Author + information
- Received August 30, 2016
- Revision received October 31, 2016
- Accepted December 2, 2016
- Published online March 29, 2017.
- Bjarke Risgaard, MD, PhDa,∗ (, )
- Bo Gregers Winkel, MD, PhDa,
- Reza Jabbari, MD, PhDa,
- Thomas Hadberg Lynge, BMa,
- Mads Wissenberg, MDb,
- Charlotte Glinge, MDa,
- Stig Haunsø, MD, DMSca,c,
- Elijah R. Behr, MD, DMScd,
- Anders Fink-Jensen, MD, DMSce,
- Gunnar Hilmar Gislason, MD, PhDb,f and
- Jacob Tfelt-Hansen, MD, DMSca,c
- aThe Department of Cardiology, The Heart Centre, Copenhagen University Hospital, Rigshospitalet, Copenhagen, Denmark
- bDepartment of Cardiology, Copenhagen University Hospital, Gentofte, Denmark
- cDepartment of Medicine and Surgery, University of Copenhagen, Copenhagen, Denmark
- dCardiology Clinical Academic Group, St George's University of London, London, United Kingdom
- ePsychiatric Centre Copenhagen, Copenhagen University Hospital and Laboratory of Neuropsychiatry, Department of Neuroscience and Pharmacology, University of Copenhagen, Copenhagen, Denmark
- fThe National Institute of Public Health, University of Southern Denmark, Copenhagen, Denmark
- ↵∗Address for correspondence:
Dr. Bjarke Risgaard, Department of Cardiology, The Heart Centre, Copenhagen University Hospital, Rigshospitalet, Section 2143, Blegdamsvej 9, 2100 Copenhagen Ø, Denmark.
Objectives This study sought to describe the use of pharmacotherapy in a nationwide cohort of young patients with sudden cardiac death (SCD).
Background Several drugs have been associated with an increased risk of SCD and sudden arrhythmic death syndrome (SADS). It remains unclear how pharmacotherapy may contribute to the overall burden of SCD in the general population.
Methods This was a nationwide study that included all deaths that occurred between 2000 and 2009 and between 2007 and 2009 in people aged 1 to 35 years and 36 to 49 years, respectively. Two physicians identified all SCDs through review of death certificates. Autopsy reports were collected. Pharmacotherapy prescribed within 90 days before SCD was identified in the Danish Registry of Medicinal Product Statistics.
Results We identified 1,363 SCDs; median age was 38 years (interquartile range: 29 to 45 years), and 72% (n = 975) were men. Autopsy was performed in 55%. Overall, 58% of SCD cases (n = 786) received at least 1 drug within 90 days before death. The most common drugs were analgesic drugs (n = 239, 18%), antihypertensive drugs (n = 234, 17%), and antibiotic drugs (n = 218, 16%). After multivariable adjustment, prescription of “brugadogenic” drugs or >1 QT-prolonging drug was associated with an increased risk of SADS compared with explained SCD (odds ratios: 2.16 [95% confidence interval: 1.12 to 4.17] and 2.91 [95% confidence interval: 1.46 to 5.81], respectively).
Conclusions Pharmacotherapy was identified in 58% of the SCD cases. After multivariable adjustment, there was a 2- and 3-fold increased risk of SADS compared with explained SCD in patients receiving brugadogenic drugs or >1 QT-prolonging drug, respectively. Identification of high-risk patients is warranted to lower the burden of SCD.
Sudden cardiac death (SCD) in the young is a devastating event, accounting for thousands of deaths each year worldwide (1–3). In the past decade, studies have investigated incidences and causes of death and how to prevent these tragic occurrences (4,5). Rational pharmacotherapy is one of the topics that has been given considerable attention for the prevention of SCD (6).
It has been reported that users of antipsychotic agents have a 2-fold increased risk of SCD compared with nonusers (7). Mood disorders have also been associated with increased risk of SCD (8), and treatment with antidepressant and antipsychotic agents has been associated with increased risk of out-of-hospital cardiac arrest (9,10). Recently, we showed that patients treated for a psychiatric disease in an inpatient or outpatient clinic had a 4.5-fold increased risk of SCD (11).
Most drugs associated with SCD have an increased propensity for prolonging the QT interval, which is considered a risk marker for the life-threatening ventricular arrhythmia torsades de pointes (TdP), or for inducing arrhythmias in patients with Brugada syndrome (12,13). A number of strategies have been suggested to reduce arrhythmias triggered by certain drugs, especially in patients with the long-QT syndrome (LQTS) and Brugada syndrome (14,15). Although there is a large burden of evidence linking certain drugs with increased risk of SCD, it is unclear how pharmacotherapy, including the use of QT-prolonging drugs and drugs to be avoided in patients with Brugada syndrome, contributes to the overall burden of SCD in a nationwide setting (16–18).
In the present study, we aimed to describe the use of pharmacotherapy before death in a nationwide unselected cohort of young SCD cases. The data may help visualize the potential magnitude of drug-induced arrhythmias in the young, improve risk stratification, and influence public health strategies in the future.
The nationwide identification of SCD cases included in this study has been described in detail previously (17,18). In brief, this is a retrospective study using the availability of all death certificates and the registration of all inpatient and outpatient activities in Danish Hospitals and emergency departments, together with access to all medical records and autopsy reports. All deaths in persons 1 to 35 years of age between 2000 and 2009 and all deaths in persons 36 to 49 years of age between 2007 and 2009 were included. The study was approved by the local ethics committee (H-KF-272484), the Danish Data Protection Agency (2011-41-5767), and the Danish National Board of Health (7-505-29-58/6).
Review of death certificates
All death certificates were retrieved digitally and read independently by 2 physicians to identify deaths that were sudden and unexpected. In case of disagreement, a consensus was reached after the circumstances surrounding the death were re-evaluated, including a review of the previous medical history, which was taken into account in each case. Danish death certificates can only be issued by a medical doctor and are informative and suitable to identify sudden unexpected deaths because of a large supplemental information field (see supplemental data of Winkel et al. ). This field is mandatory in all medicolegal external examinations (external examinations), including cases in which it is decided not to conduct an autopsy. In cases in which citizens or patients are found dead or the death is sudden and unexpected, external examinations are mandatory by Danish law. Information from these examinations, describing circumstances surrounding the death, can be found in the supplemental information field on the Danish death certificate.
Conduct of autopsies
In Denmark, an autopsy is performed when an external examination concludes that the mode of death cannot be established. All forensic autopsies are supervised by another forensic pathologist and follow a protocol in which all organs are thoroughly examined. Toxicology screens are performed in unexplained adolescent and adult cases of sudden unexpected deaths. In Denmark, there are 3 forensic departments that conduct approximately 1,200 autopsies per year, but hospital autopsies can be requested by a physician and the relatives when (after external examination) it is decided not to perform a forensic autopsy.
SCD in autopsied cases was defined as a natural unexpected death of unknown or cardiac cause; in witnessed cases as an acute change in cardiovascular status with time to death being <1 h; and in unwitnessed cases as a person last seen alive and functioning normally <24 h before being found dead. Autopsied SCD was subdivided in 2 groups: 1) explained SCD, in which a cardiac cause of death was established; and 2) sudden arrhythmic death syndrome (SADS), in which causes of death after autopsy remained unknown.
In nonautopsied cases, the same criteria were used in cases presumed to be of cardiac origin based on the circumstances relating to the death including all information from death certificates and discharge summaries.
Danish nationwide registries
The assignment of a unique civil registration number to all Danish citizens enables linkage of nationwide registries on an individual level. The Danish National Patient Registry holds information on all inpatient and outpatient activities in Denmark since 1978. Diagnoses are coded according to the International Classification of Diseases (ICD); before 1994, the 8th revision (ICD-8) was used, and from 1994 onward, the 10th revision (ICD-10) was used (19).
The Danish Registry of Medicinal Product Statistics is a complete nationwide register that holds detailed information on all dispensed drug prescriptions from Danish pharmacies since 1995 according to the Anatomical Therapeutic Chemical (ATC) system. All Danish pharmacies are required to register every dispensed prescription because of the partial reimbursement of drug expenses by the tax-financed health care system, which makes the register both valid and accurate (20).
Pharmacotherapy was defined as any claimed prescription ≤90 days before SCD if not otherwise specified. A QT-prolonging drug was defined as a drug that prolongs the QT interval according to the Credible Meds website (21). Drugs categorized as “known risk of TdP,” “possible risk of TdP,” and “conditional risk of TdP” were merged into 1 category if not otherwise specified. Proarrhythmic drugs in patients with Brugada syndrome were identified by use of the Brugada Drugs website (22). Drugs to avoid and drugs preferable to avoid were classified in 1 category and are herein referred to as brugadogenic drugs.
Categorized nominal data were compared with the chi-square test. If any cell values were <5 Fisher exact test was used. Medians were compared with the Wilcoxon rank sum test. Covariates for multivariable models were selected before analysis and included comorbidities and factors considered most relevant (Online Table 1). Sensitivity analyses were performed where appropriate. Model assumptions were tested during analysis and found to be valid unless otherwise addressed. Age was treated as a continuous variable. A receiver operating characteristic curve for the logistic regression model was used to measure the prediction accuracy. We considered a 2-sided p value <0.05 statistically significant. Data management and analyses were performed with SAS version 9.4 (SAS Institute Inc., Cary, North Carolina).
The mean size of the population studied in Denmark was 3.5 million inhabitants. These were followed up for 27.1 million person-years, during which we identified 14,567 deaths that were included in the study. There were 9,756 deaths among persons 1 to 35 years of age (2000 to 2009) and 5,811 deaths in persons 36 to 49 years old (2007 to 2009). After review of death certificates and autopsy reports, the study population consisted of 1,363 SCD cases (Figure 1).
Table 1 includes the clinical characteristics of all SCD cases and provides comparisons of SCDs within different age groups. The median age of all subjects was 38 years (interquartile range: 29 to 45 years); 72% were men (n = 975), and 40% of deaths (n = 485) were witnessed. Most deaths occurred at home (n = 845 [63%]), while the individual was awake and relaxed (n = 596 [56%]). Autopsy was performed in 55% of all SCDs (n = 753) (Table 1). The most common established causes of death were SADS and coronary artery disease (n = 271 [36%] and n = 198 [26%], respectively). Causes of death have been described in detail previously (17,18,23). Table 2 includes comorbidities of the SCD population according to age groups. Psychiatric and cardiac diseases were predominant and found in 22% (n = 297) and 16% (n = 223) of all SCD cases, respectively. All comorbidity frequencies increased with increasing age group (Table 2).
Overall, 58% of SCD cases (n = 786) received at least 1 drug within 90 days before SCD, and the distribution in number of drugs is depicted in Figure 2. SCD cases treated with at least 1 drug increased with increasing age group (1 to 18 years, 19% [n = 23]; 19 to 35 years, 34% [n = 173%]; 36 to 49 years, 49% [n = 356%]).
The most common pharmacotherapy groups were analgesic drugs (n = 239 [18%]), antihypertensive drugs (n = 234 [17%]), and antibiotic drugs (n = 218 [16%]). Most pharmacotherapy groups were increasingly present with increasing age, but a few exceptions were observed; antibiotic drugs and drugs used in obstructive airway diseases were as common in those aged 1 to 18 years as in those aged 19 to 35 years and 36 to 49 years, respectively (Table 3).
Proarrhythmic drugs and SCD
In 19% of all SCD cases (n = 252), a QT-prolonging drug with risk of TdP was administered <90 days before death. The Credible Meds website (21) was used to determine the distribution of drugs with a known risk of TdP, possible risk of TdP, and conditional risk of TdP, respectively, as follows: 37% (n = 92); 46% (n = 117), and 24% (n = 61). QT-prolonging drugs were rare in the youngest age groups but were increasingly prescribed with increasing age: 1 to 18 years, 3% (n = 4); 19 to 35 years, 16% (n = 83); and 36 to 49 years, 23% (n = 165). Figure 3 depicts the distribution of QT-prolonging drugs in autopsied SCD cases stratified by age group and cause of death (explained SCD vs. SADS). In explained SCDs, drugs with increased risk of TdP were prescribed in 13% of cases (n = 62), whereas the same figure was 22% (n = 59) for SADS cases (p < 0.01) (Figure 2). Brugadogenic drugs were administered <90 days before death in 9% of all SCD cases (n = 120). They were rare in the youngest age groups but were increasingly prescribed with increasing age: 1 to 18 years, 3% (n = 4); 19 to 35 years, 7% (n = 37); and 36 to 49 years, 11% (n = 79). In explained SCDs, brugadogenic drugs were prescribed in 5% of cases (n = 23), whereas the same figure was 10% (n = 27) for SADS cases (p < 0.01) (Online Table 2, Online Figure 1).
After multivariable adjustment for age, sex, and known cardiac disease, prescription of >1 QT-prolonging drug was associated with an increased risk of SADS compared with explained SCD, with an odds ratio (OR) of 2.91 (95% confidence interval [CI]: 1.46 to 5.81), whereas prescription of brugadogenic drugs was associated with SADS (OR: 2.16; 95% CI: 1.12 to 4.17) (Online Table 3).
In a sensitivity analysis that included only drugs prescribed within 30 days before death, >1 QT-prolonging drug remained associated with an increased risk of SADS compared with explained SCD (OR: 3.48; 95% CI: 1.85 to 6.52). However, prescription of brugadogenic drugs almost reached statistical significance (OR: 1.81; 95% CI: 1.00 to 3.33). The risk for SADS was not affected by the calendar year. Furthermore, a sensitivity analysis with adjustment for the 2 study periods (2000 to 2006 vs. 2007 to 2009) did not change our findings. The area under the receiver operating characteristic curve for the final logistic regression model was 0.694.
In this study, we examined the use of pharmacotherapy in a nationwide cohort of young SCD victims aged 1 to 49 years. The study has 3 main findings: First, pharmacotherapy was prescribed to 58% of all SCD victims in this age group. Second, the most common drugs used were analgesic drugs, antihypertensive agents, and antibiotic drugs. Third, proarrhythmic drugs were prescribed more often in SADS than in explained SCD cases, and there was almost a 2- and 3-fold higher risk of an arrhythmic death in patients receiving a brugadogenic drug or >1 QT-prolonging drug, respectively.
Several drugs have been associated with an increased risk of SCD (6–8,12). Although previous studies have been restricted to a selected subset of patients in a restricted geographic area, we are the first to investigate this matter in a nationwide cohort of young SCD cases. In 2009, Ray et al. (7) linked users of typical and atypical antipsychotic agents with an increased risk of SCD. The study was a retrospective cohort study of Medicaid enrollees in Tennessee, and the endpoint, SCD, was retrieved from death certificates. However, U.S. death certificates have previously been shown to have little value in SCD identification (24). The same conclusions were reached by Jolly et al. (12) in their investigation of coroner cases in selected areas of England. Others have used an out-of-hospital cardiac arrest as a proxy for SCD and associated these drugs with increased risk (9,10). Most recently, Cheng et al. (25) conducted a meta-analysis to examine the link between administration of macrolides and risk of SCD.
As illustrated above, it remains unknown how much pharmacotherapy and proarrhythmic drugs potentially contribute to the overall burden of SCD in a young, general population. We also address the increased risk of SADS compared with explained SCD among users of proarrhythmic drugs.
Pharmacotherapy and SCD
During the past decade, the number of prescribed drugs in the general population has increased significantly (26). Use of antipsychotic drugs, including anxiolytics, hypnotics, and sedatives, has nearly doubled between 1998 and 2002 (27), and the same trend has been shown for both antidepressant drugs and analgesic drugs (28,29). The increasing use of these drugs suggests that an estimate of the overall influence these drugs might have on the incidence of SCD is of considerable interest. We demonstrated that in this young SCD cohort aged 1 to 49 years, drugs were prescribed in 58% of the cases within 90 days before death. However, we also demonstrated that pharmacotherapy was nearly absent in children (1 to 18 years old), whereas in adolescence, there was a steep increase in the use of these drugs with increasing age.
Drug-induced arrhythmic death
One proposed mechanism by which most drugs are thought to induce SCD is by blocking the rapid component of the delayed rectifying potassium current (IKr). This will lead to prolonged cardiac repolarization and thereby, in turn, increased risk of TdP (6). The drug-induced effect can be measured as a prolonged QT interval on the surface electrocardiogram (ECG), and for virtually all QT-prolonging drugs, risk increases in a dose-dependent manner (15). Although the QT interval is imperfectly correlated with the risk of arrhythmias, it is currently the best available tool for risk stratification in individuals susceptible to drug-induced arrhythmia (14,15).
The true incidence of drug-induced arrhythmias is difficult to establish, despite evidence from multiple observational studies. There are 2 main explanations for this. First, TdP in outpatient settings is often unwitnessed, and by the time first responders arrive on scene with portable defibrillators, the arrhythmia is likely to have degenerated into ventricular fibrillation or even asystole (15). Second, autopsies, which could rule out other causes of lethal arrhythmias such as myocardial infarction, are not always conducted. As a result, the exact incidence of death caused by primary arrhythmia will be unattainable. Primary arrhythmias do not leave any traces after an autopsy, and these arrhythmic deaths are therefore often referred to as SADS. Notably, we found that case subjects who experienced unexplained deaths more often were treated with a QT-prolonging or brugadogenic drug before death, and in a regression model, there was nearly a 2- and 3-fold risk of SADS compared with explained SCD. We conducted a sensitivity analysis that included all proarrhythmic drugs prescribed <30 days before death. In this analysis, >1 QT-prolonging drug remained associated with an increased risk of SADS (compared with explained SCD), with an OR of 3.5, whereas brugadogenic drugs did not quite reach statistical significance. Overall, these findings suggest that these deaths in fact could have been caused by an arrhythmia that was in part explained by the exposure to a drug (Figure 3). Importantly, all drugs were found in nonlethal concentrations during toxicology screening, and hence, no cases were caused by (intentional or unintentional) intoxication.
Clinical implications and genetic susceptibility
SADS is the most common cause of SCD in the those aged <36 years in Denmark (17,18,30). The current findings, however, also suggest that in those age <19 years, the proportion of deaths that are potentially drug induced remains negligible. Although it has been repeatedly reported that use of some drugs is associated with increased risk of SCD, a causative link has been difficult to prove (7,10,12). It is clear that in patients who have psychiatric diseases such as schizophrenia or depression, a treatment-associated risk of SCD may be unavoidable (31); however, if left untreated, some of these diseases can also increase the risk of death by themselves.
Risk factors for the development of TdP in hospitalized patients have been published previously (15). The identification of high-risk patients with comorbidities such as ischemic heart disease, heart failure, pre-existing long-QT interval, or electrolyte disturbances could probably decrease the SCD risk significantly. Furthermore, in patients with manifest drug-induced QT prolongation and TdP, it has also become increasingly evident that the presence of LQTS-causing mutations must be considered (15). Three of the 12 LQTS-susceptibility genes account for nearly 75% of all congenital LQTS cases (32) and are similarly implicated in 10% of drug-induced TdP cases (33). Thus, a preventative strategy should involve ECG screening before and after initiation of QT-prolonging medications, supplemented by genetic screening as required.
This study was conducted on a Danish population, which may be relatively homogenous, and therefore, the study findings may not be widely applicable. In addition, the study has limitations inherent to any retrospective study. In witnessed cases, it was easy to extract information on whether the person was seen alive <1 h before death, but it was difficult to assess more precise time limits in unwitnessed cases. We assumed that all patients who claimed a prescription within 90 days before SCD were likely to take the medication, because they also had an economic incentive because the medical expenses were only being partially reimbursed by the Danish government–financed health care system. We arbitrarily chose to include only drugs prescribed within 90 days before SCD; by lengthening this period, we might include patients not treated at the time of death, whereas shortening it might exclude patients who were treated at the time of death. We were not able to control for alterations in drug dosages, nor were we able to control for changes in the metabolization of these drugs. No ECGs were available, and in people who received a QT-prolonging or brugadogenic drug, an altered QT segment could have preceded the SCD. We were unable to determine whether each SADS case was preceded by an arrhythmic event, but we believe the most plausible explanation for several of these deaths is a blockade of the potassium channels that resulted in prolongation of cardiac repolarization. However, other mechanisms could be involved, including autonomic effects or inhibition of other ion channels, such as the sodium channels.
Cause of death categorization (explained SCD vs. SADS) was not affected by the toxicology profile, which was delayed up to weeks after the autopsies were conducted, which resulted in a preliminary conclusion; it also was not available during review of death certificates. Although forensic pathologists follow standardized protocols, we cannot exclude the possibility that pre-existing evidence of intake of possible proarrhythmic drugs could have biased results in a few cases.
The study was strictly observational; it does not prove causation, nor does it assess the absolute risk of SCD. Nevertheless, this is the first nationwide study to assess the overall burden that pharmacotherapy potentially may have on SCD in the general population.
In a nationwide cohort of young SCD victims, we found that pharmacotherapy was prescribed to 58% of the cases. The most common drugs used were analgesic, antihypertensive, and antibiotic agents. Drugs known to prolong the QT interval and thereby increase the risk of TdP were more often prescribed to SADS cases, although very rarely to children (1 to 18 years of age). Patients receiving >1 QT-prolonging drug had nearly a 3-fold increased risk of SADS compared with explained SCD cases, whereas patients receiving brugadogenic agents had a 2-fold increased risk. These findings indicate that patients treated with these drugs have an increased susceptibility for an arrhythmic event. Our results also highlight that an increased focus on identification of individuals at high risk for SCD among patients receiving proarrhythmic drugs is warranted.
COMPETENCY IN MEDICAL KNOWLEDGE: Compared with explained SCD cases, patients receiving >1 QT-prolonging drug had nearly a 3-fold risk of experiencing SADS, whereas patients receiving brugadogenic agents had a 2-fold increased risk.
TRANSLATIONAL OUTLOOK: Identification of individuals at high risk for SCD among patients receiving proarrhythmic drugs is warranted in future studies.
For supplemental tables and figure, please see the online version of this paper.
Dr. Fink-Jensen received an unrestricted research grant from Novo Nordisk A/S. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Gislason and Tfelt-Hansen contributed equally to this work.
- Abbreviations and Acronyms
- confidence interval
- International Classification of Diseases
- long-QT syndrome
- odds ratio
- sudden arrhythmic death syndrome
- sudden cardiac death
- torsades de pointes
- Received August 30, 2016.
- Revision received October 31, 2016.
- Accepted December 2, 2016.
- 2017 American College of Cardiology Foundation
- Zheng Z.J.,
- Croft J.B.,
- Giles W.H.,
- Mensah G.A.
- de Vreede-Swagemakers J.J.,
- Gorgels A.P.,
- Dubois-Arbouw W.I.,
- et al.
- Fishman G.I.,
- Chugh S.S.,
- Dimarco J.P.,
- et al.
- Kaltman J.R.,
- Thompson P.D.,
- Lantos J.,
- et al.
- Whang W.,
- Kubzansky L.D.,
- Kawachi I.,
- et al.
- Weeke P.,
- Jensen A.,
- Folke F.,
- et al.
- Fanoe S.,
- Kristensen D.,
- Fink-Jensen A.,
- et al.
- Drew B.J.,
- Ackerman M.J.,
- Funk M.,
- et al.
- Roger V.L.,
- Go A.S.,
- Lloyd-Jones D.M.,
- et al.
- Risgaard B.,
- Winkel B.G.,
- Jabbari R.,
- et al.
- Winkel B.G.,
- Holst A.G.,
- Theilade J.,
- et al.
- Lynge E.,
- Sandegaard J.L.,
- Rebolj M.
- Kildemoes H.W.,
- Sørensen H.T.,
- Hallas J.
- ↵Credible Meds website. Available at: https://www.crediblemeds.org. Accessed June 10, 2016.
- Risgaard B.,
- Lynge T.H.,
- Wissenberg M.,
- et al.
- Chugh S.S.,
- Jui J.,
- Gunson K.,
- et al.
- Cheng Y.-J.,
- Nie X.-Y.,
- Chen X.-M.,
- et al.
- ↵Anonymous. Analyser: Sundhed og medicinforbrug–Apotekerforeningen.dk. Available at: http://www.apotekerforeningen.dk/analyser-og-fakta/analyser/sundhed-og-medicinforbrug.aspx. Accessed July 29, 2014.
- Winkel B.G.,
- Risgaard B.,
- Sadjadieh G.,
- Bundgaard H.,
- Haunsø S.,
- Tfelt-Hansen J.
- Tester D.J.,
- Will M.L.,
- Haglund C.M.,
- Ackerman M.J.
- Behr E.R.,
- Roden D.