(046) Differences in Time to Onset Between Spontaneous Reports and Cohort Studies: The Case of the COVID-19 Vaccines

Publication Number: 383

Background: Spontaneous reporting of Adverse Drug Reactions (ADRs) may be subject to recall bias to a greater extent than reporting from cohort studies. The time to onset (TTO) of an ADR could be related to the causality and may influence the decision to report and therefore be subject to recall bias.

Objectives: To investigate whether differences in TTO between spontaneous reporting and cohort reporting can provide information on recall bias in spontaneous reporting.

Methods: ADRs from the national spontaneous reporting system and from a prospective cohort event monitoring (CEM) study using patient reported outcomes related to the COVID-19 vaccination campaign in the Netherlands were included. For each ADR the TTO in days was calculated. Included variables possibly related to the reported TTO were solicited vs non-solicited, injection moment, high burden score on a 5-point Likert scale and the presence of media attention (for menstrual disorders only). TTO distributions were visually compared using cumulative distribution plots and differences in median TTOs between spontaneous and cohort reports were tested using Mann-Whitney U tests. In addition, the possible effect of an increasing TTO on the difference in median TTO between spontaneous and cohort reports was investigated.

Results: A total of 165,913 spontaneous reports and 19,979 reports from the cohort were included. Overall, there were no differences in the median TTO between spontaneous reports and cohort reports. In general, median TTOs were lower for solicited reports compare to non-solicited reports (1 day vs 3 days respectively; p < .001). Also, after stratification for injection moment and burden of the ADR, no differences were found. The presence of media attention resulted in different TTO patterns when comparing ADRs reported prior versus during / after the window of media attention (p < 0.001). Finally, differences in median TTO between spontaneous and cohort were more pronounced with increasing TTO values.

Conclusions: Based on differences in TTO, recall bias does not seem to be more pronounced in spontaneous reports compared to cohort reports in general. However, the observed differences between spontaneous reports and cohort reports with increasing TTO suggest that recall bias may be more evident in ADRs with a longer TTO. On the other hand, one could argue that those with a longer TTO from cohort reports are more likely to be coincidental, not causally related events.