Development of an algorithm to identify Hypereosinophilic Syndrome using the Clinical Practice Research Datalink linked with hospital data

Date of Approval: 
2020-11-03 00:00:00
Lay Summary: 
Hypereosinophilic syndrome (HES) is a very uncommon disease, in which a specific type of blood cells called eosinophils are overproduced for long periods, resulting in high blood eosinophil levels and body-organ damage. Patients with HES in the UK’s Clinical Practice Research Datalink (CPRD) database can be characterised using a specific diagnostic code. However, other databases, such as those that use the international classification of diseases, version 10 (ICD10), lack a specific code for HES, though a general term of ‘eosinophilia’ (D72.1) exists. We want to find a way to recognise HES disease in patients in these other databases. To do this, we will find and characterise patients with the HES code in the CPRD database, as the reference population. Then, by using a combination of clinical characteristics and medications used by these patients, we aim to develop a way to find similar patients in other databases where the specific codes for this disease do not exist yet.
Technical Summary: 
Hypereosinophilic syndrome (HES) is a very rare disease, characterised by a marked eosinophilia sustained over time that may cause organ damage and/or dysfunction. Patients with HES in the UK’s Clinical Practice Research Datalink (CPRD) database can be characterised using a specific READ/SNOMED/EMIS diagnostic code. However, other databases, such as those that use the international classification of diseases, version 10 (ICD10), lack a specific code for HES, though a general term of ‘eosinophilia’ (D72.1) exists. We want to recognise HES disease in patients in these other databases. To do this, we will find and characterise patients with the existing HES code in the CPRD Aurum database linked with hospital records, as the reference population. Then, by using a combination of ICD-10 codes, laboratory tests and medication codes, we aim to develop an algorithm that could be used in other databases to characterise these patients.
Health Outcomes to be Measured: 
Diagnoses, laboratory results and treatments used among patients with specific READ/SNOMED/EMIS codes for HES (‘HES’ reference population); Positive Predictive Value (PPV); Negative Predictive Value (NPV); sensitivity and specificity.
Application Number: 
20_000137
Collaborators: 

Gema Requena - Chief Investigator - GlaxoSmithKline - UK
Gema Requena - Corresponding Applicant - GlaxoSmithKline - UK
Nicholas Galwey - Collaborator - GSK
Rupert Jakes - Collaborator - GlaxoSmithKline - UK
Sandra Joksaite - Collaborator - GlaxoSmithKline - UK

Linkages: 
HES Admitted Patient Care;HES Outpatient