Development and validation of algorithms for the detection of statin myopathy signals from electronic medical records

    loading  Checking for direct PDF access through Ovid

Excerpt

Statins (3‐hydroxy‐3‐methylglutaryl coenzyme A reductase inhibitors) are commonly used for lowering low‐density lipoprotein‐cholesterol levels and reducing the risk of cardiovascular disease.1 The prevalence of elevated low‐density lipoprotein‐cholesterol in Singapore adults 18 to 69 years old is about 15.2%.2 In the United States, >25% of adults over 40 years old receive a statin.3 Although statins are generally safe, they can cause muscle toxicity, which may lead to cessation of therapy and consequently progression of cardiovascular disease. Statin‐induced myotoxicity has a broad clinical spectrum ranging from asymptomatic mild elevations of creatine kinase (CK) to rhabdomyolysis,4 which can lead to serious complications, such as renal failure and death. The reported incidence of mild myotoxicity ranges widely. Tolerable muscle symptoms without CK elevation have been reported in up to 33% of patients.4 In contrast, rhabdomyolysis is very rare, with a reported incidence of 0.44–10.2 per 10,000 person‐years.5 Here, we use statin‐induced myopathy (SIM) to refer to myopathy, severe myopathy, or rhabdomyolysis according to the standardized phenotype statin‐related myotoxicity classifications 3–5.4
Postmarketing surveillance of rare and serious adverse drug reactions (ADRs) is a crucial and integral responsibility of all drug regulatory agencies. In most countries, including Singapore, collection of voluntary reports has been the mainstay of ADR detection,8 but it has several drawbacks, such as underreporting, variable quality of reports, risk of reporting bias, and the lack of a denominator for calculation of incidence.9 Clinicians, even when they suspect an ADR, do not consistently report them to health regulators, especially if the ADR is a known side effect. With the widespread implementation of electronic medical records (EMRs) in clinical practice, there is increasing interest in using these resources to enhance ADR detection8 as a potentially more efficient way to identify ADR cases and to obtain more accurate estimates of incidence.
The preferred methodology for active ADR detection from EMRs is an active area of research and the specific types of databases and methods that work best may be jurisdiction specific, depending on the structure and nature of the local healthcare system. For example, administrative databases are widely used in the United States and have been helpful in estimating the incidence of statin‐induced rhabdomyolysis.5 Building on the high coverage and availability of administrative databases of major health insurance providers, the US Food and Drug Administration embarked on the Sentinel Initiative in 2008, and established “Mini‐Sentinel” to set up a nationally distributed electronic system for monitoring the safety of US Food and Drug Administration‐regulated medical products.15
In Singapore, most healthcare costs are paid by the patient, and large insurance databases that capture treatments and medication prescriptions with acceptable completeness do not exist. However, public healthcare institutions in Singapore have been moving toward EMRs over the past decade, and these databases, albeit still fragmented and formatted primarily for clinical practice, serve as potential resources for active ADR surveillance. Data in Singapore EMRs have less information on diagnosis and procedure codes for billing purposes, and more clinical laboratory results and unstructured text compared to health insurance databases that have been used in data‐mining efforts, such as Sentinel and Observational Medication Outcomes Partnership.16 Research on data‐mining in less structured data, such as those in the Singapore healthcare system, is therefore needed and would support corresponding research efforts in jurisdictions with similar data structures.
A long‐term goal of this work is to apply automated computer routines on EMRs from a broadly representative group of healthcare institutions in Singapore to capture ADRs in real time. A step toward this goal is to focus on specific ADRs of high clinical importance.
    loading  Loading Related Articles