(249) The importance of the analysis of multiple databanks for the valid estimation of thrombocytopenia incidence rates in children in BIFAP data source (Spain).

Background: Background Incidence rates (IR) of clinical outcomes are needed to contextualise potential safety signals detected with drug use or vaccine administration. However, IRs for newborns and rare events are scarce in the literature. IR estimations using electronic health records requires the identification and validation of cases tailored to the characteristics of the data source.

Objectives: To quantify the IR of thrombocytopenia and immune thrombocytopenic purpura (ITP) in children aged 0-1 years in BIFAP (Pharmacoepidemiological Research Database for Public Health Systems) and to describe its registration characteristics.

Methods: All patients aged 0-1 years registered in BIFAP between 2015 and 2019 formed our study cohort. Patients were followed from the date of registration with a primary care physician (PCP) until a record of thrombocytopenia (cases), death, disenrollment or the end of the study period, whichever came first, using data from two databank: Primary Care (PC) and hospital discharge records.
Three different strategies were used to identify potential thrombocytopenia cases, depending on the record type: 1-case finding algorithms based on broad semantic searches in PCP diagnosis records or free text clinical notes; 2-platelet counts lower than 150 x 10^9/L from PC laboratory results; 3-ICD-9/10 related codes in hospital discharge diagnoses.
A manual validation of medical records of all potential cases was performed. Valid cases were those identified with the diagnosis of interest without any additional information ruling out that diagnosis.
IR/100,000 py of first-ever thrombocytopenia or ITP in children aged 0-1 years was estimated. The distribution of cases by record type was also calculated.

Results: The study cohort included 463,083 patients (361,302 py of follow-up). Potential cases identified were 4,863. After manual review, the number of valid thrombocytopenia and ITP cases were 841 and 40 respectively, leading to a corresponding IR/100,000 py in children aged 0-1 years of 232.8 and 11.1 respectively.
Most valid thrombocytopenia cases were recorded as a diagnosis by the PCP (16.3%) or in clinical notes (68.8%). The proportion of cases identified solely by platelet counts in PC laboratory results and hospital discharge records was of 11.9% and 3.0%, respectively.
The proportion of patients with thrombocytopenia identified in more than one record type was of 7.2%, (PCP + Laboratory results: 4.0%; PCP + hospital discharge: 3.1% and all three: 0.1%).

Conclusions: The most relevant source of information to identify cases of thrombocytopenia in children 0-1 year-old in BIFAP was PC records, especially clinical notes. The use of different databanks and strategies tailored to the record types available in BIFAP datasource increases the validity of IR estimations.