Author + information
- Received November 28, 2001
- Revision received March 26, 2002
- Accepted June 4, 2002
- Published online September 4, 2002.
- Albert Schömig, MD*,†,* (, )
- Julinda Mehilli, MD*,
- Heidrun Holle, RN*,
- Karin Hösl, RN*,
- Dorejd Kastrati*,
- Jürgen Pache, MD†,
- Melchior Seyfarth, MD†,
- Franz-Josef Neumann, MD†,
- Josef Dirschinger, MD† and
- Adnan Kastrati, MD*
- ↵*Reprint requests and correspondence:
Dr. Albert Schömig, Deutsches Herzzentrum, Lazarettstr. 36, 80636 München, Germany.
Objectives We assessed the influence of statin therapy given after the procedure on one-year survival of patients treated with coronary artery stenting.
Background Coronary artery stenting is currently a common treatment option for patients with symptomatic coronary artery disease (CAD). Although several secondary prevention trials have demonstrated improved survival achieved with statin therapy in conservatively treated patients with CAD, it is not known whether this benefit can also be expected in patients undergoing percutaneous coronary interventions with intraluminal stenting.
Methods This study included 4,520 patients younger than 80 years who underwent coronary artery stenting and were discharged from the hospital in the period October 1995 through September 1999. We compared one-year mortality of 3,585 patients who received statins after stenting with that of 935 patients who did not.
Results The mortality rate at one year was 2.6% among patients who received statins and 5.6% among those who did not. Thus, statin therapy at discharge was associated with an unadjusted odds ratio (OR) of 0.46 (95% confidence interval [CI], 0.33 to 0.65), indicating a 54% reduction in the risk of death at one year. After adjusting for other covariates, the risk reduction associated with statin therapy was 49%, OR 0.51 (95% CI, 0.36 to 0.71). This reduction was observable in most of the subgroups of patients.
Conclusions The results of this nonrandomized study show that statin therapy improves survival after coronary artery stenting independent of patient characteristics recorded on the day of the intervention.
The use of 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors, statins, has consistently reduced cardiovascular risk in both primary (1,2) and secondary (3–5) prevention trials of cardiovascular adverse outcomes. The rationale for these trials derived from the relevant cardiovascular risk factor constituted by elevated plasma cholesterol levels (6) and the cholesterol-lowering effect of statins. Recent data point out that part of the beneficial effect of statins in reducing the cardiovascular risk may be explained with mechanisms independent of the lipid lowering, which may broaden the indications for their use in the clinical practice (7,8).
Secondary prevention trials included patients with stable coronary artery disease (CAD), with plasma lipid levels above a variable threshold and for whom no coronary bypass surgery or balloon angioplasty procedure had been planned (3–5). Recently, a randomized trial including patients with acute coronary syndromes without ST-segment elevation and no scheduled coronary intervention also found a significant advantage with statins (9). Although statin therapy is assumed to be beneficial in a broad range of indications, the proportion of patients who are treated with these drugs is lower than expected (10–12). Two recent studies reported that only 18% to 28% of the patients with acute coronary syndromes are currently given statins at discharge (13,14), although a trend to higher utilization rates has also been observed (15). A contemporary trial comparing coronary stenting with bypass surgery in patients with stable or unstable angina also indicated that only about 35% of these patients were treated with statins after the intervention (16).
It is still unclear what benefit statins may confer to patients who undergo a percutaneous coronary intervention (PCI) in the setting of stable angina pectoris or acute coronary syndromes. Administration of statins before and after an elective plain coronary angioplasty procedure had no effect on restenosis (17,18), yet it improved survival free of myocardial infarction (MI) in one trial (17). Therefore, we assessed the value of statins given after the procedure by analyzing the one-year mortality in this cohort study including a large series of patients treated with coronary artery stenting.
This study includes a consecutive series of patients with CAD who underwent coronary stenting in the period between October 1995 through September 1999 in Deutsches Herzzentrum and 1. Medizinische Klinik rechts der Isar. During this period the stent implantation was performed in 5,030 patients. We excluded four categories of patients: patients 80 years of age or older (n = 341), patients with malignancies and life expectancy <1 year (n = 57), patients in cardiogenic shock before intervention (n = 73), as well as those who died during the hospital stay corresponding to the index procedure (n = 39). Therefore, the study population consisted of 4,520 patients. All of the patients had given their written informed consent for the intervention and collection of their follow-up information after discharge according to a protocol approved by the Institutional Ethics Committee.
The stent placement was performed according to standard protocols. Intravenous heparin and aspirin were administered during the procedure. Patients with baseline or procedural characteristics suggesting a higher risk for stent thrombosis received additional periprocedural therapy with abciximab; the decision was always made by the responsible operator. After the intervention all patients received aspirin 100 mg twice daily indefinitely and ticlopidine 250 mg twice daily for four weeks. The concomitant therapy prescribed at discharge was left to the discretion of the attending physician.
Data were collected prospectively. Demographic and clinical data were collected at admission. Angiographic and procedural data related to the stented lesion were collected during the intervention. Lesions were classified according to the modified American College of Cardiology/American Heart Association grading system in type A, B1, B2, and C; lesions falling into the B2 and C categories were defined as complex (19). Coronary lumen dimensions were measured using an automated edge detection system. All information was entered into a central computer database. After discharge patient information was updated to include complications incurred during the hospital stay and the medication prescribed at discharge.
The postdischarge follow-up included a telephone interview at 30 days, a clinical visit at six months, and an additional telephone interview at one year. At one year all information provided by the patient, the referring physician, or the outpatient clinic as well as data derived from eventual hospital readmission records were entered into the computer database. Complete follow-up information was available for 4,419 of the 4,520 study patients.
The primary end point of the study was one-year total mortality. On the basis of the information obtained from hospital records, death certificates, or phone contact with relatives of the patient or attending physician, deaths were classified as cardiovascular or noncardiovascular. We also assessed other adverse clinical events such as MI and target vessel revascularization (percutaneous transluminal balloon angioplasty [PTCA] or aortocoronary bypass surgery) because of angiographic restenosis and symptoms or signs of ischemia. The diagnosis of MI was based on the presence of new pathological Q waves or a value of creatine kinase or its MB isoenzyme at least 3× the upper limit (20). Creatine kinase was determined before and immediately after the procedure, every 8 h for the first 24 h postprocedure and daily afterwards until discharge.
The differences between the groups with and without statins were assessed using the chi-square test for categorical data and the t test for continuous data. Survival was assessed using the Kaplan-Meier method; differences in survival parameters were tested for significance by means of the log-rank test, and odds ratios (OR) plus 95% confidence intervals (CI) were computed. The homogeneity of the treatment effect across strata was assessed by the test of Breslow and Day (21).
Special attention was given to two additional issues. First, because this study comprised patients treated in a four-year period, we tried to account for potential time-related differences between statin and nonstatin patients (e.g., technological advances, greater experience) by performing separate analyses for each one-year period (first, second, third, and fourth year from the beginning of the study) and assessing if there was a homogenous treatment effect over the years. Second, we performed a specific analysis to address the potential relation between patients’ baseline demographic and clinical characteristics and the likelihood that the physician prescribed statins at discharge. For this purpose we applied a logistic regression model including age, gender, systemic arterial hypertension, smoking habit, cholesterol level, diabetes, a history of MI or aortocoronary bypass surgery, severity of angina at admission, multivessel disease, left ventricular (LV) function, vessel in which the lesion was located, lesion complexity, restenotic lesion, lesion length, vessel size, diameter stenosis before the intervention, and length of stents implanted as covariates. Using this model we calculated a propensity score (22,23) for each patient indicating the estimated probability of being exposed to statin treatment at discharge. Then, on the basis of the propensity score, the population was divided into quartiles representing four categories, from the category with the lowest probability to that with the highest probability of having statins prescribed at discharge. The analysis of mortality according to statin status was also carried out for each quartile, separately.
We also used multivariate methods to assess the independent impact of statins on one-year mortality. For this purpose we applied a Cox proportional hazards model including statin status, age, gender, systemic arterial hypertension, smoking habit, cholesterol level, diabetes, a history of MI or aortocoronary bypass surgery, severity of angina at admission, multivessel disease, LV function, vessel in which the lesion was located, lesion complexity, restenotic lesion, lesion length, vessel size, diameter stenosis before the intervention, length of stents implanted, periprocedural administration of abciximab, and concomitant therapy with beta-adrenergic blocking agents or angiotensin-converting enzyme (ACE) inhibitors. In addition, the one-year period in which the intervention was performed as well as the propensity score for each patient were also included as potential confounders. The variables retained in the final model were determined using a fast backward factor elimination technique (24). To calculate the adjusted OR, we used the equivalent parameter of hazard ratio derived from the Cox model. The significance level was set at p < 0.05.
Of the 4,520 patients included in this study, 3,585 patients were discharged on statin treatment. The type of statin drug given to the patient after the intervention was not entered into the database, but, according to the pharmacy data on statin use in our department during the study period, simvastatin was prescribed in 49% of the cases, atorvastatin in 25%, pravastatin in 16%, lovastatin in 7%, cerivastatin in 2%, and fluvastatin in 1%. Table 1 shows the demographic, clinical, angiographic, and procedural characteristics as well as the concomitant therapy at discharge of the patients with and without statins.
Figure 1 displays the one-year cumulative mortality curves for both groups. The mortality rate at one year was 2.6% among patients with statins and 5.6% among those without statins. Thus, statin therapy at discharge was associated with an unadjusted OR of 0.46 [95% CI, 0.33 to 0.65], indicating a 54% reduction in the risk of death at one year, p < 0.001. If we consider only patients with statins, there was no difference in one-year mortality between the 1,384 patients who had already received statins before admission (2.7%) and the 2,201 patients who first received statins after admission (2.5%, p = 0.71). Sixteen patients (0.5%) who received statins and six (0.6%) patients without statins died due to noncardiovascular causes during the one-year follow-up (p = 0.44). If we also consider the beta-blocker therapy status, mortality was 2.5% among the 3,296 patients who received both statins and beta-blockers, 4.5% among the 289 patients who received only statins, 5.1% among the 412 patients who received only beta-blockers, and 6.0% among the 523 patients who received neither statins nor beta-blockers (p < 0.001).
The complication of MI was observed in 4.5% of the patients with statins and 5.1% of the patients without statins (p = 0.38). Sixty-six percent of all cases of MI occurred within the first 24 h after the intervention. By one year the combined incidence of death or MI was 6.7% in the statin therapy group and 10.1% in the group without statins, which corresponds to an OR of 0.66 [0.52 to 0.84], p < 0.001. Target vessel revascularization was required in 19.9% of the patients with statins and 18.5% of those without statins (p = 0.34).
We analyzed the effect of statin therapy for each one-year study period separately. Figure 2 shows the results of this analysis. The proportion of patients with statins varied between 64% in the first year and 90% in the third year. The mortality rate varied between 3.0% in the third year and 3.4% in the second year (p = 0.95). Notably, statin therapy at discharge was associated with a consistent risk reduction of 52% to 60% across the four periods (p = 0.96 from the test for homogeneity).
Figure 3 shows that statins were associated with a significant risk reduction for death in various subsets. Although to a lesser degree than patients with cholesterol levels >200 mg/dl, patients with lower cholesterol levels also benefit significantly from statin therapy. Based on the analysis according to the extent of CAD at the time of the index intervention, the benefit of statin therapy seems to be largely confined to patients with multivessel disease. Both patients with single- and multilesion interventions benefited from statin therapy. In addition, in the small subgroup of patients who were not treated with beta-blockers, the benefit associated with statin therapy was lower and not significant.
On the basis of background information, we constructed a logistic regression model to calculate the propensity score for each patient, that is, the likelihood that the patient is given statins (see Methods section). The variables that were significantly associated with a higher probability for the patient to receive statins were (in decreased order of significance): cholesterol level, a history of MI, systemic arterial hypertension, younger age, more severe stenosis, and a greater extent of stenting during the procedure. We divided the population in four groups with an equal number of patients (quartiles) on the basis of their propensity scores, with the lowest quartile representing those less likely to be given statins. Figure 4 shows the results of the specific analyses according to propensity score. The one-year mortality varied between 2.8% in the second, third, and fourth quartiles and 4.6% in the lowest quartile (p = 0.024). The OR associated with statin therapy show a more or less significant risk reduction in all of the quartiles, without a significant difference in treatment effect (p = 0.37 from the homogeneity test).
Using Cox proportional hazard model, we calculated the adjusted risk reduction associated with statin therapy after accounting for other covariates. Statins at discharge were associated with an adjusted risk reduction of 49%, OR 0.51 [0.36 to 0.71]. Figure 5 shows the significant predictors of one-year mortality in addition to statin therapy at discharge (in decreased order of significance): age, diabetes, LV function, extension of CAD, and extent of stenting (length of stents implanted during the procedure). In addition, we constructed separate multivariate models for the use of beta-blockers and ACE inhibitors after removing statin therapy. The use of beta-blockers was associated with a reduced risk of death as shown by an adjusted OR of 0.56 [0.39 to 0.81]. Similarly, the use of ACE inhibitors was also associated with a reduced risk of death, adjusted OR 0.67 [0.47 to 0.95]. No significant independent association was found between the use of abciximab and one-year mortality.
Statins exert multiple cellular effects including favorable effects on plasma lipoproteins, endothelial function, plaque architecture and stability, thrombosis, inflammation, and immune response (7,8,25). Part of these effects is induced by lipid-independent mechanisms, which render statins a potentially useful therapy in a wide spectrum of patients with CAD (7).
In a recent study that excluded patients with acute MI, Chan et al. (26) found that patients undergoing PTCA or stenting showed a reduction of mortality at 30 days and six months if they were taking statins. Walter et al. (27) reported on reduced adverse event rates after stenting in patients with high C-reactive protein levels or in carriers of the PlA2 allele of the platelet glycoprotein IIIa gene (28) receiving statins, but their effect on mortality could not be investigated due to the limited number of patients. The present study provides information about the value of statin therapy given after coronary artery stenting in a large, consecutive series of patients presenting with a broad spectrum of CAD. The study shows that patients who receive statins after coronary artery stenting may be provided with significantly improved chances of survival when compared with those who do not receive these agents. The improvement conveyed by statins was independent from the influence of other demographic, clinical, angiographic, and procedural factors as well as from the type of concomitant pharmacologic treatment. If combined with the results of previous secondary prevention trials including patients with CAD treated conservatively (3–5,13,14), the present findings along with those of the recent studies cited above suggest that the spectrum of patients who benefit from statin therapy is broader and includes those treated with PCI as well.
Practices relating to medication with statins are influenced by a wide range of nonclinical and clinical factors as demonstrated by the study of Stafford et al. (11) in the U.S. A more unfavorable insurance status, often reflecting a lower socioeconomic level of the patient, has been identified as the most relevant nonclinical factor associated with less use of costly services such as treatment with statins (11). This factor is, however, unlikely to have interfered with the decision to give statins after stenting in the present study, because all of our patients were insured and the costs related to this therapy were covered by the insurance company. Another possible source of bias is the inhomogeneity of clinical characteristics. Several efforts were made to reduce the risk of bias connected with the nonrandomized nature of our study. We excluded from the analysis all patients with highly compromised chances of survival by one year after the intervention such as elderly patients over 80 years, those with malignancies, and patients with cardiogenic shock. We controlled for naturally occurring differences in background characteristics between the treatment group and the control group by using the propensity score technology (22,23). This enabled the calculation of the likelihood for each patient to be given statins (propensity score) departing from the information available before the decision was taken by the attending physician. Although there were certain patient characteristics associated with a higher probability for the patients to receive statins after the intervention, two aspects of our analysis point out the strong independent role of statins in the improved one-year survival. First, the treatment effect was homogenous across the groups of patients with different propensity scores showing that, regardless of how likely the patient was to be given statins, those who actually received this medication had a lower mortality than those left without this therapy. Second, despite the apparent relation between propensity score and one-year mortality as shown by the higher risk only confined to the lowest quartile, it was the statin therapy, and not the propensity score, that emerged as an independent predictor of mortality from the multivariate model including both variables. The multivariate analysis demonstrated the independent positive impact of statins on mortality in contradistinction to the negative role of well-known risk factors such as an older age, diabetes, multivessel disease, impaired LV function, and a greater stented length.
Despite a number of efforts to correct for the potential selection bias, the present study cannot be substituted for a randomized trial nor can it account for nonrecognized baseline differences between the two treatment groups. Another limitation of the study is that it cannot provide mechanistic insights into the beneficial impact of statins on survival after stenting. The incidence of MI was not significantly different between statin and nonstatin patients. The design of the study enabled, however, a comprehensive assessment of the incidence of this event only during the first days after the procedure where cardiac enzymes were systematically determined. In fact, the majority of cases of MI were observed within the first 24 h after stenting. In addition, the similar incidence of target vessel revascularization between the two groups suggests that the reduction of mortality by statin therapy is probably not mediated by an influence on restenosis. In the same line, patients with multivessel coronary disease were those who most benefited from statins, suggesting that the predominating effect of this therapy is exerted beyond the site of balloon dilation. Although concordant with the results of randomized trials assessing the effect of statins on lumen renarrowing after balloon angioplasty (17,18), our restenosis findings are in contrast with those of a previous retrospective analysis of 525 patients after stent implantation (29). In the latter study, Walter et al. (29) found an incidence of target vessel revascularization of 27.9% in patients treated with statins and 36.7% in those without statins (p < 0.05). Although we are unable to offer an explanation for this difference, the unusually high revascularization rate in the latter study suggests that Walter et al. (29) have included a population with a particularly high risk for restenosis. An additional limitation of the present study is the absence of cholesterol levels at follow-up that could have enabled the assessment of the relation between the lipid-lowering effect of statins and their influence on outcome. Finally, a follow-up period longer than one year would have provided more useful information on the influence of statins in patients undergoing PCI.
The results of this nonrandomized study strongly suggest that statins improve survival after coronary artery stenting independent of patient characteristics recorded the day of the intervention. If combined with the findings of randomized, secondary prevention trials with statins, the results of the present study support the use of statins in all patients who undergo coronary stent placement and have no contraindication to this therapy.
- angiotensin-converting enzyme
- coronary artery disease
- confidence interval
- left ventricular
- myocardial infarction
- odds ratio
- percutaneous coronary intervention
- percutaneous transluminal coronary angioplasty
- Received November 28, 2001.
- Revision received March 26, 2002.
- Accepted June 4, 2002.
- American College of Cardiology Foundation
- Anderson K.M.,
- Wilson P.W.F.,
- Odell P.M.,
- Kannel W.B.
- Maron D.J.,
- Fazio S.,
- Linton M.F.
- Grundy S.M.,
- Balady G.J.,
- Criqui M.H.,
- et al.
- Stafford R.S.,
- Blumenthal D.,
- Pasternak R.C.
- Serruys P.W.,
- Foley D.P.,
- Jackson G.,
- et al.
- Ellis S.G.,
- Vandormael M.G.,
- Cowley M.J.,
- et al.
- Breslow N.E.,
- Day N.E.
- Joffe M.M.,
- Rosenbaum P.R.
- Harrell F.E. Jr.
- Laufs U.,
- Endres M.,
- Custodis F.,
- et al.
- Chan A.W.,
- Bhatt D.L.,
- Chew D.P.,
- et al.
- Walter D.H.,
- Fichtlscherer S.,
- Britten M.B.,
- et al.
- Walter D.H.,
- Schachinger V.,
- Elsner M.,
- et al.