Predictive Modeling of Flu Vaccinations in the District of Columbia: An Evaluation of Spatial Trends in Self-Reported Flu Vaccinations, 2001-2004

Introduction: About 36,000 people in the US die each year from influenza-related complications. Influenza is a viral infection of the lungs and airways that is also known as “the flu.” Being able to anticipate the expected number of flu vaccinations in the District could enhance the decision-making capability of public health surveillance.

Objective: This study is aimed at exploring spatial trends in self-reported flu vaccinations and using predictive modeling to determine expected flu vaccinations.

Methodology: We employed exploratory data analysis and kriging to evaluate flu vaccination trends and generate risk maps for expected flu vaccinations for ages 18-64 years and ages >64 years in the District. Kriging is a statistical interpolation method that uses data from a single data type (single attribute) to predict values of that same data type at unsampled locations.

Results: Generally, the highest distribution of flu vaccination rates for ages 18-64 (@22%-30%) and ages >64 years (@42%-56%) were in the Northwest, and the lowest rates for ages 18-64 years (@12%-16%) and ages >64 years (@42%) were in the Southeast. The trend analysis showed a slight east-west and north-south increasing trend in flu vaccinations. For ages 18-64 years, initial analyses showed the highest flu vaccinations (@ 29.6%-37.5%) in the northwest (Figure1).

Conclusions: This study will show that flu vaccination risk maps can be generated for the District of Columbia. This will enable better anticipation of flu vaccination needs and enhance intervention measures.