FormalPara Key Summary Points

Why carry out this study?

Certain medications used to treat patients with atherosclerotic cardiovascular disease (ASCVD) have demonstrated a possible association with increased risk of tendon rupture and tendinopathies (TRT).

When evaluating the potential risk of TRT associated with such medications, it is important to understand the background prevalence of TRT in this population; however, the prevalence of TRT in a large cohort of patients with ASCVD has not previously been determined.

We sought to determine the prevalence of TRT and associated risk factors in a population of over 5.5 million patients with ASCVD and over 61 million patients representative of the general adult United States population, with records in a large national medical and pharmacy claims database.

What was learned from this study?

When assessed over a 12-month follow-up period, the prevalence of TRT was higher among patients with ASCVD than in the overall population (3.4% vs. 1.9%, respectively).

The risk of TRT in patients with ASCVD was increased in the presence of several identified factors, including increasing age, obesity, and rheumatoid arthritis; no association was observed between statin or bempedoic acid use and TRT.

These results may help to guide clinical decision-making when treating patients with ASCVD, and provide an estimate of the background risk of TRT in patients with ASCVD, which may be useful when determining the risk of TRT with medications used to treat this population.

Introduction

The incidence of tendon rupture in the general adult United States (US) population is estimated to be 2.1 per 100,000 patient-years for the Achilles tendon [1] and 2.6 per 100,000 patient-years for tears of the distal biceps and triceps [2]. Rotator cuff tears are reported more frequently and show a clear age dependency, with prevalence ranging from 15 to 20% in people aged 60–69 years to 37–51% among those aged 80–89 years [3, 4]. Other potential risk factors for tendon rupture include male sex, diabetes, renal failure, obesity, and physical activity [5,6,7]. Associations with tendinopathy have also been observed for several other factors, including prior personal or family history of tendinopathies, hypercholesterolemia (particularly heterozygous familial hypercholesterolemia [HeFH]), gout, rheumatoid arthritis, and use of fluoroquinolone antibiotics, steroid injections, and statins [5, 8,9,10,11,12,13,14,15,16].

Statins, which are the recommended treatment for patients with atherosclerotic cardiovascular disease (ASCVD) [17], have a well-documented association with skeletal-muscle adverse events [18]; however, the relationship between statin use and tendon ruptures and tendinopathies (TRT) is unclear, in part because patients with ASCVD often possess several potential TRT risk factors, including metabolic conditions that are independent risk factors for ASCVD [19,20,21]. Case collections [22,23,24], a review of the US Food and Drug Administration (FDA) Adverse Event Database [25], and post-marketing surveillance for atorvastatin and pravastatin [26, 27] have suggested an association between statin use and TRT, but other studies have not [28, 29].

Bempedoic acid, a citrate lyase inhibitor, which acts upstream of statins in the cholesterol synthesis pathway [30], was approved by the FDA in February 2020 for the treatment of hyperlipidemia in patients with ASCVD and/or HeFH receiving maximally tolerated statin therapy [31]. The bempedoic acid US Prescribing Information (USPI) contains a warning for increased risk of tendon rupture or injury, based on pooled data from two 52-week phase 3 placebo-controlled studies [31,32,33]. The prescribing information was recently updated to reflect tendon rupture events observed in 1.2% of bempedoic acid-treated patients versus 0.9% of placebo-treated patients in the CLEAR Outcomes study [31]. The CLEAR Outcomes trial investigated the effects of bempedoic acid on cardiovascular events in 13,970 statin-intolerant patients with or at high risk for cardiovascular disease and, demonstrated that the incidence of TRT among patients treated with bempedoic acid and placebo was similar [34, 35].

Full evaluation of the potential adverse-event risk for a drug requires an understanding of the background rate of that adverse event in the indicated patient population. The rates of TRT identified in broad epidemiological studies are based on the general adult population, and do not consider the baseline demographics and characteristics of a patient population with ASCVD that may increase their risk of TRT. The aim of this study, therefore, was to determine the prevalence of TRT over a 12-month period in patients with ASCVD and in the general population, and to identify potential TRT risk factors in the ASCVD population.

Methods

Patients and Study Design

This retrospective observational study used data from the Symphony Health Integrated Dataverse (IDV®) medical and pharmacy claims database (Symphony Health, Blue Bell, PA), which includes diagnosis and procedure claims from approximately 164 million patients in the US per year and is a nationally representative sample.

Patients aged 19 years and older with an initial claim in the claims database during the identification period (January 1, 2019, through December 31, 2020), followed by 12 months of continuous enrollment after the index date, were eligible for inclusion in the overall population (Fig. 1). Continuous enrollment was defined as evidence of at least one medical, hospital, or pharmacy claim in the database every month for 12 consecutive months. The index date was defined as the first eligible claim within the identification period. Patients who were aged 18 years at the start of the identification period were included if they subsequently turned 19 before the end of the identification period and had 12 months of continuous enrollment following their initial claim. Patients who did not meet the continuous enrollment criteria or whose sex was unknown were excluded.

Fig. 1
figure 1

Study design. ASCVD atherosclerotic cardiovascular disease, ICD-10 International Classification of Diseases, 10th Revision

Full size image

For inclusion in the ASCVD cohort, patients must have had one or more International Statistical Classification of Diseases, 10th Revision (ICD-10) diagnosis codes for ASCVD during the identification period, regardless of whether the ASCVD diagnosis was their first presenting diagnosis or they had received one or more previous ASCVD diagnoses prior to the identification period. Following the initial ASCVD diagnosis during the identification period (the “ASCVD index date”), subsequent claims were not required to be related to ASCVD diagnosis or management to meet continuous enrollment criteria for inclusion in the ASCVD cohort. The ICD-10 diagnosis codes for ASCVD used to identify patients for this analysis included those associated with a cerebrovascular, coronary, or peripheral vascular event. The full list of ASCVD diagnostic codes used is presented in Supplementary Table S1, and the process for evaluating the ASCVD cohort is presented in Supplementary Table S2.

The follow-up period began on each patient’s index date and ended 12 months later. The primary outcome was evidence of TRT in the 12 months following the ASCVD index date (the “ASCVD follow-up period”). To allow comparisons with the ASCVD cohort over the same duration of follow-up, TRT prevalence in the 12 months following the index date was also evaluated in the overall population. A follow-up period of 12 months was selected as it was deemed sufficient to capture the majority of medication-related TRT events without introducing additional confounding factors that may increase the risk of TRT, such as aging. TRT was defined as an ICD-10 and/or a Current Procedural Terminology code consistent with tendon rupture or a tendon-related diagnosis. Experts in orthopedics and medical coders familiar with orthopedic codes produced a final list of codes to indicate TRT. Diagnostic codes for bicipital tendinitis, calcific tendinitis, impingement syndrome, bursitis of the shoulder, shoulder lesions, muscle strains, or codes that were likely to be associated with tendon rupture due to acute physical injury (e.g., laceration-related codes) were not included in the diagnostic codes. The full list of codes for TRT utilized in this analysis is presented in Supplementary Table S3.

TRT prevalence was also evaluated among various subgroups of interest within the ASCVD cohort: sex (male vs. female), age (19–34, 35–44, 45–64, 65–74, and ≥ 75 years), comorbid medical condition status (type 2 diabetes, obesity, and rheumatoid arthritis [yes vs. no]), and use of systemic fluoroquinolones or corticosteroids, statins, or bempedoic acid at any time during the ASCVD follow-up period (yes vs. no). The presence of a comorbid medical condition was based on having an International Statistical Classification of Diseases, 9th Revision, or ICD-10 diagnosis code consistent with that condition. Medication use was defined as having a pharmacy claim for a prescription within the ASCVD follow-up period, either at any time in relation to the TRT diagnosis or (to explore the potential impact of medication use on TRT) prior to the first instance of TRT. In the latter analysis, use of statins, fluoroquinolones, or corticosteroids (or any combination thereof) was only considered to be associated with TRT if the prescription was issued prior to the first diagnosis of TRT in the ASCVD follow-up period. Systemic corticosteroid use was defined as use for more than 7 days, as the risk of TRT increases with treatment duration [36]. For statins and systemic fluoroquinolones, no requirements were placed on the duration of medication use.

This study was exempt from independent ethics committee or institutional review board approval as it was a retrospective database study based on the secondary use of anonymous claims data in the US. Permission was obtained to utilize the database examined in this study.

Statistical Analysis

The prevalence of TRT was reported both in the ASCVD cohort and in the overall population. TRT prevalence was stratified based upon the regions of the body in which TRT occurred and the number of regions affected during the 12-month follow-up period. The prevalence of TRT was analyzed among the subgroups of interest within the ASCVD cohort using Chi-squared tests; the threshold for statistical significance was set at 0.05. Multivariate logistic regression models were constructed for the ASCVD population and for the sensitivity analysis, using TRT as the dependent variable. Odds ratios (ORs) with 95% confidence interval (CIs) were calculated for the independent variables included in the logistic regression models: sex, age, type 2 diabetes diagnosis, rheumatoid arthritis diagnosis, obesity, fluoroquinolone use, corticosteroid use, and statin use.

Results

Patient Disposition and Baseline Characteristics

A total of 5,589,273 and 61,715,843 patients were included in the ASCVD cohort and the overall population, respectively. Patients in the ASCVD cohort were older than those in the overall population (mean age, 69 vs. 57 years, respectively), and there was a greater proportion of female patients in the overall population compared with the ASCVD cohort (60% vs. 49%, respectively; Table 1); in each group, over half of the patients were White. Comorbid conditions of interest were more common in the ASCVD cohort than in the general population (type 2 diabetes, 59% vs. 31%; obesity, 45% vs. 28%; rheumatoid arthritis, 9% vs. 5%, respectively). In the ASCVD cohort, the most common initial ASCVD diagnosis within the identification period was ischemic heart disease (65%). Use of statins, corticosteroids, and fluoroquinolone antibiotics during the ASCVD follow-up period was recorded for 68%, 18%, and 17% of patients, respectively. Use of bempedoic acid, as either monotherapy or as a fixed-dose combination with ezetimibe, was recorded for 1556 (0.03%) patients in the ASCVD cohort.

Table 1 Demographic and clinical characteristics
Full size table

Prevalence of TRT

TRT during the 12-month follow-up was diagnosed in 188,899 (3.4%) patients in the ASCVD cohort and in 1,186,022 (1.9%) patients in the overall population (Table 2). Among patients in the ASCVD cohort with TRT, the majority (83.5%) had TRT recorded in only one region (Table 2). The most commonly affected TRT regions within the ASCVD cohort were the rotator cuff (48.1%) and other regions of the shoulder (29.6%; Table 3).

Table 2 TRT prevalence and number of TRT regions affected in the overall population and the ASCVD cohort
Full size table
Table 3 TRT regions among patients in the ASCVD cohort with one or more TRT regions affected (n = 188,899)
Full size table

TRT prevalence in the ASCVD cohort was higher in female patients, in patients aged 45–64 years, in patients with type 2 diabetes and rheumatoid arthritis (p < 0.0001 for all), and in patients with obesity (p < 0.001; Table 4).

Table 4 Prevalence of TRT in the ASCVD cohort by patient subgroup
Full size table

TRT prevalence was also higher among patients in the ASCVD cohort with systemic fluoroquinolone or corticosteroid use at any time during the ASCVD follow-up period (Table 5). TRT prevalence among patients in the ASCVD cohort remained higher for corticosteroids but not for fluoroquinolones when the analysis was limited to patients whose first prescription for either medication was recorded before a TRT diagnosis. TRT prevalence was lower in statin users than in non-users, regardless of the timing of the first statin prescription relative to TRT diagnosis. TRT prevalence among the 1556 patients in the ASCVD cohort with a prescription for bempedoic acid was comparable with that observed in patients with ASCVD not taking bempedoic acid (3.5% vs. 3.4%, respectively; p = 0.84). TRT prevalence was 1.5% among the 23 patients whose first TRT claim was reported after a prescription for bempedoic acid.

Table 5 TRT prevalence in the ASCVD cohort by medication usea
Full size table

Logistic Regression (ASCVD Cohort)

TRT was more likely to be observed in the 45–64-years age group compared with the 19–34-years age group, (OR 2.19; 95% CI 2.07–2.32; Fig. 2) in the multivariate logistic regression analysis that accounted for timing of medication use relative to TRT (i.e., only medication use occurring prior to TRT was considered a positive association). Other factors associated with TRT were a diagnosis of obesity (OR 1.51; 95% CI 1.50–1.53) or rheumatoid arthritis (OR 1.47; 95% CI 1.45–1.49). Reduced odds of TRT were observed among users of statins (OR 0.41; 95% CI 0.41–0.42) and fluoroquinolones (OR 0.48; 95% CI 0.47–0.48).

Fig. 2
figure 2

Multivariate logistic regression model to estimate A the odds of TRT (only statin, fluoroquinolone, or corticosteroid occurring prior to TRT considered positively associated) and B the odds of TRT (regardless of temporality of statin, fluoroquinolone, or corticosteroid use) in the ASCVD cohort (N = 5,589,273). ASCVD atherosclerotic cardiovascular disease, CI confidence interval, OR odds ratio, TRT tendon rupture and tendinopathies

Full size image

Similar trends were observed when all TRT diagnoses were included, regardless of the temporality of use of statins, fluoroquinolones, and corticosteroids (Fig. 2). Compared with the analysis that did take medication temporality into account, the odds of TRT were slightly reduced for the 45–64-years age group, and in those with obesity or rheumatoid arthritis, whereas the risk of TRT was increased with the use of statins, fluoroquinolones, and corticosteroids.

Discussion

Atorvastatin, pravastatin, and bempedoic acid [17, 26, 27, 31, 37, 38] are indicated to reduce low-density lipoprotein cholesterol in patients with ASCVD. All three medications have TRT-related side effects included in their USPI (included in post-marketing reports for atorvastatin and pravastatin, and in warnings and precautions for bempedoic acid), but the prevalence of TRT in patients with ASCVD is not defined. We used a large cohort of patients with ASCVD from the claims database to examine the prevalence of TRT in patients with ASCVD and in the overall population. This is, to our knowledge, the first report of the prevalence of TRT in patients with ASCVD. It is also, to our knowledge, the largest study to estimate the overall prevalence of TRT in a population representative of the general US adult population. Our findings serve to raise awareness among clinicians that patients with ASCVD may have underlying risk factors that predispose them to a higher prevalence of TRT compared with the general population.

During the 12-month follow-up period, the prevalence of TRT at 3.4% in the ASCVD cohort was nearly twice that observed in the overall population (1.9%). This suggests that patients with ASCVD are at greater risk of TRT than the general population; however, further investigation is warranted because many TRT-specific risk factors, such as age, overlap with those for ASCVD [21]. The patients included in the ASCVD cohort were generally older than those in the overall population and were more likely to have other risk factors for TRT, including comorbid medical conditions and medication use.

The prevalence of TRT was higher in patients aged 45–64 years compared with the other age groups, regardless of the timing of medication use relative to TRT. This is consistent with studies showing a higher incidence of TRT in middle-aged compared with younger and older populations [39, 40]. This finding may be explained by age-related structural, compositional, and biomechanical changes in tendons [41], coupled with higher levels of physical activity in middle-aged than in older people [42]. Of note, 65% of the ASCVD cohort was aged ≥ 65 years; patients aged ≥ 60 years are at increased risk of developing tendon disorders and tendon rupture than younger patients [43].

The prevalence of TRT was also higher in patients with ASCVD who also had comorbid obesity or rheumatoid arthritis. Obesity plays a clinically significant role in the development of tendinopathy, and is associated with increased risk of tendon rupture, which may be due to inflammatory mediators linked to obesity as well as increased mechanical demands due to high body weight [44, 45]. The inflammation associated with rheumatoid arthritis can extend to the tendons and cause tendon damage and ruptures, often involving the extensor tendons of the wrist, extensors and flexors of the hand, rotator cuff tendons in the shoulder, and the Achilles tendon [14, 46,47,48,49]. Various forms of Achilles and rotator cuff tendinopathy have been reported in 22–39% and 49%, respectively, of patients with rheumatoid arthritis [14, 48, 49], and the incidence of tendon rupture at the wrist has been reported to be 4%, with a risk of rupture of 51% [47]. Corticosteroid therapy, a common treatment for management of rheumatoid arthritis, may contribute to tendon rupture in patients by causing tissue degeneration. However, corticosteroids may be protective against tendinopathy by reducing inflammation [14, 48].

The prevalence of TRT was slightly higher in female patients and in patients with type 2 diabetes, but the clinical significance of these differences is unclear, particularly given the small or absent association for these subgroups observed in multivariate logistic regression analyses. These results differ from previous studies, which have shown that male patients have an increased risk of TRT [1, 7, 50, 51], and that diabetes increases the incidence and severity of TRT [52,53,54]. Some of these differences may be due to these studies not being conducted in an older ASCVD cohort and being primarily focused on the Achilles tendon.

The associations between TRT and drug use in the present study are less clear and should only be considered hypothesis-generating. Use of fluoroquinolones or corticosteroids at any time during the ASCVD follow-up period was associated with increased TRT prevalence; however, unlike corticosteroid use, fluoroquinolone use was not associated with increased risk of TRT in the multivariate analysis. There is a well-established association between fluoroquinolone use and TRT [55], so this finding is unexpected. It should be noted that, in a case-control study, fluoroquinolone-induced tendinopathy was most commonly observed in the Achilles tendon [36], a region affected in only 2.3% of patients with TRT in the present study sample. Furthermore, no minimum duration or dose of fluoroquinolone therapy was required, despite evidence that the risk of tendinopathy with fluoroquinolones may be greater with increasing treatment duration and higher doses [55]. Finally, the comorbidity profile of the ASCVD cohort contains multiple risk factors associated with TRT, which may have dampened any fluoroquinolone-specific signal. The strength of the association between corticosteroid use and TRT in the multivariate logistic regression model was reduced when only corticosteroid use prior to TRT diagnosis was considered (reduction in OR from 1.50 to OR 1.09). Several factors may have contributed to this finding, including the possibility that corticosteroids could have been prescribed both prior to or as treatment for a TRT event.

The present study did not observe an association between statin use and TRT. TRT prevalence was slightly lower in statin users (3.3%) than in non-users (3.6%), and statins were associated with a slightly lower risk of TRT in the logistic regression models. Studies on statin use and TRT have yielded variable findings, with some showing increased risk with statin use and others showing no such association [28]. A large, propensity-matched cohort study (N = 526,351 matched pairs) concluded that statin use did not increase the risk of Achilles or biceps tendon rupture [56]. However, a cohort study of Swedish patients (N = 92,933) reported a greater risk of trigger finger and shoulder tendinopathy in current statin users, and a trend toward increased risk of Achilles TRT in patients with prior or current statin use, compared with those who had never used statins [8]. A nationwide cohort study of patients in South Korea (N = 594,130) determined that statin users had a greater risk of tendinopathy, regardless of the statin type [57]. Tendon rupture and tendon disorder have also been reported during post-marketing experience with atorvastatin and pravastatin, respectively, irrespective of causality [26, 27].

The present study found no association between use of bempedoic acid and TRT. TRT prevalence among patients with a prescription for bempedoic acid was 3.5% compared with 3.4% for the ASCVD cohort overall and 1.5% in patients with a prior prescription for bempedoic acid. The number of patients with ASCVD who received bempedoic acid was small (n = 1556). This agent only became commercially available in early 2020 [31], which was after the start of the identification period. The USPI for bempedoic acid includes a warning and precaution for tendon rupture based on two 52-week trials involving 3621 patients where tendon rupture occurred in 0.5% of patients treated with bempedoic acid versus 0% of placebo-treated patients [31]. It was recently reported in the CLEAR Outcomes trial, in which 13,970 patients were followed for a median of 3.4 years, that TRT occurred in 2.0% of patients in both treatment groups; adjudicated events of tendon rupture occurred in 1.2% and 0.9% of patients treated with bempedoic acid or placebo, respectively [35]. These adjudicated events were included in the recently updated label.

Additional research on TRT with more specific parameters related to the initiation and timing of medications would allow a more robust exploration of these relationships in an observational dataset.

Limitations

The study was a retrospective analysis using medical claims data, which is the primary limitation of this study, as the results depend upon accurate coding by healthcare providers, which is subject to miscoding and underreporting [58]. Pharmacy and medical claims data in the Symphony Health IDV® database are sourced through clearing houses rather than exclusively from specific payers, hospitals, or private entities, such as closed network systems, which may result in gaps or biases in data coverage. Such gaps were minimized in this study by requiring patients to have at least one monthly medical or pharmacy claim to create a 12-month continuous follow-up period. In addition, the data in the claims database are applicable only to insured patients and do not include other patient populations.

The risk of TRT in the present study was increased in patients with the known risk factors of older age, obesity, and rheumatoid arthritis. Elevated cholesterol is also a risk factor for TRT [59, 60]; however, we were unable to control for patients’ cholesterol levels. The impact of other potential risk factors for TRT, such as concomitant medications, physical activity levels, socioeconomic status, and a number of risk factors in female versus male patients, were not investigated. Furthermore, although this study investigated medications with a potential or known association with TRT, this study did not evaluate the association with TRT of all medications used by patients with ASCVD identified in the database. TRT can present as a wide range of injuries, ranging from less severe to full tendon tear. However, the relative degree to which risk factors may be associated with more or less severe tendinopathies was unable to be evaluated using the ICD-10 codes for TRT. It is important to note that this was a cross-sectional prevalence analysis of TRT among patients with ASCVD at differing disease stages, with some patients presenting with a first-time ASCVD diagnosis and others having established disease prior to the start of the identification period. The analysis did not account for this variability in disease duration, and the follow-up period was only 12 months in duration. Unlike the “intrinsic factors,” which remain relatively constant over time, patients may have started or stopped using medications at various times during the ASCVD follow-up period, rendering analysis of the relationship between TRT and medication use challenging. Furthermore, pharmaceutical claims do not account for medication adherence; a recent population-based cohort study found that 21.9% of patients at high cardiovascular risk with indications for cholesterol lowering did not accept statins prescribed to them by their healthcare professional [61]. Future analyses evaluating the relationship between medications and TRT should therefore consider the dose and duration of medication, treatment adherence, and the timing of TRT relative to medication use.

Conclusions

The prevalence of TRT over a 12-month period in patients with ASCVD was nearly twice that observed over the same follow-up period in the overall population. These findings suggest that patients with ASCVD are at higher risk for TRT, which may be due to the overlapping risk factors associated with both ASCVD and TRT. Additional investigations using various designs are required in order to further evaluate the relationship between medication use and TRT, particularly statins, as well as the possible mechanisms underlying this effect.