The study consisted of patients seen at NUH during May–November 2009 as emergency admissions, outpatients, or inpatients whose nasopharyngeal swabs submitted for routine diagnostic testing were positive for seasonal influenza virus A (H1 and H3) or pandemic influenza A virus (H1N1) 2009. From samples taken before treatment was begun, we identified 578 patients with pandemic (H1N1) 2009 and 88 patients with seasonal influenza (11 H1 and 77 H3). Clinical characteristics of some of these patients have been described elsewhere (2). Local ethics approval (ref. no. B/09/360) was granted for this study.

Age, sex, and clinical information (i.e., days after onset of symptoms, comorbidities, clinical severity) were obtained from patient records. Comorbidities were defined as >1 of the conditions listed in Table 1. Clinical severity was defined as follows: mild, patients well enough to be treated as outpatients; moderate, patients ill enough to warrant hospital admission; severe, hospitalized patients who died or who required intensive or high-dependency care. In-house quantitative assays (Technical Appendix) were performed on archived samples previously tested as positive for pandemic (H1N1) 2009 and reported elsewhere (6).

Figure 1

Figure 1. Viral loads (in RNA copies/mL) in patients with pandemic (H1N1) 2009 (NP) and seasonal H1 and H3 (MP) influenza at time patient sought hospital care against days after symptom onset. Vertical...

Viral loads of hemagglutinin (HA) and nucleoprotein (NP) for pandemic (H1N1) 2009 ranged from 10² to 10⁹ RNA copies/mL of virus transport medium (mean 10⁵–10⁷ RNA copies/mL). Seasonal influenza viral loads ranged from 10³ to 10¹⁰ RNA copies/mL (mean 10⁶–10⁸ RNA copies/mL for seasonal influenza subtype H3 and mean 10⁵ to 10⁷ RNA copies/mL for seasonal influenza H1). Viral loads decreased with time after onset of symptoms from date the patient sought care at NUH in patients with pandemic or seasonal influenza (Figure 1).

Because of the small number of patients with seasonal influenza H1, further analysis for seasonal influenza was limited to H3. Patients infected with pandemic (H1N1) 2009, compared with those having seasonal influenza H3, were younger (p<0.0001), and a higher proportion had comorbidities (p = 0.0068; Table 1).

Figure 2

Figure 2. A) Profile plot and multivariate comparisons of the estimated nucleoprotein viral loads of pandemic (H1N1) 2009, by patient age group, against days from symptom onset in the final multiple analysis of...

For the 578 pandemic influenza cases, the multiple analysis of variance showed that viral loads were associated with number of days after symptom onset from date of presentation (p<0.0001) and with age (p = 0.0112) (Figure 2, panel A; Table 2). For the 77 seasonal influenza H3 cases, the analysis of variance showed that days after onset of symptoms from date of presentation (p = 0.0223) and presence of any comorbidities (p = 0.0249) significantly affected viral loads (Table 2). Viral loads for seasonal influenza were lower in patients with than without comorbidities (Figure 2, panel B).

One of our most striking findings was that the mean viral loads of patients visiting NUH were ≈1 log₁₀ higher for seasonal than for pandemic influenza (Figure 1). This difference persisted even after we adjusted for age. Another study demonstrated that within the first 3 days after symptom onset, historical mean viral loads of seasonal influenza exceed those of the contemporary pandemic virus by 1–2 log₁₀ (3). However, a limitation of that study is its use of viral load data for seasonal influenza that was historical rather than obtained contemporaneously with the data for pandemic (H1N1) 2009.

Approximately 30%–50% of influenza case-patients may be asymptomatic (7), and although the correlation between viral load and clinical symptoms is not well established, a viral load threshold may exist below which most persons have no clinical symptoms (although individual variation will always exist). Our analysis suggests that if such a threshold exists, it is lower for novel than for seasonal influenza viruses. For a direct virus-mediated pathologic process, this hypothesis may be understandable, given the lower prevalence of preexisting (and therefore potentially partially protective) cross-reactive immunity for this novel virus (8–10).

Viral loads for both pandemic (H1N1) 2009 and seasonal influenza tend to decrease with time after symptom onset (Figure 1). Larger studies are needed to confirm the more rapid decline of seasonal influenza H1 than of H3 viral loads. In addition, younger age groups had significantly higher viral loads for pandemic (H1N1) 2009 (Figure 2, panel A), which may not be surprising given that this Southeast Asian population appears to have little or no preexisting specific or cross-reacting antibodies to this novel virus (9,10).

Two findings are perhaps the most surprising of this analysis. First, we found no significant correlation between pandemic (H1N1) 2009 or seasonal influenza viral loads and clinical severity of illness (Table 2). Second, pandemic (H1N1) 2009 viral loads in infected patients with and without preexisting comorbidities did not differ significantly, although a significant difference was found for seasonal influenza (Figure 2, panel B; Table 2). We offer some possible explanation for these findings but note that these influenza viral loads have been measured in respiratory samples. These samples are peripheral types of specimens that may not necessarily directly affect, or be directly affected by, many of the preexisting comorbidities that involve nonrespiratory systems, unless their management involves, for example, some sort of immunosuppressive therapy.

A main limitation of this study is that these viral load measurements were performed on only 1 acute diagnostic sample from each patient at admission before treatment with oseltamivir; therefore, determining how these viral loads would have changed later during the natural course of the infection was not possible. Also, some of the patient categories (Table 1, Table 2) contained relatively few patients, e.g., the relatively low number of severe cases (Table 1, Table 2), which may have limited the statistical significance of some correlations. Finally, although influenza viral loads in various types of respiratory samples are now often reported (3,6), these are heterogeneous, peripheral samples, and such viral loads may vary considerably in the same patient during a single day, depending on individual host immune responses.

If human illness caused by influenza virus infections is mediated by host immune responses (4,5), then a more vigorous, primary immune response in the immunologically naive, otherwise healthy younger population against the pandemic (H1N1) 2009 virus may also contribute to the degree of clinical illness. The interplay between a direct viral pathologic process and a host immune-mediated pathologic process is probably unique to each person. Some recent studies investigating cytokine responses in persons with acute pandemic (H1N1) 2009 infections had contrasting findings (11–13), although postmortem investigations of some fatal cases of pandemic (H1N1) 2009 infection found substantial inflammation, which supports an immune-mediated pathologic process for at least in these cases (14). Similarly, for the more well-established seasonal H3 influenza (to which most persons have had many years of exposure) more well-established, robust, yet sufficiently individually different patterns of homologous and heterologous immune responses may contribute more (compared with similar responses to pandemic [H1N1] 2009) to the different degrees of clinical illness in infected persons with different combinations of comorbidities.