(358) Use of test-negative designs in pharmacoepidemiology: focus on vaccine effectiveness 2013-2023

Background: Vaccine effectiveness (VE) assessment poses specific pharmacoepidemiologic challenges related to definitions of vaccination exposure and infectious disease outcomes. The test-negative design (TND) is a recognized approach to assess VE. However, there is no current overview of the extent of its use nor of the type of vaccines for which it is employed.

Objectives: To examine the use of the TND to assess VE during 2013-2023 and describe characteristics of TNDs published during this period.

Methods: We conducted a literature review in PubMed, using search terms [test negative design AND vaccine effectiveness] during 2013-2023 period. Manual review was conducted to exclude publications not related to TNDs, not assessing VE, or general TND methodology and review papers.

Results: Search yielded 792 publications and 491 were deemed relevant after manual review. The annual number of TND publications increased over time, from 15 in 2013 to a peak at 108 in 2022. After a small increase in 2013-2015, this number was steady in 2016-2020 (between 32 and 43 per year) and then increased steeply from 2021 onwards, with 52% of the papers (n=255) published during 2021-2023.
The hypothesis that this increase might be due to the multiplication of Covid-19 VE studies is confirmed by the results of the search. Overall, Covid-19 was the second most frequent indication (n=162, 33%) after influenza (n=273, 56%).
All Covid-19 VE TND papers were published between 2021-2023, with a peak in 2022 (n=89) corresponding to the period of Covid-19 vaccine deployment. Conversely, influenza VE studies, for which the TND approach was primarily developed, were published regularly with an average of 24 manuscripts (9%) per year. Most (73%) of the influenza vaccine TND papers were published between 2013-2020.
The next most frequent indications were rotavirus (n=25, 5%) and pneumococcus (n=9, 2%). Other indications represented 4% of manuscripts and included cholera (n=5) and pertussis (n=5) as well as RSV (1 publication in 2023).
Our results will also describe methodologic aspects of TNDs, e.g., use of primary vs. secondary data, and potential bias related to test setting, e.g., voluntary vs. routine (hospital setting)/systemic (surveillance systems).

Conclusions: Prior to the Covid-19 pandemic, TND was mainly used to assess influenza VE. Development of specific Covid-19 vaccines triggered an increase in use of TNDs to assess their VE. TND is particularly adapted for acute diseases (e.g., respiratory viruses) for which tests exist and Covid-19 was a good candidate due to initial widespread testing. Based on the collectively acquired experience conducting TNDs, it can be expected that this approach will be applied to upcoming vaccines in new indications.