OUP user menu

Non-cardiac QTc-prolonging drugs and the risk of sudden cardiac death

Sabine M.J.M. Straus, Miriam C.J.M. Sturkenboom, Gysèle S. Bleumink, Jeanne P. Dieleman, Johan van der Lei, Pieter A. de Graeff, Jan Herre Kingma, Bruno H.Ch. Stricker
DOI: http://dx.doi.org/10.1093/eurheartj/ehi312 2007-2012 First published online: 11 May 2005

Abstract

Aims To assess the association between the use of non-cardiac QTc-prolonging drugs and the risk of sudden cardiac death.

Methods and results A population-based case–control study was performed in the Integrated Primary Care Information (IPCI) project, a longitudinal observational database with complete medical records from more than 500 000 persons. All deaths between 1 January 1995 and 1 September 2003 were reviewed. Sudden cardiac death was classified based on the time between onset of cardiovascular symptoms and death. For each case, up to 10 random controls were matched for age, gender, date of sudden death, and general practice. The exposure of interest was the use of non-cardiac QTc-prolonging drugs. Exposure at the index date was categorized into three mutually exclusive groups of current use, past use, and non-use. The study population comprised 775 cases of sudden cardiac death and 6297 matched controls. Current use of any non-cardiac QTc-prolonging drug was associated with a significantly increased risk of sudden cardiac death (adjusted OR: 2.7; 95% CI: 1.6–4.7). The risk of death was highest in women and in recent starters.

Conclusion The use of non-cardiac QTc-prolonging drugs in a general population is associated with an increased risk of sudden cardiac death.

  • Sudden cardiac death
  • Non-cardiac QTc-prolonging drugs
  • Pharmaco-epidemiology
  • Population based

Introduction

In developed countries, sudden cardiac death is one of the major causes of cardiovascular mortality.1 According to the most recent definition, sudden cardiac death is a natural death due to cardiac causes, heralded by abrupt loss of consciousness within 1 h after the onset of acute symptoms or an unwitnessed, unexpected death of someone seen in a stable medical condition <24 h previously with no evidence of a non-cardiac cause.13 Probably, the large majority of cases of sudden cardiac death is due to ventricular fibrillation.2,4 The major unanswered question in sudden cardiac death, however, is which precipitating event causes arrhythmia in an otherwise stable patient. Probably, it is a complex interplay among myocardial injury, chronic and acute coronary events, autonomic tone, electrolyte state, drugs, and genetic factors that determine the occurrence of a life-threatening arrhythmia.5,6

In the past decade, one of the most frequent causes of drug restriction or withdrawal has been prolongation of the QTc interval associated with fatal cardiac arrhythmias. Lengthening of QTc interval represents the prolongation of the action potential and is used as a surrogate marker for the prediction of this serious adverse drug effect.7 The website of the International Registry for Drug-induced Arrhythmias maintained by the Georgetown University (http://www.qtdrugs.org/medical-pros/drug-lists/drug-lists.htm) provides an up-to-date list of drugs that prolong the QTc interval and/or induce Torsades de Pointes or ventricular arrhythmia.8 The researchers have classified drugs into four categories varying from drugs that are generally considered to confer a risk of Torsade de Pointes (list 1) and drugs that, when used in usual dosages, are unlikely to increase the risk (list 4). This potentially fatal adverse reaction is not only associated with cardiovascular drugs, but also with non-cardiovascular drugs. The risk of arrhythmias and sudden cardiac death during the use of non-cardiac QTc-prolonging drugs has attracted considerable clinical and regulatory attention.7,9

Therefore, our objective was to assess the risk of sudden cardiac death and the use of non-cardiac QTc-prolonging drugs in a case–control study in a well-defined general population with complete coverage of all relevant health care information.

Methods

Setting

All data were retrieved from the IPCI project, a longitudinal observational database, containing data from computer-based medical records of a group of 150 general practitioners (GPs) in The Netherlands. In the Dutch health care system, the GP has a pivotal role by acting as a gatekeeper for all medical care. Nearly every citizen is enrolled in the practice of a GP independent of health status.10 Details of the database have been described elsewhere.10,11 Briefly, the database contains the complete medical records of ∼500 000 citizens. The electronic records contain coded and anonymous data on demographics, symptoms (in free text), and diagnoses (using the International Classification for Primary Care12 and free text) from GPs and specialists, referrals, laboratory findings, hospitalizations, and drug prescriptions, including their indications and dosage regimen. To maximize completeness of the data, GPs participating in the IPCI project are not allowed to maintain a system of paper-based records besides the electronic medical records. The project complies with European Union guidelines on the use of medical data for medical research and has been proved valid for pharmaco-epidemiological research in several validation studies that evaluated the quality of the available information.11 The Scientific and Ethics Advisory Board of the IPCI project approved this study.

Source population

The source population comprised all subjects of 18 years and older, who were registered with a GP participating in the IPCI project for at least 1 year. Subjects with a diagnosis of cancer were excluded from the source population, because in these patients the cause of death is often difficult to assess and usually not unexpected. The study period started on 1 January 1995 and ended on 1 September 2003. All subjects were followed until death, transferal out of practice, date of last data collection, or end of the study period, whichever came first.

Case and control definition

The computerized medical and demographic data were screened for deaths that occurred during the study period. The medical records of identified cases of death were reviewed manually to assess whether death could be classified as sudden cardiac death. Validation was performed independently by two physicians who were blinded to exposure (S.M.J.M.S. and G.S.B.) and in the case of discrepancy, a third expert (B.H.Ch.S.) arbitrated. Case assessment was based on the most recent definition of sudden cardiac death: a natural death due to cardiac causes heralded by abrupt loss of consciousness within 1 h after the onset of acute symptoms or an unwitnessed, unexpected death of someone seen in a stable medical condition <24 h previously with no evidence of a non-cardiac cause.13 Cases were classified as (probable) sudden cardiac death if the medical record indicated that death occurred within 1 h after the onset of cardiovascular symptoms, and if the following wording was found in the free text: ‘sudden cardiac death’, ‘acute cardiac death’, ‘mors subita’, ‘sudden death’, ‘died suddenly’, ‘died unexpectedly’, or if this was an unwitnessed, unexpected death of someone seen in ‘good health’ or in a stable medical condition <24 h previously and without evidence of a non-cardiac cause (e.g. pneumonia, convulsion, choking, or stroke). Suicides were excluded. To each case of sudden cardiac death, up to 10 controls were randomly drawn from the source population matched for age (year of birth), gender, and practice. If less than 10 controls were available, all of them were included. The index date was defined as the date on which sudden cardiac death occurred in the cases. This date was also the index date for matched controls.

Exposure definition

The exposure of interest was the use of non-cardiac QTc-prolonging drugs, as specified in the most recent version of list 1 (drugs that are generally accepted by authorities to have a risk of causing Torsades de Pointes) from the International Registry for Drug-induced Arrhythmias maintained by the Georgetown University (http://www.qtdrugs.org/medical-pros/drug-lists/drug-lists.htm)8 and comprise the following non-cardiac QTc-prolonging drugs: chloroquine, chlorpromazine, cisapride, clarithromycin, domperidone, droperidol, erythromycin, halofrantine, haloperidol, levomethadyl, mesoridazine, pentamidine, pimozide, sparfloxacin, and thioridazine. As not all drugs are licensed, marketed, or prescribed in The Netherlands, some of these drugs were not included in our analyses. Our analyses included gastro-intestinal QTc-prolonging medication: cisapride, domperidone; antibiotic QTc-prolonging medication: erythromycin, clarithromycin; and antipsychotic QTc-prolonging medication: chlorpromazine, haloperidol, pimozide, and thioridazine. In order to classify the use at the index date, we calculated the duration of each prescription, as the total number of units issued per prescription divided by the number of units prescribed daily. Exposure at the index date was categorized into three mutually exclusive groups of current-, past-, and non-use. To account for the differences in prescription patterns of non-cardiac QTc-prolonging drugs, different risk windows were specified. For non-cardiac QTc-prolonging drugs, normally prescribed for shorter periods of time (gastro-intestinal medications and antibiotics), use was defined as current if the index date fell within a period of use or within a maximum of 7 days after the end of the last prescription (to account for carry-over effects). For drugs usually prescribed for longer periods of time (antipsychotics), use was defined as current if the index date fell within a period of use or within a maximum of 30 days after the end of the last prescription. Past use was defined as discontinuation of a non-cardiac QTc-prolonging drug before the index date or discontinuation >7 days (gastro-intestinal or antibiotic medication) or >30 days before the index date (antipsychotic medication). If patients had no prescription for any type of non-cardiac QTc-prolonging drug prior to the index date they were considered non-exposed. Among current users, we evaluated the effect of duration (≤90 and >90 days continuous use), type of non-cardiac QTc-prolonging drug, and the current daily dose used in defined daily dose equivalents (DDD), as defined by the World Health Organization.13 One DDD-equivalent represents the recommended daily dose for an adult for the main indication. To evaluate dose response effects, the current daily dose of a non-cardiac QTc-prolonging drug was categorized into less than one DDD equivalent and one or more DDD equivalents.

Co-variates and risk factors

Known risk factors for sudden cardiac death and other co-variates were gathered from the medical records through computerized searches and manual validation. The co-variates that were evaluated included cerebrovascular and cardiovascular ischaemia (history of myocardial infarction, stroke, and angina pectoris), heart failure, hypertension, diabetes mellitus, arrhythmia, hypercholesterolaemia, smoking, and alcohol abuse. Cerebrovascular ischaemia, cardiovascular ischaemia, and heart failure were assessed, based on the diagnoses provided by the GP and specialists in the medical records. Hypertension was identified through the diagnoses in the medical records, the use of antihypertensive medication and/or the blood pressure measurements, meeting the guidelines of the World Health Organization (a blood pressure exceeding 140 mmHg systolic and/or 90 mmHg diastolic).14 Diabetes mellitus, arrhythmias, and hypercholesterolaemia were identified through diagnoses in the medical records from GPs and specialists and/or the use of antidiabetic, antiarrhythmic, or lipid-lowering medication. Information on smoking and alcohol abuse was obtained from the free text in the medical records. As concomitant medication, we considered among other diuretics, angiotensin-converting enzyme inhibitors, and other cardiovascular medication. Current use of concomitant medication was defined as use at the index date. We evaluated the effect of social-economic status (SES), by including a variable on health care insurance, which is a proxy for income (all below an income of about $25 000 per year have sick fund insurance and those above have a private insurance).

Statistical analysis

The relative risk for sudden cardiac death during the use of non-cardiac QTc-prolonging drugs was estimated by calculation of the odds ratios (95% CI) using conditional logistic regression analyses. We evaluated risk factors and confounders one by one, by including one factor with the exposure variable. Subsequently, we performed a bivariate evaluation for potential multicollinearity of confounders. Thereafter, co-variates that were univariately associated with sudden cardiac death (at a P<0.1 level) were included in the regression analyses. After doing this for each factor separately, we retained a list of all factors that changed the point estimate of the association between non-cardiac QTc-prolonging drugs and sudden cardiac death by >5%, which were included in the final model.15 We investigated potential effect modification by age and gender and performed sub-analyses to evaluate potential misclassification of sudden cardiac death by splitting the outcome between witnessed and unwitnessed deaths. To evaluate a possible dose-effect relation, a trend test was performed. We calculated the population attributable risk (PAR) per cent.16

Results

In the source population, 806 cases of sudden cardiac death were identified, representing an incidence rate of sudden cardiac death of almost one per 1000 person-years. As we adhered strictly to the matching criteria to ensure internal validity, the 31(4%) cases for whom no controls could be found were excluded from further analyses. Hence, the study population comprised 775 cases of sudden cardiac death and 6297 matched controls (approximate case:control ratio, 1:8). The median age of the study population was 72 years and ∼61% were male. Despite the matching for age (year of birth), the median age of cases was higher than the median age of all controls (74 and 72 years, respectively) because more controls were available for younger cases than for elderly cases (Table 1). There were 437(56.4%) witnessed and 338(43.6%) unwitnessed cases of sudden cardiac death reported. All known potential risk factors were associated with an increased risk for sudden cardiac death, notably ischaemic cerebrovascular and cardiovascular disease, hypertension, arrhythmia, diabetes mellitus, heart failure, hypercholesterolaemia, smoking, and alcohol abuse. As expected, use of cardiovascular medication was associated with sudden cardiac death as well (Table 1). There was no association between SES and sudden cardiac death.

View this table:
Table 1

Demographics, distribution of covariates, and use of concomitant medication in cases and controls

CharacteristicCases (n=775)Controls (n=6297)OR (95% CI)a
Gender
 Male465(60)3842(61)
 Female310(40)2455(39)
Age (mean±SD) (Years)71 years, 1369 years, 13
 ≤5595(12.3)1032(16.4)
 56–65108(13.9)1127(17.9)
 66–75224(28.9)2014(32.0)
 >75348(44.9)2124(33.7)
Sudden cardiac death
 Witnessed437(56.4)
 Unwitnessed338(43.6)
Co-morbidities
 Ischaemic cerebro-/cardiovascular disease246(32)1022(16)2.3(1.9–2.7)
 Hypertension444(57)3134(50)1.3(1.1–1.5)
 Arrhythmia151(19)670(11)1.9(1.6–2.4)
 Diabetes mellitus230(30)1058(17)2.0(1.7–2.4)
 Smoking98(12.6)495(8)2.2(1.7–2.9)
 Alcohol abuse87(11)536(8)1.7(1.3–2.3)
 Heart failure219(28)468(7)4.8(3.9–5.8)
 Hypercholesterolaemia47(6.1)266(4.2)1.7(1.2–2.4)
Concomitant cardiovascular drugs
 β-blocking agents113(14.6)727(11.5)1.3(1.0–1.6)
 Diuretics190(24.5)677(10.8)2.5(2.0–3.1)
 Calcium channel-blockers81(10.5)446(7.1)1.5(1.1–1.9)
 ACE-inhibitors146(18.8)590(9.4)2.2(1.8–2.8)
 ATII-blocking agents27(3.5)108(1.7)2.1(1.4–3.3)
 Lipid lowering agents45(5.8)348(5.5)1.2(0.9–1.7)
 Cardiac glycosides74(9.5)168(2.7)3.3(2.4–4.5)
 Anti-arrhythmics11(1.4)47(0.7)1.8(0.9–3.6)

Ca, Calcium; ATII, angiotensin II receptor antagonist. Values in 1st and 2nd columns represented as n(%).

aAll ORs are calculated taking matching for age, gender, practice, and calendar time into consideration.

Current use of non-cardiac QTc-prolonging drugs was associated with an almost three-fold increased risk of sudden cardiac death (Table 2). Past use of non-cardiac QTc-prolonging drugs was not associated with an increased risk of sudden cardiac death. The risk was higher among recent starters (≤90 days) of non-cardiac QTc-prolonging drugs (Table 2). The risk was significantly increased in users of gastro-intestinal medication and antipsychotics. The risk of sudden cardiac death was not significantly increased in antibiotic users, probably due to the limited number of exposed cases. Antibiotics were only used for <90 days and only in a dosage of one or more DDD equivalents. Therefore, we could not assess duration and dose–response relationships (Table 2). The risk of sudden cardiac death was increased to a larger extent in users of a higher daily dose of gastro-intestinal or of antipsychotic medication. The risk was highest in subjects using antipsychotics, predominantly haloperidol. In users of gastro-intestinal medication, the risk of sudden cardiac death was significantly increased in users of domperidone in contrast to cisapride users, although a three-fold higher risk of sudden cardiac death in cisapride users could not be excluded (Table 2).

View this table:
Table 2

Risk of sudden cardiac death and the use of non-cardiac QTc-prolonging medication

Use of non-cardiac QTC-Prolonging medicationCases (n=775)Controls (n=6297)OR (95% CI)aOR (95% CI)b
Overall
 Non-use65355441.0 (reference)1.0 (reference)
 Past use986961.2 (0.9–1.5)0.9 (0.7–1.2)
 Current use24573.4 (2.0–5.6)2.7 (1.6–4.7)
Current use of different QTc-prolonging medications
 Gastro-intestinal drugsc13432.7 (1.4–5.0)2.1 (1.1–4.2)
  Cisapride4291.2 (0.4–3.5)1.2 (0.4–3.3)
  Domperidone9155.4 (2.2–12.7)3.8 (1.5–9.7)
 Antipsychotic drugs795.1 (1.8–14.4)5.0 (1.6–15.3)
  Chlorpromazine1
  Haloperidol684.7 (1.5–14.4)5.6 (1.6–18.7)
  Pimozide1
 Antibiotic drugs455.4 (1.4-22.1)3.7 (0.9-15.7)
  Erythromycin1
  Clarithromycin353.7 (0.8–17.0)3.2 (0.7–15.0)
Duration of use in current users
 Non-use65355441.0 (reference)1.0 (reference)
 ≤90 days17284.5 (2.4–8.6)3.6 (1.9–7.2)
 >90 days7292.1 (0.9–7.2)1.7 (.07–4.2)
DDD
 Gastro-intestinal drugs1343
  <1DDD3122.2 (0.6–8.0)2.0 (0.5–7.4)
  ≥1DDDd10312.7 (1.3–5.8)2.2 (1.0–4.8)
 Antipsychotic drugs79
  <1DDD585.0 (1.5–17.0)4.8 (1.4–16.1)
  ≥1DDDd216.1 (0.3–124)6.4 (0.3–129.1)
 Antibiotic drugs45
  <1DDD
  ≥1DDD455.6 (1.4–22.5)3.8 (0.9–16.0)

aOdds ratios matched for age, gender, practice, and calendar time.

bOdds ratio matched for age, gender, practice, and calendar time and adjusted for diabetes mellitus, arrhythmias, heart failure, hypertension, smoking, alcohol abuse, and cerebrovascular and cardiovascular ischaemia, current use of diuretics and cardiac glycosides.

cSince one patient used domperidone and cisapride concomitantly, numbers do not add up.

dTest for trend P<0.005.

For witnessed cases of death, the association with current use of non-cardiac QTc-prolonging drugs was higher (OR: 2.9; 95% CI: 1.5–5.9) than those for unwitnessed cases (OR: 2.3; 95% CI: 1.0–5.4), but this difference was not statistically significant. Stratified analyses showed that the risk of sudden cardiac death in users of non-cardiac QTc-prolonging medications tended to be higher in women (OR: 3.1; 95% CI: 1.5–6.4) than in men (OR: 2.3; 95% CI: 1.0–5.1) and higher in patients older than 65 years (OR: 2.7; 95% CI: 1.5–4.9) than in patients 65 years or younger (OR: 2.3; 95% CI: 0.6–8.1), but none of these differences were statistically significant. The incidence of sudden cardiac death in our population was almost one per 1000 person-years.17 On the basis of the findings in this study, the PAR per cent of non-cardiac QTc-prolonging medication could be calculated as 2%.16

Discussion

The results of our study indicate that current use of non-cardiac QTc-prolonging drugs in a general population is associated with a significantly increased risk of sudden cardiac death. After adjustment for known confounding factors, current use of non-cardiac QTc-prolonging drugs was associated with an almost three-fold increased risk of sudden cardiac death. The risk was higher in women than in men. This is in line with earlier findings that women seem to be more susceptible to drug-induced cardiac arrhythmias than men.7,18

Drug-induced Torsades de Pointes is a significant cause of morbidity and mortality and non-cardiac drugs have been implicated in an increasing number of cases.7,1921 Prolongation of the QTc interval, ventricular arrhythmias and Torsades de Pointes have been associated with cisapride use in anecdotal reports. These case reports raised concern, but could not be confirmed in cohort and case–control analyses in the UK and Canada and in some studies no substantial increase in the QTc interval could be identified.7 Our results are in line with these findings.22 This is not surprising because life-threatening arrhythmias during cisapride use in adults are believed to occur predominantly, when there is concurrent use of inhibitors of cytochrome P450 CYP3A4, such as itraconazole, ketocaonazole, or macrolides. This was not the case in our population. Domperidone seemed an attractive, safer alternative to cisapride. Our findings of a significantly increased risk of sudden cardiac death in domperidone users, however, suggest that domperidone should not be viewed as a low risk alternative to cisapride and are in line with earlier results.23 Unfortunately, the number of exposed cases in our study prohibited us from analysing the macrolides individually. At an FDA advisory committee meeting, the relative reporting rate of Torsades de Pointes was increased in clarithromycin users and the risk was highest shortly after initiating therapy, as was also the case in our data.24 Case reports have suggested for long that erythromycin is associated with an increased risk of Torsades de Pointes and recent research has shown a two-fold increase in the rate of sudden cardiac death in current erythromycin users.25 Since the early 1960s, sudden cardiac death has been reported with antipsychotic use in case reports and epidemiological studies.2628 Also in our study, the use of antipsychotics, particularly haloperidol is associated with a significant increase in the risk of sudden cardiac death.

Many drugs can prolong the QTc interval.19,21,29,30 QT prolongation, however, is a surrogate marker with an imperfect predictive value for fatal cardiac arrhythmias and sudden cardiac death and it is difficult to predict whether a drug will cause Torsades de Pointes.21 For the vast majority of drugs known to induce QTc prolongation, it has been demonstrated that the slowing of the action potential is a consequence of the blockage of the rapid component of the delayed rectifier potassium channel (Ikr) through blockade of the human ether a go-go related gene (HERG).21,3133 The extent to which blocking of this channel results in Torsades de Pointes or sudden cardiac death, however, is highly variable among subjects.7,20,22,3436 A unifying concept ‘reduced repolarization reserve’ has been used to explain this variable risk.36 Current evidence also suggests that 5–10% of persons in whom cardiac arrhythmias occur carry a (silent) mutation in one of the genes responsible for the congenital long QT syndrome.37,38

In our population, we were able to take advantage of the fact that in the Dutch health care system, all medical information (including specialist and hospital care) are collected at practices that cover the general population instead of selected socio-economic groups. As a consequence, there was extensive information available on drug use, potential confounders, and all the circumstances surrounding death. Nevertheless, our study has some potential limitations. First, we cannot exclude that some misclassification of outcome occurred. We may have missed some deaths although this will be minimal, since death is consistently registered by GPs. Secondly, not all acute deaths may have been of cardiac origin. We determined sudden cardiac death, however, on the basis of the full medical records and all circumstances surrounding the death were available. Recently, an evaluation comparing different methods to determine the incidence of sudden cardiac death suggested that this method provides a very reliable way of determining sudden cardiac death cases.3 In addition, we could reduce misclassification by differentiating between witnessed and unwitnessed cases of sudden cardiac death. The percent of unwitnessed deaths in our population was 43.6% which is in line with earlier findings.3,39 The risk of witnessed sudden cardiac death associated with the use of non-cardiac QTc-prolonging drugs was higher than the risk of unwitnessed sudden cardiac death. This may be consistent with the fact that misclassification will occur more often in unwitnessed cases. In addition, slight misclassification of exposure may have occurred because we used outpatient prescription data and we had no information as to whether the prescription was actually filled and taken. It is likely, however, that such exposure misclassification will be random and will be evenly distributed among cases and controls. Although we adjusted for all known confounders, residual confounding may exist, but is unlikely to explain the strong association we have observed.

In conclusion, our findings suggest that the current use of non-cardiac QTc-prolonging drugs in a general population is associated with an increased risk of sudden cardiac death. Although prolongation of the QTc interval by non-cardiac drugs is not an unusual finding, potentially fatal arrhythmias and sudden cardiac death are relatively uncommon. Our results suggest that 320 cases of sudden cardiac death can be attributed to the use of non-cardiac QTc-prolonging medication in The Netherlands on a yearly basis.16,17 This is important because regulatory authorities have to evaluate the clinical significance of QTc prolongation observed in relatively small clinical trials without cases of sudden cardiac death.

References

View Abstract