Adverse events associated with administration of vasopressor medications through a peripheral intravenous catheter: a systematic review and meta-analysis

Intravenous vasopressors are commonly used to augment systemic blood pressure [1]. Current practice, especially within intensive care units (ICUs), is to administer these medications through a central venous catheter (CVC) [2, 3]. The reason for preferential use of CVCs rather than the more ubiquitous peripheral intravenous (PIV) catheters is due to the long-standing concern regarding vasopressor extravasation outside the PIV and potential for tissue damage (e.g. infiltration, ischemia and necrosis) [1, 4]. However, insertion and maintenance of a CVC is not without risk. A recent study examining CVCs in adults found that up to 2.1% of patients experienced significant mechanical complications such as pneumothorax requiring intervention, 0.5–1.4% experienced symptomatic deep-vein thrombosis directly attributable to the CVC, and another 0.5–1.4% experienced bloodstream infection [5]. Current estimates are similar, if not higher among children [6, 7]. In addition, CVC insertion can be time consuming and is associated with patient discomfort. For many patients, CVC insertion is a necessary intervention. However, for those that do not necessarily require a CVC, the adverse event profile associated with PIV vasopressor administration is less clear.

A previous systematic review of predominantly case reports and case series found that among 263 patients who received vasopressors through a PIV there were 318 adverse events, of which 86 were extravasation with no injury and 179 were skin necrosis [4]. Local tissue injury tended to occur in patients who had vasopressors infused for longer durations of time (≥ 24 h) through a PIV distal to the popliteal or antecubital fossa [4]. However, case series and case reports are at high risk for reporting bias [8], therefore the rate of adverse events associated with PIV vasopressor administration could not be estimated. In addition, larger cohort studies examining adverse events associated with PIV vasopressor administration in adults [2, 3, 9‐11] have since been published as well as multiple recent studies in the child population [12‐14]. We used systematic review and meta-analysis methodology to examine the incidence of adverse events associated with vasopressor administration through PIVs in patients of any age cared for in any acute care setting.

This systematic review and meta-analysis of adverse events associated with PIV vasopressor administration among more than 16,000 patients from 23 studies found that in both adults and children, the risk of local anatomic adverse events is low and comparable to the rates of potentially more serious complications associated with CVC insertion and maintenance. The pooled incidence of adverse events was 1.8% (95% CI 0.1–4.8%) among adults and 3.3% (95% CI 0.0–10.1%) among children. In addition to the generally low incidence, 19 of the 23 studies reported either no adverse events or only mild local tissue reactions like infiltration or extravasation, with only one study [33] reporting ischemia and skin necrosis. Importantly, a large number of studies were at some to high risk of bias, and there was significant statistical and clinical inter-study heterogeneity. In addition, although most studies reported PIV vasopressor administration that follows contemporary practice (i.e. PIVs 20 gauge or larger in veins proximal to the hands), no study directly compared the incidence of local anatomical adverse events associated with PIV compared to CVC-administered vasopressors. Therefore, the results of this meta-analysis should be interpreted as supporting the hypothesis that for many patients, PIV vasopressor administration may be safe, whilst also highlighting the need for additional high-quality research.

Our results are consistent with prior literature examining the risks associated with PIV vasopressor administration. A prior systematic review that could not estimate incidence due to the inclusion of case series and case reports suggested adverse events appear associated with longer duration of infusion and distal PIV sites [4]. Generalization of these results to modern practice regarding vasopressor and PIV management was limited by the fact that most included articles were published before 1970 [4]. More recently, Tian et al. examined adverse events associated with PIV vasopressor administration reported by seven cohort studies published between 2010 and 2018 that included 1362 adult patients in shock [38]. Most included patients were admitted to ICUs. They reported a pooled extravasation rate of 3.4% (95% CI 2.5–4.6%) with none causing limb ischemia or tissue necrosis. We included 23 studies, screening all and including five of those in the systematic review of Tian et al. but also included an additional 18 studies in patients cared for outside ICUs in the ED and OR, as well as children. Our pooled adverse event rate for adults was similar to that of Tian et al. [38]. Our pooled rate for children confirms that similar to adults, the incidence of adverse events associated with PIV vasopressor administration is low. To our knowledge, this is the first meta-analysis of PIV vasopressor administration in children.

While the incidence of adverse events associated with PIV vasopressor administration is low, it is difficult to make clinical recommendations without direct comparisons to current practice through CVCs. Designing a study to make those comparisons faces several challenges. First is defining the primary outcome of interest, wherein the adverse events associated with vasopressor administration are different for PIVs and CVCs. Medication extravasation, local tissue ischemia or sudden loss of intravenous access resulting in severe hypotension are events that would be more commonly associated with vasopressors administered through a PIV. Similarly, pneumothorax, arterial puncture, and catheter-related bloodstream infection are more commonly associated with CVCs. In an RCT done by Ricard et al., comparing all-purpose initial access with a PIV (not just for vasopressor administration) to a CVC in 263 patients admitted to ICUs in France, major complications were more common in those initially treated with a PIV [39]. However, greater than five attempts to insert a PIV and pneumothorax owing to CVC insertion were counted equally as major adverse events, which arguably have differing clinical implications [40]. The second challenge in designing an RCT to examine PIV vasopressor administration is the low rate of adverse events associated with administration of vasopressors through either a PIV or CVC that would necessitate a large sample size. Third, is the interaction of the current research question comparing use of PIVs and CVCs to also include newer, more frequently utilized forms of venous access such as PICC and midline catheters, with a midline functioning more like a PIV and the PICC more like a CVC. At first glance these two forms of venous access seem more robust than a PIV and lower risk and discomfort to patients than the traditional CVC. However, recent research suggests that critically ill patients may be higher risk for PICC-related complications such as catheter-related and deep vein thrombus [41]. None of the studies included in this review examined patients managed with midline or PICC catheters. Therefore, the research question is likely even more complex than we’ve alluded to and any future prospective studies should consider including midline and PICC catheters in what is likely to be a complex RCT design. Fourth, and perhaps most challenging may be overcoming the engrained belief among clinicians [42] that vasopressor medications ought to be administered through a CVC and the effects of this belief on clinicians’ willingness to enroll patients into an RCT that potentially randomizes patients to vasopressor administration other than through a CVC.

Though our study highlights the need for additional high-quality research, taken together with existing guidelines, our results help inform current clinical practice with regard to PIV vasopressor administration. The most recent Surviving Sepsis Campaign Guidelines for management of adults with sepsis do not make explicit recommendations regarding CVC versus PIV vasopressor administration, though it is highlighted that patients no longer require assessment of specialized data obtained from the CVC such as central venous oxygen saturation as part of their initial resuscitation [43]. Pediatric sepsis guidelines state that if a CVC is not reasonably accessible, all vasoactive medications (including epinephrine and norepinephrine) can be given through a PIV or IO to avoid delays and to transfer the infusion to a CVC as soon as possible [44]. Our sensitivity analysis excluding the large study in operative patients [11] suggests that adverse events may occur more commonly in patients in ICUs, EDs and/or step-down units where vasopressor infusions are likely administered for longer periods of time to patients with higher illness severity, though the overall rate was still low. The wide variety of studies represented in this review, the majority of which report a low incidence of adverse events, suggest that, for at least short periods of time, PIV vasopressor administration is safe provided precautions exist to reduce the likelihood of adverse events. Such safeguards include institutional policies that place limitations on PIV size and location, infusion dose and duration, require frequent PIV checks to ensure patency and always maintaining a backup PIV in case of sudden loss of intravenous access [2, 3, 10, 40].

The findings of this study should be interpreted in the context of its strengths and limitations. It was conducted by rigorously following methodological guidelines for meta-analyses, and to our knowledge, is the first meta-analysis to examine the safety of PIV vasopressor administration in children, and the largest in adults. However, most included studies were single-centre, retrospective cohort studies, inherently at risk of reporting bias, and were generally at high/some risk of bias. Subgroup and meta-regression analyses suggest a reduction in heterogeneity among those studies at lower risk of bias, however the adverse event estimate remained similar. In addition, there was significant interstudy heterogeneity. Subgroup and stratified analyses examining factors known to contribute to clinical heterogeneity such as location of clinical care, PIV gauge, and vasopressor infusion duration were identified as factors contributing to the observed statistical heterogeneity. Owing to non-uniform reporting of data within the included studies we were unable to comment on contributions of vasopressor concentration and/or dose, two factors predicted to be important determinants of the safety of PIV vasopressor administration. Sensitivity analysis excluding a large study of operative patients by Pancaro et al. wherein data was primarily obtained from an electronic database (rather than manual chart review) and adverse event rates were reported to be lower [11], resulted in a marked reduction in heterogeneity among the residual studies in ICU, ED and step down unit patients. Based on its contribution to heterogeneity, combined with the fact that the study by Pancaro et al. was the only study to obtain adverse event data from a pre-existing electronic database wherein report of more minor adverse events such as extravasation, skin blanching, and mottling may be lower, argument could be made to exclude it from the main meta-analysis. However, we elected to keep it in the main meta-analysis and report its exclusion as a separate sensitivity analysis to be consistent with our original statistical analysis plan, as well as to reflect the true state of current literature. In addition, we employed a random effects model which more evenly distributes study weighting and prevents larger studies such as that of Pancaro et al. from having undue influence on pooled effect estimates. Asymmetry in vasopressor type among the included studies is another potential limitation. Norepinephrine was the most common and this likely reflects clinical practice in most jurisdictions, however, other frequently employed drugs such as epinephrine and vasopressin were underrepresented. Attempts to examine our results in subgroups defined by vasopressor type did not yield any significant differences in the incidence of adverse events, though the analysis was likely underpowered.

The incidence of adverse events associated with PIV vasopressor administration in adults and children is low. When adverse events did occur, they tended to be minor. Additional research is required to examine the effects of PIV location and size, vasopressor type, concentration, dose and duration, and patient characteristics on the safety of PIV vasopressor administration. However, in the meantime, these results suggest that with careful monitoring, administration of vasopressors through PIVs is safe.