A Multiyear Model of Influenza Vaccination in the United States

Arnold Kamis, Yuji Zhang, Tamara KamisInternational Journal of Environmental Research and Public Health 2017 · DOI

TL;DR

This paper develops a multiyear logistic regression model to predict influenza vaccination uptake in US adults using data from 2012-2014. The model identifies key demographic and behavioral predictors, achieving a 67% correct classification rate, and suggests that understanding cognitive biases could be crucial for improving vaccination rates.

What Problem Does This Paper Address?

Despite the clear public health benefits of influenza vaccination, a significant portion of the adult population in the United States remains unvaccinated. This vaccine avoidance contributes to increased incidence of preventable diseases and can lead to severe complications, hospitalization, and even death, particularly among vulnerable groups. The challenge lies in understanding the complex interplay of factors that influence an individual's decision to get vaccinated and developing effective strategies to increase vaccination rates.

Methods

The researchers employed a standard logistic regression approach to build a predictive model. They utilized a stepwise forward selection method to identify the most significant predictors from a large set of initial variables. The model was developed using data from 2013, with a portion held out for validation and testing. To ensure the model's robustness and generalizability, it was then tested on data from 2012 and 2014. The final model was selected based on maximizing predictive power (adjusted R-squared), ensuring statistical significance of predictors, and minimizing collinearity (VIF < 5).

Datasets

The study utilized data from the Centers for Disease Control and Prevention's (CDC) Behavioral Risk Factor Surveillance System (BRFSS) for the years 2012, 2013, and 2014. The BRFSS is a nationwide, phone-based survey of adults (18 years and older) that collects data on health-related behaviors. The data was used for both training and evaluating the predictive model.

Key Insights and Ideas

  • Age Matters, and It's Consistent: Older individuals are more likely to get vaccinated, and this trend remained remarkably stable across the three years studied. This likely stems from increased awareness of health risks and more frequent interactions with healthcare providers.
  • The "Checkup Effect": The recency of a doctor's visit is a strong predictor. People who have had recent checkups are more likely to be vaccinated, suggesting that reminders and discussions during these visits are effective.
  • Health Coverage as a Facilitator: Having health insurance or other forms of healthcare coverage significantly increases the likelihood of vaccination, likely by removing cost as a barrier.
  • Beyond Demographics: Behavioral Nuances: While age and health coverage are strong predictors, factors like smoking status (less likely to smoke, more likely to vaccinate) and employment status (employed individuals showing higher rates, possibly due to workplace reminders and social interaction) also play a role.
  • Cognitive Biases as a Barrier: The paper posits that even among educated individuals, cognitive biases like the Availability Heuristic (overestimating risks based on easily recalled examples) and the Bandwagon Effect (following the beliefs of others) can contribute to vaccine hesitancy.

Key Results

The final logistic regression model identified nine key predictors for influenza vaccination: age, pneumococcal vaccination status, time since last checkup, highest education level attained, employment status, health care coverage, number of personal doctors, smoking status, and annual household income.

  • The model correctly classified 67% of cases in 2013.
  • This predictive accuracy was consistent across the three years (2012-2014), with the model classifying 67-68% of cases correctly.
  • The overall vaccination rate in the US adult population ranged from 44% to 47% during the study period.
  • Age was the most consistent and significant predictor, with odds of vaccination increasing by 1.5% for each additional year of age.
  • Individuals with health care coverage were significantly more likely to be vaccinated.
  • Those who had a recent checkup (within the last year) had higher odds of vaccination.
  • Smoking status showed an inverse relationship: less frequent smoking (former smokers, never smokers) was associated with higher vaccination rates compared to daily smokers.

Limitations and Future Work

The authors acknowledge several limitations:

  • Cross-sectional Data: The study used multiple snapshots of data, which prevents establishing causal inferences. It can show associations but not definitively prove that one factor causes another.
  • Self-Reported Data: Data from the BRFSS is self-reported, which can lead to underreporting or bias, particularly for variables with social stigma (e.g., smoking, income).
  • Exclusion of Institutionalized Populations: Individuals in hospitals, nursing homes, or prisons were not included in the dataset.

Future research directions suggested include:

  • Investigating the impact of different health information sources (doctors, online, social networks) on vaccination decisions.
  • Exploring the influence of community beliefs and trust in the healthcare system on vaccination uptake.
  • Examining the role of anxiety about vaccination and fear of needles.
  • Developing and testing communication strategies that specifically address cognitive biases to improve vaccination rates. This could involve tailoring messages to different subpopulations based on their employment status, education level, or other demographic factors.

Summary updated

Abstract

Vaccinating adults against influenza remains a challenge in the United States. Using data from the Centers for Disease Control and Prevention, we present a model for predicting who receives influenza vaccination in the United States between 2012 and 2014, inclusive. The logistic regression model contains nine predictors: age, pneumococcal vaccination, time since last checkup, highest education level attained, employment, health care coverage, number of personal doctors, smoker status, and annual household income. The model, which classifies correctly 67 percent of the data in 2013, is consistent with models tested on the 2012 and 2014 datasets. Thus, we have a multiyear model to explain and predict influenza vaccination in the United States. The results indicate room for improvement in vaccination rates. We discuss how cognitive biases may underlie reluctance to obtain vaccination. We argue that targeted communications addressing cognitive biases could be useful for effective framing of vaccination messages, thus increasing the vaccination rate. Finally, we discuss limitations of the current study and questions for future research.
BibTeX
@article{kamis2017multiyear,
  author = {Arnold Kamis and Yuji Zhang and Tamara Kamis},
  title = {{A Multiyear Model of Influenza Vaccination in the United States}},
  journal = {International Journal of Environmental Research and Public Health},
  year = {2017},
  doi = {10.3390/ijerph14080849},
}
Figure