Skowronski DM, Janjua NZ, De Serres G, Hottes TS, Dickinson JA, Crowcroft NS, Petric M, Fonseca K, Drews SJ, Rebbapragada A, Charest H, Kwindt TL, Gardy J, Mak A, Chan T, Tang P, Winter AL, Douville-Fradet M, Bontovics E, Bolotin S, Patrick DM, Brunham RC, Gubbay JB, Bastien N, Li Y, De Serres G, Boulianne N, Gilca R, Ouhoummane N, Boivin G, Fortin E, Douville-Fradet M, Landry M, Crowcroft NS, Janjua NZ, Skowronski DM, Crowcroft NS, Rosella LC, Sethi P, Willison DJ, Johnson I, McGeer A, Allison JC, Savage R, De Serres G, Janjua NZ, Skowronski DM, De Serres G, Boulianne N, Rouleau I, Charest H, Hamelin ME, Boivin G, Quach C, Flamand L, Couillard M, Bourgault AM, Skowronski DM.
Source
British Columbia Centre for Disease Control (BCCDC), Vancouver, British Columbia, Canada. danuta.skowronski@bccdc.ca
Abstract
BACKGROUND:
In late spring 2009, concern was raised in Canada that prior vaccination with the 2008-09 trivalent inactivated influenza vaccine (TIV) was associated with increased risk of pandemic influenza A (H1N1) (pH1N1) illness. Several epidemiologic investigations were conducted through the summer to assess this putative association.
METHODS AND FINDINGS:
STUDIES INCLUDED: (1) test-negative case-control design based on Canada's sentinel vaccine effectiveness monitoring system in British Columbia, Alberta, Ontario, and Quebec; (2) conventional case-control design using population controls in Quebec; (3) test-negative case-control design in Ontario; and (4) prospective household transmission (cohort) study in Quebec. Logistic regression was used to estimate odds ratios for TIV effect on community- or hospital-based laboratory-confirmed seasonal or pH1N1 influenza cases compared to controls with restriction, stratification, and adjustment for covariates including combinations of age, sex, comorbidity, timeliness of medical visit, prior physician visits, and/or health care worker (HCW) status. For the prospective study risk ratios were computed. Based on the sentinel study of 672 cases and 857 controls, 2008-09 TIV was associated with statistically significant protection against seasonal influenza (odds ratio 0.44, 95% CI 0.33-0.59). In contrast, estimates from the sentinel and three other observational studies, involving a total of 1,226 laboratory-confirmed pH1N1 cases and 1,505 controls, indicated that prior receipt of 2008-09 TIV was associated with increased risk of medically attended pH1N1 illness during the spring-summer 2009, with estimated risk or odds ratios ranging from 1.4 to 2.5. Risk of pH1N1 hospitalization was not further increased among vaccinated people when comparing hospitalized to community cases.
CONCLUSIONS:
Prior receipt of 2008-09 TIV was associated with increased risk of medically attended pH1N1 illness during the spring-summer 2009 in Canada. The occurrence of bias (selection, information) or confounding cannot be ruled out. Further experimental and epidemiological assessment is warranted. Possible biological mechanisms and immunoepidemiologic implications are considered.