Cost–consequence analysis of a hemostatic matrix alone or in combination for spine surgery patients

Abstract

Background: A five-year retrospective database analysis comparing the use of Floseal¹ flowable topical hemostat alone (F) and in combination with gelatin/thrombin (F + G/T) to achieve hemostasis and control surgical bleeding showed higher resource utilization for F + G/T cases relative to F matched pairs during spinal surgery. Lower resource use in the F group was characterized by shorter hospital length of stay and surgical time as well as fewer blood transfusions and less hemostat agent used per surgery.

Objective: To evaluate the cost–consequence of using F compared to F + G/T in minor, major and severe spinal surgery from the US hospital perspective.

Methods: A cost–consequence model was developed using the US hospital perspective. Model inputs include clinical inputs from the literature, cost inputs (hemostatic matrices, blood product transfusion, hospital stay and operating room time) from the literature, and an analysis of annual spine surgery volume (minor, major and severe) using the 2012 National Inpatient Sample (NIS) database. Costs are reported in 2017 US dollars. One-way and probabilistic sensitivity analyses address sources of variability in the results.

Results: A medium-volume hospital (130 spine surgeries per year) using F versus F + G/T for spine surgeries is expected to require 85 less hours of surgical time, 58 fewer hospital days and 7 fewer blood transfusions in addition to hemostat volume savings (F: 1 mL, thrombin: 1994 mL). The cost savings associated with the hospital resources for a medium-volume hospital are expected to be $317,959 (surgical hours = $154,746, hospital days = $125,237, blood transfusions = $19,023, hemostatic agents = $18,953) or $2445 per spine surgery.

Conclusions: The use of F versus F + G/T could lead to annual cost savings for US hospitals performing a low to high volume of spinal surgeries per year.

Keywords:

• Gelatin
• thrombin
• flowable hemostatic matrix
• spine surgery
• transfusions
• cost

JEL classification codes::

• P59
• C52

Introduction

Inadequate hemostasis during spine surgery can lead to considerable intra-operative bleeding which is associated with complications such as postoperative hematoma and increased resource utilization such as perioperative blood transfusions and hospital length of stay (LOS)¹. To control bleeding during spine surgery, non-flowable topical hemostats and flowable hemostats can be used alone or in combination^2–4.

Floseal² hemostatic matrix is a commonly used gelatin-based hemostatic matrix in US practice⁵. It comprises a self-expandable gelatin matrix component with smooth distinct round particles and lyophilized human thrombin². After application to the bleeding site or upon contact with blood, its gelatin granules swell up to 20% thereby restricting blood flow, providing tamponade and potentially reducing bleeding⁶. This combines passive and active hemostats into a single application flowable hemostatic matrix⁷. In some cases it is used alone to minimize intraoperative bleeding for a broad range of surgeries, including spinal surgery⁴.

Another medical device commonly used to control surgical bleeding is a gelatin sponge, which is often saturated with thrombin prior to use as an adjunct to hemostasis⁸. A gelatin sponge is a non-flowable medical device prepared from purified gelatin granules, but the mechanism of action, believed to induce hemostasis through physical means rather than alteration of the blood clotting mechanism, is not fully understood⁸. Randomized clinical trials have demonstrated increased clinical efficacy in achieving hemostasis and reducing patient blood loss for the hemostatic flowable matrix relative to conventional methods in spinal and other surgical specialties^9–11.

The effectiveness of using a hemostatic flowable matrix only and hemostatic flowable matrix plus gelatin/thrombin flowable hemostat in combination to achieve hemostasis in spinal surgery was only recently assessed in a retrospective analysis of a large hospital database⁴. The study by Ramirez et al.⁴ adds to the literature on the combination use of flowable and non-flowable hemostat by conducting an observational retrospective database analysis with Premier’s US Hospital Database of hospitalizations and discharges that occurred between 1 October 2010 and 30 September 2015. The authors studied 15,105 propensity-matched pairs (1:1) of spine-surgery cases who underwent severe, major or minor spine surgeries. The analysis included cases where Floseal only (F) or Floseal plus gelatin/thrombin (F + G/T) was used, patients had complete demographic/baseline values and evaluable outcome measures existed. Ramirez et al.⁴ showed that F cases compared to F + G/T cases had significantly decreased hospital length of stay (LOS) (−0.45 days, p < .0001), surgery time (−39 minutes, p < .0001) and use of hemostat agent (−12.5 mL, p < .0001). In addition, F cases required significantly fewer interoperative, perioperative, postoperative and pure blood transfusions (p < .0001)⁴.

Effective use of hemostatic agents has been shown to be cost saving in cardiac surgery^12,13 and major and severe spine surgery¹⁴. From the currently available literature, it is not clear whether clinical differences between use of F versus F + G/T amount to significant cost differences for US hospitals performing minor, major and severe spine surgeries. This cost–consequence analysis contributes to the health economic literature by determining whether the differences in clinical efficacy found in the retrospective database analysis conducted by Ramirez et al.⁴ contribute to cost differences from the US hospital perspective.

Methods

Overview

We developed a cost–consequence model in Microsoft Excel (Redmond, WA, USA) to estimate the annualized economic impact of F vs. F + G/T to achieve hemostasis during spinal surgery in a US hospital. The model is heavily based on the hemostat volumes and clinical effectiveness including hospital LOS, surgery time, blood loss complications and incidence of blood transfusions reported by Ramirez et al. in their five-year (2010–2015) retrospective analysis of the Premier Perspective hospital database (Charlotte, NC, USA)⁴. The hemostat volumes and clinical effectiveness data from Ramirez et al.⁴ is supplemented with estimates of surgical volume calculated for the cost–consequence model using the 2012 Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS)¹⁵ and costs from the literature and other sources as detailed in the following sections.

Population

The spine surgery population in the cost–consequence model includes minor, major and severe spine surgery patients to match the ICD-9 procedure codes included in the analysis by Ramirez et al.⁴ (Table 1). The selection criteria are described in detail by Ramirez et al.⁴. In brief, adult patients (ages 18+) that underwent elective, emergent or urgent spine surgeries with hospitalizations and discharges between 1 October 2010 and 30 September 2015, and had a charge of F or F + G/T on the day of surgery were included in the analysis⁴.

Table 1. Primary or secondary ICD-9 procedural codes for spinal surgeries (from Ramirez et al.⁴).

Minor spine surgery group

81.02, 81.03: Initial cervical fusion

81.32, 81.33: Repeat cervical fusion

81.04, 81.05, 81.06, 81.07, 81.08: Initial lumbar fusion

81.34, 81.35, 81.36, 81.37, 81.38: Repeat lumbar fusion

81.00: Initial fusion

81.30: Repeat fusion

Major spine surgery group

81.62: Fusion/refusion vertebrae 2–3

03.4: Tumor resection

80.89: Tumor resection

Severe spine surgery group

81.63: Fusion/refusion vertebrae 4–8

81.64: Fusion/refusion vertebrae 9+

03.4 or 80.89 Tumor resection +80.99: Corpectomy or 81.62–81.64 Fusion/refusion vertebrae 2–3, 4–8, 9+

03.53: Fracture stabilization +81.62–81.64 Fusion/refusion vertebrae 2–3, 4–8, 9+

03.53: Fracture stabilization +80.99: Corpectomy or 80.89: Osteotomy +81.62–81.64 Fusion/refusion vertebrae 2–3, 4–8, 9+

The volume of spine surgeries (minor, major and severe) for low- (25th percentile), medium- (50th percentile) and high-volume (75th percentile) US hospitals was estimated using the HCUP NIS 2012 data¹⁵. This manuscript reports results for a medium-volume US hospital performing 130 spine surgeries annually. Results for low- and high-volume hospitals performing 50 and 300 spine surgeries annually, respectively, are discussed as a secondary analysis. Additional details on the estimation of the volume of spine surgeries can be found in Supplementary Appendix Tables 1A and 1B.

Clinical inputs

The model accounts for the statistically significant differences in hemostat volume and clinical effectiveness reported by Ramirez et al. between patients treated with F and F + G/T in the Premier Perspective hospital database with the exception of blood loss complications (hemorrhage, anemia, thrombocytopenia, hematoma complicating a procedure, visual complication such as dural puncture during the operation, embolic events or venous thrombosis)⁴. The clinical inputs including average volume of hemostat used in surgery, mean difference in average length of surgery time, mean difference in hospital LOS and the percentage of cases requiring blood transfusions can be found in Table 2. It was assumed that one unit of blood product was required for each transfusion. The percentage of cases with blood loss complications, although statistically significantly different in the Premier hospital database analysis (F: 0.5%, F + G/T: 0.8%, p = .0022)⁴, are conservatively only considered in the discussion because of the wide range in potential costs.

Table 2. Clinical inputs for all spinal surgeries (severe, major, minor) in the cost–consequence model.

	Mean difference	Data source
	($\overline{FO-F+G/T}$)
Average length of surgery (in minutes)	39.028	Ramirez et al.^4
Average hospital length of stay (in days)	0.446	Ramirez et al.^4
	Difference in means	Data source
	($\overline{FO}-\overline{F+G/T}$)
Average volume of Floseal used in surgery (mL)	0.008	Ramirez et al.^4
Average volume of thrombin used in surgery (mL)	15.340	Ramirez et al.^4
Surgical cases requiring any blood product transfusions (%)^a	5.67	Ramirez et al.^4

Abbreviations. FO, Floseal only; F + G/T, Floseal + gelatin/thrombin; pRBC, packed red blood cell.

^a Blood product transfusions included intraoperative transfusions, perioperative transfusions, postoperative transfusions and pure-blood/pRBC transfusions.

Cost inputs

The costs and sources can be found in Table 3. The cost–consequence of F vs. F + G/T for a single surgery is calculated by multiplying the difference in hemostat volume and clinical effectiveness by the associated product and surgery-related costs. The per mL cost of F is the Baxter wholesale acquisition cost (WAC) per ml of 5 mL and 20 mL products¹⁶. The per mL cost of thrombin is the average hospital price per mL of 5 mL and 20 mL thrombin-JMI products¹⁷.

Table 3. Cost inputs for spinal surgery in the cost–consequence model.

Description	Cost	Data source
Unit cost (per mL) of hemostatic agent
Floseal	$27.14	Floseal WAC^16
Thrombin	$9.49	Thrombin ASP  from IMS NSP^17
Cost of blood product transfusion per unit	$2579	Tackett et al.^12
Cost of per day hospital stay	$2160	Hospital room &  board charge^19
Cost of operating room (OR) time (per hour)	$1830	Chatterjee et al.^18

Operating room, hospital stay and blood transfusion costs are from public sources and the literature. The cost of operating room time is estimated based on a national average cost per minute for operating room time across five surgical procedures from five surgical specialties in a university hospital setting¹⁸. The cost of hospital stay is approximated by the room and board per day charge for a US hospital¹⁹. Cost of blood product transfusion is the incremental hospital cost of a single-unit allogenic blood transfusion inclusive of adverse event and increased resource consumption costs associated with use of allogenic blood products¹². All costs were converted to 2017 US dollars using the Medical Care component of the Consumer Price Index (CPI) published by the US Bureau of Labor Statistics.

Sensitivity analyses

Sensitivity analyses were performed to identify the major cost drivers accounting for variability around estimates. A one-way sensitivity analysis varies the clinical and cost inputs one at a time using the lower and upper boundaries of the 95% confidence interval for clinical inputs, when available from Ramirez et al.⁴, or by ±20%. A probabilistic sensitivity analysis uses a Monte Carlo selection of all clinical and cost inputs at once from the 95% confidence interval or ±20% of each parameter according to previously defined distribution probability (repetitions = 10,000). Base values and distributions for the clinical and cost inputs for the probabilistic sensitivity analysis can be found in Supplementary Appendix Tables 2A and 3A, respectively.

Results

Cost–consequence analysis of F vs. F + G/T

By using F vs. F + G/T during spine surgery, a medium-volume hospital performing 130 spine surgeries (minor, major or severe) per year could save approximately 85 surgical hours, 58 hospital days, and 7 blood product transfusions. Along with the reduction in use of hemostatic agents (F: 1 mL, thrombin: 1994 mL), the reduction in resource utilization could lead to average annual savings of $317,959 (surgical hours = $154,746, hospital days = $125,237, blood transfusions = $19,023, hemostatic agents = $18,953) (Table 4). This is a net cost saving of $2445 per spine surgery.

Table 4. Total cost savings per hospital – all spine surgery (minor, major and severe).

Parameter	Clinical savings	Cost savings
Medium-volume hospital (130 spine surgeries annually)
Operating room time	84.56 hours	$154,746
Number of transfusions required	7.38 transfusions	$19,023
Hospital days	57.98 days	$125,237
Hemostatic agents	1995.24 mL	$18,953
Total cost savings medium-volume hospital		$317,959

Sensitivity analyses

The one-way sensitivity analysis shows that the number of spine surgeries performed annually has the biggest impact on total cost savings as expected (Figure 1). For low-volume hospitals performing 50 spine-related surgeries per year, the use of F vs. F + G/T could result in cost savings of $122,292 per hospital annually (not shown: surgical hours = $59,518, hospital days = $48,168, blood transfusions = $7317, hemostatic agents = $7290). For high-volume hospitals performing 300 spine-related surgeries per year, the use of F vs. F + G/T could result in cost savings of $733,752 annually (not shown: surgical hours = $357,106, hospital days = $289,008, blood transfusions = $43,900, hemostatic agents = $43,738).

Figure 1. One-way sensitivity analysis of total cost savings per hospital – all spine surgery (minor, major and severe).

The next largest sources of variability include the ±20% variation in the cost of operating room time (high = $348,905 annual savings, low = $287,010 annual savings, difference = $61,898), 95% CI of mean difference in hospital length of stay (high = $346,039, low = $290,160, difference = $55,879), ±20% variation in the cost of hospital length of stay (high = $343,007 annual savings, low = $292,912 annual savings, difference = $50,095), and 95% CI of mean difference in surgery time (high = $330,695, low = $305,220, difference = $25,475).

Probabilistic sensitivity analysis was conducted through a Monte Carlo simulation of all input variables simultaneously. Of 10,000 iterations, the expected cost savings for F vs. F + G/T was higher than baseline simulation in 68% of iterations (Figure 2). F vs. F + G/T was cost saving in all iterations (i.e. no simulation results with negative cost savings).

Figure 2. Probabilistic sensitivity analysis of total cost savings per hospital – all spine surgery (minor, major and severe).

Discussion

This analysis demonstrates that the avoidable costs related to hospital resource use (surgery time, hospital length of stay, blood transfusion and volume of hemostatic agent) by F vs. F + G/T is $317,959 for medium-volume hospitals (130 spine surgeries annually). However, the estimated cost savings vary widely by the expected number of spine surgeries. A high-volume hospital performing 300 spine surgeries annually would be expected to save almost three-quarters of a million dollars ($733,752) annually, and a low-volume hospital performing 50 spine surgeries annually would be expected to save $122,292 annually. Sensitivity analyses suggest that the cost savings of using F vs. F + G/T for spine surgeries is positive across a range of clinical outcomes and costs drivers.

Bleeding/bleeding-related consequences will continue to be common during spine surgery and in some cases lead to further blood loss complications such as hemorrhage, anemia, thrombocytopenia, hematoma complicating a procedure, visual complication such as dural puncture during the operation, embolic events or venous thrombosis, which are risky to patients due to risks inherent to blood transfusion such as allergic reactions, fever, acute immune hemolytic reaction and blood-borne infection (i.e. HIV, hepatitis B or C) and can be expensive to treat²⁰. This cost–consequence model adds to the clinical support recently published by Ramirez et al.⁴ for using F to achieve hemostasis during spinal surgery compared to using F + G/T.

The overall cost savings are in line with other research studies comparing the economic consequences of using hemostatic agents (Table 5). A study compared the use of F to a different flowable hemostatic matrix (Surgiflo³) in combination with thrombin in spine surgery and demonstrated a cost saving of $61,251 per year ($151 per major and $547 per severe spine surgery) for an average US hospital performing a medium number of spine surgeries (182 major and 59 severe spine surgeries)¹². These savings were driven by 3 fewer blood product transfusions and 27 fewer hours of operation room time.

Table 5. Summary of selected economic evaluations between hemostatic agents.

Study	Comparators	Surgery Type	Net Cost-Saving
Ramirez et al., 2018 (current study)	Floseal vs. Floseal + gelatin/thrombin	Spine surgery	$2445 per surgery
Makhija et al., 2017^12	Floseal vs. Surgiflo + thrombin	Spine surgery	$151 per major surgery, $547 per severe surgery
Tackett et al., 2014^13	Floseal vs. Surgicel/Gelfoam	Cardiac surgery	$8960 per case
Makhija et al., 2017^14	Floseal vs. Surgiflo	Cardiac surgery	$6256 per surgery

An additional study compared F alone to non-flowable topical hemostatic patches or sponges (Surgicel Nu-Knit⁴ and Gelfoam⁵) alone in cardiac surgery¹³. Results suggest a cost-saving of $5.38 million annually ($8960 per average case) if F was utilized exclusively in an average of 600 mixed cardiac surgeries (e.g. isolated coronary surgery, isolated valvular surgery, combined coronary/valvular surgery and aortic surgery, alone or in combination) in a single hospital, where the complications avoided (33 major complications, 76 minor complications) by using F alone are the main contributors to the cost reduction.

In another study comparing the use of F to a different flowable hemostatic matrix (Surgiflo) in cardiac surgery, the author concluded that utilization of F vs. Surgiflo in cardiac surgery procedures could potentially avoid 11 major complications, 31 minor complications, 9 surgical revisions, 79 blood product transfusions and 260.3 hours of cumulative operating time, which corresponds to a net annualized saving of $1.5 million ($6256 per surgery) for an average hospital that perform 245 cardiac surgeries per year¹⁴.

Additional comparisons between the use of hemostatic matrix alone or in combination and the use of conventional hemostatic microsurgical techniques are warranted as a future study. The growing literature on this topic could also be organized by a formal review in a future study.

Blood loss complications were conservatively not included in the base-case analysis of the cost–consequence model. However, some estimates for blood loss complications were tested in the model to determine the size of the potential expected impact on annual hospital cost savings when using F vs. F + G/T. The cost of blood loss complications estimated by Fuller et al.²¹ range from $408 per event for hemorrhage and $10,789 for venous thrombosis (inflated to 2017 USD). Even assuming all blood loss complications to be severe for a medium-volume hospital, the additional expected savings would be small ($4208 savings).

All health economic models are a simplification of complex healthcare decisions. Our model heavily relies on the retrospective analysis by Ramirez et al.⁴ and so assumes the same distribution of differences in hospital utilization for minor, major and severe spine surgeries treated with F vs. F + G/T as well as the same distribution of minor, major and severe spine surgeries across low-, medium- and high-volume hospitals for the sensitivity analysis. The model also assumes that the upfront costs such as the training of spinal surgeons to use hemostatic agents are minimal and would not substantially impact the annual cost savings of using F vs. F + G/T, particularly over the long term.

Conclusions

This cost–consequence model builds on a retrospective study that shows a surgical resource advantage during spine surgery when using F alone rather than F + G/T. F in comparison with F + G/T is expected to provide annual cost savings for medium-volume hospitals performing 130 of spinal surgeries per year of $317,959 or $2445 per surgery. The more efficient control of bleeding during spine surgery not only improves patient outcomes but reduces hospital resource utilization and, therefore, costs.

Transparency

Declaration of funding

The study was funded by Baxter Healthcare Corporation.

Author contributions: D.Y. is an author on the study this manuscript is based, and both D.Y. and M.G.R. provided clinical expertise in their review of the manuscript. X.N. developed the cost model, and both X.N. and J.E. drafted the manuscript. All authors reviewed this manuscript and provided feedback.

Declaration of financial/other relationships

M.G.R. and D.Y. have disclosed that they are paid employees and stockholders at Baxter. J.E. and X.N. have disclosed that they are employees at Stratevi, which was retained for this work.

JME peer reviewers on this manuscript have received an honorarium from JME for their review work, but have no relevant financial or other relationships to disclose.

Acknowledgements

The authors would like to acknowledge Barbara Blaylock for providing edits to the manuscript.

Notes

Additional information

Funding

This study was funded by Baxter Healthcare Corporation.

Notes

1 Floseal is a registered trade name of Baxter Healthcare Corporation, Deerfield, IL, USA

2 Floseal is a registered trade name of Baxter Healthcare Corporation, Deerfield, IL, USA

3 Surgiflo is a registered trade name of Ethicon Incorporated, Somerville, NJ, USA

4 Surgicel and Nu-Knit are registered trade names of Ethicon Incorporated, Somerville, NJ, USA

5 Gelfoam is a registered trade name of Pfizer, New York, NY, USA