness of our findings and reinforce the association between influenza vaccination and reduced stroke incidence in the population studied.

4. Discussion

COPD frequently coexists with other chronic conditions, particularly hypertension and diabetes, creating a complex clinical scenario that complicates management strategies. This co‐occurrence can lead to a vicious cycle where each condition exacerbates the other; for instance, diabetes can worsen lung function and increase the risk of respiratory infections, while COPD can complicate glycemic control due to the use of systemic corticosteroids during exacerbations [26]. The management of these comorbidities requires a multidisciplinary approach that not only addresses the individual diseases but also considers their interactions. This study and the program it is rooted in intended to adopt a holistic view in managing patients with COPD and its common comorbidities, focusing on integrated care strategies that improve quality of life and clinical outcomes.

In this extensive observational study involving a substantial cohort of over 50 000 participants, we found compelling evidence suggesting that influenza vaccination in elderly individuals with chronic illness is associated with significant reductions of the risk of stroke during influenza seasons. This association remained robust even after adjusting for various confounding variables known to impact stroke incidence. By mitigating the risk of influenza infection, vaccination may indirectly contribute to a lower likelihood of stroke occurrence [15].

Several previous studies have reported consistent findings supporting the association between influenza vaccination and a reduced risk of stroke. For instance, a matched case‐control study of 740 participants with an average age of 60.8 years found that influenza vaccination was associated with reduced odds of stroke/TIA (OR = 0.46; 95% CI, 0.28–0.77) [27]. Another retrospective cohort study involving 44 604 vaccinated and 22 463 unvaccinated individuals with type 2 diabetes in England during the 2003–2004 influenza season showed that vaccination significantly lowered admission rates for stroke (incidence rate ratios = 0.70; 95% CI, 0.53–0.91) [28].

Our findings align with the direction of the effects observed in previous research, but a crucial aspect of our study lies in the fact that all participants were patients with COPD and hypertension or with COPD and diabetes. This underscores the importance of promoting the influenza vaccine among individuals with chronic diseases. Additionally, our study employed a large‐scale cohort approach, greatly enhancing the reliability of our results. Furthermore, by including data from two distinct influenza seasons, we were able to broaden the generalizability of our findings to other seasons and settings.

However, it is worth noting that not all studies have consistently reported a reduction in outcome events associated with influenza vaccination. Differences in study design and covariates included in various models may have contributed to varying results reported in the literature [29]. Some studies have only captured the occurrence of outcomes among study participants without specifying the exact timing of these events [30]. In contrast, our study employed a more precise outcome period, leveraging influenza surveillance data to enhance our ability to detect the effectiveness of vaccination.

Several strengths are worth mentioning in our study. Firstly, the data were highly reliable on the vaccination status of participants from the Shanghai Immunization Program Information System, minimizing recall bias and referral bias that can occur with self‐reported questionnaires used in previous studies [16, 20, 31]. Secondly, we collected accurate research outcomes for all participants through the Shanghai Cardiovascular and Cerebrovascular Event Reporting Information System, which has a low underreporting rate of less than 2%. Thirdly, our study revealed that influenza vaccination provided protection against stroke even in subpopulations that were not vaccinated for pneumonia, contrary to previous reports [32] suggesting that the benefits of influenza vaccination may be due to the simultaneous administration of pneumonia vaccination.

However, there are limitations to consider in our study. Firstly, we lacked data on the following fronts: health awareness of study population, which could be problematic in a way that the patients who were willing to receive the vaccine were those who paid more attention to their health; data on other disease status such as influenza, pneumonia, or COPD status prior to incidence of stroke were also unavailable, as such the temporal association between stroke and other stroke‐inducing diseases could not be clearly established; another one was detailed smoking history data. Despite all subjects having COPD, a condition strongly associated with smoking, we were unable to collect information on prior smoking habits, including duration and amount smoked. This omission is particularly significant given that smoking is a major risk factor for stroke, even after cessation. Future studies should prioritize collecting such data to provide a more nuanced understanding of how smoking cessation impacts stroke risk in COPD populations. Secondly, our study design was observational, which implies that we cannot establish a causal relationship between influenza vaccination and reduced stroke risk. We performed sensitivity analyses to address this issue, and the results theoretically validated our findings. However, unseen factors could be inevitable, and the effect cannot be fully countered. Another concern is that the stroke rates were relatively low among the exposure groups, leading to potentially imprecise CIs. Combining with potential confounders not fully adjusted for, residual confounding influencing our findings is a possibility that cannot be entirely ruled out. Additionally, our study only included elderly individuals with chronic diseases to answer the desired research question, so the generalizability of our findings to the entire elderly population may be limited.

In conclusion, our study adds to the body of evidence supporting the positive association between influenza vaccination and a reduced risk of stroke among elderly individuals. The persistent findings, even after accounting for confounding variables, emphasize the importance of influenza vaccination as a preventive measure for stroke, particularly in high‐risk populations. Our findings shed a light on vaccination as a significant aspect of chronic disease prevention and management. Further research is necessary to delve into the underlying mechanism and explore the potential synergistic effects of influenza vaccination in relation to stroke prevention.

5. Conclusion

Compared to participants with chronic illnesses who were not vaccinated, those who received the influenza vaccine demonstrated a lower risk of stroke within one year. These findings emphasize the importance of recommending influenza vaccination for individuals aged 60 years and older who have hypertension or diabetes.

Disclosure

The sponsor had no role in the design, methods, data collection, analysis, or preparation of the article.

Ethics Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Shanghai Municipal Center for Disease Control and Prevention (SCDC), Shanghai, China.

Consent

Informed consent was obtained from all subjects involved in the study.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Data Availability Statement

The datasets analyzed for the current study are not publicly available because they contain detailed medical histories of chronic patients but are available from the corresponding author upon reasonable request.