Phlebitis in Intravenous Amiodarone Administration: Incidence and Contributing Factors

We searched the following databases for relevant articles published before February 2016: Cumulative Index to Nursing and Allied Health Literature, Cochrane Library, MEDLINE, Embase, Web of Science, and gray databases (Bielefeld, Lenus, EUGrey, RIAN, and DART). We searched for the keywords phlebitis, thrombophlebitis, or extravasation as a group and then combined with the keywords amiodarone or Cordarone. We reviewed all gray literature for potential relevance to avoid reporting bias.^58,59  Gray literature is literature not readily available to the public. We contacted 2 experts^13,14  regarding their poster presentations on amiodarone-induced phlebitis. Unfortunately, both confirmed that their research was never published. However, 1 expert¹⁴  provided via email relevant statistics pertaining to amiodarone-induced phlebitis that decreased with the implementation of nursing guidelines.

We screened the titles and abstracts of all articles retrieved by the search for relevance. We included English-language primary research reports of patients with cardiac conditions (surgical or medical) who had received peripherally administered intravenous amiodarone. We excluded studies of pediatric or nonhuman patients, secondary research reports (literature reviews and letters), and studies in which patients received amiodarone orally or via a central venous catheter.

In the studies that met the inclusion criteria, we used review management software⁶⁰  to assess the strength of association or nonindependence between phlebitis rates and its contributing factors, where appropriate. Dichotomous data were summarized with the odds ratio (OR).^60–62  An overall meta-analysis was not appropriate because of the heterogeneity of the included studies.⁶³  Studies that incorporated the same contributing factors were grouped together for analysis. Not all of the studies were randomized controlled trials, so rather than analyzing data within studies (ie, comparing interventions against controls), we compared contributing factors between studies. For instance, we compared the results of studies that did not use nursing guidelines with the results of studies that used guidelines. Likewise, we compared results of studies in which in-line filters were used with results of studies in which no filters were used. We also analyzed infusion concentration, duration, rate, and total dose between studies.

We appraised study quality using the Evidence-based Librarianship Critical Appraisal checklist.⁶⁴  This checklist employs questions with possible responses of “yes,” “no,” “unclear,” or “not applicable” to assess the external and internal validity of each section of the study process and to provide an overall score for each study. If “yes” responses are 75% or more of total responses, or if the sum of “no” and “unclear” responses is 25% or less of total responses, then the study is considered valid.⁶⁴ 

To avoid reporting bias and overrepresentation of positive or significant results,^65–69  we searched for unpublished data, but none were available. Additionally, we excluded 5 case studies^9,52–55  because they reported only extreme or unusual cases of phlebitis and were not representative of the broader population.⁶¹  Attrition bias was less than 1% and therefore did not affect our overall results.⁶⁶  Language bias was unavoidable because we included only English-language studies (no translation services were available). We included 2 systematic reviews. We reviewed 1 of these systematic reviews and appraised its quality with the PRISMA 2009 checklist.⁷⁰  We extracted 3 studies from the other systematic review and reviewed them individually; the other studies within that systematic review did not focus on phlebitis.

Our search identified 216 records. We found 2 additional studies in the reference lists of 2 literature reviews retrieved during the screening process. Seven studies linked from the Cochrane database proved to be not applicable. After removing duplicates, 192 entries remained; we screened these by reviewing their titles and abstracts. We excluded 174 articles: 10 were not in English, 97 were not applicable, 53 were not primary studies, and 14 were based on nonhuman studies or pediatric populations. We reviewed the full texts of the remaining 18 articles. We excluded 5 case studies, 1 systematic review, and 1 study that did not confirm phlebitis rate, retaining 11 studies in our systematic review. We found 9 more articles by screening the reference lists of the included studies, yielding a total of 20 included articles.

This systematic review included 5 retrospective studies,^3,6,29,33,37  1 randomized controlled trial,³⁸  and 1 descriptive study²⁵  in which amiodarone-induced phlebitis was the primary outcome. In the remaining studies (3 prospective studies,^34,42,45  1 systematic review,³⁵  and 9 randomized controlled trials^{36,39–41,43,44,46–48} ), amiodarone-induced phlebitis was a secondary outcome.

Included studies were carried out in the United States,^{3,6,25,29,37,40}  Europe,^{34,36,39,41–48}  Brazil,³³  and Iran³⁸  in various critical care, telemetry unit, and emergency department settings. Patients included in the studies had cardiac conditions (surgical or medical) that required intravenous amiodarone administration. Sample sizes varied. The systematic review conducted by Hilleman and Spinler³⁵  encompassed 550 patients. The largest study, by Halonen et al,⁴⁸  included 316 patients. The study by Boyce and Yee²⁵  was the smallest, with 12 patients. The mean (SD) number of patients in the included studies was 88 (107).

Our systematic review encompassed 2114 patients with an overall phlebitis rate of 14%. Phlebitis rates in the included studies ranged from 0% to 85% (Table 2). We grouped together studies that focused on the same contributing factor to analyze how these factors affected phlebitis rates.

Phlebitis rates were higher with amiodarone concentrations of 1.8 mg/mL or greater^{3,25,29,33,37,38}  than with an amiodarone concentration of 1.2 mg/mL^6,41  (OR, 0.09; 95% CI, 0.05-0.18; P < .001; Table 3). The 1.8 mg/mL or greater concentration was used in 7 studies; the 1.2 mg/mL concentration was used in 2 studies. The study by Mowry and Hartman⁶  was conducted in 3 parts. The 1.8 mg/mL concentration was used in parts A and C; the 1.2 mg/mL concentration was used in part B. The phlebitis rate in part B was significantly lower than in parts A and C.

The total dose of intravenous amiodarone administered to patients in the studies ranged from 0.25 g to 4.8 g. For doses of 1 g and higher, increasing doses were not associated with increasing phlebitis rates (Table 4). For example, the phlebitis rate in patients receiving a dose of 4.8 g in 1 study⁴⁰  was lower than phlebitis rates of patients receiving doses of only 1.05 g in other studies.^25,29,38,39  However, doses of 0.45 g or less^40,45,46  were associated with lower phlebitis rates than were doses greater than 1 g (OR, 0.02; 95% CI, 0.00-0.16; P = .002).^{25,29,34–39,42–44,47,48}  Norton et al³⁷  also showed that initial phlebitis rates were considerably lower when patients received an initial dose of less than 1 g of intravenous amiodarone, but once the total dose reached 3 g, phlebitis rates increased significantly by 40% (P < .001). In a 2-part study by Kowey et al,⁴⁰  the total dose in part A was 4820 mg; the total dose in part B was only 246.35 mg. Part A participants had a phlebitis rate of 2.86%, whereas part B participants had a phlebitis rate of 0%.

Intravenous amiodarone infusion rates ranged from 0.1 mg/min to 2 mg/min (Table 5). Infusion rate and phlebitis rate were not correlated. For instance, participants in the study by Cotter et al⁴³  received an infusion rate of 2 mg/min and had a lower phlebitis rate (16%) than participants in some studies with infusion rates of less than 0.75 mg/min.^{29,34,36,38,39} 

Infusion duration ranged from 1-minute boluses to 48-hour continuous infusions. We found no overall relationship between infusion duration and phlebitis rate. For example, phlebitis rates in some studies with 24-hour infusions^{29,34,36–39,42–44}  were higher than in studies with 48-hour infusions.^40,48  An exception was the study by Norton et al,³⁷  in which phlebitis significantly increased as infusion duration increased (P = .03). Studies in which amiodarone was administered as a bolus^45,46  yielded significantly lower phlebitis rates (< 2%) than studies in which amiodarone was administered as continuous infusions^{6,25,29,34–40,42–44,47,48}  (OR, 0.05; 95% CI, 0.01-0.33; P = .002; Figure 1). Additionally, in the study by Norton et al,³⁷  phlebitis rates were lower when amiodarone was administered in boluses (P < .05).

The absence of nursing guidelines increased phlebitis rates (OR, 0.13; 95% CI, 0.09-0.20; P < .001). Studies in which guidelines were implemented^6,29  had phlebitis rates ranging from 5.8% to 38%. Studies with no nursing guidelines^25,29,33,37  had phlebitis rates ranging from 40% to 85% (Table 6). The Spiering²⁹  study was divided into 2 parts. Nursing guidelines were not used in part A but were used in part B. The phlebitis rate was 85% in part A and fell to 38% in part B after guidelines were introduced.

Studies in which in-line filters were used^6,25,29  had phlebitis rates ranging from 5.8% to 67%, whereas those in which no filters were used^3,29,37  had phlebitis rates ranging from 13.9% to 85%. Phlebitis rates were significantly higher in the absence of filter use (OR, 0.23; 95% CI, 0.15-0.34; P < .001; Figure 2).

Our secondary outcome was phlebitis severity measured with the Infusion Nurses Society phlebitis scale.²⁰  Seven studies reported phlebitis severity, and some used different scales.^25,38  To facilitate analysis, we converted all of these to the Infusion Nurses Society scale.²⁰  Other studies^{34,42,45–46}  did not specify a phlebitis scale or grade but did report phlebitis symptoms, so we graded phlebitis on the basis of the descriptions (Table 1). The most common phlebitis grade overall was 0 (no phlebitis), followed by grades 1, 2, 3, and 4. Spiering²⁹  indicated that phlebitis severity decreased when nursing guidelines were implemented, and only 1 episode of grade 1 phlebitis was reported in the 2 studies of bolus administration.^45,46 

We appraised the quality of the systematic review³⁵  by using the PRISMA 2009 checklist.⁷⁰  Most steps were completed, but randomization details and the data collection process were not described and the data extraction tool was not validated. The remaining studies were valid and of high quality, with overall Evidence-based Librarianship Checklist scores ranging from 79% to 96%. Authors of 14 studies^{3,33,34,36,39–48}  failed to clarify whether their data collection tools were validated. Testing of data collection tools is essential because it confirms whether the appropriate data are collected in a valid and reliable manner.⁷¹  Martinho and Rodrigues³³  used a data collection tool based on a semistructured script, which is open to interpretation—inconsistencies that could introduce bias.⁶¹  Three studies^25,37,38  had robust data collection tools; their phlebitis scales were validated, piloted, and approved by nursing experts. We assessed the studies for detection, selection, performance, reporting, and attrition bias.

Our systematic review included 2 double-blinded studies^38,40  and 3 single-blinded studies,^36,41,47  which helped eliminate detection bias. However, the remaining studies were not blinded. Authors of 2 studies^6,25  mentioned that staff members were involved in data collection but did not confirm whether they were also involved in caring for the participants.

Five retrospective,^3,6,29,33,37  3 prospective,^34,42,45  and 1 descriptive²⁵  study used nonprobability sampling techniques, which introduce selection bias.

Two double-blinded studies^38,40  and 3 single-blinded studies^36,41,47  helped eliminate performance bias. Hofmann et al⁴⁶  confirmed that randomization was performed by using sealed envelopes but did not state whether the study was single- or double-blinded. Likewise, authors of 3 randomized controlled trials^39,43,44  did not confirm whether their studies were blinded. The remaining studies were not blinded.3,6,25,29,33,34,37,42,45,48

All included studies reported data on the outcomes under investigation.^{3,6,25,29,33–48} 

Shützenberger et al³⁹  withdrew 1 patient because of an adverse event. Kowey et al⁴⁰  stated that 14% of participants died, and Halonen et al⁴⁸  stated that 1 person died during treatment. In addition, 10 studies3,6,25,29,33,37,38,41,46,47 did not report attrition. Six studies confirmed that there was no attrition.^{34,36,42–45} 

The wide range in phlebitis rates among studies was partly due to the different contributing factors. Identifying the contributing factors that affected phlebitis rates was difficult because more than 1 of these factors were often present. For example, the phlebitis rate in the study by Kowey et al⁴⁰  was 0%. In that study both the amiodarone infusion rate (0.1 mg/min) and the total amiodarone dose (0.25 g) were low, so ascertaining which factor affected the phlebitis rate was challenging. However, among all of the included studies, total doses greater than 1 g were associated with increased phlebitis rates.^{25,29,34–39,42–44,47,48} 

Studies in which in-line filters were used^6,29  had lower phlebitis rates than did studies in which filters were not used. However, in these studies, nursing guidelines (which reduce phlebitis rates) were also used. It was therefore initially difficult to distinguish if nursing guidelines or filters improved phlebitis rates. In the study by Boyce and Yee,²⁵  filters but no nursing guidelines were used, and the authors stated that filters made no difference in reducing their phlebitis rates. Therefore, their high phlebitis rate (67%) could be due to the absence of nursing guidelines. The Boyce and Yee study²⁵  had a small sample size, inhibiting the ability to draw reliable conclusions. The Spiering²⁹  study clearly demonstrated the benefits of nursing guidelines. Even though filters were used in parts A and B of their study, the phlebitis rate dramatically decreased (from 85% to 38%) in part B, when nursing guidelines were introduced. On the other hand, Slim et al³  noted that after their study was completed, they introduced filters and their phlebitis rates decreased. The study by Martinho and Rodrigues³³  confirmed that the absence of nursing guidelines was an important extrinsic factor in increasing phlebitis rates. In summary, both nursing guidelines and filters could individually and collectively reduce phlebitis rates.

Different study designs meant that a meta-analysis was impossible because studies were not homogeneous.⁶⁷  Despite this heterogeneity, a subanalysis of groups of studies focusing on the same contributing factors was possible. These results should be viewed with caution because the wide range in phlebitis rates could also be due to different practice standards among the studies. The study by Bagheri-Nesami et al³⁸  had a high phlebitis rate (58%) even though the infusion rate was only 0.62 mg/min. This high phlebitis rate may have been due to the infusion site. Patients in this study received amiodarone infusions via veins in the hand, which tend to be fragile and prone to phlebitis.²⁸  Norton et al³⁷  also stated that some peripheral catheters were placed in hand veins, which could be a reason for their 40% phlebitis rate.

Peripheral venous catheter site and catheter size were not always noted, and a reasonable assumption is that these factors were inconsistent throughout studies. Spiering²⁹  suggested that 22-gauge catheters yielded the lowest phlebitis rates, yet Boyce and Yee²⁵  found that catheter size and peripheral venous catheter site made no difference to phlebitis rates. Dedicated catheter use was also inconsistent within the studies, potentially affecting phlebitis rates. Dedicated catheters were used in the studies by Spiering²⁹  and Mowry and Hartman,⁶  but the remaining studies were not consistent in this regard. Martinho and Rodrigues³³  highlighted that the absence of dedicated catheter use in their study increased phlebitis rates.

Different study designs yielded variation in phlebitis rates. The observation and descriptive studies introduced an element of observer bias. Spiering²⁹  and Mowry and Hartman⁶  stated that it was hard to distinguish if reduced phlebitis rates were due solely to the improvements made or due to heightened awareness during their studies. The 5 retrospective studies^3,6,29,33,37  introduced the risk of recall bias.⁶¹  Retrospective data can be incomplete or subjective because the data were not originally collected for research purposes.⁶¹  Norton et al³⁷  and Boyce and Yee²⁵  highlighted retrospective data collection as a study limitation. In the 3 prospective studies,^34,42,45  data were collected as they became available, yielding more meaningful data specific to the research question.⁶¹ 

Different sampling techniques introduced an element of selection, performance, and detection bias, which is likely to have contributed to the variation in phlebitis rates.^61,71  The wide range in phlebitis rates may also have arisen because some studies’ sample sizes were too small to draw conclusions in isolation.^25,34,39  However, despite all of these variations, the studies included in our systematic review had excellent internal validity. Pooling the studies for analysis increased the overall power of evidence.^56,62  Furthermore, the demographic profile of patients within the included studies was homogeneous. Most patients were older than 60 years old. Martinho and Rodrigues³³  suggested that patient age is an intrinsic cause of phlebitis; older patients have fragile veins.

Nurses should be aware of the potential for phlebitis, including the specific phlebitis rates for their clinical areas, and be mindful of the associated contributing factors. Understanding the contributing factors that could potentially increase phlebitis risk can prompt nurses to be vigilant and take immediate action to avoid further progression.^21,25  Frequent routine assessment and documentation of phlebitis grades and nursing interventions are important.^21,25,31  Introducing nursing guidelines and in-line filters may reduce the phlebitis rate, potentially reducing costs and shortening hospital stays.^3,25 

Our review revealed a wide range in phlebitis rates (0%-85%) that was apparently dependent on certain contributing factors. An increase in amiodarone concentration was associated with an increase in phlebitis rate. Total intravenous amiodarone doses of 0.45 g or less were associated with lower phlebitis rates than were doses greater than 1 g. We did not identify a relationship between phlebitis rate and infusion rate or infusion duration. However, the 2 studies in which amiodarone was administered as an intravenous bolus^45,46  reported extremely low phlebitis rates and minimal severity compared with studies in which amiodarone was administered as a continuous infusion.^{6,25,29,34–40,42–44,47,48}  The most common phlebitis severity grade was 0 (no phlebitis), followed by grades 1, 2, 3, and 4. The results of this systematic review suggest that using nursing guidelines and in-line filters reduces phlebitis rates, with 1 study also showing a reduction in severity. The evidence is not robust, however, because it is based on 2 retrospective studies and 1 descriptive study.^6,25,29  Increased awareness, monitoring, and education during the course of these studies probably also contributed to reduced phlebitis rates.

The wide variations in phlebitis rates were due not only to contributing factors but also to heterogeneity among studies in elements such as peripheral venous catheter location, catheter size, and dedicated catheter use. The variations in methodological approach limit the overall certainty of the evidence, highlighting the need for further research, preferably a large, well-designed randomized controlled trial. Nonetheless, this systematic review shows that educating nurses about potential contributing factors for amiodarone-induced phlebitis and implementing nursing guidelines and in-line filters will help reduce phlebitis rates. Nursing guidelines should also emphasize that higher amiodarone infusion concentrations and total administered doses require increased surveillance.