key: cord-0916198-5gi02i7x
authors: Myers, Kelly D.; Wilemon, Katherine; McGowan, Mary P.; Howard, William; Staszak, David; Rader, Daniel J.
title: COVID-19 associated risks of myocardial infarction in persons with familial hypercholesterolemia with or without ASCVD
date: 2021-05-25
journal: Am J Prev Cardiol
DOI: 10.1016/j.ajpc.2021.100197
sha: 58e40f3a93d34bff76eb88044f135183df417b26
doc_id: 916198
cord_uid: 5gi02i7x

nan

Within each group, we accounted for baseline differences between those individuals who did and did not contract COVID-19 in the exposure window by using propensity scoring (PS)-matching. Variables considered in PS-matching included demographic data (age, sex, household income, education level, and ethnicity), documented history of cardiac conditions (acute ischemic heart disease, cardiac arrest, coronary artery bypass graft, MI, and percutaneous coronary intervention), comorbidities (diabetes mellitus, hypertension, and ischemic stroke), cholesterollowering prescriptions (statin, ezetimibe, and PCSK9i claims), and selected laboratory test results (total cholesterol, low-density lipoprotein cholesterol, and lipoprotein (a)). We performed case-controlled matching without replacement using the MatchIt library [7] within the R statistical language (version 3.5.2) and balanced groups on both the number of patients with a given condition in the covariates window and the length of time the condition persisted in the patient's record. All individuals were followed through the data until they reached the end of the study period, experienced an AMI, or were lost to follow-up. Identification of AMI in the data was judged by an algorithm previously published [8] , whereby incidents of acute MI are distinguished from follow-up coding from a prior MI.

We found that AMI rates and Annualized Incidence Density Rates (AIDRs) were significantly increased in those diagnosed with COVID-19 as compared with matched individuals who did not contract COVID-19 in all six groups, including the largest group with no ASCVD and no FH ( Table 1 ). Across the study groups with a COVID-19 diagnosis, we found that rates of AMI were higher in the presence of diagnosed FH, probable FH, and ASCVD. Additionally, patients with a history of AS-CVD and diagnosed with COVID-19 had a significantly higher rate of AMI when compared to the population with COVID-19 but without AS-CVD or FH (AMI rate + 1.05%, Cl 95%; 0.99-1.11, p-value < 0.0002). Importantly, the addition of probable FH to pre-existing ASCVD represented a critical additive risk factor over ASCVD-alone, + 0.70% (Cl 95%; 0.25-1.16, p-value: < 0.0002), leading to the highest AMI rates in COVID-19 vs matched non-COVID-19 individuals. This analysis has several limitations and caveats. While the data contains records for a sizable fraction of Americans, we did not have access to a complete medical history. It is thus possible that history of ASCVD and/or FH was missed for some of the individuals, blurring the statistical differences between the groups. Similarly, the assignment of an individual into a COVID bucket is based solely on the presence of an ICD-10 diagnosis code and so may miss asymptomatic and untested individuals. Additionally, the process of PS-matching cannot account for unmeasured covariates (such as obesity, which is sparsely and poorly coded in the EHR) and so may introduce some bias related to unseen and unmeasured data. To account for this, we matched for an extensive range of lab tests, therapies, demographics, diagnosis, and procedural data, including nearly all major comorbidities directly relevant to population with cardiovascular related conditions. Finally, we were unable to answer a practical question, namely, did lipid lowering therapies (LLT) have a protective or deleterious effect on outcomes for those with FH in the COVID and No-COVID groups? Our analyses lacked statistical power for two main reasons. First, LLT rates were high in the matched groups that include diagnosed or probable FH patients (ranging from 63% to 83% of patients on one or more therapy). This was positive news for those patients but meant that a statistical comparison of event rates between individuals with and without LLT for the smallest groups in the analysis was difficult as the untreated population was small. Second, we found that patients with a history of LLT are generally also those patients with more recorded cardiac problems and comorbidities. To disentangle these competing effects of LLT therapy and higher severity individuals, we divided each of the main study groups into sub-groups containing those with a history of any LLT and those without LLT. We then PS-matched the sub-groups to directly measure the effect of LLT. Unfortunately, we found that the statistical comparisons between LLT and no-LLT was directional, but not significant.

Our analyses confirm that pre-existing ASCVD is associated with increased risk of AMI in the setting of COVID-19. Additionally, we establish that both diagnosed and probable FH are associated with increased risk for AMI in the setting of COVID-19. Critically, our data indicate that those with both ASCVD and FH are at very high risk of AMI if they contract COVID-19. Our results suggest that individuals with ASCVD and known FH should receive a COVID-19 vaccination when offered and demonstrate another reason for making greater efforts to identify and diagnose individuals with probable undiagnosed FH.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Cardiovascular implications of fatal outcomes of patients with coronavirus disease 2019 (COVID-19)

Impact of Covid-19 outbreak on clinical presentation of patients admitted for acute heart failure in India

The Covid-19 pandemic and the incidence of acute myocardial infarction

Fewer hospitalizations for acute cardiovascular conditions during the COVID-19 pandemic

Precision screening for familial hypercholesterolaemia: a machine learning study applied to electronic health encounter data

Reducing the clinical and public health burden of familial hypercholesterolemia: a global call to action

MatchIt: nonparametric preprocessing for parametric causal inference

Effect of access to prescribed PCSK9 inhibitors on cardiovascular outcomes

This analysis was funded by The FH Foundation.