Differences in patient age distribution between influenza A subtypes

Abstract

Since the spring of 1977, two subtypes of influenza A virus (H3N2 and H1N1) have been seasonally infecting the human population. In this work we study the distribution of patient ages within the populations that exhibit the symptomatic disease caused by each of the different subtypes of seasonal influenza viruses. When the publicly available extensive information is pooled across multiple geographical locations and seasons, striking differences emerge between these subtypes. We report that the symptomatic flu due to H1N1 is distributed mainly in a younger population relative to H3N2. (The median age of the H3N2 patients is 23 years while H1N1 patients are 9 years old.) These distinct characteristic spectra of age groups, possibly carried over from previous pandemics, are consistent with previous reports from various regional population studies and also findings on the evolutionary dynamics of each subtype. Moreover, they are relevant to age-related risk assessments, modeling of epidemiological networks for specific age groups, and age-specific vaccine design. Recently, a novel H1N1 virus has spread around the world. Preliminary reports suggest that this new strain causes symptomatic disease in the younger population in a similar fashion to the seasonal H1N1 strains.

Figure 1. Empirical cumulative distribution of ages for patietns with H1N1 (blue) and H3N2 (red) in New York State during the 2006–2007 and 2007–2008 influenza seasons.

The significantly low probabilities computed via Mann-Whitney (P(MN)) and Kolmogorov-Smirnov (P(KS)) tests indicate a remarkable dissimilarity between the distributions.

Figure 2. Empirical cumulative distribution of ages for patients with H1N1 (blue) and H3N2 (red), from the NCBI dataset, in the United States, between 1995 and 2008 (left) and Oceania, between 2000 and 2007 (right).

Complementary to the results from New York State (Fig. 1), the low probabilities computed via Mann-Whitney (P(MN)) and Kolmogorov-Smirnov (P(KS)) tests show a significant difference between the distributions, which is spatially and temporally consistent.

Figure 3. Empirical cumulative distribution of ages for patients with H1N1 (blue) and H3N2 (red), from the NCBI dataset in United States and Oceania combined, between 2000 and 2007.

The low probabilities computed via Mann-Whitney (P(MN)) and Kolmogorov-Smirnov (P(KS)) further confirm the results shown in Fig. 2.

Figure 4. Empirical cumulative distribution of ages for patients with H1N1 (blue) and H3N2 (red), from the NCBI dataset, in United States during the influenza seasons of 2006–2007 (left) and 2007–2008 (right).

The significantly low probabilities computed via Mann-Whitney (P(MN)) and Kolmogorov-Smirnov (P(KS)) tests during separate influenza seasons show the consistency in our results among sub-portions of the data and refute the possibility that the previous statistical results are due to a unique season.

Figure 5. Empirical cumulative distribution of birth year for patients with H1N1 (blue) and H3N2 (red), from the NCBI dataset, in the United States, between 1995 and 2008.

The significant statistical dissimilarity between the distribution for the year of birth (P(MW) = 2.04E-24 and P(KS) = 1.13E-35) hints to an existing immunity against one of the subtypes in different age groups, possibly carried over from a previous pandemic.