Estimating influenza vaccine effectiveness using data routinely available in electronic primary care records

Abstract

Background: To support timely, annual estimation of influenza vaccine effectiveness (VE), we explored the use of automated data extraction from general practice records to estimate VE over four consecutive southern hemisphere influenza seasons.

Methods: A software tool installed at 130 practices in Western Australia identified all outpatients tested for influenza by polymerase-chain-reaction (PCR) during annual influenza seasons occurring 2012-2015. Laboratory test results were collated with any existing record of influenza vaccine administered in the same year; limited patient demographic and clinical information was also collected. A case test-negative control analysis compared the odds of seasonal influenza vaccination between patients positive or negative for influenza by PCR with VE = 1 - the odds ratio.

Results: A total of 7270 influenza PCR test results were identified of which 1907 (26.2%) were positive; 9.4% of patients with a positive result had received contemporaneous influenza vaccination ≥14 days prior to specimen collection, compared to 17.9% of those with a negative result. Overall VE was 52% (95% CI, 43-60%); annual VE estimates ranged from 46% (95% CI, 22-63%) in 2012 to 60% (95% CI, 41-73%) in 2014.

Conclusion: Electronic records routinely maintained by general practice provide a promising opportunity for estimating annual influenza VE in a timely and resource-efficient manner.

Keywords: Influenza; Primary health care; Surveillance; Test-negative design; Vaccine effectiveness; Vaccines and immunisation.