Statin‐associated myopathy. Assessment of frequency based on data of all statutory health insurance funds in Germany

Abstract Aim of the study was to assess the incidence of statin‐associated myopathy (SAM) under real‐life conditions in Germany. Database: Administrative data (master data, diagnoses, prescriptions) for all individuals in Germany insured with the Statutory Health Insurance. Basic population: individuals 18 years and older who have been insured continually from 2009 to 2011 (52.9 million; 29.9 million men, 23.9 million women). Data access is provided by the German Institute of Medical Documentation and Information, DIMDI) according to the Data Transparency Regulation of 2012. Statins: identification with the ATC–Codes: C10AA, C10BA and C10BX. Study population: incident statin users in 2010 with a diagnosis of lipid disorders (ICD‐10‐GM E78, excluding patients with: E78.1, E78.3, E78.6 in eight quarters before index prescription. Definition of SAM: documentation of myopathy (ICD‐10‐GM G72.0, G72.8; G72.9, M60.8, M60.9, M79.1) in the first statin prescription quarter or in one of the three following quarters. The first event is considered for the incidence estimate. The daily doses included in a package were classified as “days under therapy” (by assuming one DDD) and taken as exposition time. SAM was found in 1.9% of 531 672 incident statin users. The percentage differs according to the patterns of statin use: the lowest incidence is observed in those with only one prescription (1.3%), the highest incidence with 5.0% is observed in those who not only stopped the treatment within 365 days, but who also had their statin changed. Administrative data including diagnoses from ambulatory care provide a realistic estimate of SAM frequency in every day practice.

Reduction in morbidity and mortality in secondary prevention is well supported by numerous studies as well as meta-analyses. [5][6][7][8] The implementation of the therapy recommendations is seen not least in the prevalence of treatment with statins, which has been increasing for years. 9,10 In Germany, lipid-lowering agents are reimbursed by the statutory health insurance if a cardiovascular disease (coronary heart disease), cerebrovascular disease (stroke) or peripheral arterial disease (PAD) is manifest or there is a high cardiovascular risk with an estimated probability of occurrence ≥20% in the next 10 years. 11 Current US guidelines recommend increased use even in primary prevention. 12 Against the backdrop of treatment rates, which are already high and are expected to continue to increase, reliable information is needed not only about benefits, but also about possible harm from the therapy, such as increased risk of diabetes, disorders of liver and kidney functions, muscle damage, as well as the frequent occurrence of malignant diseases. 4 Here, the evidence is clearly more uncertain compared to the benefits. Since cerivastatin had to be withdrawn from the market in 2001 due to cases of lethal rhabdomyolysis, muscle damage and muscle ailments have been given greater attention as a possible side effect of the statin treatment.
Nonetheless, in Germany, there is a lack of reliable data, such as how frequently muscle disorders and muscle ailments occur. The frequency estimates of statin intolerance are scattered in the literature depending on the definition and investigation method, from 0.01% to 10%. 13 Registries and observational studies identified values from 7% to 29%. 14,15 In clinical trials (RCTs), in contrast, there are only minor differences in the side effects compared to placebo. 16 However, experience in everyday health care seems to be different-a difference that has been seen not least in the controversy between the Cholesterol Treatment Trialists (CTT) Collaboration and the British Medical Journal [17][18][19] since 2014 and which raises the question of whether fewer severe side effects in clinical trials are possibly under-reported because of methodological reasons. 20,21 Statin treatment generally represents long-term therapy, but there are strong indications of insufficient adherence and premature discontinuation of treatment. 22,23 Risk for non-adherence is higher in males and increases with age, comorbidities like anxiety and depression and also with reported bad news about statins [24][25][26] Discontinuation rates in everyday care due to side effects are higher than the rates of side effects described in clinical studies.
One reason for discontinuation of therapy or for therapy interruptions may be the occurrence of muscle pain. Studies show that patients with statin intolerance have higher rates of cardiovascular events. 27,28 Since there is no information for Germany on the frequency of statin-associated myopathy (SAM) under everyday conditions, the goal of the study was to develop a method to report SAM based on routine data from the statutory health insurers and to make an estimate of the frequency of statin-induced myopathy based on that data. There are different definitions of SAM in the literature. 21,29 In the following, we generally refer to any kind of muscle-related complaints with and without CK elevation. Statin-associated does not necessarily mean that there must be causality. 29 2 | MATERIALS AND METHODS

| Database
The analysis was conducted on a Germany-wide statutory health insurance (SHI) data set, the so-called "Health Care Data Information System", which is located at the German Institute of Medical Documentation and Information (Deutsches Institut f€ ur Medizinische Dokumentation und Information, DIMDI) (hereinafter referred to as DaTraV data). This database of approx. 70 million SHI insured persons out of 80.8 million inhabitants, which was established based on a statutory provision, has been available since February 2014. 30 The basis is data that all statutory health insurers have to transmit for the morbidity-oriented risk structure compensation scheme to the authority responsible for this, the Federal Social

| Statin prescriptions
These are calculated in the data set based on the Anatomic-Therapeutic-Chemical-Classification (ATC code) C10AA for monosubstances and C10BA as well as C10BX for combinations (ATC Version 2012). 31

| Inclusion and Exclusion Criteria
The presence of a lipid metabolic disorder was assumed if one of the following diagnoses-according to ICD 10-GM 32 -was documented in the lead time or in the incidence quarter at least once in the outpatient sector (with modificator "confirmed" to indicate the These diagnoses had to have been documented either once as an outpatient (with the modifier "confirmed") or as an inpatient diagnosis. Outpatient diagnoses are documented only quarterly, inpatient discharge diagnoses at precise monthly intervals.

| Definition of statin-associated myopathy (SAM)
Since there is no specific coding for this, SAM must be determined indirectly from the SHI data. We assume SAM in cases where SAM is documented for the first time in new users of statins (in the prescription quarter or in one of the three following quarters) and where no documentation of SAM is found during the eight quarters before starting statins. Myopathies that were documented after the end of the statin therapy were not included in the analysis. Muscle pain was recorded here with following ICD 10-GM-coded diagnoses: Rhabdomyolysis is not available in the German Modification. In each case, the first event with exposure is considered.

| Statin utilization pattern and SAM
Estimation of the SAM requires that the exposure time, ie, the period of treatment with statins, be determined for every statin recipient. In the SHI routine data, there is no information on the prescribed dose from which the range of coverage of the prescribed package could be determined. For this reason, the daily doses included in a package were classified for this as "days under therapy". This approach was established as a quasi-standard for the secondary data analysis of SHI routine data.
In the case of insured persons who are given a lower dose than one daily dose per day and for whom, consequently, the range of coverage of the package would be longer, when a DDD is used to calculate the range of coverage of the package, there are interruptions in therapy due to methodological reasons. This calculative therapeutic gap was "closed" if the number of days was smaller than the range of coverage of the last package. In terms of the calculation, this    Table 2 shows the results of the duration of therapy and continuity together with the indication of whether different statins were prescribed, ie, a "change of active substances" occurred. was documented in almost 94% of the cases (cf. Table 3). The specific diagnosis of "drug-induced myopathy" was given for only 58 recipients, equivalent to a share of 0.6%.
Viewed over time, the myopathy diagnosis was documented primarily in the first quarter of treatment (cf. Table 4). treatment year and at least one statin change (5%). Consistent with other authors, 29,33 we also see that statin therapies are continued after a change in the active substance.
With respect to patients who were given continuous treatment without statin changes and were diagnosed with SAM, our interpretation is that the muscle pain occurring under therapy was discussed and documented in the doctor-patient contact in the outpatient sector. A dose reduction might have been made. However, this was not able to be investigated using the available database.
If the different diagnoses that are subsumed under SAM are broken down individually, at first glance, it is confusing that "druginduced myopathy" is named as a diagnosis, and also that unspecific diagnoses are supplemented with the description "unspecified". There are some international studies that address estimating the frequency of SAM using administrative data, 26,37-41 but they consider either only severe SAM based on hospital data (hospitalization due to SAM) or SAM restricted to rhabdomyolysis 42 and/or include laboratory values, and thus from a methodological standpoint are not comparable with the study conducted here, which observes SAM exclusively by means of diagnoses (including in the outpatient sector). Therefore, it may be assumed that in these hospital-based studies, only the most severe SAM is recorded, and the estimates of incidence rates are accordingly lower.
Chang et al. 43 also consider, in their analysis based on claims data, diagnoses from the outpatient sector, but calculate these from the plain text (myositis or rhabdomyolysis). Among 18 036 incidental statin recipients, they identified 23 cases (0.13%) with these diagnoses. These figures are around one power of ten lower than our values, which, along with possible cultural differences in the use, duration of therapy and diagnosis documentation, may also be caused by methodological differences such as the limited number of diagnoses included, the exclusion of patients with potentially interacting drugs and of patients without blood testing. Colantonio et al. 44 chose an entirely different approach which, based on administrative data, also suggested an algorithm for statin intolerance and, depending on the definition, identified a statin intolerance of 1.0% or 5.2%. The results are in the range of our study, but due to methodological differences, it has to be assumed that other patients are reported. In contrast to our approach, in their definition of statin intolerance, Colantonio et al. combined several criteria as dose reduction, switch to ezetimibe, diagnosis of rhabdomyolysis or discontinuation of therapy, "antihyperlipidemic event" after dose reduction or discontinuation of therapy or a change between three or more statins. As

| Limitations
Some limitations must be kept in mind when interpreting the results. We cannot rule out, that we might overestimate SAM as we did not apply a control group design. Our study design followed clinical practice, that is, the co-occurrence of incident myopathy together with incident statin use would be rated as SAM by physicians and patients in every day practice.
As a further limitation we have to mention that we did not analyze different dosages of the statins with regard to any dose effect relationship for myopathy. We lack information on the daily doses prescribed by the physician, which is a prerequisite for such an analysis.
When evaluating the results, documentation customs and billing rules (eg, diagnosis documentation to justify a CK value measurement) must be taken into account. It must also be considered that muscle pain is perceived differently according to the individual, and even physicians tolerate a CK value increase in their patients to different degrees.

| Strengths
The strengths of the study are that, as a database, the entire SHI

| CONCLUSION
Compared to studies that use only hospitalizations due to SAM, the results reported here, by including the outpatient sector, also include milder cases of muscle complaints and give a realistic idea of SAM under everyday conditions. Depending on the observed pattern of treatment, the shares found in the study are between 1.3% and 5.0% and thus certainly do not represent a rare event in everyday health care. On the one hand, the results show that a not insignificant share of patients tolerate the therapy despite myalgia, since the statin therapy was continued for these patients without interruptions and without changing to an alternative statin.
On the other hand, for patients who do not tolerate statins but need them due to their cardiovascular risk profile, an alternative treatment strategy must be found. There are recommendations for approaches to this issue. 14,29

FINANCIAL SUPPORT
The study was financed by an unrestricted grant to the PMV forschungsgruppe from Sanofi-Aventis.