Immune response to influenza vaccination in community-dwelling Chinese elderly persons

Vaccine 24 (2006) 5371–5380

Immune response to influenza vaccination in community-dwelling Chinese elderly persons S.L. Hui a , L.W. Chu a,∗ , J.S.M. Peiris b , K.H. Chan b , D. Chu c , W. Tsui c b

a Department of Medicine, Queen Mary Hospital, The University of Hong Kong, 102 Pokfulam Road, Hong Kong Department of Microbiology, Queen Mary Hospital, The University of Hong Kong, 102 Pokfulam Road, Hong Kong c Department of Family Medicine, Sai Ying Pun General Out-Patient Clinic, Hong Kong

Received 17 August 2005; received in revised form 7 April 2006; accepted 25 April 2006 Available online 3 May 2006

Abstract We investigated the immune antibody response to influenza vaccine in community-dwelling Chinese elderly persons in Hong Kong. One hundred and twenty-eight subjects were recruited in a single-blind, randomized, and placebo-controlled trial. There was no significant baseline difference between the vaccine and placebo groups regarding the seroprotection rates (PR) (haemagglutination inhibition [HI] titre ≥1:40) and geometric mean titres (GMT) of the HI antibody titers. The PR, GMTs and serological response rates increased significantly in the vaccinated versus placebo groups in A-H1N1 at both weeks 4 and month 6. The GMTs and serological response rates but not the PR for A-H3N2 and influenza B increased significantly in vaccinated versus placebo group at week 4 and month 6 post-vaccination. Multivariate logistic regression analyses of the seroconversion rate for A-H3N2 within the vaccinated group showed that gender, coronary heart disease and the serum albumin level were significant predictors (p = 0.018, 0.009 and 0.025, respectively). Influenza vaccination provoked a protective HI antibody response in community-living Chinese elderly persons. The mean number of unplanned hospital admissions per subject over 6 months was significantly lower in the vaccinated than in the placebo groups. Hospitalized elderly persons had poorer nutrition, 4-week post-immunization HI antibody titres and lower mini-mental state examination (MMSE) score than non-hospitalized elderly persons. Logistic regression analyses showed that chronic obstructive airway disease significantly increased the risk of hospitalization while the serum albumin level and 4-week A-H3N2 PR (HI ≥ 40) were independent predictors of a decreased risk of hospitalizations. © 2006 Elsevier Ltd. All rights reserved. Keywords: Influenza; Vaccination; Chinese; Elderly

1. Introduction Influenza and its associated complications leads to considerable morbidity and mortality in the elderly population [1–5]. Globally, the influenza virus infects about 10–20% of total population annually and is estimated to lead to more than 3 million cases of severe illness and with up to half a million deaths [5–8]. A recent systematic review of influenza vaccination in temperate regions and with Caucasian ethnic groups reports that vaccination prevents 42% deaths from pneumonia or influenza, 46% of pneumonia, 45% of hospitalizations ∗

Corresponding author. Tel.: +852 2855 3315; fax: +852 2974 1171. E-mail address: [email protected] (L.W. Chu).

0264-410X/$ – see front matter © 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.vaccine.2006.04.032

due to pneumonia or influenza and 60% of all cause mortality in the elderly living in nursing homes but only 26% of hospital admissions for influenza and pneumonia and 42% of all cause mortality among elderly living in the community [9]. The Advisory Committee on Immunization Practices (ACIP) recommended annual influenza vaccination for persons aged 50 years old and over as well as for residents of nursing homes in the US while the World Health Organization (WHO) recommended annual influenza vaccination to persons aged ≥65 years [2,6]. However, in countries with a tropical or sub-tropical climate, the impact of influenza is less well appreciated and there is a perception that influenza may have less clinical impact in warmer climates [10]. The lack of good estimates

5372

S.L. Hui et al. / Vaccine 24 (2006) 5371–5380

of the clinical disease burden of influenza and of the immunogenicity and efficacy of influenza vaccine in the tropics adds to scepticism on the efficacy and utility of influenza vaccine in these regions. In contrast to temperate regions where influenza activity is concentrated within a few weeks in the winter of each year, the more diffuse seasonality of influenza in the tropics and subtopics obscures the impact of influenza. As a consequence of this perception that the clinical disease burden of influenza is less in the tropics, vaccine utilization is less. For example, in the year 2000, Hong Kong which has a per capita GDP comparable to that of Australia used less that 1/5th the doses of influenza vaccine per 1000 population [11]. In the Chinese population, data from a nationally representative Chinese cohort of 169,871 men and women 40 years of age and older in China shows that influenza and pneumonia have an age-standardized mortality rate of 43.9 per 100,000 person-years and are the fourth leading cause of death in China [12]. The Hong Kong SAR Government has provided free influenza immunization to institutional elderly in the past 8 years (i.e. since 1998) and more recently, to persons in high-risk groups such as health care workers, the elderly and the disabled [13,14]. For elderly residents living in old aged homes, the vaccine uptake rate was reported to be over 87% [15]. In a recent quantitative review of antibody response to influenza vaccination in 31 studies, the antibody response is concluded to be considerably lower in the elderly (17–53%) than in younger adults (70–90%) [16]. Although there are many studies on the antibody immune response to the influenza vaccine in Caucasian populations in temperate regions [9,16–27], there is a paucity of information from the tropics and on non-Caucasian ethnic groups. Jianping and colleagues have reported the use of influenza vaccine in 1356 very fit Chinese from the Chinese Army. Influenza and common symptoms were reduced in all age groups but less effective in the elderly aged 60 years and over (i.e. 84.8% in children, 74.0% in adults and 68.6% in elderly people) [28]. Antibody response was not investigated in this study. Hospitalization data were not reported too. Moreover, nutritional, cognitive and functional profiles were not performed in this study. Previous Caucasian studies have reported that elderly with poor nutritional and functional status might have poor antibody responses after vaccination. Functional limitations were associated with both an increased risk of death and a decreased likelihood of vaccination [29–31]. We therefore included chronic diseases, nutritional, functional and cognitive function observations in our study. The variable seasonal pattern of influenza in the tropics and sub-tropics has also led to uncertainty about the immunogenicity and efficacy of influenza vaccines in such regions [10,32]. In a recent study in Hong Kong, we have shown that influenza-related mortality from respiratory and ischaemic heart disease in the elderly (≥65 years) was comparable to that of the USA [32]. In this study, we address the antibody response to influenza vaccine in community-dwelling elderly

Chinese persons in Hong Kong. The secondary objective was to explore the possible beneficial effect of influenza vaccination in decreasing unplanned hospital admissions.

2. Materials and methods 2.1. Design This was a randomized, single-blind, placebo-controlled study. The study protocol was approved by the Institutional Review Board of the University of Hong Kong and Hospital Authority Hong Kong West Cluster and complied with the Declaration of Helsinski. 2.2. Subjects Subjects were recruited from the geriatrics out-patient clinic at the Queen Mary Hospital and Sai Ying Pun general out-patient clinic of Hong Kong. Subjects were recruited from October 2003 to February 2004. The inclusion criteria were Chinese ethnicity, aged ≥60 years old, living in the community, could understand and be willing to comply with the study and 6-month follow-up, with or without previous year 2002–2003 vaccination and had written informed consent. The exclusion criteria were infectious diseases, fever (temperature ≥37.5 ◦ C) at baseline visit, living in institutions, known allergy to egg or any component of the vaccines, uncontrolled coagulopathy or blood disorders contraindicating intramuscular injection, known congenital or acquired immunodeficiency (including HIV infection), had received immunosuppressive treatment, active malignancy in the past 5 years, previous immunization with influenza vaccine in previous 12 months, presence of active psychiatric illness or advanced dementia, severe deafness or language barrier, previous enrolment in another clinical study and pre-menopausal women. 2.3. Outcome measures The primary outcome was an increase in the seroprotection rate (PR) after vaccination. PR was defined as the proportion of subjects with a hemagglutination inhibition (HI) antibody titers of ≥1:40 [16,26]. The secondary outcome was the cumulative numbers of unplanned hospitalizations over 6 months of follow-up. 2.4. Sample size calculation The sample size calculation was performed for the primary outcome of the seroprotection rate. Based on the data published by Govaert et al., the protection rates in the control versus vaccine groups were 3% versus 43% [for A/Singapore/6/86 (H1N1)] and 3% versus 68% [for A/Beijing/353/89 (H3N2)], respectively [26]. A total sample size of 100 subjects (with 50 each in the control and vaccine

S.L. Hui et al. / Vaccine 24 (2006) 5371–5380

groups) would achieve a 99% power to detect a 40% (i.e. 3–43%) difference in the PR (using a two-sided Chi-square test with continuity correction at a p-value <0.05). 2.5. Randomization Before the implementation of the study, a randomization schedule was generated from a random number table. Subjects were assigned consecutively according to their order of recruitment into the study. Subjects were but the nurse who administered the influenza vaccine or placebo was not blinded. 2.6. Influenza vaccine and placebo The influenza vaccine used in this study was the commercially available trivalent sub-unit inactivated influenza vaccine (VaxiGrip, Sanofi Pasteur, France) which was based on the WHO recommendation for the northern hemisphere for 2003/2004 (winter). The vaccine contained 15 ␮g of each viral strain A/Moscow/10/99 (H3N2), A/New Caledonia/20/99 (H1N1), and B/Hong Kong/330/2001. Eligible subjects who fulfilled the inclusion and exclusion criteria and had signed the written informed consent were consecutively recruited. At baseline, demographic data, history of co-morbid illness, number of regular medications, and history of influenza vaccination over the previous 4 years and assessment on cognitive status by mini-mental state examination (MMSE), functional status (Barthel Index, Instrumental Activities of Daily Living) and nutritional status (BW, BMI) were obtained. Then, a 10 ml venous clotted blood sample was taken from each subject for analysis of the HI titer and serum albumin level. They were then randomized into either the “influenza vaccination” or “placebo” group according to the randomization sequence in pre-sealed envelopes. According to the group assignment, a single dose of vaccine or placebo was administered intramuscularly in the deltoid region. A longitudinal telephone follow-up to investigate any adverse effects and outcomes on the third day after vaccination and at 4-weekly intervals for 6 months was performed. All subjects in this study were under the Hospital Authority of Hong Kong health care system, which had a fully computerized record system. The number of unplanned admissions of each subject was retrieved from this computer system. 2.7. Blinded assay of serum status antibody Venous blood was collected at baseline, week 4 and month 6 for analysis of serum albumin and HI titer. After clotting and centrifugation, the sera were stored at −20 ◦ C for future assays. Technicians who performed the antibody assay were blinded to the vaccination status of the subjects. Hemagglutination inhibition antibody assays were performed by standard microtiter techniques after removal of non-specific inhibitors in the serum with receptor destroying enzyme (RDE) (1:3), incubation overnight at 37 ◦ C followed by heat-inactivation

5373

at 56 ◦ C for 30 min. The three consecutive serum samples from each subject were tested in parallel for each of the test antigens. The 2003–2004 WHO influenza reagent kit from the WHO Collaborating Center for influenza, CDC, Atlanta, Georgia, USA was used to analyze the HI titer of the serum samples [24]. Serial two-fold dilutions of RDEtreated serum from 1:10 to 1: 10,240 were titrated against ␤-propiolactone-treated control influenza A and ether-treated influenza B antigen in the hemagglutination inhibition assays. 2.8. Statistical analyses Differences in the baseline characteristics between the vaccinated and placebo groups were examined. Chi-square statistics and Student’s t-test was employed for categorical variables and continuous data, respectively. The geometric mean titres (GMT) of HI antibodies of the two groups were compared after data transformation to log10 . The Mann–Whitney test was used to compare GMTs between the two groups. The Wilcoxon signed ranks test was used to compare GMTs within the vaccinated or placebo group. For multivariate analyses, logistic regression were used for seroprotection and seroconversion, while two-way ANOVA and ANCOVA were employed for GMT [26,29], Statistical significance was taken as p < 0.05 for all tests. All statistical analyses were performed using SPSS 13.0 for Windows.

3. Results 3.1. Subjects’ recruitment and follow-up One hundred and twenty-eight subjects were recruited from 23 October 2003 to February 2004. The overall dropout rate was 3.9% at 6 months. One subject in vaccinated group defaulted follow-up at week 4. The reason was refusal of blood taking at week 4. At 6 months, another four subjects defaulted from the study. One subject died of an accident a few days before the month 6 visit. Three subjects refused blood taking. 3.2. Baseline characteristics of subjects The mean age was 74.4 ± 7.4 years and 52.3% were males. Comparisons of baseline characteristics of the subjects in the two groups were shown in Table 1. 65 (50.4%) belonged to the vaccinated group and 63 subjects were in the placebo group. The mean age was 74.8 ± 7.6 years in the vaccinated and 74.0 ± 7.3 years in the placebo groups (p = 0.58, unpaired t-test). There were more male subjects in the placebo than vaccinated groups (63.5% versus 41.5%, p = 0.01, Chi-square statistics). Both groups were similar with regards to age, nutritional status, functional status, cognitive status, living arrangement, previous history of influenza vaccination and medical condition. Only three subjects had history of previous vaccination during 2002–2003

5374

S.L. Hui et al. / Vaccine 24 (2006) 5371–5380

Table 1 Comparison of baseline characteristics Vaccinated group (n = 65)

Placebo group (n = 63)

p-value

Gender: male, n (%) Age (years) mean ± S.D.

27 (41.5) 74.75 ± 7.6

40 (63.5) 74.03 ± 7.3

0.010a 0.583b

Nutritional status BW (kg), mean ± S.D. BMI, mean ± S.D. Hgb (g/l), mean ± S.D. Albumin, mean ± S.D.

58.8 ± 10.6 21.8 ± 3.2 13.2 ± 1.3 42.1 ± 2.4

61.3 ± 12.8 21.9 ± 3.4 13.5 ± 1.4 41.7 ± 3.0

0.218b 0.880b 0.170b 0.454b

Cognitive status MMSE (30), mean ± S.D.

25.8 ± 3.4

26.2 ± 3.3

0.515b

Functional status ADL (20), mean ± S.D. IADL (8), mean ± S.D.

19.8 ± 0.6 7.8 ± 0.6

19.6 ± 1.0 7.7 ± 1.0

0.194b 0.631b

Residence Alone, n (%) Spouse, children or relative, n (%)

8 (12.3) 57 (87.3)

10 (15.9) 53 (84.1)

0.56a

Previous history of vaccination 2002–2003, n (%) 2001–2002, n (%) 2000–2001, n (%) 1999–2000, n (%)

2 (3.1) 1 (1.5) 0 0

1 (1.6) 2 (3.2) 0 1 (1.6)

1.000c 0.616c N.A. 0.492c

Medical condition Number of co-morbid diseases, mean ± S.D. History of HT, n (%) History of coronary heart disease, n (%) History of DM, n (%) History of asthma, n (%) History of chronic obstructive airway disease, n (%) Number of regular medications, mean ± S.D.

1.48 ± 0.9 48 (73.8) 6 (9.2) 20 (30.8) 2 (3.1) 1 (1.5) 2.37 ± 1.7

1.62 ± 0.9 47 (74.6) 9 (14.3) 12 (19.0) 0 3 (4.8) 2.46 ± 1.6

0.361b 0.922a 0.374a 0.114a 0.496c 0.361c 0.753b

a b c

Chi-square statistics. Independent t-test. Fisher’s exact test.

and two of them belonged to the vaccinated group. None of the subjects was taking medications known to alter immune response (e.g. corticosteroids or other immunosuppressive agents). None of them had history of conditions which associated with immune dysfunction (Table 1). 3.3. Pre- and post-vaccination seroprotection (PR) HI titers

to a significant difference at 4 weeks post-immunization. However, at 6 months post-vaccination the vaccinated group had a significantly higher PR for A-H3N2. In multivariate analyses, gender and age (60–74 versus ≥75 years) had no significant effect on the PR while vaccination led to significantly higher proportions of PR HI antibody response than placebo for A-H1N1 at 4 weeks and 6 months (Table 2). 3.4. Pre- and post-vaccination GMTs

Results of pre- and post-vaccination immune antibody response were shown in Table 2. The HI titers of 1:≥40 was regarded as a seroprotective antibody titre against influenza. Before vaccination, there was no statistically significant difference in the proportions of subjects with PR HI antibody titers of A-H1N1, A-H3N2 and influenza B between the vaccinated and the placebo groups. Four weeks later, there was a statistically higher proportions of subjects with PR on the HI titers in the vaccinated group (85.9%) than the placebo group (42.9%) for A-H1N1 (p < 0.001, Chi-square statistics). The significant difference between the two groups remained at 6 months post-immunization. Pre-vaccination PR was already high for A-H3N2 and influenza B and consequently vaccination did not lead

The immune antibody response of the vaccinated and placebo subjects was presented also as GMTs. At baseline, there was no statistically difference in the GMT between the vaccinated and placebo groups for the three influenza viruses (A-H1N1, A-H3N2 and influenza B). Significant increases in the GMTs were observed in the vaccinated group but not the placebo group for all three influenza vaccine virus strains (all p-values < 0.001, Mann–Whitney U-test). Within each group, significant increases in the GMTs were observed between week 4 versus baseline and month 6 versus baseline (all pvalues < 0.001, Wilcoxon signed ranks test) in vaccinated but not the placebo groups for all the three vaccine virus strains. In multivariate analyses, gender and age had no significant

S.L. Hui et al. / Vaccine 24 (2006) 5371–5380

5375

Table 2 Seroprotection rate (HI ≥ 40) in vaccinated and placebo groups with multivariate adjustment for gender and age Antigen

H1N1d

H3N2d

BHKd

Groupa

A B p-value A B p-value A B p-value

Baseline

4 weeks post-vaccination

HI ≥ 40b

HI ≥ 40b

n = 127 (%) 27 (42.2) 27 (42.9) 0.939 54 (84.4) 55 (87.3) 0.636 63 (98.4) 63 (100) 1.0e

6 months post-vaccination

n = 127 (%)

Unadjusted OR (95% CI)

Adjusted ORc (95% CI)

HI ≥ 40b n = 124 (%)

Unadjusted OR (95% CI)

Adjusted ORc (95% CI)

55 (85.9) 27 (42.9) <0.001 63 (98.4) 58 (92.1) 0.115e 64 (100) 62 (98.4) 0.5e

8.1 (3.4, 19.3) <0.001 5.4 (0.6, 47.8) 0.115e NA (NA) NA

7.7 (3.2, 18.5) <0.001 4.4 (0.5, 39.6) 0.19 14500000 (0, NA) 0.997

48 (78.7) 26 (41.3) <0.001 61 (100) 54 (85.7) 0.003e 61 (100) 63 (100) NSe

5.3 (2.4, 11.6) <0.001 NA (NA) NA NA (NA) NA

5.6 (2.5, 12.6) <0.001 0.000008 (0, NA) 0.997 NA (NA) NA

Notes: Not available (NA) indicates that the OR or CI cannot be calculated (i.e. 0 subject in ≥1 of the four values in the 2 by 2 tables). a Group A: vaccinated group (n = 64); group B: placebo group (n = 63). b Chi-square statistics. c Odds ratio (OR) of vaccination versus placebo after adjustment for gender and age (60–74 versus ≥75 years) in multivariate analyses by logistic regression for seroprotection and seroconversion rates; CI: confidence interval; p > 0.05 for both gender and age in these analyses. d H1N1: A/New Caledonia/20/99; H3N2: A/Moscow/10/99; BHK: B/Hong Kong/330/2001. e Fisher’s exact test.

effect on the GMTs while the vaccinated group had significantly very much higher GMTs than placebo group for all the three strains at both 4 weeks and 6 months (Table 3). 3.5. Serological response rate after influenza vaccination

three strains at 4 weeks, which was the expected time for the peak response. At 6 months, the proportions of the serological response rate for the three strains in the vaccinated group had all decreased (Table 4). 3.6. Predictors of serological response in the vaccinated subjects at 4 weeks

The serological response rate was defined as the proportions of subjects with a four-fold or greater increase in the HI titer post-vaccination versus pre-vaccination. For all three vaccine strains, the post-vaccination serological response rates were significantly higher in the vaccinated versus placebo groups. In multivariate analyses, gender and age had no significant effect on the serological response rate while vaccination led to significantly much higher proportions of the serological response rate than placebo for all the

Predictors of the serological responses at 4 weeks (time of peak immune antibody response) for the seroprotection (HI ≥ 40) rate (PR) and seroconversion rate (≥4 folds increase in HI) were analyzed. Multivariate logistic regression analyses of the seroconversion rate for A-H3N2 showed that gender, coronary heart disease and the serum albumin level were significant predictors (p = 0.018, 0.009 and 0.025, respectively) while age showed a non-significant trend

Table 3 Geometric mean titers in the vaccinated and placebo group against viruses contained in the 2003–2004 influenza vaccine with multivariate adjustment for gender and age Antigen

Groupa

Before vaccinationb (n = 127)

4 weeks (n = 127) Vaccinec

H1N1d

H3N2d

BHKd a

A B p-value

26.3 27.5 0.633

144.9 30.1 <0.001

A B p-value

80.9 84.5 0.830

632.8 93.3 <0.001

A B p-value

120.5 125.9 0.489

584.4 128.8 <0.001

6 months (n = 124)

Genderc – 0.28 – 0.12 – 0.53

Agec

Vaccinec

Genderc

Agec

– 30.1 0.34

78.2

–

–

<0.001

0.76

0.28

–

–

0.67

0.72

–

–

0.21

0.23

– 93.3 0.98

435

– 121.3 0.06

371

<0.001

<0.001

Group A: vaccinated group (n = 64); group B: placebo group (n = 63). b Between group comparison at baseline: Mann–Whitney U-test. c Multivariate analyses by two-way ANOVA for GMT (with log GMT) with vaccination status as predictors and gender and age (60–74 versus ≥75 years) as confounders. d H1N1: A/New Caledonia/20/99; H3N2: A/Moscow/10/99; BHK: B/Hong Kong/330/2001.

5376

S.L. Hui et al. / Vaccine 24 (2006) 5371–5380

Table 4 Serological response rate in vaccinated and placebo groups with multivariate adjustment for gender and age Antigen

Groupb

4 weeks post-vaccination

6 months post-vaccination

With ≥four-fold

With ≥four-fold increasea

increasea

n = 127 (%)

Unadjusted OR (95% CI)

Adjusted ORc (95% CI)

n = 124 (%)

Unadjusted OR (95% CI)

Adjusted RRc (95% CI)

H1N1d

A B p-value

16 (25) 1 (1.6) <0.001

20.4 (2.6, 162.3) 0.004

20.4 (2.6, 162.3) 0.004

8 (13.1) 0 (0) 0.003e

0.000008 (0, NA) 0.003

0.000008 (0, NA) 0.003

H3N2d

A B p-value

30 (46.9) 1 (1.6) <0.001

54.7 (7.1, 419.0) <0.001

49.2 (6.4, 379.3) <0.001

18 (29.5) 3 (4.8) <0.001

8.4 (2.3, 30.2) 0.001

8.5 (2.3, 31.7) 0.002

BHKd

A B p-value

11 (17.2) 1 (1.6) 0.003

12.9 (1.6, 103.0) 0.016

17.6 (2.1, 148.3) 0.008

7 (11.5) 0 (0) 0.006e

0.000008 (0, NA) 0.006

0.000008 (0, NA) 0.003

Notes: NA: not available because of either 0 subject in one or both groups (cannot be calculated). a Chi-square statistics. b Group A: vaccinated group (n = 64); group B: placebo group (n = 63). c Odds ratio (OR) of vaccination versus placebo after adjusted for gender and age (60–74 versus ≥75 years) in multivariate analyses by logistic regression for seroprotection and seroconversion rates; CI: confidence interval; p > 0.05 for both gender and age. d H1N1: A/New Caledonia/20/99; H3N2: A/Moscow/10/99; BHK: B/Hong Kong/330/2001. e Fisher’s exact test.

towards a difference (p = 0.062). For A-H1N1, age, hypertension and coronary heart disease showed some non-significant trends towards a difference (p = 0.057, 0.054 and 0.073, respectively) (Table 5). Multivariate logistic regression analyses of the PR for AH3N2, A-H1N1 and B-HK showed that gender, age, history of vaccination (2002–2003), diabetes, chronic obstructive airway disease (COAD), asthma, bronchiectasis, hypertension, coronary heart disease, number of co-morbid diseases, MMSE score, BI, IADL, BMI, and the serum albumin level were all statistically insignificant as independent predictors. None of these were significant predictors of the seroconversion rate for B-HK.

3.7. Comparison of unplanned hospital admissions in the vaccinated and placebo groups Over 6 months of follow-up, 15.9% of the control subjects had been hospitalized at least once while the corresponding proportion in the vaccinated subjects was only 4.8% (i.e. 69.8% decrease, p = 0.04; Chi-square statistics). The mean number of all cause unplanned hospital admissions per subject over 6 months decreased (i.e. 68.8% reduction) significantly in the vaccinated versus that of placebo groups (0.05 ± 0.21 versus 0.16 ± 0.37, respectively, p < 0.05, independent t-test).

Table 5 Multivariate logistic regression analyses of predictor of seroconversion of influenza HI antibody at 4 weeks in vaccinated subjects (placebo subjects excluded) A-H1N1 (n = 64)

Gender (females) versus (males) Old (≥75 years) versus young (age 60–74 years) Hypertension Coronary heart disease Serum albumin level (g/dl)

A-H3N2

Odds ratio (OR) of seroconversion (HI increase ≥four-folds)

95% CI of OR

p-valuea

1.39

0.35, 5.6

0.64

0.25

0.058, 1.05

8.55 7.52 0.96

0.97, 75.72 0.83, 68.42 0.75, 1.22

95% CI of OR

p-valuea

4.84

1.31, 17.91

0.018b

0.057

0.30

0.09, 1.06

0.062

0.054 0.073 0.73

1.98 31.32 1.36

0.51, 7.72 2.37, 412.48 1.04, 1.77

0.32 0.009b 0.025b

Odds ratio (OR) of seroconversion (HI increase ≥four-folds)

a Note: Adjusted for other confounders (i.e. previous influenza vaccination 2002–2003, DM, COAD, asthma, bronchiectasis, number of co-morbid diseases, BMI, MMSE score, BI, IADL) which were all statistically non-significant (NS). b p < 0.05.

S.L. Hui et al. / Vaccine 24 (2006) 5371–5380

5377

Table 6 Comparison of baseline features in hospitalized and non-hospitalized subjects (both vaccinated and unvaccinated)

Gender: males, n (%) Age (years), mean ± S.D. Living alone BW (kg), mean ± S.D. BMI (kg/m2 ), mean ± S.D. Serum Hgb level (g/dl), mean ± S.D. Serum Albumin level (g/dL), mean ± S.D. MMSE, mean ± S.D. BI, mean ± S.D. IADL, mean ± S.D. History of vaccination 2002–2003, n (%) History of vaccination 2001–2002, n (%) History of vaccination 2000–2001, n (%) History of HT, n (%) History of coronary heart disease, n (%) History of DM, n (%) History of asthma, n (%) History of chronic obstructive airway disease, n (%) Number of co-morbid diseases, mean ± S.D.

Non-hospitalized subjects (n = 113)

Hospitalized subjects (n = 13)

p-valuea

61 (54.0) 73.9 ± 7.6 13 (11.5) 61.0 ± 11.2 22.1 ± 3.1 13.4 ± 1.4 42.2 ± 2.4 26.2 ± 3.2 19.7 ± 0.8 7.74 ± 0.85 3 (2.7) 2 (1.8) 0 (0) 83 (73.5) 12 (10.6) 28 (24.8) 1 (0.9) 1 (0.9) 1.5 ± 0.87

6 (46.2) 78.0 ± 4.6 5 (38.5) 53.2 ± 14.7 19.9 ± 4.0 12.7 ± 1.2 39.3 ± 4.2 24.0 ± 4.3 19.5 ± 1.0 7.77 ± 0.60 0 (0.0) 1 (7.7) 0 (0) 10 (76.9) 3 (23.1) 4 (30.8) 0 (0.0) 3 (23.1) 1.9 ± 0.95

0.59 0.009b 0.021b 0.023b 0.022b 0.07 <0.001b 0.025b 0.52 0.92 0.42 0.27 NS 0.79 0.23 0.65 0.64 0.002b 0.15

a Note: Gender and history of previous vaccinations, HT: heart disease; DM: asthma; chronic obstructive airway disease by χ2 statistics; other variables by independent t-test; BW: body weight; BMI: body mass index; MMSE: mini-mental state examination; BI: Barthel Index; IADL: instrumental activity of daily living; HT: hypertension; DM: diabetes mellitus. b p < 0.05.

3.8. Relationship of the 4-week serological response and unplanned hospital admission

admissions, these potential confounders were included in the multivariate logistic regression analyses. There was no significant difference between the two groups for other baseline clinical features (Table 6) and the baseline HI antibody levels (Table 7). Logistic regression analyses showed that COAD significantly increased the risk of hospitalization while the serum albumin level and 4-week A-H3N2 PR (HI ≥ 40) were independent predictors of a decreased risk of hospitalizations. However, the A-H1N1 PR, B-HK PR and the seroconversion rates of all the three influenza strains were not independent predictors of hospitalizations (Table 8).

Comparison of the baseline characteristics in both groups between the hospitalized and non-hospitalized subjects showed that hospitalized subjects were significantly older, more likely to be living alone, had chronic obstructive airway disease (COAD), lower body weight, BMI, serum albumin level and MMSE score than non-hospitalized subjects (Table 6). These were potentially confounding factors. In the assessment of the effect of the 4-week serological response (using PR and seroconversion rate) in reducing hospital Table 7 Serological response of HI antibody in hospitalized and non-hospitalized subjects Antigen

Group

Geometric mean titers(GMT)

Before vaccination H1N1

Non-hospitalized (n = 113) Hospitalized (n = 13) p-valuea

27.5 24.5 0.88

H3N2

Non-hospitalized (n = 113) Hospitalized (n = 13) p-valuea

79.4 20.2 0.13

B

Non-hospitalized (n = 113) Hospitalized (n = 13) p-valuea

125.9 102.3 0.23

4-weeks 70.8 28.8 0.039b

6-month

With ≥four-fold increase (4-week versus baseline)

With ≥four-fold increase (6-month versus baseline)

With ≥four-fold increase (6-month. versus 4-week)

4-weeks (% in each group)

6-month (% in each group)

6-month (% in each group)

50.7 30.6 0.20

15.0 0 0.046

7.2 0 0.18

1.8 0.0 0.50

251.2 166.0 0.49

190.5 281.8 0.44

26.5 7.7 0.097

16.2 23.1 0.55

9.2 23.1 0.17

295.1 141.3 0.03b

223.9 120.2 0.049b

9.7 7.7 0.81

6.3 0 0.21

2.7 0.0 0.41

Notes: HI antibody-haemagglutination inhibition antibody. a p-value for between group comparison for GMT (by Mann–Whitney U-test, two-tailed) and four-fold increase (by χ2 statistics). b p < 0.05.

5378

S.L. Hui et al. / Vaccine 24 (2006) 5371–5380

Table 8 Multivariate logistic regression analyses of seroprotection and other predictors of hospitalizations

Serum albumin level (g/dl) Chronic obstructive airway disease Four-week A-H3N2 PR HI ≥ 40 Four-week A-H1N1 PR HI ≥ 40 Four-week B-HK PR HI ≥ 40 Four-week ≥four-fold increase A-H3N2 HI Four-week ≥four-fold increase A-H1N1 HI Four-week ≥four-fold increase B-HK HI

Odds ratio (OR) of hospitalization

95% CI of OR

p-value*

0.6 150 0.078 – – – – –

0.43, 0.82 10.1, 2237.2 0.01, 0.023 – – – – –

0.001 <0.001 0.016 NS NS NS NS NS

* Note: Adjusted for other confounders (i.e. gender, age of 60–74 versus 75+, living alone, BMI, MMSE score, BI, IADL) which were all statistically non-significant (NS).

3.9. Natural influenza infection occurring between 4 weeks and 6 months Three (23.1%) of the 13 hospitalized patients and 10 (9.2%) of 113 non-hospitalized patients had a four-fold rise in antibody to H3N2 virus between 4 weeks and 6 months post-immunization (p = 0.17, χ2 square statistic), suggesting that they had a natural influenza infection which might or might not have coincided with their hospitalisation (Table 7). 3.10. Comparison of adverse effects between the vaccinated and placebo groups Mild adverse reactions were reported by 10 (15.4%) and two (3.2%) in the vaccinated and placebo groups, respectively (p = 0.018, Chi-square statistics). No fever was reported in either group. Two (3.1%) subjects in the vaccinated and three (4.8%) in the placebo groups reported headache. Muscle aches occurred in one subject (1.6%) in the placebo group but none in the vaccinated group. Nine subjects (13.8%) reported soreness, redness or swelling over the local injection site in the vaccinated group while none occurred in placebo group (p = 0.003, Fisher’s exact test).

4. Discussions We found good immune HI antibody responses to influenza vaccination in community elderly Chinese living in the sub-tropical city of Hong Kong which were comparable with previous findings from temperate regions and on predominantly Caucasian ethnic groups [9,16–27]. A higher HI antibody titre was found in vaccinated subjects for all the serological parameters examined (i.e. seroprotection rate, GMT or serological response rate). The GMT was the most sensitive parameter and showed the strongest significant association (p < 0.001) with the influenza vaccination status. Moreover, the high antibody titers in the vaccinated group remained at protective levels for up to 6 months after influenza vaccination. This is critically important when influenza virus circulation may continue to occur many months after vaccination as happens in the tropics. Therefore, a single dose of

vaccine may suffice even in this setting of the communitydwelling elderly persons in Hong Kong. On the whole, we found no significant effects on the immune responsiveness with age, gender, physical and cognitive functional status, except for the 4-week A-H3N2 seroconversion rates which were higher in females, patients with coronary heart disease and high serum albumin level (Table 5). Thus, poor nutrition as indicated by a low serum albumin level gave rise to a low rate of seroconversion. This was in agreement with a previous report [29]. We did not observe any significant effect of poor function status with the antibody response, which was different from that reported by Jackson and colleagues [31]. This might be due to the ceiling effect. On the average, our subjects were all communitydwelling ambulatory elderly with good cognitive and functional status, which were reflected by a mean Barthel Index of 19.6 out of 20 (Table 1). Our finding of a higher A-H3N2 antibody response in elderly with coronary heart disease was different from the findings of similar HI antibody response in persons with congestive heart failure versus controls [33]. Thus, our patients with mild and stable coronary heart disease appeared to have better immune response to influenza vaccination than those with congestive heart failure. Moreover, our findings of a high A-H3N2 antibody response in elderly with coronary heart disease was in line with previous reports which demonstrated clear clinical efficacy of influenza vaccination in reducing myocardial infarction and cardiovascular events [34,35]. We found influenza vaccination prevented approximately 69% of hospital admissions in our community cohort. The number to treat (NNT) to prevent one hospitalization was 9, with a 95% C.I. of 5 to 143. This is in agreement with the conclusion of a recent review on the benefit of influenza vaccination in preventing hospital admissions [9]. We believe the protective HI antibody response explained the reduction in unplanned hospital admissions in the vaccinated elderly, who had higher HI antibody titires than the non-vaccinated subjects. The 4-week HI antibody level would reflect the immune protective level against influenza infection. Hospitalized persons had a lower GMT HI antibody levels than non-hospitalized persons at 4 weeks post-immunization for the A-H1N1 and influenza B virus. The 4-week seroconver-

S.L. Hui et al. / Vaccine 24 (2006) 5371–5380

sion rates of A-H1N1 and H3N2 were lower in hospitalized subjects too (Table 7). This may reflect the poor immune competence of the hospitalized cohort of subjects. In addition, hospitalized elderly persons had poorer nutritional indices and cognitive function than non-hospitalized elderly persons. After multivariate adjustment of these confounders, we found a 92% reduction in hospitalization in subjects who had the 4-week seroprotection of A-H3N2 than those who did not (Table 8). Influenza virus circulation in 2004 peaked in March and continued at high levels through to September and the viruses circulating were predominantly A-H3N2 (93%) and influenza B (6.5%) (http://www.info.gov.hk/dh/diseases/ flu 2004.htm). We believe the 4-week A-H3N2 HI antibody response in our subjects has prevented most of these hospital admissions during this influenza season in Hong Kong (Table 8). Interestingly, 3 (23%) of the 13 hospitalized subjects experienced a four-fold or greater increase in HI antibody titres of H3N2 virus between 4 weeks and 6 months postimmunization indicating that they and had natural influenza infection (Table 7). Whether this infection was temporally associated with the hospital admission is unknown. Most of our subjects were ambulatory and non-frail with low numbers of co-morbid diseases and medications. They were comparable in terms of age and gender distribution with the subjects in another local district-based population survey [36]. Therefore, we believed the results of this study would be applicable to the community-dwelling Chinese elderly population in Hong Kong. 4.1. Limitations This study was single-blind with blinding of the subject but not the investigator who administered the influenza vaccine. Although it would better to perform a double-blind study, this study was not sponsored by a pharmaceutical company and therefore it was not possible to obtain identical-looking matched placebo. Nevertheless, the primary outcome of this study was the influenza antibody titer which is an objective laboratory parameter. Furthermore, all laboratory assays were all performed by technicians who were blinded to the vaccination status of the subjects. As we had excluded elderly persons who lived in old aged homes, our findings could not be generalized to the institutionalized elderly persons.

5. Conclusions We have demonstrated that influenza vaccination was safe and effective in eliciting protective HI antibody responses in Chinese living in a sub-tropical city, Hong Kong. Our finding also shows a reduction in unplanned hospital admissions in vaccinated community-dwelling elderly persons. Currently, the Hong Kong SAR Government’s policy primarily provides

5379

free influenza vaccination for the institutionalized elderly and those with chronic illness. Ambulatory community-dwelling elderly persons are mostly excluded from this free vaccination policy. Our findings argue for the extension of free influenza vaccination for all community-dwelling elderly persons in Hong Kong. Taken together with previous data showing that influenza is associated with significant mortality in the elderly in Hong Kong, these findings provide objective evidence that influenza vaccination should be promoted for those over 65 years of age in tropical and sub-tropical setting as they have been in temperate climates. Acknowledgements Part of the work in the present study was performed by Miss Hui Sau Lan in partial fulfilment for the degree of Master Medical Sciences (Specialised Modules in Geriatrics Medicine) of the Faculty of Medicine at the University of Hong Kong. This study was partly supported by the Vice Chancellor’s Development Fund, The University of Hong Kong. References [1] Simonsen L. The global impact of influenza on morbidity and mortality. Vaccine 1999;17:3–10. [2] Centre of Disease and Control. Prevention and control of influenza: recommendation of the Advisory Committee on Immunization Practices (ACIP). Morbid. Weekly Report; 2004. [3] Sprenger MJW, Mulder PGH, Beyer WEP, Van SR, Masurel N. Impact of influenza on mortality in relation to age and underlying disease, 1967–1989. Int J Epidemiol 1993;22:334–40. [4] Perrotta DM, Decker M, Glezen WP. Acute respiratory disease hospitalizations as a measure of impact of epidemic influenza. Am J Epidemiol 1985;122:468–76. [5] Cox NJ, Subbarao K. Global epidemiology of influenza: past and present. Annu Rev Med 2000;51:407–21. [6] World Health Organization. Adoption on global agenda on influenza. Wkly Epidemiol Rec 2002;77:191–5. [7] Stephenson I, Zambon M. The epidemiology of influenza. Occup Med 2002;52(5):241–7. [8] Couch RB. Influenza: prospects for control. Ann Intern Med 2000;133:992–8. [9] Jefferson T, Rivetti D, Rivetti A, Rudin M, Di Pietrantonj C, Demicheli V. Efficacy and effectiveness of influenza vaccines in elderly people: a systematic review. Lancet 2005;366(9492):1165–74. [10] Fitzner KA, McGhee SM, Hedley AJ, Shortridge KF. Influenza surveillance in Hong Kong: results of a trial Physician Sentinel Programme. Hong Kong Med J 1999;5:87–94. [11] van Essen GA, Palache AM, Forleo E, Fedson DS. Influenza vaccination in 2000: recommendations and vaccine use in 50 developed and rapidly developing countries. Vaccine 2003;21(16):1780–5. [12] He J, Gu D, Wu X, Reynolds K, Duan X, Yao C, et al. Major causes of death among men and women in China. N Engl J Med 2005;353(11):1124–34. [13] Department of Health. http://www.info.gov.hk/dh/diseaes/influenza. htm. [14] Hospital Authority. Influenza vaccination for elderly at general outpatient clinics (2004) [press release]. [15] Department of Health Annual Report, 2001/2002. http://www.info. gov.hk/dh/ar0002/content.htm.

5380

S.L. Hui et al. / Vaccine 24 (2006) 5371–5380

[16] Goodwin K, Viboud C, Simonsen L. Antibody response to influenza vaccination in the elderly: a quantitative review. Vaccine 2006;24(8):1159–69. [17] Bernstein E, Kaye D, Abrutyn E, Gross P, Dorfman M, Murasko DM. Immune response to influenza vaccination in a large healthy elderly population. Vaccine 1999;17:82–94. [18] Beyer WEP, Palache AM, L¨uchters G, Nauta J, Osterhaus ADME. Seroprotection rate, mean fold increase, serological response rate: which parameter adequately expresses seroresponse to influenza vaccination? Virus Res 2004;103:125–32. [19] Beyer WEP, Palache AM, Sprenger MJW, Hendriksen E, Tukker JJ, Darioli R, et al. Effects of repeated annual influenza vaccination on vaccine sero-response in young and elderly adults. Vaccine 1996;14:1331–9. [20] Brandiss MW, Betts RF, Mathur U, Douglas RG. Responses of elderly subjects to monovalent A/USSR/77 (H1N1) and trivalent A/USSR/77 (H1N1)-A/Texas/77 (H3N2)-B/Hong Kong/72 vaccines. Am Rev Respir Dis 1981;124:681–4. [21] Breeze E, Mangtani P, Fletcher AE, Price GM, Kovats S, Roberts J. Trends in influenza vaccination uptake among people aged over 74 years, 1997–2000: survey of 73 general practices in Britain. BMC Fam Pract 2004;5:8. [22] Bruijn de IA, Remarque EJ, Beyer WE, Cessie leS, Masurel N, Ligth GJ. Annually repeated influenza vaccination improves humoral responses to several influenza virus strains in healthy elderly. Vaccine 1997;15(12–13):1323–9. [23] Brydak LB, Machala M, Mysliwska J, Mysliwski A, Trzonkowski P. Immune response to influenza vaccination in an elderly population. J Clin Immunol 2003;23:214–22. [24] Christenson B, Lundbergh P, Hedlund J, Ortqvist A. Effects of a large-scale intervention with influenza and 23-valent pneumococcal vaccines in adults aged 65 years or older: a prospective study. Lancet 2001;357:1008–11. [25] Goodeve A, Potter CW, Clark A, Jennings R, Schild GC, Yetts A. A graded dose study of inactivated surface antigen influenza B vaccine in volunteers. J Hyg Camb 1983;90:107–15.

[26] Govaert TME, Sprenger MJW, Dinant GJ, Aretz K, Masurel N, Knottnerus JA. Immune response to influenza vaccination of elderly people. A randomized double-blind, placebo-controlled trial. Vaccine 1994;12(13):1185–9. [27] Gross PA, Russo C, Dran S, Cataruozolo P, Munk G, Lancey SC. Time to earliest peak serum antibody response to influenza vaccine in the elderly. Clin Diagn Lab Immunol 1997;4(4):491–2. [28] Jianping H, Xin F, Changshun L, Bo Z, Linxiu G, Wei X, et al. Assessment of effectiveness of Vaxigrip. Vaccine 1999;17(Suppl. 1):S57–8. [29] Hara M, Tanaka K, Hrota Y. Immune response to influenza vaccine in healthy adults and the elderly: association with nutritional status. Vaccine 2005;23(12):1457–63. [30] Wick G, Grubeck-Loebenstein B. Primary and secondary alterations of immune reactivity in the elderly: impact of dietary factors and disease. Immunol Rev 1997;160:171–84. [31] Jackson LA, Nelson JC, Benson P, Neuzil KM, Reid RJ, Psaty BM, et al. Functional status is a confounder of the association of influenza vaccine and risk of all cause mortality in seniors. Int J Epidemiol Dec 20 2005 [epub ahead of print]. [32] Wong CM, Chan KP, Hedley AJ, Peiris JSM. Influenza-associated mortality in Hong Kong. Clin Infect Dis 2004;39:1611–7. [33] McElhaney JE, Herre JM, Lawson ML, Cole SK, Burke BL, Hooton JW. Effect of congestive heart failure on humoral and ex vivo cellular immune responses to influenza vaccination in older adults. Vaccine 2004;22(5–6):681–8. [34] Naghavi M, Barlas Z, Siadaty S, Naguib S, Madjid M, Casscells W. Association of influenza vaccination and reduced risk of recurrent myocardial infarction. Circulation 2000;102(25):3039– 45. [35] Madjid M, Awan I, Ali M, Frazier L, Casscells W. Influenza and atherosclerosis: vaccination for cardiovascular disease prevention. Expert Opin Biol Ther 2005;5(1):91–6. [36] Chu LW, Chu M. A survey on the health and health care needs of elderly people living in Central and Western District. Elite Business Services Ltd.; 1998.