Using Real-World Evidence for Comparative Safety Studies
See the research:
Comparative safety of high-dose versus standard-dose influenza vaccination in patients with end-stage renal disease
Layton JB, McGrath LJ, Sahrmann JM, Ma Y, Dharnidharka VR, O'Neil C, Weber DJ, Butler AM.
Plotting daily incidence of influenza vaccine reactions among patients with end-stage renal disease to inform study design decisions
Layton JB, McGrath LJ, Sahrmann JM, Dharnidharka VR, O'Neil CA, Butler AM.
Generating valid real-world evidence requires thoroughly understanding the underlying real-world data. We performed a large evaluation of the comparative safety of high-dose and standard-dose influenza vaccinations among patients with end-stage renal disease, or ESRD, who were receiving hemodialysis. We wanted to understand how the unique healthcare utilization patterns of patients with ESRD could affect our ability to identify incident vaccine outcomes.
Patients with ESRD frequently have in-center dialysis sessions 3 times a week and 4 nephrologist encounters a month, 1 of which is a comprehensive visit. These patterns of regular healthcare interaction may influence when and how often diagnoses are recorded.
We identified influenza vaccinations among patients with ESRD. We looked at the 180 days before and after vaccination, and then we plotted the daily incidence rates of recorded diagnoses of several different potential adverse vaccine reactions, some milder and some more serious.
Some very clear patterns emerged. Some milder events had very clear spikes in incidence every 2 to 3 days, more than doubling over the other days; some had spikes every 7 days; and still others had very clear monthly spikes, with daily incidence rates approximately tripling every 30 days. These patterns are all consistent with diagnoses for recurrent, minor, or sustained conditions being recorded at regularly occurring healthcare interactions.
In contrast, severe events that required hospitalization didn't show clear cycles, presumably because hospitalizations do not occur with predictable frequency like dialysis sessions or physician visits.
In medically complex populations with predictable patterns of healthcare utilization, aspects of the study design such as the length of the lookback period or the follow-up period or the inclusion of the vaccination date in the follow-up period can be informed by these observed patterns. For example, in the comparative study based on this work, we extended the pre-vaccination lookback period from 30 days to 42 days to ensure that we captured the previous month's spike and 6 weeks of weekly spikes for minor events.
Follow the link to see the poster describing this work and the spotlight poster reporting the results of the full comparative safety study.