Results
Search Results
The search strategy identified 944 articles; with 943 identified using the PubMed MEDLINE and EMBASE searches (See online appendix A) as of June 2011, and one article found using a bibliography search. Through the review of titles and abstracts, 59 were found to be relevant by title or abstract, and 11 articles were specific to central line placement in the ED. Ten of these articles were excluded in our review, as five articles were reviews or letters, two were case reports, two were not randomised and one failed to have success rates reported (See figure 1). Thus, one single study by Leung et al was left which included 130 patients (65 ultrasound- vs 65 landmark-guided) selected for internal jugular vein placement.
(Enlarge Image)
Figure 1.
Flow diagram of selection process to obtain studies for review. This figure is produced in colour in the online journal—please visit the website to view the colour figure. ED, emergency department.
Study Characteristics
Table 1 contains the study characteristics for our single reviewed trial by Leung et al.
The landmark-guided technique was performed using either the anterior or posterior method, depending on operator experience or preference. The ultrasound-guided internal jugular vein catheters were inserted using a portable 10–5 MHz 38-mm linear array transducer (Sonosite 180: Sonosite Inc., Bothell, Washington, USA). The probe was placed perpendicular to the vessels at the apex of the triangle between the two heads of the sternocleidomastoid and the clavicle. The primary outcome was comparison of CVC successful insertion on first or subsequent attempts between landmark- and ultrasound-guided techniques. Failure was defined by either the inability to locate or puncture the internal jugular vein or inability to feed the guidewire or catheter. The correct position of CVC was confirmed by chest radiography at the conclusion of the procedure. Crossover to the alternative technique occurred after three missed attempts to the assigned technique. If both attempts failed, an alternate site (contralateral internal jugular, subclavian or femoral vein) was attempted. Secondary outcomes were also studied: number of attempts, access times, and complications such as haematoma, pneumothorax, carotid artery puncture and nerve injury. The operator completing the procedures collected all the data using a preformatted sheet.
Quality Assessment of the Included Studies
We used the published standard criteria for reporting of randomised clinical trials studies (CONSORT) to evaluate the quality of the studies selected for this review. We specifically focused on randomisation, concealment, blinding, intention-to-treat, baseline comparisons and follow-up. Table 2 summarises the potential sources of bias in our trial under review.
Randomisation was done by computer block randomisation. Allocation assignments were concealed in serial numbered opaque envelopes in each case. Concealment and blinding could not be possible for this study and therefore a potential for ascertainment bias existed. To mitigate ascertainment bias, patients were enrolled before randomisation and the operator was chosen before the insertion technique was known. Biases related to lack of intention-to-treat and incomplete follow-up do not appear to be significant problems for this trial.
Comparison of baseline characteristics such as age, gender, obesity, coagulopathy, previous CVC and vital signs do not appear significantly different between the ultrasound and landmark technique groups. The experience of the operators was also compared between the study groups. Experienced operators were defined as performing greater than 25 successful internal jugular catheterisations without supervision. There was no significant (p=0.60) difference in distribution of experienced operators between the landmark (49%) and ultrasound (45%) groups. Experienced compared with inexperienced operators had similar success rates on first (experienced 85% vs inexperienced 84%) and subsequent (experienced 88.5% vs inexperienced 69%) attempts at CVC insertion. The statistically insignificant difference in success rates between experienced and inexperienced operators maybe the result of a type II error. This is true for both the univariate and logistic regressions used to analyse the effects of operator experience on the primary study outcome.
The question of the study's power and the subsequent potential for bias are related to the authors' estimation of sample size and use of an early stopping rule. The authors estimated a total sample size of 360 patients were required (p=0.05; power =80%) to find a difference in CVC insertion success rates between the landmark (85%) and ultrasound (95%) techniques. An interim analysis was planned after 100 patients were entered with a p<0.01 chosen as the cut-off for ending the trial early. Sixty-five patients were obtained in each group for a total enrolment of 130 patients. It is unclear whether or not the early termination of this trial could have induced bias in the results.