Do Cost Limitations of Extended Prophylaxis After Surgery Apply to Ulcerative Colitis Patients?

Abstract

BACKGROUND:

Colorectal surgery patients with ulcerative colitis are at increased risk of postoperative venous thromboembolism. Extended prophylaxis for thromboembolism prevention has been used in colorectal surgery patients although criticized for its lack of cost-effectiveness. However, the cost-effectiveness of extended prophylaxis for postoperative ulcerative colitis patients may be unique.

OBJECTIVE:

To assess the cost-effectiveness of extended prophylaxis in postoperative ulcerative colitis patients.

DESIGN:

A decision analysis compared costs and benefits in postoperative ulcerative colitis patients with and without extended prophylaxis over a lifetime horizon.

SETTING:

Assumptions for decision analysis were identified from available literature for a typical ulcerative colitis patient’s risk of thrombosis, age at surgery, type of thrombosis, prophylaxis risk reduction, bleeding complications, and mortality.

MAIN OUTCOME MEASURES:

Costs ($) and benefits (quality-adjusted life year) reflected a societal perspective and were time-discounted at 3%. Costs and benefits were combined to produce the main outcome measure, the incremental cost-effectiveness ratio ($ per quality-adjusted life year). Multivariable probabilistic sensitivity analysis modeled uncertainty in probabilities, costs, and disutilities.

RESULTS:

Using reference parameters, the individual expected societal total cost of care was $957 without and $1,775 with prophylaxis (not cost-effective; $257,280 per quality-adjusted life year). Preventing a single mortality with prophylaxis would cost $5.0 million (number needed to treat: 6,134 individuals). Adjusting across a range of scenarios upheld these conclusions 77% of the time. With further sensitivity testing, VTE cumulative risk (> 1.5%) and ePpx regimen pricing (< $299) were the two parameters most sensitive to uncertainty.

LIMITATIONS:

Recommendations of decision analysis methodology are limited to group decision-making, not an individual risk profile.

CONCLUSION:

Routine ePpx in postoperative ulcerative colitis patients is not cost-effective. This finding is sensitive to higher than average rates of VTE and low-cost prophylaxis opportunities.

INTRODUCTION

Venous thromboembolism (VTE) after colorectal surgery remains a substantial contributor to postoperative morbidity and mortality. Even with the widespread use of postoperative prophylaxis, incidence of VTE ranges from 2–6% with a case fatality of up to 10%.^1–5 This persistent risk of VTE in the colorectal surgery patient population provides some evidence for the use of extended-duration, post-discharge chemical VTE prophylaxis (ePpx) for select high-risk patients.^6–8 With enhanced risk screening, ePpx regimens, typically 4 weeks of low-molecular weight heparin,^9,10 are frequently recommended for the elderly and patients with colorectal malignancies.^11,12

These risk assessments, however, may be incomplete. Studies have recently reaffirmed the association between inflammatory bowel disease (IBD) and VTE,¹³ and analyses from large population-based datasets have suggested that the risk of VTE following IBD surgery is as great as that from colorectal cancer.^14–16 These studies have also shown that ulcerative colitis (UC) and Crohn’s disease patients bear these risks heterogeneously.¹⁴ In the case of both forms of IBD, cost-effectiveness modeling of ePpx for those at high risk for VTE may inform societal recommendations for its broader use.

We previously examined ePpx cost-effectiveness in patients with Crohn’s disease whose long-life expectancy may have favored more aggressive use of ePpx.¹⁷ Conversely, the anticipated reduced effectiveness of ePpx secondary to the later average age of onset for UC may be offset by the disease being associated with some of the highest rates of VTE reported.¹⁴ The purpose of this study was to use existing evidence to assess the cost-effectiveness of ePpx in postoperative UC patients. We hypothesized that increased risk of VTE would make this group uniquely suitable for ePpx following colorectal surgery.

MATERIALS AND METHODS

Study Design

We adapted our previously published decision-tree model (Figure 1) using TreeAge Pro 2018, version 18.2 to assess the incremental cost-effectiveness and cost per case averted when using ePpx following transabdominal colorectal surgery for UC versus routine care (i.e., standard VTE prophylaxis through hospital discharge and no ePpx). This model replicated from our prior study was ideally suited for cross-comparison purposes as well as the overall similarities in the sequence of care and use of ePpx in both UC and Crohn’s disease.¹⁷ Decision nodes included in the model were whether to use ePpx and the impact of generic versus branded anticoagulants. Chance nodes included were the risk of postoperative bleeding, risk of post-discharge VTE, the proportion of PE (high mortality) versus deep vein thrombosis (DVT, low mortality), and the expected mortality and morbidity of VTE.

Figure 1.

Schematized representation of decision model comparing ulcerative colitis patients undergoing transabdominal colorectal surgery with and without extended venous thromboembolism prophylaxis. Note: The rectangles are decision points, and the ellipses are chance nodes. Generic versus branded ePpx arms were omitted for readability. (Figure previously published in Leeds et al., Dis Colon Rectum, 2019.)

We used the recommendations of the Second Panel on Cost Effectiveness in Health and Medicine for reporting standards (see Supplement).^18,19 We reported cost-effectiveness using the incremental cost-effectiveness ratio (ICER), incremental costs of ePpx divided by the incremental quality-adjusted life years (QALY) gained relative to no ePpx. We also assessed net monetary benefit and societal costs per VTE mortality averted to better understand the scope of the number needed to treat. We compared the ICER to a willingness-to-pay (cost-effectiveness) threshold of $150,000 per QALY, the current consensus for interventions in the United States.^20,21

We adjusted all costs to 2019 U.S. dollars by healthcare sector inflation and applied 3% per year discounting for all future costs and QALYs.^22,23

Perspective and Time Horizon

We used a United States-based societal perspective that incorporated individual plus societal costs and benefits, including total health sector costs and productivity costs. The time horizon of events incorporated all costs related to VTE events occurring within the 30 days following surgery, costs of bleeding from ePpx, long-term costs of post-thrombotic syndrome, and productivity costs through the average life expectancy of a UC patient with and without a VTE event following surgery.

Probabilities

We obtained conditional probability parameter estimates from previously published sources (Table 1).^{4,5,14–16,24–44} Authors I.L. and S.D. reviewed the primary source literature to obtain a broad range of values for sensitivity analysis. Where multiple, divergent high-quality sources were found (e.g., risk reduction estimates for symptomatic versus venographically-screened VTE),^24–29 authors I.L., S.D., and B.S. found consensus on a final parameter estimate. We also broadened sensitivity analyses around more contentious parameters to account for differences in published effect sizes.

Table 1.

Probabilities, costs, and quality-adjusted life years for cost-effectiveness analysis comparing extended venous thromboembolism prophylaxis following colorectal surgery in ulcerative colitis patients versus routine care

Parameter	Estimate	SD	Distribution	Reference
Probabilities
Probability of ePpx compliance	80.0%	7.5%a	Beta	32,33 + Assumption
ePpx-related bleeding risks
Risk of ePpx-related bleeding	1.30%	0.13%b	Beta	24–28
Death from ePpx-related bleeding	0.0%	5.0%a	Beta	24,25
VTE-related risks
Probability of VTE without ePpx	1.1%	2.5%a	Gamma	14–16,24,25,30,31
Risk reduction of VTE with ePpx	0.5	0.1b	Beta	25–29
Probability of DVT versus PE	50.0%	5.0%b	Beta	14,24
Risk of severe PTS	3.0%	0.45%c	Beta	34
Death from PE	7.4%	5.0%a	Beta	4,5,31
Death from DVT	0.0%	5.0%	Beta	Assumption
Costs (US 2019$) d
Routine ePpx costs
Cost of ePpx, branded	$879	$1,129	Gamma	24,25
Cost of ePpx, generic	$635	$912	Gamma	24,25
Postoperative complications
Hospitalization, ePpx-related bleeding	$54,387	$79,664	Gamma	24,25
Hospitalization, postoperative PE	$42,392	$53,835	Gamma	24,25
Hospitalization, postoperative DVT	$30,062	$26,240	Gamma	24,25
Lifetime costs of PTS	$5,863	$1,045	Gamma	35
Productivity (human capital lost)
Productivity costs of early death	$1,369,649	$205,447c	Gamma	37–39
Utilities (Benefits) d
QALYs from decision point
Full life	19.82	2.973	Normal	40–42
Early death	0.077	0.012c	Normal	40–42
Disutility adjustments
ePpx-related bleeding	0.0189	0.0028c	Normal	43
VTE hospitalization	0.0217	0.0033c	Normal	36
Life-long severe PTS	1.0015	0.1502c	Normal	44

^aSD: intentionally wider than reported sources by author consensus to capture all clinically realistic possibilities.

^bSD: +/− 10% of estimate.

^cSD: +/− 15% of estimate. Projected to long-term with assumed increased variability.

^dAll values discounted to present at 3% per year.

⁺Refers to the probability of compliance being both an assumption and used previously in other papers.^32,33

ePpx, extended venous thromboembolism prophylaxis; VTE, venous thromboembolism; PE, pulmonary embolism; DVT, deep vein thrombosis; QALY, quality adjusted life year; PTS, post-thrombotic syndrome

Our preliminary analysis found varying estimates of postoperative VTE in UC patients (Appendix 1). Ultimately, we chose an average 30-day, post-discharge VTE rate of 1.1% patients. Sources with higher rates of VTE included both index admission postoperative VTEs along with post-discharge VTEs or follow-up intervals beyond those where ePpx use would be relevant.^{14–16,24,25,30,31} We assume that all index admission VTEs would have occurred regardless of one’s planned post-discharge ePpx plan. Finally, we used compliance rates similar to those previously reported in prior ePpx trials.^32,33

Finally, we incorporated the risk of post-thrombotic syndrome (PTS) into the model with observational data of first-time DVT presentation and future development of PTS.³⁴ Since the costs and quality of life limitations of low- and moderate-severity disease are minimal, we only considered the effects of severe PTS in our model.⁴⁵

Costs

We used the costs of ePpx and admission for VTE care from a previously published analysis of the largest database of private payer and Medicare supplemental plans including over 245 million unique patients and generalizable to the country’s privately insured population. This analysis of costs included both payer and out-of-pocket expenses associated with ePpx as well as hospitalization costs associated with postoperative VTE in colon surgery and ePpx-related bleeding.²⁴ These cost estimates were validated with an alternative methodologic approach and data source for consistency.²⁵ We excluded costs of inpatient VTE prophylaxis as well as home health needs from both arms as they were presumed to be identical with and without ePpx. We used post-thrombotic syndrome costs based on prior U.S. data that determined the current present value of future cost of long-term complications from VTE.³⁵

In our model, we assumed that all clinically-relevant post-discharge VTE would require an inpatient readmission.³⁶ We acknowledged that not all VTE in practice are treated with an inpatient stay but that such a conservative assumption favored the model preferring the routine ePpx strategy that we were testing.

Finally, we estimated societal costs using a human capital approach for lost productivity based on lost lifetime earnings.^37–39 Although quantifying productivity cost remains controversial, we used the human capital approach as it is currently recommended over friction cost methods.¹⁹

Effectiveness (Benefits)

We assessed benefits in this study using QALYs. Using large population-based studies,^46,47 we estimated the average age of a surgical UC patient to be 45 years old. Modern studies of UC have demonstrated no decline in average life expectancy, so we estimated an additional expected 34 years of life based on U.S. life tables.^40,41,48 Each year of future life was quality-adjusted by prior estimates of disutility living with UC after surgical remission.⁴⁹ By convention, we further quality-adjusted years of life after 65 due to increased age.⁴² We describe the complete calculation of available QALYs for the average UC patient in Appendix 2 and summarize in Table 1. With our prior assumption that all events occur at no more than 30 days following surgery, we afforded deaths from ePpx-related bleeding and postoperative VTE one-twelfth (one month) of one QALY. Changing the exact day of death within the month did not result in different conclusions from the model. We assigned model events such as bleeding from ePpx and postoperative VTE a 30-day disutility for the disability and recovery time from the event.^42,43 In contrast to short-term events, the arms of the model for patients who developed post-thrombotic syndrome had an annual disutility weight applied to each year of remaining life.⁴⁴

Statistical Analysis

We conducted one-way sensitivity analysis comparing the impact of each parameter estimate on model conclusions with a pre-planned formal examination and reassessment of any parameter that changed the direction of model conclusions. We also prospectively intended to identify at what price ePpx becomes cost-effective given the great variability in prescription drug pricing in the United States.⁵⁰

We included all of the reference parameters above and performed a multivariable probabilistic sensitivity analysis (10,000-iteration Monte Carlo simulation) using probability-weighted ranges of these parameters stated in Table 1. We interpreted the output of this analysis using the probabilistic distribution of ICERs and net monetary benefits.

We used beta distributions for conditional probabilities and gamma distributions for costs. We assumed utilities to be normally distributed based on summated utilities greater than zero and adequate sampling of large population based estimates of our source data.⁵¹

RESULTS

Cost-effectiveness of ePpx Following Surgery for Ulcerative Colitis

For the base case, patients undergoing surgery for UC failed to meet conventional measures of cost effectiveness for ePpx (Table 2). The total average expected value for costs of care for ePppx were $1,775 per person versus $957 without ePpx (incremental cost, $818). The average expected QALY benefit with ePpx was 0.003179 QALYs per person. When combining these two elements, the ICER that compares cost to effectiveness was $257,280 per QALY versus a willingness-to-pay threshold of $150,000 per QALY. To prevent a single death, 6,134 individuals would need to be treated, amounting to an expenditure of $5.0 million on ePpx across the surgical UC population (Table 2).

Table 2.

Cost-effectiveness analysis for extended venous thromboembolism prophylaxis following surgery for ulcerative colitis versus routine care

Strategy	EV cost (US$ 2019)	Incremental EV cost (US$ 2019)	EV effectiveness (QALYs)	Incremental EV effectiveness (QALYs)	ICER ($/QALYs)	Net Monetary Benefit (US$ 2019)	$ per mortality averted (US$ 2019)
With extended prophylaxis	1,775	818	19.81474	0.003179	257,280	2,970,436	5,023,711
Routine care	957		19.81156			2,970,777

EV, estimated value; QALYs, quality adjusted life years; ICER, incremental cost-effectiveness ratio

Net Monetary Benefit: incremental cost minus willingness-to-pay multiplied by the incremental QALYs

Sensitivity Analysis for ePpx Following Surgery for Ulcerative Colitis

The Figure 2 Tornado diagram highlights the relative sensitivity of each parameter to variation and the reliability of our results. The model was the most sensitive to the cumulative risk (rate) of VTE for postoperative UC patients, the cost of ePpx, the cost of hospitalization for ePpx-related bleeding events, the effectiveness of ePpx (i.e., risk reduction), and the proportion of life-threatening PE within the VTEs that occur. Two sensitivity ranges ultimately affected the model’s conclusions (i.e., violated the willingness-to-pay threshold): VTE cumulative risk and price of ePpx prescription. If more than 1.5% of the population had a postoperative VTE, excluding all in-hospital VTE, then ePpx would be cost-effective for postoperative UC patients (Figure 3). Similarly, if the price – combined patient and payer contributions – of 30-day ePpx prescriptions were below $299, using ePpx would be the preferred cost-effective option (Figure 4).

Figure 2.

Tornado diagram with incremental cost-effectiveness ratio (ICER) variation by each model parameter. Note: ICER range displayed as the horizontal axis. Willingness-to-pay (WTP) threshold delineated with dashes. Shading represents sensitivity analysis for values greater (red) or less (blue) than parameter estimate. Abbreviations: VTE, venous thromboembolism; PE, pulmonary embolism; DVT, deep venous thrombosis. Parameters that contributed to less than 10% of observed variation in results are not shown.

Figure 3.

One-way sensitivity analysis of venous thromboembolism cumulative risk (event rate) on the cost-effectiveness of extended prophylaxis.

Figure 4.

One-way sensitivity analysis of extended prophylaxis drug pricing on the cost-effectiveness of extended prophylaxis.

Multivariable Sensitivity Analysis

When all parameters were varied in a simulation-based, probability-weighted fashion, multivariable sensitivity analysis demonstrated that the model’s conclusions were maintained for most scenarios. Figure 5 projects how each simulated scenario affects the incremental cost and the incremental effectiveness, which together affect the the ICER. Across 10,000 simulated scenarios, not offering ePpx was the favored cost-effective strategy 77% of the time at a willingness-to-pay threshold of $150,000 per QALY.

Figure 5.

DISCUSSION

The purpose of this study was to assess the societal cost-effectiveness of ePpx following surgery in UC patients versus routine (inpatient-only) VTE prophylaxis. We used a decision analysis model to probability-weight the increase in total costs with ePpx use relative to the increase in quality-adjusted life years. Each additional quality-adjusted life year gained through this intervention cost $257,280 versus the conventional maximum willingness-to-pay threshold of $150,000. By this measure, our model of ePpx costs and effectiveness demonstrated that a strategy of universal ePpx administration for UC patients after colorectal surgery is not cost effective. These findings were upheld throughout a comprehensive multivariable sensitivity analysis in over 75% of scenarios tested.

The most vulnerable assumptions in a minority (23%) of scenarios were the baseline VTE rate within the population and the price for which ePpx could be purchased. If a population of postoperative UC patients had a VTE rate of greater than 1.5%, then our model suggests that a recommendation to universally apply ePpx to this patient group would improve overall health of the population at a cost that is economically justifiable based on consensus standards. Similarly, if a payer was able to obtain enoxaparin ePpx regimens at total patient and payers costs less than $299, ePpx would be cost-effective for all patients of this payer.

Patients undergoing colorectal surgery for inflammatory bowel disease have an increased risk of VTE,^14–16,31 and VTE events following surgery carry a substantial mortality risk.^1–5 Extended prophylaxis after colorectal surgery for select patient groups is a recommended option when trying to mitigate this risk.^6,7 However, this approach has been challenged due to the total costs of these regimens and the decreasing frequency of VTE in surgical populations.^17,24,25,52 The study that we report offers a more balanced view for the future of ePpx in colorectal surgery, particularly with regard to UC patients.

First, this study upholds findings of our group and others’ work that when considering total costs of patients and payers, the expense of committing to a universal ePpx strategy for colorectal surgery patients does not prevent sufficient absolute numbers of VTE relative to the size of the population being treated. However, unlike prior studies, examining ePpx’s cost-effectiveness in the UC postoperative population indicates a smaller margin of error for its conclusions. UC-specific features including a disproportionately greater chance of PE formation rather than DVT¹⁴ and higher chance of mortality with a PE^4,5,31 may increase the potential benefit of ePpx in this patient population.

These differences lead to other parameters in the model being more sensitive to change. Specifically, if the VTE rate in UC patients after surgery was even marginally higher than what the best available population-based data suggests (currently, 1.1%; Appendix 1), our cost-effectiveness conclusions of ePpx may need further revision. Similarly, if combined payer and patient costs of an ePpx regimen fell to price points seen outside the United States, our ePpx cost-effectiveness model would likely favor its use as seen recently in a Canadian analysis for Crohn’s disease with substantially lower drug prices.⁵³ Ultimately, while universal use of ePpx in the UC postoperative population is not indicated under cost-effectiveness criteria, select subsets of hypercoagulable UC patients may derive a cost-effective benefit from ePpx.

A logical extension of the work reported here is to use these findings in conjunction with comprehensive VTE risk models.⁵⁴ For example, Pannucci et al. validated a postoperative VTE risk model that identifies seven weighted-factors (e.g., male gender, advanced age, history of VTE) that in combination can produce VTE risks of greater than the rate of 1.1% identified in this study as a cost-effectiveness threshold.⁵⁵ These models also exist specifically for ulcerative colitis patients,^16,56 and institutional context may define which model or combination of models provides the most accurate predictive risk.

Important aspects of clinical decision-making not included in a cost-effectiveness model are the patient’s preferences and individualized risk threshold. Cost-effectiveness provides a starting point for clinicians, policy makers, and health system administrators to assign societal value of an intervention, but for the individual patient, his or her risk aversion for VTE events is likely influenced by personal experiences and history, out-of-pocket expenses, and ease of administration.⁵⁷

The interpretation of these results is also contingent on the limitations introduced by the methodology. A strength of cost-effectiveness analyses is their ability to pool evidence from many sources, but this also requires a critical evaluation of the literature and often times expert judgment on the quality of such sources. Appendix 1 highlights the degree of variability within the expected VTE rate following surgery for UC patients, one of the most vulnerable parameter estimates to variation in our analysis. We have further attempted to mitigate any uncertainty by intentionally broadening our confidence intervals (Table 1) so that any limitations inherent to the deterministic results (Table 2) would be addressed in the probabilistic modeling (Figure 5).

Two additional considerations may influence the future interpretation of this model. First, ulcerative colitis restorative intestinal pouch procedures have a previously described association with portal vein thromboses.^58,59 The benefit of ePpx may extend beyond the conventional realm of DVTs and PEs; findings of a broader benefit would weight the model further to favoring the use of ePpx. A second limitation of our analysis is the use of low molecular weight heparin regimens as the therapeutic agent for ePpx. The increasingly popular, easy-to-use oral anticoagulants would change a number of parameters of this study and may require further investigation in the future once their risk profile for ePpx is better described.

CONCLUSIONS

Routine ePpx in postoperative ulcerative colitis patients is not cost-effective. This study’s findings are sensitive to higher than average rates of VTE and lower-cost prophylaxis options. More granular risk-stratification of ulcerative patients undergoing surgery may differentiate a hypercoagulable subset more appropriate to benefit from ePpx.