Incidence of Hospitalized Rhabdomyolysis With Statin and Fibrate Use in an Insured US Population

Kandace L Amend PhD; Joan Landon MPH; Veena Thyagarajan MPH; Steve Niemcryk PhD; Andrew McAfee MD MSc


The Annals of Pharmacotherapy. 2011;45(10):1230-1239. 

In This Article


Data Source

This study used the Normative Health Information database, which includes patient demographics, enrollment dates, reimbursement claims for physician and hospital services, and pharmacy records for members of one of the largest health insurers in the US. Data relating to approximately 14 million individuals with coverage for both medical and pharmacy benefits are available. Approximately 48% of members are from the South, 11% from the Northeast, 25% from the Midwest, and 16% from the West. The 65-year and older population is underrepresented in the database, with only 4% of individuals falling into that age group. Of the 14 million individuals, race/ethnicity and financial resource information is available for approximately 9 million (64%). Approximately 73% of members are white, 7% African American, 9% Hispanic, and 3% Asian. The average household income is $74,000. Diagnoses are recorded using the International Classification of Diseases, Ninth Revision (ICD-9), and medical procedures are coded using ICD-9 procedure, Common Procedural Terminology codes, and the Health Care Financing Agency Common Procedure Coding System codes. Medications are identified by National Drug Classification, therapeutic class, and Hierarchical Ingredient Code List codes.

Cohort Membership

Patients who initiated statins (atorvastatin, fluvastatin, lovastatin, pravastatin, rosuvastatin, simvastatin) or fibrates (fenofibrate, gemfibrozil) from January 01, 1998, to December 31, 2007, were identified. This time frame was chosen because the study was initiated in 2008 and designed to allow for 10 full years of data. Patients were included if they were continuously enrolled for at least 183 days prior to the first dispensing (baseline period), 18 years of age at cohort entry, and had complete medical and pharmacy coverage. The index date for eligible patients was the date of the first eligible statin or fibrate dispensing. Fibrate use among statin initiators and statin use among fibrate initiators were allowed during the baseline period.

Users with any claims-based rhabdomyolysis diagnosis during baseline were excluded.

Study Outcomes

Primary and secondary diagnosis codes similar to those used by Graham et al.[6] were used to identify potential cases of incident rhabdomyolysis associated with a physician visit on an inpatient claim[13] (Appendix I). Clinical consultants (internal and external) reviewed a chronological history of insurance claims (profile review) around the time of the diagnosis code to determine which potential cases warranted medical record review and to identify a provider from whom to seek records. Trained abstractors from an external abstraction firm used a standardized form to abstract medical records on the service dates corresponding to the potential outcome. Clinical consultants reviewed the medical record abstracts and determined cases to be rhabdomyolysis if: (1) a creatine kinase (CK) increase greater than or equal to 10 times the laboratory upper limit of normal (ULN) was present with concomitant muscle symptoms (eg, weakness, aching, tenderness) and no obvious acute alternate etiology; and (2) a creatinine elevation greater than or equal to the laboratory ULN or new clinical diagnosis of renal insufficiency or renal failure was present. The ULN values for CK and creatinine were laboratory specific. Hospitalization was required. Isolated myoglobinuria without concurrent increases in CK (irrespective of creatinine value) was not considered rhabdomyolysis. Myoglobinuria with concurrent increases in CK was assessed by criteria for rhabdomyolysis. The clinical consultants were blinded to statin and fibrate exposure status during the profile and medical record reviews.

For patients with multiple hospitalizations, the outcome date was defined by the first hospitalization with medical record–confirmed rhabdomyolysis. Effect estimates were derived only from confirmed cases.


Follow-up began the day after the index date and continued through the earliest of the date of health plan disenrollment, the date of confirmed rhabdomyolysis diagnosis, or December 31, 2008. Persons with dispensings of multiple statins on the same day were censored on that day to rule out possible claims errors. To reflect current standards of practice, person-time for one fibrate was truncated when another fibrate was dispensed; statins were treated similarly. Analyses were carried out on the person-time in each mutually exclusive exposure state.

Drug Exposure Definitions

Follow-up days were classified separately by current statin and fibrate exposure status, based on the medication dispensing date and days supplied. Days falling between the dispensing date and dispensing date plus number of days' supply plus 20% of days' supply of the preceding prescription were considered to be current drug exposure.

Statins were evaluated as a group and combined statinfibrate therapy was also evaluated. Each person-day was classified as 1 of 7 current exposure states or as an unexposed state: (1) no statins or fibrates, (2) statin alone, (3) fenofibrate alone, (4) gemfibrozil alone, (5) statin and fenofibrate with no gemfibrozil, (6) statin and gemfibrozil with no fenofibrate, (7) gemfibrozil and fenofibrate, or (8) statin and fenofibrate and gemfibrozil.

To evaluate monotherapy statins (atorvastatin, fluvastatin, lovastatin, pravastatin, rosuvastatin, and simvastatin), monotherapy fibrates (fenofibrate and gemfibrozil), and combined statin-fibrate therapy, each person-day had 1 of 20 current exposure states or an unexposed state.

We also evaluated concomitant statin-fibrate therapy versus statin therapy alone and statins categorized by CYP3A4 metabolism. Atorvastatin, lovastatin, and simvastatin are metabolized by CYP3A4, while fluvastatin, pravastatin, and rosuvastatin are not.[14,15]

Statistical Analysis

Patients were categorized and baseline characteristics tabulated by the initiating drug that qualified the patient for the cohort.

Incidence rates (IRs) were estimated by dividing the number of confirmed hospitalized rhabdomyolysis cases by the number of person-years in each drug exposure category. A patient could contribute person-time to multiple exposure categories. In addition, 95% CIs for the IRs were calculated using Byar's formula.[16]

Poisson regression was used to calculate crude and adjusted incidence rate ratios (IRRs) comparing different combination statin-fibrate exposures with no exposure and statin-only exposure. Final regression models were adjusted for baseline covariates if the model containing the single covariate and the drug exposure variable met the definition of a confounder by changing the rate ratio (RR) for the drug exposure variable by greater than or equal to 10% relative to the unadjusted RR for drug exposure.[17] We considered covariates evaluated previously, such as age, sex, and comorbidities (diabetes mellitus, hypothyroidism, renal or hepatic disease, and hypertension).[9,18] Geographic region, cohort entry year, dispensings of CYP3A4 inhibitors during baseline (yes/no), statin or fibrate dispensing during baseline, and health care utilization variables were also considered. Health care utilization variables may act as indirect proxies of patient status in claims data.[19,20] The number needed to harm (NNH) to observe a rhabdomyolysis case was calculated as the inverse of the attributable risk. Attributable risk was derived by subtracting the incidence in the group without exposure from the incidence in the exposed group, reflecting baseline risk and treatment effect.[21]

Data analyses were performed by OptumInsight, using SAS version 9.1 (SAS Institute, Cary, NC).

Privacy and Confidentiality

Because personal health information was used to link insurance claims to medical records, an institutional review board and a privacy board approved our protocol and privacy practices and provided a waiver of patient authorization.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: