Economic Value of the Transurethral Resection in Saline System for Treatment of Benign Prostatic Hyperplasia in England and Wales: Systematic Review, Meta-analysis, and Cost–Consequence Model

2.1. Methods

An SR was performed to identify clinical trials to form the basis of a meta-analysis. The results of the clinical meta-analysis form one part of our objective. They were also used as input for an economic analysis, the second part of our objective.

2.2. Systematic review of the clinical evidence

The SR was performed in line with guidance by the Centre for Reviews and Dissemination, University of York [22]. The detailed SR methodology, as well as results and quality assessment, is reported in the supplementary information for MTG23 [20]. We performed an update to the original search on April 20, 2015, limited to randomised controlled trial (RCT) publications from 2014 onwards. The output from this update was then combined with that from the original SR (Table 1).

2.3. Meta-analysis of the clinical evidence

Expert opinion solicited during development of the MTEP submission suggested that differences are most commonly observed between TURis and M-TURP in the incidence of TUR syndrome, blood transfusions, clot retention, and hospital stay. The evidence was synthesised using pairwise meta-analysis techniques to determine any evidence-based differences between the treatments in these outcomes, as well as in other commonly reported outcomes including urethral strictures, catheterisation time, and procedure duration. According to the Cochrane Systematic Review Handbook, a minimum of 10 studies is required to draw meaningful conclusions from random-effects calculations [5]. Because each meta-analysis undertaken involved <10 studies, fixed-effects analyses were conducted. Details of the meta-analysis methodology were previously reported [20].

2.4. Economic evaluation

The economic analysis (a cost-consequence analysis) compared costs and outcomes associated with TURis and M-TURP, respectively, for surgical treatment of BPH. The perspective of the analysis was that of the National Health Service (NHS) and Personal Social Services in England and Wales.

2.5. Patient population

The analysis considered adult men with LUTS presumed secondary to BPH, in whom surgical intervention was indicated.

2.6. Model structure

A decision-tree model was constructed in which men with BPH-related LUTS underwent surgery with either TURis or M-TURP. Patients entered the model at the point of being indicated for surgery and were allocated probabilities of experiencing procedure-related adverse events (TUR syndrome, blood transfusion, and clot retention) based on the meta-analysis output. Procedure duration was excluded from the analysis because no statistically significant difference was observed between M-TURP and TURis in the meta-analysis. The pooled reduction in catheterisation time estimated from the meta-analysis of 0.2 d for TURis versus M-TURP was statistically significant. However, given that a patient's catheter would generally be removed while in the hospital, this reduction is not expected to result in cost savings beyond those realised through the reduction in hospital stay, which was modelled. As such, catheterisation time was not modelled explicitly. The model considered a nondefined short time horizon covering the prostate resection procedure and immediately afterwards—the period during which the complications modelled are expected to occur. Capital costs assumed equipment life spans of 7 yr. Based on findings of the SR, no difference in functional outcomes (eg, efficacy) was assumed between the procedures with respect to prostate resection weight or radicality.

2.7. Model parameters and data sources

Supplementary Table 1 lists all the variables included in the analysis. Sourcing of inputs, rationale, and further details are provided in Supplement 1.

2.8. Sensitivity analysis

Sensitivity analyses (based on predefined scenarios) were conducted. The first scenario considered all-cause readmission following a TURP procedure, based on data from the single study identified, Fagerstrom et al [23], which indicated a threefold increase in all-cause readmission for M-TURP versus TURis. This study reported readmissions as a result of bleeding, infection, and “other causes.” Because the base-case analysis included clot retention, the scenario considered readmissions from infection and other causes to avoid double counting (it is assumed that “bleeding” includes clot retention). A separate scenario considered a situation in which TURis is performed as day-case surgery.

One-way deterministic sensitivity analysis was performed in which all model parameters were varied either according to published ranges or by ±25%. Results are presented as Tornado diagrams. Probabilistic sensitivity analysis (PSA) was also conducted, in which input parameters were varied according to specified probability distributions. Threshold analysis was conducted to explore the relationship between key model parameters and model results up to the point at which the main conclusions of the model might change, that is, in this case to determine the value at which TURis would be cost neutral (ie, no longer cost saving) versus M-TURP.

3.1. Clinical evidence

The original electronic searches identified a total of 1116 publications. Following title and abstract review, 783 were excluded and 14 relevant RCTs remained, following assessment of the full paper. No unpublished studies were identified from ClinicalTrials.gov. The updated search identified a total 63 further publications, of which two full-text RCT publications remained after review of title and abstract [24] and [25]. On review of the full text, it was found that both of these publications reported data from the same trial, and only the most recent publication, reporting the most up-to-date data, was included in the analysis [24]. Figure 1 shows the study selection flow diagram.

Quality assessment of the identified RCT (Supplementary Table 2) showed that the patient population was consistent with the target population of the analysis. Study end points were clear and appropriately measured according to study design and follow-up schedule.

3.2. Meta-analysis

Six studies were found to be suitable for inclusion in the pairwise meta-analysis of the incidence of TUR syndrome [23], [26], [27], [28], [29], and [30]. The total number of events reported in the studies is presented in Table 2. There were no TUR syndrome events in the TURis arm reported in any of the studies, consistent with the use of saline. The analysis estimated a pooled relative risk (RR) for TUR syndrome of 0.18 (95% confidence interval [CI], 0.05–0.61; p = 0.006) in favour of TURis.

Eight RCTs reported the incidence of blood transfusion, but only seven of these were included in the meta-analysis [23], [24], [26], [27], [28], [29], and [31] (Table 2). The study by Michielsen et al was excluded from the analysis because a higher proportion of TURis procedures was conducted by trainee surgeons compared with M-TURP (32% vs 7%, respectively), resulting in increased procedure duration in this arm as well as increased blood loss and, consequently, blood transfusions [32]. Exclusion of this study on the basis of heterogeneity was accepted by the External Assessment Centre for MTG23 [19]. The results of the meta-analysis showed a statistically significant reduction in blood transfusions with use of TURis compared with M-TURP (RR: 0.34; 95% CI, 0.18–0.61; p = 0.0003).

Six studies were identified for meta-analysis of the incidence of clot retention [24], [26], [28], [29], [31], and [32]. The pooled RR was 0.43 in favour of TURis (95% CI, 0.22–0.86; p = 0.0161), demonstrating that use of TURis is associated with fewer cases of clot retention than M-TURP (Table 2).

Meta-analysis was conducted on seven studies reporting data on urethral strictures [23], [24], [26], [28], [29], [31], and [33]. A non-significantly different pooled RR of 1.27 (95% CI, 0.85–1.9; p > 0.05) was observed for TURis versus M-TURP (Table 2). As a result, this outcome was not considered in the economic analysis.

Further meta-analyses suggested that TURis is associated with statistically significant reductions in hospital stay and catheterisation time (Table 2). For these outcomes, meta-analyses were restricted to studies reporting both the mean duration and standard deviation. From a pooled analysis of four studies reporting length of hospital stay [24], [26], [27], and [33], the mean difference for TURis versus M-TURP was −0.56 d (95% CI, −0.77 to −0.35; p < 0.0001). Analysis of four studies presenting data on catheterisation time [24], [26], [27], and [33] resulted in a statistically significant mean difference for TURis versus M-TURP of −0.23 d (95% CI, −0.38 to −0.09; p = 0.0019); however, this difference is not clinically meaningful because such a minimal reduction in catheterisation time is unlikely to make a considerable difference to patients. Six studies reported procedure duration [24], [26], [29], [31], [33], and [34] and, although not significant, the resulting pooled mean difference for TURis versus M-TURP was −0.95 min (95% CI, −3.35 to 1.46; p > 0.05).

3.3. Economic analysis

Only statistically significant differences in clinical outcomes in the meta-analysis, which directly affect costs, were considered in the base-case analysis: TUR syndrome, blood transfusions, clot retention, and length of hospital stay. Table 3 shows the results of the base-case analysis. Costs of adopting TURis differed according to whether or not hospitals owned an Olympus monopolar system. Based on 150 TURP procedures per annum, capital costs of TURis (Supplementary Table 1) for hospitals with existing Olympus systems and for those without Olympus systems were £9.68 and £29.13 per patient per year, respectively. The total cost per patient (including capital equipment, consumables, hospital stay, and complications) of using TURis for hospitals with Olympus systems in place was £881.71 versus £1085.49 for M-TURP, a cost saving of £203.78 per patient. In hospitals without Olympus systems, the total per patient cost of using TURis was £901.16 compared with £1014.58 for M-TURP, a cost saving of £113.42 per patient.

Higher equipment costs for TURis versus M-TURP were offset by savings from reductions in hospital stay and complications. Assuming a reduction in hospital stay of 0.56 d from the meta-analysis, the base-case analysis found that TURis was associated with a £165.02 reduction in the cost of hospital stay per patient versus M-TURP. TURis was also associated with a £71.82 per patient reduction in complication costs compared with M-TURP.

A further potential benefit of the TURis system is the option of conducting surgery on a day-case basis. Some UK centres are already adopting this practice [35] and [36], and the potential for further cost savings in hospital stay costs via this practice was acknowledged by NICE in MTG23 [20]. Although in clinical practice various factors influence the implementation of day-case surgery, testing a scenario reflecting this situation, in which patients undergoing TURis incurred no inpatient day costs, cost savings were £891.36 and £801.00 per patient, for Olympus and non-Olympus centres, respectively.

Considering readmissions in a further scenario, TURis was associated with annual cost savings versus M-TURP of £245.10 and £154.75 per patient, for Olympus and non-Olympus centres, respectively (Table 3). Readmission for BPH is not widely reported, thus data relating to this outcome are limited; the scenario analysis was based on a single study [23].

Sensitivity analysis also demonstrated that TURis remained cost saving versus M-TURP when each tested model parameter was varied according to its defined range. Key drivers of cost savings were reduction in hospital stay and inpatient day costs (Fig. 2). Results of PSA demonstrated that the cost-consequence analysis was robust across both types of centre. For Olympus centres, use of TURis was associated with a mean saving of £200.77 per patient (95% CI, −£115.54 to −£298.67) and remained cost saving in all 1000 simulations. For Olympus centres, the mean saving was £112.93 per patient (95% CI, £−32.94 to £−211.22), and TURis remained cost saving in 99.8% of simulations.

Results of the threshold analysis on model parameters (Table 4) confirmed the robustness of the base-case results. Existing Olympus centres would need to perform as few as seven procedures annually for TURis to become cost saving versus M-TURP. For non-Olympus centres, the number is 31. Keeping all other model parameters constant, TURis would need to be associated with significantly increased risks of blood transfusion and clot retention compared with M-TURP to cease being cost saving overall. Based on the findings of the meta-analysis, which showed statistically significant reductions in these events for TURis versus TURP, it is not considered plausible that RRs as high as those reported in Table 4 would be observed for TURis versus M-TURP.

Significant increases in the annual cost of capital equipment for TURis would be required for TURis to cease being cost saving versus M-TURP. This could only happen as a result of short equipment life span in practice because prices are generally stable over a period such as the modelled time horizon. The shortened life spans that would be required for TURis to cease being cost saving versus M-TURP (0.28 yr for existing Olympus centres and 1.30 yr for non-Olympus centres) demonstrate the robustness of the conclusions of the analysis to changes in TURis capital costs. TURis would cease being cost saving if the reduction in hospital stay for TURis versus M-TURP was removed (only for non-Olympus centres).

3.4. Discussion

This study presents current and compelling clinical and economic evidence supporting adoption of the TURis system as an alternative to M-TURP, which is the current UK surgical standard of care for the treatment of BPH. The analysis was based on an SR of RCT evidence and synthesis of those data for both systems, with a subsequent economic analysis conducted from the perspective of the NHS in England and Wales. The data synthesis (by meta-analysis) showed that TURis was associated with significant improvements in perioperative safety and in duration of hospital stay compared with M-TURP. The results of the base-case economic analysis found that the adoption of TURis could result in cost savings versus M-TURP of £204 and £113 per patient for existing and non-Olympus TURP centres, respectively. Sensitivity analysis demonstrated that reduction in hospital stay was a key driver of results. Inclusion of readmissions resulted in cost savings with TURis versus M-TURP of £245 and £155 per patient for Olympus and non-Olympus centres, respectively. Although this analysis was based on a single RCT (of good quality), NICE considered it sufficiently robust [20].

Although efficacy in prostate resection surgery for treatment of voiding symptoms presumed secondary to BPH is well established, possible perioperative complications and associated costs remain a concern. Current European guidelines state that B-TURP offers similar efficacy to M-TURP but with reduced perioperative morbidity. The guidelines recommend that the choice of B-TURP should be based on surgeon experience, availability of equipment, and patient preference [6]. Although there is currently no European consensus on the use of a specific B-TURP system, in England and Wales the recently published NICE MTG23 guidance concludes that the available evidence supports the adoption of TURis for surgical treatment of BPH and that the TURis system is clinically equivalent to M-TURP, but with the advantage of reducing the risk of perioperative complications [20]. It further states that TURis has the potential to be cost saving compared with M-TURP [20].

The SR presented here demonstrates that the TURis system is supported by strong clinical data in the form of 15 good quality RCTs. To our knowledge, this is the first study to evaluate the wealth of RCT evidence in support of TURis for conducting meta-analysis to support an economic evaluation.

The findings from the economic analysis may have important implications for decision makers. The results suggest that NHS hospitals in England and Wales adopting TURis specifically as an alternative to M-TURP can expect both cost savings and improved clinical outcomes. Based on the results of the meta-analyses, treating 56 people with TURis instead of M-TURP could prevent one case of TUR syndrome. The numbers needed to treat to avoid one blood transfusion and one case of clot retention are 22 and 40, respectively. Because Hospital Episode Statistics data indicate that approximately 14 500 TURP procedures were performed out in England in 2012–2013, approximately 260 cases of TUR syndrome might be avoided if TURis was to replace M-TURP, along with 660 blood transfusions and 360 cases of clot retention. Given the complication costs outlined in Supplementary Table 1, avoiding these complications alone might result in an annual cost savings of >£1 million in England. When reductions in hospital stay are considered, the annual cost savings exceed £3.4 million. Beyond potential cost savings, this represents efficiency improvements for hospitals, enhancing patient management and throughput.

Introduction of medical devices into clinical practice is often limited by availability of high-quality clinical evidence, an essential component of robust economic evaluation of any medical intervention. Large-scale RCTs are conventionally used as the gold standard for demonstrating clinical efficacy of drug therapies; however, medical devices face a range of challenges in RCT design that often affect quality. For example, randomisation may not be possible due to ethical constraints, and blinding (especially double blinding) may not be feasible, thus increasing the risk of bias [37] and [38]. The cohort size and timescale of follow-up and assessments may also be limited [38]. Consequently, the clinical evidence base for a medical device can often be weak. In contrast, this study benefits from high-quality RCT evidence, acknowledged in the NICE MTG23 guidance, comprising 15 studies that show the clinical evidence for TURis supports the adoption of the technology. Patients included in the studies have undergone standard M-TURP or TURis procedures and no deviations in procedure protocol were observed, providing an evidence base with low heterogeneity.

A further strength of the economic analysis is that we only included those clinical outcomes for which meta-analysis showed statistically significant differences between the two treatments. Assumptions regarding resource use for complications were based on NICE clinical guideline 97 [39] and had been validated by two clinical experts for the MTEP assessment.

The analysis is not without limitations. For meta-analysis, a random-effects model is generally considered more robust because it takes heterogeneity into account. The random-effects model is only recommended, however, if ≥10 studies are available for an end point, which was not the case here. We therefore performed a fixed-effects meta-analysis. Second, complications would vary in severity; the costs considered in the analysis are intended to reflect the typical cost of treating each event. Because the model is based on UK cost data, the results are not generalizable to other countries.

2.1. Methods

An SR was performed to identify clinical trials to form the basis of a meta-analysis. The results of the clinical meta-analysis form one part of our objective. They were also used as input for an economic analysis, the second part of our objective.

2.2. Systematic review of the clinical evidence

The SR was performed in line with guidance by the Centre for Reviews and Dissemination, University of York [22]. The detailed SR methodology, as well as results and quality assessment, is reported in the supplementary information for MTG23 [20]. We performed an update to the original search on April 20, 2015, limited to randomised controlled trial (RCT) publications from 2014 onwards. The output from this update was then combined with that from the original SR (Table 1).

2.3. Meta-analysis of the clinical evidence

Expert opinion solicited during development of the MTEP submission suggested that differences are most commonly observed between TURis and M-TURP in the incidence of TUR syndrome, blood transfusions, clot retention, and hospital stay. The evidence was synthesised using pairwise meta-analysis techniques to determine any evidence-based differences between the treatments in these outcomes, as well as in other commonly reported outcomes including urethral strictures, catheterisation time, and procedure duration. According to the Cochrane Systematic Review Handbook, a minimum of 10 studies is required to draw meaningful conclusions from random-effects calculations [5]. Because each meta-analysis undertaken involved <10 studies, fixed-effects analyses were conducted. Details of the meta-analysis methodology were previously reported [20].

2.4. Economic evaluation

The economic analysis (a cost-consequence analysis) compared costs and outcomes associated with TURis and M-TURP, respectively, for surgical treatment of BPH. The perspective of the analysis was that of the National Health Service (NHS) and Personal Social Services in England and Wales.

2.5. Patient population

The analysis considered adult men with LUTS presumed secondary to BPH, in whom surgical intervention was indicated.

2.6. Model structure

A decision-tree model was constructed in which men with BPH-related LUTS underwent surgery with either TURis or M-TURP. Patients entered the model at the point of being indicated for surgery and were allocated probabilities of experiencing procedure-related adverse events (TUR syndrome, blood transfusion, and clot retention) based on the meta-analysis output. Procedure duration was excluded from the analysis because no statistically significant difference was observed between M-TURP and TURis in the meta-analysis. The pooled reduction in catheterisation time estimated from the meta-analysis of 0.2 d for TURis versus M-TURP was statistically significant. However, given that a patient's catheter would generally be removed while in the hospital, this reduction is not expected to result in cost savings beyond those realised through the reduction in hospital stay, which was modelled. As such, catheterisation time was not modelled explicitly. The model considered a nondefined short time horizon covering the prostate resection procedure and immediately afterwards—the period during which the complications modelled are expected to occur. Capital costs assumed equipment life spans of 7 yr. Based on findings of the SR, no difference in functional outcomes (eg, efficacy) was assumed between the procedures with respect to prostate resection weight or radicality.

2.7. Model parameters and data sources

Supplementary Table 1 lists all the variables included in the analysis. Sourcing of inputs, rationale, and further details are provided in Supplement 1.

2.8. Sensitivity analysis

Sensitivity analyses (based on predefined scenarios) were conducted. The first scenario considered all-cause readmission following a TURP procedure, based on data from the single study identified, Fagerstrom et al [23], which indicated a threefold increase in all-cause readmission for M-TURP versus TURis. This study reported readmissions as a result of bleeding, infection, and “other causes.” Because the base-case analysis included clot retention, the scenario considered readmissions from infection and other causes to avoid double counting (it is assumed that “bleeding” includes clot retention). A separate scenario considered a situation in which TURis is performed as day-case surgery.

One-way deterministic sensitivity analysis was performed in which all model parameters were varied either according to published ranges or by ±25%. Results are presented as Tornado diagrams. Probabilistic sensitivity analysis (PSA) was also conducted, in which input parameters were varied according to specified probability distributions. Threshold analysis was conducted to explore the relationship between key model parameters and model results up to the point at which the main conclusions of the model might change, that is, in this case to determine the value at which TURis would be cost neutral (ie, no longer cost saving) versus M-TURP.

3.1. Clinical evidence

The original electronic searches identified a total of 1116 publications. Following title and abstract review, 783 were excluded and 14 relevant RCTs remained, following assessment of the full paper. No unpublished studies were identified from ClinicalTrials.gov. The updated search identified a total 63 further publications, of which two full-text RCT publications remained after review of title and abstract [24] and [25]. On review of the full text, it was found that both of these publications reported data from the same trial, and only the most recent publication, reporting the most up-to-date data, was included in the analysis [24]. Figure 1 shows the study selection flow diagram.

Quality assessment of the identified RCT (Supplementary Table 2) showed that the patient population was consistent with the target population of the analysis. Study end points were clear and appropriately measured according to study design and follow-up schedule.

3.2. Meta-analysis

Six studies were found to be suitable for inclusion in the pairwise meta-analysis of the incidence of TUR syndrome [23], [26], [27], [28], [29], and [30]. The total number of events reported in the studies is presented in Table 2. There were no TUR syndrome events in the TURis arm reported in any of the studies, consistent with the use of saline. The analysis estimated a pooled relative risk (RR) for TUR syndrome of 0.18 (95% confidence interval [CI], 0.05–0.61; p = 0.006) in favour of TURis.

Eight RCTs reported the incidence of blood transfusion, but only seven of these were included in the meta-analysis [23], [24], [26], [27], [28], [29], and [31] (Table 2). The study by Michielsen et al was excluded from the analysis because a higher proportion of TURis procedures was conducted by trainee surgeons compared with M-TURP (32% vs 7%, respectively), resulting in increased procedure duration in this arm as well as increased blood loss and, consequently, blood transfusions [32]. Exclusion of this study on the basis of heterogeneity was accepted by the External Assessment Centre for MTG23 [19]. The results of the meta-analysis showed a statistically significant reduction in blood transfusions with use of TURis compared with M-TURP (RR: 0.34; 95% CI, 0.18–0.61; p = 0.0003).

Six studies were identified for meta-analysis of the incidence of clot retention [24], [26], [28], [29], [31], and [32]. The pooled RR was 0.43 in favour of TURis (95% CI, 0.22–0.86; p = 0.0161), demonstrating that use of TURis is associated with fewer cases of clot retention than M-TURP (Table 2).

Meta-analysis was conducted on seven studies reporting data on urethral strictures [23], [24], [26], [28], [29], [31], and [33]. A non-significantly different pooled RR of 1.27 (95% CI, 0.85–1.9; p > 0.05) was observed for TURis versus M-TURP (Table 2). As a result, this outcome was not considered in the economic analysis.

Further meta-analyses suggested that TURis is associated with statistically significant reductions in hospital stay and catheterisation time (Table 2). For these outcomes, meta-analyses were restricted to studies reporting both the mean duration and standard deviation. From a pooled analysis of four studies reporting length of hospital stay [24], [26], [27], and [33], the mean difference for TURis versus M-TURP was −0.56 d (95% CI, −0.77 to −0.35; p < 0.0001). Analysis of four studies presenting data on catheterisation time [24], [26], [27], and [33] resulted in a statistically significant mean difference for TURis versus M-TURP of −0.23 d (95% CI, −0.38 to −0.09; p = 0.0019); however, this difference is not clinically meaningful because such a minimal reduction in catheterisation time is unlikely to make a considerable difference to patients. Six studies reported procedure duration [24], [26], [29], [31], [33], and [34] and, although not significant, the resulting pooled mean difference for TURis versus M-TURP was −0.95 min (95% CI, −3.35 to 1.46; p > 0.05).

3.3. Economic analysis

Only statistically significant differences in clinical outcomes in the meta-analysis, which directly affect costs, were considered in the base-case analysis: TUR syndrome, blood transfusions, clot retention, and length of hospital stay. Table 3 shows the results of the base-case analysis. Costs of adopting TURis differed according to whether or not hospitals owned an Olympus monopolar system. Based on 150 TURP procedures per annum, capital costs of TURis (Supplementary Table 1) for hospitals with existing Olympus systems and for those without Olympus systems were £9.68 and £29.13 per patient per year, respectively. The total cost per patient (including capital equipment, consumables, hospital stay, and complications) of using TURis for hospitals with Olympus systems in place was £881.71 versus £1085.49 for M-TURP, a cost saving of £203.78 per patient. In hospitals without Olympus systems, the total per patient cost of using TURis was £901.16 compared with £1014.58 for M-TURP, a cost saving of £113.42 per patient.

Higher equipment costs for TURis versus M-TURP were offset by savings from reductions in hospital stay and complications. Assuming a reduction in hospital stay of 0.56 d from the meta-analysis, the base-case analysis found that TURis was associated with a £165.02 reduction in the cost of hospital stay per patient versus M-TURP. TURis was also associated with a £71.82 per patient reduction in complication costs compared with M-TURP.

A further potential benefit of the TURis system is the option of conducting surgery on a day-case basis. Some UK centres are already adopting this practice [35] and [36], and the potential for further cost savings in hospital stay costs via this practice was acknowledged by NICE in MTG23 [20]. Although in clinical practice various factors influence the implementation of day-case surgery, testing a scenario reflecting this situation, in which patients undergoing TURis incurred no inpatient day costs, cost savings were £891.36 and £801.00 per patient, for Olympus and non-Olympus centres, respectively.

Considering readmissions in a further scenario, TURis was associated with annual cost savings versus M-TURP of £245.10 and £154.75 per patient, for Olympus and non-Olympus centres, respectively (Table 3). Readmission for BPH is not widely reported, thus data relating to this outcome are limited; the scenario analysis was based on a single study [23].

Sensitivity analysis also demonstrated that TURis remained cost saving versus M-TURP when each tested model parameter was varied according to its defined range. Key drivers of cost savings were reduction in hospital stay and inpatient day costs (Fig. 2). Results of PSA demonstrated that the cost-consequence analysis was robust across both types of centre. For Olympus centres, use of TURis was associated with a mean saving of £200.77 per patient (95% CI, −£115.54 to −£298.67) and remained cost saving in all 1000 simulations. For Olympus centres, the mean saving was £112.93 per patient (95% CI, £−32.94 to £−211.22), and TURis remained cost saving in 99.8% of simulations.

Results of the threshold analysis on model parameters (Table 4) confirmed the robustness of the base-case results. Existing Olympus centres would need to perform as few as seven procedures annually for TURis to become cost saving versus M-TURP. For non-Olympus centres, the number is 31. Keeping all other model parameters constant, TURis would need to be associated with significantly increased risks of blood transfusion and clot retention compared with M-TURP to cease being cost saving overall. Based on the findings of the meta-analysis, which showed statistically significant reductions in these events for TURis versus TURP, it is not considered plausible that RRs as high as those reported in Table 4 would be observed for TURis versus M-TURP.

Significant increases in the annual cost of capital equipment for TURis would be required for TURis to cease being cost saving versus M-TURP. This could only happen as a result of short equipment life span in practice because prices are generally stable over a period such as the modelled time horizon. The shortened life spans that would be required for TURis to cease being cost saving versus M-TURP (0.28 yr for existing Olympus centres and 1.30 yr for non-Olympus centres) demonstrate the robustness of the conclusions of the analysis to changes in TURis capital costs. TURis would cease being cost saving if the reduction in hospital stay for TURis versus M-TURP was removed (only for non-Olympus centres).

3.4. Discussion

This study presents current and compelling clinical and economic evidence supporting adoption of the TURis system as an alternative to M-TURP, which is the current UK surgical standard of care for the treatment of BPH. The analysis was based on an SR of RCT evidence and synthesis of those data for both systems, with a subsequent economic analysis conducted from the perspective of the NHS in England and Wales. The data synthesis (by meta-analysis) showed that TURis was associated with significant improvements in perioperative safety and in duration of hospital stay compared with M-TURP. The results of the base-case economic analysis found that the adoption of TURis could result in cost savings versus M-TURP of £204 and £113 per patient for existing and non-Olympus TURP centres, respectively. Sensitivity analysis demonstrated that reduction in hospital stay was a key driver of results. Inclusion of readmissions resulted in cost savings with TURis versus M-TURP of £245 and £155 per patient for Olympus and non-Olympus centres, respectively. Although this analysis was based on a single RCT (of good quality), NICE considered it sufficiently robust [20].

Although efficacy in prostate resection surgery for treatment of voiding symptoms presumed secondary to BPH is well established, possible perioperative complications and associated costs remain a concern. Current European guidelines state that B-TURP offers similar efficacy to M-TURP but with reduced perioperative morbidity. The guidelines recommend that the choice of B-TURP should be based on surgeon experience, availability of equipment, and patient preference [6]. Although there is currently no European consensus on the use of a specific B-TURP system, in England and Wales the recently published NICE MTG23 guidance concludes that the available evidence supports the adoption of TURis for surgical treatment of BPH and that the TURis system is clinically equivalent to M-TURP, but with the advantage of reducing the risk of perioperative complications [20]. It further states that TURis has the potential to be cost saving compared with M-TURP [20].

The SR presented here demonstrates that the TURis system is supported by strong clinical data in the form of 15 good quality RCTs. To our knowledge, this is the first study to evaluate the wealth of RCT evidence in support of TURis for conducting meta-analysis to support an economic evaluation.

The findings from the economic analysis may have important implications for decision makers. The results suggest that NHS hospitals in England and Wales adopting TURis specifically as an alternative to M-TURP can expect both cost savings and improved clinical outcomes. Based on the results of the meta-analyses, treating 56 people with TURis instead of M-TURP could prevent one case of TUR syndrome. The numbers needed to treat to avoid one blood transfusion and one case of clot retention are 22 and 40, respectively. Because Hospital Episode Statistics data indicate that approximately 14 500 TURP procedures were performed out in England in 2012–2013, approximately 260 cases of TUR syndrome might be avoided if TURis was to replace M-TURP, along with 660 blood transfusions and 360 cases of clot retention. Given the complication costs outlined in Supplementary Table 1, avoiding these complications alone might result in an annual cost savings of >£1 million in England. When reductions in hospital stay are considered, the annual cost savings exceed £3.4 million. Beyond potential cost savings, this represents efficiency improvements for hospitals, enhancing patient management and throughput.

Introduction of medical devices into clinical practice is often limited by availability of high-quality clinical evidence, an essential component of robust economic evaluation of any medical intervention. Large-scale RCTs are conventionally used as the gold standard for demonstrating clinical efficacy of drug therapies; however, medical devices face a range of challenges in RCT design that often affect quality. For example, randomisation may not be possible due to ethical constraints, and blinding (especially double blinding) may not be feasible, thus increasing the risk of bias [37] and [38]. The cohort size and timescale of follow-up and assessments may also be limited [38]. Consequently, the clinical evidence base for a medical device can often be weak. In contrast, this study benefits from high-quality RCT evidence, acknowledged in the NICE MTG23 guidance, comprising 15 studies that show the clinical evidence for TURis supports the adoption of the technology. Patients included in the studies have undergone standard M-TURP or TURis procedures and no deviations in procedure protocol were observed, providing an evidence base with low heterogeneity.

A further strength of the economic analysis is that we only included those clinical outcomes for which meta-analysis showed statistically significant differences between the two treatments. Assumptions regarding resource use for complications were based on NICE clinical guideline 97 [39] and had been validated by two clinical experts for the MTEP assessment.

The analysis is not without limitations. For meta-analysis, a random-effects model is generally considered more robust because it takes heterogeneity into account. The random-effects model is only recommended, however, if ≥10 studies are available for an end point, which was not the case here. We therefore performed a fixed-effects meta-analysis. Second, complications would vary in severity; the costs considered in the analysis are intended to reflect the typical cost of treating each event. Because the model is based on UK cost data, the results are not generalizable to other countries.

2.1. Methods

An SR was performed to identify clinical trials to form the basis of a meta-analysis. The results of the clinical meta-analysis form one part of our objective. They were also used as input for an economic analysis, the second part of our objective.

2.2. Systematic review of the clinical evidence

The SR was performed in line with guidance by the Centre for Reviews and Dissemination, University of York [22]. The detailed SR methodology, as well as results and quality assessment, is reported in the supplementary information for MTG23 [20]. We performed an update to the original search on April 20, 2015, limited to randomised controlled trial (RCT) publications from 2014 onwards. The output from this update was then combined with that from the original SR (Table 1).

2.3. Meta-analysis of the clinical evidence

Expert opinion solicited during development of the MTEP submission suggested that differences are most commonly observed between TURis and M-TURP in the incidence of TUR syndrome, blood transfusions, clot retention, and hospital stay. The evidence was synthesised using pairwise meta-analysis techniques to determine any evidence-based differences between the treatments in these outcomes, as well as in other commonly reported outcomes including urethral strictures, catheterisation time, and procedure duration. According to the Cochrane Systematic Review Handbook, a minimum of 10 studies is required to draw meaningful conclusions from random-effects calculations [5]. Because each meta-analysis undertaken involved <10 studies, fixed-effects analyses were conducted. Details of the meta-analysis methodology were previously reported [20].

2.4. Economic evaluation

The economic analysis (a cost-consequence analysis) compared costs and outcomes associated with TURis and M-TURP, respectively, for surgical treatment of BPH. The perspective of the analysis was that of the National Health Service (NHS) and Personal Social Services in England and Wales.

2.5. Patient population

The analysis considered adult men with LUTS presumed secondary to BPH, in whom surgical intervention was indicated.

2.6. Model structure

A decision-tree model was constructed in which men with BPH-related LUTS underwent surgery with either TURis or M-TURP. Patients entered the model at the point of being indicated for surgery and were allocated probabilities of experiencing procedure-related adverse events (TUR syndrome, blood transfusion, and clot retention) based on the meta-analysis output. Procedure duration was excluded from the analysis because no statistically significant difference was observed between M-TURP and TURis in the meta-analysis. The pooled reduction in catheterisation time estimated from the meta-analysis of 0.2 d for TURis versus M-TURP was statistically significant. However, given that a patient's catheter would generally be removed while in the hospital, this reduction is not expected to result in cost savings beyond those realised through the reduction in hospital stay, which was modelled. As such, catheterisation time was not modelled explicitly. The model considered a nondefined short time horizon covering the prostate resection procedure and immediately afterwards—the period during which the complications modelled are expected to occur. Capital costs assumed equipment life spans of 7 yr. Based on findings of the SR, no difference in functional outcomes (eg, efficacy) was assumed between the procedures with respect to prostate resection weight or radicality.

2.7. Model parameters and data sources

Supplementary Table 1 lists all the variables included in the analysis. Sourcing of inputs, rationale, and further details are provided in Supplement 1.

2.8. Sensitivity analysis

Sensitivity analyses (based on predefined scenarios) were conducted. The first scenario considered all-cause readmission following a TURP procedure, based on data from the single study identified, Fagerstrom et al [23], which indicated a threefold increase in all-cause readmission for M-TURP versus TURis. This study reported readmissions as a result of bleeding, infection, and “other causes.” Because the base-case analysis included clot retention, the scenario considered readmissions from infection and other causes to avoid double counting (it is assumed that “bleeding” includes clot retention). A separate scenario considered a situation in which TURis is performed as day-case surgery.

One-way deterministic sensitivity analysis was performed in which all model parameters were varied either according to published ranges or by ±25%. Results are presented as Tornado diagrams. Probabilistic sensitivity analysis (PSA) was also conducted, in which input parameters were varied according to specified probability distributions. Threshold analysis was conducted to explore the relationship between key model parameters and model results up to the point at which the main conclusions of the model might change, that is, in this case to determine the value at which TURis would be cost neutral (ie, no longer cost saving) versus M-TURP.

3.1. Clinical evidence

The original electronic searches identified a total of 1116 publications. Following title and abstract review, 783 were excluded and 14 relevant RCTs remained, following assessment of the full paper. No unpublished studies were identified from ClinicalTrials.gov. The updated search identified a total 63 further publications, of which two full-text RCT publications remained after review of title and abstract [24] and [25]. On review of the full text, it was found that both of these publications reported data from the same trial, and only the most recent publication, reporting the most up-to-date data, was included in the analysis [24]. Figure 1 shows the study selection flow diagram.

Quality assessment of the identified RCT (Supplementary Table 2) showed that the patient population was consistent with the target population of the analysis. Study end points were clear and appropriately measured according to study design and follow-up schedule.

3.2. Meta-analysis

Six studies were found to be suitable for inclusion in the pairwise meta-analysis of the incidence of TUR syndrome [23], [26], [27], [28], [29], and [30]. The total number of events reported in the studies is presented in Table 2. There were no TUR syndrome events in the TURis arm reported in any of the studies, consistent with the use of saline. The analysis estimated a pooled relative risk (RR) for TUR syndrome of 0.18 (95% confidence interval [CI], 0.05–0.61; p = 0.006) in favour of TURis.

Eight RCTs reported the incidence of blood transfusion, but only seven of these were included in the meta-analysis [23], [24], [26], [27], [28], [29], and [31] (Table 2). The study by Michielsen et al was excluded from the analysis because a higher proportion of TURis procedures was conducted by trainee surgeons compared with M-TURP (32% vs 7%, respectively), resulting in increased procedure duration in this arm as well as increased blood loss and, consequently, blood transfusions [32]. Exclusion of this study on the basis of heterogeneity was accepted by the External Assessment Centre for MTG23 [19]. The results of the meta-analysis showed a statistically significant reduction in blood transfusions with use of TURis compared with M-TURP (RR: 0.34; 95% CI, 0.18–0.61; p = 0.0003).

Six studies were identified for meta-analysis of the incidence of clot retention [24], [26], [28], [29], [31], and [32]. The pooled RR was 0.43 in favour of TURis (95% CI, 0.22–0.86; p = 0.0161), demonstrating that use of TURis is associated with fewer cases of clot retention than M-TURP (Table 2).

Meta-analysis was conducted on seven studies reporting data on urethral strictures [23], [24], [26], [28], [29], [31], and [33]. A non-significantly different pooled RR of 1.27 (95% CI, 0.85–1.9; p > 0.05) was observed for TURis versus M-TURP (Table 2). As a result, this outcome was not considered in the economic analysis.

Further meta-analyses suggested that TURis is associated with statistically significant reductions in hospital stay and catheterisation time (Table 2). For these outcomes, meta-analyses were restricted to studies reporting both the mean duration and standard deviation. From a pooled analysis of four studies reporting length of hospital stay [24], [26], [27], and [33], the mean difference for TURis versus M-TURP was −0.56 d (95% CI, −0.77 to −0.35; p < 0.0001). Analysis of four studies presenting data on catheterisation time [24], [26], [27], and [33] resulted in a statistically significant mean difference for TURis versus M-TURP of −0.23 d (95% CI, −0.38 to −0.09; p = 0.0019); however, this difference is not clinically meaningful because such a minimal reduction in catheterisation time is unlikely to make a considerable difference to patients. Six studies reported procedure duration [24], [26], [29], [31], [33], and [34] and, although not significant, the resulting pooled mean difference for TURis versus M-TURP was −0.95 min (95% CI, −3.35 to 1.46; p > 0.05).

3.3. Economic analysis

Only statistically significant differences in clinical outcomes in the meta-analysis, which directly affect costs, were considered in the base-case analysis: TUR syndrome, blood transfusions, clot retention, and length of hospital stay. Table 3 shows the results of the base-case analysis. Costs of adopting TURis differed according to whether or not hospitals owned an Olympus monopolar system. Based on 150 TURP procedures per annum, capital costs of TURis (Supplementary Table 1) for hospitals with existing Olympus systems and for those without Olympus systems were £9.68 and £29.13 per patient per year, respectively. The total cost per patient (including capital equipment, consumables, hospital stay, and complications) of using TURis for hospitals with Olympus systems in place was £881.71 versus £1085.49 for M-TURP, a cost saving of £203.78 per patient. In hospitals without Olympus systems, the total per patient cost of using TURis was £901.16 compared with £1014.58 for M-TURP, a cost saving of £113.42 per patient.

Higher equipment costs for TURis versus M-TURP were offset by savings from reductions in hospital stay and complications. Assuming a reduction in hospital stay of 0.56 d from the meta-analysis, the base-case analysis found that TURis was associated with a £165.02 reduction in the cost of hospital stay per patient versus M-TURP. TURis was also associated with a £71.82 per patient reduction in complication costs compared with M-TURP.

A further potential benefit of the TURis system is the option of conducting surgery on a day-case basis. Some UK centres are already adopting this practice [35] and [36], and the potential for further cost savings in hospital stay costs via this practice was acknowledged by NICE in MTG23 [20]. Although in clinical practice various factors influence the implementation of day-case surgery, testing a scenario reflecting this situation, in which patients undergoing TURis incurred no inpatient day costs, cost savings were £891.36 and £801.00 per patient, for Olympus and non-Olympus centres, respectively.

Considering readmissions in a further scenario, TURis was associated with annual cost savings versus M-TURP of £245.10 and £154.75 per patient, for Olympus and non-Olympus centres, respectively (Table 3). Readmission for BPH is not widely reported, thus data relating to this outcome are limited; the scenario analysis was based on a single study [23].

Sensitivity analysis also demonstrated that TURis remained cost saving versus M-TURP when each tested model parameter was varied according to its defined range. Key drivers of cost savings were reduction in hospital stay and inpatient day costs (Fig. 2). Results of PSA demonstrated that the cost-consequence analysis was robust across both types of centre. For Olympus centres, use of TURis was associated with a mean saving of £200.77 per patient (95% CI, −£115.54 to −£298.67) and remained cost saving in all 1000 simulations. For Olympus centres, the mean saving was £112.93 per patient (95% CI, £−32.94 to £−211.22), and TURis remained cost saving in 99.8% of simulations.

Results of the threshold analysis on model parameters (Table 4) confirmed the robustness of the base-case results. Existing Olympus centres would need to perform as few as seven procedures annually for TURis to become cost saving versus M-TURP. For non-Olympus centres, the number is 31. Keeping all other model parameters constant, TURis would need to be associated with significantly increased risks of blood transfusion and clot retention compared with M-TURP to cease being cost saving overall. Based on the findings of the meta-analysis, which showed statistically significant reductions in these events for TURis versus TURP, it is not considered plausible that RRs as high as those reported in Table 4 would be observed for TURis versus M-TURP.

Significant increases in the annual cost of capital equipment for TURis would be required for TURis to cease being cost saving versus M-TURP. This could only happen as a result of short equipment life span in practice because prices are generally stable over a period such as the modelled time horizon. The shortened life spans that would be required for TURis to cease being cost saving versus M-TURP (0.28 yr for existing Olympus centres and 1.30 yr for non-Olympus centres) demonstrate the robustness of the conclusions of the analysis to changes in TURis capital costs. TURis would cease being cost saving if the reduction in hospital stay for TURis versus M-TURP was removed (only for non-Olympus centres).

3.4. Discussion

This study presents current and compelling clinical and economic evidence supporting adoption of the TURis system as an alternative to M-TURP, which is the current UK surgical standard of care for the treatment of BPH. The analysis was based on an SR of RCT evidence and synthesis of those data for both systems, with a subsequent economic analysis conducted from the perspective of the NHS in England and Wales. The data synthesis (by meta-analysis) showed that TURis was associated with significant improvements in perioperative safety and in duration of hospital stay compared with M-TURP. The results of the base-case economic analysis found that the adoption of TURis could result in cost savings versus M-TURP of £204 and £113 per patient for existing and non-Olympus TURP centres, respectively. Sensitivity analysis demonstrated that reduction in hospital stay was a key driver of results. Inclusion of readmissions resulted in cost savings with TURis versus M-TURP of £245 and £155 per patient for Olympus and non-Olympus centres, respectively. Although this analysis was based on a single RCT (of good quality), NICE considered it sufficiently robust [20].

Although efficacy in prostate resection surgery for treatment of voiding symptoms presumed secondary to BPH is well established, possible perioperative complications and associated costs remain a concern. Current European guidelines state that B-TURP offers similar efficacy to M-TURP but with reduced perioperative morbidity. The guidelines recommend that the choice of B-TURP should be based on surgeon experience, availability of equipment, and patient preference [6]. Although there is currently no European consensus on the use of a specific B-TURP system, in England and Wales the recently published NICE MTG23 guidance concludes that the available evidence supports the adoption of TURis for surgical treatment of BPH and that the TURis system is clinically equivalent to M-TURP, but with the advantage of reducing the risk of perioperative complications [20]. It further states that TURis has the potential to be cost saving compared with M-TURP [20].

The SR presented here demonstrates that the TURis system is supported by strong clinical data in the form of 15 good quality RCTs. To our knowledge, this is the first study to evaluate the wealth of RCT evidence in support of TURis for conducting meta-analysis to support an economic evaluation.

The findings from the economic analysis may have important implications for decision makers. The results suggest that NHS hospitals in England and Wales adopting TURis specifically as an alternative to M-TURP can expect both cost savings and improved clinical outcomes. Based on the results of the meta-analyses, treating 56 people with TURis instead of M-TURP could prevent one case of TUR syndrome. The numbers needed to treat to avoid one blood transfusion and one case of clot retention are 22 and 40, respectively. Because Hospital Episode Statistics data indicate that approximately 14 500 TURP procedures were performed out in England in 2012–2013, approximately 260 cases of TUR syndrome might be avoided if TURis was to replace M-TURP, along with 660 blood transfusions and 360 cases of clot retention. Given the complication costs outlined in Supplementary Table 1, avoiding these complications alone might result in an annual cost savings of >£1 million in England. When reductions in hospital stay are considered, the annual cost savings exceed £3.4 million. Beyond potential cost savings, this represents efficiency improvements for hospitals, enhancing patient management and throughput.

Introduction of medical devices into clinical practice is often limited by availability of high-quality clinical evidence, an essential component of robust economic evaluation of any medical intervention. Large-scale RCTs are conventionally used as the gold standard for demonstrating clinical efficacy of drug therapies; however, medical devices face a range of challenges in RCT design that often affect quality. For example, randomisation may not be possible due to ethical constraints, and blinding (especially double blinding) may not be feasible, thus increasing the risk of bias [37] and [38]. The cohort size and timescale of follow-up and assessments may also be limited [38]. Consequently, the clinical evidence base for a medical device can often be weak. In contrast, this study benefits from high-quality RCT evidence, acknowledged in the NICE MTG23 guidance, comprising 15 studies that show the clinical evidence for TURis supports the adoption of the technology. Patients included in the studies have undergone standard M-TURP or TURis procedures and no deviations in procedure protocol were observed, providing an evidence base with low heterogeneity.

A further strength of the economic analysis is that we only included those clinical outcomes for which meta-analysis showed statistically significant differences between the two treatments. Assumptions regarding resource use for complications were based on NICE clinical guideline 97 [39] and had been validated by two clinical experts for the MTEP assessment.

The analysis is not without limitations. For meta-analysis, a random-effects model is generally considered more robust because it takes heterogeneity into account. The random-effects model is only recommended, however, if ≥10 studies are available for an end point, which was not the case here. We therefore performed a fixed-effects meta-analysis. Second, complications would vary in severity; the costs considered in the analysis are intended to reflect the typical cost of treating each event. Because the model is based on UK cost data, the results are not generalizable to other countries.

2.1. Methods

An SR was performed to identify clinical trials to form the basis of a meta-analysis. The results of the clinical meta-analysis form one part of our objective. They were also used as input for an economic analysis, the second part of our objective.

2.2. Systematic review of the clinical evidence

The SR was performed in line with guidance by the Centre for Reviews and Dissemination, University of York [22]. The detailed SR methodology, as well as results and quality assessment, is reported in the supplementary information for MTG23 [20]. We performed an update to the original search on April 20, 2015, limited to randomised controlled trial (RCT) publications from 2014 onwards. The output from this update was then combined with that from the original SR (Table 1).

2.3. Meta-analysis of the clinical evidence

Expert opinion solicited during development of the MTEP submission suggested that differences are most commonly observed between TURis and M-TURP in the incidence of TUR syndrome, blood transfusions, clot retention, and hospital stay. The evidence was synthesised using pairwise meta-analysis techniques to determine any evidence-based differences between the treatments in these outcomes, as well as in other commonly reported outcomes including urethral strictures, catheterisation time, and procedure duration. According to the Cochrane Systematic Review Handbook, a minimum of 10 studies is required to draw meaningful conclusions from random-effects calculations [5]. Because each meta-analysis undertaken involved <10 studies, fixed-effects analyses were conducted. Details of the meta-analysis methodology were previously reported [20].

2.4. Economic evaluation

The economic analysis (a cost-consequence analysis) compared costs and outcomes associated with TURis and M-TURP, respectively, for surgical treatment of BPH. The perspective of the analysis was that of the National Health Service (NHS) and Personal Social Services in England and Wales.

2.5. Patient population

The analysis considered adult men with LUTS presumed secondary to BPH, in whom surgical intervention was indicated.

2.6. Model structure

A decision-tree model was constructed in which men with BPH-related LUTS underwent surgery with either TURis or M-TURP. Patients entered the model at the point of being indicated for surgery and were allocated probabilities of experiencing procedure-related adverse events (TUR syndrome, blood transfusion, and clot retention) based on the meta-analysis output. Procedure duration was excluded from the analysis because no statistically significant difference was observed between M-TURP and TURis in the meta-analysis. The pooled reduction in catheterisation time estimated from the meta-analysis of 0.2 d for TURis versus M-TURP was statistically significant. However, given that a patient's catheter would generally be removed while in the hospital, this reduction is not expected to result in cost savings beyond those realised through the reduction in hospital stay, which was modelled. As such, catheterisation time was not modelled explicitly. The model considered a nondefined short time horizon covering the prostate resection procedure and immediately afterwards—the period during which the complications modelled are expected to occur. Capital costs assumed equipment life spans of 7 yr. Based on findings of the SR, no difference in functional outcomes (eg, efficacy) was assumed between the procedures with respect to prostate resection weight or radicality.

2.7. Model parameters and data sources

Supplementary Table 1 lists all the variables included in the analysis. Sourcing of inputs, rationale, and further details are provided in Supplement 1.

2.8. Sensitivity analysis

Sensitivity analyses (based on predefined scenarios) were conducted. The first scenario considered all-cause readmission following a TURP procedure, based on data from the single study identified, Fagerstrom et al [23], which indicated a threefold increase in all-cause readmission for M-TURP versus TURis. This study reported readmissions as a result of bleeding, infection, and “other causes.” Because the base-case analysis included clot retention, the scenario considered readmissions from infection and other causes to avoid double counting (it is assumed that “bleeding” includes clot retention). A separate scenario considered a situation in which TURis is performed as day-case surgery.

One-way deterministic sensitivity analysis was performed in which all model parameters were varied either according to published ranges or by ±25%. Results are presented as Tornado diagrams. Probabilistic sensitivity analysis (PSA) was also conducted, in which input parameters were varied according to specified probability distributions. Threshold analysis was conducted to explore the relationship between key model parameters and model results up to the point at which the main conclusions of the model might change, that is, in this case to determine the value at which TURis would be cost neutral (ie, no longer cost saving) versus M-TURP.

3.1. Clinical evidence

The original electronic searches identified a total of 1116 publications. Following title and abstract review, 783 were excluded and 14 relevant RCTs remained, following assessment of the full paper. No unpublished studies were identified from ClinicalTrials.gov. The updated search identified a total 63 further publications, of which two full-text RCT publications remained after review of title and abstract [24] and [25]. On review of the full text, it was found that both of these publications reported data from the same trial, and only the most recent publication, reporting the most up-to-date data, was included in the analysis [24]. Figure 1 shows the study selection flow diagram.

Quality assessment of the identified RCT (Supplementary Table 2) showed that the patient population was consistent with the target population of the analysis. Study end points were clear and appropriately measured according to study design and follow-up schedule.

3.2. Meta-analysis

Six studies were found to be suitable for inclusion in the pairwise meta-analysis of the incidence of TUR syndrome [23], [26], [27], [28], [29], and [30]. The total number of events reported in the studies is presented in Table 2. There were no TUR syndrome events in the TURis arm reported in any of the studies, consistent with the use of saline. The analysis estimated a pooled relative risk (RR) for TUR syndrome of 0.18 (95% confidence interval [CI], 0.05–0.61; p = 0.006) in favour of TURis.

Eight RCTs reported the incidence of blood transfusion, but only seven of these were included in the meta-analysis [23], [24], [26], [27], [28], [29], and [31] (Table 2). The study by Michielsen et al was excluded from the analysis because a higher proportion of TURis procedures was conducted by trainee surgeons compared with M-TURP (32% vs 7%, respectively), resulting in increased procedure duration in this arm as well as increased blood loss and, consequently, blood transfusions [32]. Exclusion of this study on the basis of heterogeneity was accepted by the External Assessment Centre for MTG23 [19]. The results of the meta-analysis showed a statistically significant reduction in blood transfusions with use of TURis compared with M-TURP (RR: 0.34; 95% CI, 0.18–0.61; p = 0.0003).

Six studies were identified for meta-analysis of the incidence of clot retention [24], [26], [28], [29], [31], and [32]. The pooled RR was 0.43 in favour of TURis (95% CI, 0.22–0.86; p = 0.0161), demonstrating that use of TURis is associated with fewer cases of clot retention than M-TURP (Table 2).

Meta-analysis was conducted on seven studies reporting data on urethral strictures [23], [24], [26], [28], [29], [31], and [33]. A non-significantly different pooled RR of 1.27 (95% CI, 0.85–1.9; p > 0.05) was observed for TURis versus M-TURP (Table 2). As a result, this outcome was not considered in the economic analysis.

Further meta-analyses suggested that TURis is associated with statistically significant reductions in hospital stay and catheterisation time (Table 2). For these outcomes, meta-analyses were restricted to studies reporting both the mean duration and standard deviation. From a pooled analysis of four studies reporting length of hospital stay [24], [26], [27], and [33], the mean difference for TURis versus M-TURP was −0.56 d (95% CI, −0.77 to −0.35; p < 0.0001). Analysis of four studies presenting data on catheterisation time [24], [26], [27], and [33] resulted in a statistically significant mean difference for TURis versus M-TURP of −0.23 d (95% CI, −0.38 to −0.09; p = 0.0019); however, this difference is not clinically meaningful because such a minimal reduction in catheterisation time is unlikely to make a considerable difference to patients. Six studies reported procedure duration [24], [26], [29], [31], [33], and [34] and, although not significant, the resulting pooled mean difference for TURis versus M-TURP was −0.95 min (95% CI, −3.35 to 1.46; p > 0.05).

3.3. Economic analysis

Only statistically significant differences in clinical outcomes in the meta-analysis, which directly affect costs, were considered in the base-case analysis: TUR syndrome, blood transfusions, clot retention, and length of hospital stay. Table 3 shows the results of the base-case analysis. Costs of adopting TURis differed according to whether or not hospitals owned an Olympus monopolar system. Based on 150 TURP procedures per annum, capital costs of TURis (Supplementary Table 1) for hospitals with existing Olympus systems and for those without Olympus systems were £9.68 and £29.13 per patient per year, respectively. The total cost per patient (including capital equipment, consumables, hospital stay, and complications) of using TURis for hospitals with Olympus systems in place was £881.71 versus £1085.49 for M-TURP, a cost saving of £203.78 per patient. In hospitals without Olympus systems, the total per patient cost of using TURis was £901.16 compared with £1014.58 for M-TURP, a cost saving of £113.42 per patient.

Higher equipment costs for TURis versus M-TURP were offset by savings from reductions in hospital stay and complications. Assuming a reduction in hospital stay of 0.56 d from the meta-analysis, the base-case analysis found that TURis was associated with a £165.02 reduction in the cost of hospital stay per patient versus M-TURP. TURis was also associated with a £71.82 per patient reduction in complication costs compared with M-TURP.

A further potential benefit of the TURis system is the option of conducting surgery on a day-case basis. Some UK centres are already adopting this practice [35] and [36], and the potential for further cost savings in hospital stay costs via this practice was acknowledged by NICE in MTG23 [20]. Although in clinical practice various factors influence the implementation of day-case surgery, testing a scenario reflecting this situation, in which patients undergoing TURis incurred no inpatient day costs, cost savings were £891.36 and £801.00 per patient, for Olympus and non-Olympus centres, respectively.

Considering readmissions in a further scenario, TURis was associated with annual cost savings versus M-TURP of £245.10 and £154.75 per patient, for Olympus and non-Olympus centres, respectively (Table 3). Readmission for BPH is not widely reported, thus data relating to this outcome are limited; the scenario analysis was based on a single study [23].

Sensitivity analysis also demonstrated that TURis remained cost saving versus M-TURP when each tested model parameter was varied according to its defined range. Key drivers of cost savings were reduction in hospital stay and inpatient day costs (Fig. 2). Results of PSA demonstrated that the cost-consequence analysis was robust across both types of centre. For Olympus centres, use of TURis was associated with a mean saving of £200.77 per patient (95% CI, −£115.54 to −£298.67) and remained cost saving in all 1000 simulations. For Olympus centres, the mean saving was £112.93 per patient (95% CI, £−32.94 to £−211.22), and TURis remained cost saving in 99.8% of simulations.

Results of the threshold analysis on model parameters (Table 4) confirmed the robustness of the base-case results. Existing Olympus centres would need to perform as few as seven procedures annually for TURis to become cost saving versus M-TURP. For non-Olympus centres, the number is 31. Keeping all other model parameters constant, TURis would need to be associated with significantly increased risks of blood transfusion and clot retention compared with M-TURP to cease being cost saving overall. Based on the findings of the meta-analysis, which showed statistically significant reductions in these events for TURis versus TURP, it is not considered plausible that RRs as high as those reported in Table 4 would be observed for TURis versus M-TURP.

Significant increases in the annual cost of capital equipment for TURis would be required for TURis to cease being cost saving versus M-TURP. This could only happen as a result of short equipment life span in practice because prices are generally stable over a period such as the modelled time horizon. The shortened life spans that would be required for TURis to cease being cost saving versus M-TURP (0.28 yr for existing Olympus centres and 1.30 yr for non-Olympus centres) demonstrate the robustness of the conclusions of the analysis to changes in TURis capital costs. TURis would cease being cost saving if the reduction in hospital stay for TURis versus M-TURP was removed (only for non-Olympus centres).

3.4. Discussion

This study presents current and compelling clinical and economic evidence supporting adoption of the TURis system as an alternative to M-TURP, which is the current UK surgical standard of care for the treatment of BPH. The analysis was based on an SR of RCT evidence and synthesis of those data for both systems, with a subsequent economic analysis conducted from the perspective of the NHS in England and Wales. The data synthesis (by meta-analysis) showed that TURis was associated with significant improvements in perioperative safety and in duration of hospital stay compared with M-TURP. The results of the base-case economic analysis found that the adoption of TURis could result in cost savings versus M-TURP of £204 and £113 per patient for existing and non-Olympus TURP centres, respectively. Sensitivity analysis demonstrated that reduction in hospital stay was a key driver of results. Inclusion of readmissions resulted in cost savings with TURis versus M-TURP of £245 and £155 per patient for Olympus and non-Olympus centres, respectively. Although this analysis was based on a single RCT (of good quality), NICE considered it sufficiently robust [20].

Although efficacy in prostate resection surgery for treatment of voiding symptoms presumed secondary to BPH is well established, possible perioperative complications and associated costs remain a concern. Current European guidelines state that B-TURP offers similar efficacy to M-TURP but with reduced perioperative morbidity. The guidelines recommend that the choice of B-TURP should be based on surgeon experience, availability of equipment, and patient preference [6]. Although there is currently no European consensus on the use of a specific B-TURP system, in England and Wales the recently published NICE MTG23 guidance concludes that the available evidence supports the adoption of TURis for surgical treatment of BPH and that the TURis system is clinically equivalent to M-TURP, but with the advantage of reducing the risk of perioperative complications [20]. It further states that TURis has the potential to be cost saving compared with M-TURP [20].

The SR presented here demonstrates that the TURis system is supported by strong clinical data in the form of 15 good quality RCTs. To our knowledge, this is the first study to evaluate the wealth of RCT evidence in support of TURis for conducting meta-analysis to support an economic evaluation.

The findings from the economic analysis may have important implications for decision makers. The results suggest that NHS hospitals in England and Wales adopting TURis specifically as an alternative to M-TURP can expect both cost savings and improved clinical outcomes. Based on the results of the meta-analyses, treating 56 people with TURis instead of M-TURP could prevent one case of TUR syndrome. The numbers needed to treat to avoid one blood transfusion and one case of clot retention are 22 and 40, respectively. Because Hospital Episode Statistics data indicate that approximately 14 500 TURP procedures were performed out in England in 2012–2013, approximately 260 cases of TUR syndrome might be avoided if TURis was to replace M-TURP, along with 660 blood transfusions and 360 cases of clot retention. Given the complication costs outlined in Supplementary Table 1, avoiding these complications alone might result in an annual cost savings of >£1 million in England. When reductions in hospital stay are considered, the annual cost savings exceed £3.4 million. Beyond potential cost savings, this represents efficiency improvements for hospitals, enhancing patient management and throughput.

Introduction of medical devices into clinical practice is often limited by availability of high-quality clinical evidence, an essential component of robust economic evaluation of any medical intervention. Large-scale RCTs are conventionally used as the gold standard for demonstrating clinical efficacy of drug therapies; however, medical devices face a range of challenges in RCT design that often affect quality. For example, randomisation may not be possible due to ethical constraints, and blinding (especially double blinding) may not be feasible, thus increasing the risk of bias [37] and [38]. The cohort size and timescale of follow-up and assessments may also be limited [38]. Consequently, the clinical evidence base for a medical device can often be weak. In contrast, this study benefits from high-quality RCT evidence, acknowledged in the NICE MTG23 guidance, comprising 15 studies that show the clinical evidence for TURis supports the adoption of the technology. Patients included in the studies have undergone standard M-TURP or TURis procedures and no deviations in procedure protocol were observed, providing an evidence base with low heterogeneity.

A further strength of the economic analysis is that we only included those clinical outcomes for which meta-analysis showed statistically significant differences between the two treatments. Assumptions regarding resource use for complications were based on NICE clinical guideline 97 [39] and had been validated by two clinical experts for the MTEP assessment.

The analysis is not without limitations. For meta-analysis, a random-effects model is generally considered more robust because it takes heterogeneity into account. The random-effects model is only recommended, however, if ≥10 studies are available for an end point, which was not the case here. We therefore performed a fixed-effects meta-analysis. Second, complications would vary in severity; the costs considered in the analysis are intended to reflect the typical cost of treating each event. Because the model is based on UK cost data, the results are not generalizable to other countries.

2.1. Methods

An SR was performed to identify clinical trials to form the basis of a meta-analysis. The results of the clinical meta-analysis form one part of our objective. They were also used as input for an economic analysis, the second part of our objective.

2.2. Systematic review of the clinical evidence

The SR was performed in line with guidance by the Centre for Reviews and Dissemination, University of York [22]. The detailed SR methodology, as well as results and quality assessment, is reported in the supplementary information for MTG23 [20]. We performed an update to the original search on April 20, 2015, limited to randomised controlled trial (RCT) publications from 2014 onwards. The output from this update was then combined with that from the original SR (Table 1).

2.3. Meta-analysis of the clinical evidence

Expert opinion solicited during development of the MTEP submission suggested that differences are most commonly observed between TURis and M-TURP in the incidence of TUR syndrome, blood transfusions, clot retention, and hospital stay. The evidence was synthesised using pairwise meta-analysis techniques to determine any evidence-based differences between the treatments in these outcomes, as well as in other commonly reported outcomes including urethral strictures, catheterisation time, and procedure duration. According to the Cochrane Systematic Review Handbook, a minimum of 10 studies is required to draw meaningful conclusions from random-effects calculations [5]. Because each meta-analysis undertaken involved <10 studies, fixed-effects analyses were conducted. Details of the meta-analysis methodology were previously reported [20].

2.4. Economic evaluation

The economic analysis (a cost-consequence analysis) compared costs and outcomes associated with TURis and M-TURP, respectively, for surgical treatment of BPH. The perspective of the analysis was that of the National Health Service (NHS) and Personal Social Services in England and Wales.

2.5. Patient population

The analysis considered adult men with LUTS presumed secondary to BPH, in whom surgical intervention was indicated.

2.6. Model structure

A decision-tree model was constructed in which men with BPH-related LUTS underwent surgery with either TURis or M-TURP. Patients entered the model at the point of being indicated for surgery and were allocated probabilities of experiencing procedure-related adverse events (TUR syndrome, blood transfusion, and clot retention) based on the meta-analysis output. Procedure duration was excluded from the analysis because no statistically significant difference was observed between M-TURP and TURis in the meta-analysis. The pooled reduction in catheterisation time estimated from the meta-analysis of 0.2 d for TURis versus M-TURP was statistically significant. However, given that a patient's catheter would generally be removed while in the hospital, this reduction is not expected to result in cost savings beyond those realised through the reduction in hospital stay, which was modelled. As such, catheterisation time was not modelled explicitly. The model considered a nondefined short time horizon covering the prostate resection procedure and immediately afterwards—the period during which the complications modelled are expected to occur. Capital costs assumed equipment life spans of 7 yr. Based on findings of the SR, no difference in functional outcomes (eg, efficacy) was assumed between the procedures with respect to prostate resection weight or radicality.

2.7. Model parameters and data sources

Supplementary Table 1 lists all the variables included in the analysis. Sourcing of inputs, rationale, and further details are provided in Supplement 1.

2.8. Sensitivity analysis

Sensitivity analyses (based on predefined scenarios) were conducted. The first scenario considered all-cause readmission following a TURP procedure, based on data from the single study identified, Fagerstrom et al [23], which indicated a threefold increase in all-cause readmission for M-TURP versus TURis. This study reported readmissions as a result of bleeding, infection, and “other causes.” Because the base-case analysis included clot retention, the scenario considered readmissions from infection and other causes to avoid double counting (it is assumed that “bleeding” includes clot retention). A separate scenario considered a situation in which TURis is performed as day-case surgery.

One-way deterministic sensitivity analysis was performed in which all model parameters were varied either according to published ranges or by ±25%. Results are presented as Tornado diagrams. Probabilistic sensitivity analysis (PSA) was also conducted, in which input parameters were varied according to specified probability distributions. Threshold analysis was conducted to explore the relationship between key model parameters and model results up to the point at which the main conclusions of the model might change, that is, in this case to determine the value at which TURis would be cost neutral (ie, no longer cost saving) versus M-TURP.

3.1. Clinical evidence

The original electronic searches identified a total of 1116 publications. Following title and abstract review, 783 were excluded and 14 relevant RCTs remained, following assessment of the full paper. No unpublished studies were identified from ClinicalTrials.gov. The updated search identified a total 63 further publications, of which two full-text RCT publications remained after review of title and abstract [24] and [25]. On review of the full text, it was found that both of these publications reported data from the same trial, and only the most recent publication, reporting the most up-to-date data, was included in the analysis [24]. Figure 1 shows the study selection flow diagram.

Quality assessment of the identified RCT (Supplementary Table 2) showed that the patient population was consistent with the target population of the analysis. Study end points were clear and appropriately measured according to study design and follow-up schedule.

3.2. Meta-analysis

Six studies were found to be suitable for inclusion in the pairwise meta-analysis of the incidence of TUR syndrome [23], [26], [27], [28], [29], and [30]. The total number of events reported in the studies is presented in Table 2. There were no TUR syndrome events in the TURis arm reported in any of the studies, consistent with the use of saline. The analysis estimated a pooled relative risk (RR) for TUR syndrome of 0.18 (95% confidence interval [CI], 0.05–0.61; p = 0.006) in favour of TURis.

Eight RCTs reported the incidence of blood transfusion, but only seven of these were included in the meta-analysis [23], [24], [26], [27], [28], [29], and [31] (Table 2). The study by Michielsen et al was excluded from the analysis because a higher proportion of TURis procedures was conducted by trainee surgeons compared with M-TURP (32% vs 7%, respectively), resulting in increased procedure duration in this arm as well as increased blood loss and, consequently, blood transfusions [32]. Exclusion of this study on the basis of heterogeneity was accepted by the External Assessment Centre for MTG23 [19]. The results of the meta-analysis showed a statistically significant reduction in blood transfusions with use of TURis compared with M-TURP (RR: 0.34; 95% CI, 0.18–0.61; p = 0.0003).

Six studies were identified for meta-analysis of the incidence of clot retention [24], [26], [28], [29], [31], and [32]. The pooled RR was 0.43 in favour of TURis (95% CI, 0.22–0.86; p = 0.0161), demonstrating that use of TURis is associated with fewer cases of clot retention than M-TURP (Table 2).

Meta-analysis was conducted on seven studies reporting data on urethral strictures [23], [24], [26], [28], [29], [31], and [33]. A non-significantly different pooled RR of 1.27 (95% CI, 0.85–1.9; p > 0.05) was observed for TURis versus M-TURP (Table 2). As a result, this outcome was not considered in the economic analysis.

Further meta-analyses suggested that TURis is associated with statistically significant reductions in hospital stay and catheterisation time (Table 2). For these outcomes, meta-analyses were restricted to studies reporting both the mean duration and standard deviation. From a pooled analysis of four studies reporting length of hospital stay [24], [26], [27], and [33], the mean difference for TURis versus M-TURP was −0.56 d (95% CI, −0.77 to −0.35; p < 0.0001). Analysis of four studies presenting data on catheterisation time [24], [26], [27], and [33] resulted in a statistically significant mean difference for TURis versus M-TURP of −0.23 d (95% CI, −0.38 to −0.09; p = 0.0019); however, this difference is not clinically meaningful because such a minimal reduction in catheterisation time is unlikely to make a considerable difference to patients. Six studies reported procedure duration [24], [26], [29], [31], [33], and [34] and, although not significant, the resulting pooled mean difference for TURis versus M-TURP was −0.95 min (95% CI, −3.35 to 1.46; p > 0.05).

3.3. Economic analysis

Only statistically significant differences in clinical outcomes in the meta-analysis, which directly affect costs, were considered in the base-case analysis: TUR syndrome, blood transfusions, clot retention, and length of hospital stay. Table 3 shows the results of the base-case analysis. Costs of adopting TURis differed according to whether or not hospitals owned an Olympus monopolar system. Based on 150 TURP procedures per annum, capital costs of TURis (Supplementary Table 1) for hospitals with existing Olympus systems and for those without Olympus systems were £9.68 and £29.13 per patient per year, respectively. The total cost per patient (including capital equipment, consumables, hospital stay, and complications) of using TURis for hospitals with Olympus systems in place was £881.71 versus £1085.49 for M-TURP, a cost saving of £203.78 per patient. In hospitals without Olympus systems, the total per patient cost of using TURis was £901.16 compared with £1014.58 for M-TURP, a cost saving of £113.42 per patient.

Higher equipment costs for TURis versus M-TURP were offset by savings from reductions in hospital stay and complications. Assuming a reduction in hospital stay of 0.56 d from the meta-analysis, the base-case analysis found that TURis was associated with a £165.02 reduction in the cost of hospital stay per patient versus M-TURP. TURis was also associated with a £71.82 per patient reduction in complication costs compared with M-TURP.

A further potential benefit of the TURis system is the option of conducting surgery on a day-case basis. Some UK centres are already adopting this practice [35] and [36], and the potential for further cost savings in hospital stay costs via this practice was acknowledged by NICE in MTG23 [20]. Although in clinical practice various factors influence the implementation of day-case surgery, testing a scenario reflecting this situation, in which patients undergoing TURis incurred no inpatient day costs, cost savings were £891.36 and £801.00 per patient, for Olympus and non-Olympus centres, respectively.

Considering readmissions in a further scenario, TURis was associated with annual cost savings versus M-TURP of £245.10 and £154.75 per patient, for Olympus and non-Olympus centres, respectively (Table 3). Readmission for BPH is not widely reported, thus data relating to this outcome are limited; the scenario analysis was based on a single study [23].

Sensitivity analysis also demonstrated that TURis remained cost saving versus M-TURP when each tested model parameter was varied according to its defined range. Key drivers of cost savings were reduction in hospital stay and inpatient day costs (Fig. 2). Results of PSA demonstrated that the cost-consequence analysis was robust across both types of centre. For Olympus centres, use of TURis was associated with a mean saving of £200.77 per patient (95% CI, −£115.54 to −£298.67) and remained cost saving in all 1000 simulations. For Olympus centres, the mean saving was £112.93 per patient (95% CI, £−32.94 to £−211.22), and TURis remained cost saving in 99.8% of simulations.

Results of the threshold analysis on model parameters (Table 4) confirmed the robustness of the base-case results. Existing Olympus centres would need to perform as few as seven procedures annually for TURis to become cost saving versus M-TURP. For non-Olympus centres, the number is 31. Keeping all other model parameters constant, TURis would need to be associated with significantly increased risks of blood transfusion and clot retention compared with M-TURP to cease being cost saving overall. Based on the findings of the meta-analysis, which showed statistically significant reductions in these events for TURis versus TURP, it is not considered plausible that RRs as high as those reported in Table 4 would be observed for TURis versus M-TURP.

Significant increases in the annual cost of capital equipment for TURis would be required for TURis to cease being cost saving versus M-TURP. This could only happen as a result of short equipment life span in practice because prices are generally stable over a period such as the modelled time horizon. The shortened life spans that would be required for TURis to cease being cost saving versus M-TURP (0.28 yr for existing Olympus centres and 1.30 yr for non-Olympus centres) demonstrate the robustness of the conclusions of the analysis to changes in TURis capital costs. TURis would cease being cost saving if the reduction in hospital stay for TURis versus M-TURP was removed (only for non-Olympus centres).

3.4. Discussion

This study presents current and compelling clinical and economic evidence supporting adoption of the TURis system as an alternative to M-TURP, which is the current UK surgical standard of care for the treatment of BPH. The analysis was based on an SR of RCT evidence and synthesis of those data for both systems, with a subsequent economic analysis conducted from the perspective of the NHS in England and Wales. The data synthesis (by meta-analysis) showed that TURis was associated with significant improvements in perioperative safety and in duration of hospital stay compared with M-TURP. The results of the base-case economic analysis found that the adoption of TURis could result in cost savings versus M-TURP of £204 and £113 per patient for existing and non-Olympus TURP centres, respectively. Sensitivity analysis demonstrated that reduction in hospital stay was a key driver of results. Inclusion of readmissions resulted in cost savings with TURis versus M-TURP of £245 and £155 per patient for Olympus and non-Olympus centres, respectively. Although this analysis was based on a single RCT (of good quality), NICE considered it sufficiently robust [20].

Although efficacy in prostate resection surgery for treatment of voiding symptoms presumed secondary to BPH is well established, possible perioperative complications and associated costs remain a concern. Current European guidelines state that B-TURP offers similar efficacy to M-TURP but with reduced perioperative morbidity. The guidelines recommend that the choice of B-TURP should be based on surgeon experience, availability of equipment, and patient preference [6]. Although there is currently no European consensus on the use of a specific B-TURP system, in England and Wales the recently published NICE MTG23 guidance concludes that the available evidence supports the adoption of TURis for surgical treatment of BPH and that the TURis system is clinically equivalent to M-TURP, but with the advantage of reducing the risk of perioperative complications [20]. It further states that TURis has the potential to be cost saving compared with M-TURP [20].

The SR presented here demonstrates that the TURis system is supported by strong clinical data in the form of 15 good quality RCTs. To our knowledge, this is the first study to evaluate the wealth of RCT evidence in support of TURis for conducting meta-analysis to support an economic evaluation.

The findings from the economic analysis may have important implications for decision makers. The results suggest that NHS hospitals in England and Wales adopting TURis specifically as an alternative to M-TURP can expect both cost savings and improved clinical outcomes. Based on the results of the meta-analyses, treating 56 people with TURis instead of M-TURP could prevent one case of TUR syndrome. The numbers needed to treat to avoid one blood transfusion and one case of clot retention are 22 and 40, respectively. Because Hospital Episode Statistics data indicate that approximately 14 500 TURP procedures were performed out in England in 2012–2013, approximately 260 cases of TUR syndrome might be avoided if TURis was to replace M-TURP, along with 660 blood transfusions and 360 cases of clot retention. Given the complication costs outlined in Supplementary Table 1, avoiding these complications alone might result in an annual cost savings of >£1 million in England. When reductions in hospital stay are considered, the annual cost savings exceed £3.4 million. Beyond potential cost savings, this represents efficiency improvements for hospitals, enhancing patient management and throughput.

Introduction of medical devices into clinical practice is often limited by availability of high-quality clinical evidence, an essential component of robust economic evaluation of any medical intervention. Large-scale RCTs are conventionally used as the gold standard for demonstrating clinical efficacy of drug therapies; however, medical devices face a range of challenges in RCT design that often affect quality. For example, randomisation may not be possible due to ethical constraints, and blinding (especially double blinding) may not be feasible, thus increasing the risk of bias [37] and [38]. The cohort size and timescale of follow-up and assessments may also be limited [38]. Consequently, the clinical evidence base for a medical device can often be weak. In contrast, this study benefits from high-quality RCT evidence, acknowledged in the NICE MTG23 guidance, comprising 15 studies that show the clinical evidence for TURis supports the adoption of the technology. Patients included in the studies have undergone standard M-TURP or TURis procedures and no deviations in procedure protocol were observed, providing an evidence base with low heterogeneity.

A further strength of the economic analysis is that we only included those clinical outcomes for which meta-analysis showed statistically significant differences between the two treatments. Assumptions regarding resource use for complications were based on NICE clinical guideline 97 [39] and had been validated by two clinical experts for the MTEP assessment.

The analysis is not without limitations. For meta-analysis, a random-effects model is generally considered more robust because it takes heterogeneity into account. The random-effects model is only recommended, however, if ≥10 studies are available for an end point, which was not the case here. We therefore performed a fixed-effects meta-analysis. Second, complications would vary in severity; the costs considered in the analysis are intended to reflect the typical cost of treating each event. Because the model is based on UK cost data, the results are not generalizable to other countries.

2.1. Methods

An SR was performed to identify clinical trials to form the basis of a meta-analysis. The results of the clinical meta-analysis form one part of our objective. They were also used as input for an economic analysis, the second part of our objective.

2.2. Systematic review of the clinical evidence

The SR was performed in line with guidance by the Centre for Reviews and Dissemination, University of York [22]. The detailed SR methodology, as well as results and quality assessment, is reported in the supplementary information for MTG23 [20]. We performed an update to the original search on April 20, 2015, limited to randomised controlled trial (RCT) publications from 2014 onwards. The output from this update was then combined with that from the original SR (Table 1).

2.3. Meta-analysis of the clinical evidence

Expert opinion solicited during development of the MTEP submission suggested that differences are most commonly observed between TURis and M-TURP in the incidence of TUR syndrome, blood transfusions, clot retention, and hospital stay. The evidence was synthesised using pairwise meta-analysis techniques to determine any evidence-based differences between the treatments in these outcomes, as well as in other commonly reported outcomes including urethral strictures, catheterisation time, and procedure duration. According to the Cochrane Systematic Review Handbook, a minimum of 10 studies is required to draw meaningful conclusions from random-effects calculations [5]. Because each meta-analysis undertaken involved <10 studies, fixed-effects analyses were conducted. Details of the meta-analysis methodology were previously reported [20].

2.4. Economic evaluation

The economic analysis (a cost-consequence analysis) compared costs and outcomes associated with TURis and M-TURP, respectively, for surgical treatment of BPH. The perspective of the analysis was that of the National Health Service (NHS) and Personal Social Services in England and Wales.

2.5. Patient population

The analysis considered adult men with LUTS presumed secondary to BPH, in whom surgical intervention was indicated.

2.6. Model structure

A decision-tree model was constructed in which men with BPH-related LUTS underwent surgery with either TURis or M-TURP. Patients entered the model at the point of being indicated for surgery and were allocated probabilities of experiencing procedure-related adverse events (TUR syndrome, blood transfusion, and clot retention) based on the meta-analysis output. Procedure duration was excluded from the analysis because no statistically significant difference was observed between M-TURP and TURis in the meta-analysis. The pooled reduction in catheterisation time estimated from the meta-analysis of 0.2 d for TURis versus M-TURP was statistically significant. However, given that a patient's catheter would generally be removed while in the hospital, this reduction is not expected to result in cost savings beyond those realised through the reduction in hospital stay, which was modelled. As such, catheterisation time was not modelled explicitly. The model considered a nondefined short time horizon covering the prostate resection procedure and immediately afterwards—the period during which the complications modelled are expected to occur. Capital costs assumed equipment life spans of 7 yr. Based on findings of the SR, no difference in functional outcomes (eg, efficacy) was assumed between the procedures with respect to prostate resection weight or radicality.

2.7. Model parameters and data sources

Supplementary Table 1 lists all the variables included in the analysis. Sourcing of inputs, rationale, and further details are provided in Supplement 1.

2.8. Sensitivity analysis

Sensitivity analyses (based on predefined scenarios) were conducted. The first scenario considered all-cause readmission following a TURP procedure, based on data from the single study identified, Fagerstrom et al [23], which indicated a threefold increase in all-cause readmission for M-TURP versus TURis. This study reported readmissions as a result of bleeding, infection, and “other causes.” Because the base-case analysis included clot retention, the scenario considered readmissions from infection and other causes to avoid double counting (it is assumed that “bleeding” includes clot retention). A separate scenario considered a situation in which TURis is performed as day-case surgery.

One-way deterministic sensitivity analysis was performed in which all model parameters were varied either according to published ranges or by ±25%. Results are presented as Tornado diagrams. Probabilistic sensitivity analysis (PSA) was also conducted, in which input parameters were varied according to specified probability distributions. Threshold analysis was conducted to explore the relationship between key model parameters and model results up to the point at which the main conclusions of the model might change, that is, in this case to determine the value at which TURis would be cost neutral (ie, no longer cost saving) versus M-TURP.

3.1. Clinical evidence

The original electronic searches identified a total of 1116 publications. Following title and abstract review, 783 were excluded and 14 relevant RCTs remained, following assessment of the full paper. No unpublished studies were identified from ClinicalTrials.gov. The updated search identified a total 63 further publications, of which two full-text RCT publications remained after review of title and abstract [24] and [25]. On review of the full text, it was found that both of these publications reported data from the same trial, and only the most recent publication, reporting the most up-to-date data, was included in the analysis [24]. Figure 1 shows the study selection flow diagram.

Quality assessment of the identified RCT (Supplementary Table 2) showed that the patient population was consistent with the target population of the analysis. Study end points were clear and appropriately measured according to study design and follow-up schedule.

3.2. Meta-analysis

Six studies were found to be suitable for inclusion in the pairwise meta-analysis of the incidence of TUR syndrome [23], [26], [27], [28], [29], and [30]. The total number of events reported in the studies is presented in Table 2. There were no TUR syndrome events in the TURis arm reported in any of the studies, consistent with the use of saline. The analysis estimated a pooled relative risk (RR) for TUR syndrome of 0.18 (95% confidence interval [CI], 0.05–0.61; p = 0.006) in favour of TURis.

Eight RCTs reported the incidence of blood transfusion, but only seven of these were included in the meta-analysis [23], [24], [26], [27], [28], [29], and [31] (Table 2). The study by Michielsen et al was excluded from the analysis because a higher proportion of TURis procedures was conducted by trainee surgeons compared with M-TURP (32% vs 7%, respectively), resulting in increased procedure duration in this arm as well as increased blood loss and, consequently, blood transfusions [32]. Exclusion of this study on the basis of heterogeneity was accepted by the External Assessment Centre for MTG23 [19]. The results of the meta-analysis showed a statistically significant reduction in blood transfusions with use of TURis compared with M-TURP (RR: 0.34; 95% CI, 0.18–0.61; p = 0.0003).

Six studies were identified for meta-analysis of the incidence of clot retention [24], [26], [28], [29], [31], and [32]. The pooled RR was 0.43 in favour of TURis (95% CI, 0.22–0.86; p = 0.0161), demonstrating that use of TURis is associated with fewer cases of clot retention than M-TURP (Table 2).

Meta-analysis was conducted on seven studies reporting data on urethral strictures [23], [24], [26], [28], [29], [31], and [33]. A non-significantly different pooled RR of 1.27 (95% CI, 0.85–1.9; p > 0.05) was observed for TURis versus M-TURP (Table 2). As a result, this outcome was not considered in the economic analysis.

Further meta-analyses suggested that TURis is associated with statistically significant reductions in hospital stay and catheterisation time (Table 2). For these outcomes, meta-analyses were restricted to studies reporting both the mean duration and standard deviation. From a pooled analysis of four studies reporting length of hospital stay [24], [26], [27], and [33], the mean difference for TURis versus M-TURP was −0.56 d (95% CI, −0.77 to −0.35; p < 0.0001). Analysis of four studies presenting data on catheterisation time [24], [26], [27], and [33] resulted in a statistically significant mean difference for TURis versus M-TURP of −0.23 d (95% CI, −0.38 to −0.09; p = 0.0019); however, this difference is not clinically meaningful because such a minimal reduction in catheterisation time is unlikely to make a considerable difference to patients. Six studies reported procedure duration [24], [26], [29], [31], [33], and [34] and, although not significant, the resulting pooled mean difference for TURis versus M-TURP was −0.95 min (95% CI, −3.35 to 1.46; p > 0.05).

3.3. Economic analysis

Only statistically significant differences in clinical outcomes in the meta-analysis, which directly affect costs, were considered in the base-case analysis: TUR syndrome, blood transfusions, clot retention, and length of hospital stay. Table 3 shows the results of the base-case analysis. Costs of adopting TURis differed according to whether or not hospitals owned an Olympus monopolar system. Based on 150 TURP procedures per annum, capital costs of TURis (Supplementary Table 1) for hospitals with existing Olympus systems and for those without Olympus systems were £9.68 and £29.13 per patient per year, respectively. The total cost per patient (including capital equipment, consumables, hospital stay, and complications) of using TURis for hospitals with Olympus systems in place was £881.71 versus £1085.49 for M-TURP, a cost saving of £203.78 per patient. In hospitals without Olympus systems, the total per patient cost of using TURis was £901.16 compared with £1014.58 for M-TURP, a cost saving of £113.42 per patient.

Higher equipment costs for TURis versus M-TURP were offset by savings from reductions in hospital stay and complications. Assuming a reduction in hospital stay of 0.56 d from the meta-analysis, the base-case analysis found that TURis was associated with a £165.02 reduction in the cost of hospital stay per patient versus M-TURP. TURis was also associated with a £71.82 per patient reduction in complication costs compared with M-TURP.

A further potential benefit of the TURis system is the option of conducting surgery on a day-case basis. Some UK centres are already adopting this practice [35] and [36], and the potential for further cost savings in hospital stay costs via this practice was acknowledged by NICE in MTG23 [20]. Although in clinical practice various factors influence the implementation of day-case surgery, testing a scenario reflecting this situation, in which patients undergoing TURis incurred no inpatient day costs, cost savings were £891.36 and £801.00 per patient, for Olympus and non-Olympus centres, respectively.

Considering readmissions in a further scenario, TURis was associated with annual cost savings versus M-TURP of £245.10 and £154.75 per patient, for Olympus and non-Olympus centres, respectively (Table 3). Readmission for BPH is not widely reported, thus data relating to this outcome are limited; the scenario analysis was based on a single study [23].

Sensitivity analysis also demonstrated that TURis remained cost saving versus M-TURP when each tested model parameter was varied according to its defined range. Key drivers of cost savings were reduction in hospital stay and inpatient day costs (Fig. 2). Results of PSA demonstrated that the cost-consequence analysis was robust across both types of centre. For Olympus centres, use of TURis was associated with a mean saving of £200.77 per patient (95% CI, −£115.54 to −£298.67) and remained cost saving in all 1000 simulations. For Olympus centres, the mean saving was £112.93 per patient (95% CI, £−32.94 to £−211.22), and TURis remained cost saving in 99.8% of simulations.

Results of the threshold analysis on model parameters (Table 4) confirmed the robustness of the base-case results. Existing Olympus centres would need to perform as few as seven procedures annually for TURis to become cost saving versus M-TURP. For non-Olympus centres, the number is 31. Keeping all other model parameters constant, TURis would need to be associated with significantly increased risks of blood transfusion and clot retention compared with M-TURP to cease being cost saving overall. Based on the findings of the meta-analysis, which showed statistically significant reductions in these events for TURis versus TURP, it is not considered plausible that RRs as high as those reported in Table 4 would be observed for TURis versus M-TURP.

Significant increases in the annual cost of capital equipment for TURis would be required for TURis to cease being cost saving versus M-TURP. This could only happen as a result of short equipment life span in practice because prices are generally stable over a period such as the modelled time horizon. The shortened life spans that would be required for TURis to cease being cost saving versus M-TURP (0.28 yr for existing Olympus centres and 1.30 yr for non-Olympus centres) demonstrate the robustness of the conclusions of the analysis to changes in TURis capital costs. TURis would cease being cost saving if the reduction in hospital stay for TURis versus M-TURP was removed (only for non-Olympus centres).

3.4. Discussion

This study presents current and compelling clinical and economic evidence supporting adoption of the TURis system as an alternative to M-TURP, which is the current UK surgical standard of care for the treatment of BPH. The analysis was based on an SR of RCT evidence and synthesis of those data for both systems, with a subsequent economic analysis conducted from the perspective of the NHS in England and Wales. The data synthesis (by meta-analysis) showed that TURis was associated with significant improvements in perioperative safety and in duration of hospital stay compared with M-TURP. The results of the base-case economic analysis found that the adoption of TURis could result in cost savings versus M-TURP of £204 and £113 per patient for existing and non-Olympus TURP centres, respectively. Sensitivity analysis demonstrated that reduction in hospital stay was a key driver of results. Inclusion of readmissions resulted in cost savings with TURis versus M-TURP of £245 and £155 per patient for Olympus and non-Olympus centres, respectively. Although this analysis was based on a single RCT (of good quality), NICE considered it sufficiently robust [20].

Although efficacy in prostate resection surgery for treatment of voiding symptoms presumed secondary to BPH is well established, possible perioperative complications and associated costs remain a concern. Current European guidelines state that B-TURP offers similar efficacy to M-TURP but with reduced perioperative morbidity. The guidelines recommend that the choice of B-TURP should be based on surgeon experience, availability of equipment, and patient preference [6]. Although there is currently no European consensus on the use of a specific B-TURP system, in England and Wales the recently published NICE MTG23 guidance concludes that the available evidence supports the adoption of TURis for surgical treatment of BPH and that the TURis system is clinically equivalent to M-TURP, but with the advantage of reducing the risk of perioperative complications [20]. It further states that TURis has the potential to be cost saving compared with M-TURP [20].

The SR presented here demonstrates that the TURis system is supported by strong clinical data in the form of 15 good quality RCTs. To our knowledge, this is the first study to evaluate the wealth of RCT evidence in support of TURis for conducting meta-analysis to support an economic evaluation.

The findings from the economic analysis may have important implications for decision makers. The results suggest that NHS hospitals in England and Wales adopting TURis specifically as an alternative to M-TURP can expect both cost savings and improved clinical outcomes. Based on the results of the meta-analyses, treating 56 people with TURis instead of M-TURP could prevent one case of TUR syndrome. The numbers needed to treat to avoid one blood transfusion and one case of clot retention are 22 and 40, respectively. Because Hospital Episode Statistics data indicate that approximately 14 500 TURP procedures were performed out in England in 2012–2013, approximately 260 cases of TUR syndrome might be avoided if TURis was to replace M-TURP, along with 660 blood transfusions and 360 cases of clot retention. Given the complication costs outlined in Supplementary Table 1, avoiding these complications alone might result in an annual cost savings of >£1 million in England. When reductions in hospital stay are considered, the annual cost savings exceed £3.4 million. Beyond potential cost savings, this represents efficiency improvements for hospitals, enhancing patient management and throughput.

Introduction of medical devices into clinical practice is often limited by availability of high-quality clinical evidence, an essential component of robust economic evaluation of any medical intervention. Large-scale RCTs are conventionally used as the gold standard for demonstrating clinical efficacy of drug therapies; however, medical devices face a range of challenges in RCT design that often affect quality. For example, randomisation may not be possible due to ethical constraints, and blinding (especially double blinding) may not be feasible, thus increasing the risk of bias [37] and [38]. The cohort size and timescale of follow-up and assessments may also be limited [38]. Consequently, the clinical evidence base for a medical device can often be weak. In contrast, this study benefits from high-quality RCT evidence, acknowledged in the NICE MTG23 guidance, comprising 15 studies that show the clinical evidence for TURis supports the adoption of the technology. Patients included in the studies have undergone standard M-TURP or TURis procedures and no deviations in procedure protocol were observed, providing an evidence base with low heterogeneity.

A further strength of the economic analysis is that we only included those clinical outcomes for which meta-analysis showed statistically significant differences between the two treatments. Assumptions regarding resource use for complications were based on NICE clinical guideline 97 [39] and had been validated by two clinical experts for the MTEP assessment.

The analysis is not without limitations. For meta-analysis, a random-effects model is generally considered more robust because it takes heterogeneity into account. The random-effects model is only recommended, however, if ≥10 studies are available for an end point, which was not the case here. We therefore performed a fixed-effects meta-analysis. Second, complications would vary in severity; the costs considered in the analysis are intended to reflect the typical cost of treating each event. Because the model is based on UK cost data, the results are not generalizable to other countries.

2.1. Methods

An SR was performed to identify clinical trials to form the basis of a meta-analysis. The results of the clinical meta-analysis form one part of our objective. They were also used as input for an economic analysis, the second part of our objective.

2.2. Systematic review of the clinical evidence

The SR was performed in line with guidance by the Centre for Reviews and Dissemination, University of York [22]. The detailed SR methodology, as well as results and quality assessment, is reported in the supplementary information for MTG23 [20]. We performed an update to the original search on April 20, 2015, limited to randomised controlled trial (RCT) publications from 2014 onwards. The output from this update was then combined with that from the original SR (Table 1).

2.3. Meta-analysis of the clinical evidence

Expert opinion solicited during development of the MTEP submission suggested that differences are most commonly observed between TURis and M-TURP in the incidence of TUR syndrome, blood transfusions, clot retention, and hospital stay. The evidence was synthesised using pairwise meta-analysis techniques to determine any evidence-based differences between the treatments in these outcomes, as well as in other commonly reported outcomes including urethral strictures, catheterisation time, and procedure duration. According to the Cochrane Systematic Review Handbook, a minimum of 10 studies is required to draw meaningful conclusions from random-effects calculations [5]. Because each meta-analysis undertaken involved <10 studies, fixed-effects analyses were conducted. Details of the meta-analysis methodology were previously reported [20].

2.4. Economic evaluation

The economic analysis (a cost-consequence analysis) compared costs and outcomes associated with TURis and M-TURP, respectively, for surgical treatment of BPH. The perspective of the analysis was that of the National Health Service (NHS) and Personal Social Services in England and Wales.

2.5. Patient population

The analysis considered adult men with LUTS presumed secondary to BPH, in whom surgical intervention was indicated.

2.6. Model structure

A decision-tree model was constructed in which men with BPH-related LUTS underwent surgery with either TURis or M-TURP. Patients entered the model at the point of being indicated for surgery and were allocated probabilities of experiencing procedure-related adverse events (TUR syndrome, blood transfusion, and clot retention) based on the meta-analysis output. Procedure duration was excluded from the analysis because no statistically significant difference was observed between M-TURP and TURis in the meta-analysis. The pooled reduction in catheterisation time estimated from the meta-analysis of 0.2 d for TURis versus M-TURP was statistically significant. However, given that a patient's catheter would generally be removed while in the hospital, this reduction is not expected to result in cost savings beyond those realised through the reduction in hospital stay, which was modelled. As such, catheterisation time was not modelled explicitly. The model considered a nondefined short time horizon covering the prostate resection procedure and immediately afterwards—the period during which the complications modelled are expected to occur. Capital costs assumed equipment life spans of 7 yr. Based on findings of the SR, no difference in functional outcomes (eg, efficacy) was assumed between the procedures with respect to prostate resection weight or radicality.

2.7. Model parameters and data sources

Supplementary Table 1 lists all the variables included in the analysis. Sourcing of inputs, rationale, and further details are provided in Supplement 1.

2.8. Sensitivity analysis

Sensitivity analyses (based on predefined scenarios) were conducted. The first scenario considered all-cause readmission following a TURP procedure, based on data from the single study identified, Fagerstrom et al [23], which indicated a threefold increase in all-cause readmission for M-TURP versus TURis. This study reported readmissions as a result of bleeding, infection, and “other causes.” Because the base-case analysis included clot retention, the scenario considered readmissions from infection and other causes to avoid double counting (it is assumed that “bleeding” includes clot retention). A separate scenario considered a situation in which TURis is performed as day-case surgery.

One-way deterministic sensitivity analysis was performed in which all model parameters were varied either according to published ranges or by ±25%. Results are presented as Tornado diagrams. Probabilistic sensitivity analysis (PSA) was also conducted, in which input parameters were varied according to specified probability distributions. Threshold analysis was conducted to explore the relationship between key model parameters and model results up to the point at which the main conclusions of the model might change, that is, in this case to determine the value at which TURis would be cost neutral (ie, no longer cost saving) versus M-TURP.

3.1. Clinical evidence

The original electronic searches identified a total of 1116 publications. Following title and abstract review, 783 were excluded and 14 relevant RCTs remained, following assessment of the full paper. No unpublished studies were identified from ClinicalTrials.gov. The updated search identified a total 63 further publications, of which two full-text RCT publications remained after review of title and abstract [24] and [25]. On review of the full text, it was found that both of these publications reported data from the same trial, and only the most recent publication, reporting the most up-to-date data, was included in the analysis [24]. Figure 1 shows the study selection flow diagram.

Quality assessment of the identified RCT (Supplementary Table 2) showed that the patient population was consistent with the target population of the analysis. Study end points were clear and appropriately measured according to study design and follow-up schedule.

3.2. Meta-analysis

Six studies were found to be suitable for inclusion in the pairwise meta-analysis of the incidence of TUR syndrome [23], [26], [27], [28], [29], and [30]. The total number of events reported in the studies is presented in Table 2. There were no TUR syndrome events in the TURis arm reported in any of the studies, consistent with the use of saline. The analysis estimated a pooled relative risk (RR) for TUR syndrome of 0.18 (95% confidence interval [CI], 0.05–0.61; p = 0.006) in favour of TURis.

Eight RCTs reported the incidence of blood transfusion, but only seven of these were included in the meta-analysis [23], [24], [26], [27], [28], [29], and [31] (Table 2). The study by Michielsen et al was excluded from the analysis because a higher proportion of TURis procedures was conducted by trainee surgeons compared with M-TURP (32% vs 7%, respectively), resulting in increased procedure duration in this arm as well as increased blood loss and, consequently, blood transfusions [32]. Exclusion of this study on the basis of heterogeneity was accepted by the External Assessment Centre for MTG23 [19]. The results of the meta-analysis showed a statistically significant reduction in blood transfusions with use of TURis compared with M-TURP (RR: 0.34; 95% CI, 0.18–0.61; p = 0.0003).

Six studies were identified for meta-analysis of the incidence of clot retention [24], [26], [28], [29], [31], and [32]. The pooled RR was 0.43 in favour of TURis (95% CI, 0.22–0.86; p = 0.0161), demonstrating that use of TURis is associated with fewer cases of clot retention than M-TURP (Table 2).

Meta-analysis was conducted on seven studies reporting data on urethral strictures [23], [24], [26], [28], [29], [31], and [33]. A non-significantly different pooled RR of 1.27 (95% CI, 0.85–1.9; p > 0.05) was observed for TURis versus M-TURP (Table 2). As a result, this outcome was not considered in the economic analysis.

Further meta-analyses suggested that TURis is associated with statistically significant reductions in hospital stay and catheterisation time (Table 2). For these outcomes, meta-analyses were restricted to studies reporting both the mean duration and standard deviation. From a pooled analysis of four studies reporting length of hospital stay [24], [26], [27], and [33], the mean difference for TURis versus M-TURP was −0.56 d (95% CI, −0.77 to −0.35; p < 0.0001). Analysis of four studies presenting data on catheterisation time [24], [26], [27], and [33] resulted in a statistically significant mean difference for TURis versus M-TURP of −0.23 d (95% CI, −0.38 to −0.09; p = 0.0019); however, this difference is not clinically meaningful because such a minimal reduction in catheterisation time is unlikely to make a considerable difference to patients. Six studies reported procedure duration [24], [26], [29], [31], [33], and [34] and, although not significant, the resulting pooled mean difference for TURis versus M-TURP was −0.95 min (95% CI, −3.35 to 1.46; p > 0.05).

3.3. Economic analysis

Only statistically significant differences in clinical outcomes in the meta-analysis, which directly affect costs, were considered in the base-case analysis: TUR syndrome, blood transfusions, clot retention, and length of hospital stay. Table 3 shows the results of the base-case analysis. Costs of adopting TURis differed according to whether or not hospitals owned an Olympus monopolar system. Based on 150 TURP procedures per annum, capital costs of TURis (Supplementary Table 1) for hospitals with existing Olympus systems and for those without Olympus systems were £9.68 and £29.13 per patient per year, respectively. The total cost per patient (including capital equipment, consumables, hospital stay, and complications) of using TURis for hospitals with Olympus systems in place was £881.71 versus £1085.49 for M-TURP, a cost saving of £203.78 per patient. In hospitals without Olympus systems, the total per patient cost of using TURis was £901.16 compared with £1014.58 for M-TURP, a cost saving of £113.42 per patient.

Higher equipment costs for TURis versus M-TURP were offset by savings from reductions in hospital stay and complications. Assuming a reduction in hospital stay of 0.56 d from the meta-analysis, the base-case analysis found that TURis was associated with a £165.02 reduction in the cost of hospital stay per patient versus M-TURP. TURis was also associated with a £71.82 per patient reduction in complication costs compared with M-TURP.

A further potential benefit of the TURis system is the option of conducting surgery on a day-case basis. Some UK centres are already adopting this practice [35] and [36], and the potential for further cost savings in hospital stay costs via this practice was acknowledged by NICE in MTG23 [20]. Although in clinical practice various factors influence the implementation of day-case surgery, testing a scenario reflecting this situation, in which patients undergoing TURis incurred no inpatient day costs, cost savings were £891.36 and £801.00 per patient, for Olympus and non-Olympus centres, respectively.

Considering readmissions in a further scenario, TURis was associated with annual cost savings versus M-TURP of £245.10 and £154.75 per patient, for Olympus and non-Olympus centres, respectively (Table 3). Readmission for BPH is not widely reported, thus data relating to this outcome are limited; the scenario analysis was based on a single study [23].

Sensitivity analysis also demonstrated that TURis remained cost saving versus M-TURP when each tested model parameter was varied according to its defined range. Key drivers of cost savings were reduction in hospital stay and inpatient day costs (Fig. 2). Results of PSA demonstrated that the cost-consequence analysis was robust across both types of centre. For Olympus centres, use of TURis was associated with a mean saving of £200.77 per patient (95% CI, −£115.54 to −£298.67) and remained cost saving in all 1000 simulations. For Olympus centres, the mean saving was £112.93 per patient (95% CI, £−32.94 to £−211.22), and TURis remained cost saving in 99.8% of simulations.

Results of the threshold analysis on model parameters (Table 4) confirmed the robustness of the base-case results. Existing Olympus centres would need to perform as few as seven procedures annually for TURis to become cost saving versus M-TURP. For non-Olympus centres, the number is 31. Keeping all other model parameters constant, TURis would need to be associated with significantly increased risks of blood transfusion and clot retention compared with M-TURP to cease being cost saving overall. Based on the findings of the meta-analysis, which showed statistically significant reductions in these events for TURis versus TURP, it is not considered plausible that RRs as high as those reported in Table 4 would be observed for TURis versus M-TURP.

Significant increases in the annual cost of capital equipment for TURis would be required for TURis to cease being cost saving versus M-TURP. This could only happen as a result of short equipment life span in practice because prices are generally stable over a period such as the modelled time horizon. The shortened life spans that would be required for TURis to cease being cost saving versus M-TURP (0.28 yr for existing Olympus centres and 1.30 yr for non-Olympus centres) demonstrate the robustness of the conclusions of the analysis to changes in TURis capital costs. TURis would cease being cost saving if the reduction in hospital stay for TURis versus M-TURP was removed (only for non-Olympus centres).

3.4. Discussion

This study presents current and compelling clinical and economic evidence supporting adoption of the TURis system as an alternative to M-TURP, which is the current UK surgical standard of care for the treatment of BPH. The analysis was based on an SR of RCT evidence and synthesis of those data for both systems, with a subsequent economic analysis conducted from the perspective of the NHS in England and Wales. The data synthesis (by meta-analysis) showed that TURis was associated with significant improvements in perioperative safety and in duration of hospital stay compared with M-TURP. The results of the base-case economic analysis found that the adoption of TURis could result in cost savings versus M-TURP of £204 and £113 per patient for existing and non-Olympus TURP centres, respectively. Sensitivity analysis demonstrated that reduction in hospital stay was a key driver of results. Inclusion of readmissions resulted in cost savings with TURis versus M-TURP of £245 and £155 per patient for Olympus and non-Olympus centres, respectively. Although this analysis was based on a single RCT (of good quality), NICE considered it sufficiently robust [20].

Although efficacy in prostate resection surgery for treatment of voiding symptoms presumed secondary to BPH is well established, possible perioperative complications and associated costs remain a concern. Current European guidelines state that B-TURP offers similar efficacy to M-TURP but with reduced perioperative morbidity. The guidelines recommend that the choice of B-TURP should be based on surgeon experience, availability of equipment, and patient preference [6]. Although there is currently no European consensus on the use of a specific B-TURP system, in England and Wales the recently published NICE MTG23 guidance concludes that the available evidence supports the adoption of TURis for surgical treatment of BPH and that the TURis system is clinically equivalent to M-TURP, but with the advantage of reducing the risk of perioperative complications [20]. It further states that TURis has the potential to be cost saving compared with M-TURP [20].

The SR presented here demonstrates that the TURis system is supported by strong clinical data in the form of 15 good quality RCTs. To our knowledge, this is the first study to evaluate the wealth of RCT evidence in support of TURis for conducting meta-analysis to support an economic evaluation.

The findings from the economic analysis may have important implications for decision makers. The results suggest that NHS hospitals in England and Wales adopting TURis specifically as an alternative to M-TURP can expect both cost savings and improved clinical outcomes. Based on the results of the meta-analyses, treating 56 people with TURis instead of M-TURP could prevent one case of TUR syndrome. The numbers needed to treat to avoid one blood transfusion and one case of clot retention are 22 and 40, respectively. Because Hospital Episode Statistics data indicate that approximately 14 500 TURP procedures were performed out in England in 2012–2013, approximately 260 cases of TUR syndrome might be avoided if TURis was to replace M-TURP, along with 660 blood transfusions and 360 cases of clot retention. Given the complication costs outlined in Supplementary Table 1, avoiding these complications alone might result in an annual cost savings of >£1 million in England. When reductions in hospital stay are considered, the annual cost savings exceed £3.4 million. Beyond potential cost savings, this represents efficiency improvements for hospitals, enhancing patient management and throughput.

Introduction of medical devices into clinical practice is often limited by availability of high-quality clinical evidence, an essential component of robust economic evaluation of any medical intervention. Large-scale RCTs are conventionally used as the gold standard for demonstrating clinical efficacy of drug therapies; however, medical devices face a range of challenges in RCT design that often affect quality. For example, randomisation may not be possible due to ethical constraints, and blinding (especially double blinding) may not be feasible, thus increasing the risk of bias [37] and [38]. The cohort size and timescale of follow-up and assessments may also be limited [38]. Consequently, the clinical evidence base for a medical device can often be weak. In contrast, this study benefits from high-quality RCT evidence, acknowledged in the NICE MTG23 guidance, comprising 15 studies that show the clinical evidence for TURis supports the adoption of the technology. Patients included in the studies have undergone standard M-TURP or TURis procedures and no deviations in procedure protocol were observed, providing an evidence base with low heterogeneity.

A further strength of the economic analysis is that we only included those clinical outcomes for which meta-analysis showed statistically significant differences between the two treatments. Assumptions regarding resource use for complications were based on NICE clinical guideline 97 [39] and had been validated by two clinical experts for the MTEP assessment.

The analysis is not without limitations. For meta-analysis, a random-effects model is generally considered more robust because it takes heterogeneity into account. The random-effects model is only recommended, however, if ≥10 studies are available for an end point, which was not the case here. We therefore performed a fixed-effects meta-analysis. Second, complications would vary in severity; the costs considered in the analysis are intended to reflect the typical cost of treating each event. Because the model is based on UK cost data, the results are not generalizable to other countries.

2.1. Methods

An SR was performed to identify clinical trials to form the basis of a meta-analysis. The results of the clinical meta-analysis form one part of our objective. They were also used as input for an economic analysis, the second part of our objective.

2.2. Systematic review of the clinical evidence

The SR was performed in line with guidance by the Centre for Reviews and Dissemination, University of York [22]. The detailed SR methodology, as well as results and quality assessment, is reported in the supplementary information for MTG23 [20]. We performed an update to the original search on April 20, 2015, limited to randomised controlled trial (RCT) publications from 2014 onwards. The output from this update was then combined with that from the original SR (Table 1).

2.3. Meta-analysis of the clinical evidence

Expert opinion solicited during development of the MTEP submission suggested that differences are most commonly observed between TURis and M-TURP in the incidence of TUR syndrome, blood transfusions, clot retention, and hospital stay. The evidence was synthesised using pairwise meta-analysis techniques to determine any evidence-based differences between the treatments in these outcomes, as well as in other commonly reported outcomes including urethral strictures, catheterisation time, and procedure duration. According to the Cochrane Systematic Review Handbook, a minimum of 10 studies is required to draw meaningful conclusions from random-effects calculations [5]. Because each meta-analysis undertaken involved <10 studies, fixed-effects analyses were conducted. Details of the meta-analysis methodology were previously reported [20].

2.4. Economic evaluation

The economic analysis (a cost-consequence analysis) compared costs and outcomes associated with TURis and M-TURP, respectively, for surgical treatment of BPH. The perspective of the analysis was that of the National Health Service (NHS) and Personal Social Services in England and Wales.

2.5. Patient population

The analysis considered adult men with LUTS presumed secondary to BPH, in whom surgical intervention was indicated.

2.6. Model structure

A decision-tree model was constructed in which men with BPH-related LUTS underwent surgery with either TURis or M-TURP. Patients entered the model at the point of being indicated for surgery and were allocated probabilities of experiencing procedure-related adverse events (TUR syndrome, blood transfusion, and clot retention) based on the meta-analysis output. Procedure duration was excluded from the analysis because no statistically significant difference was observed between M-TURP and TURis in the meta-analysis. The pooled reduction in catheterisation time estimated from the meta-analysis of 0.2 d for TURis versus M-TURP was statistically significant. However, given that a patient's catheter would generally be removed while in the hospital, this reduction is not expected to result in cost savings beyond those realised through the reduction in hospital stay, which was modelled. As such, catheterisation time was not modelled explicitly. The model considered a nondefined short time horizon covering the prostate resection procedure and immediately afterwards—the period during which the complications modelled are expected to occur. Capital costs assumed equipment life spans of 7 yr. Based on findings of the SR, no difference in functional outcomes (eg, efficacy) was assumed between the procedures with respect to prostate resection weight or radicality.

2.7. Model parameters and data sources

Supplementary Table 1 lists all the variables included in the analysis. Sourcing of inputs, rationale, and further details are provided in Supplement 1.

2.8. Sensitivity analysis

Sensitivity analyses (based on predefined scenarios) were conducted. The first scenario considered all-cause readmission following a TURP procedure, based on data from the single study identified, Fagerstrom et al [23], which indicated a threefold increase in all-cause readmission for M-TURP versus TURis. This study reported readmissions as a result of bleeding, infection, and “other causes.” Because the base-case analysis included clot retention, the scenario considered readmissions from infection and other causes to avoid double counting (it is assumed that “bleeding” includes clot retention). A separate scenario considered a situation in which TURis is performed as day-case surgery.

One-way deterministic sensitivity analysis was performed in which all model parameters were varied either according to published ranges or by ±25%. Results are presented as Tornado diagrams. Probabilistic sensitivity analysis (PSA) was also conducted, in which input parameters were varied according to specified probability distributions. Threshold analysis was conducted to explore the relationship between key model parameters and model results up to the point at which the main conclusions of the model might change, that is, in this case to determine the value at which TURis would be cost neutral (ie, no longer cost saving) versus M-TURP.

3.1. Clinical evidence

The original electronic searches identified a total of 1116 publications. Following title and abstract review, 783 were excluded and 14 relevant RCTs remained, following assessment of the full paper. No unpublished studies were identified from ClinicalTrials.gov. The updated search identified a total 63 further publications, of which two full-text RCT publications remained after review of title and abstract [24] and [25]. On review of the full text, it was found that both of these publications reported data from the same trial, and only the most recent publication, reporting the most up-to-date data, was included in the analysis [24]. Figure 1 shows the study selection flow diagram.

Quality assessment of the identified RCT (Supplementary Table 2) showed that the patient population was consistent with the target population of the analysis. Study end points were clear and appropriately measured according to study design and follow-up schedule.

3.2. Meta-analysis

Six studies were found to be suitable for inclusion in the pairwise meta-analysis of the incidence of TUR syndrome [23], [26], [27], [28], [29], and [30]. The total number of events reported in the studies is presented in Table 2. There were no TUR syndrome events in the TURis arm reported in any of the studies, consistent with the use of saline. The analysis estimated a pooled relative risk (RR) for TUR syndrome of 0.18 (95% confidence interval [CI], 0.05–0.61; p = 0.006) in favour of TURis.