Comparative Cardiac Safety of Selective Serotonin Reuptake Inhibitors among Individuals Receiving Maintenance Hemodialysis : Journal of the American Society of Nephrology

Journal Logo

Clinical Epidemiology

Comparative Cardiac Safety of Selective Serotonin Reuptake Inhibitors among Individuals Receiving Maintenance Hemodialysis

Assimon, Magdalene M.1; Brookhart, M. Alan2; Flythe, Jennifer E.1,3

Author Information
Journal of the American Society of Nephrology 30(4):p 611-623, April 2019. | DOI: 10.1681/ASN.2018101032

Abstract

Depression affects 25%–40% of the hemodialysis population1 and associates with a range of adverse health outcomes, including treatment nonadherence, lower quality of life, higher hospitalization rates, and increased mortality.2 With the inclusion of the Clinical Depression Screening and Follow-Up Reporting Measure in the 2018 ESRD Quality Incentive Program,3 the diagnosis and treatment of depression among United States hemodialysis patients may increase in the coming years. Selective serotonin reuptake inhibitors (SSRIs) are recommended as first-line agents for the pharmacologic management of depressive disorders,4 and in 2015, >20% of United States hemodialysis patients filled a prescription for an SSRI.5 Even though clinical trials have demonstrated that SSRIs are efficacious antidepressants in the hemodialysis population, their small sample sizes and limited follow-up time precluded adequate safety assessments.615

Electrophysiologic data indicate that all six SSRIs have QT-prolonging capabilities.1618 However, citalopram and escitalopram prolong the QT interval to the greatest extent at therapeutic doses,1618 and currently carry pharmaceutic regulatory agency–issued warnings related to their proarrhythmic potential.1924 Although these cardiac safety warnings highlight patient populations at increased risk for QT prolongation and subsequent torsades de pointes (e.g., individuals with hypokalemia or hypomagnesemia), there is no specific mention of hemodialysis patients. Individuals receiving maintenance hemodialysis may be particularly susceptible to the lethal cardiac consequences of citalopram- and escitalopram-induced QT prolongation due to their substantial cardiovascular disease burden, recurrent exposure to electrolyte shifts during dialysis treatments, and extensive use of medications that can enhance the proarrhythmic effects of citalopram and escitalopram via pharmacokinetic and/or pharmacodynamic drug interactions.

In the proarrhythmic ESRD environment, the differential QT-prolonging nature of SSRIs may render certain SSRIs safer than others. We undertook this study to investigate the comparative cardiac safety of SSRIs in a cohort of >65,000 United States hemodialysis patients. We hypothesized that individuals initiating higher (citalopram or escitalopram) versus lower (fluoxetine, fluvoxamine, paroxetine or sertraline) QT-prolonging–potential SSRIs would have a higher risk of sudden cardiac death.

Methods

This study was approved by the University of North Carolina at Chapel Hill Institutional Review Board (#17–0011). A waiver of consent was granted due to the study’s large size, data anonymity, and retrospective nature.

Data Source

The data source for this study was the US Renal Data System (USRDS) database. The USRDS is a national ESRD surveillance system that collects, analyzes, and distributes information on individuals with ESRD in the United States. The USRDS database includes the Medical Evidence and ESRD Death Notification Forms as well as Medicare standard analytic files, including the Medicare enrollment database and final action administrative claims (Medicare Parts A, B, and D).

Study Design and Population

We conducted a retrospective cohort study using an active comparator new-user design25 (Figure 1) to investigate the association between the initiation of higher (citalopram or escitalopram) versus lower (fluoxetine, fluvoxamine, paroxetine, or sertraline) QT-prolonging–potential SSRIs and the 1-year risk of sudden cardiac death among individuals receiving maintenance hemodialysis. First, we identified hemodialysis patients with Medicare coverage (Parts A, B, and D) who newly initiated SSRI therapy from January 1, 2007 to December 30, 2014 after a 180-day washout period free of documented SSRI use. To be included in the study, SSRI new-users had to receive in-center hemodialysis during the 180 days before SSRI initiation (i.e., baseline period) and also have continuous Medicare Part A, B, and D coverage during this period. We then applied the following exclusion criteria: (1) age<18 years at the start of the baseline period, (2) dialysis vintage ≤90 days at the start of the baseline period, (3) presence of an implantable automatic cardiac defibrillator, (4) receipt of hospice care during the baseline period, and (5) missing demographic data. Thus, the study cohort consisted of prevalent, center-based hemodialysis patients who were SSRI new-users.

fig1
Figure 1.:
Study design. Initiators of higher and lower QT-prolonging–potential SSRIs were defined as hemodialysis patients who had no record of an SSRI prescription in the previous 180 days (i.e., the SSRI washout period). Higher QT-prolonging–potential SSRIs included citalopram and escitalopram. Lower QT-prolonging–potential SSRIs included fluoxetine, fluvoxamine, paroxetine, and sertraline. The index date was defined as the date of SSRI initiation. Baseline covariates were identified in the 180-day period before the index date. Study follow-up began immediately after the index date. Rx, prescription.

Study Exposure, Outcomes, and Baseline Covariates

We used Medicare Part D prescription claims to identify SSRI new-users. The index date was the date of the first SSRI prescription after the 180-day washout period. We used published QT Drug Lists from the CredibleMeds website (https://crediblemeds.org/; CredibleMeds, Oro Valley, AZ) to identify the QT-prolonging potential of SSRIs marketed in the United States during the study period.26 Funded by the US Food and Drug Administration (FDA), research grants, and charitable donations, CredibleMeds is a reliable and up-to-date clinical resource that contains information about medications that can cause QT prolongation and/or torsades de pointes.26 On the basis of published medical literature, medication package inserts, and the FDA’s Adverse Event Reporting System, CredibleMeds classifies medications as having a known, conditional, or possible risk of torsades de pointes (corresponding definitions are provided in Table 1).26 Citalopram and escitalopram are classified as having a known risk of torsades de pointes. Fluoxetine, fluvoxamine, paroxetine, and sertraline are classified as having a conditional risk of torsades de pointes. Thus, we considered (1) citalopram and escitalopram new-users as initiators of a higher QT-prolonging–potential SSRI, and (2) fluoxetine, fluvoxamine, paroxetine, and sertraline new-users as initiators of a lower QT-prolonging–potential SSRI.

Table 1. - CredibleMeds definitions for medications with a known, conditional and possible risk of torsades de pointes
CredibleMeds Classification 26 Definition
Known risk of torsades de pointes Drugs that prolong the QT interval and are clearly associated with a known risk of torsades de pointes, even when taken as recommended.
Conditional risk of torsades de pointes Drugs that are associated with torsades de pointes only under certain conditions (e.g., excessive dose, in patients with conditions such as hypokalemia, or when taken with interacting drugs) or drugs that create conditions that facilitate or induce torsades de pointes (e.g., cause an electrolyte disturbance that induces torsades de pointes).
Probable risk of torsades de pointes Drugs that can cause QT prolongation but currently lack evidence for a risk of torsades de pointes when taken as recommended.

The primary outcome of interest was 1-year sudden cardiac death. We defined sudden cardiac death using the established USRDS definition, death due to cardiac arrhythmia or cardiac arrest listed as the primary cause (Supplemental Table 1).27 We obtained dates and causes of death from the ESRD Death Notification Form. We used a 1-year follow-up period to mirror clinical practice guidelines which recommend that patients starting an initial course of antidepressant pharmacotherapy are treated for a minimum of 6–12 weeks to achieve initial depressive symptom remission, followed by an additional 4–9 months of continued antidepressant treatment to prevent relapse (i.e., approximately 1 year of total pharmacotherapy).4

Baseline covariates included potential confounders and variables known to be strong risk factors for the study outcome.28 We identified covariates in the 180 days before the index date using Medicare Part A, B, and D claims. Covariates of interest included patient demographics, comorbid conditions, prescription medication use, and metrics of health care utilization. Comorbid conditions were considered present if an applicable International Classification of Diseases, Ninth Revision discharge code (located in any position) was associated with ≥1 institutional or physician/supplier claim during the 180-day baseline period (Supplemental Table 2). Medication utilization on the last day of the baseline period was determined using Medicare Part D claims and Healthcare Common Procedure Coding System codes (for medications not covered by Medicare Part D, such as intravenous drugs). Non-SSRI QT-prolonging medications of interest included drugs classified as having a known, conditional, and possible risk of torsades de pointes according to CredibleMeds (Supplemental Tables 3 and 4).26 Cytochrome P450 inhibitors of interest included medications that can reduce the hepatic metabolism of SSRIs (Supplemental Table 5).29 Use of a 180-day baseline period enabled us to: (1) maximize cohort generalizability and (2) capture patient characteristics occurring close to study medication initiation that may have influenced SSRI-prescribing decisions30,31 and were highly predictive of the study outcome.32

All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC). We described baseline characteristics across individuals who initiated higher and lower QT-prolonging–potential SSRIs as count (percentage) for categoric variables and as mean±SD for continuous variables. We compared baseline covariate distributions using absolute standardized differences. A standardized difference >10% represents an imbalance between exposure groups.33

We used an on-treatment (i.e., per-protocol) analytic approach to evaluate the association between the initiation of a higher versus lower QT-prolonging–potential SSRI and the 1-year risk of sudden cardiac death. Individuals were followed forward in historical time from the index date to the first occurrence of the study outcome, censoring event, or competing event. Censoring events included: (1) change of dialysis modality to home hemodialysis or peritoneal dialysis; (2) kidney transplantation; (3) recovery of kidney function; (4) loss of Medicare Part A, B, or D coverage; (5) discontinuation of index SSRI therapy; (6) switch to a nonindex SSRI; (7) completion of 1-year of follow-up; and (8) study end (December 31, 2014). We defined the SSRI discontinuation date as the date when the index SSRI was exhausted for >7 days (i.e., grace period) without a subsequent dispensing of the same SSRI (Supplemental Figure 1). We defined the index SSRI switching date as the date of the first prescription fill for a nonindex SSRI during follow-up. Patients were at risk for a switching event only during times of continuous index medication use (Supplemental Figure 1).

In primary analyses, we assessed the SSRI QT-prolonging potential–sudden cardiac death association in the full study cohort using Fine and Gray proportional subdistribution hazards models, treating nonsudden cardiac death as a competing risk.34 In prespecified, exploratory secondary analyses, we evaluated the SSRI QT-prolonging potential–sudden cardiac death association within clinically relevant subgroups using the same analytic approach. Subgroups of interest included individuals with and without known risk factors for drug-induced QT prolongation: advanced age, female sex, cardiac conduction disorder, heart failure, liver disease, and use of other non-SSRI QT-prolonging medications.35,36 Across all analyses, we used inverse probability of treatment (IPT) weighting to control for confounding. Briefly, we calculated the predicted probability (i.e., propensity score) of receiving an SSRI with higher versus lower QT-prolonging potential as a function of baseline covariates using multivariable logistic regression. We generated IPT weights from propensity scores using standard methods and estimated adjusted hazard ratios (HRs) by applying IPT weights in regression models.37,38

We conducted several sensitivity analyses to evaluate the robustness of our primary study findings. First, we evaluated two alternative study outcomes including: (1) a composite outcome of sudden cardiac death or hospitalized ventricular arrythmia and (2) cardiovascular mortality (Supplemental Table 1). In analyses evaluating cardiovascular mortality, noncardiovascular death was treated as a competing risk. Second, we compared citalopram and escitalopram initiators (separately) to initiators of an SSRI with lower QT-prolonging potential using methods analogous to the primary analyses. Third, because exact dates of medication discontinuation are not available in administrative claims data, we repeated primary analyses using longer grace periods of 14 and 30 days to define index SSRI discontinuation. Fourth, because long-term antidepressant therapy may be required in some patients (i.e., those with histories of multiple major depressive episodes or residual depressive symptoms), we repeated primary analyses using all available follow-up time.4 Fifth, we used an intention-to-treat analytic approach (i.e., first-exposure-carried-forward analysis) to evaluate the SSRI QT-prolonging potential–sudden cardiac death association. In these analyses, we did not consider SSRI discontinuation and switching as censoring events. Sixth, to test the specificity of our findings we evaluated the association between the initiation of a higher versus lower QT-prolonging–potential SSRI and nonsudden cardiac death, a negative control outcome that we did not expect to be influenced by SSRI type.39 In these analyses, sudden cardiac death was treated as a competing risk.

Results

Study Cohort Characteristics

Figure 2 displays a flow diagram of study cohort selection. A total of 65,654 individuals receiving maintenance hemodialysis were included in the study: 30,932 (47.1%) initiators of higher QT-prolonging–potential SSRIs and 34,722 (52.9%) initiators of lower QT-prolonging–potential SSRIs. There were 16,288 (32.3%) citalopram, 14,644 (22.3%) escitalopram, 6468 (9.9%) fluoxetine, 44 (0.1%) fluvoxamine, 7011 (10.7%) paroxetine, and 21,199 (32.3%) sertraline initiators in the study cohort. Overall, study patients had an average age of 67.0±17.2 years, 52.8% were women, 35.5% were black, 19.2% were Hispanic, and the most common cause of ESRD was diabetes (50.2%). Depression (33.2%) and anxiety (17.7%) were the most prevalent mental health conditions.

fig2
Figure 2.:
Flow diagram depicting study cohort assembly. (a) To be included in the study, SSRI new-users had to receive in-center hemodialysis during the 180 days before SSRI initiation (i.e., baseline period) and also have continuous Medicare Part A, B, and D coverage during this period. (b) Higher QT-prolonging–potential SSRIs included citalopram and escitalopram. Lower QT-prolonging–potential SSRIs included fluoxetine, fluvoxamine, paroxetine, and sertraline. Rx, prescription.

The propensity score distribution of initiators of higher and lower QT-prolonging–potential SSRIs exhibited substantial overlap (Supplemental Figure 2), indicating that study groups were highly comparable. Table 2 and Supplemental Table 6 show patient baseline characteristics stratified by SSRI QT-prolonging potential (higher versus lower). Before IPT weighting, baseline covariates were generally well balanced between treatment groups (standardized differences ≤10%), with some exceptions (e.g., race other than black or white, year of index SSRI fill, depression, and history of stroke). After IPT weighting, all baseline covariates were well balanced between treatment groups.

Table 2. - Select baseline characteristics of study patients initiating a higher and lower QT-prolonging–potential SSRI
Characteristic Unweighted Weighted
SSRI with Higher QT-Prolonging Potential,
n=30,932 SSRI with Lower QT-Prolonging Potential,
n=34,722 Std Diff a 
(%) SSRI with Higher QT-Prolonging Potential,
n=30,927 SSRI with Lower QT-Prolonging Potential,
n=34,730 Std Diff a 
(%)
Age, yr 67.6±17.2 66.5±17.2 6.2 67.0±17.2 67.0±17.21 0.0
Female 16,512 (53.4%) 18,121 (52.2%) 2.3 16,316 (52.8%) 18,324 (52.8%) 0.0
Race
 Black 11,339 (36.7%) 11,959 (34.4%) 6.3 10,967 (35.5%) 12,320 (35.5%) 0.0
 White 18,434 (59.6%) 21,005 (60.5%) 1.5 18,589 (60.1%) 20,870 (60.1%) 0.0
 Other 1159 (3.7%) 1758 (5.1%) 33.5 1371 (4.4%) 1540 (4.4%) 0.0
Hispanic 5233 (16.9%) 7346 (21.2%) 22.7 5933 (19.2%) 6654 (19.2%) 0.1
Year index SSRI was filled
 2007 3961 (12.8%) 4751 (13.7%) 6.9 4086 (13.2%) 4594 (13.2%) 0.1
 2008 4095 (13.2%) 4388 (12.6%) 4.8 3977 (12.9%) 4468 (12.9%) 0.0
 2009 4061 (13.1%) 4343 (12.5%) 5.0 3948 (12.8%) 4435 (12.8%) 0.1
 2010 4139 (13.4%) 4161 (12.0%) 11.5 3907 (12.6%) 4386 (12.6%) 0.0
 2011 4034 (13.0%) 4061 (11.7%) 11.3 3814 (12.3%) 4287 (12.3%) 0.1
 2012 3611 (11.7%) 4152 (12.0%) 2.5 3670 (11.9%) 4118 (11.9%) 0.1
 2013 3507 (11.3%) 4398 (12.7%) 11.5 3738 (12.1%) 4194 (12.1%) 0.1
 2014 3524 (11.4%) 4468 (12.9%) 12.7 3787 (12.2%) 4246 (12.2%) 0.2
Low-dose index SSRI b 27,935 (90.3%) 31,560 (90.9%) 0.6 28,011 (90.6%) 31,456 (90.6%) 0.0
Cause of ESRD
 Diabetes 15,563 (50.3%) 17,416 (50.2%) 0.3 15,542 (50.3%) 17,458 (50.3%) 0.0
 Hypertension 7941 (25.7%) 8760 (25.2%) 1.8 7848 (25.4%) 8813 (25.4%) 0.0
 Glomerular disease 3359 (10.9%) 3934 (11.3%) 4.4 3442 (11.1%) 3864 (11.1%) 0.1
 Other 4069 (13.2%) 4612 (13.3%) 1.0 4095 (13.2%) 4596 (13.2%) 0.1
Dialysis vintage, yr
 0.7–0.9 5467 (17.7%) 5903 (17.0%) 4.0 5364 (17.3%) 6023 (17.3%) 0.0
 1.0–1.9 5965 (19.3%) 6622 (19.1%) 1.1 5933 (19.2%) 6661 (19.2%) 0.0
 2.0–2.9 4631 (15.0%) 5136 (14.8%) 1.3 4594 (14.9%) 5162 (14.9%) 0.1
 ≥3 14,869 (48.1%) 17,061 (49.1%) 2.2 15,035 (48.6%) 16,883 (48.6%) 0.0
Depression 10,960 (35.4%) 10,851 (31.3%) 12.7 10,298 (33.3%) 11,567 (33.3%) 0.0
Anxiety 5590 (18.1%) 6046 (17.4%) 3.8 5492 (17.8%) 6169 (17.8%) 0.0
Arrythmia 9616 (31.1%) 10,135 (29.2%) 6.4 9327 (30.2%) 10,471 (30.1%) 0.0
Conduction disorder 2434 (7.9%) 2659 (7.7%) 2.9 2403 (7.8%) 2699 (7.8%) 0.0
Dyslipidemia 15,756 (50.9%) 17,566 (50.6%) 0.7 15,723 (50.8%) 17,652 (50.8%) 0.0
Heart failure 14,937 (48.3%) 15,884 (45.7%) 5.5 14,536 (47.0%) 16,321 (47.0%) 0.0
Hypertension 28,289 (91.5%) 31,297 (90.1%) 1.5 28,073 (90.8%) 31,522 (90.8%) 0.0
Ischemic heart disease 15,171 (49.0%) 16,216 (46.7%) 4.9 14,810 (47.9%) 16,624 (47.9%) 0.0
Peripheral arterial disease 11,332 (36.6%) 11,737 (33.8%) 8.2 10,878 (35.2%) 12,216 (35.2%) 0.0
Stroke 8225 (26.6%) 8033 (23.1%) 14.2 7668 (24.8%) 8615 (24.8%) 0.0
Valvular disease 2112 (6.8%) 2165 (6.2%) 9.8 2023 (6.5%) 2273 (6.5%) 0.1
Chronic liver disease 1385 (4.5%) 1577 (4.5%) 1.6 1400 (4.5%) 1570 (4.5%) 0.1
Diabetes 21,520 (69.6%) 23,690 (68.2%) 2.0 21,305 (68.9%) 23,927 (68.9%) 0.0
History of noncompliance 3501 (11.3%) 4072 (11.7%) 3.7 3575 (11.6%) 4013 (11.6%) 0.0
Had a cardiac pacemaker 1431 (4.6%) 1463 (4.2%) 10.6 1361 (4.4%) 1529 (4.4%) 0.0
Had cardiac surgery during the last 30 d of baseline 588 (1.9%) 614 (1.8%) 10.7 565 (1.8%) 635 (1.8%) 0.1
Had an ECG during the last 30 d of baseline 9908 (32.0%) 10,796 (31.1%) 3.0 9777 (31.6%) 10,974 (31.6%) 0.1
ACE inhibitor 5749 (18.6%) 6542 (18.8%) 1.4 5801 (18.8%) 6509 (18.7%) 0.1
ARB 3018 (9.8%) 3530 (10.2%) 4.3 3082 (10.0%) 3462 (10.0%) 0.0
β blocker 11,442 (37.0%) 13,114 (37.8%) 2.1 11,562 (37.4%) 12,981 (37.4%) 0.0
Calcium channel blocker 8768 (28.3%) 10,385 (29.9%) 5.5 9031 (29.2%) 10,141 (29.2%) 0.0
Central α agonist 3906 (12.6%) 4413 (12.7%) 0.7 3913 (12.7%) 4394 (12.7%) 0.0
Diuretic 2903 (9.4%) 3565 (10.3%) 9.5 3048 (9.9%) 3420 (9.8%) 0.1
Use of ≥1 medication with a known risk of TdP c 2919 (9.4%) 3210 (9.2%) 2.2 2891 (9.3%) 3244 (9.3%) 0.1
Use of ≥1 medication with a conditional risk of TdP c 12,634 (40.8%) 14,238 (41.0%) 0.4 12,675 (41.0%) 14,231 (41.0%) 0.0
Use of ≥1 medication with a possible risk of TdP c 3228 (10.4%) 3123 (9.0%) 15.6 3001 (9.7%) 3375 (9.7%) 0.1
Use of ≥1 CYP 1A2 inhibitor d 1154 (3.7%) 1290 (3.7%) 0.5 1155 (3.7%) 1296 (3.7%) 0.1
Use of ≥1 CYP 3A4 inhibitor d 2419 (7.8%) 2770 (8.0%) 2.1 2450 (7.9%) 2746 (7.9%) 0.2
Use of ≥1 CYP 2C9 inhibitor d 2074 (6.7%) 2385 (6.9%) 2.6 2104 (6.8%) 2360 (6.8%) 0.1
Use of ≥1 CYP 2C19 inhibitor d 8119 (26.2%) 9022 (26.0%) 1.0 8095 (26.2%) 9087 (26.2%) 0.0
Use of ≥1 CYP 2D6 inhibitor d 8777 (28.4%) 9999 (28.8%) 1.5 8845 (28.6%) 9931 (28.6%) 0.0
Hospitalized during the last 30 d of the baseline period 8701 (28.1%) 8749 (25.2%) 11.2 8248 (26.7%) 9264 (26.7%) 0.0
Had ≥1 psychotherapy visit during the baseline period 3314 (10.7%) 2738 (7.9%) 31.8 2859 (9.2%) 3211 (9.2%) 0.0
Values are given as number (percentage) for categoric variables and as mean±SD for continuous variables. Higher QT-prolonging–potential SSRIs included citalopram and escitalopram. Lower QT-prolonging–potential SSRIs included fluoxetine, fluvoxamine, paroxetine, and sertraline. All covariates were measured during the 180-d baseline period before SSRI initiation. The weighted cohort is the pseudo-population generated by the IPT weighting. Std diff, standardized differences; ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; TdP, torsades de pointes; CYP, cytochrome P450.
aA std diff >10.0% represents meaningful imbalance between groups.33
bThe definition of low-dose was on the basis of the dosing recommendations found in each SSRI’s package insert.62–67 Low doses: citalopram≤20 mg/d; escitalopram≤10 mg/d; fluoxetine≤20 mg/d; immediate release fluvoxamine ≤50 mg/d; controlled release fluvoxamine ≤100 mg/d; immediate release paroxetine ≤20 mg/d; controlled release paroxetine ≤25 mg/d; and sertraline 50 mg/d.
cLists of medications with known, conditional, and possible risks of TdP are presented in Supplemental Table 3.
dLists of medications that are relevant CYP 1A2, 3A4, 2C9, 2C19, and 2D6 inhibitors are provided in Supplemental Table 4.

Primary Analyses

Under the on-treatment analytic paradigm, the study cohort was followed for a total of 16,633 person-years (7858 person-years for initiators of a higher QT-prolonging–potential SSRI and 8775 person-years for initiators of a lower QT-prolonging–potential SSRI). A total of 51,005 individuals discontinued index SSRI therapy, and 2017 individuals switched to nonindex SSRI during follow-up. The median (interquartile range) duration of follow-up was 37 (65) days among patients initiating an SSRI of higher QT-prolonging potential, and 37 (63) days among patients initiating an SSRI with lower QT-prolonging potential. Exposure to non-SSRI QT-prolonging medications during follow-up was similar among initiators of higher and lower QT-prolonging–potential SSRIs.

During the 1-year follow-up period, 1303 sudden cardiac deaths occurred at an incidence rate of 78.3 events per 1000 person-years (702 sudden death events occurred at an incidence rate of 89.3 events per 1000 person-years among initiators of higher QT-prolonging–potential SSRIs, and 601 sudden death events occurred at a rate of 68.5 events per 1000 person-years among initiators of lower QT-prolonging–potential SSRIs). Figure 3 and Supplemental Table 7 display the SSRI QT-prolonging potential–sudden cardiac death association. Compared with individuals initiating lower QT-prolonging–potential SSRIs, individuals initiating higher QT-prolonging–potential SSRIs had a higher 1-year risk of sudden cardiac death (adjusted HR, 1.18; 95% confidence interval, 1.05 to 1.31).

fig3
Figure 3.:
Initiation of a higher versus lower QT-prolonging–potential SSRI associates with a higher 1-year risk of fatal cardiac outcomes. An on-treatment analytic approach was used in all analyses. Fine and Gray proportional subdistribution hazards models were used to estimate the association between the initiation of higher versus lower QT-prolonging–potential SSRI and the 1-year risk of fatal cardiac outcomes. Higher QT-prolonging–potential SSRIs included citalopram and escitalopram. Lower QT-prolonging–potential SSRIs included fluoxetine, fluvoxamine, paroxetine, and sertraline. Adjusted analyses controlled for baseline covariates listed in Supplemental Table 6 using IPT weighting. 95% CI, 95% confidence interval; ref., referent.

Secondary Analyses

Secondary analyses assessing the SSRI QT-prolonging potential–sudden cardiac death association within clinically relevant subgroups produced results analogous to primary study findings. Across all subgroups, initiation of a higher versus lower QT-prolonging–potential SSRI was associated with a higher 1-year risk of sudden cardiac death. However, the observed associations were more potent in patients aged ≥65 years (versus <65 years), females (versus males), individuals with (versus without) a cardiac conduction disorder, and those using (versus not using) ≥1 other non-SSRI QT-prolonging medication (Figure 4, Supplemental Table 8).

fig4
Figure 4.:
Initiation of a higher versus lower QT-prolonging–potential SSRI associates with a higher 1-year risk of sudden cardiac death within clinically relevant subgroups. An on-treatment analytic approach was used in all analyses. Fine and Gray proportional subdistribution hazards models were used to estimate the association between the initiation of higher versus lower QT-prolonging–potential SSRIs and the 1-year risk of sudden cardiac death within clinically relevant subgroups. Higher QT-prolonging–potential SSRIs included citalopram and escitalopram. Lower QT-prolonging–potential SSRIs included fluoxetine, fluvoxamine, paroxetine, and sertraline. Adjusted analyses controlled for baseline covariates listed in Supplemental Table 6 using IPT weighting. The square sizes of HR point estimates are proportional to the size of the subgroup. The larger the square size the larger the subgroup. Within each respective subgroup, the interaction P value was ≥0.05. 95% CI, 95% confidence interval; med, medication.

Sensitivity Analyses

Sensitivity analyses (1) evaluating alternative outcomes (a composite outcome of sudden cardiac death or hospitalized ventricular arrhythmia; and cardiovascular mortality, separately), (2) comparing initiators of citalopram and escitalopram (separately) to initiators of lower QT-prolonging–potential SSRIs, (3) using longer grace periods of 14 and 30 days to identify index SSRI discontinuation, (4) considering all possible follow-up time, and (5) employing an intent-to-treat analytic approach produced results that were consistent with our primary analysis (Figure 3, Supplemental Tables 7 and 9–12). In addition, sensitivity analyses evaluating the SSRI QT-prolonging potential–negative control outcome association demonstrated that the initiation of a higher versus lower QT-prolonging–potential SSRI was not associated with the 1-year risk of nonsudden cardiac death (adjusted HR, 1.01; 95% confidence interval, 0.95 to 1.09; Supplemental Table 13).

Discussion

This observational study evaluated the comparative cardiac safety of SSRIs in the hemodialysis population. We found that individuals initiating higher (citalopram, escitalopram) versus lower (fluoxetine, fluvoxamine, paroxetine, sertraline) QT-prolonging–potential SSRIs had a higher 1-year risk of sudden cardiac death. This association was robust across multiple sensitivity analyses. Furthermore, our exploratory subgroup analyses suggest that the observed SSRI QT-prolonging potential–sudden cardiac death association may be more pronounced among elderly individuals, females, patients with conduction disorders, and those treated with other non-SSRI QT-prolonging medications.

An undesirable property of SSRIs is their ability to delay ventricular repolarization, an effect that manifests as QT interval prolongation on an electrocardiogram (ECG).1618 All six SSRIs can prolong the QT interval via their antagonistic effects on myocardial human ether-a-go-go (hERG) potassium channels.4045 However, despite their common off-target effects on myocardial potassium channels, individual agents within the SSRI class have differential QT-prolonging abilities.1618 Citalopram and escitalopram (the S-enantiomer of the citalopram racemate) prolong the QT interval more than other SSRIs.1618 In fact, randomized, placebo-controlled trials conducted in healthy volunteers have demonstrated that citalopram and escitalopram can prolong the QT interval beyond the threshold of regulatory concern, an increase of 5 milliseconds from baseline, 46,47 at therapeutic doses.1921

To date, data characterizing the QT-prolonging effects and cardiac safety profiles of SSRIs in the hemodialysis population are limited. QT prolongation and ventricular arrhythmias, such as torsades de pointes, were not reported in clinical trials evaluating the antidepressant efficacy of SSRI therapy in dialysis patients.615,48 However, it is likely that cardiac safety signals could not be detected in these trials due to their small sample sizes (7–62 participants) and limited follow-up (4 weeks to 6 months).615,48 To our knowledge, published evidence linking SSRIs to adverse cardiac outcomes in ESRD is limited to a few case reports of citalopram- and escitalopram-induced QT prolongation and torsades de pointes.4951 Given the paucity of population-specific data, international experts have called for high-quality safety studies to inform antidepressant prescribing in patients receiving dialysis.52

To begin to address this evidence gap, we conducted a pharmacoepidemiologic study to assess the association between the initiation of higher versus lower QT-prolonging–potential SSRIs and sudden cardiac death in the hemodialysis population. Our findings provide initial evidence that the heterogenous QT-prolonging potential of SSRIs may differentially affect cardiac outcomes in hemodialysis-dependent ESRD. In addition, results from our exploratory subgroup analyses suggest that patients with risk factors for drug-induced QT prolongation (e.g., advanced age, female sex, cardiac conduction disorders, and use of non-SSRI QT-prolonging medications) may have heightened susceptibly to the proarrhythmic effects of higher QT-prolonging–potential SSRIs. Although our subgroup analyses should be considered preliminary in nature, our findings do align with published postmarketing surveillance data. Cases of citalopram- and escitalopram-induced QT prolongation and torsades de pointes reported to pharmaceutic regulatory agencies primarily involved individuals with risk factors for drug-induced QT prolongation (e.g., female sex, preexisting QT prolongation, or known cardiac disease).20,21

Relative to the general population, individuals receiving maintenance hemodialysis may have an increased susceptibility to the QT-prolonging effects of citalopram and escitalopram for several reasons. First, cardiovascular comorbid conditions that result in structural heart damage are common in ESRD. Cardiovascular remodeling due to coronary artery disease, left ventricular hypertrophy, and heart failure can lead to a progressive downregulation of myocardial ion channels, diminishing cardiac repolarization reserve,5355 ultimately leaving the heart vulnerable to proarrhythmic triggers such as QT-prolonging drugs. Second, patients on hemodialysis are exposed to electrolyte shifts during dialysis treatments. It is plausible that rapid dialytic removal of electrolytes, such as potassium, and the use of proarrhythmic medications have additive QT-prolonging effects. Exposure to wider versus narrower blood-to-dialysate electrolyte gradients results in more extensive dialysis-induced QT prolongation.56 Furthermore, in vitro models have demonstrated that reductions in extracellular potassium enhance the degree of drug-induced inhibition of the cardiac repolarizing ionic current IKr.57 Finally, patients receiving hemodialysis have high medication burdens, increasing the likelihood that they are exposed to interacting medications. Concurrent use of two or more QT-prolonging medications and the use of a QT-prolonging medication in combination with a metabolic inhibitor may result in QT interval lengthening.35,36

Future studies are needed to fully elucidate the cardiac safety profiles of SSRIs in the hemodialysis population, including the frequency and magnitude of SSRI-induced QT prolongation measured via serial ECG monitoring or implantable loop recorders. However, in the interim, our results suggest that SSRI therapy selection should be individualized and clinicians should consider the differential QT-prolonging properties of SSRIs, among other factors, in their prescribing choices. In particular, prescribers should be mindful of the demographic, comorbid, and medication-related risk factors for drug-induced QT prolongation, taking into account patient medical history, laboratory results, ECG findings, and concomitant medication use when selecting an SSRI. A lower QT-prolonging–potential SSRI may be a better option for patients at high risk for drug-induced QT prolongation, such as women, the elderly, individuals with cardiac conduction abnormalities, and those treated with other non-SSRI QT-prolonging medications.35,36 However, if citalopram or escitalopram therapy is unavoidable in a high-risk patient, clinicians should look to the drug safety warnings issued by pharmaceutic regulatory agencies for guidance on therapeutic monitoring.1924 Regulatory agencies, including the FDA, recommend serial ECG monitoring throughout the course of therapy in high-risk patients.19,2224 If the length of a patient’s corrected QT interval surpasses 500 milliseconds, treatment should be withdrawn gradually.19,22,23

Our study has several strengths. First, the use of Medicare claims data enabled us to conduct a large-scale safety study in a cohort of >65,000 hemodialysis patients initiating SSRI therapy. Second, we utilized a new-user study design to mitigate biases common to observational studies of prescription drugs such as selection and immortal time biases.25 Third, by performing a head-to-head comparison of SSRIs with higher versus lower QT-prolonging potential, we minimized the influence of bias due to confounding by indication because medications within the SSRI class have similar clinical indications and therapeutic roles.30,58 Furthermore, the comparison of SSRIs with a higher versus lower QT-prolonging potential reflects a clinically meaningful treatment decision encountered by prescribers in real-world practice.30,31 Finally, we performed multiple sensitivity analyses to test the robustness of our primary results.

Our findings should be considered within the context of study limitations. First, because our study was observational, residual confounding may remain. However, we adjusted for a wide-range of clinical and health care utilization metrics in our analyses to minimize confounding from difficult-to-measure factors such as ambient health status. In particular, the lack of an observed association between SSRI QT-prolonging potential and nonsudden cardiac death (the negative control outcome) suggests that the influence of unmeasured confounding in our study was minimal.39 Second, although we were able to determine whether an ECG was conducted in the month before SSRI initiation, we lacked accompanying clinical information to know whether ECG findings were used to inform SSRI-prescribing decisions. Third, although we defined our primary study outcome using the established USRDS sudden cardiac death definition, contemporary information on the validity of this outcome definition (e.g., sensitivity, specificity, positive predictive value, and negative predictive value) is limited.59,60 Because cause of death information was taken from the ESRD Death Notification Form, outcome misclassification may have occurred. However, such misclassification would likely be nondifferential (i.e., not dependent on exposure classification), biasing results toward the null.61 Reassuringly, sensitivity analyses considering broader cardiac outcomes produced consistent results. Fourth, information on missed or shortened hemodialysis treatments is not available in USRDS data. However, to minimize associated confounding, we adjusted our analyses for a history of noncompliance (International Classification of Diseases, Ninth Revision diagnosis codes V15.81 and V45.12). Fifth, information on laboratory parameters, including serum electrolytes, and patient-reported medication ineffectiveness or side effects were not available. We were unable to determine the clinical factor(s) that may have led to SSRI discontinuation or therapy switches during follow-up. Sixth, our subgroup analyses had limited power. Despite strong biologic plausibility, additional research is needed to confirm our findings. Finally, our study population was comprised of prevalent ESRD patients receiving in-center hemodialysis. Our results may not generalize to excluded populations such as incident hemodialysis, home hemodialysis, or peritoneal dialysis patients. Understanding the relative risk-benefit profiles of SSRIs in excluded patient populations is an area for future inquiry.

In conclusion, we observed that the initiation of a higher versus lower QT-prolonging–potential SSRI was associated with a higher 1-year risk of sudden cardiac death in a cohort of prevalent patients on hemodialysis. Data from our subgroup analyses suggest that the SSRI QT-prolonging potential–sudden cardiac death association may be more potent among individuals with established risk factors for drug-induced QT prolongation. When prescribing SSRIs to hemodialysis patients, providers should consider the QT-prolonging potential of these agents.

Disclosures

M.M.A. and J.E.F. have received investigator-initiated research funding from the Renal Research Institute, a subsidiary of Fresenius Medical Care, North America. M.A.B. has received research support from Amgen and AstraZeneca; served as a scientific advisor for Merck, Amgen, Genentech, Fibrogen, AbbVie, and RxAnte; and owns equity in NoviSci, LLC, a data sciences company. In the last 2 years, J.E.F. has received speaking honoraria from American Renal Associates; the American Society of Nephrology; Dialysis Clinic, Inc.; the National Kidney Foundation; and multiple universities. J.E.F. is on the medical advisory board of NxStage Medical, Inc. and has received consulting fees from Fresenius Medical Care, North America.

Published online ahead of print. Publication date available at www.jasn.org.

See related editorial, “Piecing Together the Risk of Sudden Cardiac Death on Dialysis,” on pages .

J.E.F. is supported by grant K23 DK109401 awarded by the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health.

Aspects of this study were presented in an oral abstract at the 2018 American Society of Nephrology Kidney Week meeting in San Diego, CA.

The data reported here have been provided by the US Renal Data System. The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as official policy or interpretation of the United States government.

Research idea and study design: M.M.A. and J.E.F.; data acquisition: M.M.A., M.A.B., and J.E.F.; statistical analysis: M.M.A.; data interpretation: M.M.A., M.A.B., and J.E.F.; and supervision: J.E.F. All authors approved the final version of the manuscript.

Supplemental Material

This article contains the following supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2018101032/-/DCSupplemental.

Supplemental Figure 1. Illustration of SSRI discontinuation and switching events.

Supplemental Figure 2. Propensity score distribution of patients treated with higher and lower QT-prolonging-potential SSRIs.

Supplemental Table 1. Outcome definitions.

Supplemental Table 2. ICD-9 diagnosis, ICD-9 procedure, and CPT procedure codes used to identify relevant baseline covariates.

Supplemental Table 3. List of non-SSRI QT-prolonging medications.

Supplemental Table 4. HCPCS codes used to identify QT-prolonging medications not billable to Medicare Part D.

Supplemental Table 5. List of cytochrome P450 inhibitors.

Supplemental Table 6. Full list of baseline characteristics among study patients initiating a higher and lower QT-prolonging–potential SSRI.

Supplemental Table 7. Association between the initiation of a higher versus lower QT-prolonging–potential SSRI and the 1-year risk of fatal cardiac outcomes.

Supplemental Table 8. Association between the initiation of a higher versus lower QT-prolonging–potential SSRI and the 1-year risk of sudden cardiac death within clinically relevant subgroups.

Supplemental Table 9. Association between the initiation of individual higher QT-prolonging–potential SSRIs versus lower QT-prolonging–potential SSRIs and the 1-year risk of sudden cardiac death.

Supplemental Table 10. Association between the initiation of a higher versus lower QT-prolonging–potential SSRI and the 1-year risk of sudden cardiac death when longer grace periods were used to define SSRI discontinuation.

Supplemental Table 11. Association between the initiation of a higher versus lower QT-prolonging–potential SSRI and the risk of sudden cardiac death considering all possible follow-up time.

Supplemental Table 12. Association between the initiation of a higher versus lower QT-prolonging–potential SSRI and the 1-year risk of sudden cardiac death using an intent-to-treat analytic approach.

Supplemental Table 13. Association between the initiation of an SSRI with higher versus lower QT-prolonging potential and the 1-year risk of the negative control outcome.

References

1. Palmer S, Vecchio M, Craig JC, Tonelli M, Johnson DW, Nicolucci A, et al.: Prevalence of depression in chronic kidney disease: Systematic review and meta-analysis of observational studies. Kidney Int 84: 179–191, 2013 23486521
2. Shirazian S, Grant CD, Aina O, Mattana J, Khorassani F, Ricardo AC: Depression in chronic kidney disease and end-stage renal disease: Similarities and differences in diagnosis, epidemiology, and management. Kidney Int Rep 2: 94–107, 2016 29318209
3. Centers for Medicare & Medicaid Services (CMS) End-Stage Renal Disease: (ESRD) Quality Incentive Program (QIP) Payment Year (PY) 2018 final measure technical specifications. 2016. Available at: https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/ESRDQIP/Downloads/PY-2018-Technical-Measure-Specifications.pdf. Accessed September 27, 2018
4. American Psychiatric Association: Practice guideline for the treatment of patients with major depressive disorder, Third Edition, 2010. http://psychiatryonline.org/pb/assets/raw/sitewide/practice_guidelines/guidelines/mdd.pdf. Accessed September 27, 2018
5. Saran R, Robinson B, Abbott KC, Agodoa LYC, Bhave N, Bragg-Gresham J, et al.: US Renal Data System 2017 annual data report: Epidemiology of kidney disease in the United States. Am J Kidney Dis; 71(3S1):A7, 2018
6. Levy NB, Blumenfield M, Beasley CM Jr., Dubey AK, Solomon RJ, Todd R, et al.: Fluoxetine in depressed patients with renal failure and in depressed patients with normal kidney function. Gen Hosp Psychiatry 18: 8–13, 1996 8666216
7. Blumenfield M, Levy NB, Spinowitz B, Charytan C, Beasley CM Jr., Dubey AK, et al.: Fluoxetine in depressed patients on dialysis. Int J Psychiatry Med 27: 71–80, 1997 9565715
8. Kamo T, Horikawa N, Tsuruta Y, Miyasita M, Hatakeyama H, Maebashi Y: Efficacy and pharmacokinetics of fluvoxamine maleate in patients with mild depression undergoing hemodialysis. Psychiatry Clin Neurosci 58: 133–137, 2004 15009816
9. Lee SK, Lee HS, Lee TB, Kim DH, Koo JR, Kim YK, et al.: The effects of antidepressant treatment on serum cytokines and nutritional status in hemodialysis patients. J Korean Med Sci 19: 384–389, 2004 15201504
10. Koo JR, Yoon JY, Joo MH, Lee HS, Oh JE, Kim SG, et al.: Treatment of depression and effect of antidepression treatment on nutritional status in chronic hemodialysis patients. Am J Med Sci 329: 1–5, 2005 15654172
11. Kalender B, Ozdemir AC, Yalug I, Dervisoglu E: Antidepressant treatment increases quality of life in patients with chronic renal failure. Ren Fail 29: 817–822, 2007 17994449
12. Hosseini SH, Espahbodi F, Mirzadeh Goudarzi SM: Citalopram versus psychological training for depression and anxiety symptoms in hemodialysis patients. Iran J Kidney Dis 6: 446–451, 2012 23146983
13. Yazıcı AE, Erdem P, Erdem A, Yazici K, Acar ST, Basterzi AD ., et al.: Efficacy and tolerability of escitalopram in depressed patients with end stage renal disease: An open placebo-controlled study. Klinik Psikofarmakoloji Bülteni 22: 23–30, 2012
14. Taraz M, Khatami MR, Dashti-Khavidaki S, Akhonzadeh S, Noorbala AA, Ghaeli P, et al.: Sertraline decreases serum level of interleukin-6 (IL-6) in hemodialysis patients with depression: Results of a randomized double-blind, placebo-controlled clinical trial. Int Immunopharmacol 17: 917–923, 2013 24121064
15. Zahed NS, Sharifi M, Karimi M, Nikbakht H: Impact of sertraline on serum concentration of CRP in hemodialysis patients with depression. J Renal Inj Prev 6: 65–69, 2016 28487875
16. Castro VM, Clements CC, Murphy SN, Gainer VS, Fava M, Weilburg JB, et al.: QT interval and antidepressant use: A cross sectional study of electronic health records. BMJ 346: f288, 2013 23360890
17. Funk KA, Bostwick JR: A comparison of the risk of QT prolongation among SSRIs. Ann Pharmacother 47: 1330–1341, 2013 24259697
18. Beach SR, Kostis WJ, Celano CM, Januzzi JL, Ruskin JN, Noseworthy PA, et al.: Meta-analysis of selective serotonin reuptake inhibitor-associated QTc prolongation. J Clin Psychiatry 75: e441–e449, 2014 24922496
19. United States Food and Drug Administration: FDA Drug Safety Communication: Revised recommendations for Celexa (citalopram hydrobromide) related to a potential risk of abnormal heart rhythms with high doses. 2012. Available at: https://www.fda.gov/Drugs/DrugSafety/ucm297391.htm. Accessed September 27, 2018
20. The Pharmacovigilance Working Party of the European Medicines Agency: November 2011 plenary meeting–safety concerns. 2011. Available at: https://www.ema.europa.eu/documents/report/monthly-report-pharmacovigilance-working-party-phvwp-november-2011-plenary-meeting_en.pdf. Accessed September 27, 2018
21. The Pharmacovigilance Working Party of the European Medicines Agency: October 2011 plenary meeting–safety concerns. 2011. Available at: https://www.ema.europa.eu/documents/report/monthly-report-pharmacovigilance-working-party-phvwp-october-2011-plenary-meeting_en.pdf. Accessed September 27, 2018
22. The United Kingdom Medicines and Healthcare products Regulatory Agency: Citalopram and escitalopram: QT interval prolongation. 2014. Available at: https://www.gov.uk/drug-safety-update/citalopram-and-escitalopram-qt-interval-prolongation. Accessed September 27, 2018
23. Health Canada: Health Canada Endorsed Important Safety Information on CELEXA® (citalopram hydrobromide). 2012. Available at: https://www.lundbeck.com/upload/ca/en/files/pdf/productcommunication/Celexa%20HPC_ENG_%20e-signature_20Jan2012.pdf. Accessed September 27, 2018
24. Health Canada: Antidepressant Cipralex (escitalopram): Updated information regarding dose-related heart risk. 2012. Available at: http://healthycanadians.gc.ca/recall-alert-rappel-avis/hc-sc/2012/13674a-eng.php. Accessed September 27, 2018
25. Ray WA: Evaluating medication effects outside of clinical trials: New-user designs. Am J Epidemiol 158: 915–920, 2003 14585769
26. Woosley RL, Heise CW, Romero KA: QT drugs list. 2018. Available at: www.Crediblemeds.org. Accessed September 27, 2018
27. United States Renal Data System ESRD Analytical Methods. 2018. Available at: https://www.usrds.org/2017/view/v2_00_appx.aspx. Accessed September 27, 2018
28. Brookhart MA, Schneeweiss S, Rothman KJ, Glynn RJ, Avorn J, Stürmer T: Variable selection for propensity score models. Am J Epidemiol 163: 1149–1156, 2006 16624967
29. The Flockhart Table™. 2018. Available at: https://drug-interactions.medicine.iu.edu/Main-Table.aspx. Accessed September 27, 2018
30. Velentgas P, Dreyer NA, Nourjah P, Smith SR, Torchia MM, editors: Developing a Protocol for Observational Comparative Effectiveness Research: A User’s Guide, Rockville, MD, 2013
31. Brookhart MA: Counterpoint: The treatment decision design. Am J Epidemiol 182: 840–845, 2015 26507307
32. Gilbertson DT, Bradbury BD, Wetmore JB, Weinhandl ED, Monda KL, Liu J, et al.: Controlling confounding of treatment effects in administrative data in the presence of time-varying baseline confounders. Pharmacoepidemiol Drug Saf 25: 269–277, 2016 26608680
33. Austin PC: Using the standardized difference to compare the prevalence of a binary variable between two groups in observational research. Commun Stat Simul Comput 38: 1228–1234, 2009
34. Fine JP, Gray RJ: A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 94: 496–509, 1999
35. Trinkley KE, Page RL 2nd, Lien H, Yamanouye K, Tisdale JE: QT interval prolongation and the risk of torsades de pointes: Essentials for clinicians. Curr Med Res Opin 29: 1719–1726, 2013 24020938
36. Turner JR, Rodriguez I, Mantovani E, Gintant G, Kowey PR, Klotzbaugh RJ, et al.; Cardiac Safety Research Consortium: Drug-induced proarrhythmia and torsade de Pointes: A primer for students and practitioners of medicine and pharmacy [published online ahead of print April 19, 2018]. J Clin Pharmacol 10.1002/jcph.1129 29672845
37. Cole SR, Hernán MA: Constructing inverse probability weights for marginal structural models. Am J Epidemiol 168: 656–664, 2008 18682488
38. Brookhart MA, Wyss R, Layton JB, Stürmer T: Propensity score methods for confounding control in nonexperimental research. Circ Cardiovasc Qual Outcomes 6: 604–611, 2013 24021692
39. Dusetzina SB, Brookhart MA, Maciejewski ML: Control outcomes and exposures for improving internal validity of nonrandomized studies. Health Serv Res 50: 1432–1451, 2015 25598384
40. Witchel HJ, Pabbathi VK, Hofmann G, Paul AA, Hancox JC: Inhibitory actions of the selective serotonin re-uptake inhibitor citalopram on HERG and ventricular L-type calcium currents. FEBS Lett 512: 59–66, 2002 11852052
41. Chae YJ, Jeon JH, Lee HJ, Kim IB, Choi JS, Sung KW, et al.: Escitalopram block of hERG potassium channels. Naunyn Schmiedebergs Arch Pharmacol 387: 23–32, 2014 24045971
42. Thomas D, Gut B, Wendt-Nordahl G, Kiehn J: The antidepressant drug fluoxetine is an inhibitor of human ether-a-go-go-related gene (HERG) potassium channels. J Pharmacol Exp Ther 300: 543–548, 2002 11805215
43. Milnes JT, Crociani O, Arcangeli A, Hancox JC, Witchel HJ: Blockade of HERG potassium currents by fluvoxamine: Incomplete attenuation by S6 mutations at F656 or Y652. Br J Pharmacol 139: 887–898, 2003 12839862
44. Lee SH, Sung MJ, Lee HM, Chu D, Hahn SJ, Jo SH, et al.: Blockade of HERG human K+ channels by the antidepressant drug paroxetine. Biol Pharm Bull 37: 1495–1504, 2014 25177033
45. Lee HA, Kim KS, Hyun SA, Park SG, Kim SJ: Wide spectrum of inhibitory effects of sertraline on cardiac ion channels. Korean J Physiol Pharmacol 16: 327–332, 2012 23118556
46. United States Food and Drug Administration: Guidance for industry–E14 clinical evaluation of QT/QTc interval prolongation and proarrhythmic potential for non-antiarrhythmic drugs. Available at: https://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm073153.pdf. Accessed September 27, 2018
47. European Medicines Agency: Guidance for industry–E14 clinical evaluation of QT/QTc interval prolongation and proarrhythmic potential for non-antiarrhythmic drugs. Available at: https://www.ema.europa.eu/documents/scientific-guideline/ich-e-14-clinical-evaluation-qt/qts-interval-prolongation-proarrhythmic-potential-non-antiarrhythmic-drugs-step-5_en.pdf. Accessed September 27, 2018
48. Friedli K, Guirguis A, Almond M, Day C, Chilcot J, Da Silva-Gane M, et al.: Sertraline versus placebo in patients with major depressive disorder undergoing hemodialysis: A randomized, controlled feasibility trial. Clin J Am Soc Nephrol 12: 280–286, 2017 28126706
49. Kanjanauthai S, Kanluen T, Chareonthaitawee P: Citalopram induced torsade de pointes, a rare life threatening side effect. Int J Cardiol 131: e33–e34, 2008 17919753
50. Van Asbroeck PJ, Huybrechts W, De Soir R: Case report, aetiology, and treatment of an acquired long-QT syndrome. Acta Clin Belg 69: 132–134, 2014 24724757
51. Medsafe the New Zealand Medicines and Medical Devices Safety Authority: Reminder: Citalopram and QT prolongation. 2015. Available at: https://medsafe.govt.nz/profs/PUArticles/Sep2015/Citalopram&QTReminder.htm. Accessed September 27, 2018
52. Nagler EV, Webster AC, Vanholder R, Zoccali C: Antidepressants for depression in stage 3-5 chronic kidney disease: A systematic review of pharmacokinetics, efficacy and safety with recommendations by European Renal Best Practice (ERBP). Nephrol Dial Transplant 27: 3736–3745, 2012 22859791
53. Nattel S, Maguy A, Le Bouter S, Yeh YH: Arrhythmogenic ion-channel remodeling in the heart: Heart failure, myocardial infarction, and atrial fibrillation. Physiol Rev 87: 425–456, 2007 17429037
54. Varró A, Baczkó I: Cardiac ventricular repolarization reserve: A principle for understanding drug-related proarrhythmic risk. Br J Pharmacol 164: 14–36, 2011 21545574
55. Gussak I, Gussak HM: Sudden cardiac death in nephrology: Focus on acquired long QT syndrome. Nephrol Dial Transplant 22: 12–14, 2007 17093009
56. Genovesi S, Dossi C, Viganò MR, Galbiati E, Prolo F, Stella A, et al.: Electrolyte concentration during haemodialysis and QT interval prolongation in uraemic patients. Europace 10: 771–777, 2008 18287086
57. Yang T, Roden DM: Extracellular potassium modulation of drug block of IKr. Implications for torsade de pointes and reverse use-dependence. Circulation 93: 407–411, 1996 8565156
58. Lund JL, Richardson DB, Stürmer T: The active comparator, new user study design in pharmacoepidemiology: Historical foundations and contemporary application. Curr Epidemiol Rep 2: 221–228, 2015 26954351
59. Bleyer AJ, Hartman J, Brannon PC, Reeves-Daniel A, Satko SG, Russell G: Characteristics of sudden death in hemodialysis patients. Kidney Int 69: 2268–2273, 2006 16672908
60. Pun PH, Herzog CA, Middleton JP: Improving ascertainment of sudden cardiac death in patients with end stage renal disease. Clin J Am Soc Nephrol 7: 116–122, 2012 22076878
61. Rothman KJ, Greenland S, Lash TL: Modern Epidemiology, 3rd Ed., edited by Rothman KJ, Greenland S, Lash TL, Philadelphia, Wolters Kluwer Health/Lippincott Williams & Wilkins, 2008, pp 142–143
62. Celexa® (citalopram hydrobromide) [package insert], Irvine, CA, Allergan USA Inc., 2017
    63. Lexapro® (escitalopram oxalate) [package insert], Irvine, CA, Allergan USA Inc., 2017
      64. Prozac® (fluoxetine hydrochloride) [package insert], Indianapolis, IN, Lilly USA LLC, 2017
        65. Fluvoxamine maleate [package insert], Baudette, MN, ANI Pharmaceuticals Inc., 2017
          66. Paxil® (paroxetine hydrochloride) [package insert], Research Triangle Park, NC, GlaxoSmithKline, 2017
            67. Zoloft® (sertraline hydrochloride) [package insert], New York, NY, Pfizer, 2017
              Keywords:

              hemodialysis; SSRIs; Safety; Sudden cardiac death

              Copyright © 2019 by the American Society of Nephrology