Background: Since 2004, the uptake of seasonal influenza vaccines in Latin America and the Caribbean has markedly increased. However, vaccine effectiveness (VE) is not routinely measured in the region. We assessed the feasibility of using routine surveillance data collected by sentinel hospitals to estimate influenza VE during 2012 against laboratory-confirmed influenza hospitalizations in Costa-Rica, El Salvador, Honduras and Panama. We explored the completeness of variables needed for VE estimation.
Methods: We conducted the pilot case-control study at 23 severe acute respiratory infections (SARI) surveillance hospitals. Participant inclusion criteria included children 6 months-11 years and adults ≥60 years targeted for vaccination and hospitalized for SARI during January-December 2012. We abstracted information needed to estimate target group specific VE (i.e., date of illness onset and specimen collection, preexisting medical conditions, 2012 and 2011 vaccination status and date, and pneumococcal vaccination status for children and adults) from SARI case-reports and for children ≤9 years, inquired about the number of annual vaccine doses given. A case was defined as an influenza virus positive by RT-PCR in a person with SARI, while controls were RT-PCR negative. We recruited 3 controls per case from the same age group and month of onset of symptoms.
Results: We identified 1,186 SARI case-patients (342 influenza cases; 849 influenza-negative controls), of which 994 (84 %) had all the information on key variables sought. In 893 (75 %) SARI case-patients, the vaccination status field was missing in the SARI case-report forms and had to be completed using national vaccination registers (36 %), vaccination cards (30 %), or other sources (34 %). After applying exclusion criteria for VE analyses, 541 (46 %) SARI case-patients with variables necessary for the group-specific VE analyses were selected (87 cases, 236 controls among children; 64 cases, 154 controls among older adults) and were insufficient to provide precise regional estimates (39 % for children and 25 % for adults of minimum sample size needed).
Conclusions: Sentinel surveillance networks in middle income countries, such as some Latin American and Caribbean countries, could provide a simple and timely platform to estimate regional influenza VE annually provided SARI forms collect all necessary information.