Early lipid-lowering therapy (LLT) has demonstrated clinical benefits in patients with acute coronary syndrome; however, little is known about early LLT in patients with stroke. We evaluated the effect of in-hospital initiation of LLT on the clinical outcomes of patients with stroke. The Taiwan Stroke Registry prospectively collected data from patients with acute ischemic stroke or transient ischemic attack. By July 31, 2008, 16,704 adult patients without previous LLT had been admitted and survived to discharge. The study end point was the composite outcome of recurrent stroke, ischemic heart disease, and all-cause death. We examined the effect of LLT at discharge on the clinical outcomes of these patients. A propensity analysis was done to adjust for selection biases in the prescription of LLT. At discharge, 4,032 patients (24%) had received LLT. At 6 months, 206 patients (5.1%) in the LLT group and 964 patients (7.6%) in the non-LLT group had developed ≥1 component of the composite end point (p <0.0001). On multivariate Cox regression analysis, after adjustment for the potential confounders, LLT use at discharge was associated with a lower rate of the composite end point at 6 months (hazard ratio 0.78, 95% confidence interval 0.61 to 0.98, p = 0.013). In conclusion, our data have indicated that LLT has been underused in patients with stroke. In-hospital initiation of LLT was associated with a better clinical outcome in patients with ischemic stroke or transient ischemic attack.
Lipid-lowering therapy (LLT) with statins reduces the incidence of cardiovascular events. Survivors of ischemic stroke have a significantly greater risk of recurrent stroke and myocardial infarction during follow-up. The objectives of the present study were to use the data from a nationwide hospital-based stroke registry to describe the use of LLT at discharge among patients admitted because of ischemic stroke or transient ischemic attack (TIA) and to evaluate the influence of in-hospital initiation of LLT on the prognosis of these patients.
Methods
The Taiwan Stroke Registry, launched on May 1, 2006 and engaged in 37 hospitals around Taiwan, prospectively collected the data of patients who had developed acute ischemic stroke or TIA and were hospitalized within 10 days after onset. The patients fulfilling the following criteria were recruited for the present analysis: age ≥18 years, no LLT use before admission, and surviving until discharged from the hospital. The data were systematically collected according to predetermined registry protocols and included demographic data, vascular risk factors, medications before admission, laboratory test results, treatments during admission, and follow-up information after discharge. The National Institutes of Health Stroke Scale score was used to evaluate the stroke severity at admission. Patients with ischemic stroke or TIA underwent etiology evaluations at the discretion of their attending physicians. The typical evaluations included blood examinations for fasting blood sugar and lipid panels (i.e., total cholesterol, triglycerides, low-density lipoprotein cholesterol, and high-density lipoprotein cholesterol), carotid ultrasonography, neuroimaging studies (i.e., computed tomography or magnetic resonance imaging with angiography), and cardiac studies (i.e., electrocardiography, transthoracic or transesophageal echocardiography). Hypertension or diabetes was considered present when a patient had either received antihypertensive or antidiabetic treatment before admission or hypertension or diabetes mellitus was diagnosed during the hospital stay. Ischemic heart disease was defined as a history of physician-diagnosed coronary heart disease, myocardial infarction, angina pectoris, or unstable angina. Atrial fibrillation was diagnosed when present on the admission electrocardiogram. According to their clinical symptoms and the laboratory data, the patients’ cerebral ischemia etiology was categorized according to the Trial of ORG 10172 in Acute Stroke Treatment (TOAST) classification as large artery atherosclerosis, small vessel occlusion, cardiac embolism, or other specific etiologies. The medication prescribed for long-term use after discharge was categorized as “medication at discharge.” Antiplatelet drug use at discharge was defined as documentation that the patient was given aspirin, clopidogrel, dipyridamole, or ticlopidine, alone or in combination. Antihypertensive drug use at discharge was defined as documentation that the patient was given a diuretic, calcium antagonist, angiotensin-converting enzyme inhibitor, angiotensin receptor antagonist, or β blocker, alone or in combination. Antidiabetic drug use at discharge was defined as documentation that the patient was given sulfonylurea, metformin, thiazolidinedione, an α-glucosidase inhibitor, or insulin, alone or in combination. LLT use at discharge was defined as documentation that the patient was taking a statin, fibrate, nicotinic acid, cholestyramine, or ezetimibe, alone or in combination.
After discharge, the patients were followed up at the original hospital outpatient clinics or by telephone interview by trained research assistants if they could not attend the clinic visits. The outcomes of interest for the present study were a composite end point of all-cause mortality, recurrent stroke, or the occurrence of ischemic heart disease, whichever came first. Recurrent stroke was defined as the sudden appearance of a new neurologic deficit indicating different vascular territories or a worsening of previous neurologic deficits and fitting the definition of TIA or ischemic or hemorrhagic stroke. All the recurrences were verified by neurologists. The diagnosis of ischemic heart disease included physician-diagnosed myocardial infarction, unstable angina, or angina requiring hospitalization for a coronary revascularization procedure. The medical information was obtained from all available medical records. When death occurred outside a hospital or the medical records for the death were unobtainable, death certificates were used.
Continuous variables are presented as the mean ± SD or median and interquartile ranges, and categorical variables are presented as numbers and percentages. The clinical characteristics of the groups were compared using a chi-square test for categorical variables and an unpaired Student t test or Mann-Whitney rank sum test for continuous variables. To adjust for the bias inherent to the decision to provide LLT at discharge, a propensity analysis was performed using logistic regression analysis to determine the probability of a patient receiving LLT at hospital discharge. The variables included in the propensity score model are listed in Table 1 . The c-statistic was calculated to estimate the goodness-of-fit of the propensity score model. The propensity score was compared between patients with and without LLT at discharge and subsequently forced as a continuous variable into a Cox model, along with other predictive variables associated with the outcome of interest.
| Variable | LLT at Discharge (n = 4,032) | No LLT at Discharge (n = 12,672) | p Value |
|---|---|---|---|
| Age (years) | 65.5 ± 11.8 | 68.9 ± 12.9 | <0.0001 |
| Men | 2,305 (57%) | 7,719 (61%) | <0.0001 |
| Hypertension | 3,223 (80%) | 9,629 (76%) | <0.0001 |
| Diabetes mellitus | 1,820 (45%) | 4,642 (37%) | <0.0001 |
| Current smoker | 1,275 (32%) | 3,650 (29%) | 0.001 |
| Previous ischemic stroke | 929 (23%) | 3,313 (26%) | <0.0001 |
| Previous hemorrhagic stroke | 102 (3%) | 334 (3%) | 0.007 |
| Ischemic heart disease | 426 (11%) | 1,518 (13%) | 0.008 |
| Atrial fibrillation | 280 (7%) | 1,964 (16%) | <0.0001 |
| Lipid profile | |||
| Total cholesterol (mg/dl) | 225 ± 46 | 175 ± 37 | <0.0001 |
| Triglycerides (mg/dl) | 196 ± 135 | 126 ± 85 | <0.0001 |
| Low-density lipoprotein (mg/dl) | 149 ± 41 | 110 ± 31 | <0.0001 |
| High-density lipoprotein (mg/dl) | 42 ± 13 | 41 ± 14 | 0.0574 |
| Low-density lipoprotein/high-density lipoprotein ratio | 3.8 ± 1.3 | 2.9 ± 1.1 | <0.0001 |
| Stroke subtype | |||
| Large artery atherosclerosis | 1,177 (30%) | 3,107 (27%) | <0.0001 |
| Small vessel occlusion | 1,791 (47%) | 4,558 (39%) | |
| Cardiac embolism | 242 (6%) | 1,417 (12%) | |
| Other specific etiology | 31 (1%) | 177 (2%) | |
| Undetermined | 607 (16%) | 2,402 (20%) | |
| National Institutes of Health Stroke Scale score at admission | 5 (2–7) | 6 (2–9) | <0.0001 |
| Discharge medications | |||
| Antiplatelet drugs | 3,670 (90%) | 10,466 (82%) | <0.0001 |
| Warfarin | 264 (7%) | 1,071 (9%) | 0.0001 |
| Antihypertensive drugs | 2,120 (53%) | 5,847 (46%) | <0.0001 |
| Antidiabetic drugs | 1,548 (38%) | 3,530 (28%) | <0.0001 |
| Propensity score | 0.3 (0.2–0.3) | 0.2 (0.2–0.3) | <0.0001 |
Kaplan-Meier curves were constructed and stratified by LLT use at discharge. The comparison was made using the log-rank test. Univariate and multivariate Cox proportional hazards models were used to examine the association between the clinical characteristics and LLT with the composite end point. Collinearity between LLT use at discharge with the use of other medications at discharge was assessed with Spearman’s rank correlation. Univariate Cox regression analysis was performed of all baseline characteristics. The predictors with p <0.15 on univariate analysis were then entered into a multivariate model using a forward stepwise selection algorithm with a threshold of 0.05. For the multivariate analyses, LLT and the propensity score were forced into the model. The hazard ratios and their 95% confidence intervals (CIs) were calculated. A 2-tailed p value of <0.05 was considered statistically significant. The statistical analyses were performed using Statistical Analysis Systems, version 9.1 (SAS Institute, Cary, North Carolina).
Results
During the study period to July 31, 2008, 16,704 subjects with ischemic stroke or TIA qualified for the present study. Their mean age was 68.3 years (range 19 to 101), and 60% were men. Of these 16,704 patients, 4,032 (24%) had received LLT before hospital discharge. The baseline clinical characteristics are listed in Table 1 . The serum levels of cholesterol, triglycerides, and low-density lipoprotein and the low-density lipoprotein/high-density lipoprotein ratio were greater in the patients receiving LLT at discharge. The LLT group also was more likely to be prescribed concomitant antiplatelet, antihypertensive, and antidiabetic drugs at discharge. According to the propensity score analysis, the decision to initiate LLT during the hospitalization was independently associated with hypertension (95% CI 1.08 to 1.32, p = 0.0008), large artery atherosclerosis (95% CI 1.15 to 1.46, p <0.0001), small vessel occlusion (95% CI 1.11 to 1.40, p = 0.0002), and concomitant therapy with an antiplatelet drug (95% CI 1.60 to 2.17, p <0.0001), antihypertensive drug (95% CI 1.05 to 1.24, p = 0.002), or antidiabetic drug (95% CI 1.20 to 1.56, p <0.0001). The c-statistic for goodness-of-fit of the propensity score model was 0.64, indicating that the propensity model discriminated satisfactorily between patients who were prescribed LLT at discharge and those who were not.
The follow-up rate was 94.2% at 6 months. During follow-up, 1,170 patients (7%) had developed one of the components of the composite end point, including 206 patients (5.1%) and 964 patients (7.6%) with and without LLT at discharge, respectively (p <0.0001; Table 2 ). The 6-month cumulative risk of the composite end point was 2.1% (95% CI 1.5% to 2.8%) for the LLT group and 4.8% (95% CI 4.3% to 5.3%) for the non-LLT group. The Kaplan-Meier analysis showed a significant benefit associated with using LLT at discharge ( Figure 1 ). The univariate Cox proportional hazards analyses revealed a significant reduction in the risk of the composite end point at 6 months of follow-up in patients who received LLT at discharge (hazard ratio 0.66, 95% CI 0.57 to 0.77; Table 3 ). In the multivariate analyses, after adjusting for the clinical variables and the propensity score, age, diabetes, National Institutes of Health Stroke Scale score at admission, and atrial fibrillation were selected as independent predictors of the composite end point. However, LLT at discharge was significantly associated with a lower rate of the composite end point, with an adjusted hazard ratio of 0.78 (95% CI 0.61 to 0.98; Table 4 ). No clinically relevant interaction terms improved the multivariate model; accordingly, it was judged that no significant interaction was present between LLT at discharge and age, diabetes, or the National Institutes of Health Stroke Scale score at admission. The Spearman correlation coefficients between LLT and the other concomitant drugs (antiplatelet drugs, warfarin, antihypertensive drugs, and antidiabetic drugs) was −0.023 to 0.142. Therefore, the issue of collinearity, i.e., high correlation between two variables to hinder the separate effect assessment of each variable on the outcome, was neglected.
