Evidence on cross-protection to 2009 pandemic H1N1 influenza in subjects immunised with seasonal trivalent inactivated vaccine (TIV) is controversial. We performed a retrospective cohort study on teachers at a senior school (Gold Coast, Queensland, Australia) to assess the cross-protection against 2009 pandemic H1N1 influenza due to seasonal vaccination and explored the effect of school closure on absenteeism (sick leave) of school teachers. We performed a similar study for the 2007 influenza season.
Absenteeism data were obtained for all teaching staff from May to October in 2007 and 2009 on a monthly basis. There were two potential influenza control methods that we assessed, a formal intervention, namely staff vaccination against influenza and routine school closure (for the mid-year holiday). As a proxy for influenza activity, we measured absenteeism due to illness in teachers as:
Analysis was performed comparing absenteeism in staff given influenza vaccine with that in unvaccinated staff (through Chi-squared test). Relative risk (RR) with 95% confidence interval (CI) was calculated;1 vaccine effectiveness (VE) was obtained by the formula VE= (1-RR) ×100%.
In March 2007, 51% (50/98) of the teachers were vaccinated with TIV; similar uptake was achieved in March 2009, 58% (58/99). The school was closed for three-week routine holidays in mid-2007 and 2009 (see figures). Before the 2007 school holiday, there was no significant difference in terms of proportion of school days absent between vaccinated and non-vaccinated teachers, 1.7% vs 0.9% (Figure 1). In July, after the school holiday, those vaccinated had much lower absenteeism than those non-vaccinated (0.7% vs 3.6%, p<0.001); RR was 0.18 (95% CI: 0.06–0.53). Community influenza rates were highest in mid-August (Source: personal communication, Han Wang, author). However, incidence peaked earlier (in July) in the school probably due to high transmission rates from crowding. Combining the data for July and August, there remained a significant difference with absentee rates of 2.4% (1.8–3.0%) and 4.0% (3.1–5.1%), respectively. The confidence intervals of absentee rates in other months overlapped closely. The school holiday did not obviously reduce the absentee rate in 2007.
In 2009, there was a much higher background rate of influenza-confirmed illness which peaked in the second half of July (source: personal communication, Han Wang). Compared with June (before school holiday), higher absentee rates were witnessed in both the vaccinated (5.9%) and non-vaccinated groups (6.8%) in July, and the confidence intervals for vaccinated and non-vaccinated teachers were widely overlapping (Figure 2). Higher overall absentee rates in July 2009 were reported compared with July in 2007. For all months, sick leave was not significantly different between vaccinated and unvaccinated teachers in 2009 and combining July and August data (peak influenza period) also showed no difference.
Our study shows that in this school where TIV is routinely offered to teachers, there was no evidence to support the concept that vaccination with 2009 TIV induced cross-protection to pandemic H1N1 influenza, as measured by sick leave, whereas this was clearly evident in 2007 when the annual vaccine was well-matched to seasonal strains.
Our results are consistent with findings relating to 2009 season from the US2 and Australia3 but contrasts with data from Mexico.4 In the US study, no cross-reactive antibody to the pandemic influenza A (H1N1) was found in adults (over 18 years), post administration of 2008/09 vaccine. The Mexican study showed a remarkable level of protection against pandemic influenza from TIV. However, it was likely to be affected by selection bias, e.g. because 65% of the controls had chronic medical conditions compared with 25% of cases. Canadian data from a series of epidemiological studies, based on a sentinel monitoring system of influenza effectiveness, reported a counterproductive effect of TIV.5 Although the authors looked thoroughly for potential biases, it is still controversial whether the TIV increased the risk of 2009 pandemic illness due to lack of randomisation and uncontrolled confounders which occur in observational studies, e.g. socioeconomic status was not determined.
Our data does not support a protective effect of 2009 TIV against pandemic H1N1 influenza but a systematic review of published studies is underway (personal communication, J. Kevin Yin). However, we showed that seasonal vaccination of teachers in 2007 appeared beneficial, probably because there was a close match with the seasonal strain.