The appraisal committee reviewed the evidence from a number of sources (see appendix B).
Twenty randomised clinical trials (RCTs) were identified. Of these, six evaluated audio-guided Doppler ultrasound against the landmark method, thirteen evaluated 2D ultrasound guidance against the landmark method and one evaluated both audio-guided Doppler ultrasound and 2D ultrasound guidance against the landmark method. There were no trials that compared the use of ultrasound locating devices (ULDs) against the surgical cut-down method.
Insertion sites were the IJV (15 trials), SV (4 trials) or FV (1 trial). One trial did not specify the insertion point, and 1 investigated both the IJV and the SV as insertion sites. None addressed the placement of PICCs or ports, both of which can be considered types of CVCs.
For most of the trials, the setting within the hospital where the cannulation took place was not reported clearly. In 6 of the trials the central venous catheterisation took place in an ITU or trauma unit, and in 2 trials catheterisations took place in emergency rooms. In the 7 studies involving patients scheduled for cardiac surgery, the cannulation is most likely to have taken place on the way into theatre. In only 3 of the trials does it seem likely that CVCs were inserted on wards or in clinics.
The CVC procedure was carried out by anaesthetists in 7 studies and by other medical staff in 4 studies. One study involved 2D ultrasound-guided catheterisation by junior radiologists. None of the studies involved nurses. The remaining 9 studies did not make the specialty or profession of the operator clear. The range of experience of the operator, both with respect to medical career and use of the intervention, differed greatly from study to study. Six studies described the operators as having up to 5 years' postgraduate experience, 8 described them as having more than 5 years' experience, and 2 described them as varying in experience. Four trials did not record the career experience of the operator.
Pooled results from 7 RCTs suggested that real-time 2D ultrasound guidance was significantly better than the landmark method for all 5 outcome variables measured for insertions into the IJV in adults. Compared with the landmark method, 2D ultrasound guidance was associated with reduced risks of failed catheter placements (86% reduction in relative risk, 95% confidence interval [CI] 67% to 94%, p<0.001), catheter placement complications (57% reduction in relative risk, 95% CI 13% to 78%, p=0.02), and failure on the first catheter placement attempt (41% reduction in relative risk, 95% CI 12% to 61%, p=0.009), and fewer attempts to achieve successful catheterisation (on average, 1.5 fewer attempts, 95% CI 0.47 to 2.53, p=0.004).
The difference between the 2D ultrasound method and the landmark method in the time taken to insert a catheter successfully was small and not statistically significant (2D ultrasound-guided catheterisation was 20 seconds faster, 95% CI -83 to 124 seconds). However, there was significant heterogeneity for this endpoint (p<0.01), which indicated that it might not be appropriate to pool these results. In the study which reported the longest time to achieve a successful catheterisation, the time taken to set up the ULD was also included in the outcome measurement. When the analysis was repeated, excluding this study, heterogeneity was no longer significant and the pooled result from the included trials showed that catheterisation was, on average, 69 seconds faster (95% CI 46 to 92 seconds) with the ULD than with the landmark method, which was a highly statistically significant difference (p<0.001). It is acknowledged that the importance of this endpoint will vary between clinical situations.
Three trials evaluated the effect of 2D ultrasound guidance on the cannulation of the IJV in infants. In these trials, 2D ultrasound guidance was significantly better than the landmark method in terms of reductions in the risk of failed catheter placements (85% reduction in relative risk, 95% CI 36% to 97%, p=0.01), the risk of catheter placement complications (73% reduction in relative risk, 95% CI 8% to 92%, p=0.03), and the number of attempts required before catheterisation was successful (reduced by an average of 2, 95% CI 1.2 to 2.8, p=0.001). Using 2D ultrasound guidance, successful cannulation was achieved, on average, 349 seconds (95% CI -103 to 802 seconds) more quickly than with the landmark method, although this result was not statistically significant.
Only 1 RCT was identified that analysed the effect of 2D ultrasound guidance on SV catheterisation in adults. In the trial, in comparison with the landmark method, 2D ultrasound guidance was associated with reduced risks of catheter placement failure (86% reduction in relative risk, 95% CI 43% to 96%, p=0.006) and catheter placement complications (90% reduction in relative risk, 95% CI 29% to 99%, p=0.02). However, in this trial, the operators were relatively inexperienced in both the landmark method and 2D ultrasound guidance. The failure rate with the landmark method was 55%, which is higher than that reported in trials that involved more experienced operators (around 9% to 19%).
No studies were found that investigated the effect of 2D ultrasound guidance on SV catheterisation in infants.
One study was identified that evaluated the effect of 2D ultrasound guidance on femoral catheterisation in adults. In this trial, the operators took, on average, 2.7 (95% CI 0.1 to 5.3) fewer attempts to insert a catheter using 2D ultrasound guidance than using the landmark method (p=0.04). Compared with the landmark method, 2D ultrasound guidance reduced the risk of failed catheter placement and the time to successful catheterisation, but the differences were not statistically significant. No studies in infants were found.
No studies were found that investigated the effect of 2D ultrasound guidance on FV catheterisation in infants.
Four RCTs were found that compared audio-guided Doppler ultrasound guidance with the landmark method for IJV catheterisation in adults. Pooled results from these RCTs suggest that audio-guided Doppler ultrasound guidance was significantly better than the landmark method in terms of risk of failed catheter placement (61% reduction in relative risk, 95% CI 8% to 83%, p=0.03) and the risk of failure on the first catheter placement attempt (43% reduction in relative risk, 95% CI 12% to 63%, p=0.01). With the audio-guided Doppler ultrasound method, the risk of catheter placement complications was reduced (57% reduction in relative risk, 95% CI -5% to 83%) and there were fewer attempts to achieve successful catheterisation (0.6 fewer attempts, 95% CI -0.6 to 1.8); however, the differences did not reach statistical significance (p=0.06 and p=0.40, respectively) so they could have arisen by chance. It took, on average, 35 seconds longer (95% CI -54 to 124 seconds) to successfully insert a catheter using Doppler ultrasound guidance than it did with the landmark method, although this difference was also not statistically significant.
Only 1 trial was identified that studied the effect of audio-guided Doppler ultrasound in infants. The sample size of this study was small (n=29) and so it lacked statistical power. It failed to show any differences with the landmark method.
The pooled results from 3 RCTs (all involving adults) suggest that for SV catheterisation there was a significantly increased risk of failed catheter placement when the audio-guided Doppler ultrasound method was used compared with the landmark method (48% increased in relative risk, 95% CI 3% to 114%, p=0.03). In other words, the landmark method was preferable to the audio-guided Doppler ultrasound guidance technique. In contrast, the pooled results from 2 of the trials, which reported the risk of catheter placement, showed a 43% fall (95% CI 89% to 188%) in relative risk in the audio-guided Doppler ultrasound group, although this result was not statistically significant.
Only 1 study reported the effect of audio-guided Doppler ultrasound guidance on the risk of failure of the first catheter placement in adults. There was a slight increase (4%, 95% CI -24% to 43%) in the risk of catheter placement complications associated with the use of audio-guided Doppler ultrasound guidance compared with the landmark method, although this result was not statistically significant. Only 1 study recorded the effect of audio-guided Doppler ultrasound guidance on the number of attempts required to achieve successful catheterisation. This study found that an average of 0.4 (95% CI 0.2 to 0.6) fewer attempts were needed to achieve successful catheterisation with the audio-guided Doppler ultrasound guidance method compared with the landmark method, a highly statistically significant difference (p<0.001). The same study was the only one to record the effect of Doppler ultrasound guidance on the time to achieve successful catheterisation. Catheterisation using the Doppler ultrasound guidance method was significantly (on average, 209 seconds, 95% CI 175 to 242) slower than catheterisation using the landmark method (p<0.001).
No relevant economic evaluations were identified in the literature. Furthermore, none of the submissions made to NICE included economic evaluations.
The assessment group developed an economic analysis, based on the evidence from the systematic review of RCTs, to evaluate the cost effectiveness of 2D ultrasound guidance compared with the landmark method. This model is a simple decision analytic model, and is based on a theoretical cohort of 1,000 adult patients who required IJV cannulation before surgery and who had a low to moderate risk of complications.
This model adopted a set of conservative assumptions. It was assumed that the operators were experienced in using the landmark method; the time to achieve successful puncture was the same for both methods; complications were limited to arterial puncture; there was a 10-minute delay between the prior failure and the new attempt at another insertion site; there was a 100% success rate at the second insertion site; and each machine was used for 15 procedures per week.
The results of the assessment group's model suggested that the ultrasound guidance not only avoided 90 arterial punctures for every 1,000 patients treated, but also reduced costs by an average of almost £2 per patient. In other words, the 2D ultrasound guidance method was found to be both more effective and less costly than the landmark method.
A threshold sensitivity analysis was undertaken to examine by how much key variables in the model needed to change to make the ultrasound guidance method cost-neutral instead of cost-saving. The modelled result was most sensitive to the utilisation of the ultrasound equipment. The cost-saving result was eradicated if the number of ultrasound procedures assumed per machine per week was less than around 11, or if the number of ultrasound procedures carried out by an individual trained practitioner was less than around 3 per month on average.
Given that the model used relatively conservative estimates, the assessment group concluded that the results were probably generalisable to all anatomical catheter insertion sites, to infants, and to other sites within the hospital including the clinical wards.
The committee reviewed the evidence on both the clinical effectiveness and the cost effectiveness of ULDs for placing CVCs, having also considered the evidence from clinical experts. Furthermore, the committee was mindful of the need to ensure that its advice took account of the efficient use of NHS resources.
The committee took note of the fact that the evidence on the effectiveness of CVC placement into IJVs in adult patients was more robust than that available for other insertion sites. For infants, evidence was available only from trials that evaluated central venous catheterisation of the IJV, and there was very limited evidence on the use of this technology in very small infants (those weighing less than 3 kg). In addition, the economic analysis presented to the committee was based on an evaluation of the cost effectiveness of 2D ultrasound-guided elective CVC placement into the IJV in the operating theatre prior to surgery. The assessment report provided justifications for extrapolating this analysis to other settings including ward-based management, other sites of CVC insertion and also to CVC placement in infants.
Given the constraints outlined in section 4.2.2, the committee concluded that there was evidence of both the clinical and cost effectiveness of 2D imaging ultrasound guidance as an adjunct for placing CVCs in the majority of clinical scenarios, but that the degree to which this technology would be most suitably applied would vary according to the clinical situation and the competence/previous experience of the operator. In addition, there could be potential benefits for patients arising from reduced discomfort from the procedure and reduced risk of complications compared with the landmark method, particularly for IJV insertions.
The committee found the evidence for the use of audio-guided Doppler ultrasound guidance less satisfactory, and therefore concluded that the 2D imaging ultrasound guidance should be used in preference to audio-guided Doppler ultrasound guidance.
While accepting that, from a patient's perspective, 2D ultrasound imaging guidance in CVC insertion might be more appropriate and probably superior to the traditionally used landmark method in many circumstances, the committee also considered the financial and service implications of purchasing the required equipment and of training sufficient numbers of competent practitioners.
The committee also considered that although 2D ultrasound imaging guidance in CVC placement may eventually become the routine method for placing CVCs, the landmark method would remain important in some circumstances, such as emergency situations, when ultrasound equipment and/or expertise might not be immediately available. Consequently, the committee thought it important that operators maintain their ability to use the landmark method and that the method continues to be taught alongside the 2D-ultrasound-guided technique.