490. Estimates of Influenza Vaccine Effectiveness among Highly Immunized US Military Members and Beneficiaries: Utilization of Propensity Score Matching
Session: Poster Abstract Session: Respiratory Infections: Influenza
Thursday, October 8, 2015
Room: Poster Hall

Propensity score design (PSD) in conjunction with logistic regression is an innovative approach to estimate vaccine effectiveness (VE) in large, non-randomized, observational studies. Military personnel are at increased risk for influenza, and are therefore routinely vaccinated. We evaluated influenza VE among a cohort of military members and beneficiaries with laboratory-confirmed influenza from 2009-2014.


The Acute Respiratory Infection Consortium (ARIC) conducted an observational, longitudinal study of influenza-like illness (ILI) among otherwise healthy active duty military, retirees and their beneficiaries (birth-65y) at five US military hospitals. Nasopharyngeal swabs were collected and tested for influenza by RT-PCR. Vaccination history was obtained by self-report. Odds ratios (OR) were estimated using multivariable logistic regression after implementing a matched PSD. PSD was used to balance confounder distributions (year, site, age, race, sex, number of influenza vaccinations in the previous 5y, etc.) between unvaccinated individuals and individuals >6 mo who were vaccinated at least 14 days before illness onset. The resulting distribution of VE estimates provided an overall effect estimate and 95% confidence intervals (CIs). VE estimates were defined as (1-OR) x 100%.


Of the 1396 enrollees, 887 (64%) had been vaccinated. A total of 262 (19%) were positive for influenza (A, n=206; B, n=52; not typed, n=5). After creating the PSD matched set, the overall VE was 35.2% (95%CI: -8.1, 61.1), and differed by age, 42.3% (-18.2%, 71.8%) for patients <18y and 25.4% (-55.6%, 64.2%) for patients ≥18y. VE for influenza A was 27.8% (-26.8%, 58.8%), influenza B was 56.5% (-43.2%, 86.8%). VE for H1N1 and H3N2 was 13.4% (-148%, 69.8%) and 21.9% (-72.4%, 73.8%), respectively.


In a highly immunized military population, influenza vaccine was only moderately effective in preventing disease between 2009 and 2014. VE was higher, though still moderately effective, among individuals <18y.

Christian Coles, PhD1, Elizabeth Zell, MStat2, Wei-Ju Chen, PhD3, Amy Richardson, PhD4, John Arnold, MD5, Mary Fairchok, MD6, Christina Schofield, MD FACP, FIDSA7, Kristina J. St. Clair, DO, MTM&H8, Patrick Danaher, MD, FIDSA9, Michael Rajnik, MD10, Erin Mcdonough, BS11, Deepika Mor, MS3, Michelande Ridore, MS3, Timothy Burgess, MD, MPH10, Eugene Millar, PhD12 and Henry M. Jackson Foundation for the Advancement of Military Medicine, (1)Infection Disease Clinical Research Program, Uniformed Services University of the Health Sciences, Rockville, MD, (2)Stat-Epi Associates Inc., Ponte Vedra Beach, FL, (3)Infectious Disease Clinical Research Program, Uniformed Services University, Bethesda, MD, (4)National Institute of Allergy and Infectious Diseases (NIAID), Bethesda, MD, (5)Naval Medical Center San Diego, San Diego, CA, (6)Infectious Disease Clinical Research Program, Tacoma, WA, (7)Madigan Army Medical Center, Tacoma, WA, (8)Naval Medical Center Portsmouth, Portsmouth, VA, (9)San Antonio Military Health System, Fort Sam Houston, TX, (10)Walter Reed National Military Medical Center, Bethesda, MD, (11)Naval Health Research Center, San Diego, CA, (12)Infectious Disease Clinical Research Program, Rockville, MD


C. Coles, None

E. Zell, None

W. J. Chen, None

A. Richardson, None

J. Arnold, None

M. Fairchok, None

C. Schofield, None

K. J. St. Clair, None

P. Danaher, None

M. Rajnik, None

E. Mcdonough, None

D. Mor, None

M. Ridore, None

T. Burgess, None

E. Millar, None

Findings in the abstracts are embargoed until 12:01 a.m. PDT, Wednesday Oct. 7th with the exception of research findings presented at the IDWeek press conferences.