Effectiveness of H1N1 Influenza Vaccines in Manitoba, Canada

Testing for influenza is largely discretionary because most patients presenting with influenza-like illness (ILI) could be effectively and safely managed even when the diagnosis is not confirmed. So, it is unclear whether reduction in the risk of laboratory-confirmed influenza or mild ILI among vaccinated individuals is of much clinical significance. In contrast, hospitalization due to influenza or pneumonia is usually not as discretionary, and from both clinical and public health perspectives avoiding hospitalization is more important and relevant objective for influenza control than the mere reduction of laboratory-confirmed cases.

Thus far there have been few observational studies that examined the effectiveness of the H1N1 vaccine against hospitalization, and most of them had significant limitations including small sample sizes, information limited to certain age groups (e.g., children) and the potential for selection and recall bias. Using data from a Scottish general practice sentinel surveillance network, Simpson et al reported that the adjuvanted GSK vaccine was 95% (76-100%) effective against laboratory-confirmed H1N1 infections, and 100% (∞-100%) effective against hospitalization for influenza and pneumonia. Unexpectedly, the vaccine was equally effective among 'high-risk' groups. Small number of cases limited the precision of these estimates; the above-mentioned estimate for vaccine effectiveness (VE) against hospitalization is based on a one case only, which is the reason for the wide confidence interval that is compatible with any VE estimate (∞-100%). A hospital-based case-control study from Spain showed that use of several pandemic vaccines was associated with 90% effectiveness in preventing laboratory-confirmed H1N1 infections among hospitalized patients. That study was not population-based and was limited by small number of fully-vaccinated subjects (only 7 subjects received a pandemic vaccine ≥ 7 days before symptom onset).

Two Canadian studies examined the effectiveness of the GSK adjuvanted H1N1 vaccine: one was very small (28 cases) and was limited to children under 10 years. The study by Skowronski et al. that used data from the Canadian VE sentinel network did not examine VE against hospitalization and it, too, was very small.

In the Canadian province of Manitoba, mass immunization against pandemic H1N1 commenced October 26th 2010 using primarily large scale vaccination clinics led by public health teams and lasted approximately 8 weeks. Initially, the Canadian-manufactured Arepanrix®, an AS03-adjuvanted split virion pandemic H1N1 vaccine (GlaxoSmithKline [GSK]), was used to vaccinate adults and children over 6 months of age. Later on, two nonadjuvanted vaccines, from CSL Limited and GSK, were offered to pregnant women and children over 10 years of age. The seasonal trivalent inactivated influenza vaccines (TIV)-Fluviral® (GSK) and Vaxigrip® (Sanofi Pasteur) - were less frequently administered.

All vaccines were offered free of charge, but limited vaccine supply at campaign start necessitated the development of priority groups for early vaccination. The initial priority group in Manitoba included health care workers, Aboriginal persons, pregnant women, 6-60 months-old children, individuals under 65 years with chronic medical conditions, immunocompromised individuals and residents of remote communities. On November 18,2010, the pandemic H1N1 vaccines were made available to the whole population.

This study evaluates the effectiveness of the GSK adjuvanted and non-adjuvanted pH1N1 vaccines used in Manitoba in preventing hospitalization and severe illness. The advantages of the proposed study over the existing literature are related to the availability of high-quality population-based health administrative databases that cover a well-defined population. The whole population of Manitoba is eligible and available for inclusion in the study, so selection bias is not a concern. Ascertainment of cases is virtually complete because all admitted influenza or pneumonia cases in the province are captured by the Hospital Abstract database. The availability of detailed histories of vaccination, through the unique Manitoba Immunization Registry eliminates recall bias, reduces vaccine use measurement errors (e.g., due to patient confusion about what vaccines were received) and permits detailed assessments of the effect of the timing of vaccination on the risk of hospitalization. These detailed analyses will be facilitated by the large sample size (40% of Manitoba's 1.2 million population were vaccinated).