Descrição
The EuroEVA project is the Portuguese component of the multicentre I-MOVE study and aims to obtain estimates of the seasonal and pandemic vaccine effectiveness during and after the influenza season.
Since the 2008/2009 influenza season Portugal, along with other European countries, has implemented a common protocol using a case-control study design, where influenza-like illness cases which are laboratory confirmed as influenza (ILI+) are compared to a control group consisting of ILI patients which test negative for influenza (ILI-) (Case-control Test negative design).
The results presented in this report relate to the EuroEVA 2011-2012 season and aim to estimate the seasonal influenza vaccine effectiveness for the age group 60+ years and in all age groups, using two approaches: Case-control Test negative design and Screening Method.
Materials and Methods
Test Negative Design
ILI cases were identified among patients that presented ILI symptoms to a participating EuroEVA General Medical Practitioner (GP). On a weekly basis, each GP systematically selected ILI patients (two per week with less than 60 years and all ILI patients with 60 years and more) using the EU ILI case definition. Data on potential confounding factors and effect modifiers was collected using a standardized questionnaire which included information on socio-demographic variables (age, gender, education and co-inhabitants), previous (2010-11) influenza vaccination, chronic conditions and related hospitalizations, current smoking habits, belonging to GP list and number of consultations in the previous year. An ILI patient was considered vaccinated if he/she had received one dose of the 2011/2012 trivalent influenza vaccine at least 14 days prior to onset of symptoms. VE was estimated as one minus the odds ratio of being vaccinated in cases versus controls adjusted for confounders by logistic regression. Potential confounders were investigated and included if: they changed crude OR estimate in at least 10% after adjustment by the Mantel-Haenszel method, were associated both to being a case (in the absence of the exposure factor) and to the seasonal vaccination.
Screening Method
ILI cases and ILI laboratory confirmed influenza cases were recruited in the context of the National I-MOVE case-control study (EuroEVA). Vaccine coverage in the population was obtained from a sample of 1074 households stratified by region (homogeneous allocation) selected from a dual sample frame: random digit dialling mobile and landline phones (ECOS sample). Relevant information was collected by CATI (Computer Assisted Telephone interview) – one respondent by household (proxy for the rest of the household members). VE was estimated by comparing the proportion of vaccinated cases to the vaccine coverage in the source population, using the Orenstein formula and the Farrington method to adjust for age group and target group for vaccination.
Results
In Portugal, a later beginning of the 2011/2012 influenza epidemic was observed, starting in week 4/2012 and ending at week 12/2012. In this season both influenza B and A(H3) virus types were circulating, with predominance of the later.
Test negative design
From the 59 GP’s that accepted to participate in the study, 35 effectively participated in the study by selecting patients (which corresponds to a 59% participation rate). After excluding 79 ILI cases (for not adhering to the inclusion criteria) the final sample consisted on 273 ILI patients. Of the 134 cases which tested positive for influenza, 98.5% were positive for influenza A(H3) and the remaining for type B virus. The control group, consisting of 139 ILI patients who tested negative for influenza, was statistically different (p<0.05) from the ILI+ group in the following variables:
• Clinical signs and symptoms: cough (higher in cases than in controls: 95.5% vs. 89.2%, respectively) and sore throat (more frequent in controls, 89.2%, than in cases ,75.4%);
• Age: controls were older than cases (median age in controls was 52 yrs vs. 39 yrs in cases);
• Any chronic disease: the prevalence of at least one chronic condition relevant for influenza vaccination was higher in controls (41.7% vs 29.1%);
• Seasonal vaccine in 2010-11: controls were more often vaccinated against influenza in the last season than cases (30.2% vs. 14.4%);
• Co-habitants: the median number of co-habitants was higher in cases (3 vs. 2).
Considering all population, vaccine coverage (VC) in controls was 27.5% statistically higher than in cases (VC=13.4%). Similar results were obtained for the sub-group target for vaccination by the National Health Authorities (VC cases =24.6% and VC controls=46.1%, p=0.010). These results indicate that crude VE estimates was 59.2% (95% CI: 21.1%-79.4%) in the general population and 61.8% (95% CI: 15.5% ;83.1%) in the target group for vaccination. After adjustment for co-inhabitants and month of onset of illness, VE adjusted estimates were 48.8% (95% CI: 0.0% ; 73.8%) and 51.6% (95% CI:-6.2%-77.9%) for the general population and for the target group, respectively.
Screening Method
The ECOS telephone survey was conducted during April 2012, and information was obtained from a total of 2395 individuals. According to the results, individuals were vaccinated from October 2011 through January 2012, estimating a 16.4% (95% CI: 13.6-19.6) vaccine coverage (VC) in the population. In the 60+ yrs age group, the VC was 37.3% (95% CI: 30.6-44.4) and for the individuals with chronic condition was 28.0% (95% CI: 23.0-33.7). The crude VE estimated with the Screening method for ILI+ as the outcome was 27.0% (95% CI: -19.9- 55.6) and -32.4% (95% CI: -77.7- 1.3) for ILI. Adjusted VE estimates varied from -90.4% (95% CI: -277.1- 3.9) (60+ yrs) to 6.1% (95% CI: -56.0- 43.4) (0-60 yrs) considering ILI as the outcome and from -58.6% (95% CI: -195.3- 14.8) to 56.9 (95% CI: -35.2- 86.3) for ILI positive outcomes (none were statistically significant).
Conclusions
Given the 3 years experience in conducting this study, logistical and implementation aspects were straightforward.
The 2011-2012 season adjusted VE estimates were similar for the general population (48.8%) and for the target group (51.6%), although not statistical significant. When compared to the previous season, VE point estimate for the general population was lower (VE=58% in 2010-11), although the CI overlap.
The population studied this year was older than in the last season. The time between onset of symptoms and swabbing, was also different with marginally, non significant, differences between cases and controls.
; Guiomar, Raquel 
