332. Public health implications of a mathematical method for analyzing observed household vaccination behavior
Session: Poster Abstract Session: Challenges in Vaccinology and Vaccine Exploration
Friday, October 21, 2011
Room: Poster Hall B1
Background: Age-specific individual vaccination rates have been used with apparent success in outbreak modeling and public health planning, e.g. Chao, D. et al. AmJEpi(2011); 173:1121-1130. No known population models accommodate household clustering of vaccination.

Methods: A simple method for calculating intrahousehold correlation of influenza vaccination is presented. Two parameters, mu and rho, are used to characterize household influenza vaccination patterns. Mu is defined as the proportion of vaccinated individuals within multiperson households in a population of interest; rho is the intrahousehold correlation of vaccination within the same population. 

Results: The resulting characterization of household vaccination behavior, incorporating an overdispersed betabinomial distributional model, matches actual self-reported household vaccination better than binomial or underdispersed hypergeometric models. Selected results from the 2009 Twin Cities Area Survey (household self-report for all HH members), and selected data from 2008 NHIS 2 person households are:

 

Rho

Mu

ALL HOUSEHOLDS

0.49

0.54

HH w/ children

0.48

0.52

HH w/o children

0.41

0.60

All HH members w/ health insurance coverage

0.50

0.55

Any HH member w/o health insurance

0.30

0.41

Households with elderly/no children

0.37

0.79

NHIS2008: 1 adult, 1 child (2 person HH)

0.33

0.24

Conclusion: Analysis based on a betabinomial distribution is feasible and produces useful estimates of the distribution of influenza vaccination within multiperson households. Intrahousehold correlation levels are generally moderate and clearly non-negligible. Models that do not accommodate intrahousehold correlation of vaccination underpredict the distributional tails (e.g. fully vaccinated or unvaccinated households). Household clustering of vaccination is predicted to result in exposure heterogeneities that are of evident importance for protecting vulnerable patients. Accomodating intrahousehold correlation of vaccination or other covariates in future modeling efforts could improve public health planning, disease prevention efforts, and research on understanding transmission dynamics for influenza and other contagious diseases.


Subject Category: I. Adult and Pediatric Vaccines

Mark Andrew Robien, MD, MPH, Consultant, Minneapolis, MN

Disclosures:

M. A. Robien, None

Findings in the abstracts are embargoed until 12:01 a.m. EST Thursday, Oct. 20 with the exception of research findings presented at IDSA press conferences.