Familial hypercholesterolaemia (FH) is a common inherited condition characterised by raised cholesterol and affects up to 320,000 people in the UK. Despite National Institute of Health and Care Excellence (NICE) Guidance encouraging identification of individuals with possible FH in primary care, over 80% of people with FH are still not identified. Without treatment, individuals with FH have a substantial risk of heart attack and premature death. These risks can be dramatically reduced by starting medicines to lower cholesterol levels.
Currently it is recommended for GPs to identify possible FH by examining patients who have raised cholesterol and a family history of heart disease. We, however, found this does not accurately identify people with FH and therefore results in many unnecessary referrals of individuals without FH to specialists. Using medical records from 3 million patients, we developed a new tool which identifies people with possible FH called FAMCAT in 2015.
The first version of FAMCAT is now being used in over 1000 General Practices. We have noted significant improvements in the quality of data and received feedback on incorporating new information related to FH. Based on new information and advanced analytic methods now available, we now need to update the tool to reflect current changes in patient records. Using CPRD, this study will find out if new information from patient records need to be added to the tool. Improvements to FAMCAT will further enhance its accuracy for use in health care.
Background: The algorithm for the tool to identify familial hypercholesterolaemia (FH) in routine primary care (FAMCAT) was developed and published in February 2015. Since then, the algorithm has been validated using UK primary care databases such as QResearch and RCGP database, and has shown to have consistently high performance. Based on recently available evidence, we seek to update and recalibrate the FAMCAT algorithm to the latest version of the Clinical Practice Research Datalink (CPRD) database. This will help to ensure the tool/algorithm reflects the changes in population characteristics and improvements in data quality which increases the ascertainment of FH events.
Aim: To develop and validate an updated version of the FAMCAT case identifying tool for familial hypercholesterolaemia in routine primary care.
Design: Retrospective open cohort study
Setting: General practices in UK providing data to the CPRD database.
Participants: All patients with at least 1 measurement of either total or LDL cholesterol between the baseline date (1 Jan 1999) and study end date (latest update of CPRD).
Outcomes: First (incident) diagnosis of familial hypercholesterolaemia.
Methods: The study population will be randomly split into algorithm derivation (70%) and validation (30%) cohorts. Random forest model will be developed to explore the composition and individual strength of novel risk factors/signals, with the current FAMCAT algorithm as the baseline. Logistic regression model will be used to estimate the coefficients and odds ratios associated with each potential risk factor for the diagnosis of FH for men and women separately. Prediction accuracy will be assessed by area under the receiver operating curve, sensitivity, specificity, positive predictive value and negative predictive value at various thresholds for diagnosis.
Outputs: An updated case case-finding tool to identify FH in routine primary care (FAMCAT2).
First (incident) diagnosis of familial hypercholesterolaemia.
Ralph Kwame Akyea - Chief Investigator - University of Nottingham
Ralph Kwame Akyea - Corresponding Applicant - University of Nottingham
Joe Kai - Collaborator - University of Nottingham
Nadeem Qureshi - Collaborator - University of Nottingham
Stephen Weng - Collaborator - University of Nottingham
Patient Level Index of Multiple Deprivation