Assessment of Potential Bias in the National Immunization Survey (NIS) from the Increasing Noncoverage of Non-landline Telephone Households

Learning Objectives for this Presentation:
By the end of the presentation, participants will be able to understand the potential for bias in estimates from telephone surveys that may result from exclusion of non-landline telephone households.

Background:
Prevalence of children living in households with wireless-only telephone service is increasing; these children are not currently included in landline telephone surveys such as the NIS and NIS-Teen. The potential for bias in estimates from exclusion of these children should be evaluated.

Objectives:
To assess potential bias in NIS and NIS-Teen vaccination coverage rates due to exclusion of children from non-landline households, using data from the National Health Interview Survey (NHIS).

Methods:
Data from the 2006 National Health Interview Survey (NHIS) Sample Child Module are used to estimate rates of parent-reported influenza vaccinations among children aged 1-4 and 13-17 years, stratified by telephone status (landline and non-landline households). To assess bias, overall NHIS estimates of influenza vaccination rates are compared with estimated rates based only on children from landline households. Estimates from landline households are adjusted for exclusion of non-landline households.

Results:
In 2006, 79.5% of children aged 1-4 years and 87.9% of children aged 13-17 years lived in landline households. Overall influenza vaccination rates were 29.95% (±2.41%) among children aged 1-4 years, and 11.87% (±1.51%) among children aged 13-17 years. Among children aged 1-4 years, rates were 30.82% (30.79% adjusted) for those living in landline households, and 26.58% for those living in non-landline households. Among children aged 13-17 years, rates were 12.33% (landline; 12.4% adjusted) and 8.51% (non-landline). The parent-reported NIS-Teen rate was 13.6% among 13-17 year olds.

Conclusions:
While a direct comparison of provider-reported vaccination rates between children in landline vs. all households is needed, these results suggest statistically insignificant bias in the 2006 NIS and NIS-Teen estimates resulting from exclusion of children in non-landline households.