Peer Effects and Influenza Vaccination Among Healthcare Workers
D.E. Curtis, S. Pemmaraju, L.A. Polgreen, P.M. Polgreen, A.M. Segre
3rd Biennial Conference of the American Society of Health Economists, (June 2010).

Introduction: Influenza vaccination is one of the most effective measures for preventing the transmission of influenza, and the CDC recommends influenza vaccination for all healthcare workers. However, influenza vaccination rates of healthcare workers remain unacceptably low (< 50%). Several recent investigations have shown the impact of social networks or peer effects on health-related behaviors and outcomes (e.g., smoking, obesity). In this paper we examine the effect of peer vaccination rates on healthcare workers.

Method: To examine peer effects, we construct social networks for hospital-based healthcare workers and examine the impact of neighbors’ vaccination status on the vaccination status of healthcare workers. By combining EMR login data with a thorough understanding of the hospital space, we construct social contact networks of varying densities for UIHC healthcare workers. UIHC vaccination data for the 2007 and 2008 seasons, when overlaid on the contact networks allows us to observe vaccination in a social context and to determine statistics such as an individual’s number of vaccinated neighbors.

We compare healthcare workers who were vaccinated in both years to those who were vaccinated in neither year (6904 total observations). We estimate 5 probit models with vaccination status as the dependent variable (1=not vaccinated). Independent variables are degree of connectedness, fraction of neighbors vaccinated, whether the observation is a physician, and whether she is involved in patient care. A worker’s neighbors are defined by proximity in time and space. We estimated this model with 5 data sets where we vary the definition of neighbor. A large number of healthcare workers in each data set have no discernable neighbors, and the maximum number of neighbors in each dataset ranges from 15 to 1,964.

Results: Comparing those vaccinated in both seasons, those who were vaccinated in neither season had fewer neighbors (marginal effects ranged from −0.0002 to −0.0018), were less likely to be involved in patient care (m.e. −0.03), were more likely to be a physician (m.e. 0.07), and had a lower fraction of vaccinated neighbors (m.e. −0.04 to −0.06). All marginal effects differ significantly from zero with 95% confidence. These results are robust to the different definitions of neighbor.

Conclusion: Unvaccinated healthcare workers had a lesser degree of connectedness (fewer neighbors). Thus it is possible that decisions not to become vaccinated are related to perceptions of lower risk of infection and also lower risk of infecting others. In addition, unvaccinated healthcare workers tended to have fewer vaccinated neighbors, so it is possible that they may experience less peer pressure to become vaccinated. These results may explain why it is difficult to increase vaccination rates among healthcare workers and may help design future interventions to increase vaccination rates.