5.1 The company identified 2 health economic studies in its submission (Adams 2013; Stewart 2013). Both studies were cost‑comparison studies from outside the UK healthcare system. The company noted that these studies were of low quality and limited relevance. The External Assessment Centre agreed with the company's assessment of the studies, and did not identify any additional relevant studies.
5.2 The company submitted a de novo cost analysis comparing the cost consequences of using the Sherlock 3CG Tip Confirmation System (TCS), both with and without confirmatory chest X‑ray, for both blind bedside insertion with confirmatory chest X‑ray of a peripherally inserted central catheter (PICC) and insertion using fluoroscopy. Costs were modelled from an NHS and Personal Social Services perspective. The population included in the model was adult patients needing a PICC, for whom the Sherlock 3CG TCS was suitable (that is, adult patients needing PICC insertion who had an identifiable P wave). Patients for whom it was difficult to identify a P wave (see section 2.3) were not included in the model. The model used a decision tree structure, presenting all clinical pathways of patients having PICC insertion. All patients exited the model with an accurate insertion. The model was cost‑based and did not include any health states. The time horizon was limited to the time taken to successful insertion.
5.3 The company used parameters derived from Parikh et al. (2012) and Adams (2013) and resource‑use figures presented in Walker et al. (2013) to inform its model. The model used different accuracy rates for the Sherlock 3CG TCS (96%), blind bedside insertion (93%) and fluoroscopy (100%). In cases where initial insertion was unsuccessful, all reinsertions were performed under fluoroscopy. The Sherlock 3CG TCS was considered to be suitable for 83.5% of the patient population. The company's model only considered patients for whom the technology was suitable, and not the estimated 16.5% of patients with an altered cardiac rhythm for whom the ECG component may be unreliable.
5.4 The price of the technology (£9990 excluding VAT) was calculated to be £6.39 per PICC inserted, based on the assumed patient population of 468 potential uses per year, spread over a 4‑year lifespan. The company also reported the cost of consumables (£189.91), maintenance (£1.52) and training (£1.42), and other costs for each insertion using the Sherlock 3CG TCS. The overall cost of each insertion using the Sherlock 3CG TCS with X‑ray was estimated to be £310.15, and without X‑ray to be £272.30. The company calculated the cost of blind bedside insertion to be £274.33, and insertion under fluoroscopy to be £814.93.
5.5 The results of the company's base case suggested that the Sherlock 3CG TCS without X‑ray confirmation was associated with a cost of £304.90 per patient, assuming that 96% of all placements were successful and that reinsertions were done under fluoroscopy. Based on this result, the technology was associated with a cost saving of £25.66 compared with blind PICC insertion with X‑ray confirmation and a cost saving of £510.03 compared with PICC insertion with fluoroscopy.
5.6 The company carried out extensive sensitivity analyses to test the structural assumptions underlying its base‑case model, and to identify the key drivers. The company acknowledged the limitations of the available evidence base, but considered that the extensive sensitivity analyses mitigated this somewhat. The cost of a PICC insertion and the success of placement at initial insertion were identified by the company as the key drivers of the cost model. The company's threshold analysis reported that the Sherlock 3CG TCS became cost‑incurring with less than 93% successful placement, but also became cost‑incurring if blind placement had a success rate greater than 96%. When considering the Sherlock 3CG TCS compared with insertion using fluoroscopy, the company found the Sherlock 3CG TCS to be always cost saving across the parameters considered. The company carried out scenario analyses, testing parameters such as the proportion of failed insertions, the proportion of successful reinsertions after an initial misplacement, and a variety of changes to the costs presented in the base case. The Sherlock 3CG TCS without confirmatory X‑ray remained cost saving in all scenarios identified by the company, except when the costs associated with the Sherlock 3CG TCS itself were increased by 25% (incurring an additional cost of £50.20).
5.7 The External Assessment Centre did not report any major concerns with either the structure of the company's model or its parameters, although it reported that the lack of evidence made it difficult to be confident about the cost‑model results, both from the company's analysis and its own revised analysis.
5.8 The External Assessment Centre considered that some of the clinical parameters and inputs into the company's model needed revisions to ensure their accuracy and completeness. The model did not include the setup costs of a bedside insertion service for hospitals currently using a fluoroscopy service. The External Assessment Centre noted that the scope of the company's economic submission contained a deviation from that specified by NICE and from the clinical evidence submitted. It specified that the patient population was only those for whom the Sherlock 3CG TCS is suitable, which overlooked the proportion of the population needing PICC insertion for whom the Sherlock 3CG TCS is not suitable. The External Assessment Centre also reported that a significant factor in the company's cost analysis was the time taken by a nurse to perform a bedside PICC insertion. The company's base‑case model assumed that a blind bedside insertion took the same time as a bedside insertion using the Sherlock 3CG TCS plus confirmatory X‑ray (62.49 minutes, based on Walker et al. ). Bedside insertion using the Sherlock 3CG TCS without a confirmatory X‑ray was assumed to take 39.5 minutes (Adams et al. 2013). The External Assessment Centre considered the use of 2 different data sources to inform the same procedure in different arms of the model to be irrational. In its own analysis, nurse time was adjusted to ensure parity across both treatment groups (62.49 minutes; Walker et al. 2013).
5.9 The External Assessment Centre updated the parameters in the company's model to reflect alternative assumptions made:
It incorporated the additional costs of patients needing PICC insertion who are not suitable for the Sherlock 3CG TCS (16.5% of patients). These patients had not been accounted for in the original economic model, despite being specified in the scope.
The amount of nurse time need for PICC insertion was set to be equal for both insertion using the Sherlock 3CG TCS and blind bedside insertion. The External Assessment Centre noted that results of the model for bedside procedures were strongly driven by nurse time.
It set the standard reinsertion option for unsuccessful insertions to be reinsertion using the original method, instead of fluoroscopy, to reflect the clinical experts' advice. For example, a PICC that was misplaced using the Sherlock 3CG TCS would be reinserted using the Sherlock 3CG TCS.
The malposition rate for the Sherlock 3CG TCS with no X‑ray confirmation was set to 0% instead of 4%, on the basis that there was no way to confirm a malpositioned PICC in the time horizon of the model.
Theatre costs for fluoroscopy were reset from £507.18 to £101.00.
5.10 Results of the base case in the company's model when run with the External Assessment Centre's revised parameters suggested that the Sherlock 3CG TCS without X‑ray confirmation was associated with a cost of £302.63 per patient. At this cost, it became cost incurring by £9.37 compared with blind bedside insertion. It was still associated with a cost saving compared with PICC insertion under fluoroscopy (£106.12), although this was lower than in the company's base case.
5.11 The External Assessment Centre carried out sensitivity analyses to test the impact on the costs of the technology of the accuracy of placement using both the Sherlock 3CG TCS and blind PICC placement, because there had been considerable uncertainty surrounding the clinically realistic accuracy rates of both. The External Assessment Centre also carried out a 1‑way sensitivity analysis to test the impact of varying the nurse time associated with insertion, because this had been noted to be a key driver in the model.
5.12 The results of the sensitivity analysis surrounding accuracy rates showed that, if use of the Sherlock 3CG TCS was accurate in 100% of patients (confirmed using chest X‑ray), then it would become cost incurring if blind PICC placement was accurate in just over 87% of patients. If the Sherlock 3CG without X‑ray confirmation had a 100% accuracy rate, it was cost saving if blind bedside insertion was less than 89% accurate.
5.13 The sensitivity analysis surrounding nurse times explored the impact of varying the nurse time needed for insertion of the Sherlock 3CG TCS by ±20 minutes when the nurse time needed for blind PICC insertion was 30 minutes and 80 minutes. The results showed that the factor which made the most impact was the difference in nurse times between the 2 technologies, rather than the actual length of time allocated to the procedure, and that a 10 minute difference between nurse times could make the Sherlock 3CG TCS cost saving or incurring. The External Assessment Centre reported that there was no evidence available to state with certainty that the nurse times used as inputs in the model were definitive. Expert advice reported a wide variation in nurse time depending on clinical setting and patient population.
5.14 The External Assessment Centre also carried out a separate analysis based on the results for intensive care patients presented in the studies by Johnston et al. (2013, 2014). In this analysis, the Sherlock 3CG TCS with X‑ray confirmation was compared with blind PICC placement with X‑ray confirmation to reflect the available data. PICC reinsertion was done with the original method in all cases. The External Assessment Centre used effectiveness rates based on results that met European guideline requirements as reported in Johnston et al. (2013, 2014): specifically, 79.5% for the Sherlock 3CG TCS with X‑ray, and 49.2% for blind PICC placement with X‑ray. This analysis showed that use of the Sherlock 3CG TCS with confirmatory X‑ray compared with blind insertion with X‑ray was associated with a cost saving of £41.35 per patient. The External Assessment Centre considered that intensive care patients may be a subgroup for whom the Sherlock 3CG TCS holds particular benefit, given the higher rates of malposition associated with this patient population. However, it noted that the evidence may not be generalisable, because the data were historical and from a single centre, and the actual number of repositionings was not reported.
5.15 The External Assessment Centre reported that there were numerous uncertainties in the model structure and inputs due to the lack of data available. The model was limited by the lack of available evidence, which was exacerbated by large variations in clinical practice, and different patient groups and settings. No evidence was available to the company on the impact of identified malpositions, and it was therefore unknown if PICCs were repositioned or reinserted as a result. No comparative evidence was available on the rate of complications or adverse events. The External Assessment Centre presented an alternative set of assumptions in its analysis, but stated that the lack of information did not allow for absolute certainty over which were correct. The External Assessment Centre reported that, given currently available information, use of the Sherlock 3CG TCS compared with blind PICC insertion using a chest X‑ray appeared overall to be close to cost neutral.
5.16 Following discussions at the Committee meeting, the External Assessment Centre carried out additional analysis to assess more fully the impact of an increasingly streamlined care pathway, in particular as a result of the reduced need for X‑ray confirmation. It considered potential cost savings in areas associated with this, such as portering and X‑ray interpretation. The External Assessment Centre considered a scenario in which nurse time was slightly reduced, because there was no need for interpretation of an X‑ray, and where the radiologist and portering time associated with a typical X‑ray did not need to be included. Using these parameters, use of the Sherlock 3CG TCS without X‑ray compared with blind bedside insertion was associated with a cost saving of £1.16 per patient.
5.17 The Committee recognised that the uncertainties in the economic evidence and cost modelling assumptions were substantial. It was told by the External Assessment Centre that the company had carried out substantive and appropriate sensitivity analyses to address the problem of the poor evidence base.
5.18 The Committee considered that use of the Sherlock 3CG TCS was likely to be cost saving compared with fluoroscopy‑guided PICC insertion, based on the results in both the company's base case and the results of the External Assessment Centre's revised model parameters. The Committee considered the costs presented by the External Assessment Centre to be more realistic, and accepted its estimated cost savings of £106.12 per patient to be reasonable within certain clinical settings. The External Assessment Centre noted at consultation stage that the cost savings presented may be an overestimate in a clinical setting that only uses fluoroscopy‑guided PICC insertion, because of the additional service redesign costs and the need to train staff in bedside insertion. As a result of the substantial variation in clinical settings, and the different training costs which may apply depending on the setting, the exact effect of these changes on the estimated cost saving is unknown.
5.19 The Committee considered the estimated proportion of the patient population for whom ECG tip confirmation was not used, namely those patients in whom it is difficult to identify a P wave. The External Assessment Centre presented the Committee with the full range of unsuccessful tip confirmation rates reported in the clinical evidence, ranging from 4.4% to 29.9%, and noted that varying the figure of 16.5% did not have a substantial impact on the cost modelling. The Committee noted this summary of tip confirmation failure rates, and accepted the value used by the External Assessment Centre (16.5%, Adams et al. 2013) in the cost modelling as reasonable (see section 3.6).
5.20 The Committee considered the evidence presented on the use of the Sherlock 3CG TCS in an intensive care population. It noted input from clinical experts, who confirmed that accurate PICC insertion is more difficult in intensive care patients due to problems with positioning and comorbidities. The External Assessment Centre advised the Committee that the primary driver of cost savings is the relative difference in the accuracy rates between the Sherlock 3CG TCS and bedside insertion. The Committee accepted the estimated cost saving of £41.35 obtained in a scenario analysis using the revised model with parameters from the Johnston et al. (2013, 2014) studies.
5.21 With regard to the use of the Sherlock 3CG TCS compared with blind insertion with confirmatory X‑rays, the Committee considered that the outputs of the company's model using the External Assessment Centre's updated parameters were appropriate. It was advised that the removal of X‑rays from the care pathway led to increased efficiency of service and an improved patient experience. Depending on the exact clinical context and whether or not it is used with X‑ray, the use of the Sherlock 3CG TCS in adults who need a PICC in a non‑intensive care setting ranged from slightly cost incurring (£24) to slightly cost saving (£26) compared with blind insertion with confirmatory chest X‑ray. These results led the Committee to conclude that the technology was likely to be more or less cost neutral.