From Wiki Journal Club
Jump to navigation Jump to search
Parienti JJ, et al. "Intravascular Complications of Central Venous Catheterization by Insertion Site". The New England Journal of Medicine. 2015. 373(13):1220-1229.
PubMedFull textPDF

Clinical Question

In critically ill patients requiring central venous access, how do complications vary by central venous catheter (CVC) access site?

Bottom Line

In this study with limited generalizability to US academic centers, subclavian insertion was associated with fewer bloodstream infections and DVTs compared to jugular and femoral approaches, but was balanced by more mechanical complications including pneumothorax.

Major Points

Central venous catheters (CVCs) are associated with infrequent but potentially severe complications, including catheter-related bloodstream infection (CRBSI), DVT, and pneumothorax. A 2001 French randomized study demonstrated that subclavian insertion was associated with fewer infectious and thrombotic complications than femoral insertion, but internal jugular (IJ) insertion was not studied.[1] A 2012 meta-analysis of mostly nonrandomized studies suggested that the subclavian approach may be associated with fewer infectious complications than femoral or IJ insertions. However, it concluded that a randomized trial was necessary to confirm the rates of these complications by access site.[2]

The 2015 3SITES study was a multicenter, randomized trial designed to investigate whether subclavian, IJ, or femoral access was associated with a difference in the rate of major catheter-related complications. The primary outcome of major catheter-related complications was defined as CRBSI or symptomatic DVT, and were adjudicated by a committee unaware of randomization. A total of 3,027 patients with 3,471 catheters were studied. Major complications at the subclavian, IJ, and femoral insertion sites occurred at a rate of 1.5, 3.6, and 4.6 events per 1,000 catheter-days, respectively (P=0.02). Mechanical complications, including pneumothorax, accounted for 18, 12, and 6 events in the subclavian, jugular, and femoral groups, respectively (P=0.047). Fewer mechanical complications occurred in patients whose CVC was inserted with ultrasound guidance, although use of ultrasonography was not randomized. The investigators conclude that a subclavian approach reduces the risk of catheter-associated complications.

The validity of this conclusion needs some consideration. Importantly, mechanical complications such as pneumothorax were not considered primary outcome events. As such, the 3SITES primary analysis cannot speak to whether the reduction in the primary outcome achieved by subclavian access is outweighed by an increase in mechanical complications. This was partly addressed in an analysis of the composite rate of the primary outcome plus grade ≥3 mechanical complications between insertion site groups, which found the complication rate was similar between approaches (subclavian 3.1%, jugular 3.7%, and femoral 3.3%). Ultimately, this suggests that rates of significant complications do not vary widely by insertion site and thus an optimal insertion site probably doesn't exist.

In addition to these findings, several caveats pertain to this study's generalizability. First, this study defined CRBSI as colonization of the catheter tip by ≥1000 cfu/mL of the same organism recovered from a peripheral blood culture. While this is a common research definition, it is more stringent than the published IDSA guidelines.[3] Second, while using published French and US guidelines for preventing CRBSIs, this study did not employ antibiotic-impregnated or tunneled catheters, which may reduce the frequency of CRBSIs.[4] Third, the catheters were inserted by experienced physicians who previously had placed at least 50 catheters, which differs from the routine practice at many US academic centers who allow supervised medical students, interns, and junior residents to perform these procedures as a matter of routine.

As a final observation, 3SITES suggests that the notion of the "dirty femoral catheter" is becoming a relic of the past. While 3SITES demonstrated more frequent catheter tip colonization with femoral compared to subclavian (HR 3.4; P<0.001) and jugular approaches (HR 1.6; P=0.003), this did not translate to a clinically significant increase in risk of CRBSI when compared to subclavian (HR 3.4 with a confidence interval which includes 1.0; P=0.048) or jugular approaches (HR 0.9; P=0.81). This finding is consistent with a 2012 meta-analysis.[5]


As of November 2015, no guidelines have been published that reflect the results of this trial.


  • Multicenter, randomized, controlled trial
  • N=3,027 ICU patients with 3,471 catheters
    • Subclavian (n=1,016)
    • Jugular (n=1,284)
    • Femoral (n=1,171)
  • Setting: 10 ICUs from 4 university and 5 community hospitals in France
  • Enrollment: 2011-2014
  • Median follow up: 5 days
  • Analysis: Intention-to-treat
  • Primary outcome: Major catheter-related complications (bloodstream infection or symptomatic DVT)
  • Secondary safety outcome: Major mechanical complications


Inclusion Criteria

  • Age ≥18 years
  • Admitted to the ICU
  • Required central venous access
  • Had at least 2 acceptable insertion sites (femoral, jugular, or subclavian)

Exclusion Criteria

None provided in manuscript.

Baseline Characteristics

  • Demographics: Age 63, 64% male, BMI 26
  • Comorbidities: T2 DM 19.7%, Cancer: 9.7%, AIDS: 1.3%; SAPS II 56.1
  • Antibiotic therapy: 55%; anticoagulation therapy: 30%; parenteral nutrition: 6.2%
  • Skin antiseptics: Alcoholic chlorhexidine 44%, alcoholic povidone-iodine 43%, aqueous povidone-iodine 10%
  • Use of anatomical landmarks to guide insertion: subclavian (86%) vs. femoral (74%) vs. jugular (33%)
  • Catheter insertion time quicker in femoral group (11.6 mins) vs. others (12.7 mins)
  • Insertion failure: femoral 5.3% vs. jugular 7.7% vs. subclavian 14.7%
  • Reason for removal: no longer needed 58%, death 18.5%, suspected catheter infection: 12%
  • Mean duration of catheterization: 6.3 days


  • Somewhat complex randomization to insertion site.
    • Patients with 3 available sites were randomized in 1:1:1 ratio to available site (subclavian, jugular, or femoral)
    • Patients with only 2 available sites were randomized in a 1:1 ratio to one of the 2 available sites.
  • Randomization was stratified according to ICU and according to the use of antibiotics versus no therapy
  • All ICUs used the French Haute Autorite de Sante checklist and US guidelines for preventing catheter-related infections
  • Maximal sterile barrier precautions were used
  • Catheterization was achieved by means of the Seldinger technique with the use of anatomical landmarks or ultrasound guidance, which was not randomized
  • Within 2 days of catheter removal, compression ultrasound was used to detect asymptomatic or confirm symptomatic catheter-related DVT


Primary Outcomes

Catheter-related bloodstream infection or symptomatic DVT
Femoral vs. subclavian: 25 vs. 8 (HR 3.5; 95% CI 1.5-7.8; P=0.003)
Jugular vs. subclavian: 23 vs. 11 (HR 2.1; 95% CI 1.0-4.3; P=0.04)
Femoral vs. jugular: 33 vs. 30 (HR 1.3; 95% CI 0.8-2.1; P=0.30)

Secondary Outcomes

Bloodstream infection
Femoral vs. subclavian: 11 vs. 4 (HR 3.4; 95% CI 1.0-11.1; P=0.048)
Jugular vs. subclavian: 13 vs. 6 (HR 2.3; 95% CI 0.8-6.2; P=0.11)
Femoral vs. jugular: 15 vs. 21 (HR 0.9; 95% CI 0.5-1.8; P=0.81)
Catheter tip colonization
Femoral vs. subclavian: 107 vs. 39 (HR 3.4; 95% CI 2.4-5.0; P<0.001)
Jugular vs. subclavian: 104 vs. 42 (HR 2.5; 95% CI 1.7-3.5; P<0.001)
Femoral vs. jugular: 145 vs. 121 (HR 1.6; 95% CI 1.2-2.0; P=0.003)
Symptomatic DVT
Femoral vs. subclavian: 15 vs. 5 (HR 3.4; 95% CI 1.2-9.3; P=0.02)
Jugular vs. subclavian: 10 vs. 6 (HR 1.8; 95% CI 0.6-4.9; P=0.29)
Femoral vs. jugular: 20 vs. 9 (HR 2.4; 95% CI 1.1-5.4; P=0.04)

Adverse Events

Major mechanical complications (pneumothorax, arterial injury, hematoma)
Femoral vs. subclavian: 6 vs. 18 (HR 0.3; 95% CI 0.1-0.8; P=0.03)
Jugular vs. subclavian: 12 vs. 22 (HR 0.5; 95% CI 0.3-1.1; P=0.09)
Pneumothorax: 4 vs. 14 (statistics not given)
Femoral vs. jugular: 7 vs. 13 (HR 0.5; 95% CI 0.2-1.4; P=0.19)

Subgroup Analysis

  • None of the preplanned subgroup analyses showed a significant interaction for the primary outcome
  • For mechanical complications, there was a significant interaction between the use of ultrasonography and the comparison between the femoral and jugular group (P=0.007) with the difference increasing when ultrasound was not used to guide catheter insertion
  • For mechanical complications, there was a nonsignificant trend for an interaction between the use of ultrasonography and the comparison between the femoral and subclavian group (P=0.07), with the difference increasing when ultrasound was not used to guide catheter insertion


  • By not including mechanical complications in the primary outcome, it is difficult to determine whether the reduction in the primary outcome achieved by subclavian access is outweighed by an increase in mechanical complications.
  • The short-term nature of the study precludes analysis of longer-term and potentially more relevant outcomes such as mortality and length of ICU stay.
  • Ultrasound guidance was not randomized and therefore its effect is uncertain. Results may not be generalizable to settings in which ultrasound guidance is either uniformly used or unavailable.


  • Supported by a grant from the Hospital Program for Clinical Research, French Ministry of Health

Further Reading