Vaginal hysteropexy compared with vaginal hysterectomy with apical suspension for the treatment of pelvic organ prolapse: A 5‐year cost‐effectiveness Markov model

Our objective was to perform a 5‐year cost‐effectiveness analysis of transvaginal hysteropexy (HP) via sacrospinous ligament fixation (SS) or uterosacral ligament suspension (US) versus vaginal hysterectomy (VH) with apical suspension via sacrospinous ligament fixation (SS) or uterosacral ligament suspension (US) for the treatment of uterine prolapse.


| I N TRODUC TION
Pelvic organ prolapse (POP) is a common disorder with significant impact on quality of life as well as significant cost to the healthcare system. 1 In 1997, the annual cost of surgery for POP in the USA was estimated at over $1 billion and women have nearly a 13% lifetime risk of undergoing surgery for prolapse. 2,3With reoperation rates for recurrent POP as high as 30% as well as an ageing population, this cost has exponentially increased over time. 4In the current healthcare climate, evaluating the cost-benefit of medical and surgical interventions is critically important.
Recently, there has been a growing interest in uterine conservation and hysteropexy surgical strategies for pelvic organ prolapse.Hysteropexy strategies have been shown to decrease operative time and estimated blood loss, as well as to avoid the surgical risks of hysterectomy.Lifetime risk of cervical (0.6%), endometrial (2.7%) and ovarian (1.4%) cancers are low and concomitant hysterectomy has not been shown to improve surgical outcomes after POP surgery. 5n the SAVE U (sacrospinous fixation versus vaginal hysterectomy in treatment of uterine prolapse ≥2) randomised controlled trial (RCT), patients underwent either sacrospinous (SS) hysteropexy (HP) or a vaginal hysterectomy (VH) with uterosacral (US) ligament suspension and were subsequently evaluated for recurrent prolapse of the uterus or vaginal vault stage 2 or higher, bothersome vaginal bulge symptoms or repeat surgery for recurrent apical prolapse.The authors found that HP-SS was noninferior to VH-US for recurrent apical prolapse and repeat surgery at both 1 year and 5 years after surgery. 6,7ur group previously performed a 1-year cost-effectiveness analysis comparing HP with VH with apical suspension for the treatment of uterine prolapse using data from the SAVE U trial.By weighting the different surgical approaches by their risks and costs, we found that hysteropexy strategies were the most cost-effective surgical strategies in the short-term. 8ur objective in this study was to perform a 5-year costeffectiveness analysis of hysteropexy versus vaginal hysterectomy with apical suspension for treatment of uterine prolapse.

| Population and intervention
An IRB-exempt Markov model was constructed on four different surgical treatment strategies for uterine prolapse: vaginal hysterectomy with uterosacral ligament suspension (VH-US), vaginal hysterectomy with sacrospinous ligament fixation (VH-SS), vaginal hysteropexy with sacrospinous ligament fixation (HP-SS) and vaginal hysteropexy with uterosacral ligament suspension (HP-US).We constructed a decision tree using decision analysis software (TREEAGE PRO; TreeAge Software, Inc.) using integrated empirical data from the published literature to model 5-year outcome data for a population of healthy women undergoing surgery for uterine prolapse.The input parameters of the model and assumptions are discussed below and are listed in Table 1.

| Model structure
Markov models assume that a patient is always in one of a finite number of discrete health states, and events represent transitions from one state to another.In this study, a Markov model was used as the risk of prolapse recurrence is continuous over time.Our multistate Markov model is illustrated in Figure 1 and was constructed for each of the surgical strategies after the index surgery: no prolapse recurrence, repeat surgery after prolapse recurrence and no surgery after prolapse recurrence.Each circle represents a health state and arrows represent possible transitions at the end of each yearly time cycle.Patients underwent initial treatment with either VH-US, VH-SS, HP-US or HP-SS.Prior to entering the Markov model, patients either did or did not have a complication.Afterwards, the patients entered the healthy state in the Markov model.Patients could either remain in the healthy state or experience a prolapse recurrence and move into the recurrence state.Patients in the recurrence state could either remain in the recurrence state or proceed with repeat prolapse surgery.Patients who underwent repeat prolapse surgery would then transition back into the Markov model.

| Data sources
Complications, prolapse recurrence and repeat surgery were modelled for the four different treatment options.We assumed that each of the four surgical strategies would have equivalent rates of anti-incontinence procedures, anterior and posterior colporrhaphies and perineorrhaphies.Complications and recurrence rates associated with additional procedures were assumed to be equivalent among all strategies and were not modelled.The model required input parameters for transitions between health states, treatment costs and healthrelated quality of life in each health state.We undertook an extensive search in PubMed to source appropriate parameters for the model and derive probabilities for each of the relevant outcomes.We used search terms to identify articles specific to all treatment arms, including relevant review articles.

| Complications
Complications of each surgical strategy were applied to the sub-groups prior to entering the embedded Markov model.For VH-US, the probability of experiencing a post-surgical complication was 20.2%; for VH-SS, the probability of experiencing a post-surgical complication was 21.5%.Most probabilities of complications following vaginal hysterectomy and apical suspension strategies were derived from the OPTIMAL trial, including rates of transfusion, dyspareunia, neuropathy and injury to the genitourinary (GU) tract. 19For HP-US, Pelvic organ prolapse surgery success 0.882 0.25-1.0[12]   Pelvic organ prolapse surgery failure 0.758 0.2-0.95[12]   Repeat pelvic organ prolapse surgery 0.831 0.2-0.95[12]   Dyspareunia 0.900 0.2-0.95[13]   Neuropathy 0.660 0.2-0.9[14]   Transfusion 0.760 0.4-0.9[15]   Genitourinary tract injury 0.750 0.2-0.9[14]   Model outcome percentages Dyspareunia 56.2% [17,18]   GU injury 12.3% [6,16]   Prolapse recurrence after hysteropexy 16.0% 0-100% [6]   Probability of repeat surgery after hysteropexy 3.0% 0-100% [6]  the probability of experiencing a post-surgical complication was 21.9% (neuropathy, dyspareunia, GU injury); for HP-SS, the probability of experiencing a post-surgical complication was 24.7% (neuropathy and dyspareunia).The probability of dyspareunia following hysteropexy was derived from a 2019 meta-analysis by Meriwether et al, and the probability of transfusion following hysteropexy was assumed to be 0%, reflecting the very low rates in the literature.The probability of transfusion after VH-US and VH-SS was derived from the OPTIMAL trial, where almost 75% of patients had a concurrent hysterectomy.Also using data from the OPTIMAL trial, VH-US and HP-US were assumed to have the same risk of GU tract injury, as both strategies require plication of the uterosacral ligaments, leading to increased risk of ureteral kinking.The risk of neuropathy after VH-SS and HP-SS was assumed to be similar, as was the risk of neuropathy after VH-US and HP-SS.The relative probabilities of each complication and the relative health utility scores are listed in Table 1.

| Prolapse recurrence
Prolapse recurrence was based on the 5-year observational follow-up of the SAVE U (sacrospinous fixation versus vaginal hysterectomy in treatment of uterine prolapse ≥2) randomised controlled trial.The rate of apical prolapse recurrence after HP-SS at 5 years was modelled at 16% and the rate of apical prolapse recurrence after VH-US at 26%.Of note, recurrence after the index surgery was defined as apical prolapse stage ≥2 with bothersome symptoms or repeat surgery for apical prolapse.These results were based on intention-to-treat analysis with conservative imputation.This apical recurrence criterion was chosen because it was the primary outcome of the trial and best reflected the composite outcome of success.Although HP-US and VH-SS were not performed in this trial, we assumed that apical prolapse recurrence after HP-US would be similar to the recurrence  rate after SS and that apical prolapse recurrence after VH-SS would be similar to the recurrence rate after VH-US.
This assumption was based on the similar surgical outcomes between uterosacral and sacrospinous ligament suspension from the OPTIMAL trial.

| Repeat surgery after prolapse recurrence
Repeat surgery rates after prolapse recurrence were also based on the 5-year observational follow-up of the SAVE U randomised controlled trial.The rate of repeat surgery after 5 years was 3.0% following hysteropexy and 7.0% following vaginal hysterectomy with apical suspension.Repeat surgery rates included repeat surgery in any compartment to reflect actual patient outcomes in this trial.

| Health state utility values
Health state utility values were obtained from the literature in a similar fashion for prolapse health states, complications and need for repeat surgery.Based on published estimates, we estimated the relative health utility score of apical pelvic organ prolapse prior to any surgery (0.831), resolution of prolapse symptoms after successful apical surgery (0.882) and recurrence of prolapse symptoms after failed prolapse surgery (0.758).We assumed that those who undergo a second prolapse surgery would receive the same incremental gain of Quality-Adjusted Life Year (QALY) as the primary prolapse repair (+0.05).

| Costs
Cost data for the index surgeries, complications and repeat surgeries were calculated from Stanford University Hospital institutional costs which are billed to insurance providers, as well as other cost-effectiveness literature.Institutional costs were used, as these costs were determined to be more comprehensive than Medicare costs.By using institutional costs, we were able to capture the cost of operating room (OR) time, the cost of anaesthesia, the surgical cost based on Current Procedural Terminology (CPT) code reimbursement and the cost of disposable equipment.Only costs billed to the patient were modelled in this cost-effectiveness analysis using a healthcare sector perspective.Indirect costs such as transportation and productivity losses were not modelled.All costs were evaluated in 2021 US dollars.A standard 3% discounting rate was applied, and a half-cycle correction was used to smooth costs and utilities associated with the model.

| Cost-effectiveness analysis (CEA)
Cost-effectiveness was determined using the incremental cost-effectiveness ratio (ICER), which is the ratio of the additional cost divided by the additional effectiveness of a treatment strategy compared with the next most effective strategy.Where one strategy is more effective and less costly than a comparator, the comparator is dominated.
The willingness-to-pay (WTP) threshold was set a priori at $100,000 per QALY.The WTP is typically set at two to three times per capital annual income, which would imply a US threshold of ~$100,000. 20No ICER was reported for dominated strategies, as they are not cost-effective.

| Probabilistic analysis
Probabilistic analysis was performed using the costeffectiveness acceptability curve (CEAC), which is the plot of the likelihood that an intervention is cost-effective as the value placed in the outcome is varied.The distribution for each variable was selected based on the data type.Probabilities and health utilities were modelled with a beta distribution to ensure the simulation selected a value between 0 and 1. Costs were modelled with a gamma distribution.We calculated the proportion of the 5000 simulations for which each intervention had the highest net monetary benefit (NMB across a range of values for a QALY from $0 to $100,000).The NMB is calculated by multiplying QALYs by the monetary value of a QALY and then subtracting the cost.The strategy with the highest NMB is the most cost-effective at that value of a QALY.The CEAC captures the overall impact of sampling uncertainty; the impact changes as the value of a QALY varies.

| Sensitivity analyses
In addition to the probabilistic analysis, we also report oneway sensitivity analysis which determine whether changes in the model's input parameters altered the overall outcome of the cost-effectiveness model.We conducted Tornado plots and multiple one-way and two-way sensitivity analyses.Probability values were varied across the ranges listed in Tables 1 and 2 to determine whether a threshold existed where the preferred strategy would change.Costs were varied from 20% to 200% of the base case value.

| R E SU LTS
In the base case scenario, both HP-SS and HP-US are costeffective at a WTP threshold of <$100,000 per QALY.The ICER for HP-US compared with HP-SS was $90,738.14;VH-US and VH-SS are both dominated strategies.Although both hysteropexy approaches are on the efficient frontier, HP-US is the optimal strategy with the highest effectiveness (Table 2).Tornado plots estimated that the variables that most influenced the cost-effectiveness results are the costs of the four different surgical strategies, the probability of complications following hysteropexy surgery and the probability of recurrence after hysteropexy (Figure 2).
When we varied the costs of the different strategies, vaginal hysterectomy strategies become the most costeffective option if the cost of sacrospinous hysteropexy is >$52,960 and if the cost of uterosacral hysteropexy is >$52,895.They are also cost-effective if the probability of complications after sacrospinous hysteropexy is >73% and the probability of complications after uterosacral hysteropexy is >63%.
T A B L E 2 Base case 5-year costs, effectiveness and incremental cost-effectiveness ratios (ICERs) for uterovaginal prolapse for surgical strategies ranked by cost.F I G U R E 2 Tornado plots of net monetary benefits (NMB) (WTP: $100,000).NMB, net monetary benefits; WTP, willingness to pay.Tornado plots estimated that the variables that most influenced the cost-effectiveness results are the costs of the four different surgical strategies, the probability of complications following hysteropexy surgery and the probability of recurrence after hysteropexy.

Strategy
In a two-way sensitivity analysis varying recurrent vaginal prolapse after hysteropexy and repeat surgery following hysteropexy failure, vaginal hysterectomy strategies become cost-effective along the curve shown in Figure 3. HP-US as a cost-effective strategy decays exponentially with increasing probability of prolapse recurrence and the need for repeat surgery after failed hysteropexy.For example, vaginal hysterectomy strategies are more cost effective if prolapse recurrence rates after hysteropexy are 16%, and 15% of these patients have repeat surgery.
The CEAC for the base case analysis predicts considerable certainty in favour of the sacrospinous hysteropexy strategy until reaching a willingness-to-pay threshold between $90,000 and $100,000.At this WTP threshold, the uterosacral hysteropexy strategy then becomes more favourable over most simulations.

| Main findings
Our Markov model estimated that vaginal hysteropexy strategies are a cost-effective surgical approach for the treatment of uterine prolapse.When using the available prolapse recurrence rates and repeat surgery rates in the literature, both vaginal hysterectomy strategies are dominated strategies with higher costs and lower QALYs.
F I G U R E 3 Two-way sensitivity analysis varying recurrent vaginal prolapse after hysteropexy and repeat surgery following hysteropexy failure (WTP = $100,000).In a two-way sensitivity analysis varying recurrent vaginal prolapse after hysteropexy and repeat surgery following hysteropexy failure, vaginal hysterectomy strategies become cost-effective along this curve.In the base case scenario, uterosacral hysteropexy is the most costeffective strategy when prolapse recurrence rates after hysteropexy are 16.0%, and 3.0% of these patients have repeat surgery.

| Interpretations
In this current study, both HP strategies are on the efficient CEA frontier and HP-US is the optimal surgical strategy with the highest QALYs.VH strategies only become the more cost-effective strategies if HP strategies reach higher prolapse recurrence and complication thresholds.
In our model, VH strategies become cost-effective if prolapse recurrence rates of HP strategies surpass 8%, with all patients undergoing repeat surgery.VH strategies are also more cost-effective if the prolapse recurrence rates of HP strategies surpass 23%, even if none of these patients has repeat prolapse surgery.Given the wide variability of prolapse recurrence, repeat surgery rates and definitions of 'surgical success' in the literature, more comparative studies are needed between uterine-sparing prolapse repair and vaginal hysterectomy with apical repair before strong conclusions regarding the optimal cost-effectiveness procedure can be made.
The QALY estimates for the different VH/HP surgical strategies are similar and are separated by only 0.01 QALY, which is less than the minimally important difference for utilities.This minimal difference suggests that the overall effectiveness of each surgical surgery is comparable.
Additionally, the influence of hysteropexy recurrence rates on the model must be acknowledged.2][23][24] Treatment outcomes in this subgroup have not been well described and are poorly understood.A lack of robust RCTs comparing hysteropexy with hysterectomy makes it challenging to interpret the data on the equivalency of complication rates and prolapse recurrence rates.When considering surgical options based on costeffectiveness, it is important to note that vaginal hysterectomy strategies become cost-effective at recurrence rates of >23% and can become cost-effective at lower recurrence rates if part of the cohort undergoes recurrent prolapse surgery.
In our cost-effectiveness analysis, uterosacral hysteropexy was the optimal surgical strategy for uterovaginal prolapse.Using Stanford institutional cost data, sacrospinous hysteropexy was slightly more expensive, which may be related to the cost of suture capture devices used during sacrospinous fixation.Additionally, sacrospinous fixation surgeries have a higher rate of postoperative dyspareunia, neuropathy and pain.Although these complications are infrequent and often resolve with time, the higher costs and lower QALYs in the initial postoperative period should be considered when analysing cost-effective surgical options for uterovaginal prolapse.
However, as the WTP is lowered, the optimal surgical strategy switches from uterosacral hysteropexy to sacrospinous hysteropexy.Though infrequent, patients undergoing uterosacral ligament suspension procedures are at risk for genitourinary injury, which often requires reoperation and increases the average cost of uterosacral hysteropexies.If the WTP is lowered, uterosacral hysteropexy no longer becomes cost-effective because of the low but expensive risk of a genitourinary injury.Additionally, most studies on transvaginal native tissue hysteropexy use the sacrospinous ligament as the point of fixation to the apex.More surgeons feel comfortable performing an extraperitoneal dissection to the sacrospinous ligament rather than an intraperitoneal entry to the uterosacral ligaments.After taking these considerations into account, it may be reasonable to perform a sacrospinous hysteropexy as the optimal cost-effective strategy at lower values of the WTP.
In a healthcare era that is strongly influenced by expenditures and reimbursements, choosing a surgical option that has both good outcomes as well as low cost is becoming increasingly important.Cost-effective analyses are needed to provide surgeons with the confidence that the procedures they recommend not only produce positive change in healthcare delivery but also maintain quality care.We feel our study strongly supports the consideration of uterine-sparing surgery as a standard option for any woman with uterine prolapse who is an appropriate candidate.
Our Markov model used data from one RCT which showed comparable success rates for uterine-sparing apical repair and vaginal hysterectomy and apical repair for the treatment of uterovaginal prolapse.Robust evidence on hysteropexy complications, outcomes and recurrence rates as well as additional RCTs to compare surgical approaches is needed.As more data become available, our Markov model can be updated and improved to ensure that we are homing in on the most cost-effective surgical strategies.

| Strengths and limitations
The major strength of this study was our use of data and probabilities from two large randomised controlled trials as well as well-validated utility values in the literature.We used sensitivity analyses to vary parameters as well as probabilistic modelling to determine uncertainty in our model.
Although the SAVE U and OPTIMAL trials were well designed and highly cited in the literature, randomised controlled trials may not mimic real life, limiting generalisability, and patient dropout rates may introduce bias into the results.Additionally, in the SAVE U trial, permanent sutures were used for the sacrospinous hysteropexy procedures and delayed absorbable sutures for the uterosacral ligament suspension after vaginal hysterectomy.Although there is no consensus in the literature on optimal suture type, recurrence rates and repeat surgery rates may be influenced by different surgical techniques.
As with any model, we are limited by the availability of data and the accuracy of our assumptions.Although we used data from randomised controlled trials, there are multiple definitions of surgical success and prolapse recurrence.In the literature, prolapse recurrence rates and repeat surgery rates are variable and there are only a few comparative trials of vaginal prolapse surgery.A further limitation is the relative paucity of long-term outcomes data for hysteropexy procedures, which could affect our model inputs.
Additionally, our model focused on the cost and effectiveness of apical suspension strategies considering subsequent apical prolapse recurrence.However, emerging data may show that the rates of recurrent anterior and posterior compartment prolapse differ between the two strategies, which could be a consideration in future studies.Given the variability of recurrence rates with regard to native tissue repair for prolapse, we had to make assumptions regarding surgical failure rates and rely on sensitivity analyses for better estimation.
Using institutional costs for the procedures in the model meant that the absolute monetary values are not generalisable to all practices.However, use of institutional costs is more comprehensive than a model based on Medicare costs, as they include costs such as reimbursements for anaesthesiologists, surgeons, operating room time and cost, and disposable equipment.
This study must also be taken in the context of the clinical scenario: there are situations where uterine preservation is contraindicated, and not all providers may have the surgical expertise needed to perform a hysteropexy compared with a transvaginal hysterectomy, or vice versa.
Finally, we constructed a 5-year Markov model that would not account for possible long-term costs and health effects.As this study estimated cost-effectiveness for medium length follow-up after surgery, we did not model patients who might have subsequent workup for future postmenopausal bleeding or treatment of endometrial pathology after uterine-sparing surgery, as these complications tend to occur over longer time horizons.

| CONCLUSION
Our study suggests that hysteropexy surgical strategies are cost-effective surgical strategies for the transvaginal surgical management of apical prolapse.Uterosacral hysteropexy is the optimal surgical strategy at WTP >$100,000, and at lower WTP thresholds, sacrospinous hysteropexy becomes the optimal surgical strategy.Vaginal hysterectomy with apical suspension becomes more cost-effective with increasing probability of prolapse recurrence and need for repeat surgery after failed hysteropexy.Given the variability of prolapse recurrence rates in the literature, more comparative studies are needed to understand the cost-effectiveness relation between these different surgical approaches.

AU T HOR C ON T R I BU T ION S
SLW: conceptualisation, investigation, methodology, writing original draft, draft review and editing.RS: conceptualisation, investigation, methodology, draft review and editing.KL: investigation, methodology, draft review and editing.ERS: conceptualisation, investigation, methodology, writing original draft, draft review and editing.

AC K NO W L E D GE M E N T S
ERS reports grant funding to Stanford University from the NIH, Foundation for Female Health Awareness, Cook Myo-Site, Coloplast, ACell and Cynosure.ERS owns stock in Pelvalon and received travel reimbursement from Contura.

F U N DI NG I N FOR M AT ION
There was no funding for this project.

DATA AVA I L A BI L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.

E T H IC S S TAT E M E N T
This study was exempt from Institutional Review Board Approval.

R E F E R E NC E S
Note: Model probabilities, health utility values and costs of some complications were obtained from the published literature.Cost data for the index surgeries, complications and repeat surgeries were calculated from Stanford University Hospital institutional costs that are billed to insurance providers.*Indicates conditional probabilities.T A B L E 1 (Continued) F I G U R E 1 Markov model structure.In our multistate Markov model, patients underwent initial treatment with VH-US, VH-SS, HP-US or HP-SS.Prior to entering the Markov model, patients either did or did not have a complication.Afterwards, the patients entered the healthy state in the Markov model.Patients could either remain in the healthy state or experience a prolapse recurrence and move into the recurrence state.Patients in the recurrence state could either remain in the recurrence state or proceed with repeat prolapse surgery.Patients who underwent repeat prolapse surgery would then transition back into the Markov model.