Effectiveness of the influenza vaccine at preventing hospitalization due to acute exacerbation of cardiopulmonary disease in Korea from 2011 to 2012

Abstract

There is a lack of targeted studies to validate the effectiveness of influenza vaccination on the reduction in influenza-related hospitalizations among patients with co-morbidities. In this study, we estimate the effectiveness of influenza vaccination on preventing hospitalizations in persons with cardiopulmonary disease and establish an evidence base for recommendations on influenza vaccination in this population. During the influenza epidemic in 2011–2012, we performed a multicenter, retrospective case-control study. Cases were patients hospitalized due to acute exacerbation of asthma, COPD, ischemic heart disease (IHD), and congestive heart failure (CHF). Controls were selected from outpatients who visited study hospitals but who were not hospitalized. Cases and controls were matched 1:1 based on age, gender, and date of hospital visit. Conditional logistic regression analyses were used to determine the effectiveness of vaccination. Between 25 December 2011 and 5 May 2012, 828 of each hospitalized and control subjects were identified. The influenza vaccination rate of the hospitalized and non-hospitalized patients was 54.2% and 60.4%, respectively (P = 0.006). The overall vaccine effectiveness for preventing hospitalization was 33.7% (95% confidence interval [CI] 14.0–49.0%; P = 0.002). Conditional logistic regression analysis showed that influenza vaccination significantly reduced the risk of hospitalization, especially due to acute exacerbation of IHD and CHF, in patients aged 65 y and older. The estimated vaccine effectiveness in these patients was 56.0% (95% CI 32.1–71.4%, P = 0.002). Influenza vaccination was associated with a reduction in the risk of hospitalization due to acute exacerbation of cardiopulmonary disease. We recommend the vaccine be given primarily to patients with underlying cardiovascular disease, particularly those 65 y of age and older.

Introduction

Influenza causes a considerable public health burden. During periods of influenza circulation, increases in hospitalization rates due to serious clinical manifestations of influenza infection and complications in patients with cardiopulmonary disease have been observed.^1,2 It is suggested that inflammatory process and immunologic response by influenza infection could be responsible for triggering an exacerbation of underlying cardiopulmonary disease.^3,4 For these reasons, there is consensus that influenza vaccination should be recommended to these patients.

A recent systematic review suggested that the evidence for vaccination in patients with cardiopulmonary disease is still insufficient, and benefits of vaccination in these patients remains undermined.⁵ We previously estimated the effectiveness of influenza vaccination at preventing hospitalization for acute lower respiratory infection and exacerbation of cardiopulmonary disease during the 2010–2011 influenza season.⁶ In that study, we found that influenza vaccination reduced the risk of hospitalization due to acute exacerbation primarily in patients with IHD and CHF, especially in elderly patients aged 65 y and older.

The effectiveness of the influenza vaccine might vary depending on the characteristics of the circulating strain and the degree of match with vaccine strain. Because the study was conducted during just one influenza season (2010–2011), which may have a different epidemic pattern than the current season, we planned to evaluate the vaccine effectiveness targeting patients with cardiopulmonary disease, excluding acute lower respiratory infection. As in our previous study, we conducted a multicenter, matched case-control study to estimate the effectiveness of influenza vaccination at preventing hospitalization due to acute exacerbation of cardiopulmonary disease.

Results

Overall analysis

Overall, 828 hospitalized subjects and 828 matched controls were identified. Subject characteristics are shown in Table 1. BMI and smoking status were similar between the two groups. The proportion of subjects with asthma and COPD were not different, but IHD occurred more frequently in the non-hospitalized group, and CHF occurred more frequently in the hospitalized group. There were no significant differences in comorbidities except diabetes, chronic renal disease, and solid tumors, which were more common in hospitalized subjects. The influenza vaccination coverage was 54.2% among cases and 60.4% among controls (P = 0.006). The overall effectiveness of influenza vaccination in reducing the risk of hospital admission was 33.7% (95% CI 14.0–49.0%, P = 0.002) after adjusting for confounding variables (Table 2).

Table 1. Baseline characteristics of the hospitalized cases and non-hospitalized controls.

Characteristic	Hospitalized Subjects (n = 828)	Non-hospitalized Subjects (n = 828)	P value
Male, n (%)	529 (63.9)	529 (63.9)	1.000
Age, years - median	67 ± 13	68 ± 13	0.797
BMI, kg/m2 - median	24.0 ± 3.9	24.2 ± 3.5	0.233
Smoker, n (%)	194 (23.4)	209 (25.2)	0.375
Reason for hospital visit
Asthma, n (%)	80 (9.7)	84 (10.1)	0.694
COPD, n (%)	142 (17.1)	137 (16.5)	0.576
Ischemic heart disease, n (%)	484 (58.5)	524 (63.3)	<0.001
Congestive heart failure, n (%)	121 (14.6)	77 (4.6)	<0.001
Comorbiditya	555 (67.0)	561 (67.8)	0.794
Diabetes, n (%)	256 (30.9)	174 (21.0)	<0.001
Hypertension, n (%)	464 (56.0)	466 (56.3)	0.958
Cerebrovascular disease, n (%)	56 (6.8)	45 (5.4)	0.294
Neuromuscular disease, n (%)	1 (0.1)	2 (0.2)	1.000
Chronic renal disease, n (%)	34 (4.1)	16 (1.9)	0.015
Chronic hepatic disease, n (%)	25 (3.0)	23 (2.8)	0.885
Solid tumor, n (%)	49 (5.9)	31 (3.7)	0.044
Hematologic disease, n (%)	0 (0)	0 (0)
Autoimmune disease, n (%)	6 (0.7)	3 (0.4)	0.508
Immunosuppressant user, n (%)	6 (0.7)	3 (0.4)	0.453
Influenza vaccination, n (%)	448 (54.2)	500 (60.4)	0.006

^a Some patients had more than one condition.

Table 2. Subgroup analysis of the effectiveness of the 2011–2012 seasonal influenza vaccine in preventing hospitalization stratified by age and reason for hospitalization.

Cause of hospitalization	Hospitalization, n (%)		Non-hospitalization, n (%)		Crude odds ratio			Adjusted odds ratioa
	Vaccinated	Unvaccinated	Vaccinated	Unvaccinated	OR (95% CI)	P		OR (95% CI)	P
Number of subjects (percent)
Overall hospitalization
<65 y	125 (34.8)	234 (65.2)	127 (35.5)	231 (64.5)	0.945 (0.680–1.314)	0.737	1.091 (0.714–1.667)		0.686
≥65 y	323 (69.0)	145 (31.0)	373 (79.4)	97 (20.6)	0.596 (0.442–0.805)	0.001	0.461 (0.322–0.660)		<0.001
All ages	448 (54.2)	379 (45.8)	500 (60.4)	328 (39.6)	0.733 (0.588–0.913)	0.006	0.663 (0.510–0.860)		0.002
Asthma or COPD
<65 y	36 (60.0)	24 (40.0)	29 (38.3)	31 (51.7)	1.636 (0.773–3.465)	0.198	3.168 (0.932–9.844)		0.053
≥65 y	112 (75.7)	36 (24.3)	121 (81.8)	27 (18.2)	0.679 (0.379–1.215)	0.192	0.509 (0.260–1.032)		0.051
All ages	148 (71.2)	60 (28.8)	150 (72.1)	58 (27.9)	0.949 (0.605–1.488)	0.819	0.881 (0.532–1.460)		0.623
Ischemic heart disease or heart failure
<65 y	82 (28.9)	202 (71.1)	91 (32.0)	193 (68.0)	0.847 (0.581–1.235)	0.389	0.916 (0.566–1.483)		0.722
≥65 y	197 (65.4)	104 (34.6)	233 (77.4)	68 (22.6)	0.576 (0.406–0.819)	0.002	0.440 (0.286–0.679)		<0.001
All ages	279 (47.7)	306 (52.3)	324 (55.4)	261 (44.6)	0.688 (0.532–0.910)	0.004	0.619 (0.454–0.844)		0.002

^a Conditional OR adjusted for: BMI, smoking status, comorbidities (diabetes, hypertension, cerebrovascular disease, neuromuscular disease, chronic renal disease, chronic hepatic disease, solid tumor, hematologic disease, autoimmune disease and immunosuppressant user) and influenza vaccination status.

Subgroup analysis according to age group and cause of hospitalization

We also estimated the vaccine effectiveness for each age group and cause of hospital visit (Table 2).

In subjects aged ≥65 y, the vaccine effectiveness was 53.9% (95% CI 44.0–67.8%, P < 0.001). However, the effectiveness was not seen on subjects younger than 65 y (P = 0.686). The vaccine effectiveness among patients with IHD and CHF was 38.1% (95% CI 15.6–54.6%, P = 0.002), and this preventive effect was more pronounced in patients aged ≥65 y (56.0%, 95% CI 32.1–71.4%, P < 0.001). The effect of influenza vaccine in the prevention of hospitalization was not statistically significant among subjects with asthma and COPD regardless of age group (P = 0.623).

Discussion

This matched case-control study demonstrated that influenza vaccination reduced the risk of hospitalization for acute exacerbation of cardiopulmonary disease, especially IHD and CHF, in the elderly aged 65 y and older. The findings are consistent with previous studies that suggest influenza vaccination may be protective against acute exacerbation of IHD and CHF.^7-10 A reduction of 10–20% in hospitalization risk due to IHD was observed in an observational cohort study. Furthermore, 19–27% decrease in hospitalized due to CHF was also noted.⁹ In another study, it was found that a reduction of 27% in hospitalization due to CHF after influenza vaccination.¹¹ In our study, we recorded a 56% reduction in hospitalization cases due to IHD and CHF, which was relatively higher than the reductions found in previous studies. However, these results are similar to our previous study.⁶ The difference may be due to several influences, such as the pathogenicity of circulating strains, case definitions, the methods used, the medical conditions of cases, the study period, and the type of vaccine administered.

Currently, there are a small number of clinical trials to support the implementation of universal vaccination to patients with heart disease. However, there have been studies that support the recommendation. Influenza might promote an acute inflammatory and procoagulant stimulus, altering endothelial function. Recent studies in mice have shown links between influenza infection and coronary artery remodeling.^12,13 In another study, inoculation of influenza virus into mice resulted in heavy infiltration of atherosclerotic plaques by inflammatory cells.¹⁴ Furthermore, influenza virus RNA has been found in human atherosclerotic plaques.¹⁵

Although annual influenza vaccination is recommended for all age groups with underlying cardiovascular disease, we found that vaccine effectiveness was greatest in subjects aged 65 y and older. A previous study that evaluated influenza related hospitalization showed no increase in the number of hospitalization cases among the 18–64 age group with or without the underlying cardiovascular disease.¹⁶ Another study also found that influenza vaccination was not effective in preventing cardiovascular-related hospitalization in patients younger than 65 y of age.¹⁷ There is still a need for further studies to provide clarification on the benefits.

Although influenza vaccination is usually recommended for patients with asthma and COPD, there is little data to support this recommendation. Our study showed poor effectiveness of the influenza vaccine in preventing hospitalization in this patient group. In a recent Cochrane database of systematic review, vaccination failed to show a significant preventive effect on hospitalization.^18,19 Studies that analyzed hospitalization as the main outcome were too small to show effectiveness. The available evidence is not sufficient to determine if the influenza vaccine is beneficial in preventing hospitalization among patients with asthma and COPD. It appears that a long-term, prospective controlled trial may be needed. However, this would be difficult because inclusion of subjects with these comorbidities would be unethical in placebo-controlled trials.

Our study had potential limitations. We evaluated outcomes without confirmation of influenza virus by laboratory test. Respiratory syncytial virus (RSV) epidemics usually overlap with influenza epidemics in Korea. Therefore, we were not able to exclude the possibility that other circulating respiratory viral pathogens, especially RSV influenced our results. Korea National Institute of Health (KNIH) surveillance data on the viral pathogen among patients with acute respiratory infection showed that 47.3% influenza infection and 1.4% RSV during this study period. The influence of other pathogens might be much lower than influenza. Second, it might have been different for hospitalization criteria according to each institution. Hospitalization was determined at the discretion of the medical professionals in cardiovascular and pulmonary clinics from each institute. Because this study was conducted retrospectively, hospitalization was determined without considering influenza vaccination status, so such a decision would not affect our results. Third, this observational study may be biased due to healthy vaccinee effect. Patients who choose vaccination are more likely to engage in healthy behaviors as compared with non-vaccinated patients. Therefore, the effect of vaccination could reflect these unmeasured differences and overestimate the effectiveness.²⁰ However, patients with more comorbidities who are at a greater risk of influenza hospitalization are more likely to receive an influenza vaccine. This would lead to an underestimation of the vaccine effectiveness. The healthy vaccine effect is not conclusive. We tried to overcome this influence by adjusting for comorbidities. However, the severity of comorbidities was not included and could affect the results. Fourth, vaccination history was taken from medical records. In addition, a series of telephone interviews was performed to confirm the information. Recall bias may exist as the data were collected retrospectively.

In conclusion, influenza vaccination reduced the risk of hospitalization due to acute exacerbation of cardiopulmonary disease. The vaccine should be administered primarily to patients with underlying cardiovascular disease, particularly to those patients who are 65 y and older. Additional studies investigating the effectiveness of widely used strategies, such as the use of adjuvants, high doses, and intradermal delivery are needed in high-risk patients.

Materials and Methods

During the 2011–2012 influenza epidemic, studies were performed at four university hospitals. The protocol was approved by the Institutional Review Board of each hospital, and this study was performed in accordance with the Helsinki Declaration and Good Clinical Practice.

Study design and case definition

We conducted a multicenter, matched case-control study. We retrospectively reviewed the medical records of patients who visited cardiovascular and pulmonary clinics during the influenza epidemic period. We selected patients 18 y of age or older with asthma (J45, J46), chronic obstructive lung disease (J41, J42, J43, J44), ischemic heart disease (I20, I21, I22, I23, I24), or congestive heart failure (I11.0, I13.0, I13.2, I50) based on ICD-10 diagnostic codes. We defined cases as patients who were hospitalized due to acute exacerbation of those diseases according to the diagnosis reports written by professional medical attendants. Controls were selected from outpatients who visited the same hospitals but who were not hospitalized. The cases and controls were matched 1:1 based on age, gender, and date of hospital visit. If there was no appropriate matched control, ±5 d differences from the hospital visit date or nearest age were permitted.

Study period

To define the study period, national influenza surveillance data was obtained from the Korean Influenza Surveillance System (KISS), which consists of a network of nationwide public health centers and private clinics that record the weekly proportion of patients who have influenza-like illness (ILI). KISS reports a weekly incidence of ILI per 1000 outpatients. We defined the influenza epidemic period as the week in which the ILI rate surpasses a 3.8. During the 2011–2012 influenza season, the epidemic period was from 25 December 2011 to 5 May 2012 in Korea.

Vaccine match during the study period

During the 2011–2012 epidemic period, type A(H1N1)pdm09 and A(H3N2) influenza viruses circulated, which were well matched to the corresponding strains of 2011–2012 seasonal vaccine, such as A/California/07/2009 or A/Perth/16/2009. In the case of type B strain, B/Brisbane/60/2008 (vaccine strain) and B/Florida/4/2006 circulated at the same time, and 73% of the isolated influenza B stains were antigenically similar to the vaccine strain.²¹

Data collection

The following data was collected from hospital medical records: age, gender, body mass index (BMI), smoking status, comorbidities, and influenza vaccination status in the 2011–2012 season. Comorbidities included diabetes, hypertension, cerebrovascular disease, neuromuscular disease, chronic renal disease, chronic hepatic disease, solid tumor, hematologic disease, autoimmune disease, and immunosuppressant use. To improve the accuracy of the data, a trained clinical research coordinator performed a standardized telephone interview to confirm the record information within two weeks of the hospital visit. Subjects were considered to be vaccinated if the vaccine was given less than 15 d before the hospital visit.

Statistical analyses

The characteristics of the study population are described as median ± standard deviation (SD) for continuous variables and as proportion for categorical variables. The differences among cases and controls were analyzed with McNemar’s test. To evaluate vaccine effectiveness, an odds ratio (OR) and its 95% CI were calculated by conditional logistic regression models. Confounding variables entered into the model were BMI, smoking status, comorbidities, (diabetes, hypertension, cerebrovascular disease, neuromuscular disease, chronic renal disease, chronic hepatic disease, solid tumor, hematologic disease, autoimmune disease, and immunosuppressant user) and influenza vaccination status. Adjusted ORs were estimated by including variables that were related to hospitalization with a P ≤ 0.05 in univariate conditional logistic regression model in a stepwise process. Vaccine effectiveness was defined as (1- adjusted OR) × 100. A protective effect was assumed when the OR was less than 1 and the 95% CI did not include 1. We also performed subgroup analyses to assess effectiveness according to age group (≥65y, <65 y) and disease category (respiratory disease, cardiovascular disease) for the matched sets in the same stratum. A bilateral P value of <0.05 was considered significant. These analyses were conducted using SPSS 18.0 for Windows.

10.4161/hv.26858