Despite the availability and affordability of statins, a large proportion of high risk individuals in Asia have low-density lipoprotein cholesterol (LDL-C) levels that remain above recommended treatment targets which contributes to the burden of atherosclerotic cardiovascular disease (ASCVD) [1, 2, 3]. Current guidelines from the American College of Cardiology/American Health Association and European Society of Cardiology/European Atherosclerosis Society (ESC/EAS) emphasize achieving an absolute LDL-C 1.8 mmol/L for people with ASCVD or who are at high risk of ASCVD events [4, 5].

Further reductions in LDL-C are now obtainable with the addition of ezetimibe or proprotein convertase subtilisin/kexin type 9 (PCSK9) inhibitors to statin therapy. These nonstatin lipid-modifying drugs improve cardiovascular outcomes in secondary prevention of cardiovascular events [6, 7, 8], and are recommended for patients who do not achieve LDL-C targets on maximally tolerated doses of statins [4]. Guideline recommended nonstatin therapies have long been available in Hong Kong: ezetimibe (approved July 2003), evolocumab (approved May 2016), and alirocumab (approved October 2016) [9]. However, underuse of both statin and nonstatin lipid-modifying drugs in people with acute coronary syndrome (myocardial infarction (MI) or unstable angina) is a challenge in Hong Kong public hospitals: research conducted between 2009 to 2015 indicates that 25% of individuals with acute coronary syndrome did not receive statins by discharge and there was limited use of ezetimibe [10].

Emerging evidence suggests that achieving LDL-C treatment targets is also associated with improve cardiovascular outcomes in individuals with recent percutaneous coronary intervention (PCI) [11] and stroke or transient ischaemic attack (TIA) [12, 13]. In the Stroke Prevention by Aggressive Reduction in Cholesterol Levels (SPARCL) trial, atorvastatin 80 mg daily compared with placebo reduced the risk of fatal or nonfatal stroke and overall vascular events in individuals with a recent stroke or TIA and an LDL-C of 2.6–4.9 mmol/L [12, 14]. Targeting an LDL-C 1.8 mmol/L versus a higher target of 2.3–2.8 mmol/L with a statin, ezetimibe, or both, in individuals with ischaemic stroke or TIA in the Treat Stroke to Target (TST) trial also reduced the risk of cardiovascular events [13]. A comprehensive description of achieved LDL-C levels, that includes the early period of PCSK9 inhibitor availability, is needed in Hong Kong. Therefore, we aimed to describe the distribution of LDL-C concentrations at baseline and during one year of follow-up in patients hospitalized for a MI, PCI, stroke, or TIA.

After application of the exclusion criteria, we included a total of 18417 patients with a diagnosis of MI, stroke or TIA, or PCI, between 1 January 2003 to 31 December 2016 (Fig . 1). Baseline characteristics of the included patients are shown Table 1. The majority of patients were male, Chinese, diagnosed with an index stroke or TIA while 56.4% had a diagnosis of hypertension and 31.9% had a diagnosis of diabetes. By hospital discharge 66.8% were prescribed a lipid-modifying drug. Statins were the most prescribed class of lipid-modifying drug and 70.0% of patients prescribed statins received a moderate-intensity statin.

Fig. 1.

Flowchart of patients who had a lipid test at the Queen Mary Hospital. MI, myocardial infarction; PCI, percutaneous coronary intervention; TIA, transient ischaemic attack.

For each time window, the distribution of LDL-C stratified by index event is shown in Fig. 2. Most patients had an LDL-C result available at hospital discharge and during follow-up (Supplementary Table 2). The proportion of patients with a baseline LDL-C was 75.9% among patients with MI, 52.7% with PCI, and 83.0% with stroke or TIA. Overall, 1849 (10.0%) of patients had an LDL-C 1.8 mmol/L at baseline. By one year after discharge, a total of 4591 (24.9%) achieved an LDL-C 1.8 mmol/L. The proportion of patients achieving an LDL-C 1.8 mmol/L at any time during the one-year follow-up period was 35.4% among patients with MI, 37.6% with PCI, and 16.5% with stroke or TIA. A larger proportion of patients with PCI (89.5%) or MI (83.4%) had follow-up LDL-C results available any time during follow-up when compared to patients with an index stroke or TIA (56.8%).

Fig. 2.

Distribution of low-density lipoprotein cholesterol concentrations stratified by index event. We assessed low-density lipoprotein concentrations at baseline (during the index hospitalization), and at 30, 90, 180, and 365 days after hospital discharge, or any time during one year of follow-up.

The number of patients with a hospital discharge LDL-C concentration $\ge$1.8 mmol/L who subsequently achieved an LDL-C concentration of 1.8 mmol/L by 365 days after hospital discharge are presented in Table 2. Of the patients who had any baseline LDL-C result (n = 13783), 11934 (86.6%) had a baseline LDL-C concentration $\ge$1.8 mmol/L. Overall during follow-up, 18.6% of these patients achieved an LDL-C of 1.8 mmol/L. The proportion of patients who achieved this goal by the end of 365 days varied by index event. Fewer patients with a stroke or TIA and a baseline LDL-C $\ge$1.8 mmol/L achieved an LDL-C of 1.8 mmol/L one year after hospital discharge.

Table 3 presents the baseline characteristics and percent reduction of LDL-C from baseline, stratified according to an LDL-C of 1.8 mmol/L by 365 days after hospital discharge, after excluding 5654 patients missing a follow-up LDL-C test. A larger proportion of patients who achieved an LDL-C of 1.8 mmol/L were prescribed statins at baseline as compared to patients who had an LDL-C $\ge$1.8 mmol/L (87.2% vs 67.6%). About 5% of patients in both groups used nonstatin therapies. Patients who achieved an LDL-C 1.8 mmol/L appeared to be at higher cardiovascular risk as 42.6% had $\ge$3 TRS 2${}^{\circ }$P risk factors.

Achieved LDL-C targets during follow-up, the distribution of LDL-C, and the percent reduction in LDL-C according to index year are shown in Table 4. Despite a similar percent reduction in LDL-C over the study period, the absolute mean achieved LDL-C concentrations, and the proportion of patients achieving either an LDL-C 1.8 mmol/L, $\ge$50% LDL-C reduction from baseline, or both, appears to increase over time.

Using a large cohort of Chinese individuals, we described LDL-C levels in-hospital and one year after discharge, and report important clinical characteristics such as the use of statin and nonstatin lipid-modifying drugs and the TRS 2ºP. During one year of follow-up, 75% of patients did not achieve an LDL-C 1.8 mmol/L. When examined by year of index event, average LDL-C levels declined and the proportion of patients achieving an LDL-C 1.8 mmol/L increased between 2003 to 2016. Despite the availability of multiple lipid-modifying treatments, we observed that about 1–2% of patients had follow-up LDL-C levels $\ge$4 mmol/L and 30% did not have a one-year follow-up LDL-C test result available in CDARS.

Our findings align with a previous study of patients with acute coronary syndrome who underwent PCI in Hong Kong between 2009 to 2015. Wang et al. [10] found that of these patients 11.3% received high-intensity statins, 26.8% did not receive a statin, and only 0.2% received ezetimibe. The proportion of patients with MI and PCI with a baseline and follow-up LDL-C 1.8 mmol/L was also similar to the findings of our study at about 12% and 35–40% respectively. Notably, despite that we followed-up patients until the end of 2017, 45% of patients did not receive a statin before hospital discharge, and there was limited use of ezetimibe and PCSK9 inhibitors.

We were particularly interested in the availability of in-hospital and follow-up LDL-C levels; missing lipid test patterns could identify gaps in monitoring patient response to lipid treatment. When examining the availability of LDL-C tests at baseline and during follow-up, a similar number of patients had an LDL-C level during their index hospitalization, but the proportion with follow-up levels increased from around 55% in 2004 to 80% in 2016. Further investigation is needed to understand the delay in obtaining follow-up LDL-C levels as nearly 30% of patients did not have a measured LDL-C within one year of the index event. It is possible that a group of patients with missing LDL-C values transfer their care to the private sector, leave Hong Kong, or obtain a follow-up LDL-C after 12 months, and would thus have a missing follow-up LDL-C result in our analysis.

Differences in achieved LDL-C levels exist when comparing Hong Kong to other countries. LDL-C levels during follow-up may have narrower distribution in our cohort compared with data from the United Kingdom. About 1.8% of patients in this study had LDL-C levels which remained $>$4 mmol/L during follow-up, but the proportion was generally less than those reported by Danese et al. [19], which ranged from 1.4% to 4.1%, in a similar cohort study that examined data from the Clinical Practice Research Datalink and the Hospital Episode Statistics. While difficult to make direct comparisons, a similar percentage of patients in the United Kingdom (23 to 42%) achieved an LDL-C 1.8 mmol/L at one-year after discharge [19], to those in our study (17% for patients with an index stroke or TIA to 38% for patients with an index PCI). Evidence from Italy, that assessed patients defined as very high risk according to ESC/EAS clinical guidelines between March 2016 to February 2017, demonstrates more frequent use of statins (94.1%), high-intensity statins (54.9%), and ezetimibe (14.4%) [20]. In this study, 58.1% of patients achieved a target LDL-C of 1.8 mmol/L [20]. A higher rate of LDL-C 1.8 mmol/L achievement in Italy corresponds to the greater prescribing prevalence of high-intensity statins, which was much lower in our study in Hong Kong.

This study has several limitations. Although calculated LDL-C remains the primary target for lipid modification in clinical practice guidelines, it becomes more inaccurate at low LDL-C and high triglyceride values [5]. The assessment of lipid-modifying drug effectiveness using the Friedewald formula could be confounded in some individuals because of the lack of fitness of calculated LDL-C in these contexts. The TRS 2ºP has only been validated in patients following acute coronary syndrome, and not with a diagnosis of stroke or TIA, thus the risk predictors of TRS 2ºP may not be completely applicable to all patients in our cohort. Initial lipid test data were identified only for one hospital. The Queen Mary Hospital is an important referral hospital in Hong Kong, which could cause selection bias of more severe or complex cases. CDARS only includes data from the Hospital Authority, and thus excludes private healthcare data. For example, after discharge patients could obtain follow-up prescriptions and laboratory testing at private hospitals or clinics. Finally, selection bias may also occur for patients who die after hospital discharge and prior to their first follow-up LDL-C result. However, only 913 patients died and had no recorded follow-up LDL-C level, which represents only 16% of the patients with no follow-up LDL-C results. The eligibility criteria were limited to patients who had a lipid test at a single hospital and we did not differentiate haemorrhagic and ischaemic stroke in our inclusion criteria, resulting in a more heterogeneous group of patients with stroke. To address these limitations, our research group is currently conducting an up-to-date assessment of LDL-C target achievement in all patients diagnosed with ASCVD (i.e., ischaemic heart disease, ischaemic stroke, TIA, and peripheral vascular disease) in the Hospital Authority from 2010 to 2020, regardless of whether a lipid test has been measured.

Between 2003 to 2016, there was vast under-treatment of ASCVD with lipid-modifying drugs in Hong Kong, with only 25% of patients in this study achieving a contemporary target LDL-C of 1.8 mmol/L. Local quality improvement initiatives and earlier uptake of both statin and nonstatin lipid-modifying drugs could help further reduce LDL-C and ultimately contribute to reducing the risk of recurrent cardiovascular events.

CWS, EWC, and JEB conceptualized the study; JEB did the formal analysis, and VKCY and JZ cross-checked the analysis; CWS and EWC acquired the funding, provided resources, and supervised the study; JEB, VKCY, JZ, CSLC, CWS, ICKW, and EWC interpreted the results; JEB wrote the first draft and VKCY, JZ, CSLC, ICKW, and EWC critically reviewed and edited the manuscript for intellectual content.

Ethics approval for this study was obtained from the Hong Kong West Cluster/HKU Institutional Review Board, reference number UW 14-334. Informed consent for this study was waived by the institutional review board.

JEB was supported by the Hong Kong Research Grants Council as a recipient of the Hong Kong PhD Fellowship Scheme.

This research was funded by Amgen Asia Holdings Limited, Hong Kong SAR, China, grant number RS190135. The authors drafted the initial protocol based on the funder’s objectives. The protocol underwent review and revision following feedback from the funder, and was approved by both the funder and the authors. The funder had the opportunity to review the manuscript and provide non-binding comments prior to submission for publication.

CSLC, EWC, and ICKW report grants from Amgen. CSLC received a grant from Hong Kong Innovation and Technology Commission outside of the submitted work. All other authors declare no competing interests.