We used data from the Asan BiomedicaL research Environment (ABLE) database, a data warehouse system that contains de-identified EMR from Asan Medical Center (AMC), a large tertiary care hospital in South Korea [17]. Data from January 1, 2000 to December 30, 2020 were analyzed (Fig. 1) [18]. This study was approved by the institutional review board of Asan Medical Center (Institutional Review Board number: 2021-0192), which granted permission to access and analyze the database utilized in this study. This study was conducted in accordance with the Helsinki Declaration. The requirement to obtain informed consent was waived due to the minimal-risk nature of the study.

Study schematic diagram. ASCVD atherosclerotic cardiovascular disease, HRU health resource use, LDL-C low-density lipoprotein cholesterol, MACE major adverse cardiovascular event

The study population was patients with very-high-risk ASCVD. We selected patients admitted to the AMC for an ASCVD event (ACS [MI/unstable angina], stable angina, asymptomatic coronary artery disease [CAD], IS/TIA or peripheral artery disease) between January 2000 and December 2019 (to allow for at least 1-year follow-up). This definition is based on the ESC/EAS 2019 and Korea dyslipidemia guidelines, we defined patients with any prior history of ASCVD as very-high-risk [2, 19]. The instrumental definitions used to identify patients with ASCVD (based on diagnosis codes and observations of procedures), which were validated by clinical experts, are included in Supplementary Table 1. If patients were hospitalized more than once during the study period, the first hospitalization defined the baseline ASCVD event. We restricted the study population to those with the record of baseline LDL-C. Baseline LDL-C was recorded at admission or within 6 months prior to admission (the measurement closest to the admission date was used). The study population also had to have a follow-up LDL-C screening conducted within 4–12 weeks post-discharge.

To minimize non-LDL-C-related confounding for recurrent MACEs, patients were excluded if they: (1) had a known history of hemorrhagic or lacunar stroke at any time before the baseline ASCVD event. Lacunar stroke patients were excluded to minimize confounding, as their risk of MACE is largely independent of LDL-C levels due to the different pathophysiology and primary risk factors, (2) had a systolic blood pressure (SBP) over 180 mmHg or diastolic blood pressure (DBP) over 110 mmHg at cohort entry or at index date, (3) experienced a recurrent MI or non-hemorrhagic stroke within 4 weeks of hospital discharge to exclude events potentially linked to the baseline ASCVD event, or (4) were recent/unstable LLT initiators (defined as those who had not received LLT for > 2 weeks).

We dichotomized the exposure group by whether meeting the LDL-C target goal (< 55 mg/dl) at the index date, defined as the date of the follow-up screening for LDL-C between a 4- and 12-week window for post-discharge (Fig. 1). Early target LDL-C group was defined as patients with LDL-C levels < 55 mg/dl at the index date and patients who did not meet this target served as the non-early target LDL-C group. The rationale of 4–12-week window for post-discharge LDL-C ascertainment was in accordance with clinical practice in South Korea, where outpatient visits are commonly scheduled within 8 weeks of discharge, with 4-week grace period.

The exposure criteria (< 55 mg/dl) was based on the ECDP and ESC/EAS guidelines, which recommend an LDL-C goal of < 55 mg/dl and > 50% reduction from baseline for secondary prevention in patients with very-high-risk [2, 11, 20]. Our study omitted > 50% reduction requirement to avoid distorting the association, as treatment-naïve baseline LDL-C values or LLT history before AMC admission were unavailable. Patients with a lower LDL-C at AMC admission due to prior LLT likely had a smaller percentage reduction in LDL-C (at admission vs. post-discharge). For example, a patient with an LDL-C of 75 mg/dl receiving statins before the index date might be misclassified as non-early target LDL-C group if their LDL-C decreased to 50 mg/dl (33% reduction). Sensitivity analyses were also conducted using an alternative cut-off of < 70 mg/dl, reflecting local clinical practice during the study period [19].

For the primary objective, the primary MACE outcome was recurrent MACEs, defined as a composite of MI, IS, hospitalization for unstable angina, coronary revascularization (coronary artery bypass grafting [CABG] or PCI), and all-cause mortality. The secondary MACE outcome was defined as a composite of MI, IS, and all-cause mortality.

We measured time to the earliest MACE event. From the post-discharge LDL-C screening (index date), patients were followed up to 21 years for the occurrence of the earliest recurrent MACE or were censored at the end of the study period (December 30, 2020), whichever came first. Additionally, we conducted time-at-risk analyses to measure the early outcome among patients with ACS. Fixed time-at-risk analyses were conducted to assess the recurrence of primary composite MACEs at 6, 12, 36, and 60 months after the index date in patients with ACS.

For the secondary objective to measure HRU outcomes by early LDL-C goal achieved target, HRU outcomes were frequency and length of stay of cardiovascular- (CV) related hospitalization, frequency of CV-related procedures, frequency of lipid panel tests for LDL-C monitoring, frequency of CV-related rehabilitation visits, frequency of emergency room visits, and frequency of all-cause outpatient visits (including CV-related rehabilitation visits).

For the exploratory objective, post-index longitudinal LDL-C measurements were captured throughout the follow-up period (up to 21 years).

We measured patient demographics and lifestyle variables (age, sex, smoking status, and BMI), as well as comorbidities (e.g., diabetes mellitus, hypertension, chronic kidney disease, atrial fibrillation, previous ASCVD subtype, and cancer) assessed at index date or earlier. Laboratory test results (high-density lipoprotein cholesterol, lipoprotein[a] (Lp[a]), triglycerides, estimated glomerular filtration rate, blood pressure, and albumin-to-creatinine ratio) closest to the index date were used. The use of LLTs includes statins, ezetimibe. Data on the use of PCSK9 inhibitors were not available. Comedications (aspirin or P2Y12 inhibitors, beta-blockers, antihypertensives) were also assessed as potential confounders. The list of diagnosis codes for comorbidities and Anatomical Therapeutic Chemical (ATC) codes for medications are presented in Supplementary Tables 2–3.

For comparisons of baseline characteristics, the absolute standardized difference (aSD) was calculated; an imbalance between groups was defined as an aSD value > 0.1. We estimated cumulative incidence, cumulative incidence curves, and incidence rates of MACE outcomes. We used Cox proportional hazard models to estimate the hazard ratio (HR) and 95% confidence intervals (CI) for the association between early LDL-C goal achievement and the decreased risk of recurrent MACE in the overall and subgroup cohorts, adjusting for age, sex, diabetes mellitus, smoking status, aspirin or P2Y12 inhibition, beta-blockers, RAAS inhibitors, chronic kidney disease, statin us), and non-statin LLT. Subgroup analyses were conducted based on the type of baseline ASCVD event: ACS, stable angina, asymptomatic CAD, and IS/TIA.

For the secondary objective, we compared HRU frequency between the early target LDL-C and non-early target LDL-C groups, using negative binomial regression to calculate rate ratios (RRs) with 95% CIs, using the same adjustment variables. For the exploratory objective, LDL-C trends were observed by plotting mean and median LDL-C measurements during the long-term follow-up (up to 21 years) on a simple line graph.

To test the robustness of our main results, we conducted sensitivity analyses by redefining the LDL-C target at a cut-off of < 70 mg/dl. Further sensitivity analyses were performed among patients with IS/TIA by (1) restricting the study population to patients without active cancer, given that the presence of cancer may increase risk of recurrent MACE independent of LDL-C, (2) including patients with lacunar stroke previously excluded, (3) using hemorrhagic stroke as the safety outcome, and (4) using differential LDL-C goal cut-offs (< 100 mg/dl; < 100 mg/dl and ≥ 50% reduction).

A two-tailed p value < 0.05 was considered statistically significant. No multiplicity adjustment was applied. Statistical analyses were performed using R software version 4.2.3 (R Foundation for Statistical Computing, Vienna, Austria; www.r-project.org).