A Prediction Algorithm for Familial Hypercholesterolaemia in New Zealand

Familial Hypercholesterolaemia (FH) is a genetic condition which causes an increased risk of cardiovascular disease and premature death (3-4 times the risk of early death). It is estimated to affect 1:300 people in the general population of New Zealand.
cute girl sitting in between her grandparents

Summer of Research

Project by Nick James, supervised by Dr. Patrick Gladding (University of Auckland).

Familial Hypercholesterolaemia (FH) is a genetic condition which causes an increased risk of cardiovascular disease and premature death (3-4 times the risk of early death). It is estimated to affect 1:300 people in the general population of New Zealand.

There are currently no large scale epidemiological studies from New Zealand for the prevalence of FH in the general population or within subsets such as Māori or Pacific people. Without treatment, patients with FH have a 3-4 fold increased risk of all cardiovascular diseases.

Identification and treatment of patients using statins significantly reduces cardiovascular risk to be approximately equivalent of someone without the condition. Early identification of FH patients would ensure they are on appropriate treatment to prevent them from developing cardiovascular disease, reducing the potential burden they will have on a healthcare system that is already under pressure.

This research aims to:

  • Use prediction models previously validated on other populations to estimate the probability of patients having FH within Waitemata DHB
  • Determine the differences in patient outcomes if they have a high vs low probability of having FH
  • Determine if patients who are predicted to have a high probability of having FH are on adequate medical treatment
  • Validate the prediction models using patients who are already diagnosed with FH

Using two models which have been tested on overseas cohorts and validated using genetic testing, we estimated the chances of 57,000 patients from the Waitemata District Health Board (WDHB) having FH. These patients were identified through the electronic records as having a previous echocardiogram within the WDHB since the adoption of electronic records. An end-point was created by searching unstructured electronic discharge summaries and cardiology clinic letters and finding synonymous mentions of FH.

For those patients that were predicted as likely to have FH there was a significantly decreased average age of death compared to those predicted as unlikely to have FH (67.9 vs 79.4 years, n=29 and 2847 respectively, p<0.0001). Patients who are more likely to have FH are also more likely to be on treatment for FH, being prescribed significantly higher levels of statin treatment (0.95 vs 0.68, n= 506 and 13866 respectively, p<0.0001). Both models that were tested did not perform well when predicting the few known patients who did have a confirmed diagnosis of FH, compared to other published studies using these methods.

FH can be predicted using clinical data, albeit to varying degrees of success. We found that patients who have been diagnosed with FH are likely to be more heavily medicated in the community, meaning their LDL-C levels are likely to be proportionately lower without information available to indicate a higher level of treatment. Ideally, the end-point to validate model effectiveness in New Zealand would be genetic testing, although there would be a significant cost associated with this.