Longitudinal serum quadruples from Hong Kong suggest a key role for middle-aged adults in the transmission of influenza A during the postpandemic period

Abstract

BACKGROUND: A number of populations have experienced genuine “follow-up” waves of pandemic H1N1 influenza A (pH1N1) infection (eg, winter waves of 2010-2011 in the United Kingdom and Hong Kong and the 2011-2012 winter wave in Mexico). An upward age shift has been observed in laboratory-confirmed clinical cases in these follow-up waves compared with earlier pandemic waves. Continued circulation of seasonal H3N2 influenza A (sH3N2) in the absence of seasonal H1N1 suggests that sH3N2 and pH1N1 will co-circulate for the foreseeable future. METHODS: We recruited randomly from households in Hong Kong and asked participants to attend a central clinic to answer a questionnaire and provide a blood sample. We conducted 4 rounds: round 1, July 4, 2009-September 19, 2009; round 2, November 11, 2009-February 6, 2010; round 3, December 13, 2010-March 19, 2011; and round 4, August 24, 2011-December 17, 2011. The timing of our rounds relative to waves of clinical cases was assessed using monthly subtype-specific test results from the Center for Health Protection, Hong Kong. Everyone who provided a sample in round 1, who did not explicitly refuse to be contacted again, was invited to participate in subsequent rounds. Serum samples from individuals who participated in all 4 rounds were tested in quadruples using standard hemagglutinin inhibitor assays for pH1N1 and sH3N2. An individual was assumed to have been infected between pairs of waves if the titre for the second sample was ≥4-fold greater than that of the first sample. RESULTS: We obtained quadruples of serum from 445 individuals aged 4-79 years. In general, our rounds of recruitment did not neatly bracket waves of clinical cases: round 1 was taken toward the start of the initial prolonged wave of pH1N1 and round 2 was close to the end of that period. Rounds 2 and 3 did bracket a clear clinical wave of sH3N2. Although rounds 2 and 4 bracketed the follow-up wave of pH1N1, the interval between rounds 2 and 4 was long and contained some infections from the end of the initial wave. Between rounds 1 and 2, 45 of 445 participants were infected with pH1N1. Of these 45, 33% (95% CI, 22%-49%) were aged ≤18 years (children), 24% (95% CI, 15%-40%) were aged 19-44 years (young adults), 22% (95% CI, 13%-37%) were aged 45-64 years (middle-aged adults), and 20% (95% CI, 11%-36%) were ≥65 years (older adults). Between rounds 2 and 4, 123 of 445 participants were infected with pH1N1. Of these 123, 13% (95% CI, 8%-20%) were children, 23% (95% CI, 16%-31%) were young adults, 52% (95% CI, 44%-62%) were middle-aged adults, and 11% (95% CI, 7%-18%) were older adults. Between rounds 2 and 3, 51 of 445 participants were infected with sH3N2. Of these, 13% (95% CI, 7%-26%) were children, 16% (95% CI, 8%-29%) were young adults, 49% (95% CI, 37%-63%) were middle-aged adults, and 22% (95% CI, 13%-35%) were older adults. CONCLUSIONS: There was a substantial shift in infections away from children toward middle-aged adults between the initial and follow-up waves of pH1N1 in Hong Kong. The age distribution of infections in the follow-up wave of pH1N1 was similar to that of sH3N2. Middle-aged adults may be more important for influenza A transmission during nonpandemic periods than was assumed previously. Here, because of the timing of rounds, we focused on the distribution of ages of those infected, rather than on between-round serologic attack rates. However, these data could be combined with catalytic and mechanistic models of transmission to produce accurate estimates of age-specific infection rates for individual waves.