The company conducted a search of the health economics literature on balloon sinus dilation using XprESS or equivalent systems and functional endoscopic sinus surgery (FESS). This identified 134 papers, 6 of which were included in the company's submission.
The external assessment centre (EAC) judged the company's search terms to be appropriate. However, it noted: inconsistencies in the search terms across the databases searched; that the company's submissions did not provide search terms for its searches of the Cochrane database or the NHS Economic Evaluation Database; and it considered that the company's searches would have benefited from the inclusion of a wider range of databases, such as the cost‑effectiveness registry. The EAC re‑ran the company's searches and identified no additional studies. The EAC concluded that none of the economic studies identified was relevant to the decision problem.
The company presented a decision tree model to capture costs and outcomes in the first year following sinus surgery and a Markov model out to 5 years after sinus surgery, applying a 1‑year cycle length.
Patients entered the model needing sinus surgery, and could be routed to either FESS or XprESS. The model base case used a theoretical patient with multiple sinuses treated in a single episode of care. The first phase of the decision tree captures differences in treatment costs. The next stage covers the first 3 months following surgery, during which there is sustained recovery or a need for GP visits; either scenario could need readmission to secondary care. Surgical re‑interventions and GP visits are also included from 3 months to 12 months. Irrespective of these outcomes, patients then enter the Markov model out to 5 years which consists of 2 mutually exclusive states, surgery revision or sustained recovery. Surgery revision is an absorbent state, meaning that patients cannot leave it, so it is assumed that patients could have only 1 revision surgery over the study period. Death is not included because it was expected to be very rare over the time horizon modelled.
Figures for clinical parameters were obtained from published literature, expert opinion and England and Wales audit data. The company relied heavily on the audit data published by Brown et al. (2003) to determine the base values for FESS. It then used US data reported in Chandra et al. (2016) to determine the relative values for XprESS in relation to FESS.
The cost for FESS and XprESS surgery under general anaesthesia was based on staff costs for a nurse and surgeon, bed day costs, theatre time, device and surgical consumable costs. The total cost for a FESS surgery under general anaesthesia (including equipment costs of £300) was calculated to be £2,894. The total cost for XprESS surgery (including device costs of £900) was calculated to be £1,884. The equivalent costs under local anaesthesia were calculated by applying a ratio of 0.631 to the surgical costs under general anaesthesia reported in Zilvetti et al. (2009), providing costs for FESS of £1,936 and for XprESS of £1,520. These costs were also used in the model if the patient had a revision surgery.
The company reported a base‑case per‑patient cost of £2,679 for XprESS and £3,981 for FESS, representing an average saving of £1,302 per patient.
The company presented one‑way deterministic sensitivity analyses varying the model parameters from their base‑case level by 20%. The parameters with the biggest effect on the level of cost saving were device costs and procedure time for XprESS. The results of these analyses provided a range of cost savings, from £1,044 to £1,559.
Scenario analyses were done by changing parameter values for type of anaesthetic (from general only to include local), the percentage of patients having revision surgery each year, procedure time, length of hospital stay, and unit cost of theatre time. None of these altered the direction of the cost saving for XprESS, and at worst reduced it to £367, when a unit cost for theatre time of £6.40 per minute was used.
Break‑even analyses were conducted varying the procedure time with XprESS and FESS. The company reported that XprESS was cost neutral when the XprESS procedure time was 80 minutes or cost saving when the FESS procedure time was greater than 41 minutes.
The EAC noted the assumptions in the company's model and considered them to be largely appropriate. It did note some important omissions in the model tornado diagram, such as the unit cost of a FESS procedure. The EAC was also unable to replicate results in the tornado diagram for the monthly rate of GP visits beyond 3 months with FESS. The EAC considered the company's analyses of the structural uncertainties to be limited. It judged that it would have been appropriate to run the model assuming that there was no difference in GP visits and readmission in the first 3 months following surgery.
The EAC revised the company's relative risk estimates for revision surgery, based on their limited numbers in the REMODEL study. It judged the estimates for the values up to 12 months provided in the REMODEL trial to be more appropriate than those used by the company. Based on expert opinion and Philpott et al. (2015), the EAC considered that the evidence did not show any difference in revision surgery rates between FESS and XprESS beyond 12 months.
Based on expert opinion, the EAC judged the company's base‑case estimate of 0% for the proportion of XprESS procedures done under local anaesthesia to be conservative, and revised it up to 10%. It also revised the estimate for FESS procedures done under local anaesthesia to 2%, noting that this was consistent with the company's scenario analysis.
The EAC determined the costs of FESS and XprESS surgery using a bottom‑up approach. In the absence of published data, the EAC consulted experts to determine the duration of surgery for FESS in the patient population eligible for XprESS. Based on their responses, the average procedure times were 42.5 minutes for FESS and 26.7 minutes for XprESS. The FESS figure was consistent with figures quoted in a national audit and a health technology assessment report. The EAC revised the cost of operating time to £13.65 per minute based on data for ENT surgery (2014/15) reported by the Information Services Division Scotland. It also revised the length of stay in hospital following FESS to under 5 hours (0.208 days), and for XprESS to 4.17 hours (0.174 days) based on expert responses. The EAC revised the cost per day in hospital to £370 using a weighted average of 2014/15 NHS reference costs for elective inpatient excess bed days for minor sinus procedures (CA29Z), intermediate sinus procedures (CA28Z), major sinus procedures (CA23Z) and complex sinus procedures (CA26Z). Based on these figures, the revised cost of FESS under general anaesthesia was £657, and the cost of XprESS under general anaesthesia was £428 (not including device cost).
The EAC also revised the cost of FESS and XprESS under local anaesthesia in an operating theatre using a similar bottom‑up approach. Using averages based on expert advice, it estimated procedure lengths of 30 minutes for FESS and 31.7 minutes for XprESS, and in‑hospital stays of 3.00 hours for FESS and 2.17 hours for XprESS. Information Services Division Scotland operating theatre costs of £13.65 a minute were used to calculate operation costs. The hospital bed cost of FESS was calculated using the same methodology.
The EAC revised the cost of revision surgery for FESS and XprESS by applying weightings to the cost per procedure figures. The weightings applied for FESS were 98% general anaesthetic and 2% local anaesthetic. The weightings applied for XprESS were 90% general anaesthetic and 10% local anaesthetic. This gave a cost per revision surgery for FESS of £653 and for XprESS of £432.
The EAC revised the cost of a GP visit based on expert advice, the British National Formulary and data from the Personal Social Services Research Unit. It used a value of £37.00 per GP visit, and added drug prescription costs according to the clinical indication for the visit, leading to the following total costs per visit: blocked nose (£48.91), infection (£38.97 to £39.64), and blocked nose and infection (£50.00). The mean value of these figures produced an estimate of £46.00.
The company did not include any training costs for XprESS because it provides training at no extra cost, but the EAC judged that the costs for the staff time spent on training should be included in the model. It concluded that this amounted to 7 hours of a surgeon time at a cost of £106 an hour, leading to a total of £742 per surgeon. Over the duration of the economic model this was estimated to add £5.50 to the cost of each procedure.
The EAC used a bottom‑up approach to estimate the unit cost of XprESS done in an outpatient setting. Based on expert advice it used a length of a procedure of 31.7 minutes, and a length of stay in hospital of 2.17 hours. It used NHS reference costs of £370 for a hospital bed day, the Personal Social Services Research Unit for the costs of surgeon time and nurse time, and applied £115 for the costs of gown and a tray to produce a total estimate of £251.
The analysis based on the EAC's revised parameters found that XprESS was cost incurring by £330 compared with FESS (average per‑patient costs: XprESS £1,694, FESS £1,364). The EAC conducted univariate analyses on all the model parameters, varying their value by 20%. None of these analyses changed the direction of the results, and XprESS remained cost incurring. The main factors affecting cost were the device cost of XprESS and the unit costs of a FESS and XprESS procedure under general anaesthesia. This was consistent with the company's analysis.
The EAC conducted a series of univariate sensitivity analyses on the main model parameters. Sensitivity analysis on the length of FESS procedure under general anaesthesia demonstrated that XprESS became cost saving when the duration of FESS exceeded 66.0 minutes, compared with the EAC base case of 42.5 minutes. Analysis on the length of stay in hospital after FESS found that XprESS became cost saving when hospital stay was longer than 1 day. Further analyses showed that length of XprESS procedure under general anaesthesia had to be as low as 0 before XprESS became cost saving, and that no value for length of stay in hospital after XprESS under general anaesthesia changed the direction of the result. Analysis on the unit cost of theatre time demonstrated that XprESS became cost incurring when the unit cost exceeded £34 per minute (£2,040 per hour). Varying the unit cost of hospital stay had very little effect on the results, and the cost would have to reach an unreasonably high level for XprESS to become cost saving
The EAC conducted a number of scenario analyses. In the first of these, the EAC used hospital episode statistics data for length of stay, as per the company's model, of 0.97 days. In this scenario, XprESS remained cost incurring by a smaller margin of £136 per patient. The EAC considered a scenario in which XprESS was done in an outpatient setting, without theatre costs. The total procedure cost was £251. The proportion of procedures in an outpatient setting under local anaesthesia was varied between 0% and 100%, and the results showed that XprESS remained cost incurring even at 100%. The EAC also conducted scenario analyses in which:
it used a cost ratio of 0.631 between general and local anaesthetic (as used in the company's submission)
it used an annual revision rate of 3.5% between years 2 and 5, based on figures reported by Hopkins et al. (2009)
the cost of a hospital appointment for debridement of £162 (NHS reference cost, 2014/15) was added to each FESS procedure
it used a consistent proportion of 42% for patients visiting the GP in the first 90 days after the procedure for both treatments
it varied the rate of revision surgery for XprESS at 2 to 5 years after surgery.
In all cases, XprESS remained cost incurring. The EAC considered a scenario that included an extra appointment for debridement after FESS, and in which the rate of XprESS procedures done in an outpatient setting under local anaesthesia was varied. In this scenario, XprESS was cost saving when over 80% of procedures were done in an outpatient setting under local anaesthesia and when every FESS procedure needed a single extra hospital appointment for debridement.
The EAC did additional sensitivity analyses on the price of XprESS and FESS consumables. XprESS became cost saving when the price of the device is less than £586 per patient, and the cost of FESS consumables is more than £614 per patient. The EAC did a two‑way sensitivity analysis varying the price of XprESS and the length of a FESS procedure. XprESS was only cost saving when the device cost £800 or less and the FESS procedure takes more than 60 minutes. At prices above £800, the EAC stated that the length of time the FESS procedure would need to take in order for XprESS to be cost saving was increasingly implausible.
The committee was advised that the price of the XprESS device was the main factor influencing the economic model, and thought that this should also be its main consideration in the case for adoption. It heard from experts that the cost of the technology was a barrier to current adoption in the NHS. It heard from the company that the price is negotiable based on the volume of products used. For example, XprESS is available at a lower price of £820 per unit for centres that order 50 or more in a year.
The committee considered that the length of procedure with both XprESS and FESS was integral to the outcome of the cost modelling. Expert advice indicated that estimates of procedure length should include the time taken to administer anaesthetic. Experts indicated that the length of the FESS procedure will usually be the composite of the time taken to administer general anaesthetic as well as to undertake the surgery. For XprESS, experts indicated that this will usually be the composite of the time taken to administer and wait for local anaesthesia to take effect as well as performing the balloon dilatation.
The committee heard from experts that the greater use of XprESS could change the care pathway by allowing chronic sinusitis to be treated earlier, and potentially avoiding the need for FESS. Patients who have XprESS are also able to return to work on the same day. The committee heard expert advice that these factors may result in additional cost savings that were not considered in the model.
The committee carefully considered the plausibility of the EAC scenario in which XprESS is cost saving (that is, when more than 80% of procedures are done in an outpatient setting under local anaesthesia and assuming that every FESS procedure needs an extra appointment for debridement). The committee was advised by experts that patients in the NHS do not usually have a follow‑up debridement appointment after FESS, and so concluded that this scenario is unlikely to be widely applicable.
The committee was advised that if XprESS were more widely adopted, many patients currently having FESS could instead have XprESS.
The committee considered the cost case for XprESS to be uncertain. It concluded that any cost savings were dependent on the length of the FESS procedure, the cost of the device, and the proportion of XprESS procedures done in an outpatient setting under local anaesthesia. The committee encouraged further research on the resource consequences of using XprESS for treating chronic sinusitis.