Device associated complications in the intensive care unit
BMJ 2024; 386 doi: https://doi.org/10.1136/bmj-2023-077318 (Published 13 August 2024) Cite this as: BMJ 2024;386:e077318
- Roxana Hixson, assistant clinical professor,
- Kristin Schwab Jensen, assistant clinical professor,
- Kathryn H Melamed, assistant clinical professor,
- Nida Qadir, associate clinical professor
- Correspondence to: N Qadir NQadir{at}mednet.ucla.edu
Abstract
Invasive devices are routinely used in the care of critically ill patients. Although they are often essential components of patient care, devices such as intravascular catheters, endotracheal tubes, and ventilators are a common source of complications in the intensive care unit. Critical care practitioners who use these devices need to use strategies for risk reduction and understand approaches to management when adverse events occur. This review discusses the identification, prevention, and management of complications of vascular, airway, and mechanical support devices commonly used in the intensive care unit.
Introduction
Although mortality in the intensive care unit (ICU) has declined over time, ICU associated complications remain common, affecting up to a half of critically ill patients and resulting in increased mortality, longer hospital stays, and higher healthcare costs.1234 Although both patient related and healthcare associated factors contribute to adverse events, many complications are either direct or indirect consequences of ICU practices and are thus potentially avoidable. Invasive devices used in the ICU, although crucial for the management of critically ill patients, represent a common source of complications. The mainstay of avoiding device related adverse events consists of judicious use of devices and their timely removal. However, complications can occur even with vigilant use. In this narrative review, we focus on identifying, preventing, and managing complications associated with devices frequently used in the ICU.
Sources and selection criteria
We searched PubMed, Medline, and the Cochrane Database of Systematic Reviews for articles published from 2000 to 2023. We used the following search terms: “central venous catheters”, “arterial catheters”, “endotracheal tube”, “endotracheal intubation”, “tracheostomy”, “percutaneous dilatational tracheostomy”, and “extracorporeal membrane oxygenation”. Several keywords were combined with the term “complications” or with terms representing adverse events specific to particular devices. We reviewed relevant titles and abstracts and prioritized meta-analyses, systematic reviews, international guidelines, randomized controlled trials (RCTs), and large observational studies. In some areas, available data were limited to smaller observational studies or articles published before 2000; we included these when relevant. We excluded case reports and articles not in the English language.
Epidemiology
The insertion of vascular catheters, endotracheal airways, and extracorporeal support devices constitutes some of the most common invasive procedures in intensive care. Intravascular catheters, including central venous catheters (CVCs), arterial catheters, hemodialysis catheters, and pulmonary artery catheters, are inserted in up to three quarters of critically ill patients.5 The use of airway devices is similarly widespread, with more than 1.5 million endotracheal intubations occurring outside of the operating room in the US annually and up to 24% of mechanically ventilated patients going on to receive tracheostomies.67 The use of extracorporeal membrane oxygenation (ECMO) has also grown exponentially over the past decade, with notable rises seen during both the H1N1 influenza outbreak and the covid-19 pandemic.8 The frequency of device associated complications in their entirety and their impact on morbidity and mortality in the ICU are not well established. Most large epidemiologic studies focus on infectious complications; less is known about mechanical complications. The rates, risk factors, and outcomes for such complications vary according to the type of device.
Complications of vascular devices
Although well defined indications for vascular devices have not been clearly established in clinical practice guidelines, these devices are used for several purposes, including the administration of vasoactive drugs, hemodynamic monitoring, hemodialysis, or introduction of other devices such as transvenous pacemakers. Two of the most used vascular devices—CVCs and arterial catheters—are discussed here.
Central venous catheters
CVCs are placed in up to two thirds of patients in the ICU, with considerable center and geographic variability in use,910 and are typically inserted in the internal jugular, femoral, or subclavian veins.91112 Adverse events can occur either during or after insertion. Although data from large epidemiologic studies are lacking, rates reported in RCTs and available observational studies range from 2% to 15%, with the type and frequency of complication varying by site.1213 Factors influencing the incidence of insertion related complications include the proceduralist’s proficiency, number of punctures needed, cannulation site, and use of ultrasound.121415 Clinical practice guidelines recommend the use of real time ultrasonic guidance, with meta-analyses showing an associated improvement in first pass success rate, a decrease in the number of insertion attempts, and a reduction in the risk of insertion related complications.515161718 Even with the use of ultrasound, however, limited operator experience (defined as <100 catheterizations) increases the risk of mechanical complications (odds ratio 3.11, 95% confidence interval (CI) 1.64 to 5.77).19 Insertion related complications include vascular injury with resultant hematoma and/or blood loss, pneumothorax, and air embolism (fig 1).20 Post-insertion risks include thrombosis and infection, which can be mitigated by prompt removal of the CVC once it is no longer indicated.921 Although several steps can be taken to minimize the risk of individual complications, ultimately the most important risk reduction tactic is avoiding unnecessary CVC insertion. One of the most common uses for CVCs is vasopressor administration, but growing evidence shows that short term peripheral vasopressor infusion may be safe and feasible in select patients.222324 This information, as well as an assessment of patient specific risk factors for CVC associated complications, should be factored into decision making before catheter placement. We discuss individual complications of CVCs below. Notably, although many of the principles and complications that pertain to CVCs are also applicable to hemodialysis lines and pulmonary artery catheters, each of these devices has additional specific and nuanced complications that are outside the scope of this review.
Complications of central venous catheters (CVCs). CA=carotid artery; IJV=internal jugular vein
Vascular injury
Vascular injury can be venous or arterial. Venous damage can involve posterior wall puncture, common in hypovolemic patients and typically of minimal consequence unless adjacent structures are injured, or the more serious complication of venous wall laceration. Laceration can occur as a result of inadvertent trapping of the guide wire against the vessel wall followed by attempts to advance the catheter or dilator; the resulting injury can potentially extend beyond the target vein to include the vena cava, other intrathoracic vessels, or the right atrium.25 To avoid this complication, the guide wire should not be advanced when resistance is felt.
Similarly, arterial injury can consist of either simple arterial puncture or more significant arterial cannulation. Real time ultrasonography has decreased the incidence of arterial puncture, with one meta-analysis of 130 studies showing a decrease from 69 events to 14 events per 1000 catheters (risk ratio 0.20, 95% credible interval (CrI) 0.09 to 0.44). Nevertheless, prompt identification of arterial puncture remains crucial for preventing the graver consequences of arterial cannulation.15 Although identification is usually immediate via visualization of pulsatile blood flow, this may not be a reliable indicator in cases of severe shock. If the location of the needle or wire is unclear even after visualization with real time ultrasonography, a small single lumen catheter can be inserted over the guide wire and connected to a pressure transducer to ensure venous waveform before proceeding with dilation.25
When recognized, arterial puncture can typically be managed by removing the needle and applying pressure at the access site, although this may be more difficult at the subclavian position. Clinically significant bleeding (that is, needing blood transfusion or intervention), although rare, may occur, and consequent hematoma formation can potentially compress adjacent structures or lead to hemodynamic compromise, particularly in the femoral location.
If arterial dilation or cannulation occurs, complications can include life threatening bleeding, pseudoaneurysm formation, arterio-venous fistula development, and stroke.13 Management can consist of device removal with manual compression of the arterial system, endovascular intervention, or open surgical repair. Existing data comparing these approaches are limited, with the largest systematic review limited to 150 cases, but suggest that a strategy of manual external pressure is associated with substantially higher complication rates than is endovascular intervention or surgical repair.2627 Thus, leaving the CVC in place until prompt surgical or endovascular management can be obtained may be advisable.
Blood loss
Clinically significant bleeding is rare, with a reported incidence of 0.8% in a large multicenter cohort.20 Factors increasing the risk include arterial puncture, higher number of needle passes, and pre-procedural coagulopathy.2028 The role of prophylactic transfusion in the setting of coagulopathy is uncertain. RCT data on the administration of prophylactic plasma are lacking, and no clear transfusion threshold has been established.29 With regard to thrombocytopenia, one RCT of 373 CVC placements compared a strategy of prophylactic platelet transfusion with no transfusion in patients with a platelet count of 10-50×109/L and found that withholding platelet transfusion resulted in more serious CVC related bleeding events (11.9% v 4.8%; risk ratio 2.43, 95% CI 0.75 to 7.93).30 Nevertheless, the minimal platelet count for safe catheter insertion remains unknown.2830
Pneumothorax
Pneumothorax is an infrequent but serious complication of internal jugular and subclavian vein CVC insertion, occurring in about 1% of cases.31 Risk factors include subclavian vein placement, presence of underlying lung disease, mechanical ventilation, insertion under emergency circumstances, and multiple attempts at cannulation.32 A recent meta-analysis found that insertion under ultrasound guidance decreased the incidence of pneumothorax for both the internal jugular sites (0.4 v 18 events per 1000 catheters; risk ratio 0.02, 95% CrI 0.001 to 0.28) and subclavian sites (3 v 12 events per 1000 catheters; 0.26, 0.04 to 1.60).15 Although chest radiographs are routinely ordered to rule out pneumothorax and malpositioning, both can also be detected by bedside ultrasonography, which may be more readily available. In one meta-analysis, ultrasonography showed a pooled sensitivity of 0.82 (95% CI 0.77 to 0.86) and specificity of 0.98 (0.97 to 0.99) for catheter malpositioning; the calculated sensitivity and specificity for pneumothorax was 100%, and ultrasonography also decreased the time to confirmation.3233 Management of pneumothorax consists of either observation or tube thoracostomy placement, depending on the size of pneumothorax, the stability of the patient, and the need for positive pressure ventilation.
Air embolism
Air embolism is an extremely rare but potentially fatal complication, with an incidence of 0.13% reported in one large observational study.34 Air can be introduced into the venous system during CVC placement or subsequently when accessing or removing the catheter. Although air is introduced through the venous system, air emboli can pass through a septal defect or patent foramen ovale to the arterial circulation. Hemodynamic instability is common; other symptoms can include dyspnea, chest pain, and neurologic changes.35 Although evidence based studies are lacking, CVC insertion and removal in the Trendelenburg position is often suggested to decrease the risk of occurrence.36 Management is largely supportive, although placement of the patient in the left lateral decubitus position with the head down may help by facilitating movement of air into the right atrium and away from the pulmonary circulation. Additionally, the administration of supplemental oxygen (FiO2 1.0) may accelerate reabsorption of nitrogen gas from the air embolism into the bloodstream. Finally, on the basis of limited retrospective reports and expert opinion, hyperbaric oxygen has been suggested for patients with neurologic symptoms concerning for cerebral air embolism.353738 In these cases, prompt initiation of hyperbaric oxygen (that is, within six hours of the event) may confer a higher likelihood of benefit.39
Thrombosis
The reported rates of thrombotic complications vary from 0.9% to 14%.12404142 Risk factors may be patient specific (hypercoagulable state, malignancy, previous deep venous thrombosis, critical illness) or catheter related (large relative to vessel size, multi-lumen catheter, termination proximal to the superior vena cava). In one randomized trial of 3471 catheters comparing the incidence of deep venous thrombosis at different catheter insertion sites, higher rates of symptomatic thrombosis were reported with femoral catheters than with subclavian and internal jugular CVCs (hazard ratio 3.4 (95% CI 1.2 to 9.3) and 2.4 (1.1 to 5.4), respectively); no significant difference was seen between internal jugular and subclavian vein CVCs (hazard ratio 1.8, 0.6 to 4.9).12 Duplex ultrasonography is the diagnostic modality of choice, and management consists of anticoagulation for three months in patients without contraindications.43 Clinical practice guidelines suggest leaving CVCs that remain functional and are still needed in place, with close monitoring for worsening symptoms and removal if the catheter is occluded or no longer necessary or if symptoms worsen.43
Infection
More than 23 000 central line associated bloodstream infections (CLABSIs) were reported in the US in 2022.44 Several approaches have been shown to reduce the risk of infection and are recommended by clinical practice guidelines, including the use of checklists, aseptic technique with full barrier precautions during insertion, application of chlorhexidine impregnated dressings, routine dressing changes every seven days, and prompt removal of the CVC when it is no longer needed.59214546 One study of 4416 CVC insertions found that the use of a checklist decreased the CLABSI rate (3.8 v 5.9 CLABSIs per 1000 catheter days; P=0.001).47 The subclavian vein was associated with a lower rate of bloodstream infections than either the internal jugular or femoral vein in one randomized trial of 3471 catheters (hazard ratio 2.3 (95% CI 0.8 to 6.2) and 3.4 (1.0 to 11.1), respectively).12 However, subsequent meta-analyses have found comparable CLABSI rates across the three sites.1548 Consequently, although clinical practice guidelines suggest the use of the subclavian site when feasible, the risk and benefit of different insertion sites must be considered on an individual basis.521
With regard to duration of CVC use, data from a multicenter observational study of more than 1900 CVCs show that although the daily incidence of CLABSI does not increase with catheter duration, the cumulative risk does (daily hazard ratio of colonization 1.2% on day 5, 1.6% on day 10, and 1.4% on day 15; cumulative risk for catheter related infection 1.09 per 1000 days).49 Accordingly, although timely removal of unnecessary CVCs is crucial, routine line replacement (either at a new site or via guide wire exchange) after any specific number of catheter days is not recommended.592150 CLABSI management involves both systemic antimicrobial treatment and CVC removal whenever feasible.5 Guide wire exchange is not recommended as an alternative to catheter removal. Pooled data from 12 randomized trials suggest higher rates of catheter colonization (risk ratio 1.26, 95% CI 0.87 to 1.84) and catheter related bacteremia (1.72, 0.89 to 3.33) with guide wire exchange alone.51 Rather, placement at a different site is preferable if central access is still needed.50
Arterial catheters
Arterial catheters are placed in more than a third of all patients in the ICU and more than half of those with shock in the US, totaling eight million catheters placed annually.1052535455 Catheterization is most commonly performed in the radial and femoral arteries, although the brachial, axillary, and dorsalis pedis arteries are also used.56 Similarly to CVCs, additional data from large epidemiologic studies are needed to better understand the rate of complications. Although the reported incidence of serious adverse events resulting in lasting morbidity or harm is less than 1%,40 many of the existing data have been obtained from surgical and cardiac settings and pertain to use of an arterial line for cardiac catheterization and intraoperative monitoring.4056 Current and robust data describing complications from arterial catheter use in the intensive care unit are lacking. As such, many of the complication rates may be underestimated, as less is known about arterial catheter use in the ICU, which is often more prolonged than elsewhere.4056
Complications can include vascular injury, blood loss, neurologic injury, vascular occlusion, and infection (fig 2).5657 These events can occur at all sites, although incidence and risk factors for individual complications vary by location.57 Ultrasound guidance for insertion can improve the first pass success rate (risk ratio 1.39 (95% CI 1.21 to 1.59; P<0.001) in a meta-analysis of 19 RCTs) and reduce the risk of complications (0.51 (0.28 to 0.91) in a meta-analysis of 1422 patients).585960 As with CVCs, although means of minimizing complications exist, avoiding unnecessary use of arterial catheters remains of utmost importance. Although the Surviving Sepsis guidelines suggest the use of arterial catheters for patients with septic shock,61 this is acknowledged to be a weak recommendation based on very low certainty of evidence. Multiple observational studies have found that arterial lines are not associated with improved outcomes.626364 The largest of these studies, a propensity matched cohort study of more than 60 000 patients, found no association between use of an arterial line and in-hospital mortality (odds ratio 0.98, 95% CI 0.93 to 1.03; P=0.40).64 The use of arterial catheters should hence be judicious, with careful consideration of their necessity in each case.
Vascular injury
Vascular injury can occur with any arterial puncture and typically takes one of two forms: hematoma or pseudoaneurysm formation. Hematomas are usually small but can increase the risk of vascular occlusion.56 Pseudoaneurysm is more common at the femoral site (0.3%) than at the radial site (<0.1%).56 The use of a Seldinger technique or modified Seldinger technique rather than direct puncture can reduce the risk of vascular injury.65 Although the use of ultrasound guidance can decrease the number of punctures needed to access the vessel, data are inconclusive about its effect on risk of pseudoaneurysm or arterio-venous fistula formation.6667686970
Blood loss
Blood loss occurs most commonly and significantly at the femoral site. A misguided femoral artery puncture can result in retroperitoneal hematoma and potentially life threatening blood loss.56 Insertion of femoral arterial catheters in the common femoral artery distal to the inguinal ligament can help to ensure that the puncture site remains compressible should blood loss occur.69 Additionally, puncture of the superficial femoral artery or the femoral vein can result in hemorrhage or pseudoaneurysm formation.69 Real time visualization with ultrasonography may mitigate these risks,60666971 with one meta-analysis of 19 RCTs showing that its use decreased risk of hematoma formation (risk ratio 0.40, 95% CI 0.22 to 0.72; P=0.03).60 However, its effect on pseudoaneurysm formation is uncertain.71
Peripheral nerve injury
Peripheral nerve injury can result in significant functional impairment and leave lasting paresthesias and weakness. This is most often reported with radial artery placement and consequent damage to the median nerve, through direct puncture, pressure from hematoma formation, or prolonged wrist hyperextension.7273 Although arterial line related median nerve injury has been reported in 3.1% of patients in the intraoperative setting,74 data on its incidence at other sites are lacking. Moreover, additional studies are needed to better understand the rates and sequelae of nerve injury with the more prolonged arterial catheter use typically seen in critically ill patients.
Vascular occlusion
Vascular occlusion, an often temporary event that resolves with catheter removal, occurs in 12% of cases.56 Permanent ischemic damage is much more rare, with an incidence of 0.1%.56 The radial artery, although smaller in caliber than the femoral artery, has the benefit of being part of redundant collateral circulation to the hand through the ulnar and interosseous arteries via the palmar arches.7576 Although occlusion occurs more frequently in the radial artery than in the femoral artery (19.7% v 1.5%), distal perfusion is not as often compromised, and rates of permanent ischemic damage are lower (0.09% v 0.18%).56 Proposed methods for confirming collateral hand circulation include pulse oximetry of the thumb, along with the modified Allen’s test, whereby the operator occludes both the radial and ulnar arteries until the distal hand blanches, followed by release of the ulnar artery to ensure blood return within five to seven seconds.757677 However, studies suggest that these measures do not accurately predict the risk of distal hand ischemia with placement of radial arterial lines.757678 A prospective study of 203 patients undergoing radial arterial coronary angiography found that thumb lactate concentrations did not differ between patients with a normal, intermediate, or abnormal Allen’s test (mean 1.85 (standard deviation 0.93) mmol/L v 1.85 (0.66) mmol/L v 1.97 (0.71) mmol/L; P=0.59).79 As measures for mitigating the risk of digit ischemia are limited, placement of radial arterial lines in the non-dominant hand may be prudent when feasible.
Mechanisms of vascular occlusion include vasospasm, air embolism, and thrombus formation. Vasospasm is most often seen during initial catheter insertion, whereas air embolism can occur during line flushing and in rare cases can cause life threatening cerebral air embolism.80 Thrombus formation around the catheter is the most common cause of vascular occlusion and is related to the size of the catheter relative to the diameter of the vessel.56578182 A retrospective review of 62 626 arterial lines in the operative setting showed that the risk of complications, although low overall, increased as catheter size increased from 20 gauge to 18 gauge and 5 French (2.7 (95% CI 1.5 to 4.4) per 10 000 versus 17.2 (4.7 to 43.9) per 10 000 versus 9.4 (1.1 to 34.1) per 10 000).57 Recent data in ICU populations are lacking, although older studies have had similar findings.8182 Other risk factors for vascular occlusion include female sex, history of vasculitis or connective tissue disease, low cardiac output, use of high dose vasopressor therapy, multiple arterial punctures, and hematoma formation.375676 Risk of occlusion also increases after 72 hours of placement, so catheters should be discontinued as soon as clinically feasible.56 Continuous flush running through the arterial line may prevent thrombus formation, but data comparing heparinized solutions and continuous saline flushes are inconclusive.768384 Some studies suggest that use of papaverine infusion through an arterial catheter may prevent or relieve temporary obstruction from vasospasm in pediatric and neonatal populations, adults undergoing cardiac catheterization, and intraoperative cardiac surgical patients.8586878889 However, data are lacking to support its use in critically ill adults. Should distal ischemia occur, early recognition and timely vascular surgical consultation are crucial for mitigating the risk of permanent ischemic damage to the digits or limb.76
Infection
Infections of arterial catheters are infrequent, with reported rates of 0.25-0.72% for radial sites and 0.4-1.92% for femoral sites.5556769091 As with CVCs, recommended risk reduction measures include the use of sterile technique with full barrier precautions for catheter insertion, skin sterilization with an aseptic solution (most often a chlorhexidine based preparation, although povidone iodine or 70% alcohol may be used as alternatives when necessary), and prompt removal of the catheter when it is no longer needed.5555692 Although the application of chlorhexidine impregnated dressings is recommended for prevention of infection in central venous catheters, data are less robust for their use in arterial catheters.5909394 A systematic review and meta-analysis found two studies that reported specific infection rates associated with arterial catheters, with pooled data showing a relative risk of 0.35 (95% CI 0.13 to 0.91) in favor of the use of chlorhexidine dressings.55 Another meta-analysis of nine RCTs showed that the use of chlorhexidine impregnated dressings was associated with a relative risk of catheter associated bloodstream infection of 0.60 (95% CI 0.41 to 0.88; P=0.009) compared with other line dressings, and this finding held true for the three studies that included arterial lines.93 However, this study did not specifically quantify the rate of catheter associated bloodstream infection due to arterial lines alone.
Complications of airway devices
Airway devices, including endotracheal tubes and tracheostomies, are frequently used for patients with respiratory failure. Complications of these two devices are detailed below.
Endotracheal tubes
Complications of endotracheal tubes occur in nearly half of critically ill patients and can be anatomic or physiologic in nature.9596 Although most endotracheal tube related adverse events occur during initial endotracheal intubation, complications may arise at any point during their use, with prolonged intubation increasing the risk. In general, complications can be divided into insertion related events (table 1) and post-procedural events (fig 3).
Periprocedural complications of airway procedures
Post-procedural complications of endotracheal tubes and tracheostomies. A: suprastomal tracheal stenosis. B: tracheal stenosis at site of stoma. C: infrastomal tracheal stenosis. D: vocal cord paralysis. E: tracheomalacia. F: tracheoarterial fistula (TAF) and tracheoesophageal fistula (TEF)
Insertion related complications
The most common adverse events during endotracheal intubation are physiologic complications. In one international, multicenter observational study of almost 3000 critically ill patients undergoing intubation, cardiovascular instability was seen in 43% of cases, severe hypoxemia in 9%, and cardiac arrest in 3%. Other complications included esophageal intubation (5.6%), aspiration of gastric contents (3.9%), dental injury (1%), pneumothorax (0.7%), and airway injury (0.7%); reported airway injuries included tracheal, bronchial, and laryngeal laceration.95
A higher first pass success rate reduces the likelihood of complications.95 Operator experience is associated with first pass success and successful intubation.9798 One study comparing intubation by training level found that postgraduate year 5 residents and emergency medicine specialists had an increased likelihood of successful intubation compared with postgraduate year 2 residents (n=1154; odds ratio 3.80 (95% CI 1.62 to 5.26) for postgraduate year 5 versus postgraduate year 2; 5.71 (2.07 to 18.67) for emergency medicine specialists versus postgraduate year 2).98 In addition to operator experience, a recent multicenter trial including 1417 critically ill adults suggests that the use of videolaryngoscopy compared with direct laryngoscopy results in a higher first pass success rate (85.1% v 70.8%; absolute risk difference 14.3%, 95% CI 9.9% to 18.7%)6; these findings are consistent with those of a previous meta-analysis.99
Other strategies for decreasing complications of intubation recommended by current clinical practice guidelines and detailed further below include protocolized approaches to intubation, administration of high flow nasal oxygen (HFNO) or non-invasive ventilation (NIV) for pre-oxygenation, bag-mask ventilation (as opposed to no ventilation) following induction, and the use of hemodynamically neutral induction agents.100101102103104 Of these, data on the benefit of checklists and protocolized approaches to intubation are particularly mixed: whereas one prospective controlled study suggested a decrease in life threatening complications with versus without the use of a 10 bundle management protocol (n=144; 21% v 34%; P=0.03),96 another study did not show benefit.105 By contrast, data for HFNO are more consistent: on the basis of pooled evidence from 13 studies, HFNO seems to decrease the rate and degree of desaturation compared with pre-oxygenation with a face mask.100 Data comparing HFNO and NIV are less clear but may suggest a benefit of NIV in patients with more severe baseline hypoxemia (PaO2/FiO2 ≤200) (n=242; severe hypoxemia in 24% of the non-invasive ventilation group versus 35% in the high flow group; adjusted odds ratio 0.56, 95% CI 0.32 to 0.99).106 Data supporting bag-mask ventilation compared with no ventilation come from a randomized trial of 401 patients, in which bag-mask ventilation decreased the risk of severe hypoxemia (n=401; 10.9% v 22.8%; risk ratio 0.48, 95% CI 0.30 to 0.70) compared with no ventilation, without increasing the risk of aspiration.104 Finally, recent studies suggest that propofol may increase the likelihood of hemodynamic collapse (n=2760; odds ratio 1.28, 95% CI 1.04 to 1.57), and preferential use of ketamine over etomidate may be associated with decreased mortality (25% v 27%; risk ratio 0.93, CrI 0.79 to 1.08).107108
Intubation strategies that have not been shown to increase first pass success rates include the use of a tracheal tube introducer (or bougie) compared with the traditional endotracheal tube with stylet (n=1106; 80% v 83% first pass success rate; absolute risk difference −2.6%, 95% CI −7.3 to 2.2)109 and the use of a traditional sniffing position compared with a ramped position (n=513; pooled risk ratio 0.97, 95% CI 0.86 to 1.09).110
Post-insertion complications
Post-intubation complications include laryngeal injury, vocal cord dysfunction, and tracheal complications. Although laryngeal injury is rarely observed during intubation, a systematic review including nine studies and 775 patients noted a prevalence of 83% in endotracheally intubated critically ill adults,111 suggesting that it may be under-diagnosed during intubation or develop afterward. Although most of these were self-limiting grade 1 injuries (for example, laryngeal edema or erythema), grade 2 injuries (for example, ulceration or granulation tissue) and grade 3 injuries (for example, vocal fold paresis or glottic or subglottic stenosis) occurred in 31% and 13% of intubated patients, respectively.111 These injuries are typically diagnosed via laryngoscopy and can result in ongoing dyspnea and dysphonia. Risk factors include longer duration of intubation and larger sized endotracheal tubes.112113 Vocal cord paralysis is less common, occurring in 7% of patients intubated for respiratory failure in one cohort,114 but can result in impaired phonation and increased risk of aspiration. Paralysis is typically unilateral and is thought to be caused by compression of the recurrent laryngeal nerve by the cuff rather than direct injury to the vocal cords.114
Laryngotracheal stenosis is thought to be caused by excessive endotracheal tube cuff pressures resulting in ischemic injury and ulceration with consequent formation of granulation tissue and resulting stenosis. It can be progressive, and prolonged intubation is a risk factor for its development. Its frequency is unclear, with reported rates ranging from 1% to 21% in single center series; location may be supraglottic, glottic, or subglottic.115 Symptoms include dyspnea or stridor following extubation, with diagnosis made through direct bronchoscopic visualization. Most patients have no symptoms, with interventions needed in only 3-12% of cases.116 Management strategies can involve surgical or bronchoscopic interventions, and the preferred treatment depends on the location and extent of stenosis.
Other tracheal complications can include tracheomalacia and tracheal fistulization. The incidence of tracheomalacia following endotracheal intubation is unclear, with rates of 0.7-5% reported in single center studies.117118 Tracheomalacia is thought to be caused by weakening of the tracheal wall from repetitive ischemic injury, typically due to persistent cuff overinflation, especially in cases of prolonged intubation or tracheostomy use. Loss of tissue integrity can also potentially lead to tracheomegaly, resulting in difficulty with maintaining a seal with the cuff and often leading to further cuff overinflation. Symptoms can include dyspnea, cough, or failure to wean from mechanical ventilation, and diagnosis involves bronchoscopic visualization of dynamic airway collapse. Tracheomalacia can be temporarily managed with positive pressure ventilation or by positioning the tube to bypass the area of tracheomalacia; long term management may involve stenting or surgical resection.119
Fistulization of the trachea with either the esophagus or innominate artery is rare, with potentially life threatening complications occurring in <1% of patients.115120 Excessive cuff pressure and prolonged intubation are risk factors for both tracheoesophageal fistula and tracheoarterial fistula, which are typically caused by pressure exerted by the tube or cuff, resulting in erosion into the tracheal wall and fistula formation. Tracheoesophageal fistula is caused by erosion into the posterior tracheal wall and can result in respiratory distress, entry of enteric contents into the airway, and gastric distension. Esophogram or computed tomography scanning can be used for diagnosis. Initial management includes elevation of the head of the bed, gastric acid suppression, limitation of oral intake, nasogastric tube removal, and repositioning the cuff distal to the fistula; surgical repair is needed for definitive management.121 Tracheoarterial fistula, although more commonly seen after tracheostomies, can also be a complication of endotracheal tubes. It is caused by erosion through the anterior tracheal wall, typically with fistulization to the innominate artery. Tracheoarterial fistula manifests as severe bleeding from the stoma or endotracheal tube, sometimes preceded by a brief, self-limited “sentinel bleed,” and can result in asphyxiation. Management involves cuff overinflation and, for patients with tracheostomy, endotracheal intubation and application of manual pressure to the anterior tracheal wall through the stoma, followed by emergency surgical repair for definitive treatment.121
Despite the adverse effects of tracheal tube cuff overinflation, no clear guidelines on optimal cuff pressure exist.122123 Cuff pressures targets of 20-30 cm H2O are commonly reported by clinicians,122 but the frequency of cuff pressure monitoring varies considerably in clinical practice.122123124 Further research to define a safe range is much needed to reduce the risks of complication in mechanically ventilated patients.
Tracheostomy
Complications are reported to occur in 3-5% of tracheostomies, with a meta-analysis showing similar rates of mortality and serious adverse events with both percutaneous dilatational tracheostomies and those placed using open surgical technique (odds ratio 0.52 (95% CI 0.10 to 2.60; P=0.42) and risk ratio 0.93 (0.57 to 1.53; P=0.78), respectively).125126 Higher complication rates have been noted in patients with obesity.127 Similarly to those with endotracheal tubes, adverse events can be divided into insertion related and post-procedural complications.
Insertion related complications
Complications occurring during tracheostomy placement include bleeding, pneumothorax, and tracheal wall injury. Because these complications have the potential to compromise the airway, emergency airway equipment allowing for endotracheal intubation should always be immediately available during tracheostomy insertion. Bleeding, typically arising from injury of superficial blood vessels, is the most common early complication of tracheostomy placement, with a reported incidence of 2.4-8.7% in meta-analyses126128129; clinical practice guidelines recommend procedural planning with ultrasound guidance to help to mitigate this risk.125 Although application of pressure to the bleeding site is usually sufficient for achieving hemostasis, surgical exploration and revision may be needed in some cases. Pneumothorax, a rare complication occurring in <1% of patients,126128129 can have several causes, including perforation of the posterior tracheal wall, creation of a false tract anterior to the trachea, malpositioning of the tracheostomy tube, or excessive positive pressure.121 Although management is ultimately dictated by cause, the position of the tracheostomy tube should always be verified to ensure that it is not positioned in a false lumen. If the tracheostomy tube is in a false lumen, it should be removed and ventilation should be performed via an endotracheal tube.
Tracheal injury, more common during difficult insertions, can consist of tracheal ring fracture or posterior tracheal wall laceration (reported incidence 3-36% v 0.2-12.5%).116119 Although the first rarely requires intervention, the second may result in airway bleeding or pneumomediastinum, potentially necessitating surgical intervention or stent placement depending on the extent of injury.119
Post-insertion complications
Post-procedural complications include tracheomalacia, tracheal fistulization, tracheal stenosis, and reduced phonation (fig 3). Tracheomalacia and fistulization present similarly in patients with tracheostomy as in patients with endotracheal tubes and were discussed above. Tracheal stenosis is more common in patients with tracheostomies, potentially owing to their prolonged mechanical ventilatory status130; the location of stenoses also differs and can be suprastomal, stomal, or infrastomal. Suprastomal and stomal stenoses can be caused by irritation and damage to the trachea from the tube and exacerbated by excessive friction; securing the tube to minimize unnecessary movement may help. Stomal stenosis can develop because of tracheal ring fracture at the time of tracheostomy placement or as a result of repeated stomal infections. Infrastomal stenosis can be caused by sustained, excessive tracheostomy cuff pressures or by trauma from the distal portion of the tracheostomy tube itself.116 Diagnosis and management strategies are similar to those in endotracheal tube related stenoses.
Finally, loss of phonation has been identified as being among the most negative hospital experiences for patients, resulting in profound communication challenges as well as long term post-traumatic stress disorder.131132 Although voice loss is usually temporary, with reported time to phonating 12-18 days after tracheostomy insertion, it can sometimes persist, with 30% of patients unable to vocalize at hospital discharge in one cohort.132133 In a small RCT of 30 mechanically ventilated patients, a targeted early communication intervention conducted by speech pathologists and using inline Passy-Muir valves substantially decreased time to voice restoration (median difference 11 days; hazard ratio 3.66, 95% CI 1.54 to 8.68) without any increase in adverse events.132 Larger studies are needed to confirm these findings.
Complications of extracorporeal membrane oxygenation
Despite the increasing use of ECMO, mortality remains high. Although this is largely due to patient related factors and severity of illness, ECMO related complications, which can be hematologic, neurologic, infectious, vascular, or mechanical in nature, can also contribute (table 2).
Complications of extracorporeal membrane oxygenation (ECMO)8
Hematologic complications, encompassing both bleeding and thrombotic events, are the most common adverse events reported in the Extracorporeal Life Support Organization registry, occurring in 40% of patients with venovenous ECMO and 44% of those with venoarterial ECMO.134135 Exposure of patients’ blood to the ECMO circuit leads to the activation of prothrombotic pathways, with the most common thrombotic events including mechanical thromboses (in the circuit, pump, or oxygenator), hemolysis, and ischemic stroke.134135136 Bleeding events likely relate to platelet dysfunction, altered fibrinolysis, hemolysis, and use of systemic anticoagulation.136 The most common bleeding events include cannulation and surgical site bleeding, gastrointestinal bleeding, pulmonary bleeding, intracranial hemorrhage, and cardiac tamponade.134135
Neurologic complications, although rare, represent perhaps the most feared complication of ECMO, primarily in patients receiving venoarterial ECMO support. Ischemic and hemorrhagic stroke are most often reported, and other complications include seizures and hypoxic-ischemic brain injury. Hypoxic-ischemic brain injury warrants special attention, with brain autopsy studies suggesting significant under-diagnosis in the venoarterial ECMO population.137 Although some injury may be attributable to non-modifiable factors such as precedent cardiac arrest, some may be related to differential hypoxia or “north-south syndrome.” This phenomenon occurs in patients with femorally inserted venoarterial ECMO who have both cardiac and respiratory failure. As the cardiac function recovers, the native heart preferentially circulates blood that has not been properly oxygenated through diseased lung to the brain and upper body, while the ECMO circuit preferentially supplies the lower body with blood that has been well oxygenated through the ECMO circuit.
Infectious complications affect about a third to a half of ECMO runs.138 A high level of clinical suspicion is needed, as patients on ECMO rarely have fever owing to temperature regulation from the ECMO circuit.139 The treatment of these infections is often additionally complicated by challenges in achieving source control, as changing the ECMO circuit is not always feasible.138
Although rare, complications related to the vascular access procedure can be significant; cannulation should thus be performed by skilled operators under ultrasound guidance, following similar precautions to central venous and arterial cannulation.138 Thereafter, vascular complications most frequently occur in patients with peripheral venoarterial ECMO, with limb ischemia being most common.140 This can lead to serious downstream complications, including amputation or fasciotomy in cases of compartment syndrome. Clinical practice guidelines recommend distal perfusion cannulas as a strategy for mitigating this risk.141
Mechanical complications include circuit, pump, and oxygenator related complications; as these can be immediately life threatening, a pre-primed circuit should be maintained and readily available at all times.142 ECMO’s prothrombotic milieu underlies most of these malfunctions, but air embolism can also cause immediate circuit, pump, or oxygenator failure. Finally, cannula malposition can occur and lead to reduced flows, and cannula dislodgement presents more dramatically and typically requires immediate resuscitative measures, including massive transfusion.
Importantly, the aftermath of ECMO and its associated complications endure long after the patient has been discharged from the ICU. Studies suggest worsened health related quality of life scores in ECMO survivors compared with survivors of mechanical ventilation, with a meta-analysis of 245 patients showing a 5.4 (95% CI 4.11 to 6.68) decrement in SF36 scores among ECMO survivors.143 Given high rates of post-intensive care syndrome in ICU survivors, further investigation is needed to evaluate the degree and impact of physical, cognitive, psychological, and financial impairments in the more specific post-ECMO population.
Emerging treatments
Several emerging technologies are being explored, with evolving research focused on enhancing the operator’s performance and improving the patient’s experience. The incorporation of virtual reality and augmented reality may have potential applications for these purposes.144145 From the practitioner standpoint, augmented reality and virtual reality have been used to teach, practice, and perform various ICU procedures. In a prospective RCT examining the feasibility and efficacy of using augmented reality glasses for central line simulation compared with traditional simulation in 32 novice operators, significantly higher adherence to procedural steps was seen with augmented reality simulation. However, no difference was seen between total procedure time or time to venous cannulation.146 Results of studies on augmented reality and virtual reality for training and support for other ICU procedures, including ECMO cannulation, bronchoscopy, and percutaneous tracheostomy placement, have similarly been mixed.144147148 Nevertheless, these results should be interpreted with caution, as existing data are limited to small studies, and specific approaches for using augmented reality and virtual reality vary considerably. With regard to the patient’s experience, virtual reality has been explored as a means for managing procedural pain, with a significant reduction in pain scores seen with the use of virtual reality across a variety of medical procedures in a meta-analysis of 92 RCTs.149 However, studies assessing the efficacy of virtual reality in critically ill patients is limited, and similarly to the use of augmented reality for procedural training, specific methods of using virtual reality differ among studies; additional data are needed before its use can be adopted outside of the context of clinical trials.
Finally, the development of technology obviating the need for invasive hemodynamic monitoring may serve as the best method for reducing device associated complications. Currently available non-invasive monitoring devices typically use photoplethysmography, bioimpedance, arterial tonometry, ultrasound based technology, or a combination of these modalities.150151 Each technology has limitations, with existing data showing that commercially available non-invasive monitoring devices are not yet interchangeable with invasive methods in critically ill patients.151 Further refinement is needed, along with data validation in diverse cohorts of critically ill patients.
Guidelines
Guidelines relating to ICU devices exist,51821103125128141152153 but most primarily cover infection control and methods for device insertion. Guidance is lacking in several areas. Firstly, and most importantly, clearly defined indications have yet to be established for many devices. Administration of vasoactive support and hemodynamic monitoring, common uses for CVC and arterial line insertion, are often done without these devices, with an increasing amount of data showing its feasibility.2223 However, whether safe thresholds exist with regard to severity of shock, dosing of vasopressor, or duration of infusion above which CVCs and arterial lines should be used is unclear. Similarly, no consensus indications exist for endotracheal intubation or tracheostomy, with substantial practice variation noted in epidemiologic studies.154155 Secondly, guidance on reducing the incidence and managing the consequences of several non-infectious complications is limited. For example, regarding bleeding, although procedural methods such as the use of ultrasound can minimize risk for many procedures, best practices for managing coagulopathy are unclear. Finally, appropriate conditions for discontinuation are also poorly defined for some devices, potentially leading to unnecessarily prolonged use. Given the widespread use of these devices, closing these knowledge gaps and generating practical clinical guidance are imperative.
Future directions
Future work should focus not just on improving technology and guidance for clinicians but also on better understanding patients’ perspectives and ascertaining outcomes important to patients. Characterizing these concerns is crucial for ensuring that ongoing research and quality improvement efforts are impactful. Patients who contributed to this review identified several areas of concern relating to invasive ICU devices—long term sequelae, inadequate communication by clinicians during their use, and the concurrent usage of restraints (box 1). Little is known about long term consequences of device related complications, particularly as they relate to morbidity, disability, and health related quality of life. Post-intensive care syndrome, a collection of physical, psychological, and cognitive impairments afflicting survivors of critical illness, is being increasingly recognized, but much remains uncertain about the role that invasive procedures may play in its development.143156 Long term outcome studies are needed in this patient population to close these knowledge gaps.
Patients’ experiences with devices used in intensive care unit
Long term symptoms
My right ventricle was punctured on the first attempt to cannulate me [for ECMO]. I ended up having three sternotomies and still experience a great deal of sternum and rib pain, as well as peripheral neuropathy in my legs. All of this neuromuscular pain makes sleeping extremely difficult… My trach scar is also still painful, seven years out.
I had central lines in my jugular veins on both sides of my neck. They’re gone, but now I cannot sleep with anything touching my neck. Blankets/sheets can touch my cheek/jaw or shoulder, but not my neck.
One thing many of us have experienced is [ongoing] difficulty swallowing and choking while eating. I don’t know what in the ICU causes this, possibly being vented.
I kept feeling something very uncomfortable at the right side of my neck. I had no idea what it was as I hadn't seen myself in a mirror and had no family there, so I grabbed hold of it and pulled it out. It was a port… I still have moments of pain and stiffness in that area of my neck years later.
Communication
I remember feeling like I wasn’t a person when the decision was made to initiate the ventilator… Imagine lying in a hospital bed, feeling like you are suffocating and fearing that you are dying… The patient doesn’t really get much explanation and what is explained is in medical terms and doesn’t really make sense. It’s confusing and terrifying… To this day, my parents are still experiencing PTSD symptoms.
There is a lot of anxiety and stress from the staff that fills the room… I became even more anxious and feeling out of control, scared, helpless as everyone runs around grabbing things, poking and prodding me and everyone talks about me, but not to me. Minutes felt like hours.
I found the insertion of my PICC line to be very rough. A sterile sheet was placed over my face so it was even harder to breathe. I was terrified I would cough and my heart would be punctured. I feel like this could have been accomplished with more care and understanding about how hard and scary it was.
Restraints
Mechanical and chemical restraints—this is an area that should be considered a device complication and can result in long term issues. A lot of us experienced PTSD symptoms and physical injuries related to their use.
While I was restrained in the ICU, I managed to damage many of the tendons in both arms, including severing both of my bicep tendons and a few tendons in each forearm. I did so much damage while resisting the restraints that I pretty much destroyed both of my shoulders. This cost me my career.
Selected quotes from patients reflecting their experiences with ICU devices. Some quotes were lightly edited for clarity.
ECMO=extracorporeal membrane oxygenation; ICU=intensive care unit; PICC=peripherally inserted central catheter; PTSD=post-traumatic stress disorder.
The use of physical restraints and communication lapses in the ICU also warrant further investigation. Although often applied as a safety measure to prevent inadvertent device removal, use of physical restraints in the ICU can lead to musculoskeletal and vascular injury, as well as depression, anxiety, and post-traumatic stress disorder in ICU survivors.157 Despite these adverse effects, physical restraints are used in up to 76% of patients in ICU, although substantial variability exists.158159 One factor associated with decreased restraint use is concordance in patient-clinician language,160 suggesting that strategies for improving communication warrant further exploration. Messaging boards, apps, and voice enabling devices have been explored as options for aiding communication with critically ill patients with artificial airways. Although evidence for their benefit is uncertain, additional studies are needed to better understand their impact and ensure that the interventions are being optimally designed, tested, and implemented.161 As research is ongoing, continued efforts should be made to use available resources to facilitate clear, direct communication between clinicians and patients as much as possible.
Conclusion
Invasive procedures are a common source of complications in the ICU. Although the foundation of avoiding device related complications consists of avoiding unnecessary device use, with careful assessment of their need and prompt removal when no longer necessary, understanding modifiable risk factors for complications is essential for critical care practitioners. Additional research is needed to close gaps in available clinical guidelines and better understand long term consequences of ICU devices.
Questions for future research
What are the most appropriate indications for the insertion of intravascular catheters and airway devices in critically ill patients?
What are the best methods for managing non-infectious complications of devices and reducing their incidence?
Which device associated complications are most important to patients?
What are the long term consequences of invasive devices commonly used in the intensive care unit?
How patients were involved in the creation of this manuscript
We consulted two patients who are survivors of critical illness—Judy Eloed, MSW, LCSW, and Dennis K Gonzales, RN—in the development of this review. Both JE and DKG provided initial input on the planned content. JE then engaged members of a critical illness survivorship support group to elicit additional feedback on the planned content, specifically focusing on identifying matters of particular importance from a patient’s perspective, as well as the post-hospital admission impact of device associated complications. Seven additional patients provided written input, which was then incorporated into the manuscript. We subsequently shared the full manuscript with JE and DKG for their review and approval before submission.
Acknowledgments
We thank our patient partners for their insights and contributions to this manuscript, including Judy Eloed, Dennis K Gonzales, Kirsten Harrell, Ron Jennings, and Alisa Neal.
Footnotes
Series explanation: State of the Art Reviews are commissioned on the basis of their relevance to academics and specialists in the US and internationally. For this reason they are written predominantly by US authors
Contributors: RH conceptualized the topic, design, and structure of this manuscript. RH, KSJ, KM, and NQ did the literature review and contributed to the drafting of the manuscript. NQ edited and revised the manuscript and is the guarantor. All authors approved the final manuscript.
Competing interests: We have read and understood the BMJ policy on declaration of interests and declare the following interests: none.
Provenance and peer review: Commissioned; externally peer reviewed.