Use of Administrative Data to Estimate the Incidence of Statin-Related Rhabdomyolysis

To the Editor: Studies of rare adverse drug reactions (ADRs) are difficult because they require large numbers of exposed persons to identify an adequate number of cases. Spontaneous adverse events reports have been used to identify cases of rhabdomyolysis, a rare but serious complication of statin use.¹ A complex algorithm based on administrative data also has been used to identify cases of rhabdomyolysis and myopathy among enrollees of several large health plans.^2,3 We conducted this population-based epidemiological study to evaluate use of the diagnostic code for rhabdomyolysis introduced in 2006 as a method of estimating the incidence of statin-related rhabdomyolysis and myopathy.

Computerized pharmacy data were used to estimate the total person-years of statin use by statin type and dose among all enrollees of Group Health Cooperative (GHC) from January 2006 through December 2010. Trained abstractors reviewed the full electronic medical record (EMR) of all GHC enrollees who (1) had a statin prescription within 3 months before an outpatient or inpatient encounter with an International Classification of Diseases, Ninth Revision (ICD-9) code for rhabdomyolysis; (2) had a statin prescription within 3 months before an ICD-9 code for an ADR from a lipid-lowering drug or a creatine kinase level of 1000 U/L or higher (to convert to μkat/L, multiply by 0.0167) in the GHC laboratory database; and (3) were identified by natural language processing software as having the words “rhabdo” and “statin” appear near each other anywhere in the EMR. Statin-related rhabdomyolysis and myopathy were defined as muscle symptoms with a peak creatine kinase level 10 or more times and 5 to 10 times the upper limit of normal, respectively, in the absence of another etiology.

Incidence rates were calculated from the number of cases validated by EMR review divided by person-time of use for various doses of simvastatin and for all other statins combined. Among cases who switched statins, rhabdomyolysis was classified according to the statin being used at the time of symptom onset. Incidence rate ratios (IRR) comparing simvastatin with other statins and high with low doses of simvastatin were calculated by dividing the incidence rates for validated cases and rhabdomyolysis cases identified by ICD-9 code. Confidence intervals were estimated using Stata version 11.0 (StataCorp). This study was approved and a waiver of consent granted by the GHC institutional review board.

Among 292 statin users with an ICD-9 code for rhabdomyolysis, 22 cases of statin-related rhabdomyolysis were validated (positive predictive value of 7.5% [95% CI, 5.0%-11.1%]; Table 1). Seven additional cases were identified by other methods for a sensitivity of 76% (95% CI, 58%-88%). Common etiologies for rhabdomyolysis other than statin use included prolonged immobility, arterial ischemia, recent surgery, and severe infection.

Of the 29 validated cases of statin-related rhabdomyolysis, 26 (90%) were hospitalized and 8 (29%) had at least a doubling of serum creatinine level; none died. The median peak creatine kinase level was 7450 U/L (interquartile range, 1477-150 510 U/L). Incidence rates are listed in Table 2. The rhabdomyolysis IRR for simvastatin compared with other statins was 2.61 (95% CI, 1.03-7.84) using validated cases compared with 1.03 (95% CI, 0.80-1.34) using the ICD-9 code for rhabdomyolysis. The IRR for dose of simvastatin of 80 mg/d or greater compared with 20 to 39 mg/d was 12.2 (95% CI, 3.6-52.3) using validated cases compared with 1.77 (95% CI, 1.05-2.88) using the ICD-9 code for rhabdomyolysis.

These results confirm in a community setting findings from a recent clinical trial that prompted the US Food and Drug Administration to issue a warning about the use of high-dose simvastatin.^4,5 The ICD-9 code for rhabdomyolysis was nonspecific for this ADR, and the resulting misclassification markedly attenuated the estimated relative risk for high-dose vs low-dose simvastatin. We did not adjust for potential confounding factors and may have failed to identify all cases of statin-related rhabdomyolysis, although these limitations are unlikely to account for the large relative risks. The use of administrative data alone in studies of ADRs with multiple causes may fail to detect actionable and clinically important harms.

Author Contributions: Dr Floyd had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Floyd, Heckbert, Psaty.

Acquisition of data: Floyd, Heckbert, Carrell, Psaty.

Analysis and interpretation of data: Floyd, Heckbert, Weiss, Psaty.

Drafting of the manuscript: Floyd, Psaty.

Critical revision of the manuscript for important intellectual content: Floyd, Heckbert, Weiss, Carrell, Psaty.

Statistical analysis: Floyd.

Obtained funding: Psaty.

Administrative, technical, or material support: Carrell, Psaty.

Study supervision: Heckbert, Weiss, Psaty.

Funding/Support: This research was supported in part by grants HL078888 and HL085251 from the National Heart, Lung, and Blood Institute. Dr Floyd was supported by National Heart, Lung, and Blood Institute training grant T32 HL007902. Dr Carrell was supported by grant U01 HG006375 from the National Human Genome Research Institute.

Role of the Sponsors: The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.