Introduction
Amiodarone infusions are used frequently for a variety of atrial and ventricular arrhythmias in children. Reference Bucknall, Keeton, Curry, Tynan, Sutherland and Holt1–Reference Saul, Scott and Brown10 Amiodarone has been associated with adverse effects ranging from hypotension, thyroid dysfunction, and hepatic dysfunction to cardiovascular collapse and death. Reference Celiker, Ceviz and Ozme4,Reference Celiker, Kocak, Lenk, Alehan and Ozme11–Reference Perry, Fenrich, Hulse, Triedman, Friedman and Lamberti14 Due to reports and experiences of cardiovascular collapse associated with amiodarone, there have been changes made in how amiodarone is used and administered. Reference McGovern, Garan and Ruskin15–Reference Ng, Hampson Evans and Murdoch17 Higher dose intravenous amiodarone boluses are now avoided, and longer infusion times are utilised. While adverse events do still occur, cardiovascular collapse does appear to have decreased with time and experience.
Despite increasing experience with amiodarone, there is still a paucity of data regarding the acute haemodynamic effects of amiodarone. A study by Haas and colleagues is one of the few studies dedicated to characterising the haemodynamic effects of an amiodarone infusion over the first 24 hours after initiation. This study demonstrated a decrease in heart rate and an increase in blood pressure without the need to increase vasoactive support in the first 24 hours. Reference Haas and Camphausen18
As detailed characterisation of the acute haemodynamic effects of amiodarone infusion is important, particularly early shortly after its initiation, the primary aim of the study was to characterise the effect of amiodarone infusion on heart rate. Secondary aims of the study were to characterise the effect of amiodarone infusion on systolic blood pressure, diastolic blood pressure, arterial saturation by pulse oximetry, cerebral near-infrared spectroscopy values, renal near-infrared spectroscopy values, and central venous pressure.
Methods
Study design
This was a single-centre, retrospective study designed to characterise the clinical effects of amiodarone infusions started in the paediatric cardiac ICU. The study received institutional review board approval and was in concordance with the Helsinki Declaration.
The primary aim of the study was to characterise the effect of amiodarone infusion on heart rate. Secondary aims of the study were to characterise the effect of amiodarone infusion on systolic blood pressure, diastolic blood pressure, arterial saturation by pulse oximetry, cerebral near-infrared spectroscopy values, renal near-infrared spectroscopy values, and central venous pressure.
Clinical variables of interest
Patient characteristics collected include patient age at the time of the initiation of amiodarone, weight, principal cardiac diagnosis, and principal electrophysiologic diagnosis. Whether the patients were postoperative was also recorded.
Clinical variables for which data were collected include heart rate, systolic blood pressure, diastolic blood pressure, arterial saturation by pulse oximetry, cerebral near-infrared spectroscopy values, renal near-infrared spectroscopy values, central venous pressure, and vasoactive inotrope score. These data were collected at the following time points: baseline (immediately prior to starting amiodarone infusion) as well as 1 hour, 4 hours, 8 hours, 12 hours, 24 hours, 36 hours, and 48 hours after initiation of the amiodarone infusion.
Systolic and diastolic blood pressure were recorded in mmHg using arterial lines if available. A majority of arterial lines utilised at the institution are femoral arterial lines. If blood pressure data from an arterial line were not available, then data obtained from a blood pressure cuff were utilised. Central venous pressure was recorded in mmHg from what were primarily femoral lines. Arterial saturation was collected in percent using pulse oximetry. Near-infrared spectroscopy data were collected in percent using ForeSight (Edwards Lifesciences) tissue oximetry sensors placed on the forehead and the flank.
Calculated variables were renal oxygen extraction, which was calculated as ((arterial saturation – renal near-infrared spectroscopy)/arterial saturation × 100). Vasoactive inotrope score was calculated as previously described.
Patient selection
All paediatric patients (under 18 years of age) cared for in the cardiac ICU between 1 January 2013 and 31 December 2020 were eligible for inclusion in this study. Those on mechanical circulatory support were excluded from the study. If a patient received more than one instance of an amiodarone infusion, only data from the first infusion were included. Those on an amiodarone infusion for less than 1 hour were also excluded. In addition, those who received a bolus dose preceding the infusion were excluded. This was done as the aim of the study was to identify the effect of an amiodarone infusion.
Statistical analyses
Univariable analyses were conducted to characterise the change in the clinical variables collected across the study period. An analysis of variance was done for each clinical variable of interest across the time point.
To evaluate whether the absence of haemodynamic collapse in our cohort was statistically significant, we performed a one-sample binomial test. This assessed whether the observed rate of haemodynamic collapse was significantly different from a previously published rate of 10%.
Next, a segmented regression analysis was conducted to help characterise the relationship between heart rate and cumulative amiodarone dose. This was done by initially determining a breakpoint using the slope in the graph and then fitting a regression model before and after the breakpoint.
Next, a time series regression analysis was conducted to model heart rate with the following independent variables: amiodarone dosage, vasoactive inotrope score, patient weight, patient age, primary cardiac diagnosis, primary electrophysiologic diagnosis, and postoperative status. As this was a time series regression, time was also included as an independent variable to allow for assessing the effect of time. Goodness of fit was evaluated using the R-squared statistic. Independence, normality of residuals, and homoscedasticity were evaluated. Time series analyses include baseline data prior to an intervention and then use this as time control. The individual time points then become an independent variable. As the amiodarone dose was then included as a separate independent variable, this helps to isolate some of the natural time effect. Thus, using time series regression analyses allows for quantifying the independent effect of time.
Finally, a time series regression analysis was conducted to model renal oxygen extraction with the following independent variables: amiodarone dosage, vasoactive inotrope score, patient weight, patient age, primary cardiac diagnosis, primary electrophysiology diagnosis, and postoperative status. As this was a time series regression, time was also included as an independent variable to allow for assessing the effect of time.
All statistical analyses were conducted using the R-statistical package. A p-value of less than 0.05 was considered statistically significant.
Results
Cohort characteristics
A total of 87 unique amiodarone infusions across 87 unique patients were included in the final analyses. Of these 87, all of them were on an amiodarone infusion for at least 1 hour, 84 (97%) for at least 2 hours, 78 (90%) for at least 12 hours, 66 (76%) for at least 24 hours, 60 (69%) for at least 36 hours, and 46 (53%) for at least 48 hours.
Patient characteristics are outlined in Table 1. The most frequent principal cardiac diagnosis was atrioventricular septal defect in 16% followed by hypoplastic left heart syndrome in 13% and transposition in 10%. The most frequent principal electrophysiology diagnosis was re-entrant supraventricular tachycardia in 55%, followed by ectopic atrial tachycardia in 16%, and junctional ectopic tachycardia in 14%. The remainder of the cardiac and electrophysiologic diagnoses are outlined in Table 1.
Table 1. Cohort characteristics. Continuous variables are presented as median and interquartile range. Categorical variables are presented as absolute count and percentage
Of the 87 patients, 46 (53%) had sinus rhythm restored by 12 hours after the amiodarone infusion was initiated, 83 (95%) by 24 hours, and then all (100%) by 36 hours. Of the 87 patients, 16 (17%) had a lactate greater than 2.5 or a renal oxygen extraction greater than 40%.
Of these 87 patients, 3 patients died during the inpatient admission (3.4%), but none of these deaths occurred during amiodarone infusion.
Amiodarone dose
The average amiodarone infusion dose was 14 mcg/kg/min at 1 hour, 11 mcg/kg/min at 8 hours, 9 mcg/kg/min at 12 hours, 7 mcg/kg/min at 24 hours, 6 mcg/kg/min at 36 hours, and 4 mcg/kg/min at 48 hours. None of the patients received an amiodarone bolus. Some were started on a higher dose of amiodarone infusion to help mimic a bolus, and this is captured by the dosing capture.
Changes in clinical parameters, univariable analyses
Heart rate demonstrated significant change from baseline at the following time points: 1 hour, 4 hours, 8 hours, 12 hours, 24 hours, 36 hours, and 48 hours. Heart rate tended to decrease with time. Vasoactive inotrope demonstrated significant change from baseline at the following time points: 24 hours, 36 hours, and 48 hours (Figure 1). Vasoactive inotrope score tended to decrease with time (Figure 2). Serum lactate level demonstrated significant change from baseline at the following time points: 4 hours, 8 hours, 12 hours, 24 hours, 36 hours, and 48 hours (Figure 3).
Figure 1. Graph of heart over time.
Figure 2. Graph of vasoactive inotrope score over time.
Figure 3. Graph of lactate over time.
Systolic blood pressure, diastolic blood pressure, renal near-infrared spectroscopy, and renal oxygen extraction did not change significantly over the study period. AST and ALT also demonstrated no significant change over the study period (Figures 4 and 5).
Figure 4. Graph of systolic blood pressure over time.
Figure 5. Graph of renal near infrared spectroscopy over time.
There were no cardiac arrests in any of the patients during the study period. There was no haemodynamic collapse in any of these patients (p < 0.01).
Cumulative amiodarone dose and heart rate
A segmented regression demonstrated that while heart rate decreased with increasing cumulative amiodarone dose, there was a greater decrease in heart rate under a cumulative dose of 30,000 mcg/kg. Prior to and up to 30,000 mcg/kg of cumulative amiodarone, there was a 3 beats per minute decrease in heart rate for every 1000 mcg/kg of amiodarone. After 30,000 mcg/kg of cumulative amiodarone, there was less than a 0.3 beats per minute decrease in heart rate for every 1000 mcg/kg of amiodarone (Figure 6).
Figure 6. Graph of heart rate and cumulative amiodarone dose over time.
Amiodarone dose and heart rate, regression analyses controlling for other variables
A time series regression to model heart rate resulted in a model with an R-squared of 0.568, indicating that 56.8% of the variability in heart rate was explained by the model. A 1 mcg/kg/min increase in amiodarone infusion dose was independently associated with a 0.75 beats per minute decrease in heart rate. Each hour of time was independently associated with a 0.19 beats per minute decrease in heart rate or the equivalent of 4 beats per minute over a single day. Every 1-unit increase in the vasoactive inotrope score was independently associated with a 0.87 beats per minute increase in heart rate. A 1 kg increase in weight was independently associated with a 0.56 beats per minute decrease in heart rate. A principal electrophysiologic diagnosis of ventricular tachycardia was associated with a greater decrease in heart rate associated with amiodarone when compared to other electrophysiologic diagnoses. Being postoperative was not independently associated with heart rate.
Amiodarone dose and renal oxygen extraction, regression analyses controlling for other variables
A time series regression to model renal oxygen extraction resulted in a model with an R-squared of 0.213, indicating that 21.3% of the variability in renal oxygen extraction was explained by the model. A 1 mcg/kg/min increase in the amiodarone infusion was independently associated with a 0.02 increase in renal oxygen extraction. Each hour of time was independently associated with a 0.41 decrease in renal oxygen extraction. Being postoperative was independently associated with a 4 decrease in renal oxygen extraction. Vasoactive inotrope score, weight, and systolic blood pressure did not have a significant independent association with renal oxygen extraction.
Discussion
This study demonstrates the acute clinical effects of an amiodarone infusion in children cared for in the cardiac ICU. Amiodarone has a significant and independent effect on heart rate while only minimally impacting renal oxygen extraction. Amiodarone is associated with a maximal decrease in heart rate until a cumulative dose of 30,000 mcg/kg and seems to lead to most decrease in heart rate when used in the setting of ventricular tachycardia. No significant change in liver enzymes was observed in the first 48 hours of amiodarone infusion.
This study characterises the acute effects of an amiodarone infusion on clinical parameters, for which data have otherwise been sparse. Reference Haas and Camphausen18 These are valuable data as amiodarone is used fairly frequently in cardiac ICUs across the world. It has been associated with haemodynamic collapse and even cardiac arrest in previous reports. Reference McGovern, Garan and Ruskin15–Reference Ng, Hampson Evans and Murdoch17 Thus, a thorough investigation of amiodarone’s acute effects seemed warranted.
Amiodarone has been associated with hypotension, thyroid dysfunction, and hepatic dysfunction when administered orally or intravenously. Amiodarone has also been associated with cardiovascular collapse, specifically with higher intravenous bolus doses. Many of these adverse effects of amiodarone are dose-dependent. It is also worth noting that amiodarone has also been associated with a pro-arrhythmic effect. Reference Celiker, Kocak, Lenk, Alehan and Ozme11–Reference Haas and Camphausen18
While other studies have associated amiodarone with adverse haemodynamic effects, the current study did not demonstrate such effects. Reference McGovern, Garan and Ruskin15–Reference Ng, Hampson Evans and Murdoch17 This may be linked to the use of high-dose amiodarone boluses in some centres, historically. The institution for the current study does not utilise high-dose amiodarone boluses.
Some of the previously described negative effects are from amiodarone itself, but some are also attributed to the solvent utilised for intravenous formulations. Intravenous formulations of amiodarone utilise polysorbate 80 or benzyl alcohol. These can cause significant vasodilation and a subsequent decrease in systemic vascular resistance.
Due to its aforementioned mechanism of action, amiodarone has been demonstrated to be effective for a variety of arrhythmias. The current study found that heart rate was most decreased in the setting of ventricular tachycardia.
The current study offers additional data regarding the acute effects of amiodarone. The cohort included patients who were postoperative and those who were not as well, thus allowing for characterisation of the effects of amiodarone between the two groups. Similarly, the cohort included a variety of cardiac diagnoses and ages. The use of time regression analyses allowed for the characterisation of the effect of time on heart rate and renal oxygen extraction in the regression analyses.
While the study has strengths, it is not without its limitations. This is a single-centre study, and so centre-specific practices could impact the associations. Additionally, the study is retrospective in nature, and the clinical data were collected from the electronic medical record as recorded by bedside staff. Heart rate is inherently modulated by several factors and not just amiodarone alone. One of these that could not be well controlled for was the difference in tachycardia cycle lengths, which alters the difference in the heart rate during tachycardia versus baseline. We were unable to account for this, specifically. Additionally, sedation can also impact heart rate. Due to the number of agents used for this purpose, a combination of continuous sedation along with bolus doses, and charting considerations, these were not data that were practical to collect in a truly meaningful way. Additionally, it is not routine practice in the unit to alter sedation by large degrees within 24 hours. Generally, dexmedetomidine and fentanyl are utilised in the unit, and dexmedetomidine is not weaned until rhythm control is achieved. This is not simply a statistical limitation but something experienced during clinical decision making and thus reflects a very practical and natural clinical dilemma. It has shaped the unit practice on how analgesic agents, specifically, dexmedetomidine, are titrated while on infusions such as amiodarone. Other limitations include a relatively small sample size and the lack of a control group.
Conclusion
Amiodarone infusions are associated with a significant decrease in heart rate without greatly impacting oxygen delivery, blood pressure requiring increased vasoactive support, or transaminase levels in the first 48 hours after amiodarone initiation. Heart rate decreases most until a cumulative dose of 30,000 mcg/kg, and heart rate decrease is most pronounced in those with ventricular tachycardia.
Open access
