Date of Award

Summer 8-9-2016

Degree Type

Closed Thesis

Degree Name

Master of Public Health (MPH)


Public Health

First Advisor

Betty Lai, Ph.D.

Second Advisor

Jill Ferdinands, Ph.D.

Third Advisor

Emily Martin, Ph.D.


INTRODUCTION: With the rise of non-traditional providers offering influenza vaccination, it is becoming more of an endeavor to obtain documentation on vaccination. Researchers are relying more on self-report of vaccination, but the accuracy of differing definitions of self-reported vaccination status in the context of inpatient vaccine effectiveness studies is not well understood.

OBJECTIVES: The first objective of this study was the assess agreement among four different definitions of self-reported influenza vaccination compared against documented influenza vaccination in a group for whom documented vaccination is expected to be available. For the definition with best agreement, enrollee characteristics were examined to assess which were associated with agreement between definitions. The second objective was to assess the effect of ten different vaccination status definitions, including variations of self-report and documented vaccination, on vaccine effectiveness (VE).

METHODS: We used data from the inpatient component of the US Flu VE Network study of VE in patients hospitalized with acute respiratory illness (ARI). Variations of self-reported vaccination status were defined and compared to documented vaccination status. Agreement was assessed using the kappa statistic of agreement and other measures of agreement. VE was measured using logistic regression controlling for certain patient demographics and illness characteristics.

RESULTS: Self-report with date had the highest percent agreement and kappa statistic (79.5%; 57.7%, 95% CI: 50.1, 65.2). Those in the oldest age groups (18-49 and ≥65 years) and those influenza positive were more likely to have agreement between self-reported vaccination with date and documented vaccination status (p < 0.05 for both age group and case status). Estimated VE differed three-fold depending on definition of vaccination, with documented vaccination having the lowest VE (10%, 95% CI: -54, 48) and self-reported vaccination with date and location having the highest VE (37%, 95% CI: -10, 64).

CONCLUSION: Defining vaccination status using self-reported vaccination with date provides the most accurate classification of vaccination status. Older adults and those with lab-confirmed influenza were more accurate in their self-report of influenza vaccination. Differing definitions have an impact on estimated VE, and understanding how VE is influenced by choice of vaccination status definition is important to examine and report in studies of influenza VE.

Appendix_tables_29JUN16.pdf (335 kB)
Appendix Tables