Vascular Access (Critical Care Manual)


When establishing vascular access, hand-washing with soap and water is essential, although there is no benefit to using antimicrobial agents [NEJM 327: 88, 1992]. Use sterile gloves for all cannulations except short catheters in peripheral veins. Two studies have suggested the superiority of chlorhexidine-containing antiseptic solutions over iodine solutions or 70% alcohol [Lancet 338: 339, 1991; Crit Care 24: 1818, 1996], particularly for the eradication of gram-positive cocci, although the CDC does not think this is borne out clinically [mmwr/PDF/RR/RR5110.pdf]. If chlorhexidine is not available, povidine-iodine (Betadine) is superior to alcohol (it is both an antiseptic and detergent) and should be left for 2 full minutes before starting procedures. Shaving is not recommended as it can promote colonization – if necessary, use depilatory creams or clippers.

Catheter-over-needle devices can be frayed as they pass through skin and soft tissues, leading to endothelial damage once in the vessel of interest, thus they are limited to cannulating superficial vessels – more important vessels (arteries, central veins) are cannulated using a guidewire, although the use of a rigid dilator may nullify its apparent advantage.

Silicone catheters (Hickman, Broviac) are too flexible and are inappropriate for ICU use – polyurethane catheters are stiffer, designed for shorter use, and more appropriate. Do not use heparin-impregnated catheters: they may reduce infection by a small amount (2%, [Crit Care Med 28: 3332, 2000]) but they have not been shown to reduce thrombosis [Rosen, Latto: Handbook of Percutaneous Central Venous Catheterization; WB Saunders, 1992], the heparin washes off after several hours, and they can cause HIT [AIM 149: 2285, 1989]. Antimicrobial-impregnated catheters should be considered if the infection rate in your ICU is above the national average (3.8-5.3 infections per 1000 catheter days) – only 2 of 9 studies of chlorhexidine/silver sulfadiazine show a significant reduction [Crit Care Med 28: 3332, 2000], but comparisons of the new minocycline–rifampin catheters show that they are better than chlorhexidine [Infect Med 18: 38, 2001]

Catheter diameter is key, with a 20 Ga providing 39.5, an 18 Ga providing 60.0, and a 16 Ga providing 96.3 mL/min of PRBC [Ann Emerg Med 22: 1551, 1993]. Multi-lumen catheters should be used when possible as they minimize venipunctures but do not increase the risk of thrombosis or infection when compared to single lumen catheters [NEJM 348: 1123, 2003; Rosen, Latto]. Introducer catheters are also helpful, especially since they can be used to provide fluids during large volume resuscitation in addition to allowing multiple catheters to be placed without requiring more than one venipuncture.

Venous Access

Antecubital veins are safe and easily accessible. The basilic (medial) vein is straighter and thus preferred. If veins are not visible and access is mandatory, palpate the brachial artery 1” proximal to the antecubital crease and place the catheter at 35 – 45○. In resuscitation situations, short (5 – 7 cm) catheters of 16 – 18 Ga should be used. Peripheral venous catheters should be changed every 3-4 days, and arms are preferred over legs because of a lower risk of venous thrombosis [CDC].

PICC Lines

For PICC lines, the basilica (medial) is preferable to the cephalic (lateral) because it is larger (8 vs 6 mm) and straighter. PICCs can be left in place for 30 days without increased risk of infection compared to central lines [J Crit Illness 15: 165, 2000], however they are more prone to clot (because of their small diameter) and more prone to cause phlebitis (because of their long length). A cohort study of 239 central access (PICC vs. standard central line) showed that the rate of DVT per 1,000 catheter days 7.7 for a PICC (versus 4.4 for standard CVCs [Bonizzoli M et al. Intensive Care Med 37: 284, 2011]. The major advantages of a PICC are their lower risk of infection (documented in multiple studies) and the decreased risk for pneumothorax.

PICC vs. Standard Central Line

  • Advantages of PICC: lower rate of infection, zero risk of pneumothorax
  • Disadvantages of PICC: almost 2-fold higher rate of venous thrombosis

Central Venous Catheters


Subclavian vein: cannulation carries a 1 – 3% risk of pneumothorax and a 1% risk of hemothorax. As the average distance to the right atrium is 16.5 cm in adults, all subclavian venous catheters should be less than 16 cm in length [Crit Care Med 21: 1118, 1993]. The actual vein lies just underneath the clavicle where the lateral head of the SCM inserts into the bone, and the vein can be entered from either side. In the infraclavicular approach, identify the SCM/clavicular intersection and draw a line straight down, inserting at the inferior margin of the clavicle – point the bevel up and aim towards the suprasternal notch with the needle path parallel to the patient’s back. Once entered, point the bevel to 3 o’clock so the guidewire passes inferiorly. For the supraclavicular approach, place the needle so that it bisects the angle between the clavicle and SCM, aiming it towards the opposite nipple. Once entered (1-2 cm in, more superficial than the infraclavicular approach), turn the needle to 9 o’clock towards the SVC. Some people suggest avoiding the subclavian approach in ventilated patients because of the risk of pneumothorax but Marino disagrees. This is a safe and comfortable (for the patient) procedure and should be the first line attempted. Use the right side if possible, as the left is more likely to lead to SVC perforation.

Internal Jugular

Internal jugular vein: can be canalized, and this reduces (but does not eliminate) the risk of pneumothorax, while adding the risk of carotid artery (2 – 10% [Intensive Care Med 2: 163, 218, 1987]) or thoracic duct injury. Awake patients will complain of reduced neck mobility, and agitated patients may occlude the catheter and vein secondary to excessive neck flexion. Lastly, patients with tracheostomies may be at increased risk of infection due to spread of secretions. When the head is turned to the opposite side, the IJV forms a straight line from the pinna to the sternoclavicular joint. The right side is preferred for cannulation. In the anterior approach, palpate the carotid artery in the triangle of the SCM heads and retract medially. Insert the needle into the apex of the triangle, advancing towards the ipsilateral nipple at 45° to the skin surface. If no venous blood by 5 cm, withdraw 4 cm and advance again more laterally. For the posterior approach, insert the needle (with the bevel at 3 o’clock) 1 cm above where the external jugular crosses the lateral edge of the SCM, advancing along the underbelly of the muscle and pointing towards the suprasternal notch. The IJV should be encountered within 5 – 6 cm. If blood is red and pulsating, remove the needle and apply pressure for 5 – 10 mins. If the carotid artery has been punctured by the needle, remove it and hold pressure for 5 mins. If the carotid as been actually cannulated, the catheter should be left in place and vascular surgery should be called. The only advantages of the IJ line are for pacemaker catheters and hemodialysis catheters, because of their straight course.


Femoral vein: easiest to cannulate but leads to increased risk of venous thrombosis (10%), femoral artery puncture (5%), and limited hip flexion. Previous data suggested that the infection rate of femoral lines was no different from SC or IJV catheters [Intensive Care Med 2: 163, 218, 1987]. More recent data suggests that the insertion areas can be classified from low to high risk of extraluminal infection as follows: antecubital fossa < subclavian < femoral < jugular [Infect Control Hosp Epidemiol 15: 253, 1994]. Palpate the femoral artery just below the inguinal crease and insert the needle 1 – 2 cm medial to the palpated pulse at a 45° angle to the skin and entering within 2 – 4 cm. If problems with the guidewire, tilt the needle so that it is more parallel to the skin surface. Femoral vein catheters should be at least 15 cm long. In cases where you cannot find the femoral pulse, draw a line from the ASI crest to the pubic tubercle and divide into thirds – the junction between the medial and middle thirds is the femoral artery, so enter 1 – 2 cm medial to that – this will be successful 90% of the time [Am J Surg 138: 875, 1979]. Femoral lines are not recommended for cardiopulmonary resuscitation (transit time too long) or in patients with bleeding disorders. These should only be used when a SC or IJ are unavailable, and should be removed as soon as possible.

Thrombosis of femoral vein catheters may be as high as 20%, and subclavians as low as 2% [Crit Care Clin 19: 489, 2003]

External Jugular

External jugular cannulations: eliminate the risk of pneumothorax and hemorrhage can be easily controlled, however it is difficult to advance the catheter. The vein runs approximately from the angle of the jaw to the mid-clavicle. To visualize it, press down on the distal vein, causing engorgement. The vein must be held by tension (forefinger and thumb) as it is relatively mobile. Insert halfway between the angle and the clavicle and do not force the catheter as it may rupture the vein. This is an option when inexperienced operators are cannulating patients with a coagulopathy and would like to avoid the IJV or the SC.

Adverse Effects of Central Venous Catheters

Adverse Effect Subclavian Internal Jugular Femoral
Pneumothorax 1-3% 1% 0%
Thrombosis 1% 0% 6%
Sepsis 1-4% 0-8% 2-5%
Major Bleeding 2% 1% 1%
Arterial Punct. 1-15% 3% 9%

Arterial Access

The radial artery is superficial and accessible, its only disadvantage is its size (difficult to cannulate, possibility of vascular occlusion). The ulnar artery (medial) is actually larger, but the radial is preferred. A positive Allen test (> 14 seconds for color to return) is not a contraindication to radial artery cannulation [Crit Care Med 16: 915, 1988; Crit Care Clin N Am 8: 687, 1992]. Always extend that patient’s wrist and use a short 20 Ga catheter, inserting over a needle or guidewire. At first flash, the catheter is outside the vessel, so pass the needle completely through the vessel (until flashback stops), then pull it back in (until flashback returns) at which point it can be removed with the catheter in place. Up to 25% of cases will result in arterial occlusion but digital ischemia is rare [Crit Care Clin N Am 8: 687, 1992]

The femoral artery is easier to cannulate than the radial, and offers the same risks (thrombosis, bleeding, infection), an identical infection rate, but a lower rate of thrombosis which is only rarely of clinical import [Crit Care Clin North Am 8: 687, 1992]. Catheterize as you would the vein but of course do not go 1 – 2 cm lateral. Use the Seldinger technique, 18 gauge catheters, 15 – 20 cm long. Femorals may be preferred in immobile patients (except in the case of coagulopathy) as thrombosis is lower and the pressure readings are more accurate.

Complications and Concerns


Venous air embolism is a feared complication and can be fatal. To prevent these, place the patient in Trendelenburg with the head 15° below the horizontal plane. Also, have the patient hum (or initiate a mechanical lung inflation) when changing connections to increase intrathoracic pressure. Air emboli will present with dyspnea, hypotension, and a “mill wheel” murmur and may lead to cardiac arrest, stroke. If suspected, place the patient with their left side down and try to aspirate air from the venous line. As a last resort, place a needle directly into the right ventricle.

Pneumothorax is common and thus all patients should get a chest film, preferably upright and expiratory. When this is not possible, look for air in the subpulmonic recess and along the anteromedial border of the mediastinum. In the absence of signs and symptoms, there is little justification for serial films. It is important to note, however, that catheter-induced pneumothoraces may not be evident for 24 – 48 hours [Am J Rad 144: 901, 1985] – this does not mean serial x-rays are justified, only that shortness of breath 24 hours after a clean chest film may still be due to a pneumothorax. The catheter tip should be at the junction of the SVC and the RA or slightly above the third anterior intercostals space. Catheter tips up against the wall of the SVC should be repositioned as soon as possible. While cardiac perforation is rare [Crit Care Med 21: 1118, 1993], the RA should also be avoided – this is the level of the anterior portion of the third rib, and if this cannot be visualized, pull the catheter back to or above the carina.

All vascular catheters are flushed routinely. Intermittently used venous catheters are often hep-locked. Heparin is expensive and can cause HIT, and saline has been proven as effective as heparin in venous catheters [Heart Lung 20: 631, 1991], thus saline should be used instead. Arterial catheters are and should be flushed continuously (3 mL/hr), but saline is not always as effective in arterial lines [Am J Crit Care 2: 3, 1993] so heparin should be used unless HIT is a significant risk, in which case use 1.4% sodium citrate [Chest 103: 882, 1993]

When catheters occlude, every effort should be made to re-open them and avoid replacement. Guidewires should NOT be used, however, because of the risk of embolus. Thrombosis is the most common cause of central venous catheter occlusion [Crit Care Clin 19: 489, 2003], thus the following protocol is offered:

Drug: alteplase (rTPA) [Ann Pharmacother 37: 27, 2003] Preparation: add 50 mL sterile water to 50 mg vial alteplase for a 1 mg/mL solution.
Prepare 2 mL aliquots (approximately 25) and freeze them until needed.

  1. Thaw 2 x 2 mL aliquots (must use drug within 8 hours of thawing)
  2. Draw 2 mL into a 5 cc syringe and attach to hub of occluded catheter
  3. Inject as much as possible and then cap the hub
  4. Leave the solution in for 2 hours
  5. Attempt to flush with saline (do NOT use a tuberculin syringe, the high volume of flow can fracture the catheter hub)
  6. Repeat steps 1-4 if needed
  7. If still obstructed, try using 0.1N HCl (2 mL) for drug or calcium phosphate precipitates, or 70% ethanol (2 mL) if lipid residues are suspected. Otherwise, replace the catheter
  • Note that for non-thrombotic occlusion, HCl (drug or calcium phosphate precipitates,) or 70% ethanol (lipid residues) may be more useful.

Clinically apparent subclavian vein thrombosis occurs in ~ 1% of patients and its unmistakable sign is unilateral arm swelling. The rate of secondary PE is estimated to be between 0 and 17% [Chest 123: 1953, 2003]. Dopplers are often the first test but in the upper extremity their sensitivity and specificity are 56 and 69% [AIM 162: 401, 2002]. Contrast venography is the gold standard but rarely use. If diagnosed, remove the catheter and avoid all central venous access if possible for the next several weeks (thrombus may extend into the SVC). Many people anticoagulate but there is no evidence that heparin is beneficial in this setting [Br J Clin Pract 45: 31, 1991]. Femoral vein thrombosis has a much higher incidence, which is why these catheters should be avoided and if used, replaced after 48 hours.

SVC perforation leads to non-specific symptoms (substernal chest pain, cough, dyspnea), and should be searched for radiographically (ex. mediastinal widening, pleural effusion). In fact, any patient with a sudden pleural effusion and a central line should be considered for perforation. To diagnose this definitively, tap the pleural effusion and analyze it, also pass contrast into the catheter. If verified, remove the catheter immediately – surprisingly, this maneuver does not lead to increased bleeding [J Crit Illness 9: 101, 1994]. Also, antibiotic therapy is not needed unless there is evidence of infected pleural fluid [J Crit Illness 9: 101, 1994]. If the pleural effusion is glucose-rich, it should be drained to avoid bacterial contamination/proliferation.


All catheter insertion sites should be covered. The standard is sterile gauze and hypoallergenic tape changed every 48 hours. Clear, occlusive dressings have become popular but promote colonization of the underlying skin, although only those that are completely impermeable actually show an increased incidence of septicemia [JAMA 267: 2072, 1992]. Marino does not recommend their use. The application of antimicrobial ointment to the insertion site has not been shown to decrease infection rates and is not recommended [New Horiz 1: 271, 1993]. TegaDerm and OpSite should be reserved for catheters located near infectious sites (internal jugular vein insertion sites, etc.).

The major risk of peripheral venous catheters is phlebitis (not infection) – this risk is unchanged from 72 – 96 hours [Am J Infect Contr 26: 6, 1998], thus these catheters should be changed (using a new insertion site) every 3-4 days. According to CDC Guidelines [Centers for Disease Control and Prevention. MMWR Recomm Rep 51: 1, 2002], peripheral venous catheters should be replaced every 72 to 96 hours in adults to prevent phlebitis (Class IB data).

While central venous catheters in place longer than 3 days have an increased risk of infection [Am J Infect Contr 18: 201, 1990], replacement either over a guidewire or at a new venipuncture site does not reduce the risk of infection [Crit Care Med 18: 1073, 1990] and may actually increase the risk of infection [NEJM 327: 1062, 1992] as well as bring about mechanical complications (risk = 7% [NEJM 348: 1123, 2003]), thus they should not be replaced routinely. A meta-analysis of over 450 patients in 3 controlled trials confirms this – catheters should only be changed on clinical grounds [Infect Control Hosp Epidemiol 21: 371, 2000]

Indications for catheter replacement at a new site

  • Purulence at the catheter site (erythema alone is not an indication is it is not a absolute evidence of a catheter infection and in fact there is no correlation between infection at the insertion site and the likelihood of septicemia [Crit Care Med 30: 2632, 2002])
  • Tip of previously removed catheter showing > 15 CFU
  • Femoral vein catheter in place longer than 48 hours (reduce the risk of venous thrombosis)

Indications for guidewire exchange

  • Suspected catheter-related infection with no erythema or purulence in a septic patient with a prosthetic valve, immunocompromized state, or in severe sepsis/septic shock
  • Incidences when a catheter was placed emergently without aseptic technique

These are for the most part consistent with the CDC Guidelines [Centers for Disease Control and Prevention. MMWR Recomm Rep 51: 1, 2002], which also state that institutions should consider use of antiseptic- or antimicrobial-impregnated CVCs for adults whose catheter is expected to remain in place more than 5 days if the institution is unable to achieve goal rates of bloodstream infections by use of a comprehensive strategy to reduce infection rates (Class IB data).

To diagnose using a quantitative blood culture technique, draw 10 mL blood from both the catheter and a distal site. Suspect catheter infection if catheter blood produces > 100 CFU/mL of bacteria (must be the same species as found in the bloodstream) or catheter blood produces > 5x the CFUs of peripheral blood (sensitivity 40-50% [Clin Infect Dis 23: 1249, 2001]). This should usually be the first test, especially in cases where removing the catheter is undesirable, because up to 70% of suspected infected catheters are sterile when cultured [Clin Infect Dis 23: 1249, 2001]

Another method is to semiquantitative tip culture the distal 5 cm catheter tip, compare this to a peripheral blood culture and if the same species, 15 CFU/mL is the threshold (sensitivity 38%, specificity 100%, improved by plating the tip at the bedside) – as this method requires removal of the catheter and is relatively insensitive, it should only be used in cases of erythema/purulence when the catheter should be pulled regardless. If this method is chosen, the last 2 cm should be split and Gram stained. This method misses bacteria found in the lumen of the catheter.

Quantitative tip culture methods are more cumbersome and less commonly used. The catheter segment is either flushed with broth or sonicated in broth, followed by serial dilutions and surface plating on blood agar. A yield of 100 CFUs or higher indicates infection. The sensitivities of the three techniques are reported as follows: sonication (quantitative), 80%; roll plate method, 60%; and flush culture, 40% to 50% [J Clin Microbiol 35: 641, 1997]. In laboratories that use continuous blood culture monitoring, the length of time from blood draw to positive result for catheter-drawn versus percutaneously obtained cultures can be noted. If the time to positivity of the catheter-drawn culture is at least 2 hours less than that of the peripheral culture, catheter infection is suggested. Sensitivity and specificity of 91% and 94%, respectively, have been reported for this method [J Clin Microbiol 36: 105, 1998]

Empiric Antibiotic Therapy for Catheter-Related Infections

Pneumonia is the leading cause of sepsis in ICU patients, but catheters are 2nd. The most likely organisms are coagulase-negative staphylococci (37%), Enterococcus species (14%), Staphylococcus aureus (13%), Candida albicans (5%), Enterobacter species (5%), Pseudomonas aeruginosa (4%), Klebsiella pneumoniae (3%), Escherichia coli (2%), all other pathogens (17%) [Am J Infect Control 27: 520, 1999]. Manifestation is often fever in patients with a catheter for > 48 hours, many times with no sign of local infection.

As a first step, draw quantitative catheter blood cultures (except in patients with purulent drainage, who are neutropenic, have prosthetic valves, or are in severe sepsis/septic shock) but leave the catheter in place.

For suspected catheter-related septicemia, empiric antibiotics are recommended [Clin Infect Dis 23: 1249, 2001]. Vancomycin is well-suited for this, despite concerns about resistance. Add ceftazidime or cefipime for gram-negative coverage in patients with severe sepsis/shock. For neutropenics, add a carbapenem [Clin Infect Dis 34: 730, 2002]. In patients with prosthetic valves, an aminoglycoside should probably be added.

If positive culture results come back, start a 10-14 day course of directed antibiotics (candidal infections should be treated for 14 days AFTER negative cultures are obtained). Catheters should then be removed unless they are tunneled or the organism is S.epidermidis.

The antibiotic lock technique (allowed to dwell for hours to days) with or without systemic antibiotic therapy has been reported to result in catheter salvage in 82.6% of 167 episodes and to be significantly more likely to result in catheter salvage than parenteral therapy alone [Clin Infect Dis 32:1249, 2001]

If infection/inflammation continue after antibiotics, consider disseminated candidiasis, suppurative thrombophlebitis, or endocarditis.

Candidiasis is predisposed by chemotherapy, steroids, HIV, antibiotic therapy, prostheses, and vascular catheters. Diagnosis is difficult because blood cultures are negative in ~ 50% of cases. Serum enolase and agglutination tests for cell wall antigens can be helpful but are not reliable. Urine should always be cultured and ophthalmology consulted (endophthalmitis) in suspicious cases. All patients with persistent candidemia should have a detailed eye exam [Curr Opin Crit Care 3: 335, 1997]. Empiric therapy reserved for neutropenics, prosthetic heart valves, or progressive sepsis. In those with documented disease, liposomal amphotericin B is the standard of care, although capsofungin has been proven just as effective for invasive candidiasis [NEJM 347: 2020, 2002] and has a better side effect profile.

Suppurative thrombophlebitis is often accompanied by local inflammation and purulent drainage after removal of a catheter, and is usually treated with surgical excision (although heparin and antibiotics can sometimes work).

Endocarditis should be suspected as well – any patient with a S.aureus bactermia should be evaluated by TEE [Clin Infect Dis 23: 1249, 2001]

Controversies Regarding Infection Etiology

The implication of skin microorganisms in CRBI has been documented by both conventional and molecular biology bacterial identification techniques [J Clin Microbiol 26: 8, 1988]. Marino notes, however, that there is poor correlation between skin organisms and blood organisms in specific cases [Nutr Clin Pract 6: 43, 1991; Infect Control Hosp Epidemiol 9: 54, 1998]. Practices aimed at reducing skin microbes have had little or no impact [New Horiz 1: 271, 1993; Lancet 1: 459, 1984; Surg Gynecol Obstet 175: 33, 1992; Ann Surg 218: 206, 1993]. 50% of catheter-related septicemia cases involve organisms not commonly found on skin [Surg Clin North Am 68: 57, 1988]. And lastly, Staphylococcus are commonly found in the intestines, especially of critically-ill patients and those on abx [Ann Surg 157: 847, 1964]. In fact, S.epidermidis is one of the most commonly isolated pathogens from the bowel in patients with multi-organ failure [Arch Surg 123: 309, 1988].

There are others who believe that the endoluminal route of contamination is important [Lancet I:668, 1993; Surgery 97: 355, 1985]. In one study, applying a gauze impregnated with povidone iodine around the catheter hub proved successful in controlling an epidemic of catheter sepsis caused by coagulase-negative staphylococci [Sitges-Serra et. al. J Parenter Enteral Nutr 9: 322, 1985]. In a controlled prospective trial [Nutrition 7: 33, 1991] patients with tunneled Silastic central catheters were randomized to two groups, one (control group) managed in a standard fashion and the other managed by applying a betadine-impregnated foam around the junction between the catheter and the administration set. A 24% CRBI rate was demonstrated in the control group, while not a single case of bacteremia was found in patients whose hubs had been protected. In a clinical trial, 151 patients with central venous catheterization for a mean of 2 weeks were randomized to receive catheters with standard Luer-lock connectors or connectors protected with Segur-Lock [Ann Surg 223: 363, 1996]. The CRBI rate was higher in the control group (16% vs 4%), and this was due to the low rate of hubrelated CRBI observed in the group fitted with the new hub (1% vs 11%, p ~ 0.01).