The manufacturer did not perform a de novo clinical data submission and synthesis. In place of this, they submitted a recent, peer-reviewed systematic review publication [8]. All results can be seen in the Perera et al. [8] manuscript and will not be reproduced in this article. However, the EAC findings are generally supportive of, and in accordance with, those in the systematic review.
External Assessment Centre (EAC) Clinical Data Synthesis
An independent literature search, performed by the EAC, did not identify any new published clinical studies on the Urolift device. We excluded a single study by Delongchamps et al. [9] as it was a non-English language publication with only four patients and was not deemed pivotal. We included the Abad et al. study [10] (professionally translated by Languages For Business Ltd., Cardiff), which was originally excluded by Perera et al. [8], as it lacked standard deviations (SDs). The EAC data synthesis was able to include data lacking SDs. All included studies in the EAC analysis are listed in Table 1.
Table 1 All included studies in the External Assessment Centre (EAC) analysis
The EAC combined results from the following studies, as they reported different aspects of the same series of patients:
-
1.
Chin et al. [11] and Woo et al. [12] reported urological and sexual function outcomes, respectively, from the same case series.
-
2.
Roerhborn et al. [13, 14] and McVary [15] all report on the LIFT study.
At the time of this literature search, there were no studies comparing Urolift with either TURP or HoLEP. In order to provide some comparative context for the NICE Medical Technologies Advisory Committee (MTAC) (and more fully comply with the scope for this assessment), the EAC performed a rapid pragmatic data synthesis.
The EAC’s solution was to find a TURP versus HoLEP systematic review, and extract relevant outcome data from their identified sources. A systematic review search led to the selection of a review by Li et al. [17]; because it was a very recent systematic review (July 2014) and it is listed on the PROSPERO website at The University of York Centre for Reviews and Dissemination (CRD) [18]. The EAC took the publications in the systematic review and updated them where possible (and where reported results allowed). The studies are listed in Table 2.
Table 2 Notes on Transurethral Resection of Prostate (TURP) versus Holmium Laser Enucleation of Prostate (HoLEP) randomised controlled trial (RCT) studies identified by Li et al. [17]
Table 3 shows the baseline comparisons between these studies and those identified for Urolift. The patient age and IPSS baselines all fall within the same range. The prostate volume range is wider in the TURP/HoLEP RCT studies, particularly skewed slightly towards men with larger prostates. Similarly, the Q
max baselines are skewed slightly towards slower flow rates in the baselines of the TURP/HoLEP RCTs.
Table 3 Baselines comparison between Urolift studies and Transurethral Resection of Prostate (TURP) versus Holmium Laser Enucleation of Prostate (HoLEP) randomised controlled trials (RCTs) from Li et al. [17]—data expressed in ranges
Data from all the published studies (Urolift and the TURP/HoLEP RCTs) were extracted by one EAC researcher and independently checked by a second. Table 4 shows each outcome measure, with the minimal clinically significant differences in each. This is sourced from publications where available, but in the absence of this, the EAC also consulted Expert Advisers. Weighted mean changes from baseline in each outcome measure are reported. We used this method of presentation to retain the original units of each outcome measure for clarity.
Table 4 Overview of Urolift, TURP and HoLEP results
In order to provide the NICE advisory MTAC committee with some context to judge the results, the EAC sought out published minimally important differences in each of the reported outcome measures. These are available for questionnaires such as IPSS and IIEF, as they go through a validation and testing process during development.
Where published sources were not available or unsuitable (PVR, for example), the Expert Advisers were surveyed by the EAC for their opinion on the minimum clinically significant differences in each outcome reported.
The pragmatic indirect comparison suggests the following: From similar baseline scores, both TURP and HoLEP give much better improvement in the IPSS score (including QoL, as these scores are linked) at all time-points, with Urolift giving an improvement of −9.22 to −11.82, TURP providing −17.34 to −19.70 and HoLEP −17.68 to −20.88. BPHII scores are not reported in the TURP and HoLEP studies, but as a prostate symptom score, it should give general improvements in agreement with IPSS scores.
Q
max improvements are higher at all time points for both TURP and HoLEP, with Urolift giving a +3.53 to +4.16 ml/s improvement from baseline. TURP provides a +14.11 to +23.20 ml/s improvement, and HoLEP +15.29 to +23.10 ml/s.
TURP and HoLEP give better improvements in PVR, but this is less widely reported in both the Urolift studies and the TURP/HoLEP studies. It may be worth noting that one Expert Adviser questioned the importance of PVR as an outcome measure for Urolift, and presumably other surgical treatments for BPH. This validity of PVR as a reliable outcome measure is also questioned in NICE CG97 [3].
Sexual function is poorly reported in the TURP and HoLEP papers (their aim is symptom improvement, so sexual function is secondary, and a complication), and therefore it is difficult to ascertain the impact of these interventions on erectile and ejaculatory function. A Expert Adviser recommended the GOLIATH study for more reliable IIEF-5 reporting post-TURP up to 12 months. GOLIATH patients were measured as 13.7 ± 7.2 at baseline, and 14.1 ± 8.2 at 12 months post-TURP, showing no significant changes in a cohort of 119 patients [29]. Another Expert Adviser recommended the 6-year follow-up on HoLEP by Gilling et al. [30] for sexual function post-HoLEP; and a 76 % retrograde ejaculation rate is reported, which was confirmed by surveying our clinical advisers (estimates ranged from 70–80 %). IIEF improvement from baseline was not reported.
Complications reported should also be interpreted cautiously and in the knowledge that there are no truly comparative studies between Urolift and TURP or HoLEP. One weakness of this type of comparative approach is that the Urolift studies report a different set of complications than those reported for TURP versus HoLEP RCTs, and with good reason: Urolift complications seem to be typically mild, such as transient dysuria or haematuria. Presumably, dysuria and haematuria are mild, yet expected, occurrences with TURP and HoLEP.
Manufacturer’s Economic Submission
No published economic studies of Urolift were identified by the manufacturer or the EAC, in independent literature searches.
The manufacturer presented comprehensive de novo economic model for their economic submission. The manufacturer’s de novo model structure is a decision tree, with seven executable arms, one for each technology or comparator. Only four of these are relevant to this assessment according to the scope: Urolift, mTURP, BiTURP and HoLEP. The sponsor’s submission was from the NHS and personal social services perspective and presents a 2-year time horizon.
Following treatment, the outcomes are success or failure. Success is defined as “>10 % improvement in IPSS within 12 months”, and the probability with each in-scope treatment is: Urolift: 89.80 %, mTURP: 94.00 %, HoLEP: 96.71 % and biTURP: 94.0 %. The success category then has options for relapse or no relapse: Urolift: 0.00 %, mTURP: 0.17 %, HoLEP: 0.32 % and biTURP: 0.99 %. The relapse option then has success or failure outcomes. The failure outcome has options for re-treatment (with success or failure outcomes) or no re-treatment.
The model includes costing for the following complications: Incontinence, urinary retention, urinary tract infection (UTI), stricture, TUR syndrome, decrease in erectile function, increase in erectile function and ejaculation dysfunction.
The base case assigned a cost of £2342 per patient for Urolift (based on 2014 prices). This was slightly cost incurring, by £3, compared to monopolar TURP (£2339 per patient), by £38 compared to bipolar TURP (£2302) and by £418 compared to HoLEP (£1924 per patient). These figures are shown in Table 5 alongside the EAC’s sensitivity analysis and input testing.
Table 5 External Assessment Centre (EAC) input testing and sensitivity analysis—bold type indicates where Urolift is cost saving or cost neutral
The key drivers of the model are the number of Urolift implants used, operating time and length of stay.
Critique of the Manufacturer’s Economic Model
The EAC found many of the manufacturer’s economic inputs to be appropriate and backed by published sources. The Urolift data were taken from the LIFT study [13–15] and Chin et al. [11]. Comparator data were taken from a health technology assessment (HTA) by Lourenco et al. [31].
The manufacturer’s inputs for post-Urolift length of stay (0.5 days) and procedure time (30 min) were based on the clinical opinion of three experts. A weighted mean procedure time of 59.6 min was calculated from the Urolift publications, but we were assured by Expert Advisers that this was ‘trial conditions’, and 30 min was a more appropriate input.
The number of Urolift devices is a key driver of the model. In the base case, the manufacturer has used 4 as the number of devices per procedure [11]. The EAC calculated the weighted mean number of implants from all of the clinical studies and found this to be 4.4 devices per procedure.
Blood transfusion is not likely to be required when using Urolift, based on the clinical evidence in this assessment. The manufacturer overestimated the cost of blood transfusion as £862.17 per transfusion for the comparators. This is a top-down costing based upon NICE CG97 [3, 32]. This provides a cost of £635 in 2003, inflated by the manufacturer to current value of £826.17. This also includes an additional day’s length of stay. The EAC estimates the cost of blood transfusion as £329. One unit standard red cells = £121.85 [33]. The mean number of units per transfusion is estimated to be 2.7 units of red blood cells when transfusion is required [32]. Therefore the EAC calculates 2.7 × £121.85 = £329 per transfusion. The probability of blood transfusion for Urolift in the model is zero; therefore, this change reduces the cost of the comparators, but not Urolift.
The unit cost of hospital stay was taken from published Scottish data for urology specialty in-patient costs [34], divided by the average length of stay (3.3 days) to give the unit cost per day in hospital. The excess bed day cost used in the model is calculated from the HRG code for TURP [35], minus the procedure costs included in the model. It is not clear which procedure costs were subtracted. The result is £331 in 2012 prices, which is inflated to £344 current price. The cost used in the model for hospital stay (0.5 days) for Urolift is calculated from 0.5 × £344 = £172. For comparison the EAC found the cost of an excess bed-day from the National Schedule of reference costs 2013–14 to be £294 (Excess bed day LB25F) [35].
EAC Revisions/Sensitivity Analysis of the Manufacturer’s Economic Model
We performed a number of input tests and sensitivity analyses where the published evidence or expert advice did not agree with those inputs used by the manufacturer’s model. For each, the single input was changed to assess its impact on the model.
As discussed in Sect. 3.2.1, the EAC substituted the manufacturer’s estimate of four Urolift implants, with the weighted mean of 4.4 implants. We tested a Urolift operative time of 60 min, in line with the weighted mean procedure time from the Urolift publications. We tested an mTURP procedure time of 66 min, taken from the EAC comparator studies. We included operating theatre costs for all procedures, using the cost of a urology operating theatre from NICE CG97 [3], stated at £9 per minute. We also tested a greater post-Urolift length of stay (LOS) range, from 0.25 to 1 days.
An extra Band 5 nurse was added to the TURP procedures, as Expert Advisers stated that an additional nurse is often needed to handle irrigation fluid. The impact of an additional ‘laser operator’ Band 5 nurse was also tested for HoLEP.
The EAC changed the cost of blood transfusion in the model from £862.17, which includes double counting of one additional day in hospital to the EAC estimate of £329.
We included a £10 per procedure cost for capital equipment for TURP (total capital cost £20,799 used both mTURP and biTURP) as the manufacturer did not include the capital cost in the base case.
We updated the cost of TURP consumables to £56.80 to account for roller and ball electrodes and a return electrode plate (return plate for mTURP only). HoLEP fibres were tested in a single-use scenario, with a price of £368.61 for single-use HoLEP fibres. All prices were taken from the NHS Supply Chain. We were also able to perform a sensitivity analysis for reusable HoLEP fibres, at a cost of £1207.42 (NHS Supply Chain). This was used as an upper-limit sensitivity analysis for this input.
All of these analyses, including the manufacturer’s base case, are presented in Table 5.
Additional scenario modelled by the EAC
Urolift can be performed as a day-case, whereas TURP is performed as an inpatient procedure – this was confirmed as a realistic UK practice by our clinical Expert Advisers. This scenario relies upon a number of specific inputs, requiring only 0.125 days (3 h) length of stay in total, a 30-min procedure time for Urolift and a 66-min procedure time for TURP. The scenario includes urological theatre overhead time and the more realistic cost of blood transfusion of £329, as mentioned in Sect. 3.2.1. The model inputs are detailed in Table 6, and the EAC Scenario cost results are shown in Table 7.
Table 6 ‘Urolift as day case’ EAC scenario inputs and conditions
Table 7 External Assessment Centre (EAC) scenario cost results