Effects of antidepressants on QT interval in people with mental disorders.

INTRODUCTION: Drug-induced QT prolongation is associated with higher cardiovascular mortality.
MATERIAL AND METHODS: We conducted a protocol-based comprehensive review of antidepressant-induced QT prolongation in people with mental disorders.
RESULTS: Based on findings from 47 published randomized controlled trials (RCTs), 3 unpublished RCTs, 14 observational studies, 662 case reports of torsades de pointes, and 168 cases of QT prolongation, we conclude that all antidepressants should be used only with licensed doses, and that all patients receiving antidepressants require monitoring of QT prolongation and clinical symptoms of cardiac arrhythmias. Large observational studies suggest increased mortality associated with all antidepressants (RR = 1.62, 95% CI: 1.60-1.63, number of adults: 1,716,552), high doses of tricyclic antidepressants (OR = 2.11, 85% CI 1.10-4.22), selective serotonin reuptake inhibitors (OR = 2.78, 95% CI: 1.24-6.24), venlafaxine (OR = 3.73, 95% CI: 1.33-10.45, number of adults: 4,040), and nortriptyline (OR = 4.60, 95% CI: 1.20-18.40, number of adults: 5,298).
CONCLUSIONS: Evidence regarding the risk of QT prolongation in children is sparse.

Introduction

Observational studies provide consistent evidence that prolonged QT interval is associated with higher risk of all-cause and cardiovascular mortality [1]. Drug-induced prolongation of QT contributes to higher mortality [2, 3]. The risk of drug-induced prolongation of QT is much higher in older adults and people with multiple chronic conditions [4]. Doctors often prescribe antidepressants for licensed and off-label indications without careful assessment of baseline risk for drug-induced prolongation of QT [5, 6]. The U.S. Food and Drug Administration released several warning statements concerning the elevated risk of prolonged QT interval and potentially fatal torsades de pointes arrhythmia associated with a higher dose of citalopram during post-marketing surveillance [7, 8]. Safety of other antidepressants with respect to QT interval has been examined in cross-over trials on healthy volunteers (Supplementary Table SI) [9-21]. This rapid review focuses on all available evidence regarding the effects of antidepressants on the QT interval in children and adults with mental disorders.

Supplementary Tables

Click here for additional data file.

Material and methods

We used a standard recommended methodology in conducting systematic literature reviews and meta-analyses from the Cochrane Collaboration and the Agency for Healthcare Research and Quality [22, 23]. We developed an a priori protocol for a systematic literature review to answer the clinical question about the safety of antidepressants with respect to QT interval in children and adults with mental disorders.

We defined the target population as people with mental disorders treated with antidepressants. We excluded studies of healthy volunteers. Eligible interventions included antidepressants when compared with placebo or other psychotropic medications. Eligible outcomes included change in QT interval, prolongation of QT interval as reported in the studies including QT interval corrected to RR interval (QTc) ≥ 450 ms, QTc ≥ 480 ms, QTc ≥ 500 ms [24], torsades de pointes ventricular tachycardia, and sudden death.

We conducted a comprehensive search in PubMed, EMBASE, the Cochrane Library, www.clinicaltrials.gov, PharmaPendium (www.pharmapendium.com), and https://crediblemeds.org/ up to January 2018 to find systematic reviews, published and unpublished RCTs, and nationally representative controlled observational studies that reported adjusted effect estimates [22, 23]. All of the authors determined the studies’ eligibility. All citations found during the searches are stored in a reference database.

The data were extracted from the Clinical Trials Transformation Initiative (CTTI) (https://www.ctti-clinicaltrials.org/aact-database), checked for quality, and stored in the HPCC platform (High-Performance Computing Cluster, https://hpccsystems.com/).

We performed direct frequentist meta-analyses of aggregate data when definitions of the active and control intervention and patient outcomes were deemed similar for pooling [25]. We used random effects models to address inevitable differences in patient characteristics across primary RCTs. For each abstracted hypothesis, we calculated absolute risk difference and relative risk with 95% CI. We calculated number needed to treat and number of attributable events per 1000 treated with 95% CI based on statistically significant differences in absolute risks of the outcomes. We examined consistency in results across studies with χ2 tests and I2 statistics and concluded statistically significant heterogeneity if I2 was > 50% [22]. Statistically significant heterogeneity did not preclude statistical pooling [25]. However, we planned exploring heterogeneity with a priori defined patient characteristics, drug doses, and study quality if this information was available in the studies [25].

We used consensus method guidelines for systematic review and meta-analyses that do not recommend conducting post hoc analyses of statistical power [26-29]. Instead, we downgraded our confidence in true treatment effects based on calculated optimal information size as the number of patients required for an adequately powered individual trial [30]. Since power is more closely related to number of events than to sample size, we concluded imprecision in treatment effects if fewer than 250 patients experienced the event [30].

We used Statistics/Data Analysis, Stata software (StataCorp LP, College Station, Texas). Statistical significance was evaluated at a 95% confidence level.

We evaluated the quality of systematic reviews using the Assessment of Multiple Systematic Reviews (AMSTAR) [31]. For primary RCTs, we used the Cochrane risk-of-bias tool on a three-point scale: high bias, low bias, and unclear [32, 33]. A low risk of bias was assumed when RCTs met all the risk-of-bias criteria and a high risk of bias if one or more risk-of-bias criteria were not met. An unknown risk of bias was assigned for the studies with poorly reported risk-of-bias criteria. We assigned a high risk of bias to all observational studies.

The authors assigned the quality-of-evidence ratings as high, moderate, low, or very low, according to risk of bias in the body of evidence, directness of comparisons, precision and consistency in treatment effects, and evidence of reporting bias, using Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology [34].

A high quality of evidence was assigned to well-designed RCTs with consistent findings. The quality of evidence was downgraded to moderate if at least one of four quality-of-evidence criteria was not met; for example, moderate quality of evidence was assigned if there was a high risk of bias in the body of evidence or if the results were not consistent or precise. The quality of evidence was downgraded to low if two or more criteria were not met. We concluded a high risk of bias in the body of evidence if at least one RCT had high risk of bias. We downgraded the quality of evidence when we suspected high risk of publication bias due to unavailability of the results in clinicaltrials.gov or in journal articles.

A low quality of evidence was assigned to nonrandomized studies, but the rating was upgraded if there was a strong or dose-response association [35]. Evidence was defined as insufficient when no studies provided valid information about treatment effects. This approach was applied regardless of whether the results were statistically significant.

Results

Our comprehensive search in PubMed, EMBASE, the Cochrane Library, and clinicaltrials.gov up to January 2018 identified 2 individual patient data and one aggregate data meta-analysis, 8 reviews, 47 publications of RCTs, 3 unpublished RCTs, 14 observational studies, and 4 publications of case reports [7, 36–114]. We present the results from antidepressants with known and probable risk of QT prolongation.

Citalopram and escitalopram

Low-quality evidence suggests that escitalopram, when compared with placebo, increases the risk of QT prolongation (> 30 ms) in adults with mental disorders (Table I). Escitalopram prolongs the QT interval in adults hospitalized for acute coronary syndrome (Table I). Escitalopram does not increase the risk of clinically important prolongation of QT > 400 ms in RCTs (Table I). However, observational data suggest that escitalopram is associated with higher risk of clinically important prolongation of QT (≥ 450 ms) when compared with no antidepressant use in adults undergoing dialysis for chronic kidney disease (Table I). Observational studies also suggest a positive dose-response association between the larger doses of escitalopram and longer QT interval (Table I). Post-marketing surveillance detected 19 cases of prolonged QT interval and 76 cases of torsades de pointes tachycardia in people treated with escitalopram, among other medications, for various mental disorders (Supplementary Table SI).

Table I

GRADE summary of findings: effect of escitalopram on QT interval in adults with mental disorders

Outcome	Risk with intervention per 1000	Risk with comparator per 1000	Relative measure of association	No. of participants (studies)	Quality (GRADE)	Comments†
QTcF > 30 ms, 8 weeks	48	33Attributable events per 1000 treated 15 (2;27)	RR = 1.45 (1.05–2.00)NNT = 68 (36–474)	3689 (14 RCTs)[48, 52–55, 57, 58, 62–67, 72–74, 91]	Low	Favors placebo
QTcF > 30 ms, last assessment	55	35Attributable events per 1000 treated 20 (7;33)	RR = 1.56 (1.15–2.12)NNT = 50 (30;153)	3689 (14 RCTs) [48, 52–55, 57, 58, 62–67, 72–74, 91]	Low	Favors placebo
QTcF > 30 ms, last assessment, elderly	56	42	RR = 1.35 (0.67–2.73)	612 (2 RCTs) [48, 52–55, 57, 58, 62–67, 72–74, 91]	Low	No difference
QTcF > 60 ms, last assessment, 12 weeks	2	0	RR = 0.99 (0.04–24.24)	670 (3 RCTs) [48, 52–55, 57, 58, 62–67, 72–74, 91]	Low	No difference
QTcF > 60 ms, last assessment, elderly	7	0	RR = 5.17 (0.25–107.15)	612 (2 RCTs) [48, 52–55, 57, 58, 62–67, 72–74, 91]	Low	No difference
QTcF > 450 ms, end, 8 weeks	17	11	RR = 1.50 (0.86–2.62)	3689 (14 RCTs) [48, 52–55, 57, 58, 62–67, 72–74, 91]	Low	No difference
QTcF > 450 ms, last assessment, elderly	56	26	RR = 2.20 (0.96–5.01)	612 (2 RCTs) [48, 52–55, 57, 58, 62–67, 72–74, 91]	Low	No difference
QTcF > 480 ms, last assessment, 8 weeks	2	0	RR = 6.57 (0.34–127.10)	3689 (14 RCTs) [48, 52–55, 57, 58, 62–67, 72–74, 91]	Low	No difference
QTcF > 480 ms, last assessment, elderly	3	0	RR = 3.10 (0.13–75.78)	612 (2 RCTs) [48, 52–55, 57, 58, 62–67, 72–74, 91]	Low	No difference
QTcF > 500 ms, last assessment, 8 weeks	1	0	RR = 4.69 (0.23–97.68)	3689 (14 RCTs) [48, 52–55, 57, 58, 62–67, 72–74, 91]	Low	No difference
QTcF > 500 ms, last assessment, elderly	3	0	RR = 3.10 (0.13–75.78)	612 (2 RCTs) [48, 52–55, 57, 58, 62–67, 72–74, 91]	Low	No difference
QTc (ms), adults hospitalized for acute coronary syndrome	NR	NR	MD = 0.33 (0.07–0.58)SMD = 8.00 (1.79–14.21)	239 (1 RCT) [81]	Very low	Favors placebo
QTc > 450 ms, adults hospitalized for acute coronary syndrome	17	0	RR = 4.96 (0.24–102.21)	239 (1 RCT) [81]	Very low	No difference
QTc ≥ 450 ms, adults with predialysis CKD	NR	NR	Adjusted OR = 2.20 (1.10–4.20)	3252 (1 observational study) [112]	Low	Favors control (no antidepressants)
QT prolongation, adults with an ECG recorded after prescription of antidepressant or methadone	NR	NR	Adjusted MD = 0.58 (0.29–0.87)	38397 (1 observational study) [86]	Low	Favors control (no antidepressants)
QT prolongation, escitalopram, 10 mg vs. escitalopram, 5 mg	NR	NR	Adjusted MD = 11.00 (2.18–19.82)	38397 (1 observational study) [86]	Low	Favors lower dose
QT prolongation, escitalopram, 20 mg vs. escitalopram, 10 mg	NR	NR	Adjusted MD = 4.70 (1.56–7.84)	38397 (1 observational study) [86]	Low	Favors lower dose

Population: adults with mental disorders; Settings: any; Intervention: escitalopram at doses between 5 and 20 mg/day; Comparator: placebo or no active drug. Boldface indicates statistically significant differences at 95% CI. †We concluded that there is no difference in outcomes between active and control interventions based on p-value > 0.05 and inability to reject null hypotheses but without post hoc analysis of the statistical power to detect true differences. GRADE – Grading of Recommendations Assessment, Development and Evaluation, NNT – number needed to treat, NNTp – number needed to treat to prevent an outcome in one patient (when the outcome is more probable with control intervention), NR – not reported, OR – odds ratio, RCT – randomized controlled trial, RR – relative risk. Between studies differences in continuous outcomes: MD – mean difference in absolute values of continuous outcomes between intervention and comparator, SMD – standardized mean difference between intervention and comparator where the magnitude of the effect is defined as small (SMD, 0–0.5 standard deviations), moderate (SMD, 0.5–0.8 standard deviations), and large (SMD > 0.8 standard deviations).

Moderate-quality evidence suggests that citalopram prolongs the QT interval compared with placebo in adults with mental disorders (Table II). Citalopram does not increase the risk of clinically important prolongation of QT > 400 ms in RCTs (Table II). In contrast with RCTs, observational data suggest that citalopram is associated with higher risk of clinically important prolongation of QT (≥ 450 ms) compared with no antidepressant use in adults undergoing dialysis for chronic kidney disease (Table II). Observational studies also suggest a positive dose-response association between the larger doses of citalopram and longer QT intervals (Supplementary Table SII). Much less expected is the evidence from observational studies that a citalopram dose reduction to < 40 mg following the FDA warning is associated with a higher risk of all-cause mortality, arrhythmias, and CVD-related hospitalizations (Supplementary Table SII). Post-marketing surveillance detected 36 cases of prolonged QT intervals and 196 cases of torsades de pointes tachycardia in people treated with citalopram, among other medications, for various mental disorders (Supplementary Table SI).

Table II

GRADE Summary of findings: effect of citalopram on QT interval in adults with mental disorders

Outcome	Risk with intervention per 1000	Risk with comparator per 1000	Relative measure of association	No. of participants (studies)	Quality (GRADE)	Comments†
QTc > 30 ms, 3 weeks	74	11	RR = 6.85 (0.86–54.59)	186 (1 RCT) [96]	Very low	No difference
QTc > 60 ms	33	0	RR = 3.19 (0.14–75.49)	62 (1 RCT) [107]	Very low	No difference
QTc > 450 ms for men or > 470 ms for women, 3 weeks	32	11	RR = 2.94 (0.31–27.71)	186 (1 RCT) [96]	Very low	No difference
QTc > 500 ms	0	0	RR inestimable	62 (1 RCT) [107]	Very low	No difference
QTC [ms]	NR	NR	MD = 6.56 (0.07–13.04)	892 (6 RCTs) [7, 71, 94, 96, 107]	Moderate	Favors control
QTc ≥ 450 ms, adults with predialysis CKD	NR	NR	Adjusted OR = 1.80 (1.00–3.10)	3252 (1 observational study) [112]	Low	Favors control
QT prolongation	NR	NR	Adjusted MD = 0.10 (0.02–0.18)	38397 (1 observational study) [86]	Low	Favors control
QT prolongation	NR	NR	Adjusted OR = 4.38 (1.45–13.30)	6790 (1 observational study) [87]	Low	Favors control
Sudden death with cardiovascular disease	NR	NR	Adjusted OR = 1.81 (0.81–4.03)	4040 (1 observational study) [76]	Low	No difference
Sudden death without cardiovascular disease	NR	NR	Adjusted OR = 1.70 (0.50–5.99)	4040 (1 observational study) [76]	Low	No difference

Population: adults with mental disorders, Settings: any, Intervention: citalopram, any dose, Comparator: placebo or no active drug. Boldface indicates statistically significant differences at 95% CI. †We concluded that there is no difference in outcomes between active and control interventions based on p-value > 0.05 and inability to reject null hypotheses but without post hoc analysis of the statistical power to detect true differences. GRADE – Grading of Recommendations Assessment, Development and Evaluation, NR – not reported, OR – odds ratio, RCT – randomized controlled trial, RR – relative risk. Between studies differences in continuous outcomes: MD – mean difference in absolute values of continuous outcomes between intervention and comparator, SMD – standardized mean difference between intervention and comparator where the magnitude of the effect is defined as small (SMD, 0–0.5 standard deviations), moderate (SMD, 0.5–0.8 standard deviations), and large (SMD > 0.8 standard deviations).

Systematic reviews of published case reports found only a few cases of QT prolongation and torsades de pointes tachycardia in adults taking citalopram in doses between 20 and 60 mg/day [83, 88, 106]. The authors concluded that reporting of citalopram-induced adverse effects was incomplete and monitoring of the safety outcomes for citalopram and other antidepressants was inadequate [83, 88, 106].

We found no pediatric RCTs or high-quality observational studies that examined the effects from escitalopram or citalopram on QT interval.

Desvenlafaxine and venlafaxine

Low-quality evidence from RCTs suggests that desvenlafaxine does not increase the QT interval in adults with various mental disorders (Table III). Post-marketing surveillance detected two cases of torsades de pointes tachycardia in adult women treated with desvenlafaxine among other medications (Supplementary Table SI). Post-approval studies conducted by the pharmaceutical company suggest an association between desvenlafaxine overdose and QT prolongation in adults with concurrent alcohol and drug consumption [115]. The evidence regarding the effects of desvenlafaxine on the QT interval in children is insufficient [113].

Table III

GRADE summary of findings: effect of desvenlafaxine on QT interval in adults with mental disorders

Outcome		Risk with intervention per 1000	Risk with comparator per 1000	Relative measure of association	No. of participants (studies)	Quality (GRADE)	Comments†
Adults with major depressive disorder:
	QT prolonged, desvenlafaxine 50 mg/day	0	7	RR = 0.17 (0.01–4.07)	427 (1 RCT) [110]	Very low	No difference
	QTcF change, desvenlafaxine 200 mg/day	NR	NR	MD = 1.50 (–0.88 – 3.88)	2476 (4 RCTs) [115]	Low	No difference
	QTcF change, desvenlafaxine 600 mg/day	NR	NR	MD = –2.43 (–4.90 – 0.04)	2476 (4 RCTs) [115]	Low	No difference
Postmenopausal women with moderate to severe vasomotor symptoms:
	QT prolonged	1	0	RR = 2.96 (0.12–72.59)	2118 (1 RCT) [90]	Very low	No difference
Adults with painful diabetic peripheral neuropathy:
Ventricular tachycardia		3	0	RR = 0.85 (0.03–20.57)	412 (2 RCTs) [93]	Low	No difference

Population: adults with mental disorders, Settings: any, Intervention: desvenlafaxine, Comparator: placebo or no active drug. †We concluded that there is no difference in outcomes between active and control interventions based on p-value > 0.05 and inability to reject null hypotheses but without post hoc analysis of the statistical power to detect true differences.

Low-quality evidence from observational studies suggests that venlafaxine is associated with increased risk of sudden death in adults with cardiovascular diseases (Table IV). Observational studies suggest no association between venlafaxine and QT prolongation in adults without cardiovascular disease, in adults undergoing hemodialysis for chronic kidney failure, or in adults with deliberate overdosing of this drug (Table IV). Published case reports also suggest the association between therapeutic doses of venlafaxine and QT prolongation in older adults taking concomitant drugs [37, 105, 116]. Post-approval studies conducted by the pharmaceutical company suggest the association between venlafaxine overdose and QT prolongation in adults with concurrent alcohol and drug consumption [117]. Post-marketing surveillance detected 7 cases of prolonged QT interval and 58 cases of torsades de pointes tachycardia in people treated with venlafaxine among other medications (Supplementary Table SI). The evidence regarding the effects of venlafaxine on the QT interval in children is insufficient.

Table IV

GRADE summary of findings: effect of venlafaxine on QT interval in adults with mental disorders

Outcome	Risk with intervention per 1000	Risk with comparator per 1000	Relative measure of association	No. of participants (studies)	Quality (GRADE)	Comments†
QT change	NR	NR	Adjusted MD = 0.01 (–0.01 – 0.03)	38397 (1 observational study) [86]	Low	No difference
Sudden death with cardiovascular disease	NR	NR	Adjusted OR = 3.73 (1.33–10.45)	4040 (1 observational study) [76]	Low	Favors control, drugs without known effects on QT
Sudden death without cardiovascular disease	NR	NR	Adjusted OR = 0.67 (0.07–6.19)	4040(1 observational study) [76]	Low	No difference
QRS > 100 ms, venlafaxine deliberate self-poisoning (DSP) vs. tricyclic antidepressant overdose	NR	NR	Adjusted OR = 0.60 (0.20–1.20)	538 (1 observational study) [56]	Very low	No difference
QTc ≥ 450 ms, adults with predialysis chronic kidney disease	NR	NR	Adjusted OR = 1.10 (0.50–2.70)	3252 (1 observational study) [112]	Low	No difference

Population: adults with mental disorders; Settings: any; Intervention: venlafaxine; Comparator: placebo or no active drug. Boldface indicates statistically significant differences at 95% CI. †We concluded that there is no difference in outcomes between active and control interventions based on p-value > 0.05 and inability to reject null hypotheses but without post hoc analysis of the statistical power to detect true differences.

Fluoxetine

Very low-quality evidence suggests that fluoxetine does not cause QT prolongation when compared with placebo in adults with mental disorders (Supplementary Table SIII). Low-quality evidence suggests that fluoxetine is not associated with increased risk of cardiac death in adults with or without concurrent cardiovascular disease (Supplementary Table SIII).

Paroxetine

Moderate-quality evidence suggests that paroxetine does not increase the QT interval in adults with mental disorders (Supplementary Table SIV). Low-quality evidence suggests that paroxetine administration is not associated with higher risk of sudden death in adults with mental disorders, with or without comorbid cardiovascular disease (Supplementary Table SIV).

Very low-quality evidence from a single RCT suggests that paroxetine decreases the length of the QT interval when compared with placebo or with imipramine in children with depression or obsessive-compulsive disorder (Supplementary Table SV).

Bupropion

Low-quality evidence suggests that bupropion, 75–300 mg/day, is not associated with QT prolongation in adults with mental disorders (Supplementary Table SVI).

Duloxetine

Very low-quality evidence suggests that duloxetine, 80–120 mg/day, does not prolong QT in adults with mental disorders (Supplementary Table SVII).

Fluvoxamine

Very low-quality evidence from a small single RCT suggests that fluvoxamine decreases the QT interval when compared with placebo in adults with mental disorders (Supplementary Table SVIII).

Sertraline

Low-quality evidence suggests that sertraline prolongs the QT interval but is not associated with higher risk of sudden death (Supplementary Table SVIII).

Trazodone

Low-quality evidence suggests that trazodone is not associated with clinically important QT prolongation in adults with mental disorders and comorbid renal failure undergoing dialysis (Supplementary Table SVIII).

Amitriptyline

Very low-quality evidence from a small single RCT suggests that amitriptyline does not cause QT prolongation in children with functional gastrointestinal disorders (Supplementary Table SIX). Very low-quality evidence from a single RCT suggests that amitriptyline cream does not cause QT prolongation in adults with diabetic peripheral neuropathy (Supplementary Table SIX). Low-quality evidence from observational studies suggests that amitriptyline is associated with longer QT interval in adults with mental disorders in a dose-dependent manner (Supplementary Table SIX). Observational studies suggest no association between amitriptyline with clinically important QT prolongation or sudden death (Supplementary Table SIX).

Imipramine

Very low-quality evidence from a small single RCT suggests that imipramine increases the risk of clinically important QT prolongation compared with placebo in children with depression or obsessive-compulsive disorder (Supplementary Table SX).

Clomipramine

Low-quality evidence suggests that clomipramine is not associated with higher risk of sudden death in adults with mental disorders, with or without cardiovascular disease (Supplementary Table SXI).

Dosulepin

Low-quality evidence suggests that dosulepin is not associated with higher risk of sudden death in adults with mental disorders, with or without cardiovascular disease (Supplementary Table SXI).

Mirtazapine

Low-quality evidence suggests that mirtazapine is not associated with QT prolongation (Supplementary Table SXI).

Nortriptyline

Low-quality evidence suggests that nortriptyline is not associated with QT prolongation (Supplementary Table SXI). However, nortriptyline is associated with higher risk of sudden death (Supplementary Table SXI).

Drug class effect

The evidence from large observational studies suggests that antidepressants with conditional possible or known risk of torsades de pointes are associated with higher odds of all-cause mortality after adjustment for other confounding factors in adults with mental disorders. Tricyclic antidepressants are not associated with higher risk of sudden death except at higher doses. Selective serotonin reuptake inhibitors are associated with higher risk of sudden death specifically in adults with comorbid cardiovascular disease. The adjusted odds of sudden death are the highest after high doses of selective serotonin reuptake inhibitors. However, the adjusted odds of clinically important QT prolongation are lower after overdose of selective serotonin reuptake inhibitors when compared with overdose of tricyclic antidepressants (Table V).

Table V

GRADE summary of findings: association between antidepressant classes and QT interval and sudden death in adults with mental disorders

Outcome		Risk with intervention per 1000	Risk with comparator per 1000	Relative measure of association	No. of participants (studies)	Quality (GRADE)	Comments†
Use of antidepressants, any class vs. no antidepressants:
	All-cause mortality, 5 years	NR	NR	Adjusted RR = 1.62 (1.60–1.63)	1,716,552 (1 observational study) [108]	Low	Favors control
Antidepressants without known risk for torsades de pointes:
	All-cause mortality, 5 years	NR	NR	Adjusted RR = 0.99 (0.94–1.05)	1,716,552 (1 observational study) [108]	Low	No difference
Antidepressants with conditional risk for torsades de pointes:
	All-cause mortality, 5 years	NR	NR	Adjusted RR = 1.25 (1.22–1.28)	1,716,552 (1 observational study) [108]	Low	Favors control
Antidepressants with possible risk for torsades de pointes:
	All-cause mortality, 5 years	NR	NR	Adjusted RR = 1.63 (1.61–1.67)	1,716,552 (1 observational study) [108]	Low	Favors control
Antidepressants with known risk for torsades de pointes:
	All-cause mortality, 5 years	NR	NR	Adjusted RR = 1.53 (1.51–1.56)	1,716,552 (1 observational study) [108]	Low	Favors control
Tricyclic antidepressants:
	Sudden death	NR	NR	Adjusted OR for other drugs with known QT effect = 1.28 (0.84–1.96)	4040 (1 observational study) [76]	Low	No difference
	Sudden death	NR	NR	Adjusted OR = 1.41 (0.93–2.13)	4040 (1 observational study) [76]	Low	No difference
	Sudden death with cardiovascular disease	NR	NR	Adjusted OR = 1.34 (0.79–2.28)	4040 (1 observational study) [76]	Low	No difference
	Sudden death without cardiovascular disease	NR	NR	Adjusted OR = 1.61 (0.83–3.13)	4040 (1 observational study) [76]	Low	No difference
	Sudden death, high-dose antidepressants	NR	NR	Adjusted OR = 2.11 (1.10–4.22)	4040 (1 observational study) [76]	Low	Favors control
	Sudden death, moderate-dose antidepressants	NR	NR	Adjusted OR = 0.85 (0.42–1.73)	4040 (1 observational study) [76]	Low	No difference
	Sudden death, low-dose antidepressants	NR	NR	Adjusted OR = 1.60 (0.72–3.56)	4040 (1 observational study) [76]	Low	No difference
Selective serotonin reuptake inhibitors:
	QTc interval prolongation	NR	NR	Adjusted OR = 1.10 (0.50–2.00)	794 (1 observational study) [104]	Very low	No difference
	Sudden death	NR	NR	Adjusted OR for other drugs with known QT effect = 1.78 (1.24–2.55)	4040 (1 observational study) [76]	Low	Favors control
	Sudden death	NR	NR	Adjusted OR = 1.89 (1.34–2.69)	4040 (1 observational study) [76]	Low	Favors control
	Sudden death with cardiovascular disease	NR	NR	Adjusted OR = 2.04 (1.33–3.13)	4040 (1 observational study) [76]	Low	Favors control
	Sudden death without cardiovascular disease	NR	NR	Adjusted OR = 1.63 (0.86–3.10)	4040 (1 observational study) [76]	Low	No difference
	Sudden death, high antidepressant dose	NR	NR	Adjusted OR = 2.78 (1.24–6.24)	4040 (1 observational study) [76]	Low	Favors control
	Sudden death, moderate antidepressant dose	NR	NR	Adjusted OR = 1.55 (0.96–2.49)	4040 (1 observational study) [76]	Low	No difference
	Sudden death, low antidepressant dose	NR	NR	Adjusted OR = 1.83 (0.70–4.78)	4040 (1 observational study) [76]	Low	No difference
Selective serotonin reuptake inhibitor overdose vs. tricyclic antidepressant overdose:
	QRS > 100	NR	NR	Adjusted OR = 0.20 (0.10–0.40)	538 (1 observational study) [56]	Very low	Favors selective serotonin reuptake inhibitors

Population: adults with mental disorders; Settings: any; Intervention: tricyclic antidepressants or selective serotonin reuptake inhibitors; comparator: no active drug class. Boldface indicates statistically significant differences at 95% CI. †We concluded that there is no difference in outcomes between active and control interventions based on p-value > 0.05 and inability to reject null hypotheses but without post hoc analysis of the statistical power to detect true differences.

Discussion

Our review of clinical trials found mostly low quality of evidence that escitalopram and citalopram are associated with higher risk of QT prolongation in adults, and imipramine increases the risk of QT prolongation in children. Observational studies demonstrate that selective serotonin reuptake inhibitors (specifically venlafaxine and nortriptyline) and high doses of tricyclic antidepressants are associated with higher risk of sudden death specifically in adults with comorbid cardiovascular disease (Figure 1).

Figure 1

Increased risk of patient outcomes in association with antidepressants in adults with mental disorders

Post-marketing surveillance suggests that serotonin reuptake inhibitors and tricyclic antidepressants are associated with torsades de pointes (total of 662 cases) and QT prolongation (168 cases) in people taking antidepressants among other drugs. The direct evidence regarding comparative safety of antidepressants is insufficient.

We downgraded the quality of evidence due to the high risk of bias and small number of events in the RCTs. Most clinical studies did not have statistical power to detect higher risk of ventricular tachycardia. We further downgraded the quality of evidence due to reporting bias, because only a very small proportion of primary studies that examined benefits of antidepressants also examined drug-induced QT prolongation. Post-marketing surveillance is the major safety source, but case reporting depends on clinician opinion regarding the association between ventricular tachycardia and administration of antidepressants.

Differences in the definition of QT prolongation and correction methods precluded indirect analysis of comparative safety between antidepressants. Primary studies did not evaluate factors that increase susceptibility to QT prolongation and torsades de pointes, including electrolyte disorders, chronic inflammation, baseline cardiovascular disease, comorbidities, or concomitant medications [118-120]. The evidence is lacking regarding interaction between antidepressant use with clinical and genetic risk factors for torsade de pointes [121]. Evolving research suggests genetic predisposition associated with QT prolongation and torsades de pointes [121-124]. Stratification algorithms predicting the risk of acquired QT prolongation have been proposed [122, 123, 125]. Larger sample size and longitudinal modeling with multivariate analyses are required to develop valid prediction pharmacogenomic algorithms of acquired QT prolongation for individual patients taking antidepressants and other proarrhythmic drugs [121, 124]. Genetic mutations in hERG (human-ether-a-go-go-related gene) potassium channel kinetic abnormalities and other genes responsible for cardiac function and repolarization may contribute to complex mechanisms of antidepressant-induced QT prolongation [126-128]. Valid prediction algorithms should also address factors affecting drug clearance, thereby resulting in a relative overdose and an associated increased risk of proarrhythmia [129].

Available guidelines emphasize the importance of individual assessment of probable benefits and harms, including QT prolongation, when selecting antidepressants for people with high risk of QT prolongation and cardiac arrhythmias [130, 131]. Guidelines recommend standardized assessment of QT interval in patients who need antidepressants and are at high risk of QT prolongation and cardiac arrhythmias [131-133]. Guidelines recommend therapeutic drug monitoring of antidepressants to prevent overdose and consequential adverse effects, including QT prolongation [134]. Guidelines do not address optimal risk mitigation that may include drug dose reduction or switching as well as individualized assessment and management of comorbidities and concomitant drugs [135]. For instance, the evidence suggests that a simple adjustment of escitalopram or citalopram administration is not enough to ensure better patient outcomes [92, 109, 135]. Drug labels recommend against administration of antidepressants in combination with other drugs that are known to prolong QT interval and in people with bradycardia, hypokalemia or hypomagnesemia, or congenital prolongation of the QT interval [9, 11–21, 115, 117]. Despite these recommendations, the prevalence of polypharmacy with multiple proarrhythmic drugs is high [2, 4, 136].

Our review has limitations. Our analyses are based only on the available evidence. Despite our efforts to include all published and unpublished clinical trials, observational studies, and post-marketing data, we do not know how many unregistered and unpublished studies have been conducted. We detected a substantial reporting bias but did not contact principal investigators of all studies of antidepressants that did not report QT prolongation. Analyzing aggregate data, we could not overcome differences in outcome definitions. Despite all these limitations, we attempted a comprehensive review of all available evidence and consistent appraisal of the quality of evidence with a conservative GRADE approach.

Our review has implications for clinical practice. Baseline risk of cardiac arrhythmias should be assessed in all patients treated with antidepressants [137]. High-risk patients should be monitored for clinical symptoms indicating the occurrence of cardiac arrhythmias (e.g., dizziness, palpitations, or syncope) and should receive ECG monitoring for the prolongation of QT interval [138, 139]. Multidisciplinary coordinated care should be implemented to avoid polypharmacy with antidepressants and other proarrhythmic drugs [137, 140].

Our review has policy implications. Prescriber compliance with the licensed drug use should be routinely evaluated using electronic health records [141-143]. Proactive post-marketing pharmacovigilance applications should be implemented to decrease the risk of drug-induced QT prolongation and cardiac arrhythmias [144-149].

Our review has research implications. Future research should examine long-term comparative safety of antidepressant drugs in patients of different ages as well as those with baseline cardiovascular risk, multiple comorbidities, and concomitant drug use [118]. Future research is needed to develop a valid risk stratification algorithm that includes pharmacokinetic factors (e.g., old age and renal impairment) and pharmacodynamic factors that modulate repolarization (e.g., potassium level, genetic mutations, hypertrophy and heart failure, and concomitant use of proarrhythmic drugs) [118, 122, 125].

In conclusion, in adults with indications for antidepressants and elevated risk of QT prolongation, in order to avoid prolongation of the QT interval and reduce the risk of ventricular tachycardia, clinicians should not recommend citalopram or escitalopram,

In adults with indications for antidepressants and low baseline risk of QT prolongation, clinicians may recommend licensed doses of antidepressants in addition to close monitoring of QT interval and clinical symptoms indicating occurrence of cardiac arrhythmias (e.g., dizziness, palpitations, or syncope) to avoid prolongation of QT interval and reduce the risk of ventricular tachycardia.