Prostatectomy at high-volume centers improves outcomes and lowers the costs of care for prostate cancer

Abstract

BACKGROUND:

High-volume surgeons with ⩾250 radical prostatectomies provide superior oncological outcomes as evidenced by a lower rate of PSA recurrence (PSAR). The financial benefits of performing prostatectomies at high-volume centers (HVC) are unexplored.

METHODS:

A base case—referent scenario—where the share of prostatectomies at high- and low-volume centers were evenly divided at 50% was defined. Additional scenarios with increasing shares of prostatectomies at HVC with 10% increments were also modeled. Using a lower probability of PSAR as the only advantage of more experienced surgeons, the savings that would result from fewer recurrences, avoidance of salvage radiation therapy (SRT) and management of fewer men with metastatic cancer were calculated.

RESULTS:

The savings associated with performing 80% of radical prostatectomy at HVC were $177, $357 and $559 per prostatectomy at 5, 10 and 20 years, respectively. These savings would offset referral costs of up to $1833 per prostatectomy referral at no additional total societal costs. Given the longer average biochemical failure-free survival with prostatectomies at HVC, referral costs of more than $1833 may be cost effective.

CONCLUSIONS:

Under the conservative assumption of accounting for lower rates of PSAR as the only benefit of surgery in an HVC, performing prostatectomies at an HVC was associated with savings that may offset part of the initial referral costs.

INTRODUCTION

Prior studies have established the role of experience in improving oncological outcomes of radical prostatectomy (RP) in early-stage prostate cancer.^1,2 Estimates show that >80% of surgeons nationwide have an annual caseload of ˂10, but perform ~40% of prostatectomies.³ The impact of experience on outcomes is not limited to prostatectomy and indeed data is emerging in other pathologies such as bladder cancer and even non-cancer surgeries.^4,5 However, despite these outcomes data, the relationship between the improved outcomes and the additional costs of patient referrals to high-volume centers (HVC) remains unclear. This relationship is especially difficult to assess because of the lack of established survival advantage for prostatectomies performed in highly experienced hands.

This study aimed to evaluate, from a societal perspective, the cost implications of preferentially performing RP at HVC in view of improvements in effectiveness of the intervention primarily defined by the oncological outcomes.

Improved oncological outcomes^1,2 associated with performing RP at a HVC can result in overall savings in the costs of prostate cancer care as a result of fewer men requiring salvage radiation therapy (SRT). A lower rate of PSA recurrence (PSAR) may further translate into a lower rate of metastatic failure, assuming similar rates of progression from PSAR to metastatic disease for those that have surgery at HVC. This, in turn, may result in additional savings in costs of care for metastatic disease for patients treated at HVC. The primary objective of this study was defined as estimating the oncological outcomes and the cost savings associated with performing radical prostatectomies at high- vs low-volume centers. The secondary objective was defined as exploring the cost effectiveness of making referrals to a HVC by estimating the maximum referral costs under a willingness to pay of $50 000 per life year gained.

MATERIALS AND METHODS

We used modeling and individual level simulation (microsimulation) to measure and compare the benefits of RP for prostate cancer in the US population. The study took a societal perspective and costs and effects were discounted at the standard rate of 3 percent. The model, assumptions and analysis conformed to the best practices,⁶ and recommendations of the US Panel on Cost Effectiveness in Health and Medicine⁷ and the Agency for Healthcare Research and Quality (AHRQ).⁸

Study design and model

A Markov model was built to represent the natural history of prostate cancer after RP. RP in the model included:

1. Open RP (ORP) at a HVC

2. Minimally invasive RP (MIRP) at a HVC

3. ORP at a low-volume center

4. MIRP at a low-volume center

A HVC was defined as a surgical center where all oncologic radical prostatectomies were performed by surgeons who had individually performed ⩾250 radical prostatectomies. At this level of experience the improvement in oncological outcomes approaches a plateau for open surgery although minimally invasive surgery may need higher numbers.^1,2

In this Markov model four scenarios were devised:

1. Scenario 1—referent scenario—in which the proportion of prostatectomies performed at high- vs low-volume centers were evenly divided at 50% each^3,9

2. Scenario 2 where high- vs low-volume shares were at 60% vs 40%, respectively

3. Scenario 3 where high- vs low-volume shares were at 70% vs 30%, respectively

4. Scenario 4 where high- vs low-volume shares were at 80% vs 20%, respectively

Five states of disease were defined: prostatectomy, where simulation began, no evidence of disease (NED), PSA recurrence (PSAR), metastatic disease (Met) and Dead—see Figures 1 and 2. During the simulation, the differences between scenarios in terms of number of events, time in each state and costs were calculated.

Figure 1.

A Markov diagram depicting states and transitions among different states. Met, metastatic disease; NED, no evidence of disease; PSAR, PSA recurrence.

Figure 2.

A tree representation of the model depicting states and transitions among different states. HVC, high-volume centers; LVC, low-volume centers; Met, metastatic disease; MIRP, minimally invasive radical prostatectomy; NED, no evidence of disease; PSAR, PSA recurrence; SRT, salvage radiation therapy.

Study population

The study population was designed to be representative of the US general population in terms of age and risk of recurrence. The reduction in the rates of PSA recurrence (biochemical failure) was modeled using published data.^1,2 In keeping with standards of care, individuals with PSA recurrence were diagnosed and managed in a manner consistent with the rates of salvage therapy and development of metastatic disease.^1,2,10–13 Individual preferences for compliance with salvage and metastatic treatment were modeled using the published data.^10–13 Mortality rates were modeled based on disease-specific mortality for prostate cancer,^14–16 and US life tables.¹⁷ One hundred thousand male members of the US population between the ages of 50 and 75 years¹⁸ who would undergo RP for cancer were simulated and followed for up to 20 years or until the time of death.

Measurement and analysis of effectiveness, costs and cost effectiveness

Simulated subjects would spend time in one or more of the Markov states of NED, PSAR and Met before the end of simulation or transition to the terminal state of Dead—see Figure 1. The overall survival (OS), therefore, consists of the sum of these segments of survival time.

The primary measure of effectiveness was defined as the number of discounted years spent in NED state, where the quality of life was at maximum compared with PSAR and Met states. The secondary measure of effectiveness was defined as the discounted OS as described above. As a result of lower rates of PSAR in HVC there would be a numeric increase in the time spent in NED state and assuming all following events would occur at similar rates for high- and low-volume center subjects, this incremental survival benefit would also result in an increased OS benefit. Given the advanced age of this population, the competing causes of mortality would result in a significantly smaller realization of the NED benefit in OS measurements. Using these measures, the incremental primary and secondary effects for each scenario were calculated.

The total costs of care, exclusive of the costs of prostatectomy and management of short-term and long-term complications and side effects of surgery were calculated, per prostatectomy, for each scenario. The savings, per prostatectomy, associated with performing prostatectomies at HVC were calculated. The number of additional prostatectomies at HVC for scenarios 2, 3 and 4 were used as the number of referrals and the savings associated with performing prostatectomies at HVC were also calculated per referral.

Assuming that referral costs would exceed the savings and therefore the referral is associated with a net accrual of societal costs, the boundaries of these costs were explored by calculating the maximum referral cost that would result in an Incremental Costs Effectiveness Ratio equal to a willingness to pay of $50 000 per discounted life-years gained.^19,20 Using the discounted incremental effectiveness based on the primary and secondary measures of effectiveness, the lower and upper boundaries for referral costs were calculated.

Study assumptions

This study assumed that the sole benefit of performing RP at a HVC was a lower risk of PSAR. For minimally invasive (robotic and laparoscopic) prostatectomies, the difference in the rates of PSAR were significantly smaller because the learning curve was found to take longer to plateau compared with open RP.² It was further assumed that 80% of prostatectomies would be minimally invasive that has a higher rate of PSAR compared with ORP at a HVC. It was also assumed that the natural history of prostate cancer was the same after PSAR for all subjects, regardless of where the surgery was performed.

Additional improvements in outcomes associated with greater experience, such as lower rates of impotence, incontinence, surgical complications among others are reported in the literature and would give further advantage to RP at HVC. These differences are not well documented, and although modeled, were excluded from cost calculations.²¹

Costs of SRT, treatment of metastatic disease, and costs of last year of life were taken from literature and were based on Medicare reimbursement rates,^22–24 which tended to be lower than private payer rates.²⁵ Medicare reimbursement rates for robotic, laparoscopic and open procedures were equal.²² Costs of prostatectomy at an HVC and a low-volume center were assumed to be equal. The incremental costs of treatment failure as a result of lesser experience were calculated based on the number of PSARs, SRTs performed, and metastatic cases treated in each scenario as projected by the model. Cost savings were calculated per prostatectomy in each scenario and per referral to HVC for Scenarios 2, 3 and 4.

The assumptions are summarized in Table 1.

Table 1.

Summary of assumptions

Assumptions	Value (range)	Varied in sensitivity analysis	Source
Population characteristic
Age	50–75 years		14, 17
Share of ORP versus MIRP at HVC and LVC
ORP	20%	50%	Modeling assumption
MIRP	80%	50%	Modeling assumption
Scenarios HVC.LVC (ORP-HVC:MIRP-HVC:OPR:LVC:MIRP-LVC)
1 (Referent)	50%:50% (10%:40%:10%:40%)	Yes, based on shares of ORP versus MIRP	Modeling assumption
2	60%:40% (12%:48%:8%:32%)	Yes, based on shares of ORP versus MIRP	Modeling assumption
3	70%:30% (14%:56%:6%:24%)	Yes, based on shares of ORP versus MIRP	Modeling assumption
4	80%:20% (16%:64%:4%:16%)	Yes, based on shares of ORP versus MIRP	Modeling assumption
Probability of 5 year PSA recurrence: volume
ORP: < 50	27% (24%, 30%), N = 1402	Yes, lower and upper end of range	1
ORP: 50–99	23% (19%, 27%), N = 696	Yes, lower and upper end of range	1
ORP: 100–249	19% (17%, 22%), N = 1575	Yes, lower and upper end of range	1
ORP: 250–999	16% (14%, 17%), N = 2940	Yes, lower and upper end of range	1
ORP: >1000	8% (6%, 10%), N = 1152	Yes, lower and upper end of range	1
MIRP: < 50	21% (17%, 26%), N = 793	Yes, lower and upper end of range	2
MIRP: 50–99	22% (17%, 29%), N = 611	Yes, lower and upper end of range	2
MIRP: 100–249	20% (16%, 24%), N = 946	Yes, lower and upper end of range	2
MIRP: 250–1100	13% (10%, 16%), N = 2352	Yes, lower and upper end of range	2
Distribution of PSA recurrence by year after surgery
1–2	45%	No	13
3–5	32%	No	13
6–9	19%	No	13
>10	4%	No	13
Probability of metastatic free survival
At 5 years	67% (61%, 72%)	No	10
At 10 years	48% (40%, 56%)	No	10
Probability of death
Natural causes	US Life Table	No	17
Operative mortality	0.5%	No	30
Stage IV cancer, at 5 years	68.6% (68.0%, 69.2%)	No	14
Stage IV cancer, at 10 years, if alive at 5 years	42.7% (41.9%, 43.8%)	No	14
Prostatectomy complications, long term
Incontinence-HVC	18%	No	30–33
Incontinence-LVC	20%	No	30–33
Impotence-HVC	20%	No	30–33
Impotence-LVC	20%	No	30–33
Prostatectomy hospitalization
Transfusions-HVC-ORP	16%	No	27
Transfusions-LVC-ORP	20%	No	27
Transfusions-HVC-MIRP	1.6%	No	27
Transfusions-LVC-MIRP	2.4%	No	27
LOS-HVC-ORP	2.8	No	27
LOS-LVC-ORP	3.2	No	27
LOS-HVC-MIRP	1.8	No	27
LOS-LVC-MIRP	2.0	No	27
Salvage radiation
Probability of SRT after PSA recurrence	37% (32%, 42%)	No	11,12
Relative risk reduction for cancer death after SRT	63%	No	34
Costs
Cost of SRT	$18 000 ($11 300, $25 000)	Yes, lower and upper end of range	24
Cost of management for metastatic disease	$12 765 ($11 543, $13 986)	Yes, lower and upper end of range	23
Cost of management in the last year of life	$37 504 ($35 948, $39 060)	Yes, lower and upper end of range	23

Abbreviations: HVC, high-volume center; LOS, length of stay; LVC, low-volume center; MIRP, minimally invasive radical prostatectomy; OPR, open radical prostatectomy; SRT, salvage radiation therapy.

Sensitivity analysis

The key assumption in this study that differentiates the scenarios was the probability of PSAR depending on the experience of the surgeon. A one-way sensitivity analysis around lower and upper ends of the range of the PSAR probabilities for various settings of prostatectomy was performed. This would impact the gap in PSAR rates between high- and low-volume centers, which in turn would impact the gap in oncological outcomes between the two settings.

One-way sensitivity analyses around the lower and upper ends of range for the costs of SRT, and management costs for metastatic prostate cancer and the last year of life were also performed.

Finally, a one-way sensitivity analysis using the share of MIRP vs ORP in both high- and low-volume settings was performed by decreasing the share of MIRP from a baseline of 80 vs 20 percent to 50 vs 50 percent.

RESULTS

At 5 years 10.4% of subjects experienced PSAR in the referent scenario compared with 9.9, 9.4 and 8.8% in scenarios, 2, 3 and 4. The relative risk reduction at 5 years compared with referent scenario for PSAR was 4.9, 9.6 and 15.0% for scenarios 2, 3 and 4. These numbers for metastatic disease and prostate cancer mortality were 4.4, 9.6, 15.1% and 6.2, 13.5, 19.3%, respectively.

As the share of prostatectomies in HVC increased, the number of PSA recurrences decreased and fewer individuals required SRT. Assuming that 37% of individuals who develop PSAR would receive SRT,^11,12 the discounted savings associated with fewer PSA recurrences were calculated to be $177, $357 and $559 per prostatectomy for scenario 4 at 5, 10 and 20 years, respectively. These figures were $122, $242, $375 and $53, $116, $183 per prostatectomy for scenarios 3 and 2, respectively—see Table 2 and Figure 3a.

Table 2.

Summary of the results. All comparisons were made to the Referent Scenario (Scenario 1). Long term complications refer to complications present beyond 12 months post operatively. Costs of prostatectomy, hospitalization, transfusions, and management of complications were excluded from analysis. Under cost effectiveness analysis the upper and lower boundaries of costs per referral that would meet an incremental cost effectiveness ratio of o$50 000 per life year gained were calculated.

Scenario		Scenario 1 (Referent) : 50% at HVC, 50% at LVC				Scenario 2: 60% at HVC, 40% at LVC				Scenario 3: 70% at HVC, 40% at LVC				Scenario 4: 80% at HVC, 20% at LVC
		Year 5	Year 10	Year 15	Year 20	Year 5	Year 10	Year 15	Year 20	Year 5	Year 10	Year 15	Year 20	Year 5	Year 10	Year 15	Year 20
Number of prostatectomies	ORP-HVC	9838				11 814				13 823				15 888
	MIRP-HVC	39 764				47 938				56 198				64 212
	ORP-LVC	10 150				8236				6036				4056
	MIRP-LVC	40 248				32 012				23952				15 844
Effectiveness and cost outcomes
Hospitalization measures	LOS	2.12				2.10				2.07				2.05
(per prostatectomy)	Transfusions	0.05				0.05				0.05				0.05
Long-term complications	Incontinence	0.07				0.07				0.06				0.06
(per prostatectomy)	Impotence	0.20				0.20				0.20				0.20
	Both	0.01				0.01				0.01				0.01
Cancer-related events	PSAR	10 384	12 732	14 364	15 077	9877	12 238	13 881	14 597	9385	11 770	13 416	14 141	8829	11 245	12 904	13 623
(total events in the population of 100 000)	Met	1423	4218	6156	7181	1360	4030	5907	6909	1286	3846	5669	6663	1208	3644	5421	6386
	SRT	6072	7380	8263	8634	5826	7146	7146	8409	5502	6831	7726	8106	5245	6595	7498	7879
Mortality	Operative	498				498				498				498
(total events in the population of 100 000)	Cancer	275	1864	3556	4589	258	1774	3401	4409	238	1678	3254	4234	222	1585	3107	4059
	All cause	10 537	25 095	42 561	57 266	10 500	24 998	42 429	57 124	10 486	24 919	42 318	57015	10 466	24 826	42 189	56 894
Survival (discounted)	NED	5.20	8.15	10.19	11.29	5.21	8.17	10.23	11.33	5.22	8.20	10.27	11.38	5.23	8.23	10.31	11.43
(per prostatectomy if NED, PSAR, Met or Overall)	PSAR	1.71	3.82	4.85	5.34	1.71	3.80	4.82	5.31	1.71	3.77	4.79	5.28	1.71	3.73	4.74	5.23
	Met	0.85	1.85	2.37	2.58	0.85	1.84	2.36	2.57	0.85	1.83	2.35	2.55	0.86	1.83	2.35	2.55
	OS	5.39	8.71	11.04	12.28	5.39	8.71	11.04	12.29	5.39	8.71	11.05	12.29	5.39	8.72	11.05	12.30
Costs of care	SRT	$1004	$1191	$1300	$1340	$964	$1152	$1262	$1302	$910	$1100	$1210	$1251	$867	$1060	$1171	$1213
(per prostatectomy, excludes costs of prostatectomy)	Met	$155	$995	$1861	$2361	$148	$945	$1780	$2264	$140	$900	$1704	$2172	$133	$853	$1623	$2075
	Year of death	$91	$558	$991	$1223	$85	$531	$948	$1174	$78	$502	$906	$1125	$73	$474	$863	$1077
	All	$1250	$2744	$4152	$4924	$1197	$2628	$3989	$4740	$1128	$2502	$3820	$4548	$1073	$2387	$3658	$4365
Comparative (incremental) effectiveness and costs
Number of HVC referrals						10 150				20 410				30 498
Reduction in hospitalization measures	LOS					233.89				249.62				244.30
(per referral)	Transfusions					7.78				10				10.56
Reduction in long-term complications	Incontinence					8.18				7				6.75
(per referral)	Impotence					—				—				—
	Impotence					1.58				1				0.72
Reduction in cancer-related events	PSAR					49.95	48.67	47.59	47.29	48.95	47.13	46.45	45.86	50.99	48.76	47.87	47.68
(per referral)	Met					6.21	18.52	24.53	26.80	6.71	18.23	23.86	25.38	7.05	18.82	24.10	26.07
	SRT					24.24	23.05	110.05	22.17	27.93	26.90	26.31	25.87	27.12	25.74	25.08	24.76
Reduction in mortality (per 1000 referrals)	Operative					—				—				—
	Cancer					1.67	8.87	15.27	17.73	1.81	9.11	14.80	17.39	1.74	9.15	14.72	17.38
	All cause					3.65	9.56	13.00	13.99	2.50	8.62	11.91	12.30	2.33	8.82	12.20	12.20
Increase in survival (discounted)	NED					94.7	272.2	397.0	465.3	91.4	264.0	383.0	447.5	95.7	273.5	396.1	462.0
(per 1000 referrals)	OS					2.0	22.8	61.9	94.9	1.8	18.9	54.3	83.9	2.2	19.1	54.9	84.4
Reduction in costs of care	SRT					$401	$384	$377	$374	$462	$447	$440	$435	$450	$430	$422	$418
(per referral, excludes costs of prostatectomy)	Met					$65	$496	$800	$948	$74	$464	$771	$926	$74	$466	$$780	$936
	Year of death					$55	$264	$429	$485	$60	$273	$420	$478	$58	$275	$419	$479
	All					$521	$1144	$1606	$1807	$596	$1185	$1631	$1839	$581	$1171	$1621	$1833
Cost effectiveness analysis
Max costs for a WTP of $50 000 per LYG (ICER=WTP)	UB: NED					$5257	$14 755	$21 458	$25 073	$5165	$14 384	$20 779	$24 213	$5 365	$14 844	$21 423	$24 932
(per referral)	LB: OS					$623	$2284	$4703	$6554	$685	$2128	$4347	$6032	$689	$2127	$4364	$6053

Abbreviations: HVC, high volume center; ICER, incremental costs effectiveness ratio= incremental costs/incremental effects; LB, lower boundary; LVC, low volume center; LYG, life year gained; Met, metastatic disease; MIRP, minimally invasive radical prostatectomy; NED, no evidence of disease; ORP, open radical prostatectomy; OS, overall survival; PSAR, PSA recurrence; SRT, salvage radiation therapy; UB, upper boundary; WTP, willingness to pay.

Figure 3.

All comparisons were made to Scenario 1 (Referent Scenario). (a) Decreased costs per prostatectomy as more surgeries occur at high-volume center. As more patients were referred to high-volume centers, savings increased. (b) The savings per referral to a high-volume center. As seen in the figure, the savings per referral remained stable across scenarios and increased with time. Panels (c, d) show the lower and upper boundaries of cost effectiveness for a referral assuming a willingness to pay of $50 000 per life-year gained. These figures were stable per referral and increased with time. CE, cost effectiveness.

The discounted savings per referral to a HVC were calculated to be $581, $1171 and $1833 for scenario 4 at 5, 10 and 20 years, respectively. These figures were $596, $1185, $1839 and $521, $1144, $1807 for scenarios 3 and 2, respectively—see Table 2 and Figure 3b.

Using the observed difference in OS as the measure of effectiveness, based on a willingness to pay of $50 000 per discounted life-years gained, the lower boundary for maximum referral costs at 5, 10 and 20 years were calculated to be $689, $2127 and $6053 for scenario 4. These figures were $685, $2128, $6032 and $623, $2284, $6554 for scenarios 3 and 2, respectively—see Table 2 and Figure 3c.

Similarly, using the incremental time spent in the NED state as the measure of effectiveness, based on a willingness to pay of $50 000 per discounted incremental NED years gained, the upper boundary for the maximum referral costs was calculated to be $5365, $14 844 and $24 932 for scenario 4 at 5, 10 and 20 years, respectively. These figures were $5165, $14 384, $24 213 and $5257, $14 755, $25 073 for scenarios 3 and 2, respectively—see Table 2 and Figure 3d.

Sensitivity analysis

When the lower ends of the range for PSAR probabilities for various prostatectomy settings were used, performing prostatectomies at HVC was still associated with cost savings. As expected, the savings per referral were smaller and the calculations for lower and upper boundaries of cost effectiveness using discounted years of OS and NED revealed a narrower range. Using the upper end of the range for the probability of PSAR recurrence was associated with an increase in the savings per referral and also lower and upper boundaries of cost effectiveness.

Using the lower ends of the range for costs of SRT, treatment of metastatic disease and of prostate cancer care in the year of death at the same time also resulted in smaller values for savings per referral. As expected, using the upper ends of the range for these costs provided higher values for savings per referral and lower and upper boundaries of cost effectiveness. The results of the sensitivity analysis are listed in Table 3.

Table 3.

Summary of the sensitivity analysis varing probability of PSAR, and costs of care, exlcuding prostatectomy costs and complications costs. Scenarios 2, 3 and 4 were compared with Scenario 1 (Referent Scenario)

Scenario		Scenario 2: 60% at HVC, 40% at LVC				Scenario 3: 70% at HVC, 40% at LVC				Scenario 4: 80% at HVC, 20% at LVC
		Year 5	Year 10	Year 15	Year 20	Year 5	Year 10	Year 15	Year 20	Year 5	Year 10	Year 15	Year 20
Probability of PSAR set at upper end of range
Increase in survival (discounted) (per 1000 referrals)	NED	122.8	346.1	502.8	584.3	120.2	338.0	493.2	573.2	117.7	331.2	482.7	561.8
	OS	2.1	19.4	57.0	86.3	1.1	18.7	61.5	95.1	1.3	19.9	61.0	94.2
Reduction in costs of care (per referral, excludes costs of prostatectomy)	SRT	$603	$584	$592	$592	$615	$606	$610	$610	$601	$596	$594	$594
	Met	$95	$569	$983	$1197	$95	$596	$988	$1173	$102	$581	$973	$1153
	Year of death	$59	$309	$513	$590	$54	$325	$528	$601	$64	$335	$536	$612
	All	$757	$1463	$2089	$2379	$765	$1527	$2125	$2384	$767	$1512	$2103	$2359
Max costs for a WTP of$50000 per LYG (ICER = WTP) (per referral)	UB: NED	$6898	$18 769	$27 228	$31 594	$6772	$18427	$26 783	$31 045	$6655	$18 074	$26 239	$30 449
	LB: OS	$863	$2435	$4941	$6693	$821	$2464	$5199	$7140	$831	$2506	$5151	$7068
Probability of PSAR set at lower end of range
Increase in survival (discounted) (per 1000 referrals)	NED	80.6	224.8	319.0	367.5	78.2	223.1	319.5	368.5	78.8	232.3	335.8	390.6
	OS	1.9	17.1	47.6	70.7	0.5	15.3	46.1	71.4	0.4	14.5	44.2	70.5
Reduction in costs of care (per referral, excludes costs of prostatectomy)	SRT	$411	$375	$ $360	$356	$426	$391	$375	$373	$423	$388	$372	$370
	Met	$71	$410	$642	$740	$67	$393	$661	$779	$59	$393	$ $684	$810
	Year of death	$65	$259	$405	$414	$55	$257	$411	$428	$42	$239	$399	$427
	All	$547	$1044	$1406	$1509	$549	$1041	$1447	$1581	$524	$1020	$1455	$1607
Max costs for a WTP of$50000 per LYG (ICER = WTP) (per referral)	UB: NED	$4576	$12 286	$17 357	$19 884	$4461	$12197	$17424	$20 008	$4462	$12 633	$18 246	$21 138
	LB: OS	$642	$1897	$3787	$5043	$573	$1807	$3754	$5150	$543	$1744	$3665	$5132
Costs of SRT and Met set at upper end of range
Increase in survival (discounted) (per 1000 referrals)	NED	94.7	272.2	397.0	465.3	91.4	264.0	383.0	447.5	95.7	273.5	396.1	462.0
	OS	2.0	22.8	61.9	94.9	1.8	18.9	54.3	83.9	2.2	19.1	54.9	84.4
Reduction in costs of care (per referral, excludes costs of prostatectomy)	SRT	$557	$533	$524	$519	$642	$621	$611	$604	$624	$597	$585	$581
	Met	$71	$543	$876	$1039	$81	$509	$845	$1015	$81	$511	$855	$1026
	Year of death	$58	$275	$447	$505	$62	$285	$437	$498	$60	$287	$ $436	$498
	All	$686	$1×352	$1847	$2063	$785	$1415	$1893	$2117	$765	$1394	$1877	$2105
Max costs for a WTP of$50000 per LYG (ICER = WTP) (per referral)	UB: NED	$5422	$ $14 963	$21 699	$25 329	$5355	$14614	$21 041	$24491	$5549	$15 067	$21 679	$25 204
	LB: OS	$788	$2492	$4944	$6810	$874	$2358	$4609	$6310	$873	$2350	$4620	$6325
Costs of SRT and Met set at lower end of range
Increase in survival (discounted) (per 1000 referrals)	NED	94.7	272.2	397.0	465.3	91.4	264.0	383.0	447.5	95.7	273.5	396.1	462.0
	OS	2.0	22.8	61.9	94.9	1.8	18.9	54.3	83.9	2.2	19.1	54.9	84.4
Reduction in costs of care (per referral, excludes costs of prostatectomy)	SRT	$252	$241	$237	$235	$290	$281	$276	$273	$282	$270	$265	$262
	Met	$59	$448	$723	$858	$67	$420	$697	$837	$67	$421	$706	$847
	Year of death	$ $53	$253	$411	$464	$58	$262	$402	$458	$55	$264	$401	$459
	All	$364	$943	$1371	$1557	$414	$963	$1376	$1569	$404	$955	$1372	$1568
Max costs for a WTP of$50000 per LYG (ICER = WTP) (per referral)	UB: NED	$5099	$14554	$21 223	$24 823	$4984	$14162	$20 524	$23 943	$5188	$14 628	$21 174	$24 667
	LB: OS	$465	$2083	$4468	$6304	$503	$1906	$4092	$5762	$512	$1911	$4115	$5788
MIRP vs ORP set at 50–50 percent
Increase in survival (discounted) (per 1000 referrals)	NED	93.9	269.9	401.4	472.1	98.4	281.2	418.7	495.6	96.4	280.0	421.5	502.6
	OS	0.0	12.0	38.3	61.9	0.6	15.4	47.0	76.0	0.9	15.8	51.1	84.2
Reduction in costs of care (per referral, excludes costs of prostatectomy)	SRT	$525	$553	$562	$563	$491	$512	$529	$529	$481	$497	$510	$512
	Met	$89	$485	$851	$1058	$82	$473	$848	$1049	$74	$483	$875	$1066
	Year of death	$36	$234	$415	$494	$41	$221	$408	$487	$39	$235	$433	$516
	All	$649	$1273	$1829	$2115	$614	$1206	$1784	$2065	$595	$1216	$1819	$2094
Max costs for a WTP of$50000 per LYG (ICER = WTP) (per referral)	UB: NED	$5346	$14767	$21 897	$25 722	$5533	$15 266	$22 720	$26 843	$5417	$15217	$22 895	$27 222
	LB: OS	$637	$1873	$3746	$5208	$645	$1976	$4135	$5866	$642	$2008	$4371	$6304

Abbreviations: HVC, high volume center; ICER, incremental costs effectiveness ratio= incremental costs/incremental effects; LB, lower boundary; LVC, low volume center; LYG, life year gained; Met, metastatic disease; MIRP, minimally invasive radical prostatectomy; NED, no evidence of disease; ORP, open radical prostatectomy; OS, overall survival; PSAR, PSA recurrence; SRT, salvage radiation therapy; UB, upper boundary; WTP, willingness to pay.

DISCUSSION

This analysis suggests that performing prostatectomies at a HVC could result in decreased costs by decreasing the number of patients undergoing SRT and decreasing the costs for management of metastatic disease. The savings in the model started in the first 5 years and continued to grow in size over the duration of follow-up. This model assumed that lower probability of PSAR was the only benefit for performing surgery at a HVC. This is in fact a conservative assumption as there may be other benefits to performing prostatectomies at an HVC such as lower rates of incontinence, fewer acute complications and shorter hospital stays. These could not only translate into higher quality of care, but also lower costs and could further tip the balance in favor of HVC.

The incremental increases in the share of prostatectomies at HVC constitute referral to HVC compared with the referent scenario. The savings per referral remained constant across scenarios 2, 3 and 4 and the total savings increased by increasing the number of prostatectomies referred to a HVC. Regardless of how referral cost should be defined or what should be included in it, this analysis demonstrates that within the bounds of these conservative assumptions, each referral is associated with significant societal savings.

If the cost of a referral—the sum of all costs associated with a referral—is less than or equal to the savings per referral, then the referral would be cost-saving or cost-neutral from a societal perspective. In such cases, given the superior oncological outcomes, patients should be referred for prostatectomy to HVC.

Referral costs that exceed these numbers would be subject to cost-effectiveness analysis using a willingness to pay standard. Despite the lack of empiric evidence linking lower probability of PSA recurrence to an OS benefit, in modeling, a lower probability of PSA recurrence would result in a numerical improvement in OS. Using a willingness to pay of $50 000 per discounted incremental years in NED state or discounted life-years gained (OS) the upper and lower boundaries for maximum referral costs were calculated. These numbers indirectly indicate that beyond the savings, a longer stay NED state, where quality of life is maximized and subjects do not require treatment, might justify even higher upfront referral costs.

The results also showed that at any given follow-up interval, the savings per referral remained stable and independent of the number of referrals. Therefore, as the share of radical prostatectomies at HVC increased, the total savings also increased in a linear manner. As a result, reducing the percentage of radical prostatectomies performed at a low-volume center within a healthcare system of any scope, national or regional, would result in proportional savings.

These results assumed that the costs of prostatectomy were the same for high- and low-volume centers. This is a conservative assumption, as many factors including economies of scale would indicate that these costs might be lower at HVC.

The sensitivity analysis revealed that the savings per referral and the range for maximum referral costs shrink significantly if the probability of PSAR were set at the lower end of its range. If the outcomes were to improve in low-volume centers, the need for referral based on better oncological outcomes could be reduced or eliminated. Conversely, at the higher end of the range, the savings and cost-effectiveness boundaries increased in size.

Decreasing the costs of management of treatment failure to the lowest end of the range still indicated that referral was a viable solution with savings per referral showing a 14% decrease from $1833 to $1568 for scenario 4 at 20 years.

It is important to recognize that while at 250 ORPs outcomes reach their maximum, for MIRP this number is perhaps as high as 750.² Whereas MIRP has been rapidly adopted, the high-volume qualifying experience is still lacking and may be considerably more difficult to attain compared with ORP. Our model shows that with a threshold of 250, MIRP in high- vs low-volume centers would have a narrower outcomes gap. As a result, savings per referral grew from $1833 to $2094 at 20 years for Scenario 4, by decreasing the share of MIRP from 80 to 50%. Increasing availability of MIRP combined with a decrease in experience and caseload is part of this trend, which has raised concerns for higher complications and potentially compromising patient care.²⁶ Although this trend is associated with such benefits as shorter hospitalization and decreased rates of perioperative blood transfusions, a decline in oncological outcomes is concerning.^26,27

Centralization of prostatectomy has been a controversial topic in oncologic urology.^28,29 Although the evidence exists that outcomes do improve with higher experience, and in this study we show that these improved outcomes translate to reduced costs of care, questions remain as to how these could be realized in practice. Our view is one of incremental change in where prostatectomies are done, starting with geographies where higher experience is readily available. From a payer’s perspective, referral to a HVC outside of the network would result in savings over time that may offset some of the upfront additional costs. These approaches might help optimize the performance of the healthcare system, both financially and from a quality of care standpoint and would serve as prototypes for other instances where experience could impact outcome to such extents.

Limitations of this study include the fact that it is a model built on data from several different studies conducted during different time periods. These results represent the natural history of prostate cancer as we understand it today. As a result, conclusions drawn here can significantly change by new advances in the field that would narrow the experience gap, reduction in the costs of care or development of new and more effective therapies for prostate cancer. The observed OS difference in this study is the direct result of lower probability of PSAR and has not been empirically documented. The use of life-years gained in a segment of OS such as NED in cost-effectiveness analysis is not standard and is done here only to explore the upper boundary of the referral costs that would remain cost effective.

CONCLUSIONS

This analysis suggests that the costs of referring prostate cancer patients who need RP to a HVC may, in part, be offset by the savings associated with improved oncological outcomes.