Factors influencing infection by pandemic influenza A(H1N1)pdm09 over three epidemic waves in Singapore

Abstract

Introduction: Previous influenza pandemics had second and on occasion third waves in many countries that were at times more severe than the initial pandemic waves.

Objective: This study aims to determine the seroepidemiology of successive waves of H1N1pdm09 infections in Singapore and the overall risks of infection.

Methods: We performed a cohort study amongst 838 adults, with blood samples provided upon recruitment and at 5 points from 2009 to 2011 and tested by haemagglutination inhibition (HI) with A/California/7/2009 (H1N1pdm09). Surveys on key demographic and clinical information were conducted at regular intervals, and associations between seroconversion and these variables were investigated.

Results: After the initial wave from June to September 2009, second and third waves occurred from November 2009 to February 2010 and April to June 2010, respectively. Seroconversion was 13·5% during the first wave and decreased to 6·2% and 6·8% in subsequent waves. Across the three waves, the elderly and those with higher starting HI titres were at lower risk of seroconversion, while those with larger households were at greater risk. Those with higher starting HI titres were also less likely to have an acute respiratory infection.

Conclusions: The second and third waves in Singapore had lower serological attack rates than the first wave. The elderly and those with higher HI titres had lower risk, while those in larger households had higher risk of seroconversion.

Keywords: Epidemic waves; H1N1pdm09; haemagglutination inhibition; risk factors; seroconversion; seroepidemiology.

Figure 1

Time course of serological analyses and H1N1pdm09 cases detected by the National laboratory surveillance system. Line graph – Weekly number of H1N1pdm09 (blue), H3N2 (purple) and influenza B (black) cases detected by the National Public Health Laboratory. Bar graphs –% of serological samples with the corresponding antibody titres for each blood sample.

Figure 2

Distribution of titres by influenza vaccination experience corresponding to the sample periods. Bar graphs –% of observations with the corresponding titre. Closed diamond and open circles with error bars – geometric mean titre (GMT) for observations of respective vaccination experience as indicated on x‐axis and unvaccinated controls adjusted for age; ^†influenza vaccines before period A and in period A contain only seasonal H1N1 strains; ^‡influenza vaccines in periods B and C contain the H1N1pdm09 strain.

Figure 3

Seroconversion rates and distributions of HI antibody titres by time period and age group. Graph A shows seroconversion rate by time period and age group. Graphs B–J show, within each age group, seroconversion rates (in dots with error bars) by time period and HI distributions (in coloured bars) by sample number. Sample 2/3 refers to sample 3 when available and sample 2 when sample 3 is not available; sample 5/6 refers to sample 6 when available and sample 5 when sample 6 is not available. Observations from individuals following receipt of influenza vaccination in periods B and C were excluded.

Figure 4

Univariate and multivariate analyses of possible risk factors for serological evidence of infection with influenza H1N1pdm09. Graphs A–C show the odds ratios for seroconversion on univariate (blue diamonds) and multivariate (green circles) analyses with 95% error bars comparing the different variables to the base variable indicated in the y‐axis, for periods A–C, respectively, while graph D shows the same analysis but for all three periods combined. Multivariate models include all variables shown for the corresponding univariate analyses in the respective panels. Closed symbols and open symbols denote results significant and non‐significant at P < 0·05, respectively. Observations from individuals following receipt of influenza vaccination in periods B and C were excluded.