Long‐Term Effectiveness and Safety of Initiating Statin Therapy After Index Revascularization In Patients With Peripheral Arterial Occlusive Disease

Abstract

Background

An increasing number of patients with a peripheral arterial occlusive disease were put on statins during the past years. This study assessed whether statin therapy was effective and safe for these new users.

Methods and Results

Using health insurance claims data from Germany’s second‐largest insurance fund, BARMER, we identified patients with peripheral arterial occlusive disease who had index revascularization between 2008 and 2018 without prior statin therapy. We compared patients with and without statin therapy in addition to antithrombotics during the first quarter after discharge (new users versus nonusers). Outcomes were all‐cause mortality, cardiovascular events, and incident major amputation for effectiveness and incident diabetes mellitus and incident myopathy for safety. Propensity score matching was used to balance the study groups. All analyses were stratified into patients with chronic limb‐threatening ischemia and intermittent claudication. A total of 22 208 patients (mean age 71.1 years and 50.3% women) were included in the study. In 10 922 matched patients, statin initiation was associated with lower all‐cause mortality (chronic limb‐threatening ischemia: hazard ratio [HR], 0.75 [95% CI, 0.68–0.84]; intermittent claudication: HR, 0.80 [95% CI, 0.70–0.92]), lower risk of major amputation in patients with chronic limb‐threatening ischemia (HR, 0.73; 95% CI, 0.58–0.93) and lower risk of cardiovascular events (hazard ratio, 0.80; 95% CI, 0.70–0.92) in patients with intermittent claudication during 5 years of follow‐up. Safety outcomes did not differ among the study groups.

Conclusions

Initiating statin therapy in patients with peripheral arterial occlusive disease after index revascularization is efficient and safe with an effect size comparable to earlier studies. Awareness campaigns for evidence‐based optimal pharmacological treatment among patients are recommended.

Nonstandard Abbreviations and Acronyms

ATC

Anatomical Therapeutic Chemical

CLTI

chronic limb‐threatening ischemia

IC

intermittent claudication

ICD‐10‐GM

International Classification of Diseases, Tenth Revision, German Modification

PAOD

peripheral arterial occlusive disease

Clinical Perspective

What is New?

• This is the first study assessing long‐term benefits and harms of initiating statin therapy after lower limb revascularization for symptomatic peripheral arterial occlusive disease in a real‐world setting.

• The proportion of patients initiating statin therapy doubled throughout the study period but is still substantially below societal guideline recommendations.

• Initiating statin therapy is effective and safe in patients with intermittent claudication or chronic limb‐threatening ischemia.

What Are the Clinical Implications?

• Awareness campaigns emphasizing the importance of prescribing statins in the follow‐up of patients should involve general practitioners and other medical specialist disciplines.

• The surveillance of patients after invasive revascularizations should comprise regular interviews for optimal pharmacological treatment and patient compliance.

• Health insurance claims can be used to automatically identify patients with potential for improvement in secondary prevention.

During the past decades, various pharmacological therapies for vascular diseases became available, effectively preventing cardiovascular events. ¹ Valid guidelines consistently emphasize the importance of the prescription of statins in patients with peripheral arterial occlusive disease (PAOD) irrespective of their concomitant risk profile as a cornerstone of secondary prevention. ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ Patients with PAOD are particularly dependent on optimal pharmacological treatment because of considerably elevated risks of cardiovascular events, acute limb ischemia, and amputation markedly impairing quality of life. ⁸ , ⁹ , ¹⁰ Yet, this subgroup exhibits particularly low utilization rates of statins as compared with patients with coronary artery disease or a history of stroke. ¹¹ , ¹²

The underutilization of statin therapy has been predominantly ascribed to the lack of awareness about risks and therapy options among providers and patients ¹² , ¹³ and concerns about adverse reactions such as myopathy and onset of diabetes mellitus. ¹⁴ , ¹⁵ , ¹⁶ Given the solid evidence of the benefits of statin therapy and, at the same time, the sharp increase in hospitalizations and costs related to PAOD, experts urge providers to push efficient secondary prevention more insistently. ¹⁷

Recent observational studies confirmed that statins are effective and safe in both low‐ and high‐risk patients with PAOD ¹⁸ , ¹⁹ and offer additional benefits at high‐intensity doses. ¹⁸ , ²⁰ Although these studies differed in study design and sample composition, they arrived at similar conclusions comparable to findings from randomized controlled trials.

Fueled by intensified guideline recommendations, statin utilization rates increased throughout the past decade among patients with PAOD. ²¹ In the current study, we determine the success of the expansion of statins in PAOD treatment and the impact on major outcomes in the longer term. This may contribute to the understanding of how effects measures in randomized controlled trials translate to the heterogenous real‐world population and to what extent the benefit of the drug diminishes as prescription rates increase. This concept was recently discussed for other domains of health care. ²²

Our study employed a large nationwide database for quantifying the long‐term effectiveness and safety of initiating statin therapy in symptomatic patients with PAOD after index revascularization. We aimed to quantify to what degree the initiation of statin therapy prolongs survival, reduces the risk for major amputation and cardiovascular events, and potentially increases the risk of the onset of diabetes mellitus or myopathy.

Methods

The data that support the findings of this study are available from the corresponding author on reasonable request. Our study complies with the Declaration of Helsinki 2013. Several review boards determined that using factual anonymized data from claims or national statistics retrospectively is not considered human subject research because deidentified data sets were used. All analyses were in accordance with the European Union’s General Data Privacy Regulation, taking into account the theoretical concept of k‐anonymity. Thus, patient informed consent was not obtained for this retrospective secondary data analysis. Our study is part of a larger project on outcomes of patients with PAOD after revascularization. Further details regarding this project can be found in the published study protocol (clinicaltrials.gov NCT03909022). ²³

Sample and Database

The longitudinal data of Germany’s second‐largest insurance fund, BARMER, includes the outpatient and inpatient medical care provided to ≈9.4 million German citizens (13.2% of Germany’s population) involving >21 million hospitalizations between January 1, 2008, and December 31, 2018. The BARMER cohort is similar to Western European countries and has been widely used for research projects. ²⁴ , ²⁵ A regular random sample validation of internal and external validity is performed by the Medical Service of the Health Funds in Germany, and various peer‐reviewed validation studies have been previously published. ²⁶ , ²⁷

The diagnoses and comorbidities routinely collected in health insurance claims data follow the commonly accepted international standard for reporting diseases and health conditions using World Health Organization International Classification of Diseases, Tenth Revision, German Modification (ICD‐10‐GM), operations and procedures codes, and the German version of the international Anatomical Therapeutic Chemical (ATC) classification.

In our analyses, we created separate cohorts for Fontaine stage II labeling intermittent claudication (IC) and Fontaine stages III to IV labeling chronic limb‐threatening ischemia (CLTI) (for detailed coding see Table S1). We included patients with a primary diagnosis of IC (I70.22 until 2014 and I70.21‐22 since 2015) and CLTI (I70.22‐24 until 2014 and I70.23‐25 since 2015) or IC and CLTI as a secondary diagnosis in combination with a primary diagnosis of diabetic foot syndrome (E10.50‐51, E10.7, E11.50‐51, E11.7), other peripheral vascular diseases (I73), arterial embolism and thrombosis (I74), cellulitis of the finger and toe including acute lymphangitis (L03.01‐02, L03.11), or chronic ulcer of skin and gangrene (L98.4, R02) using the ICD‐10‐GM.

The index admission for symptomatic PAOD (denoted as index stay) was identified between January 1, 2008, and December 31, 2018, with follow‐up until December 31, 2018. We used 3‐year lookback in the BARMER data set ²⁶ to create relevant comorbidities (available data going back to 2005) and to ensure index admission for symptomatic PAOD.

Statin‐naive patients (statins: ATC coding C10AA, C10BA, or C10BX) without statin utilization for at least 3 years before index stay were selected for inclusion in our study. We further included only patients with at least 1 prescription for an antithrombotic agent (eg, acetylsalicylic acid, clopidogrel, or oral anticoagulation) during the first quarter after discharge to prevent selection bias caused by prevalent users. ⁵

The following patients were excluded: patients aged <40 years, patients with prior major amputation or recorded myopathy (outpatient or inpatient), patients discharged without revascularization (amputation only or best medical treatment only) and death, patients with major amputation, and patients with cardiovascular events (myocardial infarction, stroke or transient ischemic attack) during the first quarter after discharge. Further, we excluded patients treated with other lipid‐lowering drugs than statins or statin combinations during the first quarter after discharge to ensure that all patients were eligible for statin prescription. Few cases with missing information on age, sex, or follow‐up (≈0.5%) were removed using complete case deletion.

Study Variables

We identified new users as patients filling at least 1 prescription for statins during the first quarter after index stay. Patients not filling a statin prescription during the quarter after index stay were denoted as nonusers.

The primary outcome was all‐cause mortality during follow‐up. In German claims data, the information about the death of the insured person is complete and validated. ²⁷

Secondary outcomes were incident major amputation and cardiovascular events (myocardial infarction, stroke, or transient ischemic attack), obtained from primary and secondary inpatient diagnoses.

Safety outcomes were incident diabetes mellitus and incident myopathy. Specifically, incidence was defined as first diagnosis after discharge from index stay. For assessing the risk of developing diabetes mellitus, we further excluded patients with diabetes mellitus during the 3 years before the index stay. For measuring incident outcomes, we evaluated both outpatient and inpatient diagnoses and, in the case of diabetes mellitus, also the prescription of oral and parenteral antidiabetic agents. ²⁸ For detecting myopathy, we used the broader list of conditions previously used for the identification of statin‐associated myopathy in German claims data. ²⁹

All outcomes were recorded at 3 months after discharge from index stay until the first event or end of study time. Follow‐up times were censored after 5 years to compute robust 5‐year event probabilities.

Statistical Analysis

We summarized baseline characteristics of the patients with means and SDs for normally distributed variables, medians and interquartile ranges for nonnormally distributed variables, or percentages and standardized differences for discrete variables. Cochrane Armitage trend test was used to test the change in the proportion of statin therapy over the calendar year. To balance study groups, nearest neighbor propensity score matching was applied using the following variables: discharge year; age; sex; van Walraven score: category 0 (−19 to −1 points), category 1 (0 points), category 2 (1–9 points), and category 3 (10 points and more); congestive heart failure, cardiac arrhythmias; chronic pulmonary disease; renal failure; depression; prior stroke or transient ischemic attack; smoking; obesity; prior myocardial infarction; dyslipidemia; coronary artery disease; diabetes mellitus (complicated and uncomplicated); cancer; hypertension; prior outpatient diagnosis of PAOD; number of different prescriptions; number of previous inpatient admissions; number of prior PAOD outpatient visits; invasive procedures (peripheral vascular intervention, peripheral vascular intervention, or open‐surgical revascularization); and hospital length of stay. The linear van Walraven sum score and most of the comorbidities are based on the list of Elixhauser categories, also used in various other claims data analyses. ³⁰ We evaluated the validity of these comorbidities over time thoroughly in an earlier study. ²⁵

Incident diabetes mellitus was assessed in a reduced cohort additionally excluding patients with any inpatient or outpatient diagnosis of diabetes mellitus or prescription for antidiabetic agents during the 3 years before the index stay. Since this exclusion affected the balance of the study groups, we performed a second propensity score matching for this cohort (without diabetes mellitus as a matching variable).

Outcomes were estimated using Kaplan‐Meier curves (with log‐rank test) and Cox proportional hazards models. Using hazard ratios (HRs), we computed the 5‐year probability of each outcome with 95% CIs for each study group.

For sensitivity analyses, we estimated Cox proportional hazards models in the unmatched data using the matching variables as covariates. We estimated models adjusting for co‐medication during the 3 months after discharge (angiotensin II receptor blockers or angiotensin‐converting enzyme inhibitors [ATC code C09A‐D], calcium channel blockers [ATC code C08], β‐blockers [ATC code C07], and oral anticoagulation [ATC code B01AA, B01AE, or B01AF]), and models stratified by sex, models stratified by age [age <75 years and ≥75 years], models stratified by calendar time (2008–2012 and 2013–2018), and models stratified by statin intensity (low‐to‐moderate and high).

The data processing was performed with software SAS version 9.04 (SAS Institute Inc) and R software version 3.3.3 (package survival and MatchIt, ³¹ The R Foundation for Statistical Computing). We reported results using the Reporting of Studies Conducted Using Observational Routinely Collected Health Data statement, the Strengthening the Reporting of Observational Studies in Epidemiology statement, ³² and following international recommendations on medical device evaluation studies. ³³

Results

Unmatched Study Sample

A total of 22 208 symptomatic patients with PAOD (22.2% with CLTI, 50.3% women; Table S2) were hospitalized during the study period from January 1, 2008, to December 31, 2018, undergoing invasive revascularization (Figure 1 and Table 1). The average age was 71.1±11.6 years (median follow‐up, 1277 days; interquartile range, 616–1827). In our study sample, the annual proportion of new users after discharge increased between 2008 and 2018 from 17% to 34% in patients with CLTI (P<0.001) and from 22% to 43% in patients with IC (P<0.001) (Figure S1).

Figure 1. Study flow chart.

 

Table 1.

Baseline Characteristics of the Unmatched Study Cohort (N=22 208)

Variable	New Users, CLTI n=2367	Nonusers, CLTI n=7096	Standardized Differences*	New Users, IC n=4227	Nonusers, IC n=8518	Standardized Differences*
Age, mean (SD), y	71.64 (11.73)	76.09 (11.52)	0.382	66.44 (10.27)	69.00 (10.70)	0.245
Women, n (%)	1208 (51.0)	3969 (55.9)	0.098	1981 (46.9)	4008 (47.1)	0.004
Van Walraven score >9, n (%)	698 (29.5)	3052 (43.0)	0.284	454 (10.7)	1503 (17.6)	0.199
Congestive heart failure, n (%)	445 (18.8)	1970 (27.8)	0.213	275 ( 6.5)	841 (9.9)	0.123
Cardiac arrhythmias, n (%)	519 (21.9)	2414 (34.0)	0.272	373 ( 8.8)	1288 (15.1)	0.195
Chronic pulmonary disease, n (%)	302 (12.8)	1130 (15.9)	0.09	481 (11.4)	1154 (13.5)	0.066
Renal failure, n (%)	593 (25.1)	2336 (32.9)	0.174	511 (12.1)	1235 (14.5)	0.071
Depression, n (%)	176 (7.4)	591 (8.3)	0.033	196 (4.6)	449 (5.3)	0.029
Prior stroke or TIA, n (%)	99 (4.2)	420 (5.9)	0.079	72 (1.7)	205 (2.4)	0.05
Smoking, n (%)	448 (18.9)	882 (12.4)	0.179	1057 (25.0)	1794 (21.1)	0.094
Obesity, n (%)	206 (8.7)	674 (9.5)	0.028	304 (7.2)	712 (8.4)	0.044
Prior myocardial infarction, n (%)	127 (5.4)	293 (4.1)	0.058	101 (2.4)	196 (2.3)	0.006
Dyslipidemia, n (%)	959 (40.5)	1023 (14.4)	0.611	1919 (45.4)	1261 (14.8)	0.707
Coronary artery disease, n (%)	437 (18.5)	1439 (20.3)	0.046	434 (10.3)	1132 (13.3)	0.094
Diabetes mellitus, any, n (%)	821 (34.7)	2658 (37.5)	0.058	692 (16.4)	1787 (21.0)	0.118
Cancer, any, n (%)	120 ( 5.1)	464 (6.5)	0.063	166 (3.9)	514 (6.0)	0.097
Hypertension, n (%)	1717 (72.5)	5413 (76.3)	0.086	2771 (65.6)	5820 (68.3)	0.059
Prior outpatient diagnosis PAOD, n (%)	651 (27.5)	2288 (32.2)	0.104	1049 (24.8)	2816 (33.1)	0.183
No. of different prescriptions, median (IQR)	11.00 (5.00–17.00)	14.00 (9.00–21.00)	0.396	8.00 (5.00–13.00)	10.00 (6.00–16.00)	0.304
No. of previous inpatient admissions, total (including index), median (IQR)	2.00 (1.00–3.00)	2.00 (1.00–4.00)	0.237	1.00 (1.00–2.00)	2.00 (1.00–3.00)	0.207
No. of prior PAOD outpatient visits, median (IQR)	1.00 (0.00–3.00)	1.00 (0.00–5.00)	0.145	1.00 (0.00–2.00)	1.00 (0.00–4.00)	0.213
Invasive procedure: OSR, n (%)	914 (38.6)	2714 (38.2)	0.008	876 (20.7)	2317 (27.2)	0.152
Hospital length of stay, days, median (IQR)	12.00 (7.00–21.00)	12.00 (7.00–22.00)	0.009	4.00 (3.00–8.00)	4.00 (3.00–9.00)	0.082

CLTI indicates chronic limb‐threatening ischemia; IC, intermittent claudication; IQR, interquartile range; OSR, open surgical revascularization; PAOD, peripheral arterial occlusive disease; and TIA, transient ischemic attack.

^*Values >0.1 were deemed to indicate meaningful differences.

In the CLTI group, when compared with nonusers, new users were younger (71.6 versus 76.1 years), less often women (51.0% versus 55.9%), and more often smokers (18.9% versus 12.4%) (Table 1). Moreover, new users experienced fewer comorbidities, with a van Walraven score of >9 points in 29.5% versus 43.0% when compared with nonusers. Dyslipidemia was diagnosed more often in new users than in nonusers (40.5% versus 14.4%). New users were treated less frequently and less intensely before index admission with respect to the number of different prescriptions, prior PAOD outpatient diagnosis, previous inpatient admissions, and prior PAOD outpatient visits.

In the IC group, when compared with nonusers, new users were younger (66.4 versus 69.0 years) and more often smokers (25.0% versus 21.1%), but there were no sizable differences with respect to sex. New users experienced fewer comorbidities, with a van Walraven score of >9 points in 10.7% versus 17.6% when compared with nonusers. Dyslipidemia was diagnosed in 45.4% of the new users and 14.8% of the nonusers. New users were treated less frequently and less intensely before index admission with respect to the number of different prescriptions, prior PAOD outpatient diagnosis, previous inpatient admissions, and prior PAOD outpatient visits.

The proportion of patients undergoing open surgical revascularization (bypass, endarterectomy) when compared with endovascular revascularization was less prevalent in new users than in nonusers for IC (20.7% versus 27.2%).

Matched Study Sample

Using the propensity score, we matched 10 922 patients with PAOD: 4 224 (38.7%) patients with CLTI and 6 698 patients with IC (Figure 1 and Table S3). Demographics and comorbidities of the matched study sample are presented in Table 2. In total, 89.2% new users could be matched to nonusers and no clinically relevant standardized differences among the study groups remained after matching.

Table 2.

Baseline Characteristics of the Matched Study Cohort (N=10 922)

Variable	New Users, CLTI n=2112	Nonusers, CLTI n=2112	Standardized Differences*	New Users, IC n=3349	Nonusers, IC n=3349	Standardized Differences*
Age, mean (SD), y	72.52 (11.64)	72.67 (12.05)	0.012	67.10 (10.31)	67.34 (10.47)	0.023
Women, n (%)	1100 (52.1)	1106 (52.4)	0.006	1564 (46.7)	1589 (47.4)	0.015
Van Walraven score >9, n (%)	673 (31.9)	723 (34.2)	0.05	411 (12.3)	438 (13.1)	0.024
Congestive heart failure, n (%)	422 (20.0)	444 (21.0)	0.026	251 (7.5)	263 (7.9)	0.013
Cardiac arrhythmias, n (%)	494 (23.4)	555 (26.3)	0.067	334 (10.0)	369 (11.0)	0.034
Chronic pulmonary disease, n (%)	279 (13.2)	280 (13.3)	0.001	403 (12.0)	405 (12.1)	0.002
Renal failure, n (%)	562 (26.6)	602 (28.5)	0.042	417 (12.5)	428 (12.8)	0.01
Depression, n (%)	161 (7.6)	170 (8.0)	0.016	161 (4.8)	157 (4.7)	0.006
Prior stroke or TIA, n (%)	96 (4.5)	111 (5.3)	0.033	62 (1.9)	69 (2.1)	0.015
Smoking, n (%)	362 (17.1)	354 (16.8)	0.01	794 (23.7)	804 (24.0)	0.007
Obesity, n (%)	192 (9.1)	189 (8.9)	0.005	249 (7.4)	263 (7.9)	0.016
Prior myocardial infarction, n (%)	111 (5.3)	116 (5.5)	0.01	76 (2.3)	79 (2.4)	0.006
Dyslipidemia, n (%)	705 (33.4)	714 (33.8)	0.009	1041 (31.1)	1046 (31.2)	0.003
Coronary artery disease, n (%)	387 (18.3)	425 (20.1)	0.046	369 (11.0)	393 (11.7)	0.023
Diabetes mellitus, any, n (%)	741 (35.1)	771 (36.5)	0.03	591 (17.6)	586 (17.5)	0.004
Cancer, any, n (%)	114 (5.4)	120 (5.7)	0.012	138 (4.1)	149 (4.4)	0.016
Hypertension, n (%)	1539 (72.9)	1567 (74.2)	0.03	2182 (65.2)	2227 (66.5)	0.028
Prior outpatient diagnosis PAOD, n (%)	604 (28.6)	626 (29.6)	0.023	900 (26.9)	976 (29.1)	0.051
No. of different prescriptions, median (IQR)	11.00 (6.00–18.00)	12.00 (7.00–18.00)	0.043	9.00 (5.00–14.00)	9.00 (5.00–14.00)	0.038
No. of previous inpatient admissions, total (including index), median (IQR)	2.00 (1.00–3.00)	2.00 (1.00–3.00)	0.033	1.00 (1.00–2.00)	1.00 (1.00–2.00)	0.027
No. of prior PAOD outpatient visits, median (IQR)	1.00 (0.00–3.00)	1.00 (0.00–4.00)	0.022	1.00 (0.00–3.00)	1.00 (0.00–3.00)	0.054
Invasive procedure: OSR, n (%)	816 (38.6)	779 (36.9)	0.036	747 (22.3)	783 (23.4)	0.026
Hospital length of stay, days, median (IQR)	12.00 (7.00–22.00)	12.00 (7.00–22.00)	0.008	4.00 (3.00–8.00)	4.00 (3.00–8.00)	0.007

CLTI indicates chronic limb‐threatening ischemia; IC, intermittent claudication; IQR, interquartile range; OSR, open surgical revascularization; PAOD, peripheral arterial occlusive disease; and TIA, transient ischemic attack.

^*Values >0.1 were deemed to indicate meaningful differences.

Prescription Prevalence for Statins and Antithrombotic Agents

Among 18 095 patients with CLTI and 30 424 patients with IC, 43.4% in the CLTI group and 54% in the IC group received both antithrombotics and statins after index stay, which was 36.4% in CLTI and 39% in IC before admission (Figure 2). Neither receiving statins before or after index stay (nonusers, red flows in Figure 2) was the case in 37.3% (18.5% with and 20.8% without antithrombotics before) of patients with CLTI and 28% (9.9% with and 18.1% without antithrombotics before) in patients with IC. Initiating statin therapy after index stay (new users, green flows in Figure 2) was the case in 13.1% (4.1% with and 9% without antithrombotics before) of patients with CLTI and 13.9% (3.4% with and 10.5% without antithrombotics before) of patients with IC.

Figure 2. Alluvial diagram illustrating the proportion of new users and nonusers (n=9 463 patients with chronic limb‐threatening ischemia [CLTI] and n=12 745 patients with intermittent claudication [IC]) among all statin users meeting the inclusion criteria of the study also showing formerly and permanent use (n=18 095 patients with CLTI and n=30 424 patients with IC).

Shown is the frequency of statin therapy and prescription of antithrombotics during the 3 years before and 3 months after index revascularization for symptomatic peripheral arterial occlusive disease.

Independent Predictors of Receiving Statins in the Matched Study Sample

The most important predictors of initiating statin therapy in the CLTI group were dyslipidemia (odds ratio [OR], 4.50; 95% CI, 4.01–5.06), discharge year (OR, 1.10; 95% CI, 1.08–1.12), age (OR, 0.89; 95% CI, 0.87–0.92), number of different prescriptions (OR, 0.86; 95% CI, 0.82–0.89), and prior myocardial infarction (OR, 1.60; 95% CI, 1.22–2.10) (Figure S2). In the IC group, the most important predictors of initiating statin therapy were dyslipidemia (OR, 5.19; 95% CI, 4.73–5.68), discharge year (OR, 1.10; 95% CI, 1.08–1.11), age (OR, 0.92; 95% CI, 0.90–0.94), number of different prescriptions (OR, 0.88; 95% CI, 0.85–0.91), and open surgical repair at index stay (OR, 0.68; 95% CI, 0.61–0.77) (Figure S3).

Long‐Term Effectiveness Outcomes in the Matched Study Sample

Compared with nonusers, both in the CLTI and the IC groups, new users had a significant lower probability for all‐cause mortality (for CLTI: HR, 0.75 [95% CI, 0.68–0.84]; for IC: HR, 0.80 [95% CI, 0.70–0.92]) (Table 3). Further, statin initiation was associated with a lower risk of major amputation (HR, 0.73; 95% CI, 0.58–0.93) in CLTI and a lower risk for cardiovascular events (HR, 0.80; 95% CI, 0.70–0.92) in IC. In absolute terms, statin initiation was associated with 8.8% lower probability of dying in the CLTI group (37.3% versus 46.1%) and 3.4% lower probability of dying in the IC group (15.5% versus 18.9%). The survival benefit of new users compared with nonusers increased over time in CLTI and was stable in IC (Figure 3). The probability for major amputation was 2.9% lower in the CLTI group (8.4% versus 11.3%) and for cardiovascular events was 3.3% lower in the IC group (15.2% versus 18.5%). The amputation benefit in CLTI increased over time (Figure S4), while the benefit in respect to cardiovascular events in IC was stable (Figure 3).

Table 3.

Probability of Experiencing the Outcomes of Interest Within 5 Years After Index Revascularization in New Users Versus Nonusers of Statin Therapy

Strata	Outcomes of Interest	Probability for New Users (95% CI)	Probability for Nonusers (95% CI)	HR (95% CI)	No.	Events
CLTI	All‐cause mortality	37.3 (34.8–39.7)	46.1 (43.5–48.6)	0.75 (0.68–0.84)	4224	1315
CLTI	Major amputation	8.4 (6.9–9.9)	11.3 (9.5–13.1)	0.73 (0.58–0.93)	4224	278
CLTI	Myocardial infarction/stroke/TIA	23.3 (21.0–25.6)	25.7 (23.2–28.1)	0.89 (0.77–1.04)	4224	658
CLTI	Diabetes mellitus	20.3 (17.1–23.3)	20.8 (17.5–23.9)	0.97 (0.77–1.23)	2232	284
CLTI	Myopathy	4.6 (3.4–5.8)	4.0 (2.9–5.2)	1.15 (0.79–1.67)	4224	109
IC	All‐cause mortality	15.5 (14.0–17.0)	18.9 (17.3–20.5)	0.80 (0.70–0.92)	6698	805
IC	Major amputation	1.5 (1.0–2.0)	1.6 (1.1–2.1)	0.93 (0.58–1.49)	6698	70
IC	Myocardial infarction/stroke/TIA	15.2 (13.7–16.6)	18.5 (16.9–20.1)	0.80 (0.70–0.92)	6698	788
IC	Diabetes mellitus	15.0 (13.2–16.7)	15.2 (13.3–16.9)	0.99 (0.83–1.18)	4678	490
IC	Myopathy	6.5 (5.5–7.5)	5.4 (4.5–6.4)	1.21 (0.96–1.52)	6698	287

CLTI indicates chronic limb‐threatening ischemia; HR, hazard ratio; IC, intermittent claudication; and TIA, transient ischemic attack. All estimates are based on Cox proportional hazards models using the matched data.

Figure 3. Kaplan‐Meier curve of 5‐year all‐cause mortality (upper panel) and 5‐year probability for cardiovascular event (myocardial infarction, stroke, or transient ischemic attack; lower panel) in propensity score–matched cohorts including 95% Wald CI and log‐rank test (P value).

CLTI indicates chronic limb‐threatening ischemia; and IC, intermittent claudication.

Long‐Term Safety Outcomes in the Reduced Matched Study Sample

We did not detect significant differences in the probability for incident diabetes mellitus (in the reduced sample) or myopathy between the study groups (Table 3 and Figure S4).

Sensitivity Analyses

The results for effectiveness outcomes and safety outcomes were largely similar when fitting the Cox models directly to the unmatched data (Figure S5). Without adjustment for confounding, statin users had even more favorable effectiveness outcomes, but safety outcomes were hardly affected. The effect of statins was robust to the inclusion of other important medication groups, ie, angiotensin II receptor blockers or angiotensin‐converting enzyme inhibitors, calcium channel blockers, β‐blockers, or oral anticoagulation (Figure S6). The effect of statins did not significantly differ between men and women, except for amputation in patients with CLTI (HR in women: 0.54 [95% CI, 0.29–0.76]; HR in men: 1.10 [95% CI, 0.85–1.42]) (Figure S7). Stratifying the analysis by age revealed that older patients (≥75 years) benefit most from initiating statins for survival and diabetes mellitus in patients with IC (Figure S8). Further, there were no sizeable differences when stratifying by discharge years (Figure S9). The same was true for statin intensity (patients taking high‐intensity statins: n=415, 6.2%), where the CIs for low‐to‐moderate intensity and high‐intensity statins overlapped for all outcomes (Figure S10). We found a significant association between high‐intensity statin use and myopathy in patients with IC. No differences were detected when stratifying by procedure type at index stay (Figure S11).

Discussion

This is the first real‐world study assessing the effectiveness and safety of initiating statin therapy in symptomatic patients with PAOD after revascularization in a large nationwide cohort. Compared with nonusers, new users of statin therapy had a considerably lower relative and absolute probability of all‐cause mortality in both CLTI and IC, major amputation in CLTI, and cardiovascular events in IC. At the same time, the incidence of diabetes mellitus and myopathy was not associated with new statin prescription. As same as that documented in primary prevention, ³⁴ we found no evidence for the assumption that new patient groups benefit less from statins, emphasizing the importance of quality improvement and awareness campaigns to further promote their prescription.

Valid guidelines call for more evidence on the comparative effectiveness of pharmacological therapy along the full spectrum of clinical reality. ³⁵ Yet, existing real‐world evidence stems from smaller randomized controlled trials with short follow‐up or observational studies based on smaller registries, single centers, geographic regions, or predominantly male patients. The particular merit of routinely collected data from health insurance claims is the large sample size, long follow‐up, and high variety and completeness of information available to adjust for confounding allowing study of the full heterogeneity of patients in daily care. Especially, rare and potentially late outcomes, such as major amputations and the incidence of myopathy and diabetes mellitus, could be analyzed with sufficient statistical power. ³⁶ , ³⁷ We included these safety outcomes, while prior studies focused mostly on effectiveness. Yet, our study present the central findings both for absolute and relative risk differentials. Furthermore, we used both inpatient and outpatient data, and, for the detection of incident diabetes mellitus, corresponding prescriptions. The long lookback and follow‐up periods made it possible to minimize the risk of not detecting a large portion of adverse reactions.

Among patients not on statin therapy before index stay, the proportion of statin therapy after index stay doubled during the study period. Yet, still less than half of the patients received statins in 2018, with particularly low rates among patients with CLTI. These interesting and striking results are in line with a previous study concerning sex disparities in optimal pharmacological treatment of symptomatic patients with PAOD in Germany, where only 55% of the patients received a lipid‐lowering drug. Notably, there was also preliminary evidence that patient characteristics (eg, age, sex, and comorbidities) were more influential than healthcare variables such as the type of revascularization procedure. ³⁸ Because of the nonrandomized observational study design, all results should be considered as merely hypothesis generating. Hence, it appears challenging to explain the low utilization of statins before as well as after revascularization. Unwarranted variation in best medical treatment can be attributable to a lack of high‐level evidence or insufficient application of existing evidence. In terms of statins, similar to antithrombotics, there is good evidence available from many international guidelines. ² , ⁶ , ⁷ The relationship between patients, inpatient physicians, and general practitioners is likely affected by a multifactorial system of influencing factors. It seems reasonable to address this healthcare issue with awareness campaigns and actions to improve both prescription prevalence and patient compliance.

Our results confirm findings from a large Swedish cohort study reporting higher statin utilization in patients with IC than in patients with CLTI. ³⁹ Stavroulakis et al ⁴⁰ presumed that the insufficient use in patients with CLTI might be caused by the paucity of evidence on the benefits of statins with regard to limb outcomes. At the same time, the evidence is accumulating that the walking distance in patients with IC could be positively influenced. ⁴¹

Internationally, large variations in statin utilization rates have been documented, pointing at the role of national healthcare systems (prescription patterns and regulations). For example, only 21% of patients with CLTI in Japan with below‐the‐knee lesions received statins, ⁴² while 83% received statins in the US Veterans Affairs Health System. ¹⁸ Prescription rates probably differ between reimbursement systems. In Germany, during the study period, medications were solely prescribed within the outpatient sector while hospital physicians communicate their recommendations in the medical report at discharge. Despite continuous efforts in raising awareness for this issue, ¹⁷ missed opportunities caused by low undertreatment of patients with PAOD remain. ¹¹

Recently, Arya et al ¹⁸ reported a reduction in all‐cause mortality and amputation‐free survival of ≈20% for low‐to‐moderate statins compared with antiplatelets only, which is in line with our findings. Interestingly, our sensitivity analyses suggest that in patients with CLTI, women seem to benefit to a larger extent from statin initiation when compared with men concerning amputation risk. Women were diagnosed more often with asymptomatic or even atypical disease symptoms without appropriate and timely treatment. ⁴³ Thus, they might be more dependent on adequate secondary prevention for preventing severe limb outcomes when compared with their male counterparts.

Statins significantly reduced the risk for major cardiovascular events in most prior studies ranging from reductions in event rates between 10% and 62% (Table S4). Confirming prior reports, our results for the subgroup of patients with IC are situated in the lower end of this range, while the effects were nonsignificant in patients with CLTI. Reports from Swedish patients with PAOD who underwent revascularization also documented more pronounced effects in the IC group than in the CLTI group. ³⁹

Although many potential adverse reactions have been presumed in the literature, we focused on the established safety outcomes of incident myopathy and diabetes mellitus. ³⁷ Our study results are in line with prior evidence on the safety of statin therapy in patients with PAOD. ¹⁹ Collins et al ³⁷ estimated a minor incident diabetes mellitus risk of ≈1% for a more general population. Moreover, recent guidelines state that the frequency of statin‐induced diabetes mellitus strongly depends on the study sample. ⁴⁴ For example, we even documented a tendency for lowered diabetes mellitus risk for statin initiation among patients with IC aged ≥75 years. This seems to contradict prior evidence, and future studies may focus on the role of age as a modifier in the relationship between diabetes mellitus and statins. Also based on German claims data, Ihle et al ²⁹ reported ≈2% of statin‐induced myopathy while Collins et al ³⁷ presumed 0.05%. In our study, the increase in risk ranged in between these estimates with 1.1% in patients with IC and 0.6% in patients with CLTI, and both values being nonsignificant in the final analysis. Interestingly, we detected a significant association between statins and myopathy only for high‐intensity statin users in patients with IC in our sensitivity analysis (HR, 2.00; 95% CI, 1.17–3.41). This might be a plausible finding and proof of a dose relationship, as statin toxicity indeed increases with statin dose. ⁴⁵

Study Limitations

This is a retrospective propensity score–matched health insurance claims data analysis, so there is no possibility to randomize patients and observe them prospectively. Consequently, the results of this study should be viewed as hypothesis generating and not hypothesis testing. Our propensity score analysis can prevent bias but not fully exclude all sources of bias and residual confounding, eg, that caused by confounding by indication, as compared with randomization. The study groups differed with respect to some of the measured covariates, so that differences in unobserved characteristics that likely confounded our results cannot be ruled out. Yet, as is the case for randomized controlled trials, the quality of observational studies is crucial for assessing the validity of their outcomes. This study applied a rigorous study design with fixed lookback and follow‐up, approved methods, transparent reporting of intermediate steps, and extensive sensitivity analyses. We believe that the risk for distortion caused by residual confounding is low in our study since results are broadly in line with findings from randomized controlled trials and prior observational studies (Table S4). Our sample covered only patients insured at one of many different health insurance funds in Germany. Although slightly different from the population composition in Germany, ⁴⁶ our population‐based sample is comparable to current European populations. We, therefore, believe that our results exhibit a larger degree of external validity than veteran data, more narrowly defined subgroups in trials or data from small regional registries or single‐center studies. We were not able to address all contraindications, statin intolerance, or other adverse reactions. However, the prevalence of intolerance is unlikely to be larger than a few percent, as previous studies in patients with PAOD have demonstrated. It is therefore unlikely a potential explanation for the low utilization of statin therapy. ⁵

The inexistent association of statin use and diabetes mellitus or myopathy risk in our study sample might be caused by insufficiently differentiating by statin type. Since statins differ, inter alia, in derivation and metabolism, varying strengths and limitations of each drug are possible in heterogeneous study populations. ⁴⁷ To ensure that every patient receives the safest and most effective statin, further investigations stratified by the drug, regarding risk factors in distinct patient groups, are necessary to increase adherence and avert discontinuation.

Conclusions

We documented increased long‐term survival and freedom from amputation and cardiovascular events for initiating statin therapy after revascularization. At the same time, safety concerns about the onset of diabetes mellitus and myopathy could not be confirmed. Our findings indicate that new users of statin therapy benefit as much as common users, emphasizing the importance of quality improvement and awareness campaigns to improve prescription rates.

Sources of Funding

The presented analyses were part of the IDOMENEO project (Funded by the German Joint Federal Committee; grant number 01VSF16008). The GermanVasc research group is additionally supported by the German Stifterverband and the Corona Foundation (grant number S199/10061/2015), by the German Joint Federal Committee (RABATT study; grant number 01VSF18035), and receives research funds from Bayer Vital GmbH, B. Braun Melsungen AG, and the Foundation of the University Heart and Vascular Center Hamburg. The funders had no influence on the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the article; and decision to submit the article for publication.

Disclosures

None.

Supporting information

Tables S1–S4

Figures S1–S11

References 48 , 49

(J Am Heart Assoc. 2020;9:e018338 DOI: 10.1161/JAHA.120.018338.)