To investigate a potential relationship between four different classes of non-cardiovascular drugs and the risk of sudden cardiac death.

BACKGROUND:

There are more than 400,000 sudden cardiac deaths annually in the U.S, of which 85% or more are caused by ventricular tachyarrhythmias. Medications are an important modifiable risk factor because many have effects that can provoke lethal arrhythmias. There is increasing suspicion that several drugs in four widely used classes of non-cardiovascular medications-fluoroquinolone and macrolide antibiotics, antipsychotics, and antidepressants- are pro-arrhythmic and thus increase the risk of sudden cardiac death. Published epidemiologic studies have quantified the risk conferred by older antipsychotics and antidepressants as well as oral erythromycin. The current project will extend these studies to newer medications that are used by an estimated 20% of adults in the U.S. Studies of surrogate markers suggest that the pro-arrhythmic effects of these drugs vary markedly.

DESIGN NARRATIVE:

This retrospective cohort study has three specific aims in testing the relationship between certain non-cardiovascular medications-fluoroquinolone and macrolide antibiotics, antipsychotics, and antidepressants- and sudden cardiac death.. Specific aim 1 tests the hypothesis that there is corresponding variation in risk of sudden cardiac death. In vivo data suggest that concurrent use of study drugs with other common medications that inhibit their metabolism could markedly increase drug concentrations, and thus risk of arrhythmias. Specific aim 2 tests the hypothesis that these pharmacokinetic interactions, defined a priori, increase risk of sudden cardiac death. The hypokalemia caused by the commonly used potassium-wasting diuretics may amplify the pro-arrhythmic effects of medications. Specific aim 3 tests the hypothesis that concurrent use of study drugs and these diuretics increases risk of sudden cardiac death. The investigators will conduct a retrospective cohort study in TennCare, Tennessee's expanded Medicaid program. Computerized TennCare files, linked with death certificates, provide the information necessary to define the cohort, classify followup according to medication exposure and potential confounders, and identify cases of sudden cardiac death using a validated computer case definition we have developed. The cohort will include an estimated 800,000 persons with 15,000 sudden cardiac deaths during 5,000,000 person years of followup and thus will have excellent power for risk estimates. The study will provide data that clinicians need to prescribe these widely used medications in a way that minimizes the risk of sudden cardiac death.

Study selection criteria are not based on gender, ethnicity or race. Nevertheless, we estimated that 61% of subjects would be females, 73% would be white, and 99% would be non-hispanic/latino.

• Inclusion/exclusion criteria are designed to assure the availability of data necessary for the study and to identify a cohort of patients who, absent adverse medication effects, are at low risk for sudden death. Thus, inclusion criteria require enrollment in TennCare, including access to medications. To assure complete identification of all healthcare encounters and medication use, the study will be restricted to TennCare enrollees with active enrollment and full pharmacy benefits. We require age 30 years or older at the beginning of the study. This is the population for which arrhythmia-related deaths are of greatest concern. Finally, cohort members must have use of study or control medications, as defined by filling at least one prescription recorded in the Medicaid pharmacy files.
• Patients with life-threatening illnesses will be excluded because for such persons it is not possible to distinguish from deaths related to arrhythmias versus those that are a consequence of the underlying serious illness.

: We will conduct a retrospective cohort study of Tennessee Medicaid enrollees 30 years of age or older who meet the study inclusion-exclusion criteria. From the Medicaid pharmacy files, we will identify users of study drugs and controls. During followup, we will identify out-of-hospital deaths, most likely to be sudden deaths, based in part on linkage with Tennessee death certificate files. From these deaths, we will identify those meeting the criteria for sudden cardiac death, based upon a previously validated computer algorithm. Important potential confounders will be identified from past medical care encounters, as identified from Medicaid inpatient (augmented with the Tennessee Hospital Discharge dataset), outpatient, and nursing home files. Adjusted estimates of the relative risk will be calculated from multivariate regression analyses.