A precision medicine approach to lipid monitoring in the secondary prevention of cardiovascular disease

Brown, Rosemary Elisabeth (2021) A precision medicine approach to lipid monitoring in the secondary prevention of cardiovascular disease. PhD thesis, University of Glasgow.

Abstract

Cardiovascular disease (CVD) is the leading cause of morbidity and mortality globally. Individuals should have their risk factors monitored and clinically managed to lower this risk, and many countries adopt similar prevention strategies targeting those at highest risk. One such high-risk group are those who have previously experienced a cardiovascular event, and consequently these patients are universally identified and targeted in risk management strategies. This thesis initially compared recommendations for the management of lipids in secondary prevention populations in a systematic review of national clinical guidelines. This found that statins were consistently recommended, but there were substantial differences in the use of lipid targets and the frequency of lipid monitoring after statin initiation, although annual lipid tests were the most frequently recommended. Aside from expert opinion, however, there was little robust evidence to support the recommendations for targets and monitoring frequency.

Therefore, the frequency of lipid testing under the current guidelines is unlikely to be optimal for the management of many patients. Increased use of electronic health records could allow the development of algorithms to result in a personalised approach to lipid testing. Specifically, if most patients have lipid tests annually, the possibility that there are a group of patients that need less frequent monitoring can be assessed. The remainder of this thesis, therefore, aimed to explore this within a subgroup of the secondary prevention population, survivors of myocardial infarctions (MIs), within Greater Glasgow and Clyde (GGC), where current and previous guidelines recommend an annual lipid test.

To achieve this, the cohort was first described to facilitate comparisons with external literature and to understand the ongoing real-life clinical management of these patients. Associations between adherence and the achievement of guideline-recommended lipid targets were then investigated with further hospitalisations for MIs and mortality. Finally, factors associated with non-adherence and non-target lipids were identified and used to predict patients’ subsequent adherence and cholesterol levels, i.e. those who could receive reduced lipid monitoring.

Data was obtained from NHS GGC’s Safe Haven for 11,110 patients who experienced a non-fatal MI between 2009 and 2014, with follow up available until July 2017. Demographics were consistent with similar observational cohorts from other countries in the literature, including a greater proportion of males to females, an average age of 67 years, and approximately a fifth diagnosed with diabetes before their baseline MI. Estimated statin adherence, obtained through encashed prescribing records, found that two thirds achieved an average adherence during follow up ≥80% and 85% ≥50%. Three quarters of those with at least one lipid test achieved LDL ≤1.8mmol/l during follow up. Statin adherence did not fully account for LDL target achievement, but those with higher adherence were significantly more likely to achieve it.

High adherence and lipid target attainment were common suggesting that there was a subset of patients for whom an annual lipid test could be considered unnecessary, as a further lipid test was unlikely to change any clinical decisions. Non-adherence and elevated lipids were separately significantly associated with increased mortality within this cohort (<80% adherence, HR (95% CI): 1.4 (1.3-1.5); LDL>1.8mmol/l: 1.3 (1.2-1.4)), and with CVD mortality specifically (<80% adherence: 1.3 (1.1-1.5); LDL>1.8mmol/l: 1.3 (1.1-1.5)). Therefore, careful and accurate identification of low risk patients is needed to avoid increased mortality.

Latent class analysis, a type of mixture model for categorical variables, was implemented to explore clustering and patterns within the data associated with lipid target achievement and adherence, into latent classes. For LDL ≤1.8mmol/l, sensitivity of these classes was 83% and positive predictive value was 100%, meaning all those predicted to achieve the target did so. This positive predictive value was also observed when ≥50% adherence was considered, and
sensitivity was 99%. The class share of those predicted to have ≥50% adherence was substantially larger, with 85% predicted to do so, compared to 42% for LDL ≤1.8mmol/l. Associations between predicted classes and mortality showed that those predicted to have ≥50% adherence experienced lower rates of mortality, than those predicted not to. This was not the case for the LDL ≤1.8mmol/l, although the predictions performed no worse than those observed.

In conclusion, given lipid tests as part of an annual review can be expensive in terms of the time needed for repeat appointments and biochemistry, the purpose of these for secondary prevention CVD patients needs to be considered and clarified by guideline committees. Once a patient meets a lipid target and adherence continues to be high, clinical decisions are unlikely to change with further blood tests. The results in this thesis have shown that using previous adherence and lipid results, and demographic information, patients’ adherence can be accurately predicted and therefore could be used as a practical marker of lipid test’s necessity within a review. However, before the implementation of this approach should be considered, further validation of these results within other external observational cohorts is required, and a non-inferiority randomised controlled trial should also be implemented.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Keywords: secondary prevention, precision medicine, lipids, myocardial infarction, lipid monitoring, cholesterol.
• Subjects: R Medicine > R Medicine (General)
• Colleges/Schools: College of Medical Veterinary and Life Sciences > School of Cardiovascular & Metabolic Health
• Funder's Name: Medical Research Council (MRC)
• Supervisor's Name: Welsh, Dr. Paul, Logue, Dr. Jennifer and Lewsey, Professor Jim
• Date of Award: 2021
• Depositing User: Miss Rosemary Brown
• Unique ID: glathesis:2021-82173
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 11 May 2021 10:50
• Last Modified: 11 May 2021 10:56
• Thesis DOI: 10.5525/gla.thesis.82173
• URI: https://theses.gla.ac.uk/id/eprint/82173

Related URLs

• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Article DOI