Exploring psychotherapeutic issues and agents in clinical practice
I recently saw a medically hospitalized patient for a psychiatric consultation. The patient had dementia and was initially taking haloperidol (Haldol®) for agitation and psychosis. A psychiatric resident who had seen the patient recommended that haloperidol be switched to quetiapine (Seroquel®) “as Seroquel is less likely to prolong QTc versus Haldol.” When I asked the resident why she had made this recommendation and whether she had any literature comparing the cardiac effects of these two drugs, she told me her recommendation was based on Table 3 of a review article published by Beach, Celano, Noseworthy, Januzzi, and Huffman (2013). This article had been distributed to faculty, residents, and nursing staff who worked on the psychiatric consultation service for a journal club discussion.
In my article, I will look closely at some of the source studies discussed by Beach et al. (2013), which they used to distinguish these two drugs. In doing so, I will illustrate the potential hazards of accepting review articles at face value.
Cardiac Effects of Psychotropic Drugs
An adverse effect of many non-cardiac drugs is their ability to delay cardiac repolarization, an effect that can be measured as a QT interval prolongation on the electrocardiogram (ECG) tracing. The QT interval varies according to heart rate, and its measurement is therefore adjusted accordingly. This adjustment is referred to as the corrected QT (QTc) interval, which is expressed in milliseconds (ms). An abnormally prolonged QTc interval is associated with an increased risk of developing torsade de pointes (TdP), a type of cardiac ventricular tachyarrhythmia associated with sudden death.
A stated objective of Beach et al. (2013) was to “comprehensively review the relationship between psychotropic medications and QT prolongation, with a specific focus on antidepressants and antipsychotics” (p. 1). Their Table 3 (“QTc-Prolongation Risk Stratification for Commonly Used Antipsychotic Medications,” p. 8) summarizes the relative risks for QTc prolongation of various antipsychotic drugs by grouping them into four categories:
- High risk—thioridazine (Mellaril®), intravenous haloperidol, ziprasidone (Geodon®).
- Moderate risk—fluphenazine (Prolixin®), oral/intramuscular (IM) haloperidol, iloperidone (Fanapt®), paliperidone (Invega®), risperidone (Risperdal®).
- Low risk—asenapine (Saphris®), lurasidone (Latuda®), olanzapine (Zyprexa®), quetiapine.
- Minimal risk—aripiprazole (Abilify®).
This table was used by the resident to justify switching haloperidol to quetiapine.
Comparing Beach et al. (2013), Harrigan Et al. (2004), and U.S. Food and Drug Administration (FDA, 2000)
Beach et al. (2013) first state:
In one of the few randomized studies of its effects of QTc, haloperidol (15 mg by mouth daily) led to an average increase in QTc of 7.1 ms, which was less than the prolongation caused by thioridazine or ziprasidone but greater than that caused by olanzapine, risperidone, or quetiapine (Harrigan et al., 2004). (p. 7)
Beach et al. (2013) do not describe the mean QTc intervals of the comparative drugs, do not include the 95% confidence intervals (CIs) surrounding the mean QTc intervals, and do not comment on the clinical or statistical significance of these data. Why do they choose to highlight the effect of haloperidol when Harrigan et al. (2004) state:
The mean baseline-corrected QTc change was greatest with thioridazine (30.1 milliseconds) and least with olanzapine (1.7 milliseconds). Ziprasidone, haloperidol, and quetiapine were associated with mean changes of 15.9, 7.1, and 5.7 milliseconds, respectively. (pp. 66-67)
Figure 2 (Harrigan et al., 2004, p. 67) shows the 95% CIs, which demonstrate no clinically significant or statistically significant difference between haloperidol and quetiapine based on this randomized trial, contradicting the Beach et al. (2013) summary Table 3.
Beach et al. (2013) then state:
In a nearly identical study of the same medications, oral haloperidol 15 mg daily led to an average QTc increase of 4.7 ms, which was less than ziprasidone, olanzapine, risperidone, quetiapine or thioridazine (FDA, 2000). (p. 7)
They do not present any other data from the FDA study to provide a clinical or statistical context for understanding the relevance of these data. Again, why are they highlighting haloperidol? The study they are referring to is contained in the lengthy and detailed briefing document from the FDA (2000). The data referred to by Beach et al. (2013) in this briefing document is Study 054 (found on pp. 78–86), and the QTc data they focus on are contained in Table 28 (p. 78).
My first comment is that the quetiapine QTc change (14.5 ms; Table 28) is greater than for haloperidol (4.7 ms; Table 28), but this is not accurately reported by Beach et al. (2013). Nevertheless, the CIs overlap, so the findings are not considered statistically significant. A 10-ms difference is also not clinically meaningful, even if it were statistically significant, especially if other studies do not consistently demonstrate a large separation between drugs. According to the FDA (2005) Guidance for Industry:
While increases in QT/QTc to >500 ms or of >60 ms over baseline are commonly used as thresholds for potential discontinuation, the exact criteria chosen for a given trial will depend on the risk-tolerance level considered appropriate for the indication and patient group in question…. There is no consensus concerning the choice of upper limit values for absolute QT/QTc interval and changes from baseline. While lower limits increase the false-positive rate, higher limits increase the risk of failing to detect a signal of concern. In clinical trials, a prolongation of QTc >500 ms during therapy has been a threshold of particular concern. Multiple analyses using different limits are a reasonable approach to this uncertainty, including: absolute QTc interval prolongation (QTc interval >450; QTc interval >480; QTc interval >500); change from baseline in QTc interval (QTc interval increases from baseline >30; QTc interval increases from baseline >60). (p. 11)
My second comment is that Study 054 contains the same data published by Harrigan et al. (2004). The study by Harrigan et al. (2004) is a published version of a subset of data from Study 054. Beach et al. (2013) did not look closely at their source material in the FDA (2000) document to realize that this is not “a nearly identical study of the same medications” (p. 7). Readers of Beach et al. (2013) are incorrectly led to believe that there are two nearly identical studies (Harrigan et al. 2004; FDA, 2000) comparing haloperidol and other antipsychotic drugs.
Comparing Beach et al. (2013), Breier Et al. (2002), and Wright Et al. (2001)
Beach et al. (2013) state:
The intramuscular (IM) form of halo-peridol similarly has a relatively mild QTc-prolonging effect of up to 8 ms (following a 7.5 mg IM injection) in patients without significant medical illness (Breier et al., 2002; Wright et al., 2001). (p. 7)
The study by Breier et al. (2002) compares four IM doses of olanzapine, one IM dose of haloperidol (7.5 mg), and IM placebo. Breier et al. (2002) described their QTc findings as follows:
No patient had an increase in the QTc interval of 500 milliseconds or greater, and there were only small baseline-to-24-hour end point changes in mean ± SD QTc intervals, with none being clinically relevant (IM olanzapine at 2.5 mg, −4.3 ± 22.3; IM olanzapine at 5.0 mg, −3.1 ±23.2; IM olanzapine at 7.5 mg, −2.8 ± 19.6; IM olanzapine at 10.0 mg, −1.9 ± 31.0; IM haloperidol at 7.5 mg, 6.5 ± 24.7; IM placebo, 1.2 ± 21.5). The incidence of potentially clinically significant QTc interval values, based on the sex-specific criteria of a QTc interval of 430 milliseconds or more for men and 450 milliseconds or more for women (IM olanzapine at 2.5 mg, 0% [0/45 patients]; IM olanzapine at 5.0 mg, 9.8% [4/41 patients]; IM olanzapine at 7.5 mg, 4.4% [2/45 patients]; IM olanzapine at 10.0 mg, 7.9% [3/38 patients]; IM haloperidol, 14.3% [5/35 patients]; IM placebo, 19.0% [8/42 patients]) was greater during treatment with IM olanzapine at 5.0 mg (P = .05), IM haloperidol (P = .01), and IM placebo (P = .002) than with IM olanzapine at 2.5 mg and with IM placebo vs IM olanzapine at 7.5 mg (P = .05). (p. 446)
The QTc change for haloperidol (6.5 ms) and placebo (1.2 ms) seen in this study is not clinically or statistically significantly different. The proportion of participants in this study having prolonged QTc intervals (according to their predefined criteria) was 19% for placebo and 14% for haloperidol. These proportions are not clinically or statistically significantly different, but why would placebo even be associated with any QTc changes? Beach et al. (2013) do not address this.
The study by Wright et al. (2001) compares one IM dose of olanzapine, one IM dose of haloperidol (7.5 mg), and IM placebo. Wright et al. (2001) describe their QTc findings as follows:
At 24 hours, there were no significant QTc interval changes from baseline (mean = −3.0 msec, SD = 21.3, for olanzapine; mean = −1.2 msec, SD = 24.4, for haloperidol; mean = −3.7 msec, SD = 26.1, for placebo) (F = 0.3, df = 2, 292, p = 0.73) or significant between-group differences (t = 0.70, df = 292, p = 0.49, for olanzapine versus haloperidol; t = 0.11, df = 292, p = 0.91, for olanzapine versus placebo). (p. 1151)
As you can see, there is no difference between haloperidol and placebo, which Beach et al. (2013) fail to discuss.
Comparing Beach et al. (2013) and FDA (2007)
Beach et al. (2013) go on to state:
Despite the relatively mild QTc prolongation associated with the oral and IM forms of haloperidol, this medication has clearly been linked to TdP. A post-marketing analysis of the adverse effects of haloperidol in 2007 identified 229 cases of prolonged QTc interval, including 73 cases of TdP, although the incidence of TdP cannot be accurately determined. While the investigators noted that many of these reports were confounded by other QT-prolonging medications and medical illnesses, the sheer quantity of case reports and the knowledge of haloperidol’s propensity to prolong QTc suggest that haloperidol may well have played a role in the development of TdP in some of these cases (FDA, 2007). (p. 7)
The FDA (2007) reference is an archived FDA alert (on the FDA website) that has not been altered or updated since it was posted in September 2007. Although Beach et al. (2013) emphasize “the sheer quantity of case reports” (p. 7), they do not quote or emphasize the FDA’s conclusion, which is boldly highlighted in the alert: “Based on case reports alone, we are unable to estimate the frequency with which QT prolongation or TdP occur following administration of these drugs” (para. 5). Case reports are non-randomized, un-blinded, open-label clinical trials with a sample size of one. They are an extreme example of ascertainment bias and selection bias, and are prone to bias by confounding. On hierarchies of quality and level of evidence, case reports are the lowest level of evidence because they are lowest quality. I searched the FDA website and it does not appear that the FDA requested a thorough QT/QTc study of haloperidol be conducted, despite these case reports. The FDA also has not required a thorough QT/QTc study of quetiapine (Hasnain et al., 2014). The product labels for haloperidol and quetiapine do not substantially differ in their characterization of their respective QTc risks.
Comparing Beach et al. (2013) and Ozeki Et al. (2010)
Beach et al. (2013) continue:
The intravenous (IV) form of haloperidol may carry a higher risk of QTc prolongation and TdP than the oral form. In a cross-sectional study of 1,017 medically healthy patients with schizophrenia, IV—but not oral—haloperidol was associated with QTc prolongation (Ozeki et al., 2010). (p. 7)
A close examination of the study design, patient selection, and statistical analysis used by Ozeki et al. (2010) is instructive. You cannot validly compare findings between drugs (or between routes of administration) using their study methodology. This was a non-randomized convenience selection of patients who were hospitalized, given medication, and had an ECG. Their regression model only included three variables: age, sex, and drug dose. Drug dose was actually expressed as chlorpromazine (Thorazine®) equivalents. Concurrent conditions and medications were not included as covariates in their model. However, 69% of patients were receiving antipsychotic polypharmacy and the investigators did not account for this in their statistical analysis. The relative risks they identified for certain drugs that were statistically significant were of small magnitude. Small or weak magnitude findings in observational cross-sectional studies are likely to be false and attributable to various forms of bias (Grimes & Schulz, 2012).
Comparing Beach et al. (2013) and Ray, Chung, Murray, Hall, & Stein (2009)
Finally, Beach et al. (2013) state:
In a separate, retrospective, population-based cohort study, both typical and atypical antipsychotics were associated with an approximately 2-fold increased risk of sudden cardiac death; similar risk increases were found among all agents examined individually (thioridazine, haloperidol, olanzapine, quetiapine, risperidone, and clozapine) (Ray et al., 2009). (p. 8)
There is no difference between haloperidol and quetiapine in the data analysis by Ray et al. (2009), although there was a step-wise relationship between dose (comparing low, moderate, and high doses) and the incidence of sudden cardiac death that was evident for quetiapine but not for haloperidol (Figure, p. 232), suggesting a greater adverse cardiac risk for higher dose quetiapine than higher dose haloperidol. The risk-ratios found in this observational study were statistically significant, but of small magnitude. As mentioned above, small magnitude effects in observational studies are likely to be attributable to bias (Grimes & Schulz, 2012).
Beach et al. (2013) emphasized the “sheer number of case reports” (p. 7) (of haloperidol-associated cases of prolonged QTc and TdP from the FDA alert) and the “knowledge of haloperidol’s propensity to prolong QTc,” (p. 7) but they do not critically discuss this in the context of the Ray et al. (2009) study. Why would Ray et al. (2009) find no significant elevated risk of sudden cardiac death for haloperidol compared to other antipsychotic agents if it is especially prone to causing QTc prolongation and cardiotoxicity?
Review articles generally serve to summarize what is known about a particular topic, based on an analysis of existing work related to the topic. Good reviews are invaluable, relieving readers from the arduous task of spending time and exerting effort on finding, reading, and interpreting original source material. Whether a review article can be considered “good,” however, is itself an article of faith—faith especially in the authors that they are accurate and unbiased, and faith also in the peer-review process that it is capable of uncovering bias and inaccuracies. There are significant problems with the interpretation and presentation by Beach et al. (2013) of the findings from the studies they reviewed, resulting in a misleading characterization of the relative risks of haloperidol and quetiapine. Without going through the details of the individual studies pertaining to ziprasidone, I believe they exaggerate the relative risk of this drug as well. Carefully reading the executive summary of the FDA (2000, p. 3–8) briefing document puts this in perspective. The “risk stratification” summary Table 3 developed by Beach et al. (2013) is not an accurate representation of the relative risks of these drugs and should be ignored. Looking at individual study methodologies closely and evaluating the findings from each study critically should be done before drawing any broad or generalized conclusions about the actual and comparative risks of various drugs. Nurses are likely to field questions from patients about the cardiac safety of psychotropic drugs and they should be familiar with this complex topic. When reading review articles on any topic, Caveat lector!
- Beach, S.R., Celano, C.M., Noseworthy, P.A., Januzzi, J.L. & Huffman, J.C. (2013). QTc prolongation, torsades de pointes, and psychotropic medications. Psychosomatics, 54, 1–13. doi:10.1016/j.psym.2012.11.001 [CrossRef]
- Breier, A., Meehan, K., Birkett, M., David, S., Ferchland, I., Sutton, V. & Wright, P. (2002). A double-blind, placebo-controlled, dose-response comparison of intramuscular olanzapine and haloperidol in the treatment of acute agitation in schizophrenia. Archives of General Psychiatry, 59, 441–448. doi:10.1001/archpsyc.59.5.441 [CrossRef]
- Grimes, D.A. & Schulz, K.F. (2012). False alarms and pseudo-epidemics: The limitations of observational epidemiology. Obstetrics and Gynecology, 120, 920–927. doi:10.1097/AOG.0b013e31826af61a [CrossRef]
- Harrigan, E.P., Miceli, J.J., Anziano, R., Watsky, E., Reeves, K.R., Cutler, N.R. & Middle, M. (2004). A randomized evaluation of the effects of six antipsychotic agents on QTc, in the absence and presence of metabolic inhibition. Journal of Clinical Psychopharmacology, 24, 62–69. doi:10.1097/01.jcp.0000104913.75206.62 [CrossRef]
- Hasnain, M., Vieweg, W.V., Howland, R.H., Kogut, C., Breden Crouse, E.L., Koneru, J.N. & Pandurangi, A.K. (2014). Quetiapine and the need for a thorough QT/QTc study. Journal of Clinical Psychopharmacology, 34, 3–6. doi:10.1097/JCP.0000000000000075 [CrossRef]
- Ozeki, Y., Fujii, K., Kurimoto, N., Yamada, N., Okawa, M., Aoki, T. & Kunugi, H. (2010). QTc prolongation and antipsychotic medications in a sample of 1017 patients with schizophrenia. Progress in Neuro-Psychopharmacology & Biological Psychiatry, 34, 401–405. doi:10.1016/j.pnpbp.2010.01.008 [CrossRef]
- Ray, W.A., Chung, C.P., Murray, K.T., Hall, K. & Stein, C.M. (2009). Atypical antipsychotic drugs and the risk of sudden cardiac death. New England Journal of Medicine, 360, 225–235. doi:10.1056/NEJMoa0806994 [CrossRef]
- U.S. Food and Drug Administration. (2000, July19). Briefing document for Zeldox® capsules (Ziprasidone HCl). Retrieved from http://www.fda.gov/ohrms/dockets/ac/00/backgrd/3619b1a.pdf
- U.S. Food and Drug Administration. (2005). Guidance for industry: E14 Clinical evaluation of QT/QTc interval prolongation and proarrhythmic potential for non-antiarrhythmic drugs. Retrieved from http://www.fda.gov/RegulatoryInformation/Guidances/ucm129335.htm
- U.S. Food and Drug Administration. (2007). Information for healthcare professionals: Haloperidol (marketed as Haldol, Haldol Decanoate and Haldol Lactate). Retrieved from http://www.fda.gov/Drugs/DrugSafety/ucm085203.htm
- Wright, P., Birkett, M., David, S.R., Meehan, K., Ferchland, I., Alaka, K.J. & Breier, A. (2001). Double-blind, placebo-controlled comparison of intramuscular olanzapine and intramuscular haloperidol in the treatment of acute agitation in schizophrenia. American Journal of Psychiatry, 158, 1149–1151. doi:10.1176/appi.ajp.158.7.1149 [CrossRef]