(353) Rare adverse events of interest following vaccination: Strengths and weaknesses of a national data resource compared to the national sentinel network database

Background: Vaccine safety studies may be limited by the availability of sufficiently large datasets to investigate rare events. This is the case for thrombotic thrombocytopenia syndrome (TTS), which was associated with AZD1222 (ChAdOx1 nCoV-19) vaccination from early 2021.

Objectives: To conduct a feasibility assessment of the strengths and limitations of two databases to conduct a vaccine safety study in the English population, by comparing covariate availability and the incidence proportion of TTS across databases. The target audience is pharmacoepidemiologists.

Methods: This descriptive study compared data from the COVID-19 General practice extraction service Data for Pandemic Planning and Research (GDPPR), curated in NHS England’s Secure Data Environment (SDE); and the Royal College of General Practitioners (RCGP) Research Surveillance Centre (RSC) primary care sentinel cohort (PCSC). Both datasets were linked to national hospital episode statistics (HES), laboratory test results and immunisation data.

We included individuals aged 16+ years from 8/12/2020 to 31/10/2022. TTS cases were defined algorithmically as thromboembolism (TE) with thrombocytopenia (TCP) within 7 days using Systematized Nomenclature of Medicine Clinical Terms (SNOMED CT) and International Classification of Diseases, tenth revision (ICD-10) in primary and secondary care data, respectively. Incident TTS cases had no TE/TCP in the last year. Sociodemographic, lifestyle, and medical history variables were used to describe people with TE, TCP, and TTS.

Results: We included 7,641,136 and 47,325,090 individuals in the PCSC and GDPPR, respectively. Due to the limited availability of SNOMED terms in GDPPR, simpler code lists with lower sensitivity had to be used to identify risk groups. Several conditions including chronic liver disease and chronic kidney disease could not be ascertained in GDPPR, hampering comparisons of the vaccinated population and ascertainment of risk factors of TTS.

We observed 666 (0.3%) incident TTS cases in the PCSC and linked HES data, and ~6,700 (0.01%) in GDPPR. Due to limited SNOMED terms to define TE and unavailability of platelet count data to define TCP, no TTS was identified in GDPPR alone, with all cases ascertained using linked HES data.

Conclusions: GDPPR provides a large dataset of routine primary care data but has limited availability of SNOMED terms, prohibiting the identification of all TTS cases and ascertainment of risk factors without linkage. Linked secondary care data can help to identify TTS cases but is limited to diagnostic codes only. The reduced incidence proportion of TTS in SDE compared to PCSC suggests misclassification of incident TTS cases, which should be considered when conducting vaccine safety studies.